Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The Canadian conversion loan of 1958 : a study in debt management Christofides, Loizos Nicolaou 1973-03-02

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
831-UBC_1973_A1 C47_3.pdf [ 12.55MB ]
Metadata
JSON: 831-1.0100958.json
JSON-LD: 831-1.0100958-ld.json
RDF/XML (Pretty): 831-1.0100958-rdf.xml
RDF/JSON: 831-1.0100958-rdf.json
Turtle: 831-1.0100958-turtle.txt
N-Triples: 831-1.0100958-rdf-ntriples.txt
Original Record: 831-1.0100958-source.json
Full Text
831-1.0100958-fulltext.txt
Citation
831-1.0100958.ris

Full Text

THE CANADIAN CONVERSION LOAN OF 1958 A STUDY IN DEBT MANAGEMENT by LOIZOS NICOLAOU CHRISTOFIDES B.A. , University of Essex, 1968 M.A., University of Essex, 1969 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in the Department of ECONOMICS We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA April, 1973 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver 8, Canada ii ABSTRACT THE CANADIAN CONVERSION LOAN OF 1958  A STUDY IN DEBT MANAGEMENT Loizos N. Christofides World War II was partially financed through the issue of Victory Bonds. By 1958, Victory Bonds amounting to roughly 50% of the public debt were still outstanding, maturing at discrete intervals over the following seven years. In September, 1958, the Canadian Government launched the Conversion Loan — a successful attempt to refund the Victory Bonds. This enormous debt manage ment operation raised the average term to maturity of the public debt from 8 to 14.75 years. Debt management operations, and the Conversion Loan in particular, have received little attention in the Canadian context. The scant existing liter ature has not rigorously examined the effects of the Loan on the level and term structure of interest rates, nor has it investigated its impact on the real sector of the economy. In this thesis regression analysis and simula tion -- using the Bank of Canada RDX2 model -- were used to investigate these problems. The following conclusions were reached. There is convincing evidence that the Loan increased long rates and some less convincing evidence that it decreased short rates. In contrast to the U.S. there is no doubt that, in Canada, debt management operations significantly affect the term structure of interest rates. Other determinants of the term structure are expecta tions, monetary policy, transactions requirements, private sector wealth and the U.S. term structure of interest rates. The Loan was contractionary. Its effect during 1958 is estimated at 1% of GNE, increasing to 5% be.twee'n 1959 and 19:61, and decreasing thereafter. The overall cumulative effect is likely to have exceeded $1 billion. Contrary to conventional wisdom, it was the interest sensitivity of investment rather than the reduction in Canada's competitive position in world markets -- the Loan raised interest rates, attracted "hot capital" and led to an exchange rate appreciation --that engendered the depression. iv . TABLE OF CONTENTS List of Tables vii List of Illustrations xAcknowledgements xiiCHAPTER ONE THE CANADIAN CONVERSION LOAN OF 1958 1 Summary 13 Notes to Chapter One 14 Tables for Chapter One 6 CHAPTER TWO . THE CONVERSION LOAN: A SURVEY OF THE LITERATURE 59 c-Notes to Chapter Two 68 CHAPTER THREE EFFECTS OF THE CONVERSION LOAN ON THE TERM STRUCTURE OF INTEREST RATES 69 IntroductorySection One The Conversion Loan and Time Series of Interest Rates 70 Section Two Determinants of the Term Structure of Interest Rates 72 A Forward Rates: The Hicksian Formulation 72 B Expected Rates: Perfect Foresight 74 C(i) The Expectations Hypothesis 76 C(ii) Liquidity Premia Hypothesis 83 C(iii) Segmented Markets Hypothesis 85 C(iv) Preferred Habitat Hypothesis 86 C(-v) The General Equilibrium Approach 87 Section Three Estimating the General Equilibrium Model 90 V TABLE OF CONTENTS Section Four Estimating the Preferred Habitat Model 98 Section Five Why are Supply Variables Important in Canada but not in the U.S.? 105 Notes to Chapter Three 111 Data and Sources 114 Tables for Chapter Three 116 Graphs for Chapter Three 13CHAPTER FOUR FINANCIAL RESPONSES TO A NEW TERM STRUCTURE OF INTEREST RATES 139 Section One The Response of Asset Holders 140 Section Two The Response of Asset Issuers 141 A Introductory 14B When do Borrowers Issue Liabilities in U.S. Dollars? 144 C What Determines the Short-Long Mix of Bond Issues? 149 Section Three Implications of Section Two B for a Study of the Conversion Loan 151 Section Four On Substitutability in Financial Markets 152 Notes to Chapter Four 155 Tables for Chapter Four 8 Graphs for Chapter Four 163 CHAPTER FIVE REAL RESPONSES TO THE CONVERSION LOAN 169 Introductory 169 vi TABLE OF CONTENTS Section One Some Preliminary Calculations 170 Section Two Simulation Results 172 Section Three A Conclusions for Chapter Five 187 B Concluding Comments 189 Notes to Chapter Five 191 KDX2 Variables Referred to in this Chapter 193 Tables for Chapter Five 195 LITERATURE CITED 208 APPENDIX TO CHAPTER TWO: STUDIES OF "OPERATION TWIST" 213 Notes for Appendix to Chapter Two 217 vii LIST OF TABLES CHAPTER ONE TABLE 1 Financial Statement of the Government of Canada 16 TABLE 2 Distribution of Victory Loan Purchases 17 TABLE 3 Victory and War Loan Issues: World War II 18 TABLE 4 Dollar Bonuses Paid to Victory Bond Owners Participa ting in the Conversion Loan 19 TABLE 5 Results of Conversion Loan 20 TABLE 6 Canadian Narrow Money Supply 1 TABLE 7 Classification by Term to Maturity of Total Govern ment of Canada Securities Outstanding 22 TABLE 8 Bank of Canada's Holdings of Government of Canada Direct and Guaranteed Securities 25 TABLE 9 Total Minus Bank of Canada's Holdings of Govern ment of Canada Direct and Guaranteed Securities 27 TABLE 10 Government of Canada Accounts 29 TABLE 11 Chartered Banks: Holdings of Government of Canada Direct and Guaranteed Securities 31 TABLE 12 General Public Holdings of Government of Canada Securities by Term to Maturity 33 TABLE 13 Total Debt Minus Bank of Canada Holdings by Term to Maturity—Under Different Assumptions About the Behaviour of the Bank of Canada 36 TABLE 14 Bonds Outstanding by Issuer and Currency 37 TABLE 15 Federal Government Gross New Issues of Direct and Guaranteed Bonds 38 TABLE 16 Provincial Gross New Issues of Bonds 41 TABLE 17 Gross New Issues of Bonds by Canadian Corporations 46 TABLE 18 Average Term of Provincial Debt as at Fiscal Year Ends 52 LIST OF TABLES TABLE 19 TABLE 20 TABLE 21 CHAPTER THREE TABLE 1 TABLE 2(a) TABLE 2(b) TABLE 3 TABLE 4(a) TABLE 4(b) TABLE 5 TABLE 6 TABLE 7 TABLE 8 TABLE 9 TABLE 10 TABLE 11 TABLE 12 Municipal Debt by Province Disaggregated into Serial and Sinking Fund U.S. Government Marketable Securities Based on Treasury Survey Data U.S. Federal Government Bond Rates Signs of Partial Derivatives in the Reduced Forms for r and r Regression Results for the Model of Section 2C(v). Equations 19, 20 and 21 Implied Reduced Form. Equations 22 and 23 Estimated Reduced Form Equations for the Model of Section 2C(v). Equations 24 and 25 Regression Results for the Partial Adjustment Version of the Model in Section 2C(v). Equations 26, 27 and 28 Implied Reduced Form. Equations 29 and 30 Estimated Reduced Form Equations for the Model of Section 2C(v). Equations 31 and 32 Decomposition of Implied Reduced Form Equations 22 and 23 Decomposition of Estimated Reduced Form Equations 24 and 25 Decomposition of Implied Reduced Form Equations 30 and 29 Decomposition of Estimated Reduced Form Equations 32 and 31 Regression Results for the Model of Section 2C(iv) Equations 33, 41, 42 and 43 Decomposition of Regression Equation 42 Decomposition of Regression. Equation 43 ix TABLE 13 TABLE 14 TABLE 15 CHAPTER FOUR TABLE 1 TABLE 2 TABLE 3 TABLE 4 TABLE 5 CHAPTER FIVE TABLE 1 TABLE 2 TABLE 3 TABLE 4 TABLE 5 TABLE 6 LIST OF TABLES Standard Deviation for Each of the Four Maturity Classes in the Series Inside, Public, Sutch and FRB, and of the Ratios of Bonds Under 10 Years to Over 10 Years 129 Coefficients of Variation for Each of the Four Maturity Classes in the Series Inside, Public, Sutch and FRB, and of the Ratios of Bonds Under 10 Years to Over 10 Years 132 Regression Results for Section 5. Equations 44, 45, 46 and 47 135 Proportion of Outstanding Debt Issued in Currencies Other than Canadian. Spread Between Canadian and U.S. Federal Government Bond Yield 153 Regression Equations for Specification One—Equa tions 1 and 2—and Two—Equations 3 and 4. 159 Ratio of Serial to Sinking Fund Issues by Province 160 Regression Equations for Provinces and Corporations Pertaining to Section Two C. Equations 5 and 6 161 Regression Results Pertaining to Section Four. Equations 7, 8, 9 and 10 162 Effects of the Conversion Loan on YGNE 195 Shock-Control Values for RS, RMS, RML Generated by Equations 22-25 196 Shock-Control Values for RL Generated by Equations 22 - 25 197 Effects of the Conversion Loan on YGNE 198 Effects of the Conversion Loan on UGNE" 199 Conversion Loan Effects Under NLH1A: RS, RMS, RML, RL, RMC, RHO, RNU 200 X LIST OF TABLES TABLE 7 TABLE 8 TABLE 9 TABLE 10 TABLE 11 TABLE 12 TABLE 13 Conversion Loan Effects ABLB, PCPI. PCPICE Under NLH1 : NLH1A: Conversion Loan Effects Under M, UGPP Conversion Loan Effects Under NLH1 UBAL - XBAL$, PFX, PFXF . Conversion Loan Effects RML, RL, RMC, RHO, RNU Under NLH1B: B Conversion Loan Effects Under NLH1 ABLB, PCPI, PCPICE Conversion Loan Effects Under NLH1B: M, UGPP Conversion Loan Effects Under NLH1 UBAL - XBAL$, PFX, PFXF B. ABBCD, RABEL, C, I+IIB, X, UBAL, XBAL$, RS, RMS, ABBCD, RABEL. C, I+IIB, X, UBAL, XBAL$, 201 202 203 204 205 206 207 xi LIST OF ILLUSTRATIONS CHAPTER THREE FIGURE 1 Extrapolative and Regressive Expectations FIGURE 2 Extrapolative and Regressive Weights FIGURE 3 Expectations and the Yield Curve GRAPH 1 Estimated Distributed Lag in Equation 41 GRAFH 2 Estimated Distributed Lag in Equation 42 GRAPH 3 Estimated Distributed Lag in Equation 43 79 80 81 136 137 138 CHAPTER FOUR GRAPH 1 GRAPH 2 • GRAPH 3 GRAPH 4 GRAPH 5 GRAPH 6 Estimated Weights for Equation 3 Estimated Weights for Equation 4 Ratio of New Provincial Issues in the U.S. to their Total New Issues Ratio of New Corporate Issues in the U.S. to their Total New Issues Proportion of Provincial New Short to Provincial New Long Bonds Proportion of Corporate New .Short to Corporate New Long Bonds 163 164 165 166 167 168 CHAPTER FIVE GRAPH 1 GRAPH 2 GRAPH 3 GRAPH 4 GRAPH 5 Federal Government Debt held by the Public 0-2 Years to Maturity Federal Government Debt held by the Public 2.-5 Years to Maturity Federal Government Debt held by the Public 5-10 Years to Maturity Federal Government Debt held by the Public Over 10 Years to Maturity Federal Direct Debt Held by Resident Public and Chartered Banks. 0-3 Years to Maturity 175 175 175 175 176 LIST OF ILLUSTRATIONS Federal Direct Debt Held by Resident Public and Chartered Banks. 3-5 Years to Maturity Federal Direct Debt Held by Resident Public and Chartered Banks. 5-10 Years to Maturity Federal Direct Debt Held by Resident Public and Chartered Banks. Over 10 Years to Maturity xiii ACKNOWLEDGEMENTS In the process of writing this thesis I have incurred an enormous debt to the chairman of my thesis committee, Professor Keizo Nagatani. From the inception of this study to its completion he has been an unfailing source of ideas, criticism, encouragement and advice. My other major debt is to Professor John F. Helliwell, the second member of my thesis committee. He, too, has been an exemplary member. In addition to making a number of suggestions at various stages of this study, he has provided me with substantial help in the construction of the RDX2 simulations reported in chapter five. Professor Ronald A. Shearer joined my thesis committee relatively late but has nevertheless made a number of helpful suggestions. I also wish to acknowledge helpful conversations with Dr. David E. Bond at the initial stages of this study, and the work of John Lester at the U.B.C. Computing Centre pertaining to the RDX2 simulations. Last, but not least, reference must be made to the excellent editorial and secretarial assistance of Gay Christofides. I have received financial support from the University of British Columbia and the Canada Council. It is customary to exonerate all but the main author of a study from all responsibility, even when the author's debts are very large. My only hope is that when the Christofides Bonds reach maturity, I too will be able to reconvert them. 1 CHAPTER ONE THE CANADIAN CONVERSION LOAN OF 1958 "This is a tremendous operation requiring the enthusiastic co-operation of everybody concerned... I have received from all quarters pledges of enthu siastic and vigorous support... this is a great national undertaking. It is the concluding phase of the victory loan campaign of the war years..." D. Fleming, Then Minister of Finance. The period 1950 - 1957 was one of intense activity in the real capital markets. Unemployment, while slowly rising, was low in comparison to that of 1957 - 1962. With the exception of 1955, monetary policy was rather restrictive and after the Korean war prices rose very slowly. Partly because of monetary policy the capital account was in surplus, in contrast to the current account. The overall picture being usually one of a potential balance of payments surplus, the Canadian exchange rate was following an appreciating trend until 1957. The government's financial picture was fairly "sound". Between 1947 and 1957 there was a budget deficit only twice and even then of small magnitude. The budget surpluses were disposed of by reducing the outstand ing public debt. Parizeau^ and Fullerton-* indicate that government borrow ing in this ten-year period both by means of note-issuing and otherwise was 2 modest. Coupled with the continual decrease in the public debt was a simi lar decline in the average maturity of the debt so that by 1957 the long bond market was quite thin. Borrowing requirements for the year 1958 were expected to be heavy. During the fiscal year ending March 31, 1958, the federal government ran the first of a series of sizeable budget deficits. The deficit for the period April 1, 1958 - March 31, 1959 was as high as $609.3 million and the overall cash requirements were even higher at $1,273.3 million^. Table 1 gives further details. In addition, between 1959 and 1966 large quantities of World War II bonds were maturing. During the period 1941 - 1945 the federal government had borrowed funds in order to finance its war efforts. There were nine "Victory Loans", as they were called, amounting roughly to $12 billion. The funds had been supplied by corporations and individuals, the former being the somewhat larger creditors--Table 2 gives a detailed breakdown. By 1958 almost half the $12 billion had been repaid but there •remained the 5th to 9th Victory Loans, involving some $6.5 billion and maturing at discrete intervals"* between 1959 and 1966. To place matters in perspective it should be pointed out that in December 1957 the federal debt, excluding Canada Savings Bonds , held by the Public and chartered banks was only $8.6 billion. Thus the substantial cash requirements of the federal government and the "coming of age" of the Victory Bonds posed a new and serious problem to the authorities. Many alternative courses of action were open to the government. The short planning horizon frequently attributed to governments would have indicated a "do-it-as-you-go-along" course of action, namely issuing bonds 3 and reducing cash balances as budget deficits were implemented and Victory Bonds matured. Instead, following April 1, 1958, the authorities issued bonds to finance successive budgetary deficits and pursued a monetary policy which was at variance with their fiscal convictions. Column four of Table 1 documents the validity of the first statement, while column five of Table 1 and Table 6 shows that of the second: In 1959, for example, overall cash requirements were $1.27 billion and they were met through the issue of $1.33 billion worth of bonds.7 However, cash balances were 8 9 increased by $0.17 billion and the money supply declined. Also between July 14 and September 15, 1958, the Conversion Loan was launched. The Loan was an attempt to persuade the owners not only of the 5th and 6th Victory Bonds, but also those of the 7th, 8th and 9th to exchange their old bonds for the new Conversion Loan ones. While some of the Victory Bonds were callable as early as 1956, the authorities were not obliged to redeem them until their final maturity dates. That date for the 9th Victory Loan was as late as September 1, 1966! The ownership dis tribution of the Victory Bonds immediately before the Conversion Loan is not known with accuracy. But data on ownership when the Victory Bonds were first issued, Table 2, and other fragmentary evidence, suggest that a con siderable number of these bonds were held by private individuals, often in remote parts of Canada. This fact, along with the Loan's size of some 49% of the federal debt"^ and the concurrent need for funds to finance budget deficits, made the Conversion Loan one of the most difficult financing operations ever undertaken in Canada. Considering how inexperienced at this kind of undertaking the authorities were one feels certain that they must have advanced impressive justification for their actions! 4 The government, in various public statements, suggested at least four reasons for the Conversion Loan. First, it was thought to be "... in every sense anti-inflationary"^ since one alternative, namely redeeming maturing bonds, would involve increasing the money supply. This belief was quite well-founded but worrying about the price level during a flexible exchange rate regime is not all-important. Second, ironing out humps in the matu rity structure of the public debt facilitates rolling over the debt. This, of course, is quite true but there may be a case for doing more with the debt than merely rolling it over, namely using its maturity structure to control economic activity through the term structure of interest rates. 12 Third, another alleged merit of the Loan was that it removed uncertainty. The uncertainty referred to consisted of not knowing what the government would do with the maturing Victory Bonds. Thus, while the Loan removed this kind of -uncertainty so would any other publicly announced plan. Fourth, and the Prime Minister thought therefore, the Loan would "...add greatly to the strength of Canada's national economy... including the fullest development of our resources, more and better jobs and a higher standard of living for every Canadian."13 Straightforward application of Tobin's (67) model predicts the opposite outcome. It is noteworthy, however, that none of Tobin's papers had appeared prior to the Loan. One wonders whether other, more sound, argu ments were not disclosed! It does not appear so. Even if the reasons advanced for the Loan were sound it is difficult to see why the individual bond owner should cooperate. Cooperation was sought by various means. To begin with the government made a case for act ing collectively, through its references to the resulting higher standard of living and through appeals to patriotism. The quotation at the begin-ning of this chapter is indicative of the campaign undertaken. Where moral suasion could be used it appears to have been exercised. How else can one explain the management of the Unemployment Insurance Fund? Unemploy ment was expected to rise during the winter of 1958-1959 and hence the Fund should have held a relatively liquid portfolio in anticipation of large disbursements to its members. Yet the Fund converted its holdings of the highly liquid Victory Bonds, reducing the proportion of bonds with less 14 than three years to maturity from just under 50% to a mere 0.8%. Later on when the Fund was forced to liquidate securities at a loss it sold, not the 1961 Conversion Bonds which were relatively short, but other longer term ones! This mismanagement was pointed out by H. Scott Gordon (23). Finally, there was the unavoidable sugar-coating of the pill. All nine Victory Loans were struck at a common coupon rate of 3%. The Conversion issues had coupons ranging from 3% to 4%%. Also cash bonuses were paid. The owner of a $1000 bond, from the 5th Victory Loan, for example, received $25.00 upon converting it into Conversion Bonds maturing after 1965. Table 4 gives more details. Combining the information in Tables 4 and 5 the max imum cost to the government of the cash bonus programme can be estimated. It amounted to $93,862,500! With the exception of one firm, E. M. Saunders Ltd., all investment dealers participated in what, for them, was a very profitable venture. Just how the Loan was executed is a question that will not be dis ci cussed in detail here. Excellent expositions of this can be found in Binhammer (5), Officer and Smith (48), Wonnacott (69) and others. For all the reasons mentioned earlier around 90% of the bonds maturing were converted and the Loan was pronounced a "success"—Table 5 incorporates 6 all the information available on this score. How did the Conversion Loan affect the distribution by term to matu rity of the public debt? Table 7 gives the distribution of total holdings, which include the portfolios of the Bank of Canada and of the Government Accounts. Between 1958Q2 and 1958Q3 there is a small increase in the pub lic debt—defined here so as to exclude consols and non-market issues—of about $395 million. There is no change in the par values of Treasury Bills outstanding and so the Conversion Loan manifests itself as a substantial increase in bonds with over ten years to maturity and a decrease in bonds with maturities under ten years. The former increased by $3,518 million and the latter decreased by $3,123 million. It can be shown that these changes can be attributed almost entirely to the Conversion Loan, rather than any other government issues -- $395 million shorts were issued. However, looking at the maturity distribution of total debt outstand ing may not be very informative because it contains the accounts of the Bank of Canada and the government—both traditionally regarded as "outside" the system. While the balance sheet of the Bank does contain its holdings 16 of government bonds by maturity class no such information is available in the government accounts reported. Hence it is not possible to arrive at the maturity distribution of the government debt held "inside" the system, 18 namely by chartered banks and the Public, through this particular route. An alternative route involves aggregating the holdings of the Public and those of chartered banks by term to maturity. This is not possible either; 19 while the necessary figures are available for the Public, they are not for 20 the banks. We are therefore forced to use two alternative approximations 7 to the figures for the maturity distribution of the debt, Table 9, which 21 includes the Government Accounts and Table 12, which excludes the hold ings of chartered banks. They will both be used in chapter three, where the effects of altering the relative supplies of debt with different terms to maturity on the term structure of interest rates will be investigated. Both.series confirm the view that the Conversion Loan was a debt management operation increasing debt with more than ten years to maturity and decreas ing debt under this mark. However, they also bring out two factors that Table 7 obscures. Treasury Bills held inside the system increased from $1,124 million in 1958Q2 to $1,425 million in 1958Q3 and to $1,771 million in 1959Q4, decreasing after that last date but never dropping below their 1958Q3 level. A similar pattern exists in the Treasury Bill holdings of the Public, as Table 12 shows. Secondly, bonds with less than two years to maturity decreased in 1958Q3, but by 1959Q1 they were above their pre-Conversion Loan level. Thus, while the Loan decreased the sum of all bonds with less than ten years to maturity and left Treasury Bills unaffected, this, paradoxically, is not reflected in the time profile of each and every maturity class under the ten year mark. To explain the two irregularities just mentioned it is necessary to delve into the activities of the Bank of Canada during the period 1958Q2 - 1959Q4. Although 90% of the Victory Bonds maturing were converted, this figure was not accomplished without the active intervention of the Bank of Canada in the bond markets. More specifically, until November, 1958,the Bank sup ported bond prices. In the process it clearly had to cash all bonds that would not be held and it did so until it was realised that bond prices and quantities held by the Public could not both be pegged. The authorities 8 then chose to freeze the money supply, i.e. they bought no more bonds and they allowed interest rates to float. This choice indicates their deter mination to adhere to restrictive monetary policy as the Governor corro borates: "...by the beginning of November the strong downward movement of other bond prices...had made it clear that the prices of long-term Conversion issues could not be maintained...without a dangerous degree of monetary expansion and central bank purchases were discontinued."22 Table 6 shows that while bond prices were being pegged the money supply increased; the money supply was subsequently kept below its 1958Q4 peak until 1960Q4. Table 8 shows how the central bank "financed" its bond price support programme. Between 1958Q2 and 1958Q3 the Bank decreased its holdings of Treasury Bills and other bonds with less than two years to maturity by $.1,076 million and increased those of bonds with over two to five, over five to ten and.over.ten years to maturity by $234 million, $89 million and $917 million respectively. It is worth noting that after the price support period was over the Bank did not dispose of the long-term bonds it had acquired. Rather, beginning in 1959Q1, it gradually increased its holdings of Treasury Bills and bonds with less than two years to maturity. Quite apart from the Bank's price support programme other governmental agencies, e.g. the Unemployment Insurance Fund, were under pressure to con vert their portfolios. It is therefore of interest to analyse the combined effects of the activities of these institutions. As already indicated, little is known about the maturity composition of securities held in the Government Accounts. So in Table 13 an attempt is made to disentangle the 9 effects of the Conversion Loan from those of the price support programme pursued by the Bank of Canada. On the assumption that the behaviour of the Accounts did not change during the Loan, the figures in Table 13 indi cate the true pressures in the financial markets. The effect of dropping this assumption is indicated later. Table 13 was constructed out of Tables 7 and 8. Each row gives the difference between the total federal government debt, Table 7, and that part of the debt held by the Bank of Canada, Table 8. This difference for 195802, for example, is denoted by (Total: 58Q2)-(BOC: 58Q2) and it appears disaggregated by term to maturity. Thus row one of Table 13 gives the term to maturity structure of the debt held by the Public, the chartered banks 23 and the Government Accounts in 1958Q2. We now wish to investigate the Loan's effect on the term to maturity of the debt, abstracting from the price support activities of the Bank of Canada. It is, therefore, assumed that between 1958Q2 and 1958Q3 the Bank was completely inactive so that the 1958Q2 figures on bonds held by the Bank are also applicable in 1958Q3. Row  two describes what would have happened had a "Pure Conversion Loan" been effected. The "Pure Loan" would have decreased maturities under ten years, except Treasury Bills, and increased bonds with over ten years to maturity. However, the Bank did act: It pegged bond prices until November, 1958, thereby cashing all bonds that would not be held. This increased its hold ings of long bonds and was financed through the sale of Treasury Bills and securities under two years to maturity. Row three describes the actual effect of the Bank's activities on holdings "inside" the system. Its sale of Bills and 0-2 year shorts increased the former above their 1958Q2 level and moderated the decrease in the latter that would have occurred. Its purchase of bonds with maturities over two years accentuated the decrease in bonds with maturities between two and ten years—the Conversion Loan decreased those—and moderated the increase in bonds over ten years—the Conversion Loan increased those—that would have occurred under the cir cumstances of row two. In 1959Q1, by which time the bond price support programme had been dropped, the Bank of Canada increased its holdings of Treasury Bills at the expense of bonds with less than two years to maturity. During the 24 next few years the Bank increased its liquid holdings relative to the long ones, gradually readjusting towards the portfolio composition it had prior to the price support period. Needless to say that this trend cannot be detected in the various maturity classes of the debt that was held "inside" the system because the total quantities outstanding of all bonds and Treasury Bills were changing. Thus, rows four to seven tell what the maturity distribution of the debt "inside" the system would have been during 1959, had totals remained at their 1958Q3 levels. The tendency for Treasury Bills and the shortest bonds to increase—row three—would have been substantially reversed during 1959. In this respect a more 25 pure Loan would have been implemented. This suggests that what the Bank's activities amounted to was merely delaying the Loan by a few quar ters. This idea will be taken up later, as it turns out to be quite important. Finally, rows eight to eleven show what actually happened to holdings "inside" the system. In the above discussion the explicit assumption was made that behavi our in the Government Accounts did not change following the Loan. Dropping this assumption would mainly accentuate the effects of the behaviour of 26 the Bank of Canada. This argument cannot be pursued further given available data. Since we will later be directing our attention to the effects of the Conversion Loan on the term structure of interest rates it is as well, at this point, to take a look at what other changes were taking place in the economy. Fiscal and monetary policy as well as changes in the quantities of Treasury Bills have already been discussed. Attention has also been drawn to the small changes in the size of the federal government debt. There remain at least two areas of interest: The size and composition of provincial, municipal and corporate debt and the size and composition of the U.S. federal government debt. Table 14 more or less exhausts published information of relevance on the composition of the debts of provincial and municipal government and of corporations. The totals are dominated by a very large trend component with what appear to be few deviations from it. There is no published infor mation on the term to maturity structure of these debts. However, reliable series are here constructed on the term to maturity composition of new bond issues made by the federal and provincial governments and by cor porations. While In principle it is possible to calculate from these series the term to maturity of the outstanding stocks of provincial and corporate bonds this would be a very hazardous undertaking. Thus, we have concentrated on the composition of the flows. Tables 15, 16 and 17 tell the story. There are no striking changes in the behaviour of these vari ables. This conclusion for the case of provincial debt is corroborated by Table 18. It shows that the average term to maturity of all provin-12 cial debt was 18.8 years in 1958 and 18.1 in 1960. Municipal debt poses many more problems. While micro data similar to the provincial and cor porate ones are available the problem arose that municipalities issue a -very large proportion of their debt in serial form. The term to maturity of a serial bond is not obvious and while an average term could conceptually be calculated that would require much more information than is at present available. However, we do have information on the amounts issued in serial form and those issued in sinking fund form. This is useful because that classification corresponds roughly to a term to maturity classification: A serial bond spreading over twenty years has an average term shorter than that of a twenty year sinking fund bond. Table 19 gives yearly data on the serial/sinking fund debt structure for all municipalities, by province. It remains to consider the behaviour of some of the U.S. financial variables. In the appendix to chapter two it is shown that U.S. studies have found little relation between the maturity structure of the U.S. debt and the U.S. term structure of interest rates. Yet it is well knoxm that there is some relation between U.S. and Canadian interest rates—possibly because of common influences emanating from the demand side. For this 27 reason we concentrate on the effects, if any, that the U.S. term struc ture of interest rates has on the Canadian one. Table 21 shows that dur-28 ing the period 1958Q1 - 1959Q4 all three U.S. rates rose. This poses the problem: If Canadian long rates rose, was that because of the Loan or because U.S. rates were rising concurrently? 13 SUMMARY In this chapter the background to the Conversion Loan was briefly considered and the extraordinary nature of the financing requirements for 1958 was brought out. The Conversion Loan was then discussed along with the publicly announced reasons for it. The Loan is the most important debt management operation ever carried out in Canada, nearly doubling the average maturity of the public debt. We are therefore interested in assessing its effects on interest rates and other economic variables. How ever, it was seen that other concurrent changes did occur. i) There was the increase in Treasury Bills held by the Public and chartered banks due to the increase in the totals, the Bank's bond price support programme and its restrictive monetary policy. ii) Also noteworthy is the increase in bonds with less than two years to maturity. The reasons for this are the same as in (i) immediately above. iii) There may also have been changes in the maturity pattern of new issues of bonds by the federal, provincial and municipal governments and corporations. A.detailed examination of this possibility is undertaken in chapter four. iv) There were, finally, changes in the U.S. term structure of inter est rates. Before attributing to the Loan any changes in interest rates and/or in other variables it is necessary to evaluate the contribution to any such changes of the factors mentioned above. This task is taken up in chapter three. In the next chapter the existing studies of the Conversion Loan are examined., 14 NOTES TO CHAPTER ONE 1. Quoted in Eullerton (21), p. 242. 2. Parizeau (50), p. 24. 3. Eullerton (21), pp.45 - 48. 4. It may be objected that these figures are ex post ones. However, there are indications that they are also good approximations to the ex ante figures. See Fullerton (21), p. 237. 5. See Table 3. 6. Of central importance to this thesis is the maturity composition of the public debt. Although Canada Savings Bonds are issued with a for mal maturity date they are not marketable and the government will redeem them at face value on demand—there are some costs involved in cashing a bond prematurely, not in terms of loss of principal but rather in terms of the average effective interest rate earned. Because these bonds are redeemable on demand their term to maturity is ambigu ous. It is presumably for this reason that published tables giving the term to maturity of the public debt exclude such bonds. Since these tables are extensively used here, the concept of the public debt em bedded in them is also used for convenience. 7. See Table 1, columns three and four. 8. See Table 1, column five. 9. See Table 6. 10. In September, 1958, the total federal debt was $13,357 million (Table 7) while the Conversion Loan involved issues amounting to $6,416. million. (Table 5). 11.. The Prime Minister. Quoted by Fullerton (21), p. 241. 12. Fullerton (21), p. 241, quotes the Minister of Finance as saying on July 14, 1958, "...This large volume of early maturities overhanging the market has made it very difficult to plan an orderly program of debt management and has contributed greatly to the general feeling of uncertainty which has prevailed in our bond market for the past few years." 13. The Prime Minister. Quoted by Fullerton (21), p. 243. 14. Figures are taken from Fullerton (21) , p. 254. 15 15. In fact Fullerton mentions that most informed sources attribute the whole scheme to an investment dealer who apparently had no difficulty selling the idea to Mr. Coyne, the inflation-fearing Governor of the Bank of Canada. Professor Barber has privately suggested that one alternative to the Loan might have been conversion into bonds which, like Canada Savings Bonds, have guaranteed capital values. 16. See Table 8. 17. See Table 9. 18. This category includes all financial institutions other than chartered banks, non-financial enterprises as well as private individuals. 19. See Table 12. 20. Table 11 contains all the available information on this score. The reason for the particular disaggregation reported is simply that that is all the information banks are required to report. 21. The figures in this table will, somewhat loosely, be referred to as the debt held "inside" the system. 22. Quoted in Fullerton (21), p. 245. 23. i.e. "inside" the system. This row and all other starred rows, are the same as the corresponding ones in Table 9. They are reproduced in Table 13 for the reader's convenience. 24. See Table 8, particularly 1959Q1, 1959Q4, 1960Q3 and 1962Q3. 25. We say roughly because of the figure $2,123 million appearing under "over two to five" in row seven. This departure from row two was due to the Bank's adjustment towards a more liquid portfolio. By 1960Q3 the Bank was able to increase its holdings in this category at the expense of the "over five to ten" one and by 1962Q3 bonds in this last category were also increased, this time at the expense of bonds over ten years—see Table 8. This gradual increase in liquidity may have been effected by the passage of time alone. However, since no steps to reverse this "natural" process were initiated, we are entit led to assume that it was not objectionable. 26. Recall how the Unemployment Insurance Fund was managed. 27. Table 20 gives, for the record, the maturity structure of the U.S. debt. It is seen that during 1958 there are large changes but, if anything, the net changes are in the opposite direction from those effected by the Conversion Loan in Canada. 28. Note, however, that the spread between the U.S. long and medium rates fell. TABLE 1 Financial Statement of the Government of Canada ($000's) Fiscal Year Budget surplus (+) or deficit (-) Non-budgetary receipts (+) or disbursements (-) including changes in advances to Foreign Exchange Control Board and Exchange Fund Overall cash requirements i.e. sum of first and second columns Increase (+) in unmatured debt outstand ing Decrease (+) in cash balances 1954 +45,800 * * -234,400 -104,100 1955 -151,800 * -79,700 +128,800 1956 -33,100 A * +911,100 -339,800 1957 +257,500 * * -1,039,200 +98,600 1958 -38,600 -126,300 -164,900 -123,300 +164,700 1959 -609,300 -664,000 -1,273,300 +1,329,000 -166,000 1960 -413,100 +37,600 -375,500 + 316,000 +41,500 1961 -340,400 +46,100 -294,300 + 177,800 + 71,300 1962 -791,000 +313,400 -477,600 + 877,800 -416,900 1963 -691,600 -772,300 -1,463,900 +1,016,100 +400,200 1964 -619,200 +336,700 -282,500 + 778,300 -451,700 1965 -38,000 -384,200 -422,200 +238,100 +146,700 1966 -39,000 -120,900 -159,900 + 131,600 + 47,300 * = Not comparable t = In Canadian and foreign funds Source: Bank of Canada Statistical Summary Supplement (To be referred to as Supplement) 17 TABLE 2 Distribution of Victory Loan Purchases ($000's) Victory loan Date of issue Purchased by Total cash sales Individuals Corporations 1st June 15, 1941 279,500 450,900 730,400 2nd March 1, 1942 335,600 507,500 843,100 3rd Nov. 1, 1942 374,600 616,800 991,400 4th May 1, 1943 529,500 779,200 1,308,700 5 th Nov. 1, 1943 599,700 775,300 1,375,000 6 th May 1, 1944 641,500 763,500 1,405,000 7th Nov. 1, 1944 766,400 751,200 1,517,600 8th May 1, 1945 836,300 732,600 1,568,900 9 th Nov. 1, 1945 1,221,342 801,132 2,022,474 Source: Canada Yearbook 1957-58, p. 1162. For the 1st, 2nd, 5th and 7th loans there is a small difference between the numbers given for the totals in this table and in Table 3. This may represent the difference between cash sales and total cash and non cash sales. TABLE 3 Victory and War Loan Issues; World War II Issued Maturing $ Millions Sold Coupon Rate % 1st 1941 Dec. 15, 1946 193 2 June 15, 1951 644 3 2nd 1942 Sept. 1, 1944 150 1 1/2 March 1, 1948 270 2 1/4 March 1, 1954 670 3 3rd 1942 May 1, 1946 144 1 3/4 Nov. 1, 1956 847 3 4 th 1943 Nov. 1, 1946 19 7 1 3/4 May 1, 1957 1,111 3 5 th 1943 May 1, 1947 373 1 3/4 Jan. 1, 1959 1,197 3 6 th 1944 March 1, 1948 240 1 3/4 June 1, 1960 1,165 3 7 th 1944 Nov. 1, 1948 344 1 3/4 Feb. 1, 1962 1,316 3 8th 1945 Nov. 1, 1949 268 1 3/4 Oct. 1, 1963 1,296 3 9 th 1945 Nov. 1, 1950 336 1 3/4 Sept. 1, 1966 1,692 3 Source; Fullerton [21], Table 4.2 19 TABLE 4 Dollar Bonuses Paid to Victory Bond Owners  Participating in the Conversion Loan Conversion 3% 3 3/4% 4 1/4% 4 1/2% Loans 1961 1965 1972 1983 Victory Loans 5 th 1959 15.00 25.00 25.00 25.00 6 th 1960 12.50 22.50 22.50 22.50 7th 1962 * 12.50 12.50 12.50 8th 1963 * * 17.50 17.50 9 th 1966 * * 15.00 15.00 The 1962 issue of Victory Bonds was not eligible for exchange into the shortest conversion issue and the 1963 and 1966 Victory bonds were not eligible for conversion into either of the two shortest con version issues. Figures are dollars paid to bond owners converting $1,000 worth of Victory Bonds provided certain interest certificates had not been cashed. Source: Canada Yearbook 1959, p. 1131. TABLE 5 Results of Conversion Loan (Par Values in $ Millions) Victory Loans: Issues eligible for conversion 3% Dec. 1 1961 Converted into 3 3/4% 4 1/4% Sept. 1 Sept. 1 1965 1972 4 1/2% Sept. 1 1983 Residual Uncovered Total Victory Loan Issues 5 th 3% Jan. 1, 1956/59 654 94 58 100 42 947 6 th 3% June 1, 1959/60 366 447 172 133 46 1,165 7 th 3% Feb. 1, 1959/62 * 726 238 298 54 1,316 8th 3% Oct. 1, 1959/63 * 489 584 223 1,296 9th 3% Sept. 1, 1961/66 * * 410 1,037 245 1,692 Total 1,021 1,267 1,367 2,152 610 6,416 Notes: The 1962 issue of Victory Bonds was not eligible for exchange into the shortest conversion issue and 1963 and 1966 Victory Bonds were not eligible for conversion into either of the two shortest conversion issues. Source: Bank of Canada. Annual Report of the Governor, 1958, p. 28. o TABLE 6 Canadian Narrow Money Supply = Notes and Coin Outside  Banks + Demand Deposits + Non-Personal Term and  Notice Deposits ($ Millions) Year Qi Q2 Q3 Q4 1955 4,901 5,234 5,370 5,243 1956 5,007 5,141 5,209 5,204 1957 4,847 5,057 5,081 5,333 1958 5,213 5,508 5,959 6,035 1959 5,685 5,714 5,800 5,789 1960 5,541 5,742 5,952 6,073 1961 6,035 6,211 6,612 6,822 1962 6,390 6,677 6,747 7,071 1963 6,882 7,189 7,329 7,510 1964 7,355 7,695 7,789 8,187 1965 8,229 8,877 9,267 9,433 Major Series Revision Before 1962 the last item cannot be distinguished from the last but one. Average of Wednesday series, end of quarter. Source: Supplement. TABLE 7 Classification by Term to Maturity  of Total Government of Canada Securities Outstanding Unmatured Direct and Guaranteed securities ( issues and perpetuals) ex. non -market Per petuals Non-market issues Matured and out standing marke t issues Total out standing 2 years and Treasury Bills & notes & deposit certificates under 1 Other Over 2-5 years Over 5-10 years Over 10 years Total Average term to maturity Millions of Dollars, Par Value Yrs. Mos. Millions of Dollars, Par Value End of 1955-Q1 1,590 1,666 2,302 4,290 3,448 13,296 6 11 55 2,031 53 15,435 Q2 1,705 1,665 3,468 3,076 3,448 13,362 6 8 55 1,960 41 15,418 Q3 1,775 1,129 4,104 3,076 3,448 13,532 6 6 55 1,900 35 15,522 Q4 1,725 1,829 3,404 3,076 3,448 13,482 6 4 55 2,433 30 16,000 1956-Q1 2,100 1,769 3,403 3,111 3,358 13,741 5 11 55 2,387 30 16,213 Q2 1,690 2,714 2,406 3,108 3,358 13,276 5 11 55 2,293 35 15,659 Q3 1,730 2,320 2,150 4,800 1,916 12,916 6 7 55 2,210 29 15,210 Q4 1,575 2,170 2,150 4,800 . 1,916 12,611 6 7 55 2,541 27 15,234 195 7-Q1 1,625 3,152 2,518 3,500 1,866 12,661 6 4 55 2,436 20 15,172 Q2 1,625 3,002 2,518 3,499 1,866 12,510 6 2 55 2,315 21 14,901 Q3 1,655 2,938 2,518 3,499 1,866 12,476 6 - 55 2,213 17 14,761 Q4 1,625 2,538 2,918 3,496 1,866 12,443 6 - 55 2,649 18 15,165 1958-Q1 1,525 2,538 3,168 3,246 2,166 12,643 6 2 55 2,556 15 15,268 Q2 1,495 3,303 2,402 3,596 2,166 12,962 6 4 55 2,471 15 15,503 Q3 1,495 1,824 2,184 2,170 5,684 13,357 10 6 55 2,387 12 15,810 Q4 1,495 2,324 2,006 1,947 5,684 13,456 10 3 55 2,895 10 16,416 1959-Q1 1,595 2,297 2,105 1,947 5,684 13,628 10 _ 55 2,855 22 16,560 Q2 1,955 2,512 1,703 2,007 5,774 13,951 9 9 55 2,767 18 16,791 Q3 2,024 2,437 1,702 2,077 5,704 13,944 9 6 55 2,662 15 16,6 76 Q4 2,077 2,867 1,131 2,075 5,702 13,852 9 6 55 3,212 16 17,135 Table 7 - Continued Unmatured Direct and Guaranteed securities issues and perpetuals) (ex. non -market Per petuals Non-market issues Matured and out standing market issues Total out standing 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Average term to maturity Treasury Bills & notes & deposit certificates Other Millions of Dollars, Par Value Yrs. Mos. Millions of Dollars, Par Value End of 1960--Ql 2,125 2,755 1,343 2,075 5,802 14,100 9 5 55 3,143 12 17,310 Q2 1,965 2,259 1,731 2,355 5,724 14,033 9 6 55 3,059 27 17,174 Q3 1,965 2,259 2,997 1,088 5,724 14,033 9 3 55 3,002 20 17,110 Q4 1,985 2,226 2,806 1,160 5,895 14,072 9 5 55 3,594 25 17,747 1961--Ql 1,935 2,476 2,741 1,165 5,804 14,120 9 3 55 3,562 16 17,753 Q2 1,885 2,961 2,402 1,165 5,804 14,217 9 - 55 3,473 17 17,762 Q3 1,885 2,935 2,869 1,054 5,648 14,391 8 7 55 3,537 14 17,997 Q4 1,885 3,165 2,770 978 5,527 14,325 8 4 55 4,237 19 18,636 1962--Ql 1,885 3,222 2,820 1,028 5,440 14,395 8 - 55 4,121 29 18,600 Q2 1,885 3,140 2,633 955 5,652 14,265 8 1 55 4,016 23 18,359 Q3 2,030 2,855 2,633 2,322 4,485 14,325 8 1 55 3,929 19 18,327 Q4 2,165 2,526 2,443 2,472 5,048 14,655 8 5 55 4,719 19 19,448 1963--Ql 2,165 2,651 2,568 2,202 5,090 14,677 8 4 55 4,600 16 19,347 Q2 2,345 2,587 3,058 1,838 5,190 15,018 8 3 55 4,464 16 19,553 Q3 2,245 3,837 1,792 1,838 5,190 14,902 8 1 55 4,414 14 19,385 Q4 2,240 3,548 2,183 1,838 5,188 14,997 7 11 55 5,199 25 20,276 1964--Ql 2,2 30 3,609 2,053 2,042 5,038 14,972 7 10 55 5,099 19 20,145 Q2 2,145 3,013 2,283 2,373 5,113 14,927 8 - 55 4,988 17 19,987 Q3 2,130 3,284 2,433 2,043 5,098 14,987 7 10 55 4,948 16 20,006 Q4 2,140 3,000 2,413 2,313 5,096 14,961 7 10 55 5,701 16 20,733 Table 7 - Continued Unmatured Direct and Guaranteed securities (ex. non-market issues and perpetuals) Per petuals Non-market issues Matured and out standing market issues Total out standing 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Average term to maturity Treasury Bills & notes & deposit certificates Other Millions of Dollars, Par Value Yrs. Mos. Millions of Dollars, Par Value End of 1965--Ql 2,140 2,510 2,363 2,588 5,095 14,696 7 11 55 5,600 14 20,365 Q2 2,140 2,657 2,390 2,394 5,086 14,668 7 9 55 5,467 13 20,204 Q3 2,150 2,212 2,660 2,436 5,145 14,603 7 11 55 5,431 31 20,120 Q4 2,150 2,388 2,410 2,796 4,830 14,574 7 9 55 6,034 18 20,681 End of Quarter. Source: Supplement. TABLE 8 Bank of Canada's Holdings of Government of Canada Direct and Guaranteed Securities ($ Millions) 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Treasury Bills Other 1955-Q1 165 1,161 398 265 151 2,139 Q2 29 7 1,155 392 271 163 2,278 Q3 235 868 59 7 386 202 2,290 Q4 263 1,021 355 517 213 2,368 1956-Q1 456 510 449 624 200 2,239 Q2 456 585 799 329 149 2,318 Q3 535 506 673 448 216 2,377 Q4 505 520* 6 30* 507* 232* 2,394* 1957-Q1 477 628 612 314 236 2,256 Q2 519 694 608 325 230 2,376 Q3 428 781 615 323 231 2,378 Q4 46 7 779 667 301 213 2,428 1958-Q1 480 894 664 240 131 2,409 Q2 371 1,126 371 374 296 2,537 Q3 70 351 605 463 1,213 2,701 Q4 36 245 552 463 1,326 2,622 1959-Q1 161 92 521 452 1,325 2,551 Q2 251 162 388 467 1,351 2,619 Q3 29 7 257 361 434 1,322 2,672 Q4 306 515 61 425 1,315 2,621 1960-Q1 399 417 29 386 1,315 2,546 Q2 392 449 93 491 1,185 2,609 Q3 336 518 377 207 1,187 2,625 Q4 404 353 527 218 1,187 2,690 1961-Q1 304 331 576 215 1,184 2,610 Q2 277 438 550 213 1,184 2,662 Q3 327 424 607 273 1,181 2,812 Q4 312 514 548 266 1,186 2,826 1962-Q1 232 437 564 . 342 1,185 2,760 Q2 178 338 419 370 1,187 2,493 Q3 399 301 335 805 639 2,478 Q4 455 447 507 791 683 2,883 Table 8 - Continued 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Treasury Bills Other 1963-Q1 370 510 572 630 698 2,779 Q2 434 503 696 571 684 2,887 Q3 338 837 442 571 752 2,939 QA 466 688 559 570 752 3,035 1964-Q1 476 603 552 621 707 2,957 Q2 403 390 549 875 708 2,925 Q3 519 375 810 611 702 3,017 Q4 479 349 779 711 747 3,064 1965-Q1 483 263 715 731 773 2,965 Q2 470 39 3 847 628 833 3,170 Q3 426 364 917 628 834 3,169 Q4 608 478 820 643 868 3,417 *Major Series Revision Last Month in Quarter Source: Supplement. TABLE 9 Total Minus Bank of Canada's Holdings of Government of  Canada Direct and Guaranteed Securities ($ Millions) 2 years and under Treasury Over 2-5 Over 5-10 Over 10 Bills Other years years years Total 1955-Q1 1,425 505 1,904 4,025 3,297 11,157 Q2 1,408 510 3,076 2,805 3,285 11,084 Q3 1,540 261 3,507 2,691 3,246 11,242 04 1,462 808 3,049 2,560 3,235 11,114 1956-Q1 1,644 1,259 2,954 2,487 3,158 11,502 Q2 1,234 2,129 1,607 2,780 3,209 10,958 Q3 1,195 1,814 1,477 4,352 1,700 10,539 Q4 1,070 1,650 1,520 4,293 1,684 10,217 1957-Q1 1,148 2,524 1,906 3,186 1,630 10,405 Q2 1,106 2,308 1,910 3,174 1,636 10,134 Q3 1,227 2,157 1,903 3,176 1,635 10,098 Q4 1,158 1,759 2,251 3,195 1,653 10,015 1958-Q1 1,045 1,644 2,504 3,006 2,035 10,234 Q2 1,124 2,177 2,031 3,222 1,870 10,425 Q3 1,425 1,473 1,579 1,707 4,471 10,656 Q4 1,459 2,079 1,454 1,484 4,358 10,834 1959-Q1 1,434 2,205 1,584 1,495 4,359 11,077 Q2 1,704 2,350 1,315 1,540 4,423 11,332 Q3 1,727 2,180 1,341 1,643 4,382 11,272 Q4 1,771 2,352 . 1,070 1,650 4,387 11,231 1960-Q1 1,726 2,338 1,314 1,689 4,487 11,554 Q2 1,573 1,810 1,638 1,864 4,539 11,424 Q3 1,629 1,741 2,620 881 4,537 11,408 Q4 1,581 1,873 2,279 942 4,708 11,382 1961-Q1 1,631 2,145 2,165 950 4,620 11,510 Q2 1,608 2,523 1,852 952 4,620 11,555 Q3 1,558 2,511 2,262 781 4,467 11,579 Q4 1,573 2,651 2,222 712 4,341 11,499 1962-Q1 1,653 2,785 2,256 686 4,255 11,635 Q2 1,707 2,802 2,214 585 4,465 11,772 Q3 1,631 2,554 2,298 1,517 3,846 11,847 Q4 1,710 2,079 1,936 1,681 4,365 11,772 Table 9 - Continued 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Treasury Bills Other 1963-Q1 1,795 2,141 1,996 1,572 4,392 11,898 Q2 1,911 2,084 2,362 1,267 4,506 12,131 Q3 1,907 3,000 1,350 1,267 4,438 11,963 Q4 1,774 2,860 1,624 1,268 4,436 11,962 1964-Q1 1,754 3,006 1,501 1,421 4,331 12,015 Q2 1,742 2,623 1,734 1,498 4,405 12,002 Q3 1,611 2,909 1,623 1,432 4,396 11,970 Q4 1,661 2,651 1,634 1,602 4,349 11,89 7 1965-Q1 1,657 2,247 1,648 1,857 4,322 11,731 Q2 1,670 2,264 1,543 1,766 4,253 11,498 Q3 1,724 1,848 1,743 1,808 4,311 11,434 Q4 1,542 1,910 1,590 2,153 3,962 11,157 Last Month in Quarter Source: Tables 7 and 8 above. TABLE 10 Government of Canada Accounts (i.e. Securities Investment Account; Purchase Fund; Unemployment Insurance Fund and Other) ($ Millions) Treasury Bills Other Total 1955-Q1 32 1,204 1,236 Q2 1 1,209 1,210 Q3 6 1,355 1,361 Q4 36 1,455 1,491 1956-Q1 0 1,950 1,950 Q2 3 1,604 1,607 Q3 3 1,419 1,422 Q4 40 1,478 1,518 1957-Q1 0 1,490 1,490 Q2 13 1,348 1,361 Q3 11 1,370 1,381 Q4 59 1,308 1,367 1958-Q1 1 1,286 1,287 Q2 1 1,129 1,130 Q3 6 1,215 1,221 Q4 89 1,170 1,259 1959-Q1 28 916 944 Q2 9 998 1,007 Q3 8 982 990 Q4 30 893 923 1960-Q1 13 820 833 Q2 13 766 779 Q3 57 850 907 Q4 56 810 866 1961-Q1 6 721 727 Q2 2 731 733 Q3 1 729 730 Q4 4 640 644 1962-Q1 62 474 536 Q2 181 558 739 Q3 6 613 619 Q4 47 623 670 Table 10 - Continued Treasury Bills Other Total 1963-Q1 41 428 469 Q2 36 408 444 Q3 34 433 467 04 51 465 516 1964-Q1 73 402 475 Q2 16 398 414 Q3 20 558 578 Q4 61 708 769 1965-Q1 10 454 464 Q2 16 496 512 Q3 16 484 500 Q4 12 544 557 Last Month in Quarter. Source: Supplement. TABLE 11 Chartered Banks; Holdings of Government of Canada Direct and Guaranteed Securities ($ Millions) Treasury 2 years and Over 2 Bills under years 1955-Q1 435 681 2,482 Q2 376 665 2,579 Q3 369 401 2,775 Q4 427 475 2,157 1956-Q1 593 398 1,922 Q2 772 557 1,398 Q3 786 526 1,322 04 740 406* 1,269* 1957-Q1 805 538 1,227 Q2 784 493 1,251 Q3 915 479 1,241 Q4 805 410 1,425 1958-Q1 800 403 1,643 Q2 882 710 1,736 Q3 1,096 757 2,024 Q4 950 826 1,736 1959-Q1 902 856 1,756 Q2 1,009 619 1,532 Q3 919 420 1,475 Q4 974 657 1,169 1960-Q1 968 658 1,270 Q2 959 569 1,399 Q3 1,076 540 1,443 Q4 967 615 1,472 1961-Q1 1,112 827 1,371 Q2 1,141 915 1,325 Q3 1,217 911 1,554 Q4 1,157 1,089 1,551 1962-Q1 1,164 1,150 1,567 Q2 1,013 1,080 1,384 Q3 1,018 569 1,335 Q4 1,127 754 1,487 Table 11 - Continued Treasury 2 years and Over 2 Bills under years 1963-Q1 1,272 825 1,502 Q2 1,318 922 1,554 Q3 1,233 1,408 1,127 Q4 1,282 1,335 1,325 1964-Q1 1,226 1,421 1,279 Q2 1,240 1,219 1,357 Q3 1,193 1,269 1,213 Q4 1,257 1,126 1,336 1965-Q1 1,294 991 1,539 Q2 1,262 1,077 1,399 Q3 1,382 907 1,439 Q4 1,357 955 1,423 *Major series revision Last Month in Quarter Source: Supplement. TABLE 12 General Public Holdings of Government of Canada Securities by Term" to Maturity Unmatured Direct and Guaranteed securities (ex. Canada Savings Bonds and Perpetuals) Per petuals Canada Savings Bonds Matured and out standing market issues Total out-s tanding 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Average term to maturity Treasury Bills Other Millions of Dollars, Par Value Yrs. Mos. Millions of Dollars, Par Value End of 1955-Q1 286 390 873 2,128 2,555 6,232 9 3 52 2,031 53 8,369 Q2 352 418 1,357 1,518 2,515 6,160 8 11 52 1,960 41 8,212 . Q3 486 306 1,511 1,470 2,455 6,227 8 5 52 1,900 35 8,214 Q4 494 677 1,389 1,462 2,433 6,455 7 11 52 2,433 30 '8,969 1956-Q1 546 713 1,429 1,477 2,371 6,536 7 6 51 2,387 30 9 ,004 Q2 453 1,150 1,056 1,485 2,368 6,511 7 4 52 2,293 35 8,890 Q3 401 1,088 1,006 2,630 1,198 6,324 7 6 51 2,210 29 8,614 Q4 285 1,079 985 2,612 1,186 6,146 7 6 51 2,541 27 8,766 1957-Q1 337 1,676 1,182 1,978 1,157 6,329 7 2 51 2,436 20 8,836 02 304 1,649 1,169 1,966 1,141 6,288 6 11 51 2,315 21 8,616 Q3 29 7 1,498 1,165 1,965 1,139 6,064 6 11 51 2,213 17 8,345 Q4 289 1,223 1,340 1,970 1,153 5,975 7 51 2,649 18 8,693 1958-Q1 239 1,113 1,336 1,910 1,470 6,067 7 11 51 2,556 15 8,689 Q2 239 1,341 939 2,060 1,354 5,933 8 51 2,471 15 8,470 Q3 319 487 446 720 3,534 5,507 14 9 51 2,387 12 7,956 Q4 415 1,010 413 666 3,509 6,012 13 4 50 2,895 10 8,968 1959-Q1 501 1,325 517 692 3,537 6,572 12 3 50 2,855 22 9,499 Q2 6 70 1,619 475 738 3,596 7,098 11 4 50 2,767 18 9,934 Q3 786 1,687 543 815 3,573 7,404 10 10 51 2,662 15 10,132 Q4 755 1,610 671 838 3,572 7,446 10 10 51 3,212 16 10,725 u> LO Table 12 - Continued Unmatured direct and guaranteed securities (ex. Canada Savings Bonds and Perpetuals) Per petuals Canada Savings Bonds Matured and out standing market issues Total out-s tanding 2 years and under Over 2-5 years Over 5-10 years Over 10 years Total Average term to maturity Treasury Bills Other Millions of Dollars, Par Value Yrs. Mos. Millions of Dollars, Par Value End of I960--Ql 735 1,657 824 880 3,669 7,765 10 7 51 3,143 12 10,971 Q2 591 1,212 1,059 1,027 3,774 7,663 11 - 51 3,059 27 10,800 Q3 488 1,047 1,523 545 3,790 7,393 11 2 51 3,002 20 10,466 Q4 549 1,147 1,200 559 3,954 7,409 11 6 51 3,594 25 11,080 1961- -01 504 1,255 1;171 568 3,928 7,426 11 3 51 3,562 16 11,055 Q2 459 1,537 896 580 3,922 7,394 11 1 51 3,473 17 10,9 35 i 0 3 333 1,536 1,014 525 3,856 7,264 11 - 51 3,398 14 10,728 Q4 405 1,503 952 485 3,828 7,173 10 10 51 4,080 19 11,323 1962- -yi 420 1,578 968 46 7 3,809 7,241 10 5 51 4,059 29- 11,380 i 02 505 1,650 1,077 435 3,870 7,537 10 - 50 3,988 23 11,599 | Q3 591 1,9 36 1,204 1,236 3,369 8,335 9 5 50 3,851 19 12,255 Q4 523 1,276 862 1,337 3,784 7,782 10 9 50 4,620 19 12,472 1963- -Ql 470 1,272 868 1,307 3,880 7,797 10 11 50 4,588 16 12,451 Q2 546 1,120 1,168 1,028 3,985 7,847 11 - 50 4,464 16 12,377 Q3 628 1,535 655 1,008 3,888 7,713 10 8 50 4,385 14 12,163 Q4 4 30 1,471 746 1,008 3,873 7,528 10 8 50 5,133 25 12,736 1964- -Ql 444 1,518 700 1,108 3,809 7,579 10 7 50 5,099 19 12,747 Q2 4 76 1,355 888 1,147 3,873 7,738 10 6 50 4,988 17 12,79 3 Q3 388 1,481 884 1,113 3,868 7,734 10 5 50 4,905 16 12,705 Q4 332 1,255 937 1,155 3,786 7,465 10 6 50 5,613 16 13,144 4> Table 12 - Continued Unmatured direct and Savings guaranteed securities Bonds and Perpetuals) [ex. Canada Per petuals Canada Savings Bonds Matured and out standing marke t issues Total out-s tanding 2 years and under Over 2-years -5 Over 5-10 years Over 10 years Total Average term to maturity Treasury Bills Other Millions of Dollars, Par Value Yrs. Mos. Millions of Dollars, Par Value End of 1965--Ql 342 1,195 881 1,289 3,759 7,465 10 4 50 5,557 14 13,086 Q2 381 1,076 885 1,224 3,691 7,256 10 3 50 5,426 13 12,746 Q3 313 912 1,072 1,265 3,726 7,289 10 3 50 5,324 31 12,695 Q4 157 921 964 1,550 3,404 6,995 10 4 50 5,866 18 12,929 End of Quarter Source: Supplement. OJ TABLE 13 Total Debt Minus Bank of Canada Holdings by Term to Maturity (Under Different Assumptions About the Behaviour of the Bank of Canada. Starred row numbers give actual figures, non-starred ones hypothetical figures). ($ Millions) 2 years and under Row No. Treasury Bills Other Over 2-5 years Over 5-10 years Over 10 years Total * 1. (Total:58Q2) - (B0C:58Q2) = 1,124 2,177 2,031 3,222 1,870 10,425 2. (Total:58Q3) - (B0C:58Q2) = 1,124 698 1,813 1,796 5,388 10,820 A 3. (Total:58Q3) - (B0C:58Q3) = 1,425 1,473 1,579 1,707 4,471 10,656 4. (Total:58Q3) - (B0C:59Q1) = 1,334 1,732 1,663 1,718 4,359 10,806 5. (Total:58Q3) - (B0C:59Q2) = 1,244 1,662 1,796 1,703 4,333 10,738 6. (Total:58Q3) - (B0C:59Q3) = 1,198 1,567 1,823 1,736 4,362 10,685 7. (Total:58Q3) - (B0C:59Q4) = 1,189 1,309 2,123 1,745 4,369 10,736 * 8. (Total:59Ql) - (B0C:59Q1) 1,434 2,205 1,584 1,495 4,359 11,077 A 9. (Total:59Q2) - (B0C:59Q2) = 1,704 2,350 1,315 1,540 4,423 11,332 A 10. (Total:59Q3) - (B0C:59Q3) = 1,727 2,180 1,341 1,643 4,382 11,272 A 11. (Total:59Q4) - (B0C:59Q4) 1,771 2,352 1,070 1,650 4,387 11,231 Source: Tables 7 and 8. u> ON TABLE 14 Bonds Outstanding on December 31 by Issuer  and Currency ($ Millions) Government Direct and of Canada Guaranteed Provinces Direct and Guaranteed Municipalities Direct and Guaranteed Corporations Cdn. $ Other Total Cdn. $ Other Total Cdn. $ Other Total Cdn. $ Other Total 1955 15,449 551 16,000 3,161 913 4,074 1,790 413 2,203 3,594 833 4,427 1956 14,799 435 15,234 3,509 1,107 4,616 1,930 49 7 2,427 4,178 1,045 5,223 1957 14,798 367 15,165 4,014 1,156 5,170 2,111 599 2,710 4,750 1,443 6,193 1958 16,051 365 16,416 4,484 1,304 5,788 2,318 720 3,038 5,225 1,633 6,858 1959 16,922 213 17,135 4,815 1,556 6,371 2,529 841 3,370 5,320 1,650 6,970 1960 17,535 212 17,747 5,263 1,593 6,855 2,793 947 3,740 5,636 1,549 7,186 1961 18,479 157 18,636 6,594 1,617 8,211 3,129 928 4,058 5,446 1,662 7,108 1962 19,184 264 19,448 7,205 1,846 9,051 3,339 1,024 4,363 5,706 1,967 7,673 1963 19,893 383 20,276 7,986 2,220 10,206 3,726 1,027 4,753 5,869 2,113 7,982 1964 20,350 383 20,733 8,577 2,578 11,155 4,050 1,142 5,193 6,433 2,322 8,755 1965 20,303 378 20,681 9,063 2,826 11,889 4,302 1,162 5,464 7,310 2,652 9,963 Source: Supplement. LO TABLE 15 Federal Government Gross New Issues of Direct and Guaranteed Bonds Par Values ($ Millions) 0-2 vears Over 2-5 vears Over 5-10 vears All under 10 vears Over 10 vears Total Cdn.$ only micro-series Total U.S. $ micro-series Total Cdn. $ only Supple ment series Total U.S. $ Supple ment series 1 2 ' 3 4 5 6 7 8 9 1955-Q1 Q2 Q3 Q4 0 0 0 0 0 700 0 0 0 0 0 0 0 700 0 0 0 0 0 0 0 700 0 0 0 0 0 0 0 0 700 670 0 0 0 0 1956-Q1 Q2 Q3 Q4 0 0 0 400 0 0 0 0 0 0 0 0 0 0 0 400 0 0 250 0 0 0 250 400 0 0 0 0 32 19 260 1,216 0 0 0 0 1957-Q1 Q2 Q3 Q4 0 0 0 950 0 0 0 400 0 0 0 0 0 0 0 1,350 0 0 0 0 0 0 0 1,350 0 0 0 0 21 10 6 2,566 0 0 0 0 1958-Ql Q2 Q3 Q4 0 200 400 900 0 400 1,021 0 0 0 1,267 0 0 600 2,688 900 300 350 3,519 0 300 950 6,207 900 0 0 0 0 300 950 6,206 1,744 0 0 0 0 1959-Q1 Q2 Q3 Q4 200 0 238 260 100 0 0 450 0 60 0 0 300 60 238 710 0 90 0 0 300 150 238 710 0 0 0 0 350 170 247 2,126 0 0 0 Table 15 - Continued Total Total Cdn. $ Total ..v • Cdn. $ Total only U.S. $ * only U.S. $ Supple Supple Over 2-5 Over 5-10 All under Over 10 micro- micro- ment nient 0-2 years years years 10 years years series series series series 1 2 3 4 5 6 7 8 9 1960-Q1 0 300 0 300 100 400 0 457 0 Q2 0 389 * 80 469 ' 0 469 0 502 0 Q3 0 0 0 0 0 0 0 31 0 Q4 300 300 75 675 175 850 0 1,676 0 1961-Q1 175 250 0 425 0 425 0 494 0 Q2 390 110 0 500 0 500 0 540 0 Q3 225 0 175 400 0 400 0 913 0 Q4 175 250 100 525 0 525 0 1,480 0 1962-Q1 300 0 100 400 0 400 0 489 0 Q2 100 0 100 200 0 200 0 267 0 Q3 0 0 80 80 120 300 0 274 0 Q4 0 400 250 650 135* 785 135 2,277 135 1963-01 125 225 0 350 135* 350 135 442 135 Q2 475 175 0 650 100 750 0 811 0 Q3 0 0 0 0 0 0 0 71 0 Q4 500 391 0 901 0 901 0 1,977 0 1964-Q1 170 130 0 300 50 350 0 447 0 Q2 250 0 325 575 75 1,350 0 707 0 Q3 200 0 0 200 50 250 0 334 0 Q4 325 200 350 875 0 875 0 1,894 0 Table 15 - Continued Total Total Cdn. $ Total Cdn. $ Total only U.S. $ only U.S. $ Supple Supple Over 2-5 Over 5-10 All under Over 10 micro- micro- ment ment 0-2 years years years 10 years years series series series series 1 2 3 4 5 6 7 8 9 1965-Q1 175 0 275 450 0 450 0 539 0 Q2 175 0 0 175 0 175 0 241 0 Q3 405 270 0 675 100 775 0 876 0 Q4 150 100 50 200 0 200 0 1,218 0 *U.S. $ Notes on the construction of Table 15  Supplement Series, i.e. Columns 8 and 9. "Series cover all publicly announced issues and some private placements not publicly announced. New issues are based on delivery rather than offering dates. Foreign currencies have been converted to Canadian dollars at market noon rates on the date of delivery." Bank of Canada Statistical Summary Supplement, 1960, p. 84. Treasury bills are not included but Canada Savings Bonds are. The reasons for the slight difference between the Supplement series and the micro-series are: 1. The micro-series was built from offering dates - consistent with the issuers intentions. 2. The private placements are not reported hence could not be included in the micro-series. 3. Canada Savings bonds are included in the Supplement series but not the other one. Micro-series, i.e. Columns 1 to 7. Built from the reports on new issues contained in various issues of the Supplement. They exclude Treasury bills and Canada Savings Bonds - hence 3 above. 4> O TABLE 16a Provincial Gross New Issues of (Direct and Guaranteed) Bonds Par Values ($000's) Total: 1 to 7 All under Cdn. $ Total 10 years only U.S. $ . Over 2-5 Over 5-10 Other under Cols. 1+2 Over 10 micro- micro-0-2 years years years 10 years +3+4 years Other series series 1 2 3 4 5 6 7 8 9 1955-Q1 0 6,849 28,000 0 34,849 121,000 1,024 156,873 0 Q2 2,500 0 23,250 0 25,750 36,000 8,174 69,924 0 Q3 3,000 0 50,000 4,115 57,115 63,000 253 120,368 0 Q4 0 0 0 0 0 10,000 0 10,000 75,000 1956-01 0 0 16,700 0 16,700 62,000 230 78,930 0 Q2 0 0 0 0 0 67,400 800 68,200 95,000 03 0 0 0 0 0 78,500 4,456 82,956 0 Q4 550 0 35,950 0 36,500 154,970 482 191,952 15,750 1957-Q1 0 13,957 0 0 13,957 139,869 1,051 154,877 0 Q2 0 0 18,400 0 18,400 142,600 7,654 168,654 2,000 Q3 0 20,646 31,731 0 52,377 85,523 956 138,856 0 Q4 0 1,591 13,700 0 15,291 162,208 9,736 187,235 0 1958-Q1 0 5,000 38,500 6 43,506 40,500 177 84,183 70,000 Q2 0 11,375 40,000 0 51,375 53,625 2,544 107,544 100,000 Q3 74,000 10,000 0 0 84,000 18,000 0 102,000 0 Q4 25,000 40,000 5,000 4,276 74,276 98,001 1,900 174,177 0 1959-Q1 6,500 15,000 0 5,000 26,500 95,500 2,551 124,551 75,000 Q2 • 0 0 250 0 250 32,250 5,774 38,274 50,000 . Q3 2,350 35,616 36,000 0 73,966 94,000 9,399 177,365 65,000 Q4 0 5,580 25,000 0 30,580 75,600 645 106,825 59,000 Table 16a - Continued Total 1 to 7 All under Cdn. $ Total 10 years only U.S. $ Over 2-5 Over 5-10 Other under Cols. 1+2 Over 10 micro- micro-0-2 years years * years 10 years +3+4 years Other series series 1 2 3 4 5 6 7 8 9 1960-Q1 0 5,000 16,000 0 21,000 42,721 1,250 64,971 37,000 Q2 1,850 6,000 28,500 18 36,368 150,026 2,030 188,424 40,000 Q3 10,000 25,000 16,100 0 51,100 118,900 170 170,170 0 Q4 20,000 15,019 2,301 1,250 38,570 67,472 4,728 110,770 0 1961-Q1 0 13,000 27,800 1,828 42,628 194,047 2,389 239,064 15,000 Q2 0 13,000 25,700 0 38,700 139,300 1,150 179,150 0 Q3 6,000 153,000 8,000 243 167,243 71,657 29 238,929 0 Q4 0 18,460 36,425 0 54,885 183,875 6,882 245,642 0 1962-Q1 0 20,000 45,600 0 65,600 19 7,184 0 262,784 0 Q2 10,000 0 31,500 1,544 43,044 132,200 665 175,909 0 Q3 0 53,976 10,000 0 63,976 69,500 8,480 141,956 8,000 Q4 0 0 43,500 0 43,500 171,500 5,804 220,804 96,500 1963-Q1 15,000 0 13,500 0 28,500 58,500 0 87,000 373,000 Q2 2,000 0 177,289 0 179,289 130,000 4,673 313,962 6,225 Q3 5,000 65,000 0 0 70,000 57,500 273 127,773 0 Q4 15,000 12,000 0 0 27,000 180,600 0 207,600 0 1964-Q1 0 0 39,861 5,000 44,861 131,500 2,968 179,329 34,225 Q2 15,000 0 26,532 0 41,532 103,000 1,620 146,152 30,000 Q3 0 20,000 0 0 20,000 70,500 1,428 91,928 10,000 Q4 15,000 0 0 0 15,000 212,000 0 227,000 107,500 Is) Table 16a - Continued Total 1 to 7 All under Cdn. $ Total 10 years only U.S. $ Over 2-5 Over 5-10 Other under Cols. 1+2 Over 10 micro- micro-0-2 years vears years 10 years +3+4 vears Other series series 1 2 3 4 5 6 7 8 9 1965-Q1 0 0 0 0 0 162,000 77,688 239,688 47,000 Q2 15,000 0 0 0 15,000 178,500 0 193,500 65,000 Q3 0 0 0 100,000 100,000 100,000 0 200,000 35,000 Q4 0 0 0 0 0 126,000 455 126,455 0 .Notes on the Construction of Table 16a Micro-series 1. Built from individual issues reported in three sources: The Financial Post, confidential data made available by the Bank of Canada and Moody's annual statements. The Bank's data series include more issues than the Financial Post, hence the usual understatement of columns 1 and 2. The Financial Post series, being the only apparently unchanged series (and from 1955-57 the only one available), was used as the basis from which the series reported here was constructed. The data from the Bank and Moody's were used to check and enrich the reports in the Financial Post. 2. With the exception of two quarters the Financial Post does not report Treasury bill issues by Manitoba and Saskatchewan - they each issue (before 1962, the Bank's data consist of handwritten sheets and make no mention of Treasury bill issues) roughly $ 4 M per month. Note that it is a constant for the period after 1962. 3. Some issues in more than one currency would appear under Cdn. $. It is not clear what the Bank does in this respect. 4. Information on coupons and yields is available. 4> TABLE 16b Provincial Gross New Issues of (Direct and Guaranteed) Bonds Par Values ($000's) Cdn. $ only Other currencies Supplement Supplement series series 1 2 1955-Q1 159,000 0 Q2 76,000 0 Q3 114,000 0 Q4 23,000 0 1956-Q1 79,000 92,000 Q2 79,000 50,000 Q3 94,000 39,000 Q4 168,000 34,000 1957-Q1 126,000 63,000 Q2 148,000 46,000 Q3 108,000 0 Q4 252,000 24,000 1958-Q1 101,000 69,000 Q2 119,000 97,000 Q3 141,000 0 Q4 199,000 0 1959-Q1 125,000 104,000 Q2 100,000 57,000 Q3 193,000 81,000 Q4 155,000 81,000 1960-Q1 111,000 42,000 Q2 230,000 41,000 Q3 221,000 11,000 Q4 122,000 0 1961-Q1 272,000 23,000 Q2 214,000 0 Q3 382,000 0 Q4 275,000 10,000 1962-Q1 284,000 0 Q2 282,000 0 Q3 172,000 9,000 Q4 466,000 104,000 45 Table 16b -.Continued Cdn. $ only Other currencies Supplement Supplement series series 1 2 1963-Q1 134,000 186,000 Q2 462,000 81,000 Q3 172,000 61,000 Q4 339,000 2,000 1964-Q1 233,000 84,000 Q2 322,000 146,000 Q3 182,000 30,000 Q4 356,000 156,000 1965-Q1 299,000 21,000 Q2 310,000 121,000 Q3 208,000 83,000 Q4 348,000 46,000 Notes on the Construction of Table 16b Supplement series 1. Before 1960 they exclude provincial Treasury Bills, e.g. Manitoba and Saskatchewan. Beginning in 1960 they include those sold publicly. 2. They include some bonds issued in exchange for shares, e.g. when B. C. took over B. C. Electric Co. 3. The series is not available by term to maturity. 4. Retirements and, therefore, Net New Issues are also available in the Supplement. TABLE 17a Gross New Issues of Bonds by Canadian Corporations  Par Values ($000's) Total: 5 to 7 All under Cdn. $ Total 10 years only U.S. $ Over 2-5 Over 5-10 Other under Cols. 1+2 Over 10 micro- micro-0-2 years years years 10 years +3+4 vears Other series series 1 2 3 4 5 6 7 8 9 1951-Q1 0 0 1,650 15,950 17,600 74,900 0 92,500 0 Q2 0 0 15,000 35,375 50,375 227,715 7,240 285,330 0 Q3 0 0 11,898 8,600 20,498 40,450 7,625 68,573 600 Q4 0 0 47,500 23,600 71,100 125,750 0 196,850 0 1956-01 0 0 8,500 1,500 10,000 187,525 0 197,525 0 Q2 10,000 0 20,750 15,600 46,350 108,350 0 154,700 132,500 Q3 0 7,000 24,250 0 31,250 55,350 0 86,600 0 Q4 0 750 34,000 0 34,750 158,125 0 192,875 50,500 1957-Q1 0 / 0 53,845 250 54,095 246,509 15,000 315,604 137,165 Q2 0 10,500 13,650 0 24,150 210,000 0 234,150 137,450 Q3 0 0 1,526 0 1,526 50,600 102,500 154,626 27,250 Q4 0 1,201 2,500 0 3,701 100,020 20,050 123,771 0 1958-Ql 0 0 7,000 0 7,000 189,790 850 197,640 50,000 Q2 0 2,900 10,400 26,900 40,200 182,250 1,200 223,650 37,000 Q3 0 2,000 0 600 2,600 96,726 800 100,126 9,600 Q4 0 0 1,150 400 1,550 63,250 0 64,800 0 1959-Q1 0 0 0 13,000 13,000 109,600 2,500 125,100 0 Q2 0 0 1,550 2,650 4,200 43,700 0 47,900 28,500 Q3 0 0 3,000 450 3,450 42,325 0 45,775 0 Q4 0 0 2,675 250 2,925 109,350 0 112,275 0 Table 17a - Continued Total: 5 to 7 All under Cdn. $ Total 10 years only U.S. $ Over 2-5 Over 5-10 Other under Cols. 1+2 Over 10 micro- micro-0-2 years years years 10 years +3+4 years Other series series 1 2 3 4 5 6 7 8 9 1960-01 0 1,000 1,550 250 2,800 174,050 4,142 180,992 45,000 Q2 0 0 6,500 500 7,000 129,816 500 137,316 10,260 Q3 0 0 500 1,350 1,850 101,825 460 104,135 30,000 Q4 0 1,150 850 600 2,600 46,9 75 0 49,575 0 1961-Q1 0 0 500 750 1,250 98,825 6,050 106,125 5,000 Q2 0 4,500 1,410 6,650 12,560 243,500 23,240 279,300 98,000 Q3 0 0 950 1,040 1,590 68,750 500 71,040 13,000 Q4 0 2,000 750 2,500 5,250 66,725 6,100 78,075 0 1962-Q1 0 0 8,650 300 8,950 96,025 2,800 107,775 3,000 Q2 0 0 2,900 17,200 20,100 136,757 9,000 165,857 100,000 Q3 0 1,000 1,250 0 2,250 175,580 1,000 178,830 21,000 Q4- 0 2,000 20,000 5,325 27,325 85,600 0 112,925 39,000 1963-Q1 0 20,000 0 0 20,000 96,234 6,000 122,234 93,009 Q2 0 800 10,050 11,600 22,450 241,571 11,000 275,021 86,250 Q3 500 0 1,500 2,000 4,000 23,950 1,500 29,450 37,000 Q4 0 8,000 8,800 0 16,800 89,300 6,400 112,500 0 1964-Q1 0 7,500 5,250 1,500 14,250 130,350 5,000 149,600 22,000 Q2 0 0 6,000 8,100 14,100 212,964 5,005 232,069 97,000 Q3 0 0 0 300 300 96,550 12,000 108,850 0 Q4 0 12,072 1,950 5,000 19,022 221,824 0 240,846 112,500 Table 17a - Continued Total: 5 to 7 All under Cdn. $ Total 10 years only U.S. $ Over 2-5 Over 5-10 Other under Cols. 1+2 Over 10 micro- micro-0-2 years years years 10 years +3+4 years Other series series 1 2 3 4 5 6 7 8 9 1965-Q1 0 400 2,550 0 2,950 220,225 2,000 225,175 44,600 Q2 0 0 10,000 1,800 11,800 403,620 5,500 420,920 113,000 Q3 0 1,000 2,000 2,800 5,800 98,500 3,500 107,800 125,000 Q4 0 10,000 35,000 0 45,000 123,450 12,000 180,450 112,000 Notes on the Construction of Table 17a Micro-series 1. Built from individual issues in three sources: The Financial Post, confidential data made available from the Bank of Canada and Moody's annual statements. The Bank's data series include more issues than the Financial Post until 1962. At that time the Bank series changes somewhat so that while less issues (only large ones) are reported, a residual containing smaller issues is also reported, thereby inflating the totals. In most cases issues reported in the Financial Post were also reported in the Bank's series. This series builds on the reports in the Financial Post, uses Moody's and the Bank's data to enrich these reports, but in the few cases where agreement could not be reached the Financial Post "wins." 2. Some issues in more than one currency would appear under mainly Cdn. $. The reports on each issue are not sufficiently detailed for a more accurate series to be constructed. These issues are believed to be of little significance. 3. Some issues enable the lender to acquire company stock, etc. Such information is ignored. 4. Information on coupons and yields is usually available too. 4> oo Table 17a - Continued Micro-series - Continued 5. Some issues by non-Canadian corporations are included in the Financial Post. They are not incorporated in these series. 6. There are a handful of issues that were reported as Cdn. $ in the Financial Post and as U.S. $ in Bank's dat Experience indicated that the Bank was always right (Moody's was also consulted), and therefore this was the dat used. TABLE 17b Gross New Issues of Bonds by Canadian Corporations Par Values ($000* s) Total Total Cdn. $ only Other currencies Supplement series Supplement series 1 2 1955-Q1 190,000 0 Q2 207,000 0 Q3 72,000 9,000 Q4 218,000 0 1956-Q1 228,000 22,000 Q2 202,000 50,000 Q3 167,000 66,000 Q4 215,000 90,000 1957-Q1 238,000 130,000 Q2 324,000 189,000 Q3 74,000 66,000 Q4 165,000 25,000 1958-Ql 199,000 61,000 Q2 276,000 93,000 Q3 153,000 40,000 Q4 . 158,000 13,000 1959-Q1 92,000 9,000 Q2 131,000 7,000 Q3 74,000 27,000 Q4 123,000 18,000 1960-Q1 206,000 33,000 Q2 198,000 41,000 Q3 136,000 18,000 Q4 81,000 13,000 1961-Q1 118,000 53,000 Q2 192,000 91,000 Q3 163,000 78,000 Q4 139,000 28,000 1962-Q1 134,000 17,000 Q2 198,000 152,000 Q3 116,000 13,000 Q4 174,000 83,000 51 Table 17b - Continued Total Total Cdn. $ only Other currencies Supplement Series Supplement series 1 2 1963-Q1 91,000 31,000 Q2 360,000 225,000 Q3 53,000 11,000 Q4 150,000 42,000 1964-Q1 145,000 14,000 Q2 330,000 73,000 Q3 132,000 40,000 Q4 380,000 172,000 1965-Q1 233,000 48,000 Q2 509,000 144,000 Q3 258,000 146,000 Q4 255,000 161,000 Notes on the Construction of Table 17b Supplement series The source of Table 17b is the Bank of Canada Statistical Summary Supplement. The reader should compare Table 17b with cols. 8 and 9 of Table 17a. The comparison gives an indication of the accuracy of the micro-series. The Supplement gives information on retirements and hence net new issues but in no case is there a disaggregation by term to maturity made available. TABLE 18 Average Term of Provincial Debt as at Fiscal Year Ends (in years) 1954 19.5 1956 19.5 1958 18.8 1960 18.1 1962 19.8 1963 19.6 1964 19.7 1965 19.8 Source: DBS Annual Provincial Government Finance: Debt for the Year 1967 (68-209). TABLE 19 Municipal Debt by Province Disaggregated into Serial and Sinking Fund. The former is the first entry under each year and province, while the latter is the second"] ($000's) 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 Nfld. 6,165 4,600 6,890 4,600 8,930 6,100 10,036 5,565 10,573 5,580 11,404 5,481 13,305 5,310 14,600 5,068 15,476 .4,960 17,298 4,804 18,120 4,585 P.E.I. 2,026 3,959 2,417 3,918 2,747 4,894 2,645 4,721 2,832 5,029 2,722 5,608 2,972 6,379 3,841 7,651 3,728 8,431 3,358 8,804 3,410 8,937 N.S. 49,451 13,451 53,882 13,262 57,075 12,350 63,269 11,837 69,377 10,945 78,519 10,592 89,206 7,254 98,517 7,004 107,239 6,479 110,241 6,507 111,410 5,655 N.B. 36,897 16,164 38,523 17,660 58,899 16,322 62,681 18,814 68,137 17,016 74,523 16,374 77,042 16,096 71,344 14,173 76,206 14,358 76,533 14,236 76,115 13,159 Ont.* 728,234 39,063 759,840 120,115 814,438 203,590 884,218 285,302 938,515 366,488 1,015,257 446,573 1,078,358 499,634 1,133,128 56 7,100 1,180,798 650,112 1,253,548. 731,507 1,323,456 790,591 Man. * 46,173 33,678 48,206 43,718 52,884 44,571 54,180 47,432 60,506 51,245 65,960 58,118 76,844 57,269 88,997 53,042 105,799 54,460 119,718 63,063 • 128,738 69,969 Sask.* 32,601 17,248 38,46 7 21,288 44,186 27,973 47,283 36,126 52,683 42,101 60,348 46,744 67,035 51,899 75,078 55,557 82,763 60,526 87,968 66,555 99,734 68,071 Alta.* 154,137 10,601 180,844 10,596 216,078 10,596 237,600 10,596 274,606 10,596 291,722 10,512 316,549 9,150 338,299 9,142 355,129 9,467 365,236 9,467 391,606 9,467 B. C. 195,500 90,170 205,640 97,599 224,205 104,223 238,219 113,017 26 7,824 127,647 303,025 140,673 322,694 148,137 353,174 149,959 388,539 151,464 407,888 151,034 441,420 154,297 Table 19 - Continued 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 Que. * 238,004 t 253,401 12,717 297,052 11,717 327,693 10,306 386,730 40,306 445,264 94,108 488,040 117,722 532,115 114,269 52,786 110,993 49,579 106,626 Yukon t t 1,046 0 1,004 0 1,016 0 987 0 955 0 924 0 890 0 856 0 820,000 0 N.W.T. t t 105 0 101 0 203 0 218 0 212 0 186 0 213 0 19 3 0 227,000 0 Not reported. There remains some portion of total indebtedness of starred provinces which could not be classified. Source: DBS Municipal Government Finance (68-204). 55 TABLE 20 U.S. Government Marketable Securities Based on Treasury  Survey Data. Par Values ($ Millions) Public Holdings  U.S. Government agencies + trust funds + FRB excluded Within 1 year 1-5 years 5-10 years Over 10 years Total 1955-Q1 Q2 32,224 34,216 32,16 7 29,269 127,876 Q3 39,335 30,382 32,165 30,010 131,892 Q4 39,467 36,320 29,925 28,518 134,230 1956-Q1 37,329 35,481 29,945 28,505 131,260 Q2 37,545 30,410 29,864 28,485 126,304 Q3 42,814 28,874 27,647 28,464 127,799 Q4 45,516 39,940 16,562 28,436 130,454 1957-Q1 45,700 40,875 16,556 28,419 131,550 Q2 49,649 37,293 13,687 26,550 127,179 Q3 50,395 41,843 13,679 26,532 132,449 Q4 51,705 43,334 10,955 27,621 133,615 1958-Ql 50,045 38,276 15,207 29,213 132,741 Q2 43,873 38,492 21,991 30,235 134,591 Q3 45,584 45,482 14,881 30,135 136,082 Q4 50,900 46,741 17,267 27,710 142,618 1959-Q1 47,168 54,920 13,402 28,441 143,931 Q2 51,341 51,253 16,680 25,709 144,983 Q3 54,194 52,917 16,658 25,687 149,456 Q4 54,867 53,176 21,066 21,219 150,328 1960-Q1 54,711 61,812 18,233 17,722 152,478 Q2 48,527 64,472 18,490 18,056 149,545 Q3 53,297 60,566 19,517 18,024 152,204 Q4 57,125 59,156 15,903 21,331 153,515 1961-Q1 57,703 51,398 23,441 18,982 151,524 Q2 63,287 47,924 21,718 18,463 151,392 Q3 65,151 51,404 18,062 21,124 155,741 Q4 65,526 55,763 15,961 21,350 158,600 56 Table 20 - Continued Within 1 year 1-5 years 5-10 years Over 10 years Total 1962-Q1 67,843 49,463 19,365 21,653 158,324 Q2 68,508 47,378 21,564 20,389 157,839 Q3 66,047 46,686 27,297 18,752 158,782 Q4 67,952 49,381 29,158 16,061 162,552 1963-Q1 62,056 49,231 32,831 17,931 162,049 Q2 61,955 48,073 32,299 18,034 160,361 Q3 62,296 45,424 33,843 18,942 160,505 Q4 64,979 47,919 30,525 18,666 162,089 1964-Q1 63,175 49,326 30,792 19,156 162,449 Q2 61,573 48,814 30,090 19,098 159,575 Q3 61,055 42,689 37,758 20,048 161,550 Q4 65,331 48,021 31,477 18,435 163,264 1965-Q1 62,162 47,490 32,509 20,121 162,282 Q2 59,222 43,782 34,174 20,043 157,221 Q3 61,458 43,992 30,234 19,726 155,410 Q4 67,198 43,349 30,214 19,639 160,400 End of quarter figures. The first and second columns should and do agree with Okun's [ 49 ] data - they only extend till 1959-Q4. Treasury Bills are included in column 1. Source: Federal Reserve Bulletin. TABLE 21 U.S. Federal Government Bond Rates 90 day Medium term Long term Treasury Bill bond yield bond yield 1955-Q1 1.23 2.06 2.69 Q2 1.48 2.34 2.76 Q3 1.86 2.60 2.89 Q4 2.34 2.67 2.85 1956-Q1 2.33 2.72 2.86 Q2 2.57 3.03 2.96 Q3 2.58 3.28 3.10 Q4 3.03 3.47 3.30 1957-Q1 3.10 3.40 3.26 Q2 3.14 3.60 3.43 Q3 3.35 3.96 3.63 Q4 3.30 3.55 3.56 1958-Q1 1.76 2.56 3.25 Q2 0.96 2.07 3.15 Q3 1.68 2.92 3.57 Q4 2.69 3.59 3.75 1959-Q1 2.77 3.90 3.91 Q2 3.00 4.26 4.06 Q3 3.54 4.70 4.16 Q4 4.23 4.89 4.17 1960-Q1 3.87 4.67 4.22 Q2 2.99 4.22 4.11 Q3 2.36 3.48 3.82 Q4 2.31 3.54 3.91 1961-Q1 2.35 3.40 3.83 Q2 2.30 3.38 3.80 Q3 2.30 3.80 3.97 Q4 2.46 3.61 4.00 1962-Q1 2.72 3.61 4.06 Q2 2.71 3.37 3.89 Q3 2.84 3.49 3.98 Q4 2.81 3.32 3.88 58 Table 21 - Continued 90 day Medium term Long term Treasury Bill bond yield bond yield 1963-Q1 2.91 3.39 3.91 Q2 2.93 3.53 3.98 Q3 3.29 3.77 4.01 Q4 3.50 3.90 4.10 1964-Q1 3.53 4.03 4.16 Q2 3.48 4.05 4.16 Q3 3.50 3.96 4.14 Q4 3.68 4.06 4.14 1965-Q1 3.89 4.17 4.15 Q2 3.87 4.14 4.14 Q3 3.86 4.19 4.20 Q4 4.16 4.53 4.35 Source: RDX1 Source: IMF Source: IMF Data Tape Financial Financial Statistics Statistics The IMF data are quarter averages. 59 CHAPTER TWO THE CONVERSION LOAN: A SURVEY OF THE LITERATURE The current paradigm on the effects of the Conversion Loan is as follows: The Loan caused an outward shift in the demand for money which, given the rather restrictive monetary policy of the time, led to an increase in the interest rate. This development may have discouraged domestic investment expenditures but it had its greatest effect by attracting foreign capital which in turn led to an exchange rate appreciation. At a time when the current account was not particularly healthy this was quite undesirable. And so it is concluded that "An addition to our GNP amounting to several billion dollars has been lost forever."! The widespread acceptance of this view would lead one to believe that many long and involved studies of the Loan have been undertaken, all of which point to the same conclusion. What is the evidence on which current opinion^ is based? Although there exists a large body of literature which pertains to the Loan, there appear to be only two papers that concern themselves with this episode directly: Barber's submission to the Royal Commission of 1962 (4) and Shearer's 1964 paper (57). Part of the circumstantial evidence has been noted in chapter one, the remaining being mainly Shearer (56), Wonnacott (68), Johnson and Winder (34), Breton (8), Goodhart (22), Shearer (58) and Miles (43). The discussion will be centred around the two papers directly concerned with the Loan. 60 The paper by Barber is really the foundation for the current paradigm and so it is necessary to examine it in detail. The evidence for the rise 3 in interest rates is contained in a table which gives time series on interest rate differentials between specific, long, issues of Canadian and U.S. Government securities. What should this increase in rates be attribut to? "It seems clear that one of the important effects of the conversion loan was to increase substantially the public's demand for cash. Individuals and financial institutions who were induced to exchange short term securities for the much longer term and less liquid conversion loan bonds would naturally want to hold more cash in their portfolio to prevent a serious decline in their liquidity position. This need not have . resulted in a serious rise in interest rates had the Bank of Canada been willing to allow a substantial increase in the money supply. But except for a short period during, and immediately after, the loan, the Bank of Canada was not willing to provide the basis for such an increase."4 How does Barber substantiate his claims of a shift in the demand for money function? He plots the yield on long-term government securities against the ratio of money supply to GNP, where money is defined both narrowly and widely. The observations for the period 1958Q4 - 1961Q3 lie above those for the pre-Loan time span. This is more pronounced when the money component in his independent variable, the inverse of velocity, is defined narrowly. Barber strengthens his argument that an excess demand for money, cause by a shift in the demand curve, is responsible for the increase in interest rates by attempting to discount two other possibilities. He simply dismisses^ the possibility that the alleged excess demand for money may 61 have been caused by restrictive monetary policy, i.e. a shift in the supply curve rather than the demand curve. The possibility that the alleged shift in the demand curve for money may have been caused by expectations of inflation is also examined. He reasons: "It has sometimes been suggested that the rise in interest rates during this period reflected a shift in investor preferences away from bonds and towards equities as a result of a fear of continued inflation and that higher interest rates in effect today contain a premium to offset an expected long term rise in the price level. If this view were correct one would expect to find a similar development in the United States." To test his conjecture Barber inspects graphs similar to those discussed above but constructed using U.S. data and concludes that, although post-Loan observations lie above the pre-Loan ones, we would not be justified in concluding that the U.S. demand curve for money shifted too.1 There fore, it is concluded that the Conversion Loan has been identified as the cause of the rise in interest rates. The view that high interest rates led to exchange rate appreciation by developing a balance of payments surplus through the capital account is Q also dealt with briefly in Barber's paper. It is my contention that: A. Some of the empirical assertions made in Barber's paper have not been statistically validated. B. The model used by Barber is not made explicit. When a suitable model is explicitly employed some of Barber's arguments would appear to be of doubtful validity. 62 C . Barber implicitly accepts the Expectations Hypothesis of the term structure of interest rates as evidenced by his concern for the behaviour of the rate of interest. This approach prevented him from exploring some interesting and important issues. To be more specific; concerning point A, the following observations can be made. A(i) The rise in interest rate differentials between Canada and the U.S. has not been shown to be statistically significant. A(ii) Also, the rise in the Canadian rate has really been taken for granted. A(iii) The extent to which increases in Canadian rates, such as there occurred, are due to a shift in the demand curve for money is ques tionable. Presumably the mechanism Barber had in mind runs somewhat like this: An increase in the average maturity of the outstanding debt of the federal government held by the Public, m, increases the price variability of portfolios^ and hence decreases liquidity.^ Then, if m is an argu ment in the demand for money function, the Conversion Loan could exert an influence in the money market by changing m. This is all fairly plaus ible but Barber does not present empirical evidence on such a demand for money function. It is, therefore, necessary to turn to other sources for evidence on this score. Such evidence is not unambiguous. Breton (8), provides what would appear to be pertiment evidence. He estimates, using Canadian data for 1935 - 1959,^ 63 V = 1.983 + 0.3236R - 0.0973m* R = 0.859 (0.0606) (0.0166) where V s Y/M, Y = Roughly as GNP, M = Currency in circulation + demand and savings deposits held by the Public as well as by federal, provincial and municipal governments, R = Rate of Interest on long-term federal government bonds and m* = Average maturity of the outstanding debt of the federal government. The significance of m* indicates that, when plotting 1/V against R and an increase in m* occurs, the relationship between 1/V and R shifts out, as Barber has found. However, Breton's results have been effectively criticised. Goodhart's attempts (22) to extend the estimation period after 1959 failed disasterously. Moreover, Johnson and Winder (34) pointed out that the choice of the interest rate variable is not without consequences: Using Barber's data wherever possible they concluded that r, the Treasury Bill rate, is more appropriate. They thereby highlighted the dilemma that any study using one interest rate must face: Which interest rate? No respectful reference to Keynes, or the work of previous resear chers, can resolve the problem! In their study Johnson and Winder used m, rather than m*. Finally, Shearer (58), using r, pointed out that the definition of M is an issue of substance. Thus, the function implicitly used by Barber is at least controversial. So much for observation A(iii). A(iv) Finally, Barber produces no empirical evidence to support his important point that the increase in Canadian rates led to increased capital inflows which appreciated the exchange rate and led to a loss in potential GNP. This omission is particularly important: Since Barber's argument does not rest on the usual mechanism, that increases in interest rates decrease investment and income, the argument for a loss in potential GNP must rely heavily on the link between exports and imports and the exchange rate. Turning to point B, we note that, even if we accept Barber's test for establishing the presence of inflationary expectations, namely examining whether there was a shift in the U.S. relationship between 1/V and R, Barber's conclusion that there was no such shift may be unwarranted --he presents no evidence on this score. But, more importantly, the whole argument that inflationary expectations shift outwards the demand^ function for money is unconvincing when viewed from the viewpoint of the models proposed by Tobin in (65), (66) and (67). In the usual Keynesian model, which Barber is presumably using, money is juxtaposed with bonds and real capital, as Tobin clearly points out.13 if bonds and capital are perfect substitutes then it would appear pointless for an investor to get out of bonds and into equity. Additional to this is the difficulty that lies with the Keynesian model itself, rather than Barber's usage of it: Neither bonds nor money will immunise their holder from inflation, while real capital will. Tobin's model, in which money and bonds are juxtaposed with capital^, would" seem more appropriate, but then the shift in the l/V function would be towards the origin, not outwards. The above objections deal with the assertion that an excess demand 65 for money developed because of a shift in the demand curve. It is also possible, however, that such an excess demand may have arisen because of supply considerations. The quotation from Barber on page 60, and the evidence in chapter one, indicate that monetary policy was restrictive after the price support programme was abandoned. Thus, the interest rate may have risen because the supply curve for money shifted to the left. If so, we should be discussing a movement along the demand curve for money, not a shift in it. Point C relates to what is, perhaps, the most serious criticism that can be made of the analysis in Barber's paper. The survey of theories of the term structure of interest rates in the next chapter contains two well-known predictions of the Expectations Hypothesis: Firstly, that the "effective rate of return" on assets with different terms to maturity is the same,and, secondly, that the relative supplies of such assets have no effect on the term structure of interest rates -- unless they affect the way in which expectations are formed. Then we need only concern our selves with the determination of the "effective rate" and this is essen tially what Barber does — he is implicitly using a mechanism that deter mines the effective rate. However, it is conceivable that the degree of market segmentation is far more serious than Barber assumes. If so, the Conversion Loan may have led to increases in long rates and decreases in short ones, leaving an average rate unaffected. It is also conceivable that short and long rate elasticities in investment and capital flow functions differ substantially so that, given an average rate, different yield curves imply 66 different levels of income and employment. Thus an examination of the behaviour of rates on bonds with different terms to maturity becomes essential. It is concluded that while the Barber paper raised some fascinating issues it did not go far enough. This, of course, is hardly surprising given the state of the art in the 1950's. The paper by Shearer (57), the only other study explicitly dealing with the Conversion Loan, will now be examined. Shearer (57) argues that at least part of the interest rate increase was due to the highly restrictive monetary policy. He quotes Smith's (59) views on debt management operations, namely that such policies are unlikely to have liquidity effects which bear on expenditures directly and that small quantitative differences in the importance of short and long rates on the real sector can be found. Rather, wise debt management operations that maintain a long debt -- leaving few highly liquid assets -- improve the potency of monetary policy. This is precisely what the Conversion Loan did and hence, Shearer concludes, post-Loan monetary policy may have been highly effective. It is worth noting, however, that monetary policy was not restr ictive until after the abandonment of the bond price support programme of the Bank of Canada, i.e. until after 1958Q4. The reader will recall that Barber simply dismissed this alternative explanation. Another line of argument in Shearer revolves around the performance of the Bank of Canada following the Conversion Loan. It was noted in chapter one that bond prices were pegged throughout October 1958, an action that 67 involved the Bank in purchases of securities. In the following months, however, bond prices were left free to fluctuate and the Bank started increasing its holdings of longs relative to shorts. Shearer treats this as something distinct from the Conversion Loan but the question could conceivably be viewed as one of timing. In other words, the Conversion Loan was spread over the last two quarters of 1958 and possibly 1959Q1. This will be the approach taken here. This discussion has covered the two most direct contributions to the problem and has found that widely accepted views on the effects of the Loan are largely based on circumstantial evidence and the bare minimum of analysis. Important questions remain unanswered: Did interest rates increase signifi cantly? Do rates on bonds with different terms to maturity behave differently following the Loan? What determines the spread between long and short-term rates? Is it the relative supply of assets, monetary policy, or some other factor? Does a change in the term structure of government bond yields lead to adjustments in the issuing patterns ofother debtors e.g. provinces, munici palities and corporations? Did the Loan really forestall an addition to our GNP, as Barber contends? What were the channels through which the Loan led to an economic contraction? The reader will concede that such questions must be examined. Before proceeding it will be instructive to consider how the effects of Operation Twist, a close relative of the Conversion Loan, have been examined. This will provide guidance and a standard against which Canadian experience can be compared. A survey of all studies on Operation Twist known to this writer appears as an appendix to this chapter, on pages 205 - 209. 68 NOTES TO CHAPTER TWO 1. Barber (4), p. 3. 2. It is as well to give some examples of such opinion. Barber takes the position stated in the opening paragraph of this chapter -- see Barber (3) and (4). Officer and Smith (48) state that the Conversion Loan increased interest rates and, therefore, contributed to the exchange rate appreciation -- see p. 35. Boreham e_t al (7), take essentially the same view but specifically attribute some of the capital inflows to provincial and municipal borrowing in the U.S., p. 578. Like Barber, they believe that the Conversion Loan led to an increase in unemployment. O'Brien and Lermer (47) draw our attention to the possibility that the Loan may have discouraged new investment projects by increasing the interest rate on long bonds -- p. 336. Finally, Bond and Shearer (6), rather cautiously, report Barber's views. 3. Barber (4), Table 2, p. 4. 4. Barber (4), p. 5. 5. Barber (4), P. 5. 6. Barber (4), P. 6. 7„ Barber (4), P. 6. 8. Barber (4), P. 9. 9. This is because "For a given change in yield from the nominal yield, changes in bond prices are greater the longer the term to maturity." Malkiel (39), p. 54, gives a simple proof that follows from the mathematics of bond pricing. 10. A liquid asset is one whose par value is, according to Keynes "More certainly realisable at short notice without loss." Keynes (36), Vol. II, p. 67. Since long bonds fluctuate in price more than short ones the probability of realising the par value of a long bond without--or alternatively with a given--loss is smaller than that of a short one. Hence, long bonds are less liquid. 11. Breton (8), p. 453. Unfortunately he does not report any other statistics. 12. So is, therefore, Barber's test. 13. See Tobin (66), pp. 158 - 167. 14. See Tobin (66), pp. 159 - 160. 69 CHAPTER THREE EFFECTS OF THE CONVERSION LOAN ON THE TERM STRUCTURE OF INTEREST RATES INTRODUCTORY Chapter one showed that the Conversion Loan affected significantly the maturity composition of the public debt, increasing bonds with more than ten years to maturity and decreasing those under that mark. The survey of the literature in the second chapter indicated that, while concern has. been expressed about the effects of the Loan on interest rates, little has been statistically established about their behaviour. More specifically, the following questions remain unanswered. i) Did interest rates rise over and above trend values? ii) Are we to attribute increases such as there occurred to the Loan •—Barber's contention—or to restrictive monetary policy—Shearer's argu ment -- or to some other factors? iii) Finally, since the Loan was a debt management operation, did it affect the term structure of interest rates? This chapter attempts to answer these questions. The discussion opens with section one in which the behaviour of time series of various interest rates is considered. The second and third issues are also con sidered there, albeit rather briefly. A more extensive treatment of 70 them can be found in sections three and four which follow a brief survey of the literature on the determinants of the term structure of interest rates in section two. The results obtained en route to answering questions two and three above do not conform to those obtained in the U.S. The differences are so striking that it was felt necessary to deal with this paradox separately—in section five. There, two explanations of the paradox are considered. SECTION ONE: THE CONVERSION LOAN AND TIME SERIES OF INTEREST RATES Following the Conversion Loan did interest rates increase, as often alleged? The answer to this question is in the affirmative but it is hardly illuminating: It is well-known that in the post-war era there has been a marked upward trend in all interest rates. Under those circum stances a more appropriate question would be: Following the Conversion Loan, did interest rates increase above trend levels? To obtain an answer a number of interest rates were in turn regres sed against time. The ensuing residuals Y-Y became time series that could be -visually inspected. Moreover, this procedure made available the usually useful notion of a confidence interval for Yij-Yij, where i stands for the ith observation in the equation for the jth rate. We say usually because a confidence interval is of little value when the estimated equation j, for example, is not reliable in terms of goodness of fit."'" Both intra-sample and extra-sample confidence intervals were constructed for a large number of interest rates in the periods 1951Q1 - 1967Q4 and 1951Q1 - 1958Q2 2 respectively. These calculations indicated that: 71 i) The simple average of rates for all maturities, SA4, increased 3 above its trend values during 1959Q2 - 1960Q1, thus confirming the view that following the Conversion Loan interest rates in general increased. It is noteworthy that this interest rate did not significantly increase until well after the Bank of Canada had stopped supporting bond prices. ii) When interest rates on bonds with varying terms to maturity were examined some interesting patterns emerged. In view of the substantial changes in the maturity composition of the debt effected by the Conver sion Loan it would be reasonable, according to some theories of the term structure of interest rates, to expect the yield curve to change. Since longs were increased at the expense of shorts some theorists and most market analysts would expect long rates to rise and short ones to fall. This belief was partially corroborated by experience. For approximately five quarters, i.e. between 1958Q2 and 1959Q3, short rates rose signifi cantly above their trend values, but they declined significantly during and following 1959Q4. This decrease below trend after 1959Q3 was eviden ced most clearly in the behaviour of the Treasury Bill rate, the most -volatile of the short rates. Long rates rose significantly when the Loan was implemented. iii) There is some evidence that the bond yield on securities with five to ten years to maturity behaves like the over ten rate. This may suggest that although the quantities of bonds in the five to ten year category were decreased, the movement in the five to ten year rate does not reflect this because bonds in the two categories are close substi tutes—recall that the increase in bonds in the over ten slot is larger than the decrease in bonds with five to ten years to maturity. However, too much can be made of this point since the evidence for it comes from 72 the regression of r,. ^ on time for the period 1951Q1- 1958Q2. When the estimation period is extended to 1967Q4 and the intra-sample confidence intervals constructed the post 1959Q4 observations lie below trend. Thus, the view in the literature that the Conversion Loan increased the interest rate is true but overly simplistic. When time trends have been accounted for there is some evidence that, after 1959Q3, short rates decreased while long ones rose. During the five-quarter period between the announcement of the Loan and 1959Q3 both short and long rates increa sed. This may be due to changes in variables other than supply ones, as Shearer (57) has suggested. Question (i) above has now been dealt with and the issues raised by the second and third ones given some airing. In the following sections they are considered explicitly. Before doing so it is necessary to review the existing literature on the determinants of the term structure of inte rest rates. This review will establish just how to proceed in answering questions two and three. More specifically, we will see whether and how debt management, monetary policy and other factors can affect the level and term structure of interest rates. Then in sections three and four an attempt will be made to quantify the relative importance of such influences. SECTION TOO: DETERMINANTS OF THE TERM STRUCTURE OF INTEREST RATES A. FORWARD RATES: THE HICKSIAN FORMULATION This survey begins with Hicks' (29) treatment of the problem. He analyses the functioning of an economy from the general equilibrium point of view. There are naturally the commodity and money markets, as always, but in addition Hicks considers the markets for short-term bonds and the forward markets for short loans ranging from two ton periods into the future. Given a numeraire, the system of demand and supply equations determines commodity prices, and the variables of particular interest to us, namely the short rate and the n-1 forward rates. However, the task is not yet complete in that there exist variables such as the market rate on a loan which extends over i periods, R_^, about which little has been said. At this point the reader should recall that the assets corresponding to loans of duration i periods are treated as identical in all respects except maturity, a convenient assumption. Hicks also makes the useful point that a long loan can be conceptually decomposed into a short one-period loan plus a number of forward loans of the same length. If so, an investor should be indifferent as to whether he holds a two-year bond, 2 which in two years pays him (l+R^j.) , or a one-year bond paying, in a year's time, the sum Cl+R^t), which he immediately reinvests according to a forward contract to reap (1+R^ ) (l+2rlt^ at t*ie en^ °^ t*ie two Peri°ds. The symbol 2rit rePresents> in the usual notation, the forward rate on a one-year loan, agreed upon in period one, but not commencing till period two. In fact, there is more to it than that: If [1] Cl+K2t)2 * (1+Rlf;) (l+2rlt) there are gains to be made through arbitrage and the market will ensure that in equilibrium an equality holds. In general, [2] (1+Rnt)n= (1+RU) d+2rit> <1+3rlt) •' * (1Vlt> The last n-1 equations complete the Hicksian system since they determine the n-1 long rates R„ , R„ ...R . ° 2t 3t nt It is important to note that the only statement about behaviour that equations [2] make is that investors are profit maximizers, ready to engage in arbitrage should the opportunity arise. However, this is often assumed. In that case equations [2] become merely equilibrium conditions. The only behavioural relations are to be found in the determination of R It and' r...... r. . In what follows it will be assumed that investors are 2 It n It profit maximizers. It should be emphasized that a concise statement of the problem is not to be found in Hicks. Rather, the discussion above condenses his treatment in chapters 10, 11 and 12 of (29) in a manner that hopefully does not misrepresent him. An important implication of this analysis is that the term structure depends crucially on demand and supply considerations since it is such influences that determine the short rate R^fc and the n-1 forward rates ^r^. . -^j.- Also associated with Hicks is the argument that investors may require a premium in order to hold long bonds. This issue will be examined later on. B. EXPECTED RATES: PERFECT FORESIGHT While Hicks' construct as presented above appears internally consis tent and appealing, formulations that have evolved from that model have recently gained more currency. Although it is entirely possible to engage in a forward lending contract through a judicious combination of borrowing 75 and lending, forward markets for loans do not in fact exist. It is pos sibly for this reason that attempts have been made to replace the use of forward rates by expected ones in theoretical discussions on the matter. We now turn to these. Here again we are considering securities identical in all respects except maturity and, as before, transactions costs involved in getting in and out of specific maturities are assumed not to exist. These assump tions will be made throughout what follows. If now we also assume perfect foresight then, it is widely agreed, the term structure will be determined according to the principle that all assets should yield the same rate of -return—inclusive of capital gains. Given that no coupon payments are made -until maturity,i.e. interest is compounded, equations [2] describe the term structure. The meaning of the small r's is slightly changed. There are now no forward markets but investors happen to know what the •rate on a one-period security will be one year, two years...n years into the future—this is the meaning of perfect foresight. Note that mere foreknowledge of these rates is sufficient to make equations [2] hold; the process through which this happens is the same as in the Hicksian model. No behavioural statement is to be found in the perfect foresight model if we assume, as we have,profit maximizing investors. This was pointed out by Meiselman. When the assumption of perfect foresight is dropped, opinions on how the term structure is determined begin to diverge. At this point it is usually assumed that imperfect foresight can be associated with uncertainty 76 about the future level of short rates. In the presence of such uncer tainty the term structure is thought to be determined according to one of five hypotheses: Expectations, Liquidity Premia, Segmented Markets, Pre ferred Habitat and the General Equilibrium Approach. As Meiselman put it6 "Alternative hypotheses of the determination of the term structure revolve about the central analytical and empirical problem of how the market copes with interest rate uncertainty." C(i). THE EXPECTATIONS HYPOTHESIS (= EH) When deprived of the omniscience implicit in the perfect foresight model we must supply a mechanism that generates 2 r±t' 3rlt""'nrit' ^ this point it is usually assumed that.investors hold "firm and uniform" expectations about these rates.'' A number of hypotheses concerning the formation of such expectations have been advanced in the literature and they are considered later. If investors generate firm and uniform expecta tions, then the EH predicts that equations [2] hold in equilibrium. It should be noted that the symbolnow stands neither for the forward rate, nor the divinely known one, but for the one-year bond rate firmly expected by all now to prevail in the ith period. The EH infuses economic content into equations [ 2 ] over and above what was to be found there before. It asserts that expected rates are unbiased estimators of the rates in models A and B. This prediction is very difficult to test. To begin with, the rates ,r. of models A and B. • I lt are not known. Also, if we were to compare 0r1 with next year's short rate, i.e. R^t+-^> we would, according to Meiselman, not be conducting a proper test of the hypothesis because "...anticipations may not be realised yet still deter mine the structure of rates in the manner asserted by the theory."8 Mieselman's statement makes some sense when it is recalled that in Hicks' framework is determined through the juxtaposition of money and bonds, a process quite different from that generating the , as will be seen below. It may be instructive at this point to count equations and unknowns. The problem is to determine n-1 rates ^2t'''^nt* T° tnat en(^ t*ie EH introduces n-1 short rates' r. ... r. and then uses them along with R. / It n It t and equations f21 to determine R_ ...R L J 2.x. nt How are these expected rates determined? The literature on this issue is relatively limited in scope and volume. Two issues have been been rai sed. The first one revolves around the concept of the elasticity of expec tations discussed by Hicks (29). It gives the percentage change in expec ted future short rates given a percentage change in the current short rate, i.e. given functions 13] ±rlt-f(Rlt) i = 2...n the elasticity of expectations is given by r/1 Arlt Rlt . [4] e. = l = 2...n 1 9R-. • ri. lr I It The discussion pertaining to this concept appears in "The Working of the 78 Dynamic System", the last part of (29), and it compliments the discussion with which this survey opened. The elasticity of expectations provides a good way of summarizing what has been called the Keynes-Duesenberry controversy on the formation of expectations. Associated with Keynes is the hypothesis that the market expects rates to return to some normal level N should they chance to move away from it -- regressive expectations. In symbols ^ t+irit = Rlt + ^"V 0 < k < 1 By contrast, associated with Duesenberry is a statement to the effect that the market expects movements in interest rates away from an expected rate E to continue in the same direction -- extrapolative expectations. One possible formalization is W t+irit = Rit + d(Rit"E) 0< d< 1 Clearly the two types of expectations require different values of the elasticity of expectations. For extrapolative expectationse ^> 1, while regressive expectations requiree ^< 1 regardless of the value of • 9 The figure below illustrates this. When R, increases from R, to R' lt lt It' j? extrapolative expectations give r > R' while regressive expecta-t+1 lt lt R tions would lead to r < R1 t+1 It • It 79 FIGURE 1: Extrapolative and Regressive Expectations t+irit Extrapolative A word about the constants k and d is in order. These may be viewed, somewhat heretically, as speed of adjustment coefficients. To illustrate, if k is close to unity, the line RR, whose slope is (1-k), will be relative ly flat. The flatter the RR line i) the smaller the changes in t+^r^t given a change in R^fc and ii) if .,r = R i.e. anticipations are realised, the faster the t+1 It lt+1 return of R^t to its normal level, N. Turning to the constant d, the larger d is the steeper EE will be, since 80 its slope is (1+d) and hence i) the larger the change in r given a change in R and & 6 t+1 It t ii) if anticipations are realised, the more quickly will R explode. It is sometimes argued that the normal rate involved in the regressive case is a weighted average of R ., i = l...n with roughly equal weights assigned to each R . In the extrapolative case, however, the more recent R, . the greater its weight is assumed to be. Figure 2 illustrates lt-i & a one possibility. FIGURE 2: Extrapolative and Regressive Weights Weights ts Time n It is now necessary to examine briefly the effects of the two mechan isms on the term structure of interest rates. The following assumptions are made: The yield curve is flat to begin with, there exist only two periods, expectational machanisms are common to all and anticipations are not necessarily realised. Let there be an increase in R, . Then extra-J lt polative expectations lead to an upward sloping yield curve while regres sive expectations lead to a downward sloping one, as figure . 3 illustrates. 81 FIGURE 3: Expectations and the Yield Curve Rates R It R2t-1 _ Rlt-1 Term Note that if we, rather unreasonably, assume that anticipations are realised then the yield curve shifts up parallel to itself and stays there if £2=1, it becomes steeper as time elapses if > 1 and conversely for < 1. When the two-period assumption is dropped and we allow to have a different value depending on i the yield curve could take any shape. The second issue concerning the formation of expectations was raised by Meiselman who investigated how expectations are revised. He considered the hypothesis that 82 m (irit - i^t-^ = f(Rit - trit-i> While this approach may say something about how expectations are revised its empirical verification or falsification has no bearing on the validi ty of the EH per se despite. Meiselman's and others' assertations to the contrary. It was mentioned earlier that the n-1 expected rates, along with R and equations [2] determine the term structure of interest rates. We have as yet said nothing about the process that determines R^t« It is usually presumed that, if we assume a short enough period so that no appreciable change in the capital value of the shortest asset can occur, R^t will be Conard's "effective yield". This is the rate of return that includes capi tal gains and losses and, given the EH, it is the same for all assets. It is then argued that the general theory of interest determines R through the juxtaposition of money and bonds. Thus, given expected rates and R the rate structure is determined through equations [2]. What predictions does the EH make? Given arbitrage they are i) That expected future short rates are unbiased predictors of for ward rates. ii) The "effective yield" on all assets is the same^. iii) The hypothesis is consistent with any shape for the yield curved iv) Abstracting from the problem of how expectations are formed, when short rates are expected to rise the yield curve will be upward sloping. v) The relative supplies of assets do not influence the term struc ture. 83 vi) However, the position and shape of the yield curve will be affec ted by monetary policy. This follows from the mechanism that is supposed to determine R and from prediction 5. A decline in the money supply, for example, will increase But, since longs fluctuate less than shorts, assuming that expectations remain unchanged, R2 will increase by less than AR^^. Thus, the yield curve will be displaced upward and the spread between longs and shorts decrease algebraically. This is a fairly lengthy list of predictions but the theory is diffi cult to test. This may appear curious in view of the large number of em pirical studies undertaken, but there is no doubt that such studies either assume that anticipations are realised, or they conduct a joint test of the EH and a particular expectations formation mechanism. Even testing for the significance of supply variables will not do as a test of the EH. Should they prove significant, as they are in Canada, that may be because they affect the way in which expectations are formed—not because equations [2] do not hold. C(ii). LIQUIDITY PREMIA HYPOTHESIS (= LPH). Hicks qualified the analysis presented earlier on by an argument that 12 is by now well-known. While the expectations theory argues that in the presence of uncertainty investors behave as if they are indifferent to risk, Hicks maintained that they are risk averters. Since a long contract can be decomposed into a short contract and a series of forward short ones it is easy to see that more risk is involved in a long contract than in a short one. This is because future short rates are unknown, though uniform 84 and firm expectations about them may be held. Hicks then asserts that "...The forward market for loans may be expected to have a constitutional weakness on one side...If no extra return is offered for long lending, most people... would prefer to lend short...But this situation would leave a large excess of demands to borrow long which would not be met. Borrowers would thus tend to offer better terms in order to persuade lenders to switch over into the long market."x3 The explicit recognition of demand and supply forces in the argument is worth emphasizing. Forward rates are higher than they would be and hence, through equations [2], long rates are higher than the corresponding ones in the absence of risk aversion. The modern statement of the argument is somewhat different, consistent with the use of the term expected rates rather than forward ones. It is then argued that we should rewrite equa tions [2] as [8] (1+R )n = (1+R )Cl+9rn +L.)...(1+ r +L ) nt It z It JL n It n where L^, i = 2,...n are liquidity premia. The economic content of [8] is now that the expected future short rates are biased estimators of the true, or forward rates, where now [9] .p. = .r. +L. i = 2.. .n X It X It X are the forward rates under risk aversion and a constitutional weakness. While the Hicks statement is quite consistent all the problems dis cussed in the section on the expectations hypothesis carry over to this one. There are, however, certain implications buried in the modern state--ment of the LPH which one does not see treated in the literature. It is often implied that just as demand and supply were unimportant forces under the EH, they are of no consequence in the LPH either. This would appear to be quite wrong. The idea that there exist was lifted right out of Hicks but with them goes the mechanism through which they are determined. One of the predictions of the LPH would then be that the extent to which expected rates underestimate the ^P-^t's depends, given i, on demand and supply for forward, or, equivalently, long loans. Moreover, given demand and supply considerations, expected rates will increasingly underestimate the ^P^t's as 1 varies from 2 to n. The popular statement of this is L_ < L0 <... L . Of course, if Meiselman's rules of the game are accep-2 3 n ° • r ted this too is an untestable hypothesis. Particularly since the seve rity of the constitutional weakness argument is an empirical matter— 14 Meiselman, for example, makes an empirical statement ' contrary to Hicks'. C(iii) . SEGMENTED MARKETS HYPOTHESIS ( = SMH) It has been argued by Culbertson (18), that in the presence of risk, investors in general, but certain important financial institutions in particular, hedge rather than speculate: An institution committed to a stream of payments at future dates can only insure itself against income and capital loss risk by holding assets which expire on the date on which future payments are due. If it were to hold shorter term assets it would be taking an income risk since funds can only be reinvested at what now is an uncertain rate. If, on the other hand, it invested in assets matur ing after the payment dates it would have an assured income but not an assured capital value. An implication of the SMH is that changes in the structure of bond supplies outstanding will affect the yield curve. The SMH predicts the shape and position of the yield curve, ascribes the variability of short rates relative to longs to the more frequent and substantial intervention of the authorities at the short end, e.g. Bills Only and, finally, dis agrees with the EH on the latter's prediction that the effective rate of return on all assets will be the same. The SMH is appealing in that, for a change, we now have an hypothesis that makes testable statements about the state of the world. C(iv). PREFERRED HABITAT HYPOTHESIS (= PHH) According to the authors of this theory (44), this is a blend of the preceding three. It basically accepts the LPH but introduces elements of -market segmentation in the process by which the yield curve is determined. The most satisfactory way of presenting the PHH, and this is not the pro cedure used by Modigliani and Sutch, is to recall earlier comments that in Hicks, demand and supply determine the L^. Hicks made his theory rather specific by assuming that, on the whole, people preferred to lend short and, hence, the subsequent convention that < < ...< More generally, however, there are investors who have funds available for i periods and who, in pursuit of insurance against both income and capital value risk, would prefer to hold a bond with exactly i periods to maturity. Then the struc ture of the will depend on demand and supply in each of the n-1 markets. In a sense the PHH, as interpreted here, is really the SMH applied not to the p 's but to part of them, namely the L.. The expectation formation i It x mechanisms discussed in C(i) are relied upon to produce the . r 's in a 87 somewhat roundabout way.'*'"' C(v). THE GENERAL EQUILIBRIUM APPROACH (= GEA) This theory has not been advanced as an explanation of how the term structure is determined but rather as a way of examining the workings of an economy at the macro level. However, it does involve a theory of the term structure which In a sense has much to share with the preceding hypo theses. It is useful to proceed by way of an example which, while not given by the proponents of this approach, namely Tobin and Brainard, is -undoubtedly in the spirit of the GEA. We are now considering a financial sector which contains for simpli city money M, shorts S and long securities L, all of which are the liabili ties of sectors "outside" the system and are treated as exogenous. Money is regarded as demand debt bearing a fixed nominal rate of return r'^. Income and the price level are treated as exogenous to this sector and the latter is normalized at unity. We may then write the following demand functions: [10] X. = X. (R,Y,W) rM = r'M -p e , e rs " r S _P t e rL = r L "P i = M, S, L 6 where R is the vector of real rates of returnrM, r and r^, p is the expected rate of inflation, r' , r' , r', are nominal rates. All nominal Mb" rates, p and Y are exogenous. Also 88 [11] W = M + S + L where W is to be understood as the wealth of the private sector. We have two independent equations in [io! and two unknowns r and r . The system essentially determines the term structure and the predictions of the model are given in Table 1 . Supply and demand forces are clearly important. TABLE ONE: Signs of Partial Derivatives in the Reduced Forms for rs and ri, Exogenous Endogenous M S L L-S Y rs - + ? - + + TL - + + + + The following assumptions are sufficient to yield these results. i) Partials with respect to own rates are positive, cross ones are negative subject to the condition ii) E SXi : 9z i 0 where X stands for an asset, i = M, S, L and Z = r , r , iii) || > 0, || < 0, || < 0 subject to l |f- = 0 i dXi iv) Partials with respect to wealth are positive subject to X T;— = 1 89 v) The wealth variable in the individual and hence aggregate demand functions for the three assets is defined at market values. This usual assumption leads to conceptual difficultueswhich are ignored here. Doing so does not change the results qualitatively. We now turn to the connections between the GEA and the other theories of the term structure. The strongest and most obvious connection is to be found with the SMH and PHH: The common reliance on market forces to determine wholly, or for the PHH partially, the yield curve. The relation to.the LPH is less obvious but still strong. The income variable in the demand functions for assets carries with it the notion of liquidity. Suppose, for example, that originally the yield curve was flat, i. e. r = r„ = rT. Let there be an increase in Y. This increases an asset M S L holder's demand for liquidity and hence his demand for money. The result ing "constitutional weakness" forces r and r above their previous levels, as the last column in Table 1 indicates. Finally, on expectations: This link is not entirely clear. Nevertheless, one of the features of the EH was its treatment of the various assets as perfect substitutes. While the GEA does not go that far, it does allow for the cross elasticity of demand to be large subject,of course, to the condition contained in assumption (ii) above. . Other hypotheses or emendations of the ones already discussed have been proposed but no attempt will be made to cover them, primarily because the common denominator between them and the five above is not large. There is also the vast empirical literature on the matter. It is not examined here because the primary concern is with the Conversion Loan, not testing 90 the various hypotheses advanced. The preceding discussion provides guidance in the following attempts to examine the effects of the Conversion Loan on the level and term struc ture of interest rates. The two most general hypotheses, namely the GEA and the PHH will be used to examine the effects of the Conversion Loan and other contemporary changes on interest rates. SECTION THREE: ESTIMATING THE GENERAL EQUILIBRIUM MODEL In this section we report on an attempt to estimate the model descri bed earlier as belonging to the GEA. The results presented will, subject to the difficulties discussed below, shed some light on the behaviour of time series of interest rates examined in the previous section. It was seen there that, while both short and long rates increased substantially between 1958Q2 - 1959Q3, following the last date, long rates increased and 17 short rates decreased. This section investigates why that might be so and in the process appraises Barber's and Shearer's contentions concerning the behaviour of interest rates. It also implicitly deals with question (iii) of the introduction to this chapter. That issue is dealt with more explicitly in section four. It is now appropriate to discuss some of the difficulties encountered while estimating the GEA model. The first question is the extent to which the constraints ought to be applied. Does private wealth equal M + S + L alone or are there any other assets involved? One may decide that char tered banks should be treated as being, along with the government, outside the system, exogenously supplying just demand deposits to the private sec tor which holds them, i.e. M, along with the liabilities of the government 18 namely S and L . Even so, there still remains the point that data limita tions force us to treat financial institutions other than banks as "inside" the system. They issue near moneys which are very similar to the liabili ties of banks. To the extent that the authorities are able to affect the size of these near moneys they might best be treated as exogenous along with the other policy instruments M, S and L. Also there are other assets such as real capital which should be part of the portfolio of the "inside" sector and which are not taken into account. The exclusion of all other assets may be justified on the following grounds. Consider Tobin's model (67) where the portfolio of the "inside" sector consists of M, B, and K, where B stands for all government bonds and K for real capital. Let there be a partition of this portfolio into (M+B) and K. Nov? suppose that the term structure of rates within the (M+B) class does not affect the choice between (M+B) and K, although it does affect the choice; within (M+B). The choice between (M+B) and K may be affected by the level of the average rate in the (M+B) class. It could then be argued that model 2C(v) des cribes the choice within the (M+B) part of the portfolio. Clearly, this imposes stringent conditions on the nature of the asset demand functions, but they are necessary given the terms of reference of this thesis. There is another problem that must be dealt with before we can pro ceed. We are concerned with rates of return and, therefore, in empirical estimation we should be concerned with reduced form rather than structural equations. Yet the constraints on the partial derivatives of the demand functions, discussed in section 2C(v), cannot be computationally implemen-ted unless complex programming is resorted to. An alternative is to esti-19 mate the constrained demand functions and then solve two independent equations for rc and r . This was one of the methods used. Because of all the above difficulties it was felt necessary to also estimate the reduced forms implied by the GEA model without any constraints. Thus, a measure of the importance of all the problems mentioned above can be had. The results obtained are reported in Tables 2 and 3. Table 2(a) gives the results from constrained estimation of structural equations while Table 2(b) reports the reduced form equations implied by those in Table 2(a). Table 3 gives the reduced form equations that were estimated directly. The following comments can be made about these results: i) All variables in Tables 2(a) and 3 have the expected signs and are on.the whole significant. ii) The goodness of fit in all equations is fairly good, as the above point implies. iii) Since no quantitative restrictions were placed on the partial derivatives in the model of section 2C(v), it follows that the implied -reduced forms will also have the expected signs. This is borne out as a comparison of Tables 1, 2(b) and 3 shows. iv) The coefficients in the estimated reduced forms of Table 3 are 20 considerably smaller than those in the implied reduced form equations. v) It is possible to have both income and wealth variables in the demand functions for assets with signs that make sense. Whether income, wealth, or both variables should be included in the demand for money func-93 tion is a question that has received considerable attention in the litera ture. MeltzerC42), Brunner and Meltzer (9), Hamburger (24), Laidler (. 37) and others explicitly examined this problem for the U.S. and they all con cluded that a) .there was no scope for both variables in the demand for money function in the sense that when both were used one was insignificant. b) .thewealth variable somehow measured—permanent income is con sidered as a proxy for wealth in the literature—is the most important of the two variables. Unless the concept of transactions demand for cash balances is empirically unimportant in the U.S., both income and wealth should be significant in 21 an appropriate demand for money function. It is, therefore, gratifying to find that, at least in Canada, both variables are important. vi) Turning to the quantitative aspect of the results three points are noteworthy. a) The short rate appears to be more important than the long one^ in the demand for money function—equation 19. The performance of supply varia bles in equations 24 and 25 is on the whole poor. b) The coefficient on the income variable in the demand function for shorts is, in absolute terms, smaller than its counterpart in the demand function for longs. It would be- more plausible to see shorts rather than longs used as a buffer against changes in transactions requirements. No evidence exists on the matter. See, however, the result (X) obtained below. c) It is not always true that the own coefficients in equations 20 and 21 are larger than the cross ones in absolute terms. There has been 23 some U.S. evidence to the contrary. In the implied reduced form 94 equations 22 and 23, and in equation 25, the partial ifrdH5* iffjl where i, j=S, Landi^j vii) There is evidence of positive serial correlation in all equations. Concerning point (vii) above, in studies of asset demand functions a 24 partial adjustment mechanism is often incorporated. This is theoretically appealing and it has the property of reducing serial correlation in the cal culated residuals. However, incorporating such a mechanism in the model of section 2C(v) is not straightforward: Suppose that there is an increase in W which leads to new asset demand levels. In the present model such increa ses in wealth are exhausted by the requirement that £ ~- = 1 for i = M, S, L. i But, if people adjust slowly what happens to that part of the wealth which is not immediately "desired"? A consistent re-specification of the model 25 emerges if we assume that the change in the holdings of an asset depends on the difference between desired and actual stocks, with a common speed of adjustment for all assets, and a term that decides the temporary allocation of remaining changes in wealth i.e. ii2] cx^-x^) = d(x*jt - x.^) + ywt-Vl) where, as before [13] X~. = a*j + EB*r. + Y*Y + S*W + U. with U ~ N(0,o\) jt J ± ±3 it 'j t j t jt Jt 3 so that [14] Xjfc = a. + (1-d) X.^ + ZB. .r.t 4- y.Yt + (6. + A.) Wfc - A.W^ + Vjt 95 with a 6*d, V.„ = U. d, V. ~N(0,d2a2), J Jt jt jt J and the constraints [15] la. = 0 [16] 0 [17] K&. + I.) = j J 3 1 and [18] EX. = (1 - d) . 3 The estimated equations appear in Table 4(a). Table 5 gives the directly estimated reduced forms implied by this model, while Table 4(b) reports the reduced form equations implied by those of Table 4(a). The comments on the viii) All variables have the expected signs except for the long rate in the demand for money equation. Its sign is positive but the coefficient itself is not significantly different from zero. ix) The t statistics for most variables are absolutely higher. The goodness of fit of all equations is improved owing to the inclusion of more variables. x) Short term bonds are now a better buffer against changes in transac tions requirements than long term bonds are, as one would expect. xi) The Durbin-Watson statistics in Table 4(a) indicate no positive serial correlation. This is not true of those in Table 5. However, in order to retain the comparability of equations 29, 30 and 31, 32, the matter was not pursued further, particularly since dw is not a very reliable indicator of serial corr elation when lagged endogenous variables are included in an equation. xii) Slow adjustments to desired asset positions are observed — d = 0.39. xiii) Supply variables in equations 31 and 32 are now more important. results made above hold here too with the following emendations. 96 The results reported in Tables 2 to 5 will now be used to account for the behaviour of interest rates during the post-Conversion Loan period and to assess the relative validity of the Barber and Shearer arguments. To that end equations 22 to 25 and 29 to 32 are decomposed by variable, thus making it possible to evaluate the importance of each argument individually. Changes in the debt held by the Public will, somewhat loosely, be referred to as the Conversion Loan. Why did rates in general increase following 1958Q2? One answer can be had with the aid of Table 8. Consider the changes occurring between 1958Q2 and 1958Q3 in the long rate: The biggest single source of its increase to a pre dicted 5.06%, was the Conversion Loan -- it contributed 1.94% to its rise. Column 3 of Table 8 shows that the bond price support programme of the Bank considerably relieved upward pressures on the long rate: Had the money supply remained at its 1958Q2 levels the long rate would have been higher by 0.66%. A minor source of upward pressure on the long rate was the behaviour of the maturity composition of the debt in the last quarter; as columns 6, 7 and 8 show, this force raised r^ by 0.27%,. The short rate fell by 1.23% to a predicted 2.62%.26 Equation 29 ascribes a decrease in predicted rg of 52 basis points to the Conversion Loan and a further decrease of 91 basis points to the expansionary monetary policy concom itant with the price-support programme of the Bank of Canada. The behaviour of the debt in the previous quarter mitigates the tendency of the short rate to fall -- columns 7 and 8 show that rg would have risen by 0.34%. Beyond 1958Q4, rates increased for several reasons: i) The quantity of shorts outstanding was increased very fast indeed --97 longs did not increase until 1960. The consequences of these changes can be seen in columns 4 and 5, Table 8 -- and of course Tables 6, 7 and 9. Had S remained at its 1958Q3 level, the short rate would have been lower by 1.44%, and the long rate would have been lower by 0.80% in 1959Q1. ii) The nominal money supply expanded during the bond price support period, but it was subsequently held in the neighbourhood of its 1958Q2 level, until 1960Q3. The seasonally adjusted real money supply peaked in 1959Q1, declined for the next three quarters, and started increasing as early as 1960Q1 -- column 3, Tables 6, 7, 8, 9. Thus, after the temporary relief provided by the price support programme there is an upward pressure on and r^ until 1960, when the money supply expanded. The effects of the Loan without the price support programme were, therefore, not fully felt until 1959. iii) Increased transactions requirements due to increases in real income following 1959 also placed an upward pressure on interest rates. This force on estimated rs and r^ amounted to 0.94%, and 0.67%, respectively during 1958Q4 to 1959Q1 -- column 3, Table 8. Thus, the generally held view that the Conversion Loan alone accounted for the rise in interest rates requires qualification. The Loan per se only led to increases in the long rate -- it decreased the short rate. Over the longer haul, changes in the size of the debt, and increases in transactions requirements placed overwhelming upward pressures on both rates. Contractionary monetary policy over and above undoing the effects of the price support programme never really happened: At no time has the nominal seasonally unadjusted money supply ---- Table 6, chapter one -- fallen below its 1958Q2 level. Though Table 8 has been used in this discussion, conclusions reached hold regardless of the table -- 6 to 9 -- used. 98 SECTION FOUR: ESTIMATING THE PREFERRED HABITAT MODEL The survey of studies pertaining to "Operation Twist", i.e. the appen dix to chapter two, indicated that little, if any, importance was to be attached to debt management operations. The preceding section gave ample evidence to the contrary. Because of this apparent paradox it is necessary to delve more deeply into this issue. In this section the analytical proce dures used by Modigliani and Sutch (EMS) (44) , adapted to fit the peculiari ties of our problem, are used. The first point made by MS is that in assessing the effects of debt management operations on the term structure one ought to look at the spread between the yields of bonds in the relevant categories. In the present case the relevant spread is between long rates—over ten years—and short 27 rates—under ten years—i.e. Spread = r -r . Except for the period till J-l o 1959Q3, this spread rises as one would expect. The next step is to note the importance of the business cycle. In recessions, such as during 1958, the Spread increases because, while both rates fall, the short rate by virtue of its larger variance, declines more than the long one. For this reason it is necessary to purge the Spread from the influence of the cycle. To that end, the Spread is regressed against the short rate—this is equation 33 in Table 10. It is used to predict the Spread for the period 1958Q3- 1965Q4. To the extent that the business cycle is captured by the short rate, we may expect the predicted Spread to lie consistently below the observed one, since the Loan should increase the long and decrease the short rate. This is in fact the case. The residuals are in all cases greater than twice the standard error of estimate—a rough but indicative test. It is noteworthy that when MS carry out this test for "Operation Twist" they find no evidence that the spread, in that case, decreased below what it would have done during the course of the cycle. One problem with these results is that the hypothesis of zero serial correlation in the residuals must be rejected. While coefficient estima tors are unbiased, the usual formulae for the calculation of the variance-covariance matrix of the estimated coefficients are no longer appropriate. Moreover the estimator of the residual variance may be biased—a particu larly serious problem here, since we are interested in prediction. To circumvent this problem iterative procedures are often resorted to—their ^rationale being that they maximise a likelihood function. However, these procedures can be abused if used in cases where there is a blatant omis sion of relevant independent variables. Thus, a second way out of the autocorrelation box is a better theory. Modifying the MS specification somewhat it is hypothesized that the long rate r is a linear function of the short rate r , and the expected future short rate rG [34] r = a + b'r + cr6 b > 0 , c > 0 Equation [2] is the analytical justification of the above specification. Moreover, rg captures the influence of the business cycle. Using equa tions [5] and [6] of section two, ve may argue that the market centa:\ns both regressive and extrapolative elements so that [35] reg = rg + k(N-rg) + d(rg-E) where, in accord with the discussion in. section two [36] N « I KT 1=1 ™ E = ^ Eirst-i 1=1 Then n m rec = r + k( E N r - r ) + d(rc, - Z E r ) S S . - l St-i S S . . x St-i 1=1 1=1 [38] = qrc + E Z',r^_. i St-i where q = 1 + d - k i=l Z'. = kN.-dE. and l ii p = max (ro, n) substituting [38] into [34] P rL= a + b'r + c(qr + Z Z« r ) i=l P [39] = a + br + E Z.r where b = b' + cq and i=l Z, = cZ! 1 I Substracting r^ from both sides of [39] 101 p [40] Spread = r -r = a + E W.r where W. = b-1 for i = 0 x=0 W. = Z. for i 4 0 1 x A stochastic version of [40] was estimated using the Almon (1) technique of estimating distributed lags which constrains the W_^ to lie on a polyno mial of specified degree. A third degree polynomial was used and the first Almon variable was dropped thereby constraining the polynomial to have a Wp = 0, a zero slope at that point and a maximum of one peak. This last restriction was placed because theory justifies only one peak satisfactor ily and because earlier work indicated that when a fourth degree polynomial is allowed for the coefficients W_^ beyond the point where the function crosses the horizontal axis for the first time are not significantly dif ferent from zero. The estimated equation is reported as number 41 in Table 10, and the estimated polynomial appears as Graph 1. Here as in equation 33, all variables have the anticipated signs and are significantly differ ent from zero. Moreover, in equation 41 the Durbin-Watson statistic indi-cates no serial correlation. Looking at the Y-Y values of Spread reveals that the post-Loan observations are greater than twice the standard error of estimate. Given the fit of the equation this test is telling. Note that it was partly on the basis of this test that MS had concluded that "Operation Twist" did not affect the term structure of interest rates. To implicate the Conversion Loan more explicitly it is necessary to 28 introduce supply variables into equation [40]. The difficulties associ ated with doing so are well known and have been extensively discussed 29 elsewhere. Two supply variables were considered: DI and D2. The for mer is the ratio of shorts to longs held by the public, financial interme-102 diaries including the chartered banks and the Government Accounts. The latter is the ratio of federal government shorts to longs held by the public and financial intermediaries excluding the chartered banks. The corresponding ratio of assets held by the public alone—a preferable 'variable—cannot be obtained without making assumptions about the maturity composition of the federal government bonds held by chartered banks and 30 the Government Accounts. Also included in estimated equations was the liquid asset ratio of 31 chartered banks i.e. the ratio of Canadian Liquid Assets to Canadian Dollar Deposit Liabilities times 100. This variable is a proxy for two important influences on interest rates, namely monetary policy and the port folio adjustments of chartered banks. When monetary policy becomes tight, for example, banks are forced to liquidate short-term assets. This dec reases prices and raises short rates, including r thereby decreasing the 32 Spread. There is also evidence that chartered banks will adjust their portfolios away from short-term assets and into loans when it is safe and profitable for them to do so. Such responses, whilst motivated by interest 3' orate differentials, in fact exaggerate them thereby affecting the Spread. ~ 34 The U.S. equivalent of Spread, i.e. U.S.S., was included in the stochastic version of [40] in order to capture some of the open economy aspects of Canada. Doing so within the Tobin framework requires drastic respecification of the model and so it was not pursued there. One may conceive of this continent as one large economy in which all assets of the same class to maturity are perfect substitutes. Then it may be argued that it was a change in U.S.S. that caused the increase in the Spread following 103 1959Q3. Equations 42 and 43 in Table 10 and Graphs 2 and 3 report the •results obtained. The restrictions on the polynomial for the weights W in equation 41 apply here too. The following points are noteworthy. i) There appear to be significant expectational forces at work. Regressive and extrapolative elements cannot be distinguished but both may be present. This comment is made because given the estimated weight pat terns, a purely regressive, or a purely extrapolative mechanism i.e. E^=0, all i or N^=0 all i, appears implausible. Taylor, in a recent paper (64), attempts to make some inferences on the underlying expectational mechanisms but he should fail to convince the reader. No such attempt is made here, ii) Supply variables DI and D2 have the correct signs and are signifi-cant. Thus, the Canadian authorities can alter the term structure of 35 interest rates. There are, however, quantitative considerations here. To increase the Spread by about 20 basis points the authorities must dec rease the D2 ratio by one unit—this ratio was decreased from 3.21 to 0.47 during the Conversion Loan thereby increasing the Spread by a predicted 58 basis points. iii) The behaviour of certain financial institutions affects security prices importantly as indicated by the correctness in sign and signifi cance of the liquid asset ratio. iv) The variable U.S.S. has the correct sign but is only significantly different from zero at the 5% level. This is very curious. Under flex-ble exchange rates the exchange risk may effectively segment the two eco nomies. Thus, the same equations were reestimated for the flexible and for the fixed exchange rate periods ecpecting to find the U.S.S. at least 104 more important during the latter period. This hypothesis did not square with the facts. v) The Durbin-Watson test gave inconclusive results, as is often the case. vi) Equations 42 and 43 remained essentially unchanged when dif ferent variable series and estimation periods were used. Tables 11 and 12 correspond to Tables 6 to 9 of the previous section. They are of value in that they isolate the importance of each variable in affecting the Spread. Examining the absolute size of the coefficients in equations 42 and 43 clearly does not do that. The tables show that the immediate rise in predicted Spread in 1958Q3 was almost entirely due to the change in the composition of the debt. Expectational forces, i.e. anticipated rises in the short rate, which decrease the Spread, did not set in until the Bank stopped supporting bond prices. The predicted Spread 36 was expected to fall by 10 basis points during 1958Q4 when the Bank was changing its policy and by 17 basis points in 1959Q1. Further decreases in the Spread occurred because of the expected relative rises in short rates during 1959Q1 - 1959Q3. Column 5 in the two tables indicates the impor tance of the short rate. The fact that the Spread became negative during early 1959 appears to be entirely due to the influences contained in col umns 5 and 6. The above conclusions do not depend on whether Table 11 or 12 was chosen. 105 SECTION FIVE: WHY ARE SUPPLY VARIABLES IMPORTANT IN CANADA BUT NOT IN  THE U.S? In the previous two sections it was shown beyond doubt that supply -variables do affect the term structure of interest rates in Canada, at least during the estimation period. This finding is in accord with frag-37 tnentary evidence In the Canadian economic literature. However, it has been assumed by economists that debt management is not an important policy tool. This belief is based on the series of empirical studies surveyed in 38 the appendix to chapter two. Thus, the apparent paradox emerges that in Canada supply variables matter while in the U.S. they do not. This sec tion offers an explanation. One possible explanation of this paradox has been suggested, namely that the Federal Reserve and the Treasury often pursue conflicting poli cies. This argument was examined for the Operation Twist period, one of the most publicised debt management operations, in the appendix to chapter two. The result of such policies would be, the argument continues, to re duce the variability of quantities in the various maturity categories of the U.S. debt below what it would have been had either the Federal Reserve or the Treasury not existed. While the variance that would have prevailed under those hypothetical circumstances is not known, the implication is that the maturity composition of the debt does not vary a lot. But what is the meaning of "a lot" in this context? One answer is relative to the -variability of the maturity composition of the Canadian public debt. We know that in Canada we have had large debt management operations, e.g. the Conversion Loan, while in the U.S. their existence is disputed. Are supply variables more important in Canada because the composition of the country's 106 debt varies more than that of the U.S.? The answer can be found by calculating a measure of dispersion for the -various maturity classes in the two debts. There is, of course, the question: What is the appropriate measure? If we believe, that an increase in the amount of long debt of $X should have the same effect on the term structure in both countries, then the standard deviation of the relevant series would be a good measure. However, the U.S. economy is far larger than the Canadian one and so is its public debt. A $X increase in longs in both countries could have substantial effects on the small economy's term structure and no appreciable effect on that of the larger economy. Hence, the coefficient of variation, which takes this size factor into account, might be a more appropriate measure. Both are reported. Two alternative debt series are considered for each country: The Inside and Public series for Canada and the Sutch and FRB series for the U.S. The data from R. Sutch's Ph.D. thesis were supplied to him by the Federal Reserve. The various maturity classes have been blurred some what by not assigning securities their full weight while in a particular class. Thus a bond with four years to maturity is partly included in the category Short, partly in Medium (I) and partly in Medium (II). When time comes, for it to cross the boundary into the shorter class the transition is smoothed by i) No longer including it partly in Medium (II), ii) By still preserving part of it in Medium (I) and iii) By assigning it a greater weight in Short, where it now properly belongs. Naturally the weights applied to a bond in different categories 107 must at each point in time sum up to one. These transformations clearly ought to reduce the variance of bonds outstanding in each class. Tables 13(a) to 13(d) corroborate this conjecture. For a more complete explana tion of the construction of the Sutch series see (63), p. 336. The series Inside was taken from Table 9, chapter one. Similarly the series Public appeared in Table 12 of the same chapter. The Federal Reserve Bulletin series was taken from that publication. The various maturity classifications for the four series were made as comparable as they could possibly be. The observations from which measures of dispersion were calculated spanned the period 1955Q1 - 1965Q4. The ratio shorts/longs is, in fact, Dl and D2 for the two Canadian series. For the U.S. series Sutch and FRB they are [Short + Medium (I) + Medium (II)] / Long and [0 to 1 + 1 to 5 + 5 to 10 ] / Over 10 respectively. The results obtained appear in Tables 13 and 14. The following observations can be made. i) The FRB series usually used by U.S. researchers has a higher standard deviation than either of the Canadian series for all four maturi ty classes and for the ratio of shorts to longs. ii) The Sutch series, because of the method used in its construction, has a lower standard deviation than the FRB series. But even so, only in two cases, Table 13(c) and 13(d) is the standard deviation of a maturity class smaller than the corresponding number for the Canadian series. In Table 13(e) which contains the kind of variable used in the empirical studies of debt management, the standard deviation of the Sutch series is T.30, equal to that of the Inside series and greater than that of the second Canadian series. 108 iii) While in absolute terms debt management operations in Canada and the U.S. were roughly equally strong, in percentage terms, as measured by the coefficient of variation, such operations were of far greater impor tance in Canada. This then may be one reason why in Canada supply vari ables "matter". Another explanation may be that the assumption of perfect substituta-bility underlying the expectations hypothesis is more valid in the U.S. than in Canada. Hence, the predictions of the expectations hypothesis, one of which is that supply variables do not affect the term structure, are -more likely to hold in the U.S. than in Canada. To test this hypothesis the following procedure was used. First cal culate the implied short rate expected last period to prevail in this cur-rent one, i.e. r ^ using the formula (1 + rT)2 t-1 (1 + rs) i.e. assuming that the expectations hypothesis holds. The Canadian expec-e e ted rate is denoted-by r while the U.S. one by R .. Then run the t-1 J t-1 • e e regression of r on x and R on R . These regressions will tell ot t X ot t L how well the market is capable of anticipating the future short rate. In the extreme case where the market is capable of predicting perfectly well the estimated slope should be equal to unity and the constant should be equal to zero. It is important to note at this point that this procedure is fairly 109 controversial. Meiselman, for example, has argued that anticipations may determine the term structure ex ante and yet not be realised ex post. 39 Conard, however, took the opposite view that "...It is unreasonable to presume the market is so consistently and grossly wrong in its expectations that poor foresight could wholly explain these obser vations ." The position taken here is that, in view of the difference in the mechan-40 e isms determining r and r , it would be unreasonable to expect anti-cipations to completely materialise. This means that we cannot use the above regressions as tests of the expectations—and in this case the null 41 —hypothesis. But we can use it as a means of establishing the relative degree of substitutability between short and long federal government bonds in Canada and the U.S. if the following assumption is made: That the fac tors causing anticipations to depart from subsequent experience are of the same quantitative importance in Canada as in the U.S. With the above caveats we turn to the results obtained. They are reported as equations 44 and 46 in Table 15. They show that i) 'jhe estimated U.S. line--equation 46-- conforms much more closely to the 45° line than the Canadian one—equation 44 --does. ii) The explanatory power of equation 46 is considerably higher than that of equation 44 . iii) Positive serial correlation is present in both equations but since we- are merely interested in the estimated coefficients, which are unbiased, and R^ this problem is not very serious here. 110 The previous results show that in Canada the degree of substitution among shorts and longs is considerably smaller than in the U.S. That is another reason why supply variables are important in Canada but not in the U.S. To justify these differences in substitutability among'assets between the two countries an examination of the effects of unit ^competi tive) versus branch (=oligopolistic) banking and of other institutional differences is called for. That task is beyond the scope of this thesis. Ill NOTES TO CHAPTER THREE 1. For a discussion of these procedures see Christ (14), pp. 549 - 564. 2. The regression equations and plots of actual and predicted interest rates are not reported here in order to economize on space. The section on Data and Sources at the end of this chapter describes each rate used in detail. 3. That is, Y-Y was greater than -- roughly twice the standard error of estimate. 4. See Meiselman (41), p. 4. 5. See Conard (15), pp. 302 - 303. 6. See Meiselman (41) , p. 9. 7. See Conard (15), p. 300. 8. See Meiselman (41), p. 12. 9. For simplicity let Rlfc = N = E. 10. See Conard (15), pp. 307 - 308 for a proof. 11. The discussion concerning diagram 3 is relevant here. 12. Hicks (29), pp. 146 -147. 13. Hicks (29), pp. 146 - 147. 14. See Meiselman (41), pp. 14 - 16. 15. Modigliani and Sutch (44), pp. 185 - 187. 16. See Tobin (67) , p. 18. 17. All references to "increases" and "decreases" in interest rates are in fact to increases above trend and decreases below trend. This less cumbersome terminology is used in what follows. 18. See the section on Data and Sources. 19. This can be effected by using a routine in Massager (40), a programme written by M.C. McCracken. This routine also utilizes the efficient features of Zellner's method of estimating seemingly unrelated regressions when disturbances across structural equations are contem poraneously correlated. 112 20. There are many reasons why the estimated reduced forms and the implied ones should differ, as the following discussion shows. There is, to begin with, a simultaneous equations problem because in the equation for each asset some of the independent variables, namely rg and rL, are not independent of the error term in the equation. This can be seen by finding the covariance of rg and Ug, or r-^ and for that matter, where Ug and U*L are the error terms in the structural equations for S and L respectively. This amounts to multiplying the reduced form for rg -- or r-^ -- by Ug -- or U^ -- and taking the expected value of the resulting expression -- the usual assumption that E(Ug) = 0 and E(UL,) = 0 is made. The resulting covariance will not be equal to zero leading to bias and inconsistency. Given that, the OLS estimates in equations 19, 20 and 21 are inconsis tent. Other characteristics of those estimates are that the constraints dictated by the GEA were implemented and that the Zellner method of estimating seemingly unrelated regressions was used. By contrast, in the exactly identified system of equations 19, 20 and 21, the reduced form equations 24 and 25 are consistent but do not have the second and third characteristics of the estimated structural equations. There are also the discrepancies that might arise in small samples. Turning to the system of equations 26, 27 and 28, we note that any two independent equations are overidentified. In this case the implied reduced forms have the advantage of satisfying a priori overidentifying restrictions, while equations 31 and 32 do not. Of course, had equations 27 and 28 -- say -- been estimated with 2SLS and had they then been used to solve for the implied reduced forms this argument would not apply. This latter procedure is more appropriate -- see Christ (14), pp. 464 - 481. The comments in the previous paragraph with respect to differences arising due to the use or not of the constraints and the Zellner procedure apply here too. Here again, discrepancies might arise due to small samples. 21. Feige (20), has argued that permanent income may be viewed as an optimal predictor of measured income -- a_ l_a Muth (46) -- rather than a proxy for wealth. While his argument is appealing it leaves his money demand function without a wealth variable. This leads to conceptual problems. 22. This is also the case in the U.S. -- See Laidler's discussion (37), p. 108. 23. See Hamburger (24), pp. 105 - 106. 24. See, for instance, Hamburger (24) and Feige (20). 25. We follow the way operation 57 in Massager (40) was constructed. 26. Although equation 29 predicts this turning point correctly, it exaggerates the fall in this rate somewhat -- by 44 basis points. 27. See the discussion in the previous section. 113 28. See Modigliani and Sutch (44). 29. Malkiel has a good exposition of these difficulties. Sjee (39), pp. 221 - 226. 30. Chapter one contains a discussion of these problems. The model builders of RDX2 (27) have made one set of such assumptions. Use of their data indicated that the results obtained are very similar indeed. DI and D2 are used here to preserve continuity. 31. See Data and Sources at the end of this chapter. 32. See, for example, the Annual Report of the Governor, Bank of Canada, particularly the 1958 and 1959 issues, pp. 36 - 38 and pp. 44 - 45 respectively. 33. To the extent that federal government bond rates are correlated with loan rates, a simultaneous equations problem may be present here. 34. See Data and Sources. 35. In the same paper, Taylor (64) also reports that he failed to identify any significant supply variables in the U.S. during the pre-Accord period. This, of course, is in line with other U.S. evidence. However, no investigator should expect to find such evidence during a period when bond prices were pegged! 36. See column 6, Table 11. 37. See, for example, chapter two, p.63 and the financial sector of RDX2 in (27), particularly equation 17.2, in Part 2, p. 107. 38. Paradoxically there has been a revival in theoretical interest on the matter which is most manifest in Tobin's writings. See particularly (67). 39. Conard (15), p. 339. 40. See section 2C(i). 41. Hickman (28) has used this procedure as a test of the expectations hypothesis per se. He essentially compared the results from regressions discussed above with those got from the inertia hypothesis that rSt ~ a + brst-l + Ut* For curiosity's sake, equations 45 and 47 in Table 15 were also estimated. Their explanatory power, in both Canada and the U.S., is higher than that of equations 44 and 46. For reasons explained earlier we do not draw the conclusion that Hickman might have drawn, namely that the expectations hypothesis is not in accord with experience. 114 DATA AND SOURCES SECTION ONE The following interest rates were used for the period contained by the quarters 1951Q1 and 1967Q4: Treasury Bill Rate. Bank of Canada Statistical Summary Supplement (= Supplement) . Government bond yield 0-2 years to maturity. This is the simple average of individual Direct and Guaranteed bond yields. They are reported in the Supplement. Government bond yield 2-5 years to maturity. Constructed as the previous rate from data in the Supplement. Government bond yields on bonds with 1 - 3, 3 - 5, 5 - 10 and over 10 years to maturity. All four rates were taken from the data tape for the RDX2 model of the Canadian economy (27) which is available at the University of British Columbia Computing Centre. These too are average rates of individual bond yields. This source will be referred to as RDX2 data tape. Several simple averages were constructed out of the previous rates, for example: SA4 is the simple average of rates on bonds with 0 - 2, 2 - 5, 5 -10 and over 10 years to maturity and rg, often referred to as the short rate, is the simple average of rates with 0 - 3, 3 - 5 and 5-10 years to maturity. r-^, often referred to as the long rate, is in fact, the rate on bonds with over 10 years to maturity. SECTION THREE In addition to r<, and r^ the following variables were used: P, the Consumer Price Index, was used to deflate Y, M, S and L below. Its source is the RDX2 data tape. Y =GNE/P. The source of GNE is the RDX2 data tape. 115 M, currency in circulation plus OTHER deposits with chartered banks, all divided by P. The source of the numerator is the Supplement. S, quantity of short government bonds -- under 10 years -- held by the Public, deflated by P. The source of the numerator is the Supplement. See also the discussion in chapter one. L, quantity of long government bonds -- over 10 years -- held by the Public, deflated by P. The source of the numerator is the Supplement. W = M + S + L. SECTION FOUR The following variables were used: Spread = rT - r DI, the ratio of short bonds held by all but the Bank of Canada to long bonds held by all but the Bank of Canada. Source, Table 9, chapter one. D2, the ratio of short bonds held by the Public -- i.e. all but the Bank of Canada, the Government Accounts and the chartered banks --to long bonds held by same. Source, Table 12, chapter one. LAR, Liquid Asset Ratio for Canadian chartered banks. It is the ratio of Canadian liquid assets -- defined as cash reserves plus Day to Day Loans plus Treasury Bills -- to Canadian dollar liabilities times 100. Its source is the Supplement. Between 1955Q1 and 1956Q2 this ratio increased by about 4% following an informal agreement reached by the banks and the authorities. While this, change followed a change in the.constraints under which the banks operate, it nevertheless should exert the same influence on the Spread as when the ratio is changed following parametric changes under a given set of constraints. USS, long U.S. Government bond yield minus a medium term bond yield. Source is the IMF Financial Statistics. All the above variables are seasonally unadjusted with the exception of nominal M and GNE where strong seasonal factors exist. TABLE 2(a): Regression Results for the Model of Section 2C(v) Estimation Period Dependent Variable • Constant rs rL Y W W-l Lagged Dependent Variable R2 dw Eqn. // 1955Q1 to 1965Q4 Real Money Stock M -4.468 (-0.98) - 2.405 (-3.35) -0.897 (-1.03) 0.563 (10.29) 0.225 (3.42) 0.96 0.65 19 1955Q1 to 1965Q4 Real Short Bonds S 51.068 ( 3.94) 11.477 (5.63) -16.403 (- 6.6.2) - 0.013 (-0.09) 0.096 (0.51) 0.59 0.64 20 1955Q1 to 1965Q4 Real Long Bonds L -46.600 (-4.51) - 9.072 (-5.59) 17.300 (8.76) - 0.549 •(- 4.45) 0.680 (4.58) 0.81 0.75 21 TABLE 2(b) : Implied Reduced Form Equations Endogenous Variable Constant Y M S L M-l S-l- L-l Eqn. ii Nominal Yield on Shorts (Under 10 Yrs) = rg -2.394 0.186 -0.257 0.091 0.073 22 Nominal Yield on Longs (Over 10 Yrs) = r 1.471 . 0.129 -0.174 0.008 0.057 23 TABLE 3: Estimated Reduced Form Equations for the Model of Section 2C(v) Estimation Period Dependent Variable Constant Y M s L R2 See dw Eqn. # 1955Q1 to 1965Q4 rs 1.983 (1.57) 0.129 (8.05) -0.164 (-6.46) 0.001 (0.05) 0.015 (0.88) 0.68 0.44 0.92 24 1955Q1 to 1965Q4 rL 2.024 (2.72) 0.102 (10.80) -0.124 (-8.29) -0.008 (-0.69) 0.036 (3.63) 0.88 0.26 1.02 25 TABLE 4(a): Regression Results for the Partial Adjustment Version of the Model of Section 2C(v) Estimation Period Dependent Variable Constant rs rL ,Y W W-l Lagged Dependent Variable R2 dw Eqn. // 1955Q2 to 1965Q4 Real Money Stock M 3.185 (1.19) -2.270 (-5.64) 0.297 (0.60) 0.322 (8.28) 0.338 (6.12) -0.343 (-6.22) 0.612 (11.03) 0.99 1.65 26 1955Q2 to 1965Q4 Real Short Bonds S -4.079 (-0.39) 9.161 (6.17) -10.613 (-5.79) -0.235 (-2.04) 0.477 (2.31) -0.093 (-0.52) 0.612 (11.03) 0.80 2.09 27 1955Q2 to 1965Q4 Real Long Bonds L 0.894 (0.10) -6.891 (-5.30) 10.316 ( 6.22) -0.088 (-0.82) 0.185 (1.00) -0.177 (-1.15) 0.612 (11.03) 0.89 2.09 28 TABLE 4(b): Implied Reduced Form Equations Endogenous Variable Constant Y M S L M-l S-l L-l Eqn. // Nominal Yield on Shorts (Under 10 Yrs) E rg 1.526 0.158 -0.322 0.161 0.175 0.133 -0.163 -0.171 29 Nominal Yield on Longs (Over 10 Yrs) = rL 0.930 0.114 -0.233 0.089 0.196 0.106 -0.091 -0.157 30 TABLE 5: Estimated Reduced Form Equations for Model of Section 2C(v) Estimation Period Dependent Variable Constant Y M S L M-l S-l L-l R2 See dw Eqn. # 1955Q2 to 1965Q4 rs 2.144 (1.55) 0.124 (7.73) -0.201 (-5.77) 0.082 (2.34) 0.099 (2.30) 0.056 (1.91) -0.092 (-3.24) -0.099 (-2.52) 0.71 0.39 1.00 31 1955Q2 to 1965Q4 rL 2.118 (2.23) 0.102 (9.28) -0.139 (-5.82) 0.006 (0.25) 0.050 (1.71) 0.018 (0.90) -0.018 (-0.93) -0.018 (-0.68) 0.86 0.27 0.95 32 VO 120 TABLE 6: Decomposition of Implied Reduced Form Equations 22 and 23 5 Col. 6 = E col. i i=i SHORT RATE Constant 1 0.186 Y 2 -0.257 M 3 0.091S 4 0.073 L 5 Predicted r 6 b 1958Q1 -2.39 15.80 -14.76 4.14 1.12 3.91 Q2 -2.39 15.97 -14.77 4.08 1.02 3.91 Q3 -2.39 15.82 -15.50 1.55 2.66 2.15 Q4 -2.39 15.59 -15.23 1.95 2.62 2.54 195901 -2.39 16.70 -15.85 2.37 2.65 3.48 Q2 -2.39 16.78 -15.26 2.65 2.70 4.48 Q3 -2.39 16.59 -14.93 2.83 2.66 4.76 Q4 -2.39 16.30 -14.40 2.86 2.63 5.00 1960Q1 -2.39 17.70 -15.26 3.11 2.72 5.88 Q2 -2.39 16.95 -15.10 3.04 2.79 5.29 Q3 -2.39 16.97 -15.15 2.86 2.79 5.08 Q4 -2.39 16.56 -14.93 2.64 2.88 4.74 LONG RATE Constant 1 0.129 Y 2 -0.174 M 3 0.008 S 4 0.057 L 5 Predicted r 6 L 1958Q1 1.47 10.96 -9.99 0.36 0.87 3.68 Q2 1.47 11.07 -10.00 0.36 0.80 3.70 Q3 1.47 10.97 -10.49 0.14 2.08 4.17 Q4 1.47 10.81 -10.31 0.17 2.05 4.19 1958Q1 1.47 11.58 -10.73 0.21 2.07 4.60 Q2 1.47 11.64 -10.33 0.23 2.11 5.12 Q3 1.47 11.50 -10.11 0.25 2.08 5.20 Q4 1.47 11.30 -9.75 0.25 2.05 5.33 1960 Ql 1.47 12.28 -10.33 0.27 2.12 5.82 Q2 1.47 11.76 -10.22 0.27 2.18 5.45 Q3 1.47 11.77 -10.26 0.25 2.18 5.41 Q4 1.47 11.48 -10.11 0.23 2.25 5.33 121 TABLE 7: Decomposition of Estimated Reduced Form Equations 24 and 25 5 Col. 6 = £ col. i i=i SHORT Constant 0.129 Y -0.164 M 0.001 S 0.015 L Predicted r„ RATE 1 2 3 4 5 6 S 1958Q1 1.98 10.93 - 9.40 0.05 0.23 3.79 Q2 1.98 11.04 - 9.40 0.05 0.21 3.88 Q3 1.98 10.94 -9.87 0.02 0.54 3.62 Q4 1.98 10.78 - 9.70 0.02 0.53 3.63 1959Q1 1.98 11.54 - 10.09 0.03 0.54 4.01 Q2 1.98 11.60 - 9.72 0.03 0.55 4.45 Q3 1.98 11.47 - 9.51 0.03 0.54 4.52 Q4 1.98 11.27 -9.17 0.03 0.54 4.65 1960Q1 1.98 12.24 - 9.71 0.04 0.55 5.10 Q2. 1.98 11.72 - 9.61 0.04 0.57 4.69 Q3 1.98 11.73 - 9.65 0.03 0.57 4.67 Q4 1.98 11.45 - 9.50 0.03 0.59 4.55 LONG Constant 0.102 Y -0.124 M -0.008 S 0.36 L Predicted rT RATE 1 2 3 4 5 6 1958Q1 2.02 8.65 -7.11 -0.37 0.55 3.75 Q2 2.02 8.74 -7.12 -0.36 0.51 3.79 Q3 2.02 8.66 -7.47 -0.14 1.32 4.40 Q4 2.02 8.54 -7.34 -0.17 1.30 4.35 1959Q1 2.02 9.14 -7.64 -0.21 1.31 4.63 Q2 2.02 9.19 -7.35 -0.24 1.34 4.96 Q3 2.02 9.08 -7.20 -0.25 1.32 4.98 Q4 2.02 8.92 -6.94 -0.26 1.30 5.05 1960Q1 2.02 9.69 -7.35 -0.28 1.35 5.43 Q2 2.02 9.28 -7.28 -0.27 1.38 5.14 Q3 2.02 9.29 -7.30 -0.26 1.38 5.14 Q4 2.02 9.06 -7.19 -0.24 1.43 5.09 TABLE 8: Decomposition of Implied Reduced Form Equations 29 and 30 8 Col. 9 = Z col. i i=l SHORT RATE Constant 1 0.158 Y 2 -0.322 M 3 0.161 S 4 0.175 L 5 -0.133 M 6 -1 -0.163S. 7 "i -0.171L 8 Predicted r 9 b 1958Q1 1.53 13.43 -18.49 7.33 2.69 7.13 -7.75 -2.07 3.79 Q2 1.53 13.57 -18.51 7.22 2.45 7.64 -7.42 -2.62 3.85 Q3 1.53 13.44 -19.42 2.75 6.38 7.64 -7.31 -2.39 2.62 04 1.53 13.24 -19.08 3.44 6.29 8.02 -2.78 -6.24 4.42 1959Q1 1.53 14.18 -19.85 4.19 6.36 7.88 -3.49 -6.14 4.66 Q2 1.53 14.26 -19.12 4.69 6.47 8.20 -4.25 -6.22 5.57 Q3 1.53 14.09 -18.71 5.01 6.39 7.90 -4.75 -6.33 5.12 Q4 1.53 13.84 -18.04 5.07 6.31 7.73 -5.07 -6.24 5.12 1960Q1 1.53 15.04 -19.11 5.49 6.52 7.45 -5.13 -6.16 5.62 Q2 1.53 14.40 -18.92 5.38 6.69 7.89 -5.56 -6.37 5.04 Q3 1.53 14.41 -18.98 5.07 6.70 7.81 -5.45 -6.54 4.55 04 1.53 14.07 -18.70 4.66 6.90 7.84 -5.13 -6.55 4.62 Continued N5 TABLE 8 (Continued) LONG RATE Constant 1 0.114 Y 2 -0.233 M 3 0.089 S 4 0.196 L 5 0.106 M 6 -0.091 S 7 -0.157L 8 Predicted r 9 L 1958Q1 0.93 9.69 -13.38 4.05 3.01 5.68 -4.32 -1.90 3.76 Q2 0.93 9.79 -13.39 3.99 2.74 6.09 -4.14 -2.41 3.60 Q3 0.93 9.70 -14.05 1.52 7.15 6.09 -4.08 -2.20 5.06 04 0.93 9.56 -13.81 1.90 7.04 6.39 -1.55 -5.73 4.74 1959Q1 0.93 10.23 -14.37 2.32 7.12 6.28 -1.95 -5.64 4.94 Q2 0.93 10.29 -13.83 2.59 7.25 6.54 -2.37 -5.71 5.68 Q3 . 0.93 10.17 -13.54 2.77 7.15 6.29 -2.65 -5.81 5.31 Q4 0.93 9.99 -13.06 2.80 7.06 6.16 -2.83 -5.73 5.33 • 1960Q1 0.93 10.85 -13.83 3.04 7.30 5.94 -2.86 -5.66 5.70 Q2 0.93 10.39 -13.69 2.97 7.49 . 6.29 -3.11 -5.85 5.44 Q3 0.93 10.40 -13.74 2.80 7.50 6.23 -3.04 -6.00 5.08 Q4 0.93 10.15 -13.51 2.58 7.73 6.25 -2.86 -6.01 5.23 TABLE 9: Decomposition of Estimated Reduced Form Equations 31 and 32 8 Col. 9 = |j col. i SHORT RATE Constant 1 0.124 Y 2 -0.201 M 3 0.082 S 4 0.099 L 5 0.056 M 6 -0.092 S 7 -0.099 L 8 1 Predicted r 9 S 195801 02 Q3 Q4 2.14 2.14 2.14 2.14 10.53 10.65 10.55 10.39 -11.55 -11.56 -12.13 -11.92 3.75 3.70 1.41 1.76 1.51 1.38 3.60 3.54 2.98 3.19 3.20 3.35 -4.35 -4.17 -4.11 -1.56 -1.20 -1.52 -1.39 -3.61 3.83 3.82 3.27 4.11 1959Q1 Q2 03 Q4 2.14 2.14 2.14 2.14 11.13 11.19 11.06 10.86 -12.40 -11.94 -11.69 -11.27 2.15 2.40 2.57 2.60 3.58 3.65 3.60 3.55 3.30 3.43 3.30 3.23 -1.96 -2.38 -2.67 -2.85 -3.56 -3.60 -3.66 -3.62 4.39 4.88 4.65 4.66 1960Q1 Q2 Q3 Q4 2.14 2.14 2.14 2.14 11.80 11.30 11.31 11.04 -11.94 -11.82 -11.86 -11.68 2.81 2.76 2.60 2.39 3.67 3.77 3.78 3.89 3.12 3.30 3.27 3.28 -2.88 -3.12 -3.06 -2.88 -3.57 -3.69 -3.79 -3.79 5.16 4.64 4.39 4.38 Continued TABLE 9 (Continued) LONG RATE Constant ' 1 0.102 Y 2 -0.139 M 3 0.006 S 4 0.050 L 5 0.018 M 6 -0.018 S 7 -0.018L 8 Predicted r 9 L 1958Q1 2.12' 8.67 -7.80 0.28 0.77 0.96 -0.86 -0.22 3.72 Q2 2.12 8.77 -8.01 0.27 0.70 1.03 -0.82 -0.28 3.78 Q3 2.12 8.68 -8.40 0.10 1.83 1.03 -0.81 -0.26 4.30 Q4 2.12 8.56 -8.25 0.13 1.81 1.08 -0.31 -0.67 4.46. 1959Q1 2.12 9.16 -8.59 0.16 1.83 1.06 -0.39 -0.66 4.70 Q2 2.12 9.21 -8.27 0.18 1.86 1.10 -0.47 -0.67 5.06 Q3 2.12 9.10 -8.09 0.19 1.83 1.06 -0.53 -0.68 5.01 Q4 2.12 8.95 -7.81 0.19 1.81 1.04 -0.56 -0.67 5.07 1960Q1 2.12 9.72 -8.27 0.21 1.87 1.00 -0.57 -0.66 5.42 Q2 2.12 9.30 -8.18 0.20 1.92 1.06 -0.62 -0.68 5.13 Q3 2.12 9.31 -8.21 0.19 1.92 1.05 -0.60 -0.70 5.08 04 2.12 9.09 -8.09 0.18 1.98 1.06 -0.57 -0.70 5.06 TABLE 10: Regression Results for the Model of Section 2C(iv) Es timation Period Dependent Variable Constant rs Max R2 at lag DI D2 LAR USS R2 See d w Eqn. # 1951Q1 to 1958Q2 Spread 1.93 (19.17) -0.51 (-16.77) 0.91 0.11 0.44 33 1955Q1 to 1958Q2 Spread 1.58 (37.34) -0.50 (-40.45) Six Quarters 0.99 0.03 1.96 .41 1955Q1 to 1965Q4 Spread 0.62 (2.81) -0.35 (-10.39) Six Quarters -0.19 (-13.7.0) 0.06 (3.11) 0.19 (2.65) 0.93 0.11 1.20 42 1955Q1 to 1965Q4 Spread (0.28V (1.28) -0.33 (-9.71) S ix Quarters - 0.21 (-13.30) 0.07 (3.51) 0.16 (2.09) 0.93 0.11 .1.24 43 127 TABLE 11: Decomposition of Regression Equation 42 e CoL 7 = £ col. i. Column 6 is calculated residually i=l Constant 1 -0.19 DI 2 0.06 LAR 3 0.19 USS 4 -0.35 rs 5 Expecta tions 6 Predicted Spread 7 1958Q1 0.62 -0.78 0.96 0.13 -1.29 0.76 0.40 Q2 0.62 -0.81 1.01 0.21 -1.15 0.81 0.69 Q3 0.62 -0.21 1.03 0.12 -1.08 0.78 1.26 Q4 0.62 -0.21 0.97 0.03 -1.39 0.68 0.70 1959Q1 0.62 -0.21 0.95 0.00 -1.64 0.51 0.22 Q2 0.62 -0.20 0.96 -0.04 -1.81 0.45 -0.02 Q3 0.62 -0.21 0.93 -0.10 -1.93 0.51 -0.19 Q4 0.62 -0.20 0.97 -0.14 -1.86 0.63 0.02 1960Q1 0.62 -0.20 0.99 -0.09 -1.88 0.78 0.21 Q2 0.62 -0.21 1.02 -0.02 -1.61 0.84 0.64 Q3 0.62 -0.21 1.03 0.07 -1.43 0.92 0.99 Q4 0.62 -0.20 0.97 0.07 -1.53 0.92 0.85 128 TABLE 12: Decomposition of Regression Equation 43 6 Col. 7 = E col. i. Column 6 is calculated residually i=i Constant 1 -0.21D2 2 0.07LAR 3 0.16USS 4 -0.34 rs 5 Expecta tions 6 Predicted Spread 7 1958Q1 0.28 -0.63 1.12 0.11 -1.23 0.79 0.44 Q2 0.28 -0.68 1.18 0.17 -1.10 0.83 0.69 Q3 0.28 -0.10 1.20 0.10 -1.03 0.79 1.24 Q4 0.28 -0.13 1.14 0.02 -1.33 0.69 0.68 1959Q1 0.28 -0.15 1.11 0.00 -1.57 0.53 0.20 Q2 0.28 -0.17 1.13 -0.03 -1.72 0.48 -0.03 Q3 0.28 -0.18 1.09 -0.08 -1.85 0.54 -0.20 Q4 0.28 -0.18 1.14 -0.11 -1.77 0.66 -0.01 1960Q1 0.28 -0.19 1.16 -0.07 -1.80 0.81 0.18 Q2 0.28 -0.18 1.20 -0.02 -1.54 0.87 0.61 Q3 0.28 -0.17 1.20 -0.05 -1.36 0.94 0.95 Q4 0.28 -0.15 1.14 0.06 -1.46 0.94 0.81 129 TABLE 13 Standard Deviation for Each of the Four Maturity Classes in the Series Inside, Public, Sutch and FRB, and of the Ratios of Bonds Under 10 Years to Over 10 Years 13(a) ^^JJata Series Maturity ^-v. Inside Public Sutch FRB 0-2 Years 651.70 0-2 Years 379.21 Short 1135.20 0 - 1 Years 9716.51 13(b) ^^^JData Series Maturity ^-v. Inside Public Sutch FRB 2-5 Years 530.79 2-5 Years 282.85 Medium CI) 629.28 1-5 Years 8303.60 130 13(c) NSSSs\pata Series Ma t ur i tyVNVNNNXXv\ Inside Public Sutch FRB 5-10 Years 989.66 5-10 Years 581.00 Medium (II) 662.90 5 - 10 Years 7465.77 13(d) Maturity ^^N,^^ Inside Public Sutch FRB Over 10 Years 1059.57 Over 10 Years 1019.74 Long 314.56 Over 10 Years 4622.97 131 13(e) Data Series Inside Public Sutch FRB Shorts Longs 1.30 1.22 1.30 1.79 132 TABLE 14 Coefficients of Variation for Each of the Four Maturity Classes in the Series Inside, Public, Sutch and FRB, and of the Ratios of Bonds Under 10 Years to Over 10 Years. 14(a) ^sData Series Maturity Inside Public Sutch FRB 0-2 Years 31.20% 0-2 Years 30.24% . Short 18.50% 0-1 Years 17.87% 14(b) ^vData Series Maturity Inside Public Sutch FRB 2-5 Years 27.18% 2-5 Years 28.86% Medium (I) 16.19% 1 -"5 Years 17.95% 133 14(c) ^\Data Series Maturity Inside Public Sutch FRB 5-10 Years 50.28% 5-10 Years 47.72% Medium (II) 28.72% 5-10 Years 31.71% 14(d) \ Data Series Maturity Inside Public Sutch FRB Over 10 Years 28.39% Over 10 Years 33.00% Long 12.98% Over 10 Years 19.90% 134 14(e) Data Series Inside Public Sutch FRB Shorts Longs 54.54% 74.04% 24.62% 31.48% TABLE 15: Regression Results for Section 5 Estimation Period Dependent Variable Constant rst-i Rst-i R2 See d w Eqn. # 1955Q1 to 1965Q4 rst 2.31 (4.07) 0.38 (3.45) 0.20 0.69 0.38 44 1955Q1 to 1965Q4 r St 0.76 (2.47) 0.83 (11.59) 0.76 0.38 1.54 45 1955Q1 to 1965Q4 Rst 0.17 (0.31) 0.89 (6.25) 0.47 0.45 0.56 46 1955Q1 to 1965Q4 R St 0.68 (2.46) 0.83 (10.78) 0.73 0.32 1.54 47 LO 0.10 0.00 -0.10 -0.20 -0.30 GRAPH 1 136 Estimated Distributed Lag in Equation 41 1 / 2 3 4 5 6 Quarters Coefficient t Ratio -0.50 -40.45 -0.14 -41.87 0.04 9.42 0.10 19.92 0.08 23.22 0.03 24.80 0.00 0.00 137 GRAPH 2 0.00 -0.10 -0.20 -0.30 Estimated Distributed Lag in Equation 42 1/23 4 5 6 Quarters Coefficients t Ratio -0.35 -10.39 -0.08 .-6.73 0.06 4.58 0.09 7.18 / 0.07 7.96 0.02 8.31 0.00 0.00 GRAPH 3 0.30 139 CHAPTER FOUR FINANCIAL RESPONSES TO A  NEW TERM STRUCTURE OF INTEREST RATES In the previous chapter the determinants of the Canadian term structure of interest rates were examined. Of particular interest is the finding that the composition of the federal government debt does affect the term structure of government bond yields. Given that much, debt management operations such as the Conversion Loan, can twist the yield curve. The question then arises: What are the effects of such changes on the holding and issuing patterns of lenders^ and borrowers? In this chapter these problems are examined. Information on these matters is desirable per se. It has been suggested, for example, that as long-term rates increase relative to short-term ones, cost minimizing asset issuers would intensify their use of two alternatives. First, to.the extent that they are constrained to the issue of bonds rather than, say, stock, they may issue more short-and fewer long-term bonds. Second, they may float more bonds in foreign currencies. In section one, the reasons why the response of lenders -- to the changes in the term structure that the Conversion Loan effected -- cannot be examined are stated. In section two, the response of borrowers is considered. The extent to which the two alternatives stated above are utilized is examined 140 in parts B and C of this section. In section three, the implications of section two for some of the effects of the Conversion Loan are discussed. Finally, section four examines the assumption made in what follows that government and other -- provincial, municipal and corporate -- bonds of the same term to maturity are perfect substitutes. SECTION ONE: THE RESPONSE OF ASSET HOLDERS Concerning the demand side of the problem, i.e. the holding patterns of lenders, little will be said. The reason is the extreme paucity of available information. What one aims for is time series of balance sheets for the various sectors. Moreover, these accounts must report government and other bonds in sufficient detail: Such bond holdings must be disaggre gated by terra to maturity. The Flow of Funds Accounts published by the Dominion Bureau of Statistics -- now Statistics Canada -- do contain such time series of sectoral balance sheets. There are, however, two major problems. The sectoral bond holdings are, naturally, disaggregated by issuer but not by term to maturity. Secondly, the Flow of Funds Accounts were not published during the 1950's. A pioneering study for the Royal Commission on Canada's Economic Prospects2 gives some information for the period 1947_l955. But this too does not disaggregate bonds by term to maturity and in any case a gap for the years 1955- 1962 still remains. Alternative sources of information were sought. Other published material is not helpful and personal inquiries at the Bank of Canada proved sterile. More information may become available in the future as the Flow of Funds Accounts series becomes more established. At that time a study of the 141 demand side of the problem may prove feasible. More information is available on changes in ownership of Canadian securities held by foreign residents and foreign securities held by Canadians. However, since capital flows have been extensively studied by Helleiner (25), Penner (51), Powrie (52), Lee (38), Caves and Reuber (13) and RDX2 (27), attention is paid only to the new issues component of such flows. SECTION TWO: THE RESPONSE OF ASSET ISSUERS A. INTRODUCTORY Time series of balance sheets provide information not only on the demand side, but also on the supply aspects of the problem. That is, they would help answer questions like: i) If the spread between Canadian and U.S. interest rates widens, do borrowers become more inclined to incur liabilities in U.S. funds? ii) If the spread between Canadian long-and short-term rates widens do borrowers issue shorter term securities? iii) Do high interest rates discourage borrowing? Given the paucity of the information contained in existing time series of balance sheets it might appear that such questions might remain unanswered. Fortunately an alternative source of information is available concerning issuing patterns of borrowers. 142 The Financial Post publishes an annual record of new financial issues. It gives, for every month, a list of bond and stock issues floated by the federal, provincial and municipal governments and by corporations. Each item in the list tells the face value of an instrument, whether in Canadian or U.S. funds, the coupon rate and yield, date of issue and maturity and various other less important details. This information is very accurate. This statement is based on a comparison of these data with unpublished material kindly made available by the Bank of Canada. Thus, while it is not possible to examine the term to maturity composition of the liabilities of the main bond issuers, it is possible to construct tables giving the maturity composition of new issues of bonds by the federal and provincial governments and by corporations. These tables were reported in chapter one. It should be noted that in studying such data no identification problems arise. The reason is that our information is not about quantities traded -- the usual kind of information -- but rather it reflects true borrower intentions -- points on supply curves. Municipal governments are fairly important bond issuers but they are usually not at liberty to adjust their issuing patterns quite as much as other borrowers when market conditions change. Municipal borrowing is regulated by the province concerned. Frequently municipalities are obliged to issue serial bonds so that interest and principal are repaid annually.^ For this reason the term to maturity of a serial bond is ambiguous and so is therefore the maturity composition of municipal debt. In principle some idea about it can be obtained: One needs detailed information on each 143 issue. Then for each particular bond an average term to maturity can be calculated. When this is done for all bonds some idea about the term to maturity of municipal debt can be obtained. Since a provincial by-law is usually required, authorizing each municipal issue, detailed information on each bond issue is, in fact, available. But the computational work involved is formidable. Because of this problem a table for Canadian municipalities similar to Tables 15, 16 and 17 was not constructed. It may be argued, however, that, since the average term to maturity of an X year serial bond is smaller than that of an X year sinking fund bond, a municipality wishing to issue shorter term liabilities may switch away from the latter to the former. This kind of possibility is investigated using another source of information -- Table 19 of chapter one. The extent to which municipalities tap U.S. funds when it is profitable to do so will be briefly examined using annual data -- Table 14, chapter one. The new issue data on municipalities could also have been used but were not for two reasons. To begin with, comparing the Financial Post reports with those of the Bank of Canada indicated a substantial number of disparities --this was not the case with provincial and corporate issues. A second, related, problem was the large number of rather small issues appearing in each month. This makes the clerical work involved quite substantial. The questions mentioned three paragraphs earlier are now investigated. It was seen earlier that Boreham e_t al (7) had argued that the Conversion Loan increased the interest rate differential between Canada and the U.S., thereby inducing borrowers to issue bonds in U.S. dollars. The influx of 144 this capital is claimed to have appreciated the exchange rate and hurt the economy by handicapping our export industries. The extent to which borro wers issue U.S. dollar bonds under those circumstances is first examined. Implications for the effects of the Conversion Loan are stated later. B. WHEN DO BORROWERS ISSUE LIABILITIES IN U.S. DOLLARS? In discussions of the openness of the Canadian economy and its links with that of the U.S. the connections between the financial sectors of the two economies are emphasized. One such connection arises out of the alleged willingness of asset issuers in one country to float issues in the currency of the other country if the terms are right. The terms that a borrower must consider include interest rate differentials and the relation between the spot rate now and that prevailing at appropriate future dates. The latter prices are, of course, unobservable, the individual issuer must form expectations about them. A Canadian issuer, for example, will be more likely to float issues in U.S. dollars the higher the interest rate differential between Canada and the U.S. (CR-USR), and the higher the difference between the amount of Canadian dollars required to buy $1.00 U.S. and the spot rate expected to prevail in the future (S-Se). These considerations underly the modern version of the interest rate parity theory.-^ Using Se rather than the forward rate (S) established on the market may be necessary for two reasons. i) Individual issuers may or may not wish to cover themselves with forward contracts, if appropriate forward markets exist. ii) In fact such markets are not adequate. Many of the provincial 145 issues, for example, are as long as twenty years' The province concerned must think not only of repaying the principal but also of the interest payments due between the time of issue and repayment. Since it cannot cover itself by buying U.S. dollars forward it must speculate -- this involves constructing an Se. Using these considerations an attempt is now made to analyse some of the available information. Table 14 of chapter one gave annual data on the liability structure of federal, provincial and municipal governments, corporations, and other institutions. There, the distinction drawn is between Canadian dollar and other currency liabilities. Table 1 below gives the proportion of total liabilities issued in other currencies for the four main groups of borrowers. It shows that provinces, municipalities and corporations issue a substantial -- about 0.23 -- proportion of their liabilities in currencies other than Canadian dollars. This is not true of the federal government which does so for only 0.02 of its bond issues. The same table also shows that the variance of each proportion is quite small. No attempt is made to apply regression analysis to the data because the relevant period contains so few observations. Instead, the relationship between each of columns 1 to 4 and column 5 -- the spread between the Canadian government bond yield CR and the corresponding U.S. one, USR i.e. CR-USR -- was examined on graphs not appearing here. One would expect this relationship to be a positive one: As the spread increases so does the proportion of debt denominated in other currencies. This appears to be partially true for municipalities and corporations and untrue for the federal and provincial governments. 146 There are, of course, two lots of assumptions implicit in the constru ction of those graphs. Firstly, the U.S. bond yield is used as a proxy for the bond yields in other countries generally. This is not unreasonable in view of the relative importance of U.S. dollar issues in the other currency category. Secondly, the assumption is made that government and other securities -- provincial, municipal and corporate -- are perfect substitutes. This problem will be examined in detail later on. More evidence on this issue can be had from the new issue data discussed earlier on. It has already been mentioned that new issue data on municipal debt were, for various reasons, not constructed. Table 15, chapter one, shows that between 1955Q1 and 1965Q4 the federal government issued bonds in U.S. dollars on two occasions only. This makes it difficult to infer anything about its behaviour in this respect. Thus, we concentrate on provinces and corporations and utilize the data of Tables 16 and 17, chapter one, to examine whether these bodies will issue liabilities in U.S. dollars^ when it is advantageous for them to do so. In accord with earlier discussion the proportion of new provincial issues and new corporate issues in the respective totals is regressed against a constant, the spread CR-USR, a variable reflecting the availa bility of credit in Canada -- namely the nominal, narrow, money supply M -- and the difference between the spot rate S and the expected future spot rate Se. Two alternative specifications of Se were made. In specification one, Se was set equal to the only forward rate (F) available, the 90-day one. The rationale is that if a province or corporation wished to hedge its loan and it borrowed on a 90-day basis, 147 the 90-day forward rate would be the rate that it would use. In specifica tion two, Se was made a function of past spot rates.. The reason for doing so is that the underlying expectational framework is the same as the one generating expectations about future short interest rates. This mechanism was extensively discussed in the last chapter. The main point is that if the spot rate has been rising, extrapolative expectations would have it continue rising in the future, while regressive expectations see it falling to a normal level. Since different provinces and corporations may have a different view of the future, a combination of both regressive and extrapo lative elements may be necessary in order to explain observed behaviour. In this specification the number of relevant past spot rates as well as the weights attached to each one of them is determined empirically. A third degree polynomial was specified in the context of a modified Almon procedure and the first Almon variable was dropped. This imposes further restrictions on the shape of the polynomial describing the weight pattern so that only one turning point in it can occur, in accord with theoretical considerations discussed in chapter three. The results obtained are consistent with the conjectures made on the basis of the annual data considered earlier on. Table 2 and Graphs 1 and 2 give the necessary details. They show that: i) The overall explanatory power of either specification is very low so that the maintained hypothesis, that the vector of coefficients is equal to the zero vector, must be accepted. This means that the following state ments are made quite informally. ii) There is some evidence that the hypothesis more accurately 148 describes the behaviour of corporations than it does that of provinces. iii) Specification two provides a better representation of how expecta tions concerning future spot rates are formed. The coefficients for (S-F) do not have the anticipated signs. iv) The credit availability variable, M, in the equation for corporate placements in the U.S. does not have the expected sign. The remaining variables do. How can these results be rationalized? Firstly, it may be argued that a quarter is too fine a period of time for looking at new issue data. This is because new issues by both provinces and corporations are fairly sparse. Whatever variance there may be in the dependent variables^ may, therefore, be of no economic significance. This statement may be consistent with the one made earlier to the effect that the alternative, annual, data examined above varied over a small range. Statement (ii) may be consistent with informal, but widely held, views that business firms are better cost minimizers than government agencies. Statement (iii) should come as no surprise. Although provinces and corporations may wish to hedge in their dealings with U.S. markets the opportunities for doing so are quite limited. There is no possibility of covering a twenty-year contract, as many of the bonds issued are. Then expectations about future spot rates must be formed, at least partially, out of current and past experience with the behaviour of the spot rate. This hypothesis is fairly consistent with corporate behaviour, as Graph 2 shows. Finally, statement (iv) is consistent with at least two thoughts. First, that corporations possess more means of finance than provinces do, so that credit availability is less likely to 149 affect their operations. Second, if monetary policy is effective, when M decreases firms are likely to reduce their risky means of financing projects -- i.e. borrowing in the U.S. -- before they reduce the less risky ones -- i.e. borrowing in Canada. If so, the sign of M should in fact be positive. It is noteworthy that the behaviour of bond issues in foreign funds has proved a difficult "nut to crack". Helleiner, for example, using both delivery data and alternative contract-data, reports results no more encouraging than those presented here. Also, the explanatory power of the analogous equations in the RDX2 model of the Canadian economy is about the lowest in the entire model.^ So much for this issue. The extent to which provinces and corporations switch to short-term financing as the spread between the long and short government bond yields increases and as interest rates in general^ increase will now be examined. This constitutes the second possibility of adapting issuing patterns to changed costs of borrowing. It was shown in chapter three that the Conversion Loan increased the spread between long and short rates. Did bond issuers subsequently adjust their financing patterns? This issue is examined first. Implications for the Conversion Loan are again confined to a separate section. C. WHAT DETERMINES THE SHORT-LONG MIX OF BOND ISSUES? In perfect this part of section 2, as in the last one, the assumption of substitutability between government and other bonds is maintained. 150 The response of municipalities, provinces and corporations is examined. In line with previous comments it is hypothesized that as the spread between the government long and short bond yield (= Spread) increases, municipali ties issue more serial and less sinking fund bonds. In Table 3, the ratio of serial to sinking fund municipal issues by province, is given for the years 1955 - 1965. A trend can be detected in each column but it is not always in the same direction: Serial bonds have become more popular with municipalities in some provinces and less so with others. The last column gives the ratio, for municipalities in all provinces. There is a downward trend here. This last column was plotted against the Spread. The graph revealed, if anything, a negative relationship between the two, contrary to what one might expect. Turning to the new issue -- quarterly -- data for provinces and corpora tions, the ratio of new short issues to new long ones is regressed against a constant, the Spread, and the rate CR -- a simple average of the Canadian government bond yields over and under 10 years. The variable Spread should carry a positive sign and so should variable CR. Graphs 5 and 6 depict the two dependent variables and Table 4 gives the estimated regression equations. The following comments•may be made. i) Here, as in the previous section, the maintained hypothesis cannot be rejected. As a result the following points are made informally. ii) The sign of the Spread variable is different in equations 5 and 6. iii) There is some evidence that high interest rates coincide with decreases in the ratio of new corporate shorts to their new long ones.' 151 At first these results may appear implausible, but this may not be so on reflection. The argument that,as the Spread increases cost minimizing implies issuing more shorts, is too simplistic. Similarly, the argument that as CR increases only short-term commitments will be undertaken omits important forces. Both arguments ignore the importance of expectations. The first argument ignores expectations about future short rates. If the expectations hypothesis on the term structure of interest rates holds, then at any moment in time there is an expected future short rate implied^ by the market r*. An individual bond issuer will have his own expectations about that rate, J let us say that he expects it to be re. Then he will issue: longs if re>r" e ~k shorts if r <r and be indifferent if re = r* This will hold regardless of the shape of, or changes in, the yield curve. The argument leading to an a priori sign on CR ignores expectations about the future level of interest rates in general. An increase in CR will not deter investors from committing themselves to high interest payments if even higher CR values are expected to prevail in the future. For these reasons the results in Table 4 are not too implausible. SECTION THREE: IMPLICATIONS OF SECTION TWO B FOR A STUDY OF THE  CONVERSION LOAN It was seen in section two B that Boreham et al_ (7) have claimed that 152 the Conversion Loan induced borrowers to issue bonds in U.S. dollars. In his 1962 paper, Barber (3) had made the more general statement that the Loan attracted capital from abroad. He did not specify what particular forms of capital were involved and hence the suggestion by Boreham et al maybe at least part of what Barber had in mind. The results presented above indicate that this argument is not supported by the evidence. Also, the relevant equations in RDX2 show that such flows have not been found to be sensitive to Canadian-U.S. interest rate differentials, as already indicated. This contradicts the point by Boreham e_t a_l and requires that, for Barber's argument to ho Id, some other capital flows must be sensitive to such differentials. There is, in fact, ample evidence that this is so. The studies by Helleiner (25), Penner (51), Powrie (52), Lee (38), Caves and Reuber (13) and the RDX2 researchers (27) all point to that direction.^ Use has been made thus far of the assumption of perfect substitutability between government securities and those issued by others -- when the term to maturity is held constant. It is now necessary to question this assumption. SECTION FOUR: ON SUBSTITUTABILITY IN FINANCIAL MARKETS It is very difficult to supply a viable definition of perfect substi tutability between government bonds and other bonds of the same term to maturity. Price theory definitions cannot be applied here: Since the maturity composition of other debt is not known -- only the maturity composition of new issues between 1955 - 1965 is known -- demand functions 153 cannot be estimated and cross elasticities must remain unknown. . An alternative definition may be that the rate of return on a government bond with X years to maturity is identical to that for other bonds of the same term. Accordingly, the government bond yield under ten years was regressed on the provincial and corporate new issue yields on bonds with less than ten years to maturity. Similarly for government bond yields over ten years and the provincial and corporate new issue yields on bonds with more than ten years to maturity.^ If the assump tion of perfect substitutability holds, 45° lines should be estimated. Thus, zero intercept and a slope equal to unity becomes the null hypothesis. Table 5 presents the results obtained. i) All constants except that in equation 10 are not significantly different from zero at the 1% level of significance. ii) All slope coefficients are not significant]}7 different from unity at the YL level. iii) The explanatory power of the provincial equations 7 and 9 is higher than that of the corporate ones 8 and 10. iv) The d w statistic indicates positive serial correlation in equations 8, 9 and 10-. There is no positive serial correlation in equation 7. The significance of- the constant term in equation 10 requires comment. In footnote 15, mention was made of the findings of the two NBER studies on the spread between new and seasoned long corporate bond yields. If this spread exists in Canada too, then running the equations in the form 154 used here would result in a negative intercept. Thus, equation 10 provides some evidence corroborating the NBER results. On the whole the hypothesis of perfect substitutability is quite consistent with evidence. Thus, a fair amount of confidence can be invested in the results of sections two B and two C. 155 NOTES TO CHAPTER FOUR 1. In Canada the main bond suppliers are the federal, provincial and municipal governments and corporations. The main bond demanders are financial institutions, some governmental bodies, such as the Unemploy ment Insurance Fund and private individuals. 2. See the appendix in Hood (31). 3. Tables 15, 16 and 17. 4. This is believed to minimize the possibility of mismanagement by the allegedly relatively inexperienced municipal treasurers. 5. For good expositions see Kesselman (35) and Stoll (62). 6. Mainly U.S. dollars. Note that the range of the variables in the first three columns of Table 1 is rather small. 7. The new issue data indicate that these were the only issues in currencies other than Canadian dollars. 8. See Massager (40), Operation 53. 9. See Graphs 3 and 4 for an indication of such information. 10. See Helleiner (25), pp. 386-387 and RDX2 (27), equations 19.5 and 19.6, in Part 2, pp. 119 - 120. 11. See Wonnacott (69). p. 143. 12. See section one, chapter three. 13. If firm and uniform expectations are held then re = r*. 14. It may be worthwhile to outline some of the main differences between the study in this thesis and the one by Caves and Reuber (13) -- the most extensive and recent of those mentioned above. They are: A. DEPENDENT VARIABLE DIFFERENCES i) Caves and Reuber -- CR -- use balance of payments data which refer to deliveries whereas the Financial Post data used here are offer-data. It is well-known, and as CR imply -- CR, pp. 35 - 36 -- offer-data are preferable since they more accurately reflect borrowers' intentions. ii) The data used in this chapter include only corporate and provincial issues --see page 146 for reasons -- whereas the CR data 156 presumably include federal and municipal issues. The inclusion of issues by the federal government should make little difference, since, in the relevant period, it issued hardly any U.S. - dollar bonds. However, the inclusion of municipal issues could make a substantial difference: There is some evidence that the proportion of municipal debt outstanding issued in currencies other than Canadian dollars is positively related to the Canada-U.S. interest rate differential --see p. 145.. Then the CR results are likely to arise. In what follows, the sensitivity of portfolio capital flows to interest rate differen tials observed by CR will be referred to as the "CR results". iii) The data used here is "gross-new" whereas theirs is "net". To the extent that Canadians retire fewer securities -- thereby increasing the net inflow of capital -- as the Canadian-U.S. differential increases the CR results are again more likely to arise. iv) CR include U.S. and other foreigners' issues in Canadian dollars and, of course, their retirements. These issuers may be more responsive to interest rate differentials than Canadians are. v) Unlike the data used here, the CR data include stocks. They had included a yield-on-capital-differential variable -- i.e. DRK in CR, pp. 58 - 59' -- which they regard as an acceptable proxy of the appropriate rates of return, even though it is not statistically significant at the customary 5% level. However, it is possible that their CL and USL rates are better proxies for the appropriate rates of return. If so, and if net flows of "stock-capital" are sensitive to the Canadian-U.S. differential, then the CR results might occur. vi) Finally, there are the differences between the Financial Post data and reality as presented in government statistics. A rough indication of these differences is available in Tables 16b and 17b of chapter one. B. INDEPENDENT VARIABLE DIFFERENCES i) Caves and Reuber use CR and USR, rather than the differential CR-USR used in this study. Their specification captures, they argue, expectations of future changes in these rates. But their argument is couched in terms of regressive expectations alone. Moreover, they still feel it necessary to include a separate expectational variable. ii) Their expectational variable CTS assumes that expectations are regressive and that they are realised -- both assumptions are question able. The apparent significance of CTS can be otherwise accounted for. iii) As CR point.out, of their availability variables, only NNCS, i.e. net new issues sold to Canadians, is vaguely acceptable. Of course a good proxy of credit availability must reflect excess demand, but such proxies are hard to come by. 157 Thus, the main difference between the two studies is that they use a much higher level of aggregation. For the purposes of analysing the conjecture made by Boreham et_ al_ -- see p. 155 -- this study is adequate. 15. This procedure requires the assumption that a new bond with X years to maturity is a perfect substitute for an (X+Y) year bond issued Y years ago. Conard (16) and Conard and Frankena (17), present evidence that the yield on the former is usually above that on the latter. They did not examine whether this "premium" differed according to the size of X: The bond yields examined are those on very long-term bonds --around 26 years. See Conard (16) p. 106. 158 TABLE 1 Proportion of Outstanding Debt that has Been issued  in Currencies Other than Canadian. Spread. Between Canadian and U.S. Federal Government Bond. Yield Year Federal Government Provincial Government Municipal Government Corporations CR-USR 1 2 3 4 5 1955 0.000 0.224 0,187 0.188 0.114 1956 0.000 0.240 0.205 0.200 0.509 1957 0.024 0.224 0.221 0.233 0.800 1958 0.023 0.225 0.237 0.238 0.686 1959 0.012 0.244 0.250 0.237 0.764 i960 0.012 0.232 0.253 0.216 0.923 1961 0.008 0.197 0.229 0.234 0.833 1962 0.014 0,204 0.235 0.256 1.134 1963 0.019 0.218 0.216 0.265 0.840 1964 0.018 0.231 0.220 0.265 0.779 1965 0.018 0.238 0.213 0.266 0.666 Sources: Columns 1-4; Table 14, chapter one. Column 5f CR is the Canadian Rate — a simple average of the rates SA2 and ru--see Data and Sources, chapter three 0 USR is the U0SS rate — a simple average of U.S. federal government bond yields constructed by Rc Sutch (63 ), TABLE 2 Regression Equations for Specification One (Equations 1 and 2) and Two (Equations 3 and 4) Equ. # Estimation Period Dependent Variable Constant CR-USR S-F M ^Wi St-i R2 See d w 1 1955Ql-65Q^ Ratio of New Prov incial Issues in the U.S. to their Total New Issues 0.20 (0.91) Ooll (0.85) -0.62 (-0.03) -0.00002 (-0.53) 0.00 0.23 1.76 2 1955Q1-65Q4 Ratio of New Cor porate Issues in the U.S. to their Total ITew Issues -0.28 (-2.01) 0.08 (1.04) -18.30 (-1.22) 0.00006 (2.61) 0.16 0.15 2.40 3 Ratio of New Pro vincial Issues in the U.S. to their Total New Issues -0.44 (-0.37) 0.12 (1.00) -0.00004 (-0.76) See Graph 5 0.00 0.24 1.78 4 1955Q1-65Q4 Ratio of New Cor porate Issues in the U.S. to their Total New Issues -1.11 (-1.50) 0.14 (1.90) 0.00001 (0.23) See Graph 6 0.19 0.15 2.46 TABLE 3 Ratio of Serial to Sinking Fund Issues by Province. The last Column gives the Ratio of all Serial to all Sinking Fund Issues by all Provinces. Year N PEI NS NB 0 M S A BC Q Y NWT All . 1955 1.34 0.51 3.68 2.28 18.64 1.37 1.89 14.54 2.17 •** * * . 5.47+ 1956 1.50 0.62 4,06 2.18 .6.31 1.10 1.81 17.07 2.11 •** * * 4.01+ 1957 1.46 O.56 4.62 3.61 4,00 1.19 1.58 20.39 2.15 19.93 * 3.91 1958 1.00 O.56 5.35 3.33 3.10 1.14 1.31 22.42 2.11 25.35 • * * 3.48 1959 1.90 O.56 6.34 4.00 2.>6 1.18 1.25 25.92 2.10 31.80 * * 3.21 i960 2.08 0.49 7.41 4.55 2.27 1.13 1.29 27.75 2.15 9.60 # 2.93 1961 2.51 0.47 12,30 4.79 2.16 1.34 1.29 34.60 2.18 4.73 * * 2.?8 1962 2.88 0.50 14.06 5.03 2.00 1.68 1.35 37.00 2.36 4.15 * * 2.70 1963 3.12 0.44 16,55 5.31 1.82 1.94 1.37 37.51 2.57 4.66 * * 2.65 1964 3.60 0.3R 16.94 5.38 1.71 1.90 1.32 38.58 2.70 0.48 * 2.14 1965 3.95 O.38 19.70 5.78 1.6? 1.84 1.4? 41.36 2.86 0.47 * •* 3.00 Source: Table 19, chapter one. * Yukon and North West Territories do not issue sinking fund debentures. These two figures exclude issues by ** The data for these two dates are not reliable, the Province of Quebec, See ** above. TABLE 4 Regression Equations for Provinces and Corporations  Pertaining to Section Two C Equation # Estimation Period Dependent Variable Constant Spread CR E 2 See d w 5 1955Q1-65Q> Ratio of New Prov incial Short Issues to their New Long Ones. 0.77 (1.04) 0.41 (1.60) -0.09 (-0.53) 0.02 0.77 1.81 6 1955Q1-W Ratio of New Cor porate Short Issues to their New Long Ones. 0.64 (5.21) -0.07 (-1.76) -0.11 (-4.00) 0.27 0.13 1.46 TABLE 5 Regression Results Pertaining to Section Four Equation No. Estimation Period Dependent Variable Constant zl Z2 z3 z4 I2 SEE d.w. 7 1955Q1-65Q4* Canadian Federal Government Bond Yield Under 10 Years 0.34 (0.63) 0.82 [0.11] 0.60 0.48 8 1955Q1-65Q4+ Same as above 0.69 (0.66) 0.59 [0.17] 0.25 0.68 0.90 . 9 1955Q1-65Q4 Canadian Federal Government Bond Yield Over 10 Years -0.24 (-0.97) 0.94 [0.05] 0.90 0.23 0.74 10 1955Q1-65Q4 Same as above -2.37 (-4.89) 1.19 [0.08] 0.83 0.31 0.90 Numbers in rounded 'orackets are t ratios. Those in square brackets are standard errors. Zl z Bond yield on new provincial issues under 10 years to maturity, Z2 5 Bond yield on new corporate issues under 10 years to maturity, Z3 s Bond yield on new provincial issues over 10 years to maturity. z4 5 Bond yield on new corporate issues over 10 years to maturity, * Excluding the following quarters during which no new provincial issues were made: 55Q2, 55Q4, 5°Q2, 643,4, 65QI. + Excluding the following quarters during which no new corporate issues were made: 55Q2, 57Q3, 58Q1, 59Q1» 62Q4, 64Q2, 64Q3. 163 164 GRAPH 3 Ratio of New Provincial Issues in the U.S GRAPH 4 GRAPH 5 Proportion of Provincial New Short  to Provincial New Long Bonds GRAPH 6 Proportion of Corporate New Short to Corporate New Long Bonds 1955 1960 169 CHAPTER FIVE  REAL RESPONSES TO THE CONVERSION LOAN INTRODUCTORY In chapter three the effects of the Conversion Loan on the level and term structure of interest rates were examined. In chapter four certain financial responses to the new term structure were investigated. It is now time to look at the effects on aggregate economic activity. It will be recalled from chapter two that important claims have been made about its effects. Specifically, it has been argued that by reducing liquidity the Loan increased the rate of interest, led to a capital account surplus, an exchange rate appreciation and a consequent decrease in economic acti vity. Barber in fact ventured a guess that GNP would have been higher by an amount in the order of billions of dollars. In proceeding, no guidance can be had from the studies of Operation Twist. As mentioned on page 213, no one has as yet investigated the claim of its proponents that, for example, it would break the trade-off between unemployment and a sound balance of payments position. This chapter utilizes published econometric information on the Cana dian economy to investigate the effects of the Loan on economic activity. In section one, some"back-of-the-envelope" calculations are presented using information from the Stewart (61 ) model of the Canadian economy. In sec tion two, the results of simulations using the Bank of Canada RDX2 model (27) are discussed. The final section contains concluding remarks. 170 SECTION ONE: SOME PRELIMINARY CALCULATIONS It will be instructive to quantify the arguments suggested in the open ing paragraph of this chapter. To that end, the Stewart (61) models-is first utilized. In that model, the average term to maturity of the federal debt held by the Public (A) enters the demand for money function much as Barber had argued it should. Estimates of changes in endogenous variables follow ing the increase in A can be had from the table of impact multipliers and o his data. The Conversion Loan increased A by 81 months-* and hence raised the 3-month Treasury Bill rate (rsc) by 187 basis points. This increase led to a rise in the average yield on Government of Canada securities over 12 years (rlc) of 11 basis points. The exchange rate^ (ERs) appreciated by $0,008. These are, of course, impact effects. The impact effects on real variables were as follows. The endogenous components of the national income identity — in real terms -- are the fami liar C, I, X and M.^ The Loan apparently had no effect on exports and it decreased imports by $7,792 million — more will be said on this later. Consumption expenditures were decreased by $9,388 million. The Loan, through its effects on rlc, reduced investment very substantially -- by $86,751 million. The sum-total of these changes is $88,347 million, although the effect on real GNP minus accrued net income of farm operators from farm production (Ygnp-nf), as given by the impact multiplier is only $57,429 million. The above discussion leaves something to be desired. Although impact tipliers take into account the complete interdependence of most variables 171 in the system they fail to capture effects that manifest themselves with a time-lag. This problem is particularly acute when lagged endogenous — and to an extent exogenous — variables play an important role in the model. It may be instructive to illustrate this argument. The equation for rlc is^ rlc = 0.330 + 0.895 rlc , + 0.056 rsc t t-1 t Looking at this equation in isolation from the rest of the model it can be seen that an increase in rsc by 100 basis points will lead to an immediate increase in rlc of only 6 basis points. Such an increase would only dis courage investment (I) by $20 million in 1958Q3 — the rlcfc coefficients in the equations for residential (Ire) and non-residential (Ibc) construc tion are -33 and -303 respectively, while other components of I are not sensitive to rlct« Turning to the long-run form of this equation, obtained by successive substitution of the expression for rlc^ ^, i.e. rlc = 0.314 + 0.533 rsc it is clear that the ultimate effect on rlc of such a change is 53 basis points. The contractionary effect on I, for example, would now be con siderably higher, namely $178 million. Thus, the long-run effect on rlcfc and hence aggregate demand given by the impact multipliers is understated. The moral is twofold: On the one hand, this model allows no possibility for the authorities to affect the long rate directly -- rlcfc is tied to rsc^, which is in turn determined in the money market. Since rlc features more prominently than rsc in the real sector of the model, the Stewart model may be underestimating the impact effect of the Conversion Loan. On the other hand, 172 the impact on real income of the increase in A considered above does not tell the whole story. Further increases in the long rate and decreases in investment can be expected. However, no attempt was made to carry out simulations using the Stewart model. Instead, use was made of another more recent and far more disaggregated model. SECTION TWO: SI>IULATION RESULTS Two sets of experiments were conducted using the Bank of Canada RDX2 7 model of the Canadian economy. In the first set, the model was asked to hold the composition of the --exogenous -- federal government debt at levels that might have prevailed in the absence of the Conversion Loan — three no-Loan hypotheses were examined. The effects of this "shock" on the endogenous variables were calculated over the following thirty quar ters and compared to the "control" values of these variables; that is the values predicted by the model given that the Conversion Loan in fact occurred. Thus, a measure of the effect of the Loan on endogenous vari ables was derived. This simulation showed that the Loan had very weak effects on all variables, primarily because the RDX2 model leaves very little scope for any possible effects from debt management operations on the level and term structure of interest rates. For this reason it was thought fit to introduce some of the results from chapter three of this thesis into the financial sector RDX2. In that chapter, the effects of the Loan on r<, and r were derived within the context of a portfolio model. When the predicted — "control" — values of r and r are subtracted from the values for these rates obtained by holding the composition of the debt at the hypothesized levels -- the "shock" values — the resulting figures 173 give an indication of the effects of the Loan on interest rates. The shock-control values for interest rates are then incorporated in the equa tions for such rates in RDX2 and their effects on the rest of the system are traced out through simulation. These simulations indicate that the Conversion Loan had quite substantial effects on real variables. It is note worthy that in the two sets of simulations the exchange rate was assumed to remain flexible throughout the simulation period. The purpose of this procedure was to avoid imposing upon the model shocks additional to the hypothesized no-Loan ones, such as a structural change of the foreign exchange market. In fact, the control solution for the flexible exchange rate tracks the history of the pegged rate extremely well until 1966Q4 -— this is one reason why we only report simulation results till 1965Q4. Before giving a detailed account of the results it is necessary to briefly remind the reader of the changes in the composition of the debt that the Conversion Loan brought about and to speculate on what would have happened to it in the absence of the Loan— this is an essential element of counterfactual methodology. It will be recalled that the com position of direct and guaranteed debt held by non-governmental agencies is not available and so the quantities held by the Public are used instead — see chapter one, pp. 6-7. Graphs one to four show the composition of the federal government debt held by the Public between 1958Q1-1961Q4. Looking at Graph 4, it is clear that the Loan simply increased the number of bonds in the over 10 year category — for the moment ignore all but the solid lines. Graph 3 tells a similar story. Following 1958Q3, there is no appreciable change in the 174 value of bonds in this category until 1960Q3, at which time bonds worth approximately $500 million were reclassified into the 2-5 year category. This change is, of course, reflected in Graph 2 which also shows a small increase in bonds with 2-5 years to maturity during 1959Q4-1960Q3. The pic ture with bonds under 2 years to maturity is far more complicated. Follow ing the Conversion Loan there was a decrease of bonds in this category. Chapter one showed that this decrease was not nearly as great as would have occurred had the Bank of Canada not sold short bonds — in order to pur chase those long bonds which the Public did not wish to hold at 1958Q3 interest rates. Beyond 1958Q3, bonds in this category increased. In 1958Q4 and 1959Q1 they increased for two reasons. To begin with, the Bank g was still reducing its holdings of bonds in this category. Secondly, the total value of 0-2 year bonds was also increased. In 1959Q2, the Bank began increasing its holdings of these bonds, but the larger increase in the totals outstanding raised the value of bonds held by the Public. For the remainder of 1959, changes in Bank holdings and in the totals outstand ing just about cancel each other. Beyond 1960Q1, bonds in this category decreased.^ It is now necessary to speculate on the alternative' course of history, assuming that the Conversion Loan-did not occur. In particular, how would the public debt have behaved in the absence of the Loan? Three possibili ties are considered: First No-Loan Hypothesis (NLH1). It is assumed here that in the absence of the Loan the four debt categories would have behaved as they did his torically plus a constant adjustment for the shock imposed by the Loan. Since the Loan affected the four categories differently, the adjustments 175 $M Federal Government Debt Held by the Public (Table 12, Chapter One) Craph 1 0-2 Years to Maturity 1600 1000 400 Graph 2 2-5 Years to Maturity 1600 1000 400 2200 1600 1000 400 Graph 3 5-TJ Years to Maturity * X- X X X X Graph 4 Over 10 Years to Maturity 3500 2900 200C 176 SM Federal Direct Debt held by Resident Public + Chartered Banks (RDX2 Variables LCFRIC) Graph 5 0-3 Years to Maturity 3300 I.GFR1C Graph 6 3-5 Years to Maturity >™ LGFR2C v/cc Graph 8 Over 10 Years Time 177 also vary. They are +$854 million,+$493 million, +$1340 million and -$2180 million, corresponding to the debt categories under 2 years, 2-5 years, 5-10 years and over 10 years to maturity."^ The resulting hypothe tical time series are indicated by the x's in Graphs 1-4, and they tell what the four debt categories might have been if the Conversion Loan and "other" changes had not occurred. "Other" changes include the price support programme of the Bank of Canada, the response of other governmental agencies''"''' and the induced changes in chartered bank portfolios. The re sults of these "other" responses in terms of pressures felt by the Public were to change the Conversion Loan from a pure debt management operation of $3518 million to a decrease in shorts of $2687 million and an increase 12 in longs of $2180 million — a "scale effect" and a "shortening effect". Second No-Loan Hypothesis (NLH2). Had "other" changes been more symmetric in their effects on shorts and longs held by the Public a more pure debt management operation would have been felt "Inside" the system. It is assu med here that longs held by the Public would have increased by $2687 mil lion. The resulting hypothetical time series are exactly the same as in NLH1, except for longs — indicated by the z's in Graph 4. They tell what the debt composition would have been like had a debt management operation of $2687 million been implemented. This hypothesis eliminates the "shor tening effect" that "other" changes brought about. Since the authorities wished to preserve orderly markets, I did not attempt to examine the hypo thesis that in the absence of the Loan shorts and longs would have been higher and lower respectively by the full $3518 million. Third No-Loan Hypothesis (NLH3). It is assumed here that without the Con version Loan the debt levels would have continued at their 1958Q2 values. The implied time series are indicated by the broken lines in Graphs 1-4. 178 This is a more naive hypothesis. Three more points must be raised before discussing the results of the simulations. First, the constraints imposed upon government behaviour by the no-Loan hypotheses: It is implicitly assumed that bonds are issued in order to make the various debt categories implied by the NLHl-3 viable. Also, in the context of the RDX2 model, the no-Loan hypotheses imply that Treasury Bills, which are excluded from the shortest category, become the source of any residual finance dictated by the values of the variables in the model. Second, the short rate r in chapter three is the simple aver-age of the RDX2 variables RS, RMS, RML; they correspond to the rates on the three maturity classes 0-3, 3-5, 5-10 years. In the second set of simula tions below, the shock-control values for r calculated from chapter three equations are used for all three RDX2 variables. Clearly this procedure preserves the relationship between r and the three RDX2 variables. Third, there is another problem relating to the difference between variables used in chapter three and the RDX2 model: The relevant data on the composition of the debt used in chapter three were taken from Table 12, chapter one, which excludes chartered bank holdings. The RDX2 series does include chartered bank holdings, but it excludes guaranteed federal issues. The two sets of simulations are now discussed in greater detail. The first set of simulations made use of the RDX2 model only. The model was asked to set the exogenous levels of the four debt categories equal to those suggested by the three hypotheses NLHl-3 and calculate the resulting shock values of endogenous variables. These were then compared to the control solution values thereby giving a measure of the effects of 179 the Conversion Loan on endogenous variables. These effects turned out to be minute. Table 1 reports the shock-control values for nominal GNE (YGNE). The Conversion Loan as specified here had . no impact on the GNE deflator (PGNE) and hence the values reported are effectively in real 13 terms. Using NLH1, for example, the cumulative effect on YGNE by 1961Q4 is $66,730 million — the ensuing contractionary cycle reduces this effect to $18,878 million by 1965Q4. The maximum impact in any one quarter never exceeds one tenth of 1% of real GNE. The results displayed in Table 1 indicate a cyclical response to the shock. There are major cycles — lasting between fourteen and sixteen quarters — each containing smaller cyclical patterns. There is also other evidence indicating that the amplitudes of major cycles beyond 1965Q4 may be increasing. The reason why the results are so negative becomes obvious when we look at the financial sector of the model. The maturity composition of the federal government debt, as distinct from its size and changes in its size, does not feature very prominently in the model. The only place where supply variables are at all important is in equation 17.2 for the long rate, RL. There, the change in the ratio of bonds over ten years to those under 14 three affects the long rate positively. The change brought about by the Loan in this ratio''"'' was 1.5394 and the coefficient being 0.0580, the equa tion predicts that the Loan increased RL by a mere 9 basis points. It is noteworthy that beyond 1958Q3 there is no scope for equation 17.2 to increase the predicted RL through the ratio in question, since this ratio in fact declined. Conducting the same exercise using equations 23 and 25 of chapter 180 three, provides insight into the distinctly different nature of the results in the second set of simulations reported below. Equations 23 and 25 pre dict that the Loan increased the long rate in 1958Q3 by 106 and 103 basis points respectively. The substantive issue is not whether equation 17.2 in RDX2 is in an overall sense better or worse than equations 23 and 25 of chapter three. Rather, the point is that equation 17.2 offers no scope for debt management to affect the level and term structure of interest rates. It was, therefore, thought desirable to incorporate some of the features of equations 22 to 25, chapter three, into the equations for RS, RMS, RML, and RL in RDX2. Equations 22 to 25 were first used to establish what the short and long rates, rg and r^,would have been under the no-Loan hypotheses discussed above. Then the federal debt categories were again held at levels consis tent xvi-th NLHl-3 in order to derive "shock" solutions for the endogenous variables in RDX2. Finally, the intercepts in the equations for RS, RMS, RML, RL were altered so that the shock-control values for these variables were equal to those calculated using equations 22 to 25. In this step, the shock values of interest rates were exogenized. Since rg is the simple average for RS, RMS, and RML, the shock-control values of these variables are all equal. This simulation then answers the question: How would the economy have behaved under a no-Loan hypothesis, if equations 22 and 23, or 24 and 25, correctly estimate the effects of debt management on the level and term structure of interest rates? Equations 22 to 25, chapter three, are reproduced below for the readers convenience: 181 [22] r_ = -2.394 + 0.186Y - 0.257M + 0.091S + 0.073L [23] rT = 1.471 + 0.129Y - 0.174M + 0.008S + 0.057L [24] r_ = 1.983 + 0.129Y- 0.164M + O.OOIS + 0.015L [25] rT = 2.024 + 0.102Y - 0.124M - 0.008S + 0.036L Since there are two equations for each of r and r and three no-Loan O Li hypotheses, six simulations were carried out. The superscript A denotes use of equations 22 and 23 to construct shock-control values for r and r , D Li while superscript B denotes use of equations 24 and 25. The effects of the no-Loan hypotheses (NLH) on the term structure of interest rates are repor ted in Tables 2 and 3. Table 4 shows the effects of the Conversion Loan on YGNE under the six NLH, while Table 5 the effects on UGNE*, where Shock-Control UGNE .. , YGNE UGNE* = ,.«.„• TTP_M17 x 100 , and UGNE •= Control UGNE ' PGNE ' Tables 6-9 and 10-13 report in greater detail the results of two out of the A B six simulations conducted, namely NLH1 and NLH1 . The impact effect of the Loan on GNE given by row 1 of Table 4 is remarkably similar in all simulations, ranging between $41.945-$57.969 million, or 0,391-0.537% of UGNE*-- Table 5. However, over a longer peri od, different results are reported. In terms of their implications for the effects of the Conversion Loan on UGNE*, the A simulations rank as follows: NLH3 > NLH2 > NLHl. The reasons are provided in Tables 2 and 3 which give the impact of the NLH on RS, RMS, RML and RL. NLH2 implies a bigger change in longs than NLHl. Given the coefficients for S and L in equations 22 and 23, NLH2 implies a greater decrease in RL and a smaller increase in the 182 three short rates than NLHl. Hence its effects on real income are larger. Turning to NLH3, although its impact on RL is always smaller than that of NLH2, it has a very different effect on rg: As Graphs 1 and 2 show, NLH3 implies that for long periods of time the Conversion Loan increased the quantities of bonds with less than 5 years to maturity.' When the 5-10 year category is included, NLH3 still posits an increase in shorts, but a much smaller one. This results in a lower rg and is, of course, expan sionary. Turning to the B simulations, the pattern is NLH2 > NLHl > NLH3. Whereas in equation 22 the S coefficient is greater than the L coefficient, the opposite is true in 24. This means that without the Conversion Loan rg would have been lower. Since NLH2 decreases L by more than NLHl does, this source of expansion is stronger in NLH2. This is also the reason why NLH2 -has a greater impact on r than NLHl, despite the fact that the difference between the S and L coefficients in [23] exceeds absolutely that in [25] --they are -0.049 and -0.042 respectively. NLHl has a greater impact on GNE than NLH3 because it lowers r^ more: The negative coefficient on S in [25] reinforces the tendency of to fall under NLHl; but since NLH3 posits a considerably smaller increase in shorts, it yields a milder overall reduc tion in r^. B A Turning to another cross-classification, note that NLHl >NLHl B A and NLH2 > NLH2 . The reason is again the configuration of S and L coeff icients in [22] and [24] . The latter equation implies lower r„ without the Conversion Loan which is expansionary. However, NLH3A > NLH3B. This arises because of the assumed small increase in S: It does not reinforce the tendency under a NLH of r to fall given the -0.008 coefficient on S in [25] ; 183 nor does it counteract the strong negative effect onrg, imparted by the large 0.073 coefficient on L in [22],with the even larger 0.091 coefficient A R on S in the same equation. The simulation results for NLHl and NLHl will now be discussed in greater detail. These are probably the two most inter esting simulations. Tables 6-9 report on NLHl . Table 6 shows that under this hypothesis short rates are higher and RL is lower. The large number of interest rates in RDX2 are interconnected and a decrease in RL decreases the conventional mortgage rate (RMC). The effects on the supply price of capital (RHO) are more complex. Under a NLH lower long rates initially reduce RHO. An intui tive explanation is as follows. Given that the relative supplies of real capital and government debt are unchanged, a shock that reduces RL increases the desirability of real capital in portfolios. The market ensures that the existing stocks of government debt are held by reducing RHO. The reduction in RHO is checked and, after 1961, reversed by the increase in corporate profits, inflationary expectations and the rise in the market value of capital assets brought about by increased economic activity under NLH3^. The supply price of capital in real terms (RHOR) declines throughout the simulation period because of the substantial increases in inflationary expec tations (PCPICE) during 1961 - 1964 -- Table 7, column 5. Table 7 displays some of the consequences of exogenizing RS, RMS, RML and RL in the shock simulations. Given the reaction function,^ the shock increase in RS is effected with a reduction in chartered bank personal (ABLP) and business and miscellaneous general loans (ABLB), which is in turn caused 184 by a reduction in Bank of Canada deposits held by chartered banks (ABBCD) --Table 7, columns 1 and 3, 1958 and parts of 1959. As time elapses, the need for tight monetary policy is obviated by the growth in government debt, the increases in the consumer price index (PCPI) and the growth of RS itself. In fact, after 1959Q2 ABBCD and hence ABLB increase. The credit availabi lity variable, RABEL -- column 2, Table 7 -- behaves somewhat more erratically. The real sector feels the expansionary forces very early -- Table 8. The initial decrease in RHOR stimulates most components of consumption. With the subsequent improvement in incomes, further induced increases in consump tion demand occur until the end of 1963. The most powerful increase in aggregate demand comes from the rise in business investment in machinery, equipment and inventories, and the increases in residential and non-residen tial construction -- their sum is shown in column 2, Table 8. The rise in these demand components is due to the rise in consumption, the decrease in RL, RHO and RMC, the increased credit availability after 1959Q1 and the inc reased loans to business after 1959Q3. Export demand stimulates the economy only moderately but trade as a whole (X-M) is contractionary until the end of 1961. Despite the increase in short rates implicit in NLHl , capital inflows (UBAL-XBAL$) decrease throughout all but a few quarters in the simulation period. The balance of payments surplus (UBAL) decreases until 1961Q3, but despite this the exchange rate (PFX) appreciates slightly during 1958 and 1959. During 1960 and 1961 the increased economic activity maintains imports at a high level, thereby keeping the current account (XBAL$) in the red despite an exchange rate 185 depreciation of as much as 7 cents in 1962Q4. Beyond 1962 the current account surplus stimulates aggregate demand, but it is not long before these injections are swamped by the contractionary cycle that sets in. It is noteworthy that this account of the significance of the openness of the Canadian economy is essentially different from that in the convention al wisdom on the effects of the Loan -- recall that the Loan is felt to have led to a capital account surplus which appreciated the exchange rate, led to a current account deficit, which in turn brought about a depression. Under NLHl^ this argument is valid only during 1962Q1 - 1964Q4. Columns 3 and 4 of Table 8 indicate the effect of the no-Loan hypothesis on exports and imports in real terms. Column 5 in that table gives the effect on gross private real business product. Table 9 documents these effects on trade and capital flows in nominal terms, as well as those on PFX and the 90-day forward rate PFXF. The effects of the expansion in the labour market are summarized by the unemployment rate(RNU)-- column 5, Table 6. It shows that the maximum effect of NLHlA oc curs in 1960Q4, when the unemployment rate is lowered by 2.083%. Figures not shown indicate that gains in employment were secured despite increases in the labour force -- induced by higher wage rates. Average weekly hours worked also increase. After 1961 RHO increases and by 1963Q2 decreases in investment and consumption set in -- columns 1 and 2, Table 8 -- reversing the expansionary 186 cycle. The ensuing cycle is not complete by the end of the simulation period. Here, as in the first set of simulations minor cyclical fluctua tions exist within the major cycles. Tables 10 - 13 document the results from NLH1B. Although the results are in many respects similar, some interesting differences exist. For TJ reasons already indicated, NLHl implies a decrease in short rates -- column 1, Table 10. As can be seen from column 1, Table 11, this assumption about short rates does not call for tight monetary policy and so it increases the expansionary impact of NLHl . Of course this greater effectiveness calls for an earlier increase in RHO -- shock-control RHO becomes positive in 1959Q4 under NLHl . With lower short rates under NLHl , capital inflows are-lower, at least during the early part of the simulation period, and despite a smaller current account deficit the exchange rate depreciates throughout 1958Q3 - 1965Q2. This last observation is even less favourable to conventional wisdom than the analogous one under NLHl . Although the Conversion Loan did attract hot capital and appreciate PFX, its contractionary nature checked the tendency of the current account to be in deficit.1 Remarks made earlier on concerning the cyclical nature of the results apply here too. It is rather unfortunate that the length of the major cycles did not make it possible to get a more precise idea about the stability of the model. It appears unlikely that the length of these cycles is a simple function of the size of the shock imposed: Simulation NLH1B was conducted 187 reducing the shock-control values of RS, RMS, RML and RL to one tenth of what they originally were. The expansionary cycle in YGNE finished in exactly the same quarter -- 1964Q1 -- and the size of the shock-control YGNE values were greater than one tenth of those appearing in column 2, Table 4. SECTION THREE: CONCLUSIONS A: CONCLUSIONS FOR CHAPTER FIVE Several somewhat different estimates of the effects of the Conversion Loan on economic activity have been presented. It is now time to draw some informal tentative conclusions on this score. I will concern myself only with effects on GNP (or GNE), as one proxy for economic welfare , It will be recalled that the following estimates of the effects of the Loan on GNE have been given. For 1958Q3 only, the impact effects predicted by the Stewart model are in the region of $61 - 94 million.^ Turning to the RDX2 model, the first set of simulations, using RDX2 only, yield estim ates around $4 million -- Table 1 -- while the second set of simulations, using RDX2 plus chapter three, yield the range $42 - 58 million -- Table 4. In view of the substantial lags in RDX2 the Stewart range does not appear unreasonable and so the figure of $60 million -- or roughly 0.67* of GNE --188 is chosen. Beyond 1958Q3 the effects become by all accounts stronger. In Table I, the effect very nearly doubles while in Table 4 it ranges around three times A B the impact effect -- the range there is $117 - 163 million. NLHl and NLHl report an average loss in GNE of $126 million. Recalling the back-of-the-envelope long-run effect on I in the Stewart model of $178 million, infuses more credibility to this result. Thus, in the last two quarters of 1958 approximately 17„ of GNE was lost because of the Conversion Loan. Any statements made for the effects of the Conversion Loan beyond 1958 are made with considerable apprehension. In Table 1 the effect of NLHl A V, stays roughly at its 1958Q4 level until 1961. In Table 4, NLHl and NLHl indicate that it increases to about five times its1958Q4 level until at least the end of 1961, declining thereafter. Thus, the loss in YGNE during this A period increases to around 5% in 1961Q1, declining beyond that date -- NLHl , Table 4, is used. A The figures given by NLHl are probably more reasonable than those by NLHl because the latter compounds the effects of the Conversion Loan with those of a monetary expansion -- indicated in column 1, Table 11. whereas A R NLHl permits some monetary expansion this is not as serious as in NLHl , though it still results in some overestimation of the Conversion Loan and "other" changes per se. It should also be remembered that we have only been able to report on part of one of the major cycles that NLH bring about: The cumulative effect of the Loan is not equal to the sum of the positive entries 189 under column 1, Table 4. However, Barber's guess that "An addition to out GNP amounting to several billion dollars has been lost forever. "^ is not outside the realm of possibility. Finally, it should be remembered that the Conversion Loan without the "shortening effect" of the price support programme of the Bank of Canada would have had considerably greater effects -- perhaps as high as the Tk indicated by NLH2A,or NLH2B,in 1961Q1? B. CONCLUDING COMMENTS In chapter one of this thesis the problem at hand was extensively discussed. Chapter two surveyed existing literature on the problem and found that the following questions had, in some cases, not been posed and certainly not answered. These questions were: i) Did the Loan significantly increase interest rates? ii) Did the Loan alter the term structure of interest rates? If so, what are the determinants of the term structure? iii) Following changes in interest rates, did borrowers such as provinces municipalities and corporations change their issuing patterns in an attempt to minimize costs? iv) Was the Loan contractionary? v). How much GNP was "lost forever"? vi) If the Conversion Loan was contractionary, what were the channels 190 through which this was brought about? The answers to them were given in chapters 3, 4 and 5 and are as follows: i) When time trends have been accounted for, there is evidence that long rates rose and some evidence that short ones declined. ii) Hence, the Loan did affect the term structure. There is unquest ionable evidence that the composition of the federal government debt affects the term structure of interest rates. Other determinants are expectations, monetary policy and the behaviour of chartered banks, transactions require ments, private sector wealth and the U.S. term structure. iii) Although the proportion of new short issues to new long ones and also the proportion of total new issues made in U.S. funds do vary, little success must be reported on attempts to determine just how these ratios vary. These ratios have also defied several other investigators, iv) The Loan was certainly contractionary, v) The effects of the Loan on GNE during 1958 are estimated at 1% of GNE, increasing to possibly 5% during 1959 to 1961, decreasing thereafter. The cumulative contractionary impact on YGNE by 1964Q4 exceeds $10 billion, but the expansionary part of the first cycle decreases this figure. vi) Contrary to conventional wisdom, the Loan was contractionary not so much because it affected our trading position, but because high interest rates discouraged investment. vii) More generally, the Canadian authorities -- unlike the U.S. authori ties -- can "twist" the yield curve. They can also use monetary policy to determine the level of interest rates. Such policies can have real effects. However, it is not so obvious that the authorities can also break the trade off between employment and the balance of payments: The import component of Canadian production is rather substantial. 191 NOTES TO CHAPTER FIVE .1. Stewart's model was chosen for various reasons. First, it is not very disaggregated so that a quick intuitive grasp of how it works can be developed. Second, the specification of the various equations is very much in line with Canadian economic thinking during the 1950's. A good example of this is the specification of the money market, noted below. Third, Stewart publishes his data. 2. See Stewart (61), pp. 163 - 172 and 121 - 131 respectively. 3. See Stewart (61), p. 121, column 1. 4. Throughout this chapter the exchange rate is defined as the amount of Canadian, dollars required to buy one U.S. dollar. 5. That is, consumption (= Cd + Cnd + Cs), investment (= Ibc + Ime + Ire + Iinv-nf), exports (= Xgs) and imports (s Mgs). 6. See Stewart (61), equation 33, p. 115. 7. I wish to gratefully acknowledge John Helliwell's very substantial help in constructing the simulations and John Lester's work at the U.B.C. Computing Centre. 8. See Tables 8 and 13, chapter one. 9. Graphs 5-8 show that the RDX2 variables for the composition of the federal government debt behave almost exactly like those of chapters one and three -- compare Graphs 1-4 with Graphs 5-8. 10. These numbers are suggested by Table 12, chapter one -- compare 1958Q2 with 1958Q3. The corresponding numbers for the RDX2 data are very similar indeed, namely +$784 million, +$402 million,+$1172 million and -$2221 million. 11. See chapter one, p. 5. 12. The real counterpart of these numbers is $27,776 million and $22,484 million respectively. 13. The list of variables at the end of this chapter defines allRDX2 variables mentioned. A more detailed discussion of some interrelationships in RDX2 appears in connection with the second set of simulations below. 192 14. The coefficient on this variable is not significant at the 5% level. 15. The RDX2 data series for the various debt categories were used here. 16. Equation 17.1 in RDX2. 17. In the Stewart model price indeces are used with 1957 as the base year. However, price indeces in the RDX2 model use 1961 as the base year. In order to make the GNP figures comparable, the impact effects given in section one -- i.e. $57,429 million and $88,347 million -- were multi plied by ^QQ'Q . The value 106.6 corresponds to the average value during 1961 of the GNP deflator (pgnp) in the Stewart model -- see Stewart (61), p. 127. 18. See Barber (4), p. 3. 193 RDX2 VARIABLES REFERRED TO IN THIS CHAPTER ABBCD = Bank of Canada deposits held by chartered banks -- millions of current dollars. ABLB = Chartered bank business and miscellaneous general loans -- millions of current dollars. ABLP = Chartered bank personal loans -- millions of current dollars. C = CNDSD + CS + CMV + CDO, where CNDSD = Consumer expenditure on non-durables and semi-durables --millions of 1961 dollars. CS = Consumer expenditure on services -- millions of 1961 dollars. CMV = Consumer expenditure on motor vehicles and parts -- millions of 1961 dollars. CDO = Consumer expenditure on durables, excluding CMV -- millions of 1961 dollars. I = IME + INRC + IRC IME = Business investment in machinery and equipment -- millions of 1961 dollars. INRC = Business investment in non-residential construction --millions of 1961 dollars. IRC = Business investment in residential construction -- millions of 1961 dollars. IIB = Change in non-farm business inventories -- millions of 1961 dollars. M = Imports of goods and services -- millions of 1961 dollars. PCPI = The consumer price index -- 1961 = 1.00. PCPICE = Expected annual rate of change in PCPI. PFX = Spot exchange rate -- Canadian dollars per $1 U.S. 194 RPX2 VARIABLES (CONTINUED) PFXF = 90-day forward exchange rate -- Canadian dollars per $1 U.S. PGNE = Price deflator for gross national expenditure -- 1961 = 1.00. RABEL = Earning liquid asset ratio of chartered banks. RHO = An approximation to the nominal supply price of capital -- % per annum. RHOR = RHO - PCPICE. RL = Average yield on Government of Canada bonds, over 10 years -- % per annum. RMC = Conventional mortgage rate -- 7o per annum. RML = Average yield on Government of Canada bonds, 5-10 years -- °L per  annum. RMS = Average yield on Government of Canada bonds, 3-5 years -- % per  annum. RNU = The unemployment rate -- °/0. RS = Average yield on Government of Canada bonds, 0-3 years -- °L per annum. UBAL = Net balance of payments on current and long-term capital account --millions of current Canadian dollars. UGNE = Gross national expenditure -- millions of 1961 dollars. UGPP = Gross private business product, excluding agriculture and non commercial services -- millions of 1961 dollars. X = Exports of goods and services -- millions of 1961 dollars. XBAL$ = Net balance on current account -- millions of current Canadian dollars. YGNE = Gross national expenditure -- millions of current dollars. TABLE 1 Effects of the Conversion Loan on YGNE Shock-Control Values (i.e. No Loan minus Conversion Loan Values) NLHl NLH2 NLH3 1958Q3 3.715 4.191 3.719 QA 6.176 7.039 5.898 1959Q1 6.254 7.238 5.504 Q2 7.023 8.203 5.852 Q3 5.734 6.828 4.270 Q4 5.934 7.145 4.441 1960Q1 5.703 6.953 4.379 Q2 5.211 6.383 4.129 Q3 6.070 7.297 5.227 Q4 5.148 6.258 3.996 1961Q1 3.574 4.426 2.477 Q2 2.758 3.488 1.410 Q3 2.141 2.793 0.484 Q4 1.289 1.855 -0.461 1962Q1 -0.109 0.188 -1.984 Q2 -0.539 -0.297 -2.383 Q3 -0.953 -0.730 -3.098 Q4 -2.082 -2.012 -4.457 1963Q1 -2.395 -2.566 -3.637 Q2 -3.348 -3.703 -4.344 Q3 -3.664 -4.199 -3.867 Q4 -4.582 -5.324 -4.562 1964Q1 -4.434 -5.227 -3.965 Q2 -4.785 -5.754 -3.875 Q3 -5.379 -6.512 -4.555 Q4 -5.098 -6.293 -3.859 1965Q1 -3.863 -4-914 -2.336 Q2 -3.195 -4.234 -1.191 Q3 -2.367 -3.352 -0.121 Q4 -1.059 -1.910 1.426 196 TABLE 2 Shock-Control Values for RS, RMS, RML  Generated by Equations 22-25 NLH1A NLH1B NLH2A NLH2B NLH3A NLH3B 1958Q3 0.886 -0.309 0.500 -0.389 0.886 -0.309 Q4 0.886 -0.309 0.500 -0.389 0.533 -0.305 1959Q1 0.886 -0.309 0.500 -0.389 0.078 -0.317 Q2 0.886 -0.309 0.500 -0.389 -0.251 -0.329 Q3 0.886 -0.309 0.500 -0.389 -0.394 -0.324 Q4 0.886 -0.309 0.500 -0.389 -0.394 -0.317 1960Q1 0.886 -0.309 0.500 -0.389 -0.723 -0.338 Q2 0.886 -0.309 0.500 -0.389 -0.731 -0.352 Q3 0.886 -0.309 0.500 -0.389 -0.557 -0.351 Q4 0.886 -0.309 0.500 -0.389 -0.413 -0.366 1961Q1 0.886 -0.309 0.500 -0.389 -0.497 -0.365 Q2 0.886 -0.309 0.500 -0.389 -0.515 -0.365 Q3 0.886 -0.309 0.500 -0.389 -0.518 -0.355 Q4 0.886 -0.309 0.500 -0.389 -0.352 -0.347 1962Q1 0.886 -0.309 0.500 -0.389 -0.399 -0.344 Q2 0.886 -0.309 0.500 -0.389 -0.551 -0.352 Q3 0.886 -0.309 0.500 -0.389 -1.245 -0.286 Q4 0.886 -0.309 0.500 -0.389 -0.707 -0.336 1963Q1 0.886 -0.309 0.500 -0.389 -0.745 -0.349 Q2 0.886 -0.309 0.500 -0.389 -0.681 -0.361 Q3 0.886 -0.309 0.500 -0.389 -0.458 -0.340 Q4 0.886 -0.309 0.500 -0.389 -0.455 -0.337 1964Q1 0.886 -0.309 0.500 -0.389 -0.477 -0.326 Q2 0.886 -0.309 0.500 -0.389 -0.551 -0.333 Q3 0.886 -0.309 0.500 -0.389 -0.592 -0.330 Q4 0.886 -0.309 0.500 -0.389 -0.411 -0.316 1965Q1 0.886 -0.309 0.500 -0.389 -0.366 -0.308 Q2 0.886 -0.309 0.500 -0.389 -0.126 -0.293 Q3 0.886 -0.309 0.500 -0.389 -0.160 -0.293 Q4 0.886 -0.309 0.500 -0.389 -0.075 -0.248 A: Equations 22 and 23 were used for shock-control r and r . B: Equations 24 and 25 were used for shock-control r„ and r. 197 TABLE 3 Shock-Control Values for RL Generated by Equations 22-25 NLH1A NLH1B NLH2A NLH2B NLH3A NLH3B 1958Q3 -1.059 -1.032 -1.361 -1.222 -1.059 -1.032 Q4 -1.059 -1.032 -1.361 -1.222 -1.062 -0.977 1959Q1 -1.059 -1.032 -1.361 -1.222 -1.124 -0.955 Q2 -1.059 -1.032 -1.361 -1.222 -1.186 -0.954 Q3 -1.059 -1.032 -1.361 -1.222 -1.173 -0.920 Q4 -1.059 -1.032 -1.361 -1.222 -1.150 -0.901 1960Q1 -1.059 -1.032 -1.361 -1.222 -1.240 -0.923 Q2 -1.059 -1.032 -1.361 -1.222 -1.291 -0.964 Q3 -1.059 -1.032 -1.361 -1.222 -1.278 -0.982 Q4 -1.059 -1.032 -1.361 -1.222 -1.323 -1.043 1961Q1 -1.059 -1.032 -1.361 -1.222 -1.325 -1.031 Q2 -1.059 -1.032 -1.361 -1.222 -1.326 -1.028 Q3 -1.059 -1.032 -1.361 -1.222 -1.290 -0.998 Q4 -1.059 -1.032 -1.361 -1.222 -1.254 -0.995 1962Q1 -1.059 -1.032 -1.361 -1.222 -1.247 -0.981 Q2 .. -1,059 -1.032 -1.361 -1.222 -1.281 -0.985 Q3 -1.059 -1.032 -1.361 -1.222 -1.081 -0.705 04 -1.059 -1.032 -1.361 -1.222 -1.231 -0.918 1963Q1 -1.059 -1.032 -1.361 -1.222 -1.280 -0.953 Q2 -1.059 -1.032 -1.361 -1.222 -1.319 -0.996 Q3 -1.059 -1.032 -1.361 -1.222 -1.235 -0.961 Q4 -1.059 -1.032 -1.361 -1.222 -1.222 -0.951 1964Q1 -1.059 -1.032 -1.361 -1.222 -1.185 -0.916 Q2 -1.059 -1.032 -1.361 -1.222 -1.214 -0.929 Q3 -1.059 -1.032 -1.361 -1.222 -1.204 -0.914 Q4 -1.059 -1.032 -1.361 -1.222 -1.146 -0.894 1965Q1 -1.059 -1.032 -1.361 -1.222 -1.115 -0.876 Q2 -1.059 -1.032 -1.361 -1.222 -1.048 -0.859 Q3 -1.059 -1.032 -1.361 -1.222 -1.053 -0.857 Q4 -1.059 -1.032 -1.361 -1.222 -0.887 -0.732 A: Equations 22 and 23 were used for shock-control r and r. s B: Equations 24 and 25 were used for shock-control r„ and r. 198 TABLE 4 Effects of the Conversion Loan on YGNE Shock-Control Values (i.e. No Loan minus Conversion Loan Values) NLH1A NLH1B NLH2A NLH2B NLH3A NLH3B 1958Q3 Q4 41.945 116.527 46.457 135.449 57.969 163.480 55.566 161.547 41.949 118.629 46.461 133.121 1959Q1 Q2 Q3 Q4 187.324 274.719 350.148 421.980 223.598 338.379 450.813 544.004 266.297 397.453 521.285 630.027 266.734 404.016 541.195 654.078 198.789 309.805 426.641 538.035 216.137 323.797 425.062 506.254 1960Q1 Q2 Q3 Q4 453.273 483.324 519.926 570.906 596.258 651.219 722.008 777.297 683.055 739.504 813.551 886.828 716.375 784.426 876.398 948.004 615.848 708.164 828.324 929.105 551.168 599.898 667.484 723.598 1961Q1 Q2 Q3 Q4 555.828 576.125 588.270 595.980 • 751.555 775.918 792.047 775.617 858.590 888.008 905.445 904.855 918.785 953.129 976.406 961.379 936.578 997.645 1049.145 1052.676 704.871 733.832 755.336 741.465 1962Q1 Q2 Q3 Q4 524.668 545.281 554.094 534.645 655.105 659.891 651.215 591.691 780.078 800.926 804.297 756.004 814.699 822.336 812.730 742.539 928.098 956.730 967.473 921.855 630.418 635.695 613.793 544.559 1963Q1 Q2 Q3 Q4 441.801 434.523 409.125 350.012 437.223 395.762 329.543 227.098 588.559 .551.160 485.258 370.426 547.703 493.582 409.684 278.680 744.305 727.949 681.313 540.785 395.055 352.988 295.441 199.332 1964Q1 Q2 Q3 Q4 254.047 210.797 143.680 68.883 83.953 -8.273 -124.051 -225.074 204.285 108.883 - 16.117 -129.824 93.945 -23.957 -169.973 -291.695 314.641 186.703 24.730 -142.246 62.480 -25.984 -145.418 -251.027 1965Q1 Q2 Q3 Q4 -9.062 - 69.563 -137.121 -194.809 -307.652 -403.066 ' -500.895 -549.086 -235.246 -340.957 -453.652 -520.891 -395.066 -514.215 -635.758 -690.066 -310.629 -476.863 -663.605 -812.469 -333.121 -431.805 -535.387 -597.105 A: Equations 22 and B: Equations 24 and 23 were used for shock-control r„ and r . ^ L 25 were used for shock-control r and rT. S >-• TABLE 5 Effects of the Conversion Loan on UGNE* Shock-Control Values (i.e. No Loan minus Conversion Loan Values) NLH1A NLH1B NLH2A NLH2B NLH3A NLH3B 1958Q3 0.391 0.427 0.537 0.510 0.391 0.426 Q4 1.210 1.343 1.663 1.601 1.228 1.320 1959Q1 2.091 2.379 2.908 2.836 2.202 2.297 Q2 2.776 3.251 3.923 3.878 3.084 3.107 Q3 3.028 3.679 4.367 4.399 3.588 3.464 Q4 3.903 4.698 5.601 5.613 4.776 4.363 1960Q1 4.494 5.535 6.506 6.601 5.852 5.102 Q2 4.479 5.642 6.552 6.728 6.267 5.184 Q3 4.007 5.183 5.929 6.189 6.074 4.789 Q4 4.710 5.890 6.846 7.042 7.242 5.489 1961Q1 4.898 6.102 7.070 7.299 7.853 5.750 Q2 4.563 5.629 6.514 6.741 7.534 5.371 Q3 4.031 4.933 5.688 5.909 6.853 4.773 Q4 4.080 4.652 5.559 5.587 6.719 4.536 1962Q1 3.702 3.935 4.861 4.730 6.044 3.886 Q2 3.402 3.390 4.341 4.072 5.426 3.368 Q3 2.788 2.594 3.447 3.104 4.320 2.486 Q4 2.641 1.986 2.968 2.362 3.791 1.796 1963Q1 2.163 1.074 2.032 1.216 2.818 0.891 Q2 1.775 0.451 1.316 0.413 2.124 0.290 Q3 1.256 -0.128 0.557 -0.315 1.319 -0.197 Q4 0.787 -1.087 -0.501 -1.592 0.039 -1.098 1964Q1 0.322 -1.840 -1.379 -2.590 -1.224 -1.822 Q2 0.012 -2.280 -1.926 -3.175 -2.056 -2.223 Q3 -0.305 -2.559 -2.336 -3.539 -2.706 -2.521 Q4 -0.808 -3.334 -3.247 -4.558 -4.079 -3.307 1965Q1 -1.069 -3.595 -3.598 -4.896 -4.828 -3.594 Q2 -1.287 -3.793 -3.887 -5.154 -5.434 -3.804 Q3 -1.390 -3.683 -3.863 -4.994 -5.614 -3.725 Q4 -1.795 -4.110 -4.458 -4.188 -6.827 -4.220 A: Equations 22 and 23 were used for shock-control rq and r B: Equations 24 and 25 were used for shock-control rc and r 200 TABLE 6 Conversion Loan Effects Under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) RS, RMS, RML RL RMC RHO RNU 1958Q3 0.886 -1.059 -0.127 -0.872 -0.084 Q4 0.886 -1.059 -0.233 -0.802 -0.296 1959Q1 0.886 -1.059 -0.313 -0.706 -0.578 Q2 0.886 -1.059 -0.374 -0.582 -0.936 Q3 0.886 -1.059 -0.424 -0.443 -1.253 Q4 0.886 -1.059 -0.461 -0.300 -1.573 1960Q1 0.886 -1.059 -0.490 -0.190 -1.808 Q2 0.886 -1.059 -0.507 -0.107 -1.990 Q3 0.886 -1.059 -0.509 -0.047 -2.049 Q4 0.886 -1.059 -0.503 -0.015 -2.083 1961Q1 0.886 -1.059 -0.495 -0.010 -2.016 Q2 0.886 -1.059 -0.482 -0.020 -1.887 Q3 0.886 -1.059 -0.468 -0.029 -1.676 Q4 0.886 -1.059 -0.458 -0.011 -1.394 1962Q1 0.886 -1.059 -0.452 0.008 -1.041 Q2 0.886 -1.059 -0.447 0.062 -0.675 Q3 0.886 -1.059 -0.448 0.138 -0.340 Q4 0.886 -1.059 -0.452 0.204 0.061 1963Q1 0.886 -1.059 -0.459 0.230 0.438 Q2 0.886 -1.059 -0.468 0.211 0.826 Q3 0.886 -1.059 -0.480 0.143 1.156 Q4 0.886 -1.059 -0.494 0.069 1.524 1964Q1 0.886 -1.059 -0.512 0.018 1.728 Q2 0.886 -1.059 -0.527 -0.035 1.910 Q3 0.886 -1.059 -0.545 -0.105 1.993 Q4 0.886 -1.059 -0.564 -0.203 2.107 1965Q1 0.886 -1.059 -0.584 -0.298 2.030 02 0.886 -1.059 -0.602 -0.424 1.940 Q3 0.886 -1.059 -0.621 -0.574 1.823 Q4 0.886 -1.059 -0.640 -0.717 1.729 NLHl : NLHl when equations 22 and 23 are used. Conversion Loan Effects Under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) ABBCD RABEL ABLB PCPI PCPICE 1958Q3 -3.323 -0.544 -3.128 0.000 -0.002 Q4 -6.645 -0.483 -10.712 0.000 -0.006 1959Q1 -6.347 -0.209 -17.368 -0.001 -0.012 Q2 -3.958 0.153 -19.197 -0.001 -0.015 Q3 0.686 0.271 -11.673 0.000 -0.007 Q4 3.960 0.337 5.725 0.001 0.014 1960Q1 8.261 0.216 36.036 0.002 0.046 Q2 7.283 -0.478 84.691 0.004 0.084 Q3 4.587 -0.683 114.824 0.006 0.123 QA 8.738 -0.442 122.426 0.009 0.155 1961Q1 10.172 -0.560 122.911 0.011 0.184 Q2 7.602 -0.629 123.734 0.014 0.212 Q3 8.609 -0.478 117.953 0.016 0.243 Q4 10.092 -0.289 106.738 0.018 0.289 1962Q1 9.083 -0.158 100.559 0.020 0.354 Q2 15.406 0.256 97.723 0.022 0.433 Q3 21.205 0.405 98.508 0.024 0.515 Q4 22.755 0.305 100.930 0.025 0.590 1963Q1 19.357 0.128 111.371 0.026 0.648 02 21.947 -0.062 115.555 0.027 0.682 Q3 17.651 -0.109 108.543 0.027 0.685 Q4 15.166 -0.201 90.844 0.026 0.655 1964Q1 7.819 -0.354 71.984 0.026 0.596 Q2 2.919 -0.423 .40.555 0.025 0.515 Q3 -7.301 -0.649 -0.141 0.023 0.416 Q4 -18.532 -0.814 -52.168 0.021 0.307 1965Q1 -25.412 -0.768 -106.613 0.018 0.194 02 -29.650 -0.536 -183.055 0. 015 0.082 Q3 -42.613 -0.742 -258.137 0.013 -0.024 Q4 -52.557 -0.627 -329.937 0.010 -0.125 NLHl : NLHl when equations 22 and 23 are used. i 202 TABLE 8 Conversion Loan Effects under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) C I + IIB X M UGPP 1958Q3 Q4 -25.692 37.241 27.737 68.791 -0.656 -4.390 4.823 31.741 38.059 109.844 1959Q1 Q2 Q3 Q4 80.811 138.092 161.816 202.844 131.212 185.953 241.560 281.722 -8.339 -14.307 -20.487 -17.199 55.151 90.802 108.021 123.248 174.672 249.992 310.656 366.195 1960Q1 Q2 Q3 Q4 186.920 201.757 195.920 224.665 322.678 324.312 308.366 305.614 -13.117 -10.049 -0.203 7.164 114.725 115.036 94.969 102.444 387.711 405.191 418.379 442.410 1961Q1 Q2 Q3 Q4 197.493 208.799 189.400 200.573 290.257 254.730 213.886 186.567 16.416 29.100 51.270 45.404 79.575 70.222 48.957 48.625 425.867 423.949 410.973 390.059 1962Q1 Q2 Q3 Q4 150.709 140.755 112.584 102.685 148.440 112.013 83.976 61.217 46.149 61.691 85.809 66.540 16.295 1.301 -9.183 -16.781 332.684 324.965 309.570 264.559 1963Q1 Q2 Q3 Q4 51.502 32.465 16.179 -0.043 39.922 -5.142 -38.744 -72.163 58.922 75.098 96.406 72.113 -40.722 -58.908 -55.837 -73.629 200.195 174.137 141.633 74.188 1964Q1 Q2 Q3 Q4 -36.408 -59.381 -77.580 -112.683 -88.930 -140.088 -170.780 -188.882 61.610 74.542 89.166 65.728 -86.715 -108.007 -99.962 -118.345 19.148 -15.172 -56.410 -114.191 1965Q1 02 Q3 Q4 -134.067 -156.844 -149.031 -166.789 -183.242 -217.296 -241.398 -254.517 52.370 61.491 69.830 52.148 -125.983 -138.244 -119.538 -138.962 -138.805 -173.473 -207.008 -242.957 NLHlA: NLHl when equations 22 and 23 are used. TABLE 9 Conversion Loan Effects Under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) UBAL XBAL$ UBAL-XBAL$ PFX PFXF 1958Q3 -109.367 -3.212 -106.155 -0.007 -0.005 QA -133.935 -32.469 -101.466 -0.011 -0.010 1959Q1 -157.626 -58.851 -98.775 -0.013 -0.012 Q2 -188.241 -97.629 -90.612 -0.012 -0.011 Q3 -202.397 -120.338 -82.059 -0.007 -0.007 Q4 -206.330 -133.288 -73.042 -0.001 -0.001 1960Q1 -187.784 -125.337 -62.447 0.007 0.007 Q2 -162.265 -125.501 -36.764 0.016 0.017 Q3 -123.229 -94.340 -28.889 0.026 0.026 Q4 -117.096 -93.826 -23.270 0.035 0.036 1961Q1 -75.849 -65.788 -10.061 0.044 0.045 Q2 -36.413 -45.876 9.463 0.051 0.053 Q3 18.473 1.470 17.003 0.057 0.059 Q4 8.784 -1.265 10.049 0.063 0.064 1962Q1 32.765 29.740 3.025 0.066 0.068 Q2 58.477 61.855 -3.378 0.068 0.070 Q3 65.166 100.635 -35.469 0.069 0.072 Q4 64.663 89.485 -24.822 0.069 0.071 1963Q1 54.503 101.627 -47.124 0.068 0.071 Q2 89.862 138.361 -48.499 0.066 0.069 . Q3 92.676 162.098 -69.422 0.064 0.066 Q4 76.231 156.506 -80.275 0.060 0.063 1964Q1 72.516 158.836 -86.320 0.057 0.059 Q2 98.864 195.659 -96.795 0.052 0.055 Q3 112.112 205.723 -93.611 0.047 0.050 Q4 52.365 198.050 -145.685 0.043 0.045 1965Q1 60.432 191.695 -131.263 0.038 0.041 02 55.036 213.691 -158.655 0.034 0.036 Q3 54.513 205.542 -151.029 0.030 0.032 Q4 36.301 204.166 -167.869 0.026 0.028 NLHlA: NLHl when equations 22 and 23 are used. 204 TABLE 10 Conversion Loan Effects Under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) RS, RMS, RML RL RMC RHO RNU 1958Q3 Q4 -0.309 -0.309 -1.032 -1.032 -0.133 -0.246 -0.813 -0.689 -0.101 -0.354 1959Q1 Q2 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.333 -0.396 -0.435 -0.457 -0.527 -0.319 -0.104 0.102 -0.695 -1.139 -1.557 -1.952 1960Q1 Q2 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.472 -0.477 -0.470 -0.459 0.259 0.375 0.458 0.506 -2.255 -2.500 -2.605 -2.616 1961Q1 Q2 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.447 -0.430 -0.414 -0.401 0.519 0.507 0.485 0.483 -2.497 -2.304 -2.019 -1.602 1962Q1 Q2 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.392 -0.383 -0.383 -0.389 0.470 0.488 0.520 0.539 -1.090 -0.556 -0.075 0.522 1963Q1 Q2 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.398 -0.409 -0.427 -0.450 0.506 0.410 0.244 0.084 1.082 1.660 2.144 2.675 1964Q1 Q2 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.477 -0.502 -0.532 -0.565 -0.048 -0.192 -0.364 -0.558 2.973 3.247 3.382 3.523 1965Q1 02 Q3 Q4 -0.309 -0.309 -0.309 -0.309 -1.032 -1.032 -1.032 -1.032 -0.596 -0.624 -0.655 -0.686 -0.738 -0.937 -1.142 -1.324 3.365 3.192 2.976 2.730 NLHl : NLHl when equations 24 and 25 are used. 205 TABLE 11 Conversion Loan Effects under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) ABBCD RABEL ABLB PCPI PCPICE 1958Q3 6.677 0.677 0.992 o:ooo 0.001 QA 18.619 1.262 7.625 0.000 0.004 1959Q1 28.385 1.656 29.787 0.000 0.009 Q2 42.604 2.053 74.131 0.001 0.021 Q3 54.882 2.213 140.820 0.002 0.046 QA 66.885 2.068 227.043 0.004 0.082 1960Q1 75.043 1.851 331.336 0.006 0.126 Q2 80.325 0.461 471.085 0.009 0.176 Q3 82.569 -0.230 576.777 0.013 0.233 QA 88.523 -0.523 632.664 0.017 0.290 1961Q1 87.992 -0.809 658.924 0.020 0.352 Q2 86.467 -1.193 689.019 0.024 0.416 Q3 85.872 -1.365 682.059 0.028 0.483 QA 83.438 -1.048 645.527 0.031 0.560 1962Q1 75.547 -1.113 608.480 0.034 0.649 Q2 79.094 -0.830 588.941 0.036 0.743 Q3 77.729 -0.456 557.824 0.038 0.830 Q4 70.222 -0.661 514.527 0.039 0.904 1963Q1 64.856 -0.699 481.824 0.040 0.954 Q2 62.539 -0.607 450.832 0.040 0.974 Q3 56.524 -0.578 406.355 0.040 0.953 QA 50.166 -0.572 350.914 0.039 0.888 1964Q1 39.115 -0.290 300.801 0.037 0.784 Q2 36.503 -0.195 243.086 0.035 0.647 Q3 25.246 -0.604 187.977 0.031 0.488 QA 9>465 -0.775 118.402 0.027 0.315 1965Q1 -0.036 -0.682 46.992 0.023 0.136 02 -1.824 -0.321 -45.680 0.019 -0.041 Q3 -18.382 -0.566 -130.137 0.014 -0.209 QA -26.153 -0.484 -211.277 0.010 -0.368 NLHl*5: NLHl when equations 24 and 25 are used. 206 TABLE 12 B Conversion Loan Effects under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) I + IIB M UGPP 1958Q3 Q4 1959Q1 Q2 Q3 Q4 1960Q1 Q2 Q3 Q4 1961Q1 Q2 Q3 Q4 1962Q1 Q2 Q3 Q4 1963Q1 Q2 Q3 Q4 1964Q1 Q2 Q3 Q4 1965Q1 02 Q3 Q4 -25.958 35.940 74.770 128.011 150.890 190.994 180.065 197.621 197.350 229.035 209.209 220.586 196.822 200.703 .144.480 121.942 81.680 54.419 -7.400 -44.558 -66.656 -103.904 -142.729 -182.502 -198.092 -251.703 -261.703 -290.414 -267.498 -291.191 25.718 66.260 135.808 196.705 264.133 325.327 380.791 384.401 370.945 382.090 359.704 304.983 247.330 212.524 146.118 85.576 42.312 1.797 -43.030 -107.556 -157.183 -209.273 -238.682 -304.705 -346.741 -374.724 -361.408 -394.226 -420.444 -427.901 0.591 3.166 5.873 11.392 21.795 22.782 27.234 40.463 65.354 55.420 56.249 76.130 108.037 80.944 71.778 90.478 118.234 82.792 67.484 82.788 100.772 69.057 54.411 . 62.403 67.897 46.244 31.184 34.399 32.153 23.666 -0.779 22.590 41.223 121.105 44. 379 197. 629 74. 086 290. 719 95. 511 374. 934 121. 264 437. 082 115. 745 473. 242 115. 307 506. 516 100. 830 537. 219 119. 878 549. 371 92. 089 527. 184 76. 435 520. 898 51. 156 501.941 52. 046 443. 328 6. 421 353. 148 -18. 171 323. 473 -31.928 287. 836 -45. 689 197. 332 -76. 729 95. 227 101. 854 37. 203 -97. 080 -26. 383 •125. 622 -133.230 136. 941 -207. 840 •163. 967 -274. 078 •150. 216 -344. 000 •178. 097 -415. 063 •179. 089 -421. 598 -188.703 -473. 527 •160. 823 -518. 824 •185. 725 -535 699 NLHl : NLHl when equations 24 and 25 are used. 207 TABLE 13 Conversion Loan Effects under NLHl Shock-Control Values (i.e. No Loan minus Conversion Loan Values) UBAL XBAL$ UBAL - XBAL$ PFX PFXF 1958Q3 -105.126 0.252 -105.378 0.005 0.004 QA -127.177 -20.937 -106.240 0.011 0.010 1959Q1 -148.378 -41.586 -106.792 0.017 0.016 Q2 -167.710 -68.442 -99.268 0.025 0.024 Q3 -173.818 -78.544 -95.274 0.034 0.033 QA -186.879 -102.307 -84.572 0.044 0.042 1960Q1 -163.234 -97.363 -65.871 0.054 0.053 Q2 -117.007 -87.235 -29.772 0.064 0.063 Q3 -53.263 -43.640 -9.623 0.073 0.072 QA -58.119 -67.758 9.639 0.082 0.080 1961Q1 -5.013 -43.796 38.783 0.088 0.087 Q2 61.294 -8.324 69.618 0.092 0.092 Q3 140.092 57.388 82.704 0.094 0.094 QA 110.878 35.651 75.227 0.095 0.095 1962Q1 135.327 70.194 65.133 0.093 0.094 Q2 170.697 117.434 53.263 0.090 0.090 Q3 174.512 163.814 10.698 0.085 0.086 Q4 167.745 143.734 24.011 0.079 0.080 1963Q1 137.786 155.045 -17.259 0.072 0.073 Q2 172.884 199.812 -26.928 0.064 0.065 Q3 156.077 218.019 -61.942 0.056 0.056 QA 131.967 216.226 -84.259 0.047 0.048 1964Q1 111.118 213.111 -101.993 0.038 0.039 Q2 122.572 251.329 -128.757 0.030 0.030 Q3 112.578 242.853 -130.275 0.021 0.021 QA 23.873 245.885 -222.012 0.014 0.014 1965Q1 36.421 233.1-71 -196.750 0.007 0.007 02 0.143 243.800 -243.657 0.002 0.002 Q3 -22.599 210.963 -233.562 -0.002 -0.003 QA -37.237 223.632 -260.869 -0.005 -0.006 NLHl : NLHl when equations 24 and 25 are used. LITERATURE CITED 208 1. Almon, S. "The Distributed Lag Between Capital Appropriations and Expenditures," Econometrica, January 1965, pp. 178-196. 2. Annual Report. (Ottawa: Bank of Canada). 3. Barber, C. L. "Austerity 1962,'" Toronto Daily Star, September 1962. 4. Barber, C. L. The Canadian Economy in Trouble: A Brief to the Royal  Commission on Banking and Finance. Mimeograph, 1962. 5. Binhammer, H. H. "Canada's Foreign Exchange Problems," Kyklos, No. 4, 1964, pp. 636-652. 6. Bond, D. E. and R. A. Shearer. The Economics of the Canadian Finan  cial System: Theory, Policy and Institutions. (Scarborough: Pren tice-Hall of Canada, 1972). 7. Boreham, G., E. Shapiro, E. Solomon, W. L. White. Money and Banking: Analysis and Policy in a Canadian Context. (Holt, Rinehart and Win ston of Canada, 1968). 8. Breton, A. "A Stable Velocity Function for Canada?" Economica, Nov ember 1968, pp. 451-453. 9. ;- Brunner, K. and A. H. Meltzer. "Predicting Velocity: Implications -for Theory and Policy," Journal of Finance, May, 1963, pp. 319-354. 10. Canada. Dominion Bureau of Statistics: Canada Yearbook. Annual. (Ottawa: Queen's Printer). 11. Canada. Dominion Bureau of Statistics: Municipal Government Finance. Occasional (68-204). (Ottawa: Queen's Printer). 12. Canada. Dominion Bureau of Statistics: Provincial Government Finance. Annual (68-209). (Ottawa: Queen's Printer). 13. Caves, R, E. and G. L. Reuber. Capital Transfers and Economic Policy:  Canada, 1951-1962. (Cambridge: Harvard University Press, 1971). 14. Christ, C. F. Econometric Models and Methods. Cowles Foundation for Research in Economics. (John Wiley and Sons Inc., 1966). 15. Conard, J. W. An Introduction to the Theory of Interest. (Berkeley and Los Angeles: University of California Press, 1963). 16. Conard, J. W. The Behaviour of Interest Rates: A Progress Report. National Bureau of Economic Research. (New York: Columbia Univer sity Press, 1966). 17. Conard, J. W. and M. W. Frankena. "The Yield Spread Between New and Seasoned Corporate Bonds, 1952-63," in Essays on Interest Rates, Vol. 209 1, J. M. Guttentag and P. Cagan (eds.), National Bureau of Economic Research, (New York: Columbia University Press, 1969). 18. Culbertson, J. M. "The Term Structure of Interest Rates," Quarterly  Journal of Economics, November 1957, pp. 485-517. 19. Federal Reserve Bulletin. Monthly. (Washington, D. C.: Board of Governors, Federal Reserve System). 20. Feige, E. "Expectations and Adjustments in the Monetary Sector," American Economic Review, Papers and Proceedings, May, 1967, pp. 462-473. 21. Fullerton, D. H. The Bond Market in Canada. (Toronto: The Carswell Co., 1962). 22. Goodhart, C. A. E. "A Stable Velocity Function for Canada? A Note," Economica, August 1969, pp. 314-315. 23. Gordon, H. S. The Economists Versus the Bank of Canada. (Toronto: Ryerson Press, 1961). 24. Hamburger, M. J. "Household Demand for Financial Assets," Economet- rica, January 1968, pp. 97-118. 25. Helleiner, G. K. "Connections Between United States and Canadian Capital Markets, 1952-1960," Yale Economic Essays, Fall 1962, pp. 351-400. 26. Helliwell, J. F., L. H. Officer, H. T. Shapiro and I. A. Stewart. The  Dynamics of RDX1. Bank of Canada Staff Research Studies, No. 5. (Otta wa: Bank of Canada, 1969). 27. Helliwell, J. F., H. T. Shapiro, G. R. Sparks, I. A. Stewart, F. W. Gorbet, D. R. Stephenson. The Structure of RDX2. Bank of Canada Staff Research Studies, No. 7. (Ottawa: Bank of Canada,. 1971). 28. Hickman, W. B. The Term Structure of Interest Rates: An Exploratory  Analysis. National Bureau of Economic Research. (New York: Columbia University Press, 1943). 29. Hicks, J. R. Value and Capital, 2nd ed. (London: Clarendon Press, 1946). 30. Holland, T. E. "'Operation Twist' and the Movement of Interest Rates and Related Economic Time Series," International Economic Review, October 1969, pp. 260-265. 31. Hood, W. C. Financial Economic Activity in Canada. A Study for the Royal Commission on Canada's Economic Prospects. (Ottawa: Queen's Printer, 1957). 32. International Financial Statistics. Monthly. (Washington, D.C: In ternational Monetary Fund). 210 33. Johnson, H. G. "An Overview of Price Levels, Employment and the U.S. Balance of Payments," Journal of Business, July 1963, pp. 279-289. 34. Johnson, H. G. and J. W. L. Winder. Lags in the Effects of Monetary  Policy in Canada. Working paper prepared for the Royal Commission on Banking and Finance, November 1962. 35. Kesselman, J. "The Role of Speculation in Forward-Rate Determination: The Canadian Flexible Dollar 1953-60," Canadian Journal of Economics, August 1971, pp. 279-298. 36. Keynes, J. M. A Treatise on Money. (London: 1930). 37. Laidler, D. E. W. The Demand for Money: Theories and Evidence. (Scranton: International Textbook Co., 1969). 38. Lee, C. H. "A Stock-Adjustment Analysis of Capital Movements: The United States-Canadian Case," Journal of Political Economy, July/ August 1969, pp. 512-523. 39. Malkiel, B. G. The Term Structure of Interest Rates. (Princeton, N. J.: Princeton University Press, 1966). 40. McCracken, M. C. Massager. (Computel Systems Ltd., 1970.) 41. Meiselman, D. The Term Structure of Interest Rates. (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1962). 42. Meltzer, A. H. "The Demand for Money: The Evidence from the Time Series," Journal of Political Economy, June 1963, pp. 219-246. 43. Miles, P. L. "Some Empirical Evidence of Interest Rate Expectations on Financial Behaviour," Paper presented to the Canadian Economic Association, June 1968. Mimeograph. 44. Modigliani, F. and R. Sutch. "Innovations in Interest Rate Policy," American Economic Review, Papers and Proceedings, May 1966, pp. 178-197. 45. Modigliani, F.and R. Sutch. "Debt Management and the Term Structure of Interest Rates: An Empirical Analysis of Recent Experience," Journal of Political Economy, August 1967, pp. 139-159. 46. Muth, J. F. "Optimal Properties of Exponentially Weighted Forecasts," Journal of the American Statistical Association, June 1960, pp. 299-306. 47. O'Brien, J. W. and G. Lermer. Canadian Money and Banking. 2nd ed. (Toronto: McGraw-Hill Co. of Canada, 1969). 48. Officer, L. H. and L. B. Smith. Canadian Economic Problems and Poli  cies. CToronto: McGraw-Hill Co. of Canada, 1970). 211 49. Okun, A. M. "Monetary Policy, Debt Management and Interest Rates: A Quantitative Appraisal," Stabilization Policies, Commission on Money and Credit. (Englewood Cliffs, N. J.: Prentice-Hall, 1963). 50. Parizeau, J. Debt Management, A Preliminary Report. Study prepared for the Royal Commission on Banking and Finance, 1962. 51. Penner, R. G. "The Inflow of Long-term Capital and the Canadian Busi ness Cycle, 1950-1960," Canadian Journal of Economics and Political  Science, November, 1962, pp. 527-542. 52. Powrie, T. L. "Short-term Capital Movements and the Flexible Exchange Rate," Canadian Journal of Economics and Political Science, February, 1964, pp. 76-94. 53. Roosa, R. V. "Reconciling Internal and External Financial Policies," Journal of Finance, March 1962, pp. 1-16. 54. Ross, M. H. "'Operation Twist': A Mistaken Policy?" Journal of  Political Economy, April 1966, pp. 195-199. 55. Scott, R. H. "Liquidity and the Term Structure of Interest Rates," Quarterly Journal of Economics, February 1965, pp. 135-145. 56. Shearer, R. A. Monetary Policy and the Current Account of the Balance  of International Payments. Working paper for the Royal Commission on Banking and Finance, November 1962. 57. Shearer, R. A. "A Note on Bank of Canada Operations and the Deflation ary Effects of the Conversion Loan of 1958," Mimeograph, 1964. 58. Shearer, R. A. "The Income Velocity of Money in Canada, 1960-68: A Further Comment," Ecoriomica, November 1970, pp. 409-419. 59. Smith, W. Staff Report on Employment, Growth and Price Levels. Con gress of the U.S. Joint Economic Committee (86th Congress, 1st Session, 1959). 60. Statistical Summary Supplement. Annual. (Ottawa: Bank of Canada). 61. Stewart, I. A. "A Quarterly Econometric Model of the Canadian Economy 1951-1962." Unpublished Ph.D. dissertation, Cornell University, 1966. 62. Stoll, H. R. "An Empirical Study of the Forward Exchange Market Under Fixed and Flexible Exchange Rate Systems," Canadian Journal of Econo  mics , February 1968, pp. 56-64. 63. Sutch R. "Expectations, Risk and the Term Structure of Interest Rates," Unpublished Ph.D. dissertation, Massachusetts Institute of Technology, 1968. 64. Taylor, J. H. "Debt Management and the Term Structure of Interest Rates," Journal of Money , Credit and Banking, August 1971, pp. 702-708. 212 65. Tobin, J. "Money, Capital and Other Stores of Value," American  Economic Review, Papers and Proceedings, May 1961, pp. 26-37. 66. Tobin, J. "An Essay on Principles of Debt Management," Research Study 3, Fiscal and Debt Management Policies. Commission on Money and Credit. (Englewood Cliffs, N. J.: Prentice-Hall, 1963), pp.143-218. 67. Tobin, J. "A General Equilibrium Approach to Monetary Theory," Jour  nal of Money, Credit and Banking, February 1969, pp. 15-29. 68. Wonnacott, P. The Height Structure and Significance of Interest Rates. Working Paper for the Hoyal Commission on Banking and Finance,November 1962. 69. Wonnacott, P. The Canadian Dollar, 1948-1962. (Toronto: University of Toronto Press, 1965). 213 APPENDIX TO CHAPTER TWO  STUDIES OF "OPERATION TWIST" The early 1960's found the U.S. with a balance of payments deficit and high unemployment. It was then suggested by President Kennedy's economic advisers that the familiar trade-off could be dodged. This was to be achieved with the aid of "Operation Twist".''- The Federal Reserve was to sell short securities and buy long ones, thereby increasing short and decreasing long rates. The increase in short rates would forestall capital outflows and perhaps reverse them, while leaving inventory invest ment practically unaffected. The decrease in long rates would bring about an increase in long-term and hence overall investment, and, therefore, income and employment. It was also hoped that the increase in long-term investment would, through increased productivity, improve the competitive position of U.S. merchandise abroad. At the risk of repetition it must be emphasized that OT was not an attempt to shift yield curves per se. Its ultimate aim was to avoid the familiar trade-off. In the words of one Government official, 2 "My own thesis is that all these commitments can be met, that they need not, as some would have it, be mutually contradictory; but that with determined effort they can become instead... mutually reinforcing."3 Was OT successful? A decade has elapsed since then. Yet, remarkably, no study has attempted to examine the rather bold claims of those suppor ting it. We still cannot give the old trade-off a decent burial for fear 214 it might still be alive. But let us be more specific. What existing studies do is to try to establish whether the term structure has been twisted. It should be clear that while this issue may logically precede others it is not the end of the road: We still have to know whether twisting the yield curve will, in fact, improve the balance of payments at no cost to domestic employment. Nevertheless, looking at the effects on the term structure is a good starting point. An early paper by Roosa (53) does little more than make the assertions presented in the second paragraph above. Okun's CMC study (49) does not deal with OT explicitly, though it does present quantitative evidence indicating that debt management^ is practically ineffective. His conclu sion has been challenged by Scott (55) who would attribute more importance to it when a more sensitive measure of average maturity is used. Ross (54) argues that by overlooking the interest elasticity of short term, inventory, investment the effects of OT have been exaggerated. Modigliani and Sutch (44) examine the extent to which the term structure has been twisted. They give the following figures: In 1961Q1, the spread between the government long rate and the bill rate was +1.48% and that between AQO corporate bonds 3.3-and the commercial short paper +1.26%. In 1965Q3, by contrast, the former was down to +0.35%. and the latter to +0.12%,. This would appear to be impressive evidence suggesting that OT did twist the yield curve. But, though they do not discuss this, their data show that, aside from the spread, the actual level of both the long-term government bond rate and the Aaa one were higher in 1965Q3 than in 1961Q1. The reason that Modigli ani and Sutch advance for withholding judgment on this score is that, in 215 recovery, such as presumably 1965Q3, the spread usually becomes more narrow. The question then is what part of the decreased spread was due to recovery and what, if any, to OT? This is not an easy question to answer. They advance an hypothesis explaining the spread between the two rates and estimate the functional form that their hypothesis"* suggests, using data prior to the OT period. Then they predict the spread for the OT period and find that, although the actual spread, after OT was initi ated, was always below the computed one^ the difference was not very large. In another equation the authors add a dummy variable that takes on the value of 1 after 1962 to allow for the introduction of negotiable Time Certificates of Deposit.^ This shifts the predicted spread line down so that actual spread is usually above the computed one. In their conclusion it is stated that "The spread between long and short rates in the government market since the inception of OT was on average some twelve base points below what one might infer from the pre-OT relation. This discrepancy seems to be largely attributable to the successive increase in the ceiling rate under Regulation Q which enabled the newly invented CD's to exercise their maximum influence."8 The remaining papers are not as important and can be dealt with briefly. Holland (30) simply runs regressions to "explain" the U.S. government long-term bond yield index and that on three-month TB's, using 1953 - 1961 data. Little justification can be found for the inclusion of particular variables in his equations. Also, although R is generally high, many of the variables are insignificant and some have the wrong signs. Nevertheless these equations are used to predict the two rates for 216 1962 - 1964. The predictions for the long rate are very good, indicating that OT did not succeed in making them lower than they would be, but the predicted short rate lies below the actual one, thus indicating some degree of success for OT. Modigliani and Sutch presented further evidence on the effectiveness of DM in a more recent paper (45). Again using the Preferred Habitat Theory of the term structure they introduce additional independent variables, such as average maturity and the proportion of a particular term in total government debt, in an attempt to evaluate the importance of DM. They argue that they do not expect such measures to be very effective because almost all the variance in the long rate is explained by the current and lagged values of the bill rate. Indeed, they find little evidence substantiating the importance of DM. Malkiel examines, in a 9 thorough manner, the implementation of the project, i.e. the size of Treasury operations and the concurrent activities of the Federal Reserve. It has been argued that the combined activities of those institutions resulted in changes in the maturity composition of the federal debt that were not consistent with declared policy objectives, namely OT.^ Malkiel points out that, while the average term to maturity did indeed increase during OT, there was also a substantial increase in short-term issues --less, than 6 months to maturity.-- outstanding. Thus, the overall effect on the term structure would depend on the relative magnitude of the increase in long and short rates needed to accommodate the increases in both long and very short maturities. This appears to exhaust the studies that deal in a fairly direct manner with OT. 217 NOTES FOR APPENDIX TO CHAPTER TWO 1. Hereafter referred to as OT. 2. These commitments are listed immediately above the quotation and include, among others, balance of payments equilibrium along with a high growth and employment rate. 3. Roosa (53), p. 2. 4. Hereafter referred to as DM. 5. They argue that the term structure is determined according to the Preferred Habitat Hypothesis. The empirical formulation of the model typically takes the form n where Rfc = Long Rate at t; r = Bill Rate at t and n is determined by the data. More on this appears in chapter three. 6. OT is effective. 7. Abbreviated to CD's. 8. Modigliani and Sutch (44), p. 196. 9. See Malkiel (39), pp. 232 - 233. 10. Johnson, for example, writes i t-i + U t "As a result, primarily of Treasury funding operations, the maturity of the debt in public hands has in fact been lengthened appreciably, instead of shortened as the policy would require. it See Johnson (33), p. 286. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0100958/manifest

Comment

Related Items