UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Quantitative analysis of quota trading behaviour at the end of the quota year Sterelyukhin, Alex 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2008_spring_sterelyukhin_alex.pdf [ 2.83MB ]
Metadata
JSON: 24-1.0066744.json
JSON-LD: 24-1.0066744-ld.json
RDF/XML (Pretty): 24-1.0066744-rdf.xml
RDF/JSON: 24-1.0066744-rdf.json
Turtle: 24-1.0066744-turtle.txt
N-Triples: 24-1.0066744-rdf-ntriples.txt
Original Record: 24-1.0066744-source.json
Full Text
24-1.0066744-fulltext.txt
Citation
24-1.0066744.ris

Full Text

QUANTITATIVE ANALYSIS OF QUOTA TRADING BEHAVIOR AT THE END OF THE QUOTA YEAR  by Alex Sterelyukhin  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in The Faculty of Graduate Studies (Agricultural Economics) (THE UNIVERSITY OF BRITISH COLUMBIA) (Vancouver) April 2008  © Alex Sterelyuhkin 2008  ABSTRACT  The Canadian supply management system offers some lessons for the design of a domestic permit trading system. One of the objectives of the domestic trading system is minimizing costs and maximizing the system's efficiency for participants and system administrators. This paper suggests that a permit trading system designed with a longer permit period and without a grace period can be more efficient than a system designed with a shorter permit period and a grace period for compliance. This study is based on Canadian Supply Management System experience and examines the Canadian dairy industry, where two different schemes (monthly and annual) have been used. Under the annual scheme, a strong compliance mechanism stimulates participants to exchange quotas during the dairy year (permit period) and does not require burdensome, noncompliance procedures after the permit period ends. The monthly scheme is characterized by a short permit period and a grace period for compliance. This study examines how these two schemes affect participants' behavior on the quota exchange. The empirical results show evidence of the influence of different schemes on farmers' behaviors regarding the quota exchange. As a conclusion, the paper recommends the use of a permit trading mechanism with a longer permit period and without a grace period for the design of a carbon trading system. The results support Barichello (2002), who developed the Canadian domestic permit trading scheme on the basis of receiving an offset from agricultural soil carbon sinks.  ii  TABLE OF CONTENTS ABSTRACT ^ ...ii TABLE OF CONTENTS ^ iii LIST OF TABLES ^ iv LIST OF FIGURES ^ .v ACKNOWLEDGEMENTS ^ ......vi 1. Introduction ^ .1 1.1 Objectives and proposed solutions .1 1.2 A review of the Canadian dairy supply management system .7 1.3 Differences in marketing board's regulations ^ 8 2. Literature Review ^ 12 2.1 Current pollution permit trading schemes ^ 12 2.2 Analyzing the price and quantity of common stocks on the Stock Market ^ .19 2.3 Canadian supply management studies ^ 20 3. Theoretical Part ^ .21 3.1 Analyzing farmer behavior in dairy quota exchange ^ .21 3.2 Primary market, profit maximization approach ^ .24 3.3 Secondary market, quota capitalization approach ^.27 3.4 Transaction costs of the trading systems and their effect on participant behavior in the market exchange ^ .29 3.5 Suggested Method for Quantitative Quota Analysis at the End of the Year 30 4. Empirical Model ^ 32 4.1 Introduction to the empirical model 32 4.2 Including the transaction cost of quota trading ^ 33 4.3 Problems with the regression model ^ .34 4.4 Omitting penalties from the empirical model ^ 36 4.5 Data ^ 37 4.6 Method of estimation ^ 41 4.7 Results ^ .43 4.8 Discussion ^ .46 4.9 Summarizing research findings 48 5. Conclusion ^ 49 REFERENCES ^ ..51  iii  LIST OF TABLES Table 1 A comparison of European Union's proposed carbon trading system with proposed domestic carbon trading system based on the Canadian supply management scheme ^  3  Table 2 Data summary statistics for the annual scheme, Ontario, August 1980 to July 1994 ^  39  Table 3 Data summary statistics for the monthly scheme, Ontario, August 1997 to May 2005 ^  40  Table 4 Results of estimations ^  43  iv  LIST OF FIGURES Figure 1 A comparison of monthly quantities of TPQ under the monthly scheme and adjusted MSQ exchanged under the annual scheme ^  .10  Figure 2 Monthly dummies for the dairy year for the control and treatment groups, Ontario Aug. 1980 to May 2005 ^  45  ACKNOWLEDGEMENTS  It was a great pleasure for me to work with my supervisor, Rick Barichello. First of all, I would like to express my deep appreciation to him for organizing my education in agricultural economics at UBC. Without a previous background in economics and empirical analysis, it was not easy for me to gain the appropriate knowledge in such a short period—two years. I understand that it was not easy for him to help me obtain those skills. I am very grateful for all his efforts and also for his patience during my study period. In addition, I would like to mention that I was lucky to have Prof. Barichello as my supervisor. As a person, he is very kind and helpful by nature, and I received a lot of advice, not only in my academic field, but in my day-to-day new life in Canada. Also, I am very thankful to him for an opportunity to be a research assistant and to receive funding for that, and I hope my small efforts have helped him develop his research projects.  I also would like to express many thanks to Prof. Sumeet Gulati for his support as part of my supervisory committee and his open attitude and answers for the many questions that I had.  I would like to say many thanks to Prof. Kathy Baylis, who explained many principles in economics, particularly in the beginning of my study, when I desperately needed them. Without the help I received from Prof. Vadim Marmer, I would not have completed the econometric part of this thesis. His course, ECON 527, sealed my knowledge and gave me confidence that I could use econometric tools and methods. I would like to express many thanks to him for all the time he spent with me in his office voluntarily teaching me basic econometric principles.  Finally, I would like to thank our graduate students, Mario Anda, Shinan Kassam, Stephen Peplow, and Andrei Stoyanov, for their support for putting up my work in good order, answering all of my sometimes unusual questions about economics, and for their patience and friendly human attitude.  vi  Chapter 1 Introduction  In 1997, the governments of many developed countries signed the Kyoto Protocol in an attempt to reduce greenhouse gas (GHG) emissions. When the Russian government ratified the Kyoto Protocol on February 16, 2005, participating nations started to proceed with their own domestic systems of greenhouse gas emission reduction. As a part of their systems, many nations considered developing domestic emissions trading schemes prior to implementation of the Protocol. These schemes often included GHG offsets gained by agricultural soil carbon (ASC) sinks. Boehm (2003) defines ASC as an agricultural activity of sequestering atmospheric carbon into the soil. Carbon sequestered by one participating country could be traded to offset emissions in another country in order to meet its international commitments (Marland, McCarl & Schneider, 2001). Canada, a participating country, is currently developing a domestic emissions trading system that includes a domestic offset system (DOS). Although many different schemes have already been suggested, this study is another argument in favor of the research conducted by Barichello (2002), who suggests developing the Canadian domestic permit trading scheme on the basis of getting an offset from agricultural soil carbon sinks using the experience of the Canadian supply management system. Large final emitters (LFE) could buy such offsets from farmers' ASC rather than following the potentially more costly process of actually reducing their emissions. Farmers who practice no-till or any other tillage-limiting production technique can earn extra income by selling offsets that reward growers who sequester carbon in the soil.  1.1 Objectives and Proposed Solutions  When designed properly, a domestic emission permit trading system allows participants to achieve the objectives of environmental emission reduction at minimum cost. In addition, minimizing design costs of the domestic permit trading system is an important issue for policy makers. Most designed permit trading schemes have used a fixed permit period with a grace period for compliance. According to the National Round Table on the Environment and the Economy (NRTEE), Issue Paper 9 (Policies that could complement a domestic emissions trading system for greenhouse gases), Canadian environmental enforcement 1  officials have restricted the selection of enforcement responses for implementation of the domestic emissions trading system. The main enforcement tool is prosecution in the criminal court system, a process that is cumbersome, time consuming, and often inappropriate for minor violations of an emissions trading program. In addition, these regulations require long grace periods to put all participants' accounts in order after the permit period ends and burdens participants with non-compliance procedures. This approach adds delays and does not exploit the possibility that regulations can stimulate participants to manage their accounts without special grace or settling-up periods. One of the alternatives is administrative penalties that may provide very effective incentives for compliance and that can be imposed by government agencies rather than the courts. Moreover, imposed penalties can shorten or even lead to nonexistence of a grace period at the close of the permit period. Ideally, penalties should be non-discriminatory, set at the outset, remain static, and be as simple as possible. Simplicity in an enforcement mechanism of the trading permit system is one of the key factors in reducing system operational costs. Currently suggested proposals for the design of domestic permit trading systems include a long grace period when the permit year ends, which adds operational costs for participants. hi the table below, as an example, the European Union's (EU) proposed carbon trading system is compared with a proposed domestic carbon trading system based on the Canadian supply management scheme.  2  Table 1 A Comparison of European Union's Proposed Carbon Trading with Proposed Domestic Carbon Trading System Based on the Canadian Supply Management Scheme  The proposed European Union carbon trading scheme, with a long grace period (for industry) 90-day grace period for compliance after the permit year ends Penalties for compliance apply after the grace period Allowance of the carbon permit rental Audit and participants' certification needed Behavior of Carbon permit purchase/sales can participants be delayed for 90 days Rules and regulations  May increase likelihood of appeals Follow-up disputes and court actions  The proposed domestic carbon permit trading system based on the Canadian supply management scheme (for agriculture) No grace period for compliance Enforced penalties apply during the permit period Possibility of carbon permit rental/purchase Audit and participants' certification needed Participants are encouraged to put all accounts in order during the permit period Fact: minimum numbers of appeals on over-producing issues No future obligations, after "end of the year," participants are involved only in the "new year's" issues  From the table above, it can be seen that the system with a grace period for non-compliance would increase operational costs of a domestic permit trading system: the system burdens carbon emitters, system administrators, and taxpayers. It would increase processing times for approvals, the time taken to determine compliance, and the time taken to enforce penalties, all causing increased operational costs. Increasing processing times also increase transaction costs. Lower transaction costs may be achieved by eliminating the grace period, which is accompanied by time spent with evaluation and invoicing processes, reconciling transactions, and a never ending exchange of letters between participants.  3  Thus, a carbon permit trading mechanism designed without a grace period would reduce costs for system administrators, trade organizations, and taxpayers. On the other hand, the grace period is a relaxation for participants, the period of time in which participants can gradually overcome the stochastic unforeseen problems during the permit period. Thus the grace period can reduce participants' production costs in a short run. So, to design a domestic emissions permit trading system properly, participant behavior under that system has to be analyzed. Much research has been done in the design of domestic carbon permit trading mechanisms, but some has been related to the analysis of how a permit period's design affects participant behavior. The dilemma is immediately apparent to the system designer: Do we need to design the carbon permit trading system with a strict incentive mechanism and without a grace period to reduce system operational costs, or do we need to design a system that will diminish the negative effect at participant production from an imposed penalty mechanism? Lessons learned from more than 30 years of experience in the Canadian supply management system might help answer that question and build an efficient domestic carbon permit trading system with minimum design costs. Most supply management systems have a specific penalty mechanism for overproduction of the regulated product. The penalties involve a short grace period or none at all and strongly encourage farmers to put their accounts in order by the end of the annual quota period. Mechanisms exist to do this despite random shocks to production levels due, for example, to weather, such as the buying and selling of market quotas. The major principles of the supply management system quota exchange also are similar to current proposed and developed permit trading mechanisms. So, we can find the similarities of farmer behavior under the supply management system and participant behavior under the designed permit trading system. Analysis of participant behavior under the supply management system can reveal the need for a grace period for permit trading participants and answer the question, Should the designed permit trading system have a grace period or not? Consequently, there are some other questions that need to be answered: If the Canadian supply management system can be a guideline for building an efficient permit trading scheme, how does this scheme work? When do farmers trade quotas? How do farmers use 4  the quota market to avoid penalties and get their production equal to their quota allocation? Do different system rules make a difference in their behavior on the quota exchange? The Canadian dairy industry is one sector of the agriculture industry that operates under the supply management system. The Canadian dairy supply management system has some unique features. One of them is the provincial level of administration. Many operational rules in this regime are set by provincial marketing boards and vary accordingly by province. Comparing of different provincial schemes and analysis of results could answer these questions. One of the typical examples is the "annual scheme" with a strict penalty mechanism for milk overproduction and without a grace period. During a quota period, which is one year (it is called a dairy year), farmers have two assessments for matching allocated quota with actual milk production. The penalties strongly encourage farmers to put their accounts in order by the end of the quota period. Given inherent uncertainty in production levels under the annual scheme when the end of the quota year approaches farmers must plan to adjust production or to buy/sell quotas. Sometimes, they can buy quotas as an alternative way of reducing production. In general, if the total quantities of exchanged quotas have increased at the end of the dairy year, suggests the regime stimulates farmers to exchange quotas. This is what has been observed because under the annual scheme, farmers have a lengthy one-year period to react to any stochastic factors and can fix a problem without quota stock adjustments to the latter part of the quota period to match end-of-year production levels. If they do not comply with their quota level and production, they can buy or sell quotas at the end of the dairy year. Thus, the period of complying is an important issue for analysis, because the length of the quota period affects a farmer's decision to buy or sell quotas. Moreover, if a period is long enough, such as with the annual scheme, the grace period might be not needed for farmers. Another quota trade mechanism is the "monthly scheme," where participants have monthly assessment deadlines with a grace period of 20 days for overproducing and 30 days for underproducing. Under the regulations of this scheme a farmer has to adjust his/her quota level to milk production every month. Under the monthly scheme, with 12 quota assessments 5  during the dairy year, farmers have less reaction time to adjust their production to any stochastic changes, and they may go to the quota exchange more often throughout the year to buy or sell quota. Thus, the quantity of quotas exchanged during the entire dairy year could be greater under the monthly scheme compared to the annual scheme. Also, a grace period may be needed for every end-of-the month under the monthly scheme due to the short assessment period (one month). In this case, the scheme should not cause an increase in the quantity of quotas exchanged at the end of the dairy year, because the actual quota period ends every month with a relatively lengthy 20-30-day grace period. Thus, to answer the above questions and understand a farmer's behavior on the quota exchange we can compare two systems - monthly and annual. This comparison will reveal how they affect farmer behavior on the quota exchange and will explain farmers' reactions to different regulations. In addition, we can explain how the length of the quota period affects farmer behavior, and determine the necessity of grace periods after the quota period. Finally, one of the ways of estimating the effect of system regulations is to measure the difference in quantities of quotas exchanged and compare them at the end of the dairy year for both annual and monthly schemes. Consequently, the research question finally arises: "If the end of the year affects a farmer's behavior to buy or sell quotas under the particular supply management system, can we estimate the quantity of quotas exchanged as a result of that effect?" This question raises the importance of understanding how farmers react to an imposed penalty when the permit year ends. Thus, the institutional framework is an important issue in designing an efficient workable permit trading system. This research could show the effectiveness of the rules and help the design of a simple but effective domestic carbon permit trading scheme using lessons from the Canadian supply management system. To understand how the system works in practice, the following section introduces the dairy supply management system.  6  1.2 A Review of the Canadian Dairy Supply Management System  The supply management sector in Canada covers the dairy and poultry sectors,. The dairy supply management system has the power to restrict imports, control domestic supply, and set the domestic price for milk. There are two types of quotas. The first is a farm-level quota, which sets the limit on a farmer's output that can be marketed each year. The farm quota can be considered as a permanent asset, tradable among farms within one province. Another type of the quota is the import quota, which provides quantitative limits on import access across different final consumer products. Since the Uruguay Round Agreement on Agriculture came into effect in 1995, these quotas have become tariff rate quotas (TRQs). Different levels of government agencies are involved in the quota management system. The federal government determines legislation, sets up national regulatory agencies, and is represented on federal—provincial committees. For example, in the dairy industry, the federal government has set up the Canadian Dairy Commission (CDC), which recommends milk production targets, sets the price, and operates the offer-to-purchase price support scheme to underwrite industrial milk prices across the country. The Canadian Milk Supply Management Committee sets aggregate quota levels and allocates provincial quotas to provincial marketing boards. They, in turn, regulate quota levels and prices in each province. Most of the initial quota allocations occurred in mid-1970s, and quotas were distributed gratis to farm producers who were in the industry at that time, based on historical patterns of  production. If growth in consumption occurred, new quota was added at the national level first, then distributed across provinces, and finally granted to individual farms. The mechanism for allocating new quotas to provinces has often been pro rata (equal percentage I increases. If quota transfers are unrestricted and accessible, any newcomer can enter the industry, assuming that a person can finance the quota purchase. Quota transferability also encourages less efficient farmers to leave the industry. It is well known that the high rents generated by increased domestic farm product prices are capitalized into high quota values. The price of a quota in the dairy industry is now $25,000-30,000 per cow, making the cost of purchasing enough quotas for a hundred-cow farm equal to $2.5 million, effectively excluding the dairy market from new entrants. 7  The Canadian supply management system has evolved gradually over time so that quota is now transferable among farmers in almost all provinces via permanent purchases and sales of the asset. At the same time, within-province restrictions that differ by province sometimes still exist. To understand better the difference of provincial regulations, the following section compares Ontario and British Columbia marketing board schemes. 1.3 Differences in Marketing Board's Regulations There are quota measurement differences across provinces and differences in frequency of quota assessments during the dairy year. In British Columbia (BC) the quota is measured in kilograms of butterfat per year, while in Ontario it is measured in kilograms of butterfat per day. In Ontario the quota is divided into monthly levels, and a farmer must adjust production of milk monthly relative to the assigned quota level. In the case of overproduction, a farmer has a grace period of up to 20 days to adjust production to the quota level and 30 days in the situation of underproduction (production less than the monthly quota).. Thus, a farmer can carry forward a small amount of unused quota in case of underproduction or borrow up to 30% of quota from the future in case of overproduction and only for a short period. Therefore, an Ontario farmer meets a quota assessment every month and has to regulate milk production buying or selling quotas on a monthly basis. Conversely, under the BC system, a farmer has two deadlines for quota assessment during the year: on January 31 and at the end of the dairy year on July 31, when a farmer has to match milk production with quota levels. Compared to the Ontario scheme, a farmer under the BC scheme has the longer response period of up to 6 months to react to any stochastic changes in the short run. In other words, the BC farmer does not need to go to the quota exchange every month to adjust the quota level; he or she can try to adjust production first without selling or buying quotas in the current year. The Ontario farmers are encouraged by the regulations to sell or buy quota in the current year. Thus, the Ontario scheme does force a farmer to exchange quotas more frequently, within the year relative to the BC scheme. A similar comparison can be observed within the Ontario scheme. Until 1997 in Ontario, there were two quota exchange markets concerning industrial milk production: one for used quotas and another for unused quotas. The marketing board operated the industry using the 8  used—unused scheme (annual scheme). The main difference between the two types of quotas is that the farmer can use any unused quota in the same dairy year when he or she bought it, while a farmer who bought the used quota could use it only in the next year. For the annual scheme, the Ontario Marketing Board reported data on sales and purchases of marketing shared quota (MSQ). In 1997, the Ontario province, along with other Eastern provinces, introduced the "P5" pooling system, in which they operated the new exchange system (monthly scheme) by measuring the quotas on a daily basis. This scheme eliminated two quota markets (used and unused) and launched a single exchange market for total production quotas (TPQs), which included both fluid and industrial milk quotas. To compare the two schemes visually, the reported data of the MSQ for the annual scheme were converted into the equivalent of total production quotas by multiplying by a coefficient of 1.4 (the rate of fluid milk to the total produced amount of milk in Ontario). That conversion allowed us to qualitatively estimate the effect of implementation of the two different schemes (monthly and annual) in one province. The figure 1 below compares the average of monthly quantities of TPQ exchanged during one dairy year period under the monthly scheme and adjusted MSQ exchanged under the annual scheme in kilograms for Ontario.  9  ▪ ▪^  Figure 1 Quantities of TPO Exchanged Aug 87 - May 05 under the Monthly Scheme (Average Monthly), Compare to the Adjusted MSO Exchange under Annual Scheme Aug 80- Aug 94. Ontario Province --o— TPO exchanged monthly Daity scheme ^—a—Adjusted MSQ exchanged monthly U&U scheme  >7.  450000  0 2^400000  •  c •0o  350000  .c o „c x^300000 o c  sg  • .er 0.  250000  "Es cm 200000  Z" Q  as  0 0  150000 100000 50000 0  451 co' 09  ,s9,  q9"^  c0^ 0^ osb'  ce^  c  ce  00^oti  ,"  cto  Years  The graph shows that the amount of the total quotas exchanged has increased almost twice since the province began to operate under the monthly scheme. This result shows that the annual and monthly schemes have different effects on fanner behavior in the exchange quota market. Due to the much shorter assessment period under the monthly scheme, farmers have a limited time period to react to stochastic factors and are more likely to have to go to the quota exchange market to buy or sell quotas. Even with a grace period under the monthly scheme, the regulations may encourage farmers to go to the quota exchange market more often. Nevertheless, the results must be examined, because of the differences in transaction costs for the two systems. In Ontario, the 15% assessment levy was imposed for all transactions under the annual scheme and just prior (1996) to the introduction of monthly scheme, this levy was eliminated. So we are not sure how much of the increase in quota exchange transactions were due to decreased transaction costs or stimulation from the new system Moreover, the actual length of the assessment period substantially differs across the two schemes. Obviously, for the monthly scheme, farmers need a grace period to comply with 10  their production and quota allowance because of the short amount of time to overcome unforeseen stochastic factors. As for the annual scheme, farmers have a much longer assessment period, and it gives them an opportunity to adjust production to meet the quota levels without buying or selling quota. Correspondingly, we might see an accumulation of quota exchanged quantities at the end of the dairy year if our hypothesis is correct. In addition, if the quota accumulation exists only for the last months of the dairy year, we can declare that farmers firstly want to adjust their production to the quota level and then only change the quota level if necessary at year-end. If this is the case, the regulations do not push farmers to exchange continuously during the quota year, and it gives a relaxation time for farmers, which allows them to minimize their costs and adjust production in the event of some random shock, such as weather conditions, animal health, and other problems. Also, to understand a farmer's behavior regarding quota exchange, we need to examine the quantities of quotas exchanged at the end of the dairy year for the annual scheme. Also, the quantitative measurement of the effect of the quota year approaching can estimate the effectiveness of the imposed rules and show how the system affects farmer behavior in the dairy quota trading market. Finally, an examination of the Canadian quota exchange mechanism will make clear the necessity or lack thereof of a grace period, which will help designers build an efficient permit trading system for both system administrators and participants.  11  Chapter 2 Literature Review In designing a Canadian permit trading system, the issue of management is very important. The stakeholders want a carbon permits exchange system without a criticism from public officials that the market for the permit exchange works inefficiently and requires extra costs to manage permit transfers. On the other hand, program participants desire to use the permit trading mechanism for minimizing their costs while still meeting their emission requirements under GHG regulations. One of a policymaker's key concerns permit trading mechanism use is the design of a transparent and flexible system that is not a burden and is an effective permit trading mechanism for all participants. Barichello (2002) suggested that the lessons learned from managing quotas under the supply management system could be used with great profit to guide the design and implementation of a domestic emissions permit trading regime Also, pollution permit trading systems have been used in the USA since the 1990s, and those schemes have many similarities with a currently designed domestic carbon permit trading system, and many lessons from their successful experience can be taken as well. To understand better how the carbon permit trading system works, the following sections briefly describe existing, well-known pollution allowance trading systems, such as the Sulfur dioxide (SO2) emissions program (the Acid Rain Program), the NOx Budget Program for ground-level ozone in the northeastern United States, and the Los Angeles pollution trading systems, also called the RECLAIM NOx and SO2 program; all these systems have characteristics similar to developing carbon emission permit trading schemes. 2.1 Current Pollution Permit Trading Schemes Sulfur dioxide (SO2) Emissions Program (the Acid Rain Program) A. Principles and Achievements The U.S. Acid Rain Program, created by Title IV of the 1990 Clean Air Act Amendments, has achieved great emission reductions in a short time. The program consists of two phases. Phase I began in 1995 and affected 445 units at 110 mostly coal-burning electric utility plants 12  located in 21 eastern and midwestern states. Phase II, which began in 2000, tightened the annual emissions limits imposed on these large emitting plants and also set restrictions on smaller, cleaner plants fired by coal, oil, and gas, encompassing more than 2,000 units in all. During the entire five years of Phase I, emissions were reduced by twice as much as was required to meet the Phase I cap. On a yearly basis of a compliance period, the annual emissions reduction has increased steadily from 3.9 million tons in the first year in 1995, to 4 4 million tons in 1999, the last year of Phase I, and to 6 9 million tons in 2002, a 77% increase in abatement by the eighth year (Ellerman 2003). Sulfur dioxide (SO2) emissions have fallen significantly, and costs have been even lower than the designers of the program expected. The "Cap and Trade" scheme was used to control emissions that were causing severe acid rain problems over very large areas of the country. A cap and trade program establishes an aggregate emissions cap that specifies the maximum quantity of emissions authorized from sources included in the program. The regulating authority of a cap and trade program creates individual authorizations ("allowances") to emit a specific quantity (e.g., 1 ton) of a pollutant. The total number of allowances equals the level of the cap. To be in compliance, each emitter must submit allowances equal to its actual emissions. Participants of the program may trade allowances with other participants of the program. Each participant can design its own compliance strategy: either emissions reductions or allowance purchases or sales to minimize its compliance costs, in response to changes in technology or market conditions, without government approval. Researches found three features of the successful environmental performance of the Acid Rain Program. First, a large reduction of emissions was accomplished relatively quickly—in the fifth year of Phase I. Second, the schedule of emissions reduction was accelerated significantly as a result of banking. Third, no exemptions or relaxations from the program's requirements were granted (Ellerman 2003)  13  B. Compliance Mechanism The program reported 100% compliance in all years of Phase I; this compares with the 80% compliance typical under other federal air programs (Swift 2001). Banking allowance was one of the specific design features of this system (Ellerman & Montero, 2002). Banking allows sources to carry over unused allowances for use in a later compliance period when there might be more restrictive requirements or higher expected costs to reduce emissions. Banking gives some flexibility for participants to manage the emissions reduction over time. Most of the reduction observed in 1995 was due to banking Also, 100% compliance reflects another feature of the simple, property rights system that Congress established with Title IV. Compliance became cheaper than seeking the various forms of relaxation that characterize conventional regulatory programs (Ellerman 2003). C. Compliance Period At the end of each year under the Acid Rain Program, each participant is granted a 60-day grace period, during which SO2 allowances may be purchased, if necessary, to cover each unit's emissions for the previous year. At the end of the grace period, the allowances a unit holds in its compliance account must equal or exceed the annual SO2 emissions recorded by the unit's monitoring system and verified by the US Environmental Protection Agency (EPA). Any remaining allowances may be sold or banked for use in future years. If annual emissions exceed the number of allowances held, the owners or operators of delinquent units must pay a penalty ($2,000 in 1990 dollars) per excess ton of SO2 emissions. The NO Budget Program  A. Principles and Achievements The NO Budget Program is a multi-state, regional program that was formed for the purpose of attaining the National Ambient Air Quality Standard (NAAQS) for ground-level ozone in the northeastern United States. Negotiations among these states led to a Memorandum of Understanding in 1994 that established a three-phase program of control of NO emissions from electric utilities and large industrial boilers. The NO Budget Trading Program (NBP) is 14  a market-based cap and trade program created to reduce emissions of nitrogen oxides (NO.) from power plants and other large combustion sources. The NBP was designed to reduce NO emissions during the warm summer months, referred to as the ozone season, when ground-level ozone concentrations are highest. Phase 1 began in 1995 on the basis of a conventional, command-and-control system, but it was recognized that further NO emission reductions would require a trading mechanism. Phases 2 and 3, beginning in 1999 and 2003, consisted of a cap-and-trade program during the ozone season (May through September) when the formation of ground-level ozone occurs. Beginning in 2004, the third phase was extended to cover most of the states east of the Mississippi River, in what is known as the NO SIP Call (EPA 2006). The EPA (2005) reported that ozone-season NO emissions are 60% below 1990 baseline levels, but two thirds of this reduction was accomplished in the first phase, which did not involve emissions trading. The second phase cap has reduced emissions by about 30% over the level achieved by the first phase, and the third phase will likely effect another 35% reduction. A cap-and-trade mechanism was chosen for the second and third phases, because regulators had come to recognize the limits of the conventional prescriptive approach for controlling air emissions, so they turned to the most practicable and less expensive alternative to achieve the desired reductions in sources of pollution (Ellerman 2003). B. Compliance Mechanism Under the NO. Budget Trading Program, initial allowances are distributed to all affected units based on historical heat input. There are two major allocated units: existing, new units, and opt-in units. One allowance is equal to one ton of NO emissions and may be used to authorize NOx emissions during the ozone season for which it was allocated or for a subsequent ozone season. By the NOx allowance transfer deadline (two months after the end of the ozone season), participants must hold sufficient allowances for each of their units in the appropriate accounts to cover their total seasonal emissions.  15  Although the phase 2 emissions trading program differs in important aspects from the Acid Rain Program, for instance in placing limits on the use of banked allowances, this phase has been successful in reducing NO emissions in the Northeast. Also, under Phase 3 (the NO SIP Call) of the NO. Budget Program, participants are allowed to obtain offsets for their NO emissions reduction from alternative sources. C. Compliance Period The length of the compliance period of the NO Budget Program is the five-month period each year from May to September because of the seasonality of the effect. The trading system allows participants to trade allowances in the market. Unused allowances are not transferable to another period, but the participants have a grace period until November 30 of each year, which is two months after the end of the ozone season. The grace period helps to obtain sufficient allowances to cover participant emissions.  The Regional Clean Air Incentives Market (RECLAIM, Los Angeles Basin Program)  A. Principles and Achievements The RECLAIM NO and SO2 programs provide a cap-and-trade approach to achieve emissions reduction in the Los Angeles Basin. The program was introduced in late 1993 on two phased-in, cap-and-trade programs, one for NO and the other for SO2, that would achieve the desired level of aggregate emissions in ten years. The environmental effectiveness of the RECLAIM programs has been comparable to those of the other cap-and-trade programs. NO emissions declined 60 percent, from approximately 25,000 tons in 1994 to 10,000 tons in 2004 (or 75 percent from initial allocations of approximately 40,000 NO x RTCs. SO2 emissions declined 50 percent, from approximately 7,000 tons in 1994 to 3,500 tons in 2004 (or 65 percent from initial allocations of approximately 10,000 SO2 RTCs). The SO2 cap has been met in each of the years since the program started. The NO x cap was exceeded in 2000 by 3,294 tons (16%) and in 2001 by 28 16  tons (0.25%) as a result of the electricity market problems in California in these years (EPA Clean Air Markets Division 2006). Nevertheless, the RECLAIM program received some critique for allocating too many NO„ RECLAIM Trading Credits (RTCs) to facilities in the early years. NO and SO2 RTC markets do not have a significant volume of trades, automatic financial penalties are not imposed, and there are high transaction costs for the cap-and-trade program due to a large amount of small-scale participants. B. Compliance Mechanism RECLAIM facilities must comply, such that their NO and SO2 emissions do not exceed their RTC holdings. Participants can trade RTCs to cover emissions in excess of allocation levels. According to annual program audits, the RECLAIM program has exhibited compliance rates of 84 percent in the first year of the program and of 97 percent in 2002 and 2003. Compliance in the SO2 program is generally higher than in the NO program. The 2004 overall compliance rate was 96 percent, with 13 facilities exceeding their NO RTC holdings (SCAQMD 2006). C. Compliance Period The limited amount of banking and borrowing allowed through the use of overlapping cycles in the RECLAIM does not explicitly include banking; each RTC expires at the end of its 12month term. There are two compliance cycles. A facility is assigned to one of these cycles and allocated RTCs accordingly. Cycle I facilities have a compliance year that runs January 1 to December 31; Cycle II facilities' compliance year runs July 1 to June 30. The two compliance cycles are intended to provide flexibility, promote a liquid market, and guard against price swings that might occur if all RTCs expired at the same time. Credits can be exchanged between the cycles, so two vintages of credits are available for time periods within each compliance year. This creates some opportunity for limited banking. Additionally, RECLAIM facilities are divided into two zones based on geographic location; coastal (Zone 1) and inland (Zone 2). Coastal zone reductions are more valuable due to 17  upwind impacts from coastal to inland areas. As a result, coastal zone facilities can only use coastal zone RTCs (except in limited circumstances), while inland zone facilities can use RTCs from both zones. Finally, the NO and SO2 RTC markets are separate; there is no interpollutant trading in RECLAIM. Assessments occur both quarterly and annually. RTCs can be purchased at any time throughout the year, and following each compliance year, a facility has a 60-day grace period to ensure RTC holdings are sufficient to cover emissions for that year. Facilities that fail to hold sufficient RTCs are required to surrender future-year RTCs at a ratio of 1:1 to cover any surplus and are also subject to substantial civil financial penalties determined on a case-bycase basis. In summary, all pollution trading mechanisms include grace periods after the compliance period ends. At the end of each year, each affected unit is granted a 60-day grace period, during which allowances may be purchased, if necessary, to cover each unit's emissions for the previous year. At the end of the grace period, the allowances a unit holds in its compliance account must equal or exceed the annual emissions recorded by the unit's monitoring system and verified by authorized agencies. Any remaining allowances may be sold or, for some schemes, banked for use in future years. If annual emissions exceed the number of allowances held, the owners or operators of delinquent units must pay a penalty. The length of the compliance period is linked to the environmental problem. If the environmental problem is continuous and long-term, as in the case of acid rain or RECLAIM programs, the compliance periods are continuous, covering all months of the year. If the problem is seasonal, as is the case with ground-level ozone in the eastern United States (the NO Budget Program), then the compliance period is seasonal, such as the five-month compliance period each year. The administrative burden depends on the frequency (quarterly, annually, or less frequently) of assessments imposed by authorized agencies. A short compliance period puts a larger administrative burden on both the regulating authority and participants. A longer compliance period allows more flexibility for the participants to achieve compliance and reduces the administrative burden for the regulating authority (EPA 2006). To test these statements, the Canadian supply management system might help. We can compare several different quota 18  trading schemes and examine the effect of the different period lengths on participant behavior in the quota exchange. Nevertheless, according to the Guide to Design and Operation of a Cap and Trade Program (2003), at the end of the compliance period, the participants should be given enough time to verify emission data for the period and to submit them for compliance. The period should not be so short as to cause the emission sources to submit data that has not been properly quality assured, but not so long as to unreasonably delay compliance assessment. It should also allow enough time for the regulating authority to finish conducting the compliance determination well before the end of the subsequent compliance period. This guide does not consider the possibility of using a strict compliance mechanism without a grace period for a permit trading system design that can, in one turn, reduce design costs and system operational costs. Analyzing the supply management system regulation mechanism will reveal the importance of grace period and length of permit period on participant behavior in the permit trading system. 2.2 Analyzing the Price and Quantity of Common Stocks on the Stock Market Much economic research has been done on the relationship between price and permit quantities. This problem is important for permit trading scheme designs due to uncertainties under a cap-and-trade scheme of management. Price spikes might occur if all permits expired at the same time. An imposed grace period can protect against price swings, as can an increase in the frequency of a compliance period, such as the monthly scheme. The annual scheme without the grace period and with a strict penalty mechanism can increase not only quantities of quotas exchanged at the end of the dairy year, but also will increase volatility of quota prices. These price shocks at the end of the compliance period are not very welcome for participants and regulating authorities, and avoiding them will benefit both sides. Some studies have been done in high returns during the end-of-the-year phenomena in the stock market related to the relationship between price and quantities at the end of the year. According to Givoly, Ovadia (1983), the seasonality in the stock market at the end of the calendar year exists due to tax-induced sales. The price of many stocks over the last 35 years was temporarily depressed in December but recovered in the following January. This price 19  recovery is a major contributor to the high returns observed in January. Their analysis indicated that tax-induced sales are the sole contributor to the high January returns. Dyl (1980) studied the effects of capital gain taxes on trading volumes of common stocks and concluded that "capital gain taxes affect investor's year-end portfolio decisions, and that effect is, in turn, reflected in the trading volumes of common stocks in December." Those studies explained the relationship of prices and quantities traded at the approaching tax deadlines, at the end of a calendar year, but the authors do not reveal the effect of the compliance period on the participants' behavior in the stock market. 2.3 Canadian Supply Management Studies  There are a number of studies on the Canadian supply management system regarding analysis of quota price versus quantity, but I have not found an end-of-the-year quantitative quota analysis. Also, I have not found any studies on how duration of a quota period affects farmer behavior in the quota exchange market. Nevertheless, I reviewed many applicable articles and found related information and methodological approaches for this study. The articles where authors develop procedures to obtain the values of welfare and marginal costs for Canadian milk production based on a profit optimization approach are closely related to the method used in this study. One of the problems of the profit maximization approach is difficulties with marginal costs determination. Briefly, the profit maximization model is usually used in performing welfare analysis based on knowledge about farmers' marginal costs. Marginal cost values can be derived using both the utility or profit maximization techniques. Barichello and Stennes (1994) computed average costs from cost-of-production survey data to approximate the marginal cost of milk production. However, survey methods, based on farm level averages, can only provide rough approximations of marginal costs. Moschini (1987) estimated econometrically the marginal cost of milk production from farm-level cost information. Dairy federal—provincial agencies used that information also to set the price of milk Since the milk producers participating in the cost-of-production surveys know this, there is an incentive for them to overstate their costs. 20  Another approach is to derive the marginal cost based on the difference between quota price and quota rental value. It can be derived from a single output, static model under the assumption of competitive markets (Moschini 1989, Babcock and Foster 1992). When quotas are rented freely, the rental value is directly observable and so is the marginal cost. Usually in the Canadian dairy industry, a quota rent is prohibited, so to obtain the rental value of quotas from capitalized values, two approaches are often used. Both approaches treat production quotas as a capital asset that provides a stream of annual returns. The capital asset pricing model is then used to determine the rental value. Arcus (1978), Albon (1979), Barichello (1981, 1984), and Veeman (1982) obtained the rental value by dividing the capitalized value of a quota by a discount factor. The choice of an appropriate discount rate is crucial to this approach. Since expected capital gains, interest rates, and the degree of risk inherent in the asset are unknown, the choice of a discount rate is largely arbitrary. Alternatively, Moschini and Meilke (1988) assumed a constant expected rental value based on estimates of long-run marginal costs of milk production. They then computed discount rates for unused and used industrial milk quotas. The derived discount rate was found to vary over time and across closely related assets. Barichello (1984) suggested that prices are established for two types of industrial milk quotas in Ontario: unused and used. It was recommended that the rental value for a quota can be inferred from the difference between the capitalized values of unused and used quotas, without any arbitrary choice of a discount factor. Barichello's rental value formula was adopted in Hickling's (1990) report to Industry, Science and Technology Canada to obtain the rental value of quotas and the marginal cost of milk production. Kevin Chen and Karl Meilke (1997) argue that Barichello's (1984) rental value formula (derived as a difference between the capitalized values of unused and used quotas) can be applied only under the assumption of a perfectly competitive quota action. Even though the formula was adopted in Hickling's report to Industry, their empirical results suggested that caution should be exercised in using the difference between unused and used quota prices because of the strongly regulated dairy market in Canada.  21  In the Canadian dairy industry, which prohibits quota renting, marginal costs are not observable. This is a problem for welfare analysis, which is out of the scope of this research. Also, since the marginal costs are a key element of the profit optimization approach, and the Ontario daily scheme does not rely on unused and used quotas, getting data for marginal cost estimations is not simple. Using a method suggested by Wooldridge (2003) for a policy analysis with an assumption that the time difference in marginal costs for one province converges to zero, we can solve the serious problem even with unavailable marginal costs. To my knowledge, this is the first study to examine quantitative quota analysis at the end of the dairy year in Canada.  22  Chapter 3 Theoretical Part  3.1 Analyzing Farmer Behavior in Dairy Quota Exchange  An efficient domestic emissions permit trading system should be designed for both parties— participants and system administrators. The lessons from the Canadian dairy supply management system can be learned with great success in terms of understanding farmer behavior in quota exchange. Also, the comparison of two schemes—monthly and annual— will explain farmer reaction to different regulations and estimate the effectiveness of those regulations by measuring the difference in quantities of quotas exchanged. Thus, the analysis of the dairy quota exchange mechanism will make clear the necessity or lack thereof of a grace period with relation to building an efficient permit trading system for both system administrators and participants. For this study, I proposed a profit optimization method in general frames, adding the speculative element of the quota price. The method shows a way for solving a farmer's profit maximization problem under uncertainty and can be developed for building a model for quantitative quota assessment at the end of the dairy year. The uncertainty comes from the fact that the milk quota price for the Canadian dairy industry is a risky investment, because of a possibility of quota system elimination or rent reduction, at any time by, for example, the federal government's international obligations. This method consists of two different approaches to analyze farmer behavior in the dairy quota exchange market.  The first is the profit maximization approach based on milk production under a quota constraint. This method can be used to analyze a farmer's behavior in the primary milk market where a farmer maximizes his or her profit-producing milk output and minimizes inputs. Solving the profit maximization equation leads to an inclusion of milk price and a farmer's marginal cost variables into the model.  23  The second is income capitalization approach, where a farmer is looking for some potential gains from milk quota capitalization. According to Barichello (1996), in a quota-protected market, economic rents most likely exist, and a producer receives benefits from an increase in quota price. When the market shows potential for a steady growth in quota prices, it indicates low risks of political changes in the near future, but on the other hand, it is uncertain if quota prices are inconsistent. Thus, a farmer's behavior in quota exchange depends on the price of the quota as well. Therefore, a dual approach that combines profit maximization in the primary market (milk production), and income capitalization in the secondary market (potential gains from quota capitalization) will model a farmer's behavior in Canadian quota exchange. Building a comprehensive duality model is not the purpose of this research and is out of the scope of a master's thesis; for simplicity, I provide a logical analytical path for development of a duality model. Nevertheless, these analyses explain the general idea of the duality model and verify the inclusion of the particular variables into the empirical model used in this thesis. 3.2 Primary Market, Profit Maximization Approach  The profit maximization approach can be used for analysis of a farmer's behavior under the Canadian daily supply management system, where a farmer continually faces quota constraints with a threat of being penalized for under or overproduction relative to the level permitted by quota. A provincial marketing board has to encourage a farmer to produce a constant quantity of milk in order to prevent the dairy market from overproduction or underproduction situations. Because of a tendency for a continuous increase in milk productivity per cow in Canada, a fanner is more likely to end up overproducing rather than underproducing at the end of the dairy year. This will increase demand for quota purchases in the short and long run. Provincial marketing boards impose a quite mitigated regulation on farmers who turn out to be underproducing, whereas for the overproduction situations, the regulations are stricter. On the other hand, there are some problems that might suddenly arise for a farmer, such as unpredictable weather conditions, business problems, sickness of family members, etc. Faced 24  by some unforeseen factors, a farmer will be forced to sell some quota to avoid losing quota or being penalized. Those factors usually affect a farmer's behavior in the short run and sometimes encourage a farmer to sell quota. Long-run quota sales are typically caused by an intention to leave the industry in the interests of other businesses or reduce milk production. Hence, following marketing board regulations, a farmer should have a constant marginal production during the dairy year to get to the optimum production level. In order to avoid penalties and maximize profits, a farmer has to adjust milk production or buy or sell needed quota. This also implies the importance of the provincial marketing board's regulations and correspondingly their tools — penalties - which are imposed on farmers. Regulations and their implementation are not the same across the provinces, and a particular scheme of a quota regulation might affect a farmer's intention to buy or sell quotas in a different way. Thus, penalties and regulations are important factors affecting farmer behavior in quota exchange. Canadian dairy farmers obtain profit from milk production activities under production quota constraints. A farmer's net manufacturing income is the difference between revenue from milk products and their costs. The level of production is subject to stochastic influences that raise or lower production from the planned level. If a farmer produces below or above the possessed quota limit, he or she is subject to penalties. Starting to build a theoretical model, profit maximization approach is the first step. At the beginning of the dairy year, the total quota amount  Q  is distributed among farmers by  marketing boards. Let N denote the number of farmers in a province. Initially, a farmer possesses the amount of quota  Q , so a total quota allocation for all fanners in a province:  Q =E  N  n^  (1)  There were two quota auctions, one for used and one for unused quotas for the annual scheme. The main difference between used and unused quotas is that the former can be used 25  only in the next year, while the latter can be used in the current dairy year. For simplicity, we will consider only unused quotas for the annual scheme in our analysis, and we will assume that an auction is held one time per year for both schemes. The Canadian dairy industry has strict regulations to avoid over- and underproducing situations. The marketing board penalties, which can be imposed on over- or underproduction relative to quota levels, are different. Under these circumstances, fanner reactions will depend on the current marketing board's penalizing policies.  Y <Q 1^1 Yi > Q -›  Z2  (2)  Where Z i and Z 2 are marginal penalties (per unit of output), for over- or underproduction situations, and y is actual milk output. i  Given penalties for a farmer for the failure to meet the quota constraint under stochastic influences, it may be cheaper to buy/sell quotas instead of reducing/increasing production in the short run. In other words, there can be large adjustment costs to change production, for example, by buying or selling cows, or by changing the feeding regime, especially within the tight time constraints of the last few months of the dairy year. So, the production constraint for farmers is:  y =g+AQ+E•^ (3) i  where Q is allowed milk production for the i-th farmer per dairy year, A  Q  is an  additional amount of quota bought (positive value) or sold (negative value) by a farmer on the quota auction to adjust milk production in the current dairy year, and E is the stochastic element of the actual milk output, which represents the unexpected changes in weather conditions, feed technology, animal health, and others. The net income for i-th farmer over the one-year period is:  26  zi=Pm*E C,^*/ Z2*e *(1. I) (4) -  -  -  Where Pm is a milk price, ci is an increasing marginal cost function of the i th farmer, -  z i is the net income of the i th farmer, Y. is the actual milk output for one dairy year, and -  / is an indicator variable:  {14Y <Q 1 41} 0 ifY >a > Z2 -  I  ,  -  3.3 Secondary Market, Quota Capitalization Approach  In addition, a quota price influences a farmer's decision to buy or sell quota in the short and in the long run. If quota price is reasonably low, then a fanner is more likely to buy quota instead of adjusting milk production. On the other hand, if the quota price is high, a farmer might prefer to decrease milk production and avoid buying the extra quota. Equation (4) shows only the net income flow from production, and it does not take into account the costs of buying or selling quota A Q . In other words, it ignores the cost of one of the input factors. Thus, it is necessary to include the quota adjustment expenses to the net income function for the current dairy year. These quota adjustment expenses do not reflect a speculative factor if fanners consider milk quota as an input, they show only purchasing/selling costs of the A  Q quota in the current year. So, the future gains from a  quota price increase are subject to the capital asset pricing model (CAPM). According to Barichello (1996) and Eaton and Gersovitz (1981), the CAPM for the supply management regime should include policy risks, because a fanner does not know when the regime will end. To build the dual approach, first we need to recall the fact stated above that a quota exchange auction is held once per year, and farmers can arrange their production buying or selling  27  A Q quota to avoid penalties. Second, the assumption that variable A  Q is relatively small  compared to the allocated quota Q is crucial, too. With these assumptions, in general the demand for A Q is a function of marginal benefits from milk production, including gains from reducing penalties Z 1 and Z 2 , marginal costs of buying the quota A Q including risk and discount factors in the current year, and future gains from A Q quota purchases in the long run:  AQ = 4r: (Z1Z2)P171(r g ) 1)0r g^(5)  Where: z i ,Z 1Z are marginal benefits from milk production, Pq2 1 is the quota price (  -  (with positive sign for buying and negative sign for selling),  r  is a normal interest rate,  including a systematic risk factor that current policy might change,  g is a discount factor,  including any expectations of future growth returns or capital gains, and future gains from the price increase of milk quota A  Pq00(r. g) is  Q , (see Barichello, 1996, for reference).  Solving the profit maximization problem with respect to A  Q will bring the following  general equation:  AQ F (pm KC (•),Pq HZ1,' Z23  ^  Where A Q is the quantity of quota exchanged on the market,  (6)  Pm (.) is the milk price,  Pq () is the quota price, c 6) is the marginal costs function, and^  Z 22 are marginal  penalties for over- and underproduction.  28  3.4 Transaction Costs of the Trading Systems and their Effect on Participant Behavior in Market Exchange The main idea of the designed permit trading systems is to minimize costs for participants and system administrations. However, the cost savings will depend on the particular mechanisms of a trade system and the transaction costs related to it. According to Gangadharan (2000), some transaction costs can be defined in the pollution market exchange systems, such as information costs—costs to enter the market, presence of brokers or auctions, and search costs looking for a partner. Transaction costs for the pollution market can be found in different steps of permit exchange. Before entering the market, the participants have to study the market rules and estimate the consequences of entering the market as a buyer or seller Finally, a company has to make a decision to sell or buy permits. Also, the permit traders are heterogeneous in nature and do not use the same input and output market, and not all of them have cooperation and business relationships; therefore searching for a partner is very costly and time consuming. In addition, if a system does not have a centralized auction, participants have to find a way to enter the market and find a partner, which increases transaction costs. For example, the Los Angeles pollution permit system (RECLAIM) does not have a central market for exchange, and participants have to search for a broker or partner to get into the market, which increases the transaction costs for all participants. In comparison to the Los Angeles pollution system, the Canadian Dairy Commission established provincial auctions for quota trading. The supply management mechanism includes an auction that reduces the transaction costs for farmers, while holding auctions every month adds operational costs. On the other hand, dairy farmers can use an electronic bid and offer submission mechanism based on the Internet that also reduces transaction costs for them in comparison to the RECLAIM trading system. These operational transaction costs exist but do not affect participants as assessment levies, which can be high as 15% of all traded quota. So, to simplify the problem, we can examine only the largest part of all the transaction costs for the supply management system—assessment levies. The largest transaction impact on farmer behavior under the dairy supply management system is a 15% assessment levy imposed on all exchanged quota due to setting aside some quantities of 29  quota for the special entry programs. Sellers have to give up 15% of their total quota sold, to the Board, when making a sale. (DFO 2004). The probability exists that if transaction costs are non-zero, some farmers will not participate in the permit exchange market. So, we need to distinguish two different uncertainties in the permit exchange market—production risk due to production/market uncertainties, and default policy risk that the current policy will end. Default policy risk leads to a higher threshold value for the investment in quotas in comparison with the situation where there is only production risk (Gangadharan 2003). Moreover, the high quota price affects a farmer's decision to exchange due to production risks; therefore, it increases transaction costs as well. If transaction costs are high enough with a high default policy risk that the system will end, it leads to a higher threshold value for investment in the quota, and as a result, it might reduce the volume of exchanged quotas. Thus, the transaction costs have to be included into the model as an independent variable. Finally, the equation (6) can be rewritten with the transaction cost variable Tr as:  A Q F Pm C (011 q (e)Z1 ,Z 2 , T r (•)) ( 7) )  3.5 Suggested Method for Quantitative Quota Analysis at the End of the Dairy Year  Under the monthly scheme, a farmer has twelve assessments, which smooth out the quantities of the exchanged quotas during the dairy year. A farmer has the same response periods at the end of every month for over- and underproduction situations from the beginning until the end of the dairy year. Therefore, the quantities of the exchanged quota are distributed independently from the end period of the dairy year (one year), and the end-ofthe-year quota exchange results will not show evidence of the effect of the end of the dairy year. Conversely, under the annual system with only two measurement points, a farmer does not need to adjust milk production monthly and might keep unused quota longer (up to 6 months) with an opportunity to fix the underproduction situation before the assignment period approaches. At the same time, if a farmer produced over the quota limit in the short run, he or 30  she could wait and try to reduce production to meet a quota constraint in the current year. Accordingly, the large amount of exchanged quota might accumulate at the end of the dairy year for unused quota under the annual scheme. Thus, the quantities of exchanged quota should not show the evidence of the end of the year under the monthly scheme, whereas under the annual system, the quantities of quota exchanged might be concentrated at the end of a dairy year. The Ontario marketing board reported data for both schemes, so to get the estimation of the end of the dairy year effect on the quota exchange market we need to compare two schemes for one province. Other factors, such as weather, any seasonal fluctuations, farmer costs, or quota capitalization, can be assumed to be the same for both systems in one province. Finally, assuming that the monthly scheme does not affect a farmer's decision to exchange quota at the end of the dairy year, we can state that comparing two schemes of supply management regimes within one province will answer our research question, and we can estimate the quantity of quota exchanged at the end of the dairy year under the annual scheme. Adding monthly dummies to the models for the annual and monthly schemes will give a quantitative assessment of the effect of the dairy year on the annual scheme. Thus, building two models for two schemes with dummies allows us to find a difference between dummies for the same month. The logarithm specification of the model can help interpret the dummies as a percentage of the changes and examine the effect of the dairy year for two schemes along with an appropriate model specification, which will be discussed in the next chapter. This quantitative assessment of the two schemes will indicate the impact of regulations imposed by the Ontario milk marketing board (DFO). Moreover, the results will show when and how farmers exchange quota and help us understand how long compliance and grace periods are needed for the design of a permit trading system.  31  Chapter 4 Empirical Model 4.1 Introduction to the Empirical Model The objective of the empirical analysis is to confirm the theoretical hypothesis that the end of the dairy year affects the quantity of quota exchanged under the annual scheme. According to the statement made in the theoretical part above, the monthly scheme does not affect a farmer's decision to buy or sell quotas at the end of the dairy year, whereas under the annual scheme, the exchanged quotas might accumulate at the end of the dairy year (Chapter 3). The Ontario marketing board provincial data provides a unique opportunity to compare two different systems - the monthly scheme and the annual scheme - within one province. The Ontario marketing board (DFO) used the annual scheme from August 1980, until July 1997, and the monthly scheme from September 1997 to this day. In contrast to the monthly scheme with monthly assessments, the annual scheme has far fewer assessment deadlines (for example, the current BC marketing board scheme has only two assessments—one each in January and July), which affect a farmer's behavior to sell or buy quotas at the end of a dairy year. Other factors, such as any seasonal fluctuations, farmer's costs, quota capitalization, profit gains, and imposed restrictions are relatively similar for the two systems from the Ontario province. Thus, to estimate the quantities of quota exchanged at the end of the dairy year, the monthly scheme can be considered as a control group and the annual scheme as a treatment group for the purpose of the statistical  analysis. Monthly dummies in the regression models (with quota quantities in kilograms of butterfat as a dependant variable) will show a difference in the quota exchange for each month of the dairy year, keeping other factors constant Finding the difference between the estimates of dummy coefficients for the same months for two systems at the end of the dairy year—April, May, June, and July—will allow quantitative assessment of the effect of the end of the dairy year on quota exchange. At that point, if a control group is not affected by the end of the dairy year, then monthly dummies for the last 3-4 months for the monthly scheme should not be significantly different from zero. Moreover, using a pooled regression and special dummies as an interaction of monthly dummies with control group dummies will give a quantitative assessment of the dummy differences for the two systems and provide their standard error estimations as well (Wooldridge 2003). 32  In addition, using a logarithm functional form, one can estimate the difference between monthly dummies in percentage terms. Introducing a logarithm transformation of a dependant variable (quantities of exchanged quota), which is recommended in general for any time-series data with all-positive observations, allows the estimation of the quantitative differences between the two schemes. 4.2 Including the Transaction Cost of Quota Trading  Some problems in including transaction costs into the regression have arisen. First, the transaction costs are not fully observable. According to DFO data (2004), under the dairy scheme, only 3-4 % of allocated quotas are transferred through the market exchange every year. From 1979 to 1996, the Ontario marketing board used a 15% assessment levy for sellers for all transferred quotas, including ongoing and within-family transactions (e.g., family quota transfers from parents to children). Under the monthly scheme, all transactions go without assessment (we used data for the monthly scheme after 1997). When the monthly scheme was introduced in 1997, the levy had already been eliminated in 1996. However, I excluded from the regression analysis data from August 1994, to August 1997 because of an adjustment period for quota measurement (DFO 2004). Thus, we have data available for periods before the regulation changes and after only. Second, according to our hypothesis, we expect to see that the monthly scheme will encourage farmers to exchange more, so an increase of exchanged quota quantities from the annual to the monthly scheme for Ontario will contain two elements: an increase of quota exchange by imposed new rules for the monthly scheme ("regulations" effect) and an increase of the eliminated levy ("transaction costs" effect) together. Because of the absence of available data for determination of the transaction effect, some assumptions have to be made. First, I assume that a levy or transaction cost will affect the quantity of market traded quota and ongoing and within-family exchanged quotas for the two schemes equally or at least with the same proportion. In other words, the ratio of market traded quota over ongoing and 33  within-family exchanged quota will stay the same or close, if there is no change in regulations. If so, this ratio for the annual and monthly schemes will not change much if only a levy will be imposed for both schemes, without changing regulations. Unfortunately, the regulations were changed. So we need more assumptions. The second assumption is that regulation changes will affect only a quantity of marked traded quota. We can suppose also that ongoing operation and within-family quota quantities of exchanged quota are relatively stable by nature; farmers exchange these quotas on a planned basis, and their decision to exchange within family or ongoing operation does not depend on imposed marketing board regulations. Thus, the quantity of market traded quota will be affected by regulations and transaction costs, whereas the ongoing operation and withinfamily quota quantities will be affected by transaction costs only. Now let's recall the first assumption, that the ratio of market traded quota over ongoing and within-operation quota is the same for the annual and monthly schemes if regulations are not changed. So, we can state that the ratio for the annual scheme differs from the ratio of the monthly scheme only for the regulations effect. In other words, if the annual scheme ratio is larger than the monthly ratio, the regulations effect exists only. If the ratios are close to each other, only transaction costs affect the monthly scheme. If the annual scheme ratio is smaller than the monthly ratio, then our assumptions are wrong. So, to include a transaction costs variable in the empirical model, we can use a proxy variable as an ongoing operation and within-family quota quantities with the same pattern. If regression coefficients of this variable for the monthly scheme are less than for the annual, transaction costs and regulations effects exist in quota exchange. If the coefficients are the same for both schemes, it means only transaction costs affect the trade, not regulations. If the monthly scheme coefficients are larger than the annual scheme, it shows that our assumptions are wrong. To get the net of these coefficients, a pooled regression is used for the standard error estimation.  4.3 Problems with the Regression Model Solving the endogeneity problem of the quota price Pq and milk price Pm variables is one of the problems with using this regression specification. The second problem is related to 34  the unavailability of marginal costs data  C (). The third problem involves the difficulty of  including penalty variables for two different schemes. As a solution for the first problem, a lagged variable the market clearing quota prices  Pq t - 1 , as an instrumental variable of  Pq is used to solve the endogeneity problem. The Pq t _ j  variable could be considered as an exogenous variable for the behavioral model. The intuition behind this assumption comes from the idea that a fanner makes a decision to buy or sell quotas after observing the quota price from the previous quota auction results. At the same time, that last month market clearing price is not affected by the farmer's current decision-making process, so it is possible to assume that the lagged  (t -1) quota price is an  exogenous variable. Thus, according to our model specification, we can run 2SLS regression with including the logged quota price variable  Pq t . ] as the instrumental variable for Pq .  The milk price Pm is regulated by the marketing board and calculated using a cost-ofproduction (COP) formula. In general, COP formulas include variable input costs (feed, labor, etc.), fixed input costs (depreciation, plant and administration, overhead, etc.), levies paid by producers to operate national agencies and provincial boards, and a return on labor and investment (DFO statistical handbook 2004-2005). The Ontario marketing board does not provide aggregate prices for industrial and fluid milk on a monthly basis. At the same time, the Ontario marketing board reported the farmers' monthly gross returns for fluid and industrial milk in Canadian dollars per hectoliter, so it is reasonable to include it into regression as a proxy variable for a milk price. Accordingly, the proxy of the milk price variable does not bring an endogeneity problem to our empirical model because of the regulated nature of the milk price. According to the equation  (6) quantity of quota exchange at the auctions, A Q  function of the marginal costs  is a  C 6), which are not observable from our data sources.  Moreover, Canadian farmer costs are subject to a seasonal effect because of the small scale of production. Canadian farmers have lower costs at the beginning of the spring and summer 35  seasons (they can use new pasture to feed animals) and higher costs in winter. So, the seasonal cost difference may affect a farmer's behavior in buying and selling quotas on the quota market. Also, it is difficult to derive marginal costs based on capitalized quota values in the Canadian dairy industry; Barichello (1984), Moschini (1989), and Babcock and Foster (1992) suggested using two types of industrial milk quotas in Ontario—used and unused—to calculate an economic rent. The problem remains that the monthly scheme does not operate on the used and unused quota basis, and a rent calculation using that method is impossible. In spite of data difficulties, we have to include the cost variable  C  (.) into an empirical  analysis. Excluding the marginal costs variable from the models leads to bias of the parameter estimates. This will be the case only if omitted variable (i.e., costs) are correlated with other explanatory variables, and this correlation is most likely to exist with the milk and quota prices. To replace the cost variable with a proxy variable, I used a farm price index from Statistics Canada sources. The data were available on a quarterly and annual basis only. I discontinued the indices by CPI as for all variables, including in the empirical model, along with taking a logarithm for easier interpretation. The proxy variable does not produce an exogenesis problem from this inclusion. 4.4 Omitting Penalties from the Empirical Model Penalties  Zl'Z2, are different across provinces. Moreover, a provincial marketing board  change particular polices from time to time, which affects a farmer's behavior in quota exchange. It is difficult to compare the effect of penalty regulations for two regimes that use different measurements of quota accounts (for example, the Ontario annual scheme and monthly scheme). Nevertheless, supposing that we need to compare two different regimes within one province and we are interested in analyzing the quantity of quota exchanged at the end of the dairy year, in this case, the effect of penalties might be considered almost the same for the two regimes The penalties were introduced by one marketing board with the same management, and it is reasonable to assume that the designers tried to build a new regime 36  with similar regulation effects on farmers. In addition, it is difficult to design 7  1,  2  2  variables for the empirical model and collect realistic data, so it seems possible to omit the penalty variables from our empirical analysis. Therefore, the behavioral empirical model for a quantitative analysis of the milk quota should consist of a net income, including profit gained from milk production, costs of buying or selling additional quotas, future gains from dairy quota capitalization, and particular market regulations. Thus, a quota price, a milk price, marginal costs, and transaction costs affect a farmer's decision to buy or sell quotas on the quota exchange. Finally, the explained variable in RHS can be denoted by the quantity of total quotas exchanged.  4.5 Data I used provincial data for the annual scheme quota exchange system from Ontario marketing board published sources (DFO "Dairy farmers of Ontario" 2006). The monthly data covers the period from August 1980, to July 1994, and includes unused and used MSQ (marketing shared quota) prices measured in current Canadian dollars per kilogram of butterfat, as well as quantities of quota exchanged in kilograms of butterfat. For the monthly quota system, the provincial marketing board reports TPQ (total production quota) prices in current Canadian dollars per kilogram of butterfat per day and quantities of quota sold in kilograms of butterfat from July 1997, to May 2005. In the period from August 1994, to July 1997, the annual scheme was used, but the Ontario marketing board converted data to the monthly quota system and reported quota prices per kilogram of butterfat per day (the monthly scheme). Also, since September 1997, the Ontario marketing board has reported TPQ instead of MSQ and fluid quota separately. For this reason, I excluded the time series from August 1994, until August 1997, from the data set used in estimation. For the milk price Pm , I used the proxy variable: gross fluid and industrial returns in Canadian dollars per hectoliter of sold milk available from the Ontario marketing board sources. The data were reported on a monthly basis for the period from August 1982, to July 37  1983, and August 1984, to July 1994, for the annual scheme (126 observations), and from August 1998, to July 2003, for the monthly scheme (60 observations), totaling 186 observations. To capture the transaction cost effect, I included a proxy variable as quantities of quota transferred within family and ongoing operation expressed in kilogram of butterfat per year. To find a net between monthly and annual schemes, I introduced the interaction variables for pooling regression. To replace the not observable marginal cost variable for both schemes, I used a proxy variable: the farm input price index taken from Statistics Canada on a quarterly basis, discounted with CPI. The following tables (2 and 3) present the summary statistics of the variables, excluding a time trend and dummy variables.  38  Table 2 Data Summary Statistics for the Annual Scheme, Ontario, Aug. 1980 to July 1994 Variables 1  2  3  4  5  6  Gross returns for fluid and industrial milk. Real values, $ CAD per HL Gross returns for fluid and industrial milk, norm values, $ CAD per FIL, 1992 = 100% Total unused quotas exchanged monthly, kg of BF Unused quota price $ CAD per kg of BF, real values Unused quota price $ CAD per kg of BF, norm values 1992 = 100% Quotas transferred within family and ongoing operation, kg of BF per year, annually Farm input price index, Statistics Canada adjusted with CPI  Minimum  Maximum  53.06  Standard deviation 3.74  42.09  59.07  126  69.92  4.39  58.88  76.96  126  92,681.9  41,716.9  22,320.0  280,183.0  126  22.28  7.26  5.86  44.9  126  30.45  8.30  11.06  51.31  126  2,841,686  470,720  2,190,065  3,640,414  126  1.47  0.095  1.26  1.84  Observations, Aug. '80 — July '94 126  Mean  39  Table 3 Data Summary Statistics for the Monthly Scheme, Ontario, Aug. 1997 to May 2005 Variables 1  2  3  4  5  6  7  Gross returns for fluid and industrial milk, real values, $ CAD per HL Gross returns for fluid and industrial milk, norm values, $ CAD per HL, 1992 = 100% Total quotas exchanged monthly, kg of BF Quota price $ CAD per kg of BF per day, real values Quota price $ CAD per kg of BF per day, norm values 1992 = 100% Quotas transferred within family and ongoing operation, kg of BF per year, annually Farm input price index, Statistics Canada adjusted with CPI  Minimum  Maximum  61.62  Standard deviation 2.61  54.64  61.62  60  53.59  2.01  48.68  57.58  60  370,259.4  141,075.2  149,613.5  833,076.0  60  57.69  12.59  37.26  80.55  60  50.63  8.24  35.76  66.13  60  4,904,968  617,217  4,182,275  6,169,930  60  1.27  0.035  1.19  1.36  Observations, Aug. '97 — May '05 60  Mean  Thus, the annual scheme gives 126 monthly observations in total. Because of the shortage of gross returns for fluid and industrial milk data and three missing observations for unused prices in the beginning of dairy years, 42 observations in total were dropped. 40  Because of 33 missed gross returns for fluid and industrial milk data, the monthly scheme provides 60 observations overall.  4.6 Method of Estimation The methodology of the regression model was taken from an econometric policy analysis (Wooldridge 2003). The pooling regression will give a difference in particular months as a coefficient of the interaction term between the control group dummy and monthly dummies. It also provides a standard error for these estimators:  -1-0(Qm)=a -FAL94Pm)±132 1-0(Pq) -FALPITA -FALPAC)±tid, +Mc  +6ocidi + 7 60,0,1Tr)+E.  ^  (8)  ,  1=9  where:  QM is the quantity of quota exchanged, monthly, in kilograms of butterfat;  Pm, is gross returns for the fluid and industrial milk variable, in $/hectoliter; Pq  is the quota price, in $/kg of butterfat, the instrumental variable (  Pq t is the lagged  quota price that was used for the 2SLS method);  Tr is the proxy variable for transaction costs, defined as the quota transferred within family and ongoing operation, in kg of butterfat;  C is the proxy marginal costs variable, farm input price index, from Statistics Canada adjusted with CPI;  di are the monthly dummy variables, I = 1,2...11;  if  is the control group dummy, which takes a value of one for the control group (monthly  scheme) and zero for the treatment group (annual scheme);  ao is an intercept;  A,11133/34 ix 4y are parameters to estimate; 41  9 is the parameter of interest on the interaction terms that shows the difference in monthly dummies for the last four months between the annual and monthly schemes; is the net of the transaction variable coefficients for the monthly and annual schemes; and E is the error term. Three regressions were implemented. One is for the annual scheme with 2SLS and robust standard errors; the second is for the monthly scheme with 2SLS robust standard errors, and the third one is a pooling regression with 2SLS robust errors. Parameters efor the last four months of pooling regression indicated the magnitudes of a difference between monthly dummies for the two schemes at the end of the dairy year. The method allowed us to obtain confidence intervals for the interaction terms as well. To check the regressions for autocorrelation, the Durbin—Watson test was implemented for both schemes. The Cook— Weisberg test for heteroskedasticity was implemented as well. The results are reported in the following section. 4.7 Results  The results of the regressions of the quantity of quota exchanged over independent variables and dummies (Equation 8) are shown in Table 4. For the annual scheme, the unused quantities and prices of MSQ have been used, whereas for the monthly scheme, the total amount of TPQ and prices have been used.  42  Table 4  Results of Estimations Dependent Variable: quantity of quota exchange, kg of butterfat (in logarithm)  Variable  2SLS robust estimation, pooling  2SLS robust for monthly scheme separately  2SLS robust for annual scheme separately  Quota price with instrumental variable Log(Pq,_ 1 ),$ per kg of butterfat Log (Pm) - gross milk returns for fluid and industrial milk, $ per hectoliter Log(Tr) - proxy transaction costs variable. Quota transferred within family and ongoing operation, kg of BF Log (C) - proxy variable for costs. Input price index, adjusted with CPI Trend  -0.454a (0.185)  -0.053 (1.454)  -0.392b (0.225)  2.37 b (1.129)  1.418 (1.184)  3.966b (2.072)  0.353 (0.238)  0.653 (0.626)  0.361 (0.232)  0.820 (0.820)  -0.895 (1.664)  1.462 (1.069)  0.007a (0.002) 0.133 (0.166)  -0.005 (0.010) -0.220 (0.228)  0.010b (0.003) 0.326 (0.212)  0.089 (0.179) 0.122 (0.170) 0.063 (.151) 0.175 (0.143) 0.366b (0.174) 0.527 a (0.137) 0.553 a (0.143)  -0.450 b (0.172) -.0038 (0.137) -0.309 (0.187) 0.151 (0.179) 0.387b (0.200) 0.339b (0.159) 0.504a (.189)  0.333 (0.239) 0.169 (0.247) 0.213 (0.204) 0.178 (0.193) 0.349 (0.241) 0.598a (0.182)  Dummy, Sep D2 Dummy, Oct D3 Dummy, Nov D4 Dummy, Dec D5 Dummy, Jan D6 Dummy, Feb D7 Dummy, Mar D8 Dummy, Apr D9  0.62e (0.171) 43  Variable  Dummy, May D10 Dummy, Jun Dll Dummy, Jul D12 Interaction dummy, Apr D9*Dc Interaction dummy, May D10*Dc Interaction dummy, Jun Dll*Dc Interaction dummy, Jul D12*Dc Interaction Log(Tr) *Dc Const  2SLS robust estimation, pooling 0.691a (0.143) 0.847 a (0.177) 0.651 a (0.170) 0.149 (0.168)  Table 4 to be continued 2SLS robust for 2SLS robust for monthly scheme annual scheme separately separately 0.784a 0.3626 (0.197) (0.166) 0.98e 0.194 (0.206) (0.181) 0.785' .0.025 (0.144) (0.184) -  0.129 (0.148) -  0.427 b (0.217) -  0.464a (0.150) -  0.683 (0.454) -3.81 -3.78 -11.544 (5.99) (8.815) (9.308) (15, 125) = 1.20 Durbin-Watson d- (20, 186) = 1.33 (15, 60) = 1.84 statistic 60 126 Quantity of 186 observations R2 0.828 .621 0.378 Notes: Standard errors are reported in parenthesis with a, b and c standing for 1%, 5% and 10% significance levels, respectively  The Durbin—Watson statistic for the polling regression and for the annual scheme showed that Ho of no autocorrelation has to be rejected. Nevertheless, the Durbin—Watson statistic for the monthly scheme was close to 2 (1.84), and the autocorrelation almost did not exist. The  dU= 2.06. Thus, to correct the variances of estimators, the robust options of 2SLS were used for all regressions. Nevertheless, with the omission of the marginal costs  C  we may  get biased results in the presence of autocorrelation. The robustness procedure will not correct this bias if autocorrelation is in place.  44  The differences between April to July dummies (interaction dummies for the pooling system) show negative values for all of them and for the last two months (June and July); also, the differences are statistically and economically significant. The negative sign for all four interaction dummies depends on choosing the  "dc dummy" as a control group dummy and  shows that under the annual scheme, farmers exchanged more quotas in kilograms at the end of the dairy year than under the monthly scheme. The magnitudes of these interaction coefficients show that the effect of the end of the dairy year exists for the annual scheme and can be counted in percentage terms according to the chosen specification of the regression. The graph below represents the magnitudes of dummies during one dairy-year period: Figure 2  Monthly Dummies for the Dairy Year for the Control and Treatment Groups, Ontario Aug. 1980 to May 2005 Values of Monthly Dummies of Quota Exchanged in Ontario under Monthly (97-05) and Annual (8094) Schemes [  1  .-  Annual Scheme^Monthly Scheme  Sep Oct Nov Dec Jan Feb Mar Apr May Jun J ul  Dairy Year, Months  Starting in January, both schemes express the seasonal effect. March, April, and May dummies for both annual and monthly schemes have positive signs, and the graph has a similar "shape". The difference in magnitudes of these coefficients can be seen at the end of 45  the dairy year. Beginning in April, the monthly dummies of the monthly scheme lined down in values compared with dummies from the annual scheme. All dummy coefficients for the annual scheme are significant from March until July, whereas the monthly scheme coefficients become insignificantly different from zero for the last two months (June and July). This confirms the proposition outlined in the theoretical section of this work that the annual scheme does affect a farmer's decision to buy or sell quotas at the end of the dairy year. The marginal cost variable was neither economically nor statistically significant, which can be explained in two ways: first, that the marginal costs do not affect a farmer's decision to exchange, or second, the proxy variable does not sufficiently correlate with non-observable marginal costs. The net of the transaction cost variables for the annual and monthly schemes almost hits the 10% significance level (P-value = 0.135), and a magnitude of the coefficient is economically large and positive (0.683). It shows that the difference between the two schemes exists and is statistically significant. Results show also, for separate annual and monthly schemes, that the coefficients of the transaction cost variables are not similar. The coefficient for the annual scheme is almost statistically significant, with a 10% significance level (P-value = 0.124), while for the monthly scheme, the transaction cost variable statistically is not significant from zero. Also, the monthly scheme transaction costs coefficient is larger than for the annual scheme in magnitude, and it contradicts the proposed hypothesis that transaction costs coefficients for the monthly scheme should be lesser than for the annual scheme. The small sample size (60 observations) may be an explanation, but the magnitude has a positive sign, and it is economically significant from zero. The hypothesis stated above of including the transaction costs is still not answered, and further research with a larger sample size could reveal the effect of transaction costs on farmer behavior in quota exchange.  4.8 Discussion The empirical results confirmed the theoretical hypothesis and showed evidence of the influence of different schemes (monthly and annual) on farmer behavior in quota exchange. 46  Moreover, the regression specification allowed estimation of the effect of the end of the dairy year on the annual scheme in percentages, assuming that the monthly scheme does not affect farmer behavior in quota exchange at the end of the dairy year. The interaction dummies for the pooling regression are significant only for the last two months, which confirmed the hypothesis that farmers tried to adjust production first, and then only went to the quota exchange market when necessary. If the quota period is long enough, as in one year or longer, farmers have enough reaction time to adjust the production first, without going to the quota exchange. Thus, at the end of the dairy year under the annual scheme, farmers exchanged quotas 43% more in June, (1% significance level) and 46% more in July (1% significance level) in comparison with the monthly scheme, assuming it is not binding with the dairy year's restrictions. The results of a 44% increase in quota exchange for two the last months are a reasonable outcome, described by Barichello (1996). The results indicate that the system affects a farmer's decision to exchange. Also, the length of a quota period and existence of a grace period is an important issue for a properly functioning trading mechanism. If farmers have a longer quota period, for example under the annual scheme, they do not need to exchange quotas every month, and they can get some relaxation in production planning to overcome any stochastic problems in the short run. Under the annual scheme, the results show that only in the last two months do quota exchange activities increase, so if a grace period is needed, it has to be not longer than 2 months. In fact, under the annual scheme, farmers do not even need a grace period to put all accounts in order. In other words, nothing will change in terms of a farmer's production plans in the short run when increasing the length of a quota period for two months, for example. One year or 14 months is not a crucial difference in the period's length, and farmer behavior in quota exchange will not be significantly different. At the same time, two months of grace period time will increase operational costs for system administrators and, correspondingly, for taxpayers. Nevertheless, the annual scheme may be a cause of price increases at the end of the quota year, which will overall increase annual volatility in price changes, due to quota exchange quantity accumulations in the last two months. 47  The research shows also that the proposed hypothesis that the monthly scheme penalty regime can stimulate farmers to exchange quotas more is still not answered. The transaction costs coefficient for the monthly scheme is larger in magnitude than for the annual scheme but not statistically significant from zero. We expected to see an opposite situation to confirm the hypothesis, in which we stated that transaction costs have the same effect on a farmer's decision to exchange either within family or ongoing operation or through market-sold quota exchange. It can be explained that reducing transaction costs such as an assessment levy encourages farmers to exchange within family or ongoing operations more than on the open traded market, and we cannot conclude that regime stimulates farmers to exchange. No one has done empirical research in this area, so it is difficult to compare the results with other studies. Moreover, the dummy magnitudes might be biased. Feasibly assuming that the correlation between the milk price, lagged quota prices, and monthly dummies is not significant, it seems reasonable to accept the result of this research—that we estimated the changes of quota exchanged quantities at the end of the dairy year under the annual scheme. 4.9 Summarizing Research Findings  Many proposed permit trading systems require sufficiently long grace periods (up to 3 months) to put all participants' accounts in order after the permit period ends, which delays procedures for compliance and does not use sufficiently enough the regime's stimulation. Based on the analysis of the Canadian supply management system we can formulate four main study findings: - The regression specification allowed the estimation of the quantitative effect of the quotas exchanged at the end of the dairy year with a strong compliance mechanism and without a grace period—the annual scheme. The results showed that farmers try to adjust production first, and only in the last two months of the dairy year would they consider changing their quota level. The coefficients of dummy variables show that the quantities of quota exchanged in the last two months increased by 44% at the end of the dairy year. - The length of the quota period and the presence of a grace period affect farmer behavior in quota exchange. First, longer periods provide more flexibility to farmers to adjust their 48  production to comply with the quota allowance. Moreover, a longer permit period eliminates the necessity of a grace period, which can reduce operation costs for system administrators. Also, a two month grace period seems long enough for farmers to put their accounts in order, according to the two month activity at the end of the dairy year under the annual scheme. -The monthly scheme with 12 assessments does not have a peak of exchanged quota volume at the end of the quota year. Also, the monthly scheme with a single quota exchange market seems to be easier to operate and more logical for participants because of the clarity and transparency of its rules. - Research did not confirm the stated hypothesis that the monthly scheme (with a short compliance period and a grace period) would encourage farmers to exchange more quotas during the dairy year, compared to the annual scheme, with a longer quota period Finally, it is possible to state that an annual assessment period (one year) or longer for a developing permit trading regime might be used without a grace period. In the carbon permit trading market, participants will have enough time to adjust their emission to comply with their permit allowance. Using an efficient penalty mechanism without a grace period will reduce operational costs for system administrators and possibly not increase short run production costs for participants.  5. Conclusions This thesis uses a theoretical approach and an empirical analysis to examine the effect of the end of the dairy year and its influence on the milk quota exchange market under the supply management regime in the Canadian milk industry. In the theoretical part, I proposed a dual farmer's behavioral model (a profit optimization method with a speculative element of the quota price) on the quota exchange auctions, and I explained why this model has been chosen. I built a hypothesis that the monthly scheme does not encourage a farmer in exchanging at the end of the dairy year, in contrast to the annual scheme, which does affect it. 49  In the empirical section, I verified the theoretical hypothesis that the monthly scheme does not encourage more quota transfers on the quota exchange at the end of the dairy year, whereas the annual scheme does. The empirical results showed no significance of the June and July dummies (end of the dairy year) for the monthly regression, which confirmed the assumption that the monthly scheme does not influence quota trading at the last two months of the dairy year. Conversely, the annual scheme is characterized by the large size and statistical significance of June and July dummies. The interaction dummies for a pooling regression for June and July might be interpreted as the percentage of the difference in the quota exchange between the two schemes. Finally we can state that the results supported the research in developing the Canadian domestic permit trading scheme using lessons from the Canada's supply management system. The research found that the length of the quota period affects farmer behavior in quota exchange. Also, the research results showed that under the annual schemes, farmers do not need a grace period. Thus, this compliance mechanism reduces costs for system administrators and does not increase farmer production costs in the short run. Policymakers can use this experience to design an efficient domestic permit trading system with minimal costs for all participants in the system.  50  6. References  Albon, R.P. "The Real Cost of the British Columbia Milk Marketing Board: Is It Correctly Measured?" Canadian Journal of Agricultural Economics (1979)7: 44-51 Arcus, P.L. "The Value of Milk Quota in British Columbia" Canadian Journal of Agricultural Economics (1978) 26: 62-71  Babcock, B.A. and W.E. Foster "Economic Rents under Supply Controls with Marketable Quota" American Journal of Agricultural Economics (1992):630-637.  Barichello R, "Managing Domestic Emissions Permit Trading: The Relevance of Canada's Supply Management Quotas" (working papers) UBC, 2002  Barichello, R.R. "Analyzing an Agricultural Marketing Quota". Discussion Paper #454 Economic Growth Center Yale University 1984 Barichello, R. "The economics of Canadian dairy regulation" Technical report # E/12 Ottawa: Economic Council of Canada 1981. Barichello R.R. "Capitalizing government program benefits: Evidence of the risk associated with holding farm quotas". In the Economics of Agriculture: Papers in Honor of D. Gale Johnson, Vol .2. edited by J. Antle and D. Summer, pp. 283-99. Chicago: The University of Chicago Press 1996. Barichello R. and B.Stennes "Cost Competitiveness of the Canadian Dairy Industry: A Farm Level Analysis", in Supply Management in Transition Towards the 21st Century". Proceedings of Conference Held at McGill University, Ste. Anne de Bellevue, Quebec. June1994. Boehm Marie "Afforestation on the Prairies -Opportunities and Challenges:Agriculture and Carbon Sequestration 2003" http://www.mbforestryassoc.ca/pdf/CarbonSequestrationBoehm.pdf  51  British Columbia Milk Marketing Board "Milk Production for the 2005/2006 Dairy Year, TPQ Management 2005".  http://www.milk-bc.com/publications/index.php/download/274/0506prod.pdf DFO "Dairy farmers of Ontario" Policy Booklet 2004  http://www.milk.org/publications/index.html Ellerman, A. Denny Juan-Pablo Montero. 2002. The Temporal Efficiency of SO 2 Emissions Trading, MIT-CEEPR Working Paper 02-003  Ellerman Denny "Are cap-and-trade programs more environmentally effective than conventional regulation?" 2003 Center for Energy and Environmental Policy Research  http://web.mit.edu/ceepr/www/2003-015.pdf EPA Clean Air Markets Division, 2006, "An Overview of the Regional Clean Air Incentives Market (RECLAIM)" http://www.epa.gov/airmarket/articles/reclaimoverview.pdf Lata Gangadharan, Cason,Timothy N., 2003. Transactions Costs in Tradable Permit Markets: An Experimental Study of Pollution Market Designs. Journal of Regulatory Economics 23, no. 2:145-165  Hickling Report. "International Competitiveness of Dairy Food Processing in Quebec and Ontario" Industry Science and Technology Canada, Food Policy Task Force and Subsidy Analysis Branch Reference 3569 (1990).  "A Guide To Designing and Operating a Cap and Trade Program For Pollution Control" United States Environmental Protection Agency Office of Air and Radiation EPA430-B-03002 www.epa.gov/airmarkets June 2003  Jeffrey M. Wooldridge. "Introductory econometrics: a modern approach." 2003.  52  Kevin Chen and Karl Meilke "The simple analytics of transferable production quota: implications for the marginal cost of Ontario milk production." Canadian Journal of Agricultural Economics (1997) 42 (1):37-52 Marland, G., B. A. McCarl, and U. A. Schneider "Soil carbon: Policy and economics. In: Storing Carbon in Agricultural Soils: A Multi-Purpose Environmental Strategy". N.J. Rosenberg and R.C. Izaurralde (eds.). Kluwer Academic Publishers, Boston, MA, (2001) 111-117. Moschini, G. "Modeling the supply response of supply-managed industries: A review of issues" Canadian Journal of Agricultural Economics (1989) 37(3): 379-392. Moschini, G. and K. D. Meilke "Sustainable Rate of Return for Milk Quotas in Ontario" Canadian Journal of Agricultural Economics (1988) 36: 119-126 The National Round Table on the Environment and the Economy (NRTEE), Issue Paper # 9 (Policies that could complement a domestic emissions trading system for greenhouse gases) http://www.nrtee-brnee.ca/eng/index e.htm  Organization for Economic Co-operation and Development (OECD), 2002 http://www.oecd.org/home/0,2605,en 2649 201185 1 1 1 1 1,00.html  South Coast Air Quality Management District (2005), "Staff Report on Proposed Amendments to Regulation XX — RECLAIM", January www.aqmd.gov/hb/2005/050125a.html. Swift, Byron 2001. "How Environmental Laws Work: An Analysis of the Utility Sector's Response to Regulation of Nitrogen Oxides and Sulfur Dioxide under the Clean Air Act," Tulane Environmental Law Journal, 14:309 (summer) United Nations Framework Convention on Climate Change (UNFCCCB), 2004 http•//unfccc.int/2860.php  53  The US Environmental Protection Agency http://www.epa.gov/ Veeman, M. M.. "Social costs of supply restricting marketing boards Canadian Journal of Agricultural Economics". (1982) 30 (1): 21-36.  54  

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.24.1-0066744/manifest

Comment

Related Items