UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Interindustry variation in salaries : an empirical study Appelbe, Trent Woods 1969-05-18

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

Item Metadata


831-UBC_1970_A8 A66.pdf [ 3.65MB ]
JSON: 831-1.0102022.json
JSON-LD: 831-1.0102022-ld.json
RDF/XML (Pretty): 831-1.0102022-rdf.xml
RDF/JSON: 831-1.0102022-rdf.json
Turtle: 831-1.0102022-turtle.txt
N-Triples: 831-1.0102022-rdf-ntriples.txt
Original Record: 831-1.0102022-source.json
Full Text

Full Text

INTERINDUSTRY VARIATION IN SALARIES: AN EMPIRICAL STUDY by TRENT WOODS APPELBE B.A..; University of British Columbia, I96I A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in the Department of Economics We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September, 1969 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia., I agree that the Library shall make it, freely available for reference and Study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department pr by his representatives. It is understood that copying or publication of this thesis for financial gain sh9.ll not be allowed without my written permission. Department of ECONOMICS  The University of British Columbia Vancouver 8, Canada Date September 22, 1969 ABSTRACT The purpose of this study was to empirically investigate the determinants of the variation in white-collar salaries among United States manufacturing and mining in dustries. A model was formulated, hypothesizing that inter industry salary variation is primarily a function of industry monopoly power, capital-labour ratio and growth. Linear regression analysis was to be used to test the model with data from the U.S. Bureau of the Census and the Internal Revenue Department. Conceptual and statistical problems were encountered in measuring the industry variables. Although concentration and barriers to entry were to be used together in measuring monopoly power, in the final analysis it was concentration that had to be relied upon. The measure of industry growth, based upon change in employment, turned out to be inappropriate for its intended purpose. The regression results indicated that the amount of capital in an industry has no effect upon the level of salaries. Problems of multicollinearity made It impossible to isolate the capital-labour and capital requirements effects. i ii The evidence further indicated that industry concentration is a factor in explaining salary rates. However, the results were not decisive enough to allow any firm conclusions concerning the effects of monopoly power, particularly in view of the fact that concentration is an imperfect measure of this variable. . . Thus, one assertion of the hypothesis was tentatively accepted, another rejected, and for the third, the evidence was inconclusive. CONTENTS Page LIST OF TABLES . . .'  . . iv LIST OF FIGURES v CHAPTER I. INTRODUCTION . 1 II. THE MODEL ' . . . 8 III. DATA SOURCES AND MEASUREMENT OF REGRESSION VARIABLES 18 IV. REGRESSION RESULTS . . . . • 36 V. CONCLUSIONS 50 BIBLIOGRAPHY 2 APPENDICES I. MATCHING OF INDUSTRY DATA ........ 56 II. DETAILED REGRESSION RESULTS 65 iii LIST OP TABLES Page TABLE , I. Industry Variables 21 II. Occupational.Groupings for Regression Analysis 33 III. Multiple Regression Results - Coefficients of Industry Variables 4-0 IV. Comparison with Regression Results of Weiss' Study 47 V. Basic Industry Data . 59 - VI. Detailed Regression Results 6? VII. Regression Results for Unionization ... 69 iv LIST OF FIGURES Page FIGURE I. Wage Determination Under.Monopoly and Competition 14 v CHAPTER I INTRODUCTION The subject of interindustry differences in the earnings of salaried employees has received little attention in economic literature. Most of the research on the deter mination of earnings has centered about the analysis of the wages of hourly paid workers. In addition, there have been some scattered attempts to explain the variation in executive salaries. Such studies, however, are of limited value in contrib uting to a general analysis of salary levels. There exist important factors contributing to the determination of earnings that are unique for these different .classes of employees. Notably, unionization, which is given considerable attention in the study of wage levels, is likely to be of less Interest a's a determinant of the earnings of salaried employees. Converse ly, the existence of discretionary profits should exert its maximum Influence upon the earnings of top executives, having less effect for the rest of salaried employees and little or 1 none for wage earners. ^Discretionary profits have been defined as "the difference between actual profits and minimum profits demanded", O.E. Williamson, "Managerial Discretion and Business Behavior", American Economic Review. Vol. 53J> No. 5 (December 1963) j> p. 1035-This subject is explored in greater det3.il in Chapter II. - 2 -Thus, a gap exists in that there'has been no thorough empirical study of the industry characteristics that determine salary levels. In view of the trend away from unionized wage earners towards salaried employees, knowledge in this area 2 becomes increasingly important. Hence, this study will investigate the economic variables that influence the inter industry salary structure. To 'accomplish this, U.S. Census data on earnings of salaried employees will be matched with data on a cross section of U.S. manufacturing and mining industry. This will provide the input for multiple regressions to determine the industry variables that are responsible for any salary variation. The results of the regressions are presented in Chapter IV. The remainder of this introductory chapter will be devoted to a summary of the literature relevant to this study and to the implications of the findings of the study. Chapter II explains the theoretical model which the regression analysis is to test, and Chapter III examines the data that is the basis of the study. In Chapter V, the conclusions are put forth. In the period from 19^-7 - 19^5 white-collar workers in the U.S. (professional, managerial, office and sales workers) increased by 9-6 million while blue-collar workers (craftsmen, operatives and labourers) were decreasing by 4 million. In 1900 white-collar workers were 17.6 per cent of the labour force and 44.5 per cent in 1965. ' Figures from Manpower Report of the President and A Report on Manpower Requirements, Resources, Utilization and Training, United States Department of Labor, March, 1966. A. Other Studies on Interandustries Differences in Earnings There has been considerable research done on the interindustry variation in wage rates. The objective of these studies has been to relate wage levels to certain key industry variables. A study based on the hypothesis that wage levels are determined by a firm1s "ability to pay" was made by David R. Brown. Ability to pay was measured by six variables --labour productivity, concentration, ratio of labour costs to total costs, male - female ratio, seasonal stability and growth in employment., Analysis indicated that all of these industry variables except growth, were significant in- determining wage levels. Studies by Garbarino, Allen, and Goldner and Ross investigated the industry factors affecting wage changes, rather than absolute wage levels. The findings of these three studies indicate that high degrees of unionization and concentration are D.R. Brown, Expected Ability to Pay and Interindustry Wage Structure in Manufacturing", Industrial and Labor Relations Review, Vol. 16, No. 1 (October, 1962), pp. 45 - 62. J.W. Garbarino, "A Theory of Interindustry Wage Structure Variation", Quarterly Journal of Economics, Vol. 64, No. 3; (August, 1965), pp. 282 - 305-~ W. Goldner and A.M. Ross, "Forces Affecting the Interindustry Wage Structure", Quarterly Journal of Economics, Vol. 64, No. 2 (May, 1950), pp. 254 - 289. B.T. Allen, """"Market Concentration and Wage Increase: U.S. Manufacturing, 194? - 1964", Industrial and Labor Relations  Review, Vol. 21, No. 1 (April, 1968), pp. 353 - 555. _ 4 -two major factors in promoting rising wages. However, these results are of limited applicability to this -study as they do not deal with the actual level of wages. The effects of these two variables, concentration and unionization, upon wage levels, were also investigated in an 5 empirical study by Weiss. Regression results indicated that these two variables were significant in explaining wage levels for most occupational groups tested but only in the absence of other industry variables and personal characteristics. When individual personal characteristics such as education, age, race and area of residence were introduced into the regressions, the significance of the industry variables in most cases disappeared. Weiss concludes that in paying higher wages, concentrated industries get a better quality of worker and hence there is little or no monopoly profit accruing to labour. In surveying this literature, no firm conclusions can be drawn, except perhaps that the determination of wage levels is a complex process. No model with one or two simple variables can succeed in explaining wage variation. The studies on the determination of executive compen sation have been fewer in number and more limited in scope. One popular empirical exercise has been to compare firm size -^L.W. Weiss, "Concentration and Labor Earnings", American Economic Review, Vol. 56, No. 1 (March, 1966), pp. 96~^~117.. - 5 -and firm profitability as variables explaining the level of 6 executive, incomes. The results indicate that of these two, firm size is the more Important explanatory variable. This finding, however, is of little interest In the absence of a credible model. It is not surprising that firm size is related to salaries as it is likely a proxy for such variables as monopoly power, capital-labour ratio, profitability and unionization. In another study, the compensation of the top executive was found to be significantly and positively related to monopoly power, as measured by concentration and entry 7 barriers. This finding is of limited interest here for it deals only with, the single top executive in each firm. How ever, the hypothesis that was tested is based on the theory that, in the modern corporation, executives have the discre tionary power to distribute profits "to employees as well as to stockholders. The fact that the findings support this theory indicate that it may be a line of thought worth pursuing. D.R. Roberts, Executive C ompensat ion, Illinois, Free Press of Glencoe, 1959, p. 50, 62>. J.W. McGuire, J.S.Y. Chin, and A.O. Elbing, "Executives, Sales and Profits", American Economic Review, Vol. 52, No. 4 (September, 1962), pp. 753" 61. O.E. Williamson, Managerial Discretion and Business Behavior", American Economic Review, Vol. 53, No. 5 (December, 1963), pp. 1033 - 1057. - 6 -B. Implieations of This Study The possible implications for economic theory that could he drawn from the findings of the present study are several. If it is found that firms possessing monopoly power operate at greater than competitive costs, due to higher salary payments, then conclusions- could 'be drawn about the cost to society of imperfect competition. There would be a transfer of income from the consumers of the monopoly product to the salaried employees. There would also be a misallocative effect, greater than that commonly attributed to monopoly. Harberger, for instance, in his estimate of the'misallocative cost of monopoly, assumed that there exists no difference in costs g under monopoly or competition. If in fact monopolistic firms incur higher costs, then the misallocation of resources due to monopoly will be even greater than Harberger calculated. Such a finding would also cast light upon the opposing managerial discretion and profit maximizing theories of firm behavior. If potential monopoly profits are absorbed by the firm in the form of higher salaries, then it could be said that management is exercising its discretion over profits. Failure to find a significant relationship between salaries and monopoly power would lend support to the theory of profit maximization. 8 A.C. Harberger, "Monopoly Power and Resource Allocation", American. Economic Review, Vol. 44, No. 2 (May, 1954), pp. 77 ~:r^7. - 7 -Finally, if evidence is found of significant interindustry variation in salaries, conclusions could be drawn about executive quality and mobility between industries. If salary variation was found to be completely accounted for by personal characteristics of the employees, then it could be said that the industries paying higher salaries are getting an employee of higher quality. If there is no significant interindustry variation, then employee quality can be said to be' uniform and mobility high. CHAPTER II THE MODEL The hypothesis to he tested is that for given occupational groups the interindustry variation in salaries is a function primarily of three industry variables -- monopoly power, growth, and the capital-lab our ratio. A. Monopoly Power The reason for believing that industries possessing monopoly power, and hence the potential for monopoly profits, pay higher salaries, lies in the theory of managerial discre tion. There Is considerable support for the view that the management of the modern corporation is responsible to the shareholders for a minimum level of profits only, and beyond that level has considerable discretion in dealing with profits.' The phenomenon that is responsible for this departure from con ventional profit maximizing theory of firm behavior is the ^O.E. Williamson, The Economics of Discretionary  Behavior: Managerial Objectives in A Theory of the Firm, Englewood Cliffs, Prentice-Hall, 19b~5~. Carl Kays en, '""The Corporation: How Much Power? What Scope?", in The Corporation  in Modern Society, Edward S. Mason, ed., Cambridge, Harvard-University Press, 1959- W.J. Baumol, Business Behavior, Value and Growth, New York, Macmillan, 1959-- 8 -_ 9 -separation of ownership from management. Over the last several decades the trend has "been away from corporate ownership by one or more powerful individuals. Today the change is almost total with the vast majority of corporate stock in the hands of 2 financial institutions and small investors. Both these groups are primarily interested in maximizing their income through stock market dealings. Neither is in the business of managing corporations. Only if earnings fall below a certain minimum acceptable level can these stockholders be aroused. Provided this profit.level is maintained, management enjoys the interference free operation of the firm. It follows that any profits above the minimum level can be dealt with at the discretion of management. Thus, one obvious ivay for top management to exercise this discretion would be to pay themselAres higher salaries. There is reason to believe that this would in turn have a positive effect upon the earnings of white-collar employees 2 For evidence on the change from priva,te ownership to management control of U.S. corporations, see R.J. Larner, "The 200 Largest Nonfinaneial Corporations", American Economic Review, Vol. 56, No. 4, Part I (September, 1966), pp. 777 - T&7. 3 ^Baumol indicates that the minimum acceptable profit level is determined by competitive conditions In the capital market. It must permit sufficiently high divided payments and reinvestment in the firm to ensure that investors are content to continue holding the stock. Baumol, op. cit., p. ^1. 10 -in the lower stratas. Another way would be to pay generous salaries for the purpose of purchasing employee goodwill and public approval. Firms with considerable monopoly power might be particularly conscious of the value of a good public image. It also follows that the function of management in hiring and retaining good employees is simplified through the paying of higher .salaries. The pressure to keep the salary budget down will be less than in-competitive industries and hence these costs might be allowed to creep up either deliberately or through inattention. 1. Measurement of Monopoly Power For the purposes of this study, monopoly power will be measured by concentration and barriers to entry. Generally the higher the concentration ratio the stronger will be the recognition of mutual dependence among firms. Hence concen tration is a good measure of the ability of firms to tacitly agree to charge a monopoly price. Obviously it is not an infallible measure. Other industry variables such as the homogeneity of the product, the price leadership situation, ^H.A. Simon, "The Compensation of Executives," Sociometry, Vol.- 20, No. 4 (December, 1957),.pp. 32 - 35. ^The theory of managerial discretion is concerned with the ability of the firm to reap monopoly profits, rather than the action of doing so. Management may pass on discre tionary profits .to the consumer as well as to the employees, hence less than the profit maximizing price would be charged. The objective of such strategy might be to maximize sales or else to purchase public approval. - 11 -the state of inter-firm communications and the threat of new entrants, all will have an effect upon the pricing policy of the industry. Of these, the threat of entry Is the only variable that lends itself to measurement. In Joe Bain's authoritative work on this subject, 6 barriers to entry are classified into four types. These are product differentiation, economies of scale, absolute cost advantages .and capital requirements. The absolute cost barrier, which includes such established firm advantages as patent control and monopolization of resource supplies, is the least important of the four and probably the most difficult to 7 measure. Hence, an attempt will be made only to measure the other three types of-barriers, although as it turns out, this attempt will be only partially successful. This will be explained in detail in Chapter III. ^J.S. Bain, Barriers to New Competition, Cambridge, Harvard University Press,, 1963, pp. 167 - 169. 7Ibid., p. 149. B. Industry Growth Industry growth is thought to he a determinant of higher salaries for two reasons. Firstly, growth usually means higher profits. When demand is rising for the product of an industry, entry of new firms is never instantaneous. Hence, supply is to a greater or lesser degree inelastic. Thus, until supply catches up with demand, the firms producing in the growth Industry are in a position to reap higher than normal profits through their ability to charge higher prices. Manage ment has the discretion to expend these profits on higher salaries in the same way it does with monopoly profits. The second reason that growth results in higher salaries follows from the increased derived demand for employ ees in the growth industry. To the extent that workers are not perfectly mobile, the growth industry will have to pay higher salaries to entice them to move. Of more importance in determining higher salaries will be the lag in supply of, relative to demand for, workers trained in the skills of the growth industry. Thus, the price of the workers will be bid up until supply catches up with demand. Industry growth will be measured by change in employment over a ten-year period. • - 13 -C. Capital Labour Ratio Those industries with a high capital labour ratio may be expected to pay higher salaries because the pressure to keep wages and salaries down will be weaker. Payroll costs will be of less concern to management compared to costs of securing capital or of purchasing equipment than they would be in a more labour .intensive industry. Those firms that are. capital intensive will be more concerned with avoiding inefficiency or disruptions due to employee dissatisfaction. Hence, it is probable that salary payments will be more generous. This variable will be measured by the ratio of total assets to number of employees. D. Conflicting Theory It may be true that, because of their large size, firms with monopoly power in the product market also possess 8 monopsonistic power in purchasing labour. In this situation, neoclassical theory would lead to the conclusion that monopolis tic firms pay lower, not higher, wages. The firm equates the marginal revenue product of labour to the marginal factor cost of labour in determining the quantity to employ. For the 8 Joan Robinson, among others, believed that monopsony power might often exist at the level of the firm. See Joan Robinson, The Economics of Imperfect Competition, London, Macmillan, 1961, p. 296. ' — - 14 -monopolist-monopsonist both these values will be less than under competition; hence he will pay lower wages. This is shown in Figure 1 where, under competition, a quantity OC of labour is employed at a wage of OD. The monopolist employs only OM units of labour at wage OP. Price of Labour , MFC Supply Q M C • Quan"ki"ky or> Labour FIGURE 1 Wage Determination Under Monopoly and Competition This prediction of Lower wages under monopoly is based upon the existence of an upward sipping supply curve of labour to the firm. This in turn depends mainly upon the assumption of considerable immobility of labour. It is likely that, among white-collar workers, a sizeable proportion, sufficient to ensure elastic supply curves, will be mobile. - 15 -•Neoclassical theory also depends upon the assumption of profit maximization. It has been alleged earlier in this chapter, that in general the behavior of monopolistic firms diverges from strict profit maximization. Under these conditions the determination of wages and salaries cannot be completely explained by marginal productivity theory. Probably a more serious threat to this study's proposed model of salary determination is the question of mobility of workers. The extent of interindustry salary variation will be inversely related to the degree of mobility of white-collar workers. A sufficiently high degree of mobility may cause the salary variation to be statistically insignificant. E. Unionization No study of the determination of earnings can ignore the variable of unionization. However, for salaried employees, its effect Is difficult to assess. Salaried employees are not themselves unionized to a significant extent. One study on the subject estimates unionization of white-collar workers at about 11 per cent of Q potential. This figure Is roughly confirmed by a more thorough q vAdolf Sturmthal, White-Collar Trade Unions: Contemporary Developments in Industrial Societies, Urbana, University of- Illinois Pres"s^ 1966^ p] 3387 - 16 -study that estimates unionization figures for the white-collar groups that are analyzed in this study. The figures are: 18 per cent for Clerical and Kindred Workers, 9 per cent for Professional, Technical and Kindred Workers, and 12 per cent for Sales Workers.10 These figures are negligible, for the purposes of this study, because the majority of these unionized white-collar workers are not in manufacturing industries. Most of them are In the retail trade or telephone industry, or they are musicians, railway clerks or post office workers. None of these occupations or industries are within the scope of this study. The problem of unionization, however, does not end here. It is possible that the degree of unionization of wage earners in an industry may affect the salary levels of the white-collar workers. Unfortunately, neither theory nor empirical work has much to say about this problem. Conclusions on the importance of this relationship can only be tentatively drawn from what is known about the effect of unionization upon interindustry wage levels. One of the most complete studies on unionism and wages indicates that unionism is not a major factor In explain ing the interindustry wage structure. It is stated that most of B. Solomon and R.K. Burns, "Unionization of White-Collar Employees: Extent., Potential and Implications", Journal of Business, Vol. 36, No. 2., (April, 1963) , pp. 151. - 17 -the wage dispersion among industries must he accounted for by-other factors.11 If unionism has only a minor effect upon the interindustry wage structure, It can be assumed that its effect upon salaries will be even less. Thus, although some tests of the Influence of unionism will be attempted, it is assumed to be a factor of negligible importance in the determination of salaries among industries. 11H.G. Lewis, Unionism and Relative Wages in the United States. Chicago, University of Chicago Press, ±9b~3, ^oii '—'—-— CHAPTER III DATA SOURCES AND MEASUREMENT OF REGRESSION VARIABLES A. General Description of Data Figures for salary, occupation, industry and various personal characteristics are reported for 180,000 individuals in the l/lOOO sample of the i960 U.S. Census of Population.1 This data is available on magnetic tape and when matched with the relevant industry variables, provides the regression input. The study covers a cross section of 63 U.S. manufac-2 turing and mining industries, all unregulated. The data for the industry variables is drawn from three sources. Figures for growth and concentration are taken from the appendix of Weiss' article. This appendix is not printed with the journal article, 3 but may be obtained from the author. Advertising sales and 4 assets data come from U.S. income tax data.. Finally, figures "'"Bureau of the Census, U.S. Census of Population and Housing: i960 l/lOOO and l/lOOOO, Two National Samples of the Population of the United States. 2 See Table I for a list of these industries. •5 See Footnote in Weiss, _op. cit., p. 101. Internal Revenue Service, Statistics of Income  1959-1960, pp. 58 - 60. - 18 -for number of employees and number of establishments are taken from the 1958 U.S. Censuses of Manufacturing and of Mining.D It is evident that there is a discrepancy in that the years for which the various data is reported do not exactly coincide. The figures from the population census are for 1959; while those from the manufacturing census are from 1958. The remaining industry figures are reported for the accounting periods ending between July 1959 and June i960. Presumably then, the average reporting period for these figures is the year 1959- . Another minor discrepancy exists because salary figures are for 1959; while all other data for the individuals refers to April i960, the time at which the census was taken. Thus, if an individual changed occupation or industry in 1959, or the first four months of i960, the data would be in error. A more serious problem with the data, is that some figures are reported for a detailed industry classification, but others are available only for a major industry breakdown. This problem arises because the population census classifies industries according to a unique three digit code that does not correspond to the Standard Industrial Classification 5 ^Bureau of the Census, United States Census of  Manufacturers: 1958, Volume I. Bureau of the Census, United  States Census of Mineral Industries: 1958, Volume I. - 20 coding used "by the other sources. Since there exists no other comparable source of earnings data, there is no choice but to attempt to match industry figures on the three digit basis. Table I shows the results of this matching. It can-be seen that concentration and growth figures are available for all 63 industries. However, for the other variables, figures for the major industry groups must be used for the more detailed classifications. While this inconsistency in industry classification obviously detracts from the quality of the data, the variables should still be meaningful. It does not seem unreasonable, for instance, to attribute the advertising-sales .or 0340113,1-13,13our ratios for all textiles to the individual textile industries. Further detail on the development of the industry variables is given in Appendix I. B. Detailed'Description of Variables Following is a detailed account of.the measurement and content of each of the variables to be used in the regressions. TABLE I INDUSTRY VARIABLES IND NAME 126 Metal Mining 136 Coal Mining 146 Crude Pet. & Nat. Gas 156 Non-Metalic Mining CONC 58 29 32 27 GROWTH 102 44 112 114 ADV 0.0003 0.0009 0.'0020 0.0038 CAP-LAB $'000 45.53 n.o4 22.85 20.18 CAP/REQS $'000 '1886. 284. 386. 323. Lumber and Wood Products 206 207 208 209 Sawmills, Planing Mills, Millwork Misc. Wood Products Furniture and Fixtures 38 15 21 15 102 74 98 117 o.oo4o o.oo4o o.oo4o 0.0122 10.29 10.29 10.29 7.05 i4i, i4i, i4i. 24l, ro Stone. Clay and Glass Products 216 Glass and Glass Products 59 120 0. 0070 15. 73 217 Cement, Concrete, Gypsum and Plaster 32 168 0. 0070 15. 73 218 Structural Clay Products 29 . . 108 0. 0070 15. 73 219 Pottery and Related Products 39 • 82 0. 0070 15. 73 236 Misc. Non-Metalic Minerals 48 . 155 0. 0070 •15. 73-580, 580, 580, 580, 580, 237 238 239 Primary Metal Industries Blast Furnaces, Steel Works, Etc, Other Primary Iron and Steel Primary, Non-Ferrous Metals 56 49 53 94 io4 139 0.oo44 o.oo44 0.oo44 23-97 23.97 23-97 4077. 4077. 4077. TABLE IND NAME Fabricated Metal Products 246 Cutlery, Hand Tools, Hardware 247 Fabricated Structural Metal Prods. 248 Misc. Fab. Metal Products Machinery Except Electrical 256 Farm Machinery 257 Office, Computing & Acctg. Mach. 258 Misc. Machinery 259 Electrical Machinery 267 Motor Vehicles and Eqpt. Transportation Equipment Except Auto 268 Aircraft and Parts 269 Ship and Boat Building and Repair 276- Railroad and Misc. Transp. Eqpt. Instruments and Related Products 286 Professional Eqpt. and Supplies 287 • Photographic Eqpt. and Supplies 289 Watches/Clocks, Etc. " -296 Miscellaneous Mfgrs. (Continued) CONC GROWTH ADV CAP-LAB ' CAP/REQS, $'000 $'000 38 111 O.OO89 11.15 475. 39 135 0.0089 11.13 ' 475. 36 l8l O.OO89 11.13 475. 44 . 78 0.0105 14.94 675. 72 163 0.0105 14.94 675. 39 131 0.0105 14.94 675. 51 " 176 0.0173 12.74 1767. 72 100 0.0081 78.67 7254. 52 . 258 0.0028 9.69 2242. 45 • 144 0.0028 9.69 2242. 52 120 0.0028 • 9.69 2242, 51 222 . 0.0226 15.33 1289. 77 137 0.0226 15.33 ' 1289. 45 85 0.0226 15.33 ! 1289. 30 103 o.oi84 • 7.91 317. TABLE IND NAME Food and Kindred Products 306 Meat Products 307 Dairy Products 308 Canning and Preserving 309 Grain Mill Products 316 Bakery Products 317 Confectionery & Related Products 318 Beverages 319 Misc. Food Preparations 329 Tobacco Products Textile Mill Products 346 Knitting Mills 347 Dying and Finishing 348 Floor Coverings- Except Hard Surfaces 349 Yarn, Thread and Fabric 356 Misc. Textiles Fabricated Textiles 359 Apparel and Accessories 367 Misc. Fab. Textiles Paper and Allied Products 386 Pulp, Paper, and Paperboard 387 Paperboard Containers 389 Misc. Paper and Pulp Products ontinued) CONC GROWTH ADV ^OOQ8 K)0(f' 28 119 0.0192 12.07 499 56 154 0.0192 12.07 499 34 137 0.0192 12.07 499 49 123 • 0.0192 12.07 499 j?6 134 0.0192 12.07 499 26 101 0.0192 12.07 499 34 105 0.0488 25.45 944 49 126 • 0.0192 12.07 499 68 93 0.0547 39-14 6560 18 108 0.0062 9.90 1163 44 87 0.0062 • 9.90 1163 47 66 . 0.0062 9.90 1163 27- 67 0.0062 9.90 1163 4l 98 0.0062 9.90 1163 11 110 0.0098 3.82 154 26- 135 0.0098 3.82 154 45 128 o.oo84 17.96 1893 27 143 o.oo84 • 17.96 1893 4l 106 -• o.oo84 17.96 1893 TABLE I (Continued IND NAME . C0NC Printing and Publishing 396 Newspapers 52 398 Printing & Publishing Exc. Newsp. 26 Chemicals and Allied Products 406 Synthetic Fibres 74 407 Drugs and Medicines 4408 Paints, Varnish and Related Products 25 409 Misc. Chemicals and Related Products 42 Petroleum and Coal Products 4l6 Petroleum Refining 47 419 Misc. Pet. & Coal Products 50 Rubber and Plastics Products 426 Rubber Products 51 429 Misc. Plastic Products 2Leather and Leather Products 436 Leather Tanning, Etc. 20 437 Footwear Except Rubber 8 438 Leather Products Except Footwear 21 GROWTH ADV CAP-LAB CAP/REQS. $'000 . $'000 130 0.0076 9.43 230. 135 0.0076 9-43 230, 109 ' 0.0379 33.19 2052, 196 • 0.0379 33.19 2052, 117 0.0379 ' 33.19 2052, 134 0.0379 33.19 2052. 98 o.oo46 221.85 24719. 136 o.oo46 • 221.85 24719. 114 0.0164 i4.il 1100, 205 0.0164 i4.ii 1100, 69 0.0112 5.28 407. 97 0.0112 5.28 • 407. 90 0.0112 5.28 407. - 25 -1. Salary For each individual worker, private wage and "salary income, including bonuses, is reported by the census. This is to be the dependent variable in the regressions. Individuals were selected for the regression sample in such a way as to minimize the possibility of errors in the data. Persons with any self employment income were rejected as their' salary income could be distorted in two ways. It might be low if they are working at their salaried employment on a casual or part-time basis. Conversely, it might be inflated if they were employing themselves in a salaried position. Stigler has pointed out that profits of closely held corporations may often be withdrawn in the form of executives' salaries. Those individuals who had not worked at least 50 weeks in 1959 were also rejected. It is likely that many individuals who changed occupation or Industry in 1959 would have been voluntarily or involuntarily unemployed for at least a brief period of time. Hence, by rejecting such persons, data errors of this type will be reduced. G.J. Stigler, Capital and Rates of Return in  Manufacturing Industries. Princeton, Princeton University Press 1963, pp. 125"'- 127. - 26 -2. Concentration The figures used are 1958 four firm concentration ratios calculated on product shipments. The ratios have been scaled up for those products selling in local or regional product markets. It was found, by Weiss, that on the average, national concentration ratios for. products sold on national markets were I.98 times the national concentration ratios for those sold on regional markets and 2.67 for those sold on local markets. Hence the applicable concentration ratios were scaled up by these factors. 3. Product Differentiation This barrier to new entrants can be attributed to several sources, notably customer service, product design and advertising intensity. However, in his study of twenty industries, Bain found that advertising" is consistently the. 7 most important factor In establishing product differentiation. Therefore, it is the ratio of advertising expenditure to sales that is used as a measure of product differentiation. It should be noted that while advertising is important in establishing product differentiation, and hence monopoly power, there is also a relationship in the opposite direction. Firms possessing monopoly power can, in many cases, be expected to spend more on advertising effort than competitive firms. The greater is the product differentiation and concentration ^Bain, _op. cit. , pp. 122 - 24. - 27 -in an industry, the more each firm will stand to gain from advertising expenditure, as there will be less "spillover" to other firms. Also, due to the uncertainties of price competition, oligopolists are often hypothesized to prefer advertising and other sales increasing expenditures as a form of competition. 4. Economies of Scale Considerable effort was applied to the problem of measuring the economies of scale barrier to entry. The final conclusion reached was that a meaningful measure of this variable cannot be derived with available techniques and data. Economies of large scale become a significant barrier to entry if the new entrant, in establishing an optimal sized plant, or firm, must secure a sizeable portion of the market. In this situation, the potential entrant is faced with two choices. He. can operate a sub-optimal sized plant or firm, at higher than minimum attainable costs, or he can enter producing the optimal output and in doing so lower the selling price. Secure in the knowledge that potential competition is at a, dis advantage, the established firms can charge a price greater than minimum attainable cost. They possess a degree of monopoly control over supply. While this concept is straightforward, measurement is not. There is no agreement on criteria for choosing the optimal or minimum efficient size of plant or firm. In addition, it is virtually impossible to design a measure that - 28 -takes account of the shape of the industry cost curves. This is a serious shortcoming for knowledge of the rate of decline of unit cost as minimum efficient size is approached is essential in evaluating the importance of the scale harrier. Finally, the problem is further complicated by the concept of diseconomies of large scale. This can presumably be avoided at the plant level by duplicating optimal sized plants. However, in the .case, of firms, it is quite conceivable that diseconomies do exist. Empirical evidence to clarify this problem is non-existent.^ Despite these problems, there have been attempts to measure the barriers to entry due to economies of scale. Presumably, because of the question of diseconomies of large firms, these attempts have concentrated on plant economies of scale. Two of these measures were investigated in detail for this study. The first, by. Bain, takes advantage of the fact that the U.S. Census of Manufacturing gives industry data, grouping plants into size classes, size being measured by number of Q employees. His measure is arrived at by expressing the value added of the average sized plant .in the largest size class as a percentage of industry value added.- The weakness of this measure is that the choice of the optimal or minimum efficient g See Bain, _op. cit. , pp. 6l - 62. q _ .See Bain, _op. cit. , pp. 69, 75. - 29 -plant size is essentially arbitrary. This value was calculated for the industries in this study, hut the results appeared to he meaningless as a measure of the- economies of scale harrier. For instance, the calcula tions give a slightly higher measure of economies of scale to the major industry group textiles', than to transportation ^ 10 equipment. In the largest size class, textiles had ten establish ments producing a value added of $236,5^5,000 with total Industry value added of $4,857,638,000. This gave an economy of scale measure of .0049. For transportation equipment the corresponding figures were .134 establishments producing $9,244,433,000 value added compared to $15,283,694,000 for the industry for a measure of .0045. Textiles is the textbook example of a competitive, ease of entry industry and trans portation equipment is dominated by Industries manufacturing automobiles, airplanes and ships in which economies of scale in production, are of great importance. Clearly, Bain's measure is of no value for the industry data classifications used in this study. The other measure of economies of scale given consideration was that used in a recent study on the sources of "^See U.S. Census of Manufacturing 1958, Vol. 1, p. 2 - 3. - 30 -monopoly power. " This measure is derived by taking the average plant size among the largest plants accounting for 50 per cent of industry output and dividing by total output in the relevant market. This measure suffers from all the theoretical short comings listed above. It ignores possible economies of large firms.. Its choice of a minimum efficient plant size is apparently arbitrary. It does not attempt to measure the rate of decline of unit costs as plant size increases. In vieiv of these shortcomings and of the difficulty --if not impossibility -- of calculating the measure with available data, no attempt at calculation was made. It was concluded that the economies of scale barrier could not effectively be measured and the idea was abandoned. 5. Capital Requirements Measurement of the capital requirements barrier to entry generally follows from the economies of scale measure. It is the amount of capital required to enter the industry with a plant or firm of minimum efficient size. This is said to be a barrier.to entry on the grounds that the capital market may charge higher interest rates to potential entrants as compared to established firms. It Is thought that as the amount ^S. Comanor and T.A. Wilson, "Advertising Market Structure and Performance", The Review of Economics and  Statistics, Vol. 44, No. 4, "(November, I967), pp. 423 - 440. - 31 -of capital gets larger, this difficulty or extra cost of obtain-12 ing capital for entry increases. Without a value for the economies of scale variable, precise measurement is difficult. However, an approximation of the capital required to enter with a single plant was calculated by dividing total assets by the number of plants in 13 the industry. ' This is not meant to be an absolute measure of the capital required to enter an industry. It suffers from the shortcomings of the previously discussed economies of scale measures. However, as it measures the amount of capital required to build and operate an average sized plant, it should be a meaningful approximation of the relative capital barriers among industries. 6. Capital - Labour Ratio This is measured by dividing total industry assets by number of employees. 7. Growth The measure used is i960 industry employment divided by 1950 industry employment. 12 Bain, _op. cit., pp. 145, 146. ] 3 The U.S. Census of Manufacturing gives industry data, on establishments rather than plants. An establishment is a producing location and therefore a multi-plant site shows up as only one establishment. It is not a vital distinction and in this, as'in most studies, the two terms are used more or less interchangeably. - 32 -8. Occupation and Sex These are two of the most important factors determin ing salary levels. It is desirable to eliminate salary variation due to these two variables when regression analysis is applied to the model. This is accomplished by grouping the regression observations into samples that are homogeneous with respect to occupation and sex. Occupations to be tested were selected from the four white-collar groups in the population census. The groups are Professional Technical and Kindred Workers, Managers, Officials and Proprietors Except Farm, Clerical and Kindred Workers and Sales Workers. An occupation could be selected for testing only if it would provide a sample with individuals distributed over a reasonable number of the manufacturing and mining industries to be tested. Thus, a majority of the occupations such as architects, dentists, bank tellers and insurance agents, could not be used. Also, only occupations that would provide sufficient observations for meaningful regression analysis could be chosen. Thus, many occupations such as economists, civil engineers and statisticians could not be tested as there would have been less than ten observations in the final sample. This process of selection and elimination yielded the fifteen occupations shown in Table II. Four of these occupations, draftsmen, mechanical engineers, managers, officials and proprietors, and typists correspond to groups tested in Weiss' TABLE -II OCCUPATIONAL GROUPINGS FOR REGRESSION ANALYSIS OCCUPATION NAME Accountants Chemists Draftsmen Industrial Engineers Mechanical Engineers Sales Engineers Engineers- (NEC)* Personnel and Labor Relations Workers Professional, Technical and Kindred Workers (NEC) Managers, Officials and Proprietors Bookkeepers Secretaries St enographer s Typists Salesmen and Salesclerks (NEC) SEX NO. OBSERVATIONS Male 100 Male 53 Male llil-Male 58 Male 110 Male 25 Male 33 Male 22 Male 56 Male 532 Female 99 Female 239 Female 50 Female 8 Male 322 1871 *NEC means Not Elsewhere Classified - 3h -study. Since he was most interested in the determination of wages, rather than salaries, most of the occupations tested in his study came from the labouring groups. For each occupational group either all males or all females were selected for the regression sample. Most occu pations were almost completely homogeneous in one sex or the other, so it was decided to make them completely so. This cost little in the way of sample size and eliminated distor tions due to an important and unwanted variable. 9. Pe r s onal Charac t e ri s11c s For each individual employee, six personal character istics were run as independent variables in the regressions, in an attempt to account for some of the salary variation. The characteristics are education, age, southern residence, rural residence, size of place of residence and race. Educa tion is measured by the highest grade of school or year of university completed, but there is not a unique value of the variable for each grade and year. There is a value of three for those whose highest grade completed was one, two, three or four, a value of four for those whose highest grade was five or six, and so on up to twelve for those who finished five or more years of college. Southern residence is a dummy .variable with a, value of one if the individual resided in the south, zero otherwise. RuraT residence is a dummy variable with a value 14 Weiss, op. cit., pp. .102 - 10 J>. - 35 -of one if the individual resided outside an urbanized area, or outside a. Standard Metropolitan Statistical Area, zero other wise. Size of place is a variable with a value ranging from one for a rural farm to twelve for a place of one million or more. Race is a dummy variable with a value of one if the individual is white and does not 'have a Spanish surname, zero otherwise. 10. Unionization This measure is taken from Weiss' study. It is the percentage of employees in establishments where more than half the production workers are covered by collective bargaining agreements. CHAPTER IV REGRESSION RESULTS Multiple regression analysis was used to test the hypothesized relationship between salaries and industry monopoly power, capital-labour ratio and growth. Regression input was created by matching industry data from the various sources with the data for individuals from the population census. Each person was assigned the values of the variables for the industry - in which he was employed. The resulting data record containing salary, personal characteristics and industry parameters Is a regression observation. From the original census sample of 186,000 records, 1871 such observations were selected over .15 occupations. This constitutes a small, very select sample. As explained in Chapter III, to be selected an individual had to satisfy the conditions of being employed In one of the 63 industries in Table I, having worked 50-52 weeks in 1959, having no self-employment income, as well as being in one of the designated o c c up atIon-sex gr oups„ Two basic sets of linear regressions were run, one with only the industry variables and the other with personal characteristics added. The two regression equations were: - 36 -- 37 -'1) S = b. + bnC + b^AS + b_,CR + b,,CL + brG o 1 2 5 4 J? "r 5 (2) S = b + b.,C + b„AS + bvCR + b,,CL + bcG + x 1 o 1 2 3 4 5 b^-A + b,7E + bQRS + b„RR + b-, „SP + bn-,R o ( o 9 10 -11 where: S = Salary C = Concentration AS = Advertising - Sales Ratio CR ~ Capital Requirements CL = Capital - Labour Ratio • ' G - Growth E = Education A = Age RS = Dummy with value 1 if residence in south RR = Dummy with value 1 if rural residence SP = Size of place of residence R = Dummy with value 1 if race is white According to the reasoning of the model in Chapter II, the five industry variables should all have positive coeffi cients. Growth, the capital-labour ratio and the three monopoly power variables, concentration, advertising/sales and capital requirements, are all hypothesized to have a positive effect in the determination of industry salaries. No predictions were made as regards the coefficients •of the other variables as they were inserted in the regression equations only to account for some of the variation in salaries. However, it would be expected that the coefficients of educa tion, age, size of place of residence and race would be positive and those of the two residence dummy variables negative. The first problem encountered upon running the regressions was the discovery of a critically high degree of collinearity between the capital'requirements and capital-labour variables. For 12 of the 15 occupations, the simple correlation coefficient of these two variables was greater than .9- Collinearity was expected but not to this degree. In order to avoid extremely high standard errors, it was arbitrarily decided to drop the capital requirements variable from the regression runs. No other problems of multicollinearity were encountered. The second discovery made from the initial regression runs was that there was very little difference in the coeffi cients of the variables between equations (.1) and (2). The only important difference between the two sets of regressions was in the amount of variation accounted for. With equation (l) 2 the values of R were consistently less than .1, but with the personal character j.sties added it ranged between .1 and .5. This result was expected. In the white-collar occupational groups salaries will vary greatly according to personal characteristics, -particularly experience and education. With much of this variation accounted for by the age and years in school variables, the regression results become more meaning ful. - 39 -Along with the capital requirements variable, race and size of place were dropped from the regression runs as neither of them accounted for any of the variation In salary. Thus, the independent variables in the remainder of the regression runs were concentration, advertising/sales ratio, capital-labour ratio, growth, education, age, southern residence and rural residence. The results of the multiple regressions are summarized in Table III. For each occupation only the coefficients of the industry variables are shown. Parallel runs were made with capital requirements in place of capital-labour to check that the results would not be substantially different. The presence of one or two asterisks indicates that the coefficient is significant at the five per cent or one per cent level on a two-tailed T-test.1 Table VI In Appendix II presents these regression results in full detail showing the coefficients of all the variables along with standard errors, F- probabilities 2 and values of R . According to the model in Chapter II, the signs of the coefficients in Table III should all be positive. Clearly, the results do not offer unambiguous support for the hypothesis. The signs are mixed and few of the coefficients are significant. Careful analysis of the coefficients for each variable is 'necessary in order to formulate any conclusions. "'"See Appendix II for an explanation of the test. TABLE III MULTIPLE REGRESSION RESULTS - COEFFICIENTS OF INDUSTRY VARIABLES SALARY RELATED TO 4 INDUSTRY AND 4 PERSONAL VARIABLES OCCUPATION NUMBER OBSER. CONCEN TRATION ADVER TISING ' CAPITAL) ' (REQ. ) CAPITAL LABOUR GROWTH Ac c ountant s 100 - l.8o - 10470. - .04 - 4.20 - 6.49 Chemists 53 + 7.91 + 978.2 + .17* + 18.56* - 3.36 Draftsmen 114 + 24.69* + 18330. - .18 + .42 - 3.93 Industrial Eng. 58 + 15.74 - 99260.* + .35 + 18.76 - 4.84 Mechanical Eng. 110 + 19.24 + 26330. + .05 + 2.62 Sales Engineers 25 + 87.12 -129500. -2.13 -120. -29.0 Engineers (NEC) 33 +117.72 -130100. - .38 .- 95.51 -21.19 Personnel and Laoour Relations 22 + 30.83 + 82480. - .03 - 3.55 - 2.48 Profess., Tech. & Kindred Workers 56 + 4.58 + 27050. - -57 - 2. 23 - 4.20 Managers, Officials & Proprietors 532 - 10.47 + 17780. + .03 - 1.39 - 4.oo Bookkeepers 99 - 1.91 - 5812. - .02 - 2.19 + 5.71 Secretaries 239 + 21.35** - 5552. + .04* + 4.99* + .88 Stenographers 50 + 14.34 - 11220. - .03 - 3.5^ + 1.42 Typists 58 + 29.55* + 2105. - .04 - 3.89 - 2.25 Salesmen and Salesclerks 322 - 57.77* - 9563. + .12 + 10.63 - 5.7 (Regressions were run with Capital Requirements in place of Capital Labour - only the one coefficient is shown as there was a negligible change In the other coefficients.) - 41 -1. Cone entration This is the one variable that offers strong support for the model. The coefficient of concentration is positive 11 out of 15 times and three of these times it is significant. •A binomial test was applied to measure the probability of having at least 11 sample coefficients of the same sign.if the coefficient of concentration was in fact zero. Assuming a probability of .5 for either a positive or a negative sign, the probability of having at least 11 positive or 11 negative coefficients is .119. However, on applying a one-tailed test, that is, assuming the coefficient of concentration cannot be negative, this probability falls to .059.- The null hypothesis can be rejected at the six. per cent level of significance. The unfortunate result of a significant, negative coefficient in the Salesmen and Salesclerks sample can possibly be explained by the composition of this occupational group. Salesmen will generally earn considerably higher salaries than salesclerks. Yet this differential, is unlikely to be explained by age or education. It is possible that the firms in the more concentrated industries employ a greater proportion of sales clerks to salesmen, hence the inverse relationship between concentration and salaries. Taking this factor into consideration, along with the results of the binomial test, the regression results can be said to offer support for the hypothesis that industry concentration has a positive effect upon salaries. • 2. Advertising The results for this variable are surprisingly poor. Advertising/sales is a simple, effective measure of advertising intensity and should be a good proxy for the product differentia tion barrier. However, there are eight negative coefficients in the results and one of them is significant. The hypothesis that product differentiation is a factor in explaining salary rates must be rejected. 3. C ap 11 al Requ 1 rement s Although this variable Is twice significant with a positive sign, it has a negative sign nine out of fifteen times. The null hypothesis of no relationship between the capital requirements barrier to entry and salaries cannot be rejected. However, in view of the lack of precision in measuring this variable, no firm conclusions can be drawn from these regression results. 4. Capital-Lab our Ratio The results for this variable were the same as for capital requirements. Again, the conflicting positive and negative coefficients prevent rejection of the null hypothesis. Due to the extremely high degree of collinearity .between these two variables, their individual effects cannot be separated out. All that can be said is that it appears - 43 -that the amount of capital in an industry does not have an effect upon salaries. 5• Growth These results are strongly contrary to the hypothesi With 12 out of 15 negative coefficients, an hypothesis that growth has a negative effect upon salaries could he accepted. This is contradictory to most theory, hut the explanation for these perverse results probably lies in the method of measure ment of the variable. Change in employment was used as a measure of growth mainly because these figures were the only ones available for the detailed Industry breakdown. In retrospect, it must be considered a poor measure. It fails to take into account the introduction of labour saving innovations, thus understating the growth rate for the more progressive industries. This weakness can be illustrated through comparison with a similar growth measure based on value added. For two industries, Office, Computing and Accounting Machines, and Dairy Products, the growth rates calculated on i960 employ ment/1950 employment are 153 and 154 respectively, as shown in Table I. However, the growth for these two Industries based on 1958 value added/194? value added are 193 and 14?. 2 Value added figures are taken from Bureau of the Census, United States Census of Manufacturers: 1947 and I958. _ 44 -As a growth measure, value added may he distorted by relative price changes, but it would more accurately reflect change in demand, which is the relevant factor. Not only is change in employment a. weak measure of growth, but there is reason to think that it might be inversely related to salaries. Those industries paying higher salaries would be more intent on achieving growth with a minimal increase in employment. There would be greater emphasis on substituting other factors of production for labour. Growth measured by increase in employment would tend to be less for the higher paying industries. Thus, not only is change in employment an Inaccurate measure of growth, but it might well be biased towards showing a higher rate of growth for lower paying industries. In light of this measurement problem, little can be concluded about the effect of industry growth upon salary rates. 6. Unionization Regressions were run to determine what effect the degree of unionization of production workers has upon salaries. The results showed, as expected, that concentration and unionism are highly collinear. The correlation coefficients for these two variables ranged from .20 to .79 with a median of .50 for the 15 occupations. - 45 -When unionism was run alone against salary, results suggest that It is a significant explanatory variable. However, when unionism was run with concentration in the same multiple regression, only 8 out of 15 of its coefficients were positive and none of them significant. This problem of collinearity between concentration and unionism has hampered most research efforts to isolate 4 the effects of these-variables. The one researcher who wa„s apparently unconcerned with this problem of multIcollinearity was Weiss, He included concentration, unionism and the product of the two in his 5 multiple regressions. If is unlikely, however, that the effects of unionization and concentration can be determined through multiple regression analysis. Conclusions must be reached by-recourse to the a priori reasoning In Chapter II. There are strong reasons for believing that monopoly power and hence concentration will be important in explaining salaries/ How ever there is little theory or evidence to support the belief that the degree of unionization of production workers will See Table VII, Appendix II, for results. ^See H.G. Lewis, Unionism and Relative Wages in 'the United Stales, The University of Chicago Press, Chicago^ 1963, "'Weiss, op. cit., p. '98. - 46' -significantly affect the salaries of white-collar employees. Therefore, it seems logical to attribute, to concentration, most of the variation in salaries that is explained by these two variables. 7. Comparison With Weiss It is of interest to compare the results of this study-to those of Weiss, since the data and to a certain extent the approach, of the two studies are similar. Although Weiss concentrated on hourly paid wage earners, he did run regressions for four salaried occupations that are covered by this study. They are draftsmen, mechanical engineers, managers, officials and proprietors (NEC) and typists. Regression coefficients for concentration, the main variable that the two studies have in common, are shown in Table IV. TABLE IV COMPARISON WITH REGRESSION RESULTS OF WEISS' STUDY SALARY RELATED TO CONCENTRATION ALONE AND TO CONCENTRATION ALONG WITH OTHER INDUSTRY AND PERSONAL VARIABLES - ONLY COEFFICIENTS FOR CONCENTRATION SHOWN WEISS' STUDY OCCUPATION Draftsmen Mechanical Engineers Managers . Officials, & Proprietors Typists No. Oh s r, 150 126 693 126' Me an Salary 5763 8998 IO980 2693 C one. Al one 44.36* (13O0) 24. 32 (26.73) 31.03 (24.05) 13.78 (10.52) Cone. With Other Variables 5.773 (72.08) - .5249 (152.7) 11. 23 ; 116.40) -13.14 (78.66) No. Ob s r. 114 110 532 THIS STUDY Mean Cone. Salary Alone 58 6389 O.lLOil s 1 J • 11840 3256 24.21* (10.72) 20,43 (30.40) -12.98 (26.81) 25.41* (9.69) Cone. With Other Variables 24.6Q* (12.33) 19.24 (31.83) ' -10.47 (28.99) 29.55* (11.23) •indicates significance at the 5% level. Standard errors are shown In brackets immediately below the coefficients. - 48 -The difference in sample sizes between the two studies is largely accounted for by the fact that employees who did not work at least 50 weeks in 1959 were dropped from this study, but not from Weiss'. This explains why Weiss' mean salaries are lower. Apparently for draftsmen, mechanical engineers and typists, the inclusion of these employees did not bias the sample for or against concentration. Looking at the coefficients for concentration when it was the only Industry variable in the regression, the results of the two studies are almost identical. For the managerial occupation group, the results are noticeably different. This discrepancy may be due to the fact •that all individuals with any self-employment income were dropped by this study. Many of these excluded persons were employed by small firms in unconeentrated industries and received little or no salary income. The inclusion of these observations in the regressions would bias the results in favour of a stronger relationship between salaries and con centration. It is not clear exactly how Weiss handled this problem. If he did not drop all employees with self-employment income, then his results would be different, and perhaps biased. The results for the two studies are decidedly different, when the concentration is run with all other variables. Part of this difference is probably due to the fact that the two studies included somewhat different industry and _ 49 -personal variables. Most of the difference, however, is likely-due to the fact that Weiss included the three collinear variables, concentration, unionization and their product together in the same regression equation. This is confirmed by the very high standard error for concentration. CHAPTER V CONCLUSIONS The model that this study set out to test predicts that interindustry salary variation is determinated, to a significant extent, -by industry monopoly power, capital-labour ratio, and growth. The results of the linear regression analysis do. not lead to a simple conclusion of acceptance or rejection of the model. Rather, they suggest tentative acceptance of certain aspects of the model, rejection of others and for some of the hypotheses the evidence is simply in conclusive. This failure to establish conclusive evidence, either for or against the model, can be blamed largely on problems of measurement. While the earnings data on individual employees was good, the Industry variables caused difficulties. Of the monopoly power measures, only concentration can be considered consistently accurate and meaningful. Economies of scale could not be measured, capital requirements were only roughly approximated and while advertising was easily measured, it suffered, as did the others, from aggregation. - 50 -- 51 -The capital-labour ratio had to be measured on an aggregate basis and in addition, it was highly collinear with capital requirements. Growth was available on a detailed industry basis, but change in employment was judged a weak measure of growth in demand. This being the case, little was learned from the behaviour of the growth variable. Despite the measurement problems, tentative con clusions can be drawn from the regression results. The evidence indicates that the amount of capital in an industry has no effect upon salary levels. The same can be said for the degree of advertising intensity. Any conclusions on the effect of monopoly power will have to be based upon the performance of the concentration variable, as is the case in most studies. Regression results indicate that Industry concentration does exert a, positive effect upon salary rates. The results were neither unanimous, nor strongly significant, but the proportion of positive coefficients was high enough to suggest this relationship. Perhaps the strongest conclusion that can be drawn from the study is that the relationship between salaries and industry characteristics is a complex one. The interindustry salary structure is undoubtedly determined by a large number of interrelated variables. Much effort would be required to successfully develop and test a complete model, but it would be an important contribution to the study of economics. BIBLIOGRAPHY A. BOOKS Bain, J.S. Barriers to New Competition. Cambridge, Harvard University Press, 1956• ' Baumol, William J. Business Behavior, "Value and Growth. New York, Macmillan, 1959» Bowen, W.G. Wage Behavior in the Postwar Period: An Empirical Analysis. Princeton, Princeton University • Press, i960. Galbraith, J.K. The New Industrial State. 'Boston, Houghton Mifflin, 1967^ Lewis, H. G. Unionism and Relative Wages in the United States. Chicago, University of Chicago Press, 1963. ~ Roberts, David R. Executive Compensation. Illinois, Free Press of Glencoe, 1959-Robinson, Joan The Economics of Imperfect Compet.ltion. London, Macmillan, 196I. Stigler, George J. Capital and Rates of Return in Manufacturing Industries. Princeton, Princeton University Press, 1963. Sturmthal, Adolf White-Coliar Trade Unions: Contemporary  Developments in Industrialized Societies. Urbana, University of Illinois Press, 1966"! - 52 -- 53 -Williamson, O.E. The Economics of Discretionary Behavior: Managerial Objectives in a Theory of the Firm. Englewood Cliffs, Prentice-Hall, I96IL B. ARTICLES Allen, B.T. "Market Concentration and Wage Increase: U.S. Manufacturing, 1947 - 1964." Industrial and Labor  Relations Review, Vol. 21, No. 1 (March 1966), PP. 353 - 365. Brown, D.G.. "Expected Ability to Pay and Interindustry Wage Structure in Manufacturing." Industrial and Labor Relations Review, Vol. 16, No.~l (October igh~2) , pp. 45 -~61T. •Comanor, S. and Wilson, T.A. "Advertising Market Structure and Performance." Review of Economics and Statistics, Vol. 49, No. 4 (November I90T) , pp.423 - ^W. Garbarino, J.W. "A Theory of Interindustry Wage Structure Variation." Quarterly Journal of Economics, Vol. 64, No. T~ (August l95oyr~pp- 2&r2' - 305. Goldner, W. and Ross, A.M. "Forces Affecting the Inter industry Wage Structure." Quarterly Journal of Economics, Vol. 64, No. 2 "pfey 1950)7 pp. 254 - 281. Harberger, A.C. "Monopoly and Resource Allocation." American Economic Review. Vol, 44, No. 2 (May 1954), pp. 77 ~^7, Herrnstadt, I.L. "Comment on 'Monopoly and Wages'." Canadian Journal of Economics and Political Science, Vol. 27, No. 3 (August 196T) , pp. 428 - 4 38^ - 54 -Kaysen, Carl "The Corporation: How Much Power? What Scope?" in The Corporation in Modern Society, Edward S. Mason, ed., Cambridge, Harvard University Press, 1959-Lamer, R.J. "The 200 Largest Nonfinaneial Corporations." American Economic Review, Vol. 56, No. 4 XSeptemher 196F)7~pp. 777 - 787. McGuire, J.W., Chiu, J.S.Y. and Elbing, A.O. "Executives' Incomes, Sales and Profits." American Economic Review, Vol. 52, No. 4 (September 1962") , pp. 753 - 761. Simon, H.A. "The Compensation of Executives." Sociometry, Vol. 20, No. 4 (December 1957), pp. 32 - 35-Shwartzman, D. "Monopoly and Wages." Canadian Journal of Economics and Political Science, Vol. 26, No] 3 "(August I960), pp. 428 - 438. Solomon, B. and Burns, R.K. "Unionization of White-Collar Employees: Extent, Potential and Implications." Journal of Business, Vol. 36, No. 2 "(April 1963), pp."141 - lb~5~. Weiss, L.W. "Concentration and Labor Earnings." American Economic Review, Vol. 56, No. '1 (March 1960") , pp. 96 - 117. Williamson, O.E. "Managerial Discretion and Business Behavior." American Economic Review, Vol. 53, No. 5 (December ±9b~5T, pp. .1033 - 1057". Yordon, W.J. "Another Look at Monopoly and Wages." Canadian Journal of Economics and Political Science, Vol. 27, No. 3 (August 1961), pp. 373 - 379-C. U.S. GOVERNMENT PUBLICATIONS Bureau of the Census, United States Census of Manufacturers: . ' 1947. Vol. I. Bureau of the Census, United States Census of Manufacturers: 1958. Vol. I. Bureau.of the Census, United States Census of Mineral  Industries: 195~87 Vol. I. Bureau of the Census, U.S. Censuses of Population and Housing  i960. 1/I00"0~and 1/10000, Two National Samples of~ the Population of the United States, Description and Documentation. Internal Revenue Service, Statistics of Income: July 1959 -June i960. APPENDIX I MATCHING OF INDUSTRY DATA As indicated in Chapter III, industry data for this study comes from three sources: 1. Statistics of Income, 1959 - I960, published by the Internal Revenue Service, U.S. Treasury Department. This source provides data on industry sales, assets and advertising. 2. U.S. Census of Manufacturers: '1958, and U.S. Census of Mineral Industries: 1958, both published by the Bureau of the Census, U.S. Department of Commerce. Figures for number of establishments and number of employees are taken from here. 3. "Concentration and Labor Earnings" by L.W. Weiss In the American Economic Review, March 1966, pp. 96 - 117. The appendix of this article provided figures for growth and concentration. The growth data originally came from the 1950 and i960 population censuses. The original sources for concentration data were Table IV of Concentration Ratios in  Manufacturing Industries, Part I (Senate Judiciary, 1962) 'and - 56 -- 57 -U.S. Census of Mineral Industries: 1958. The industry variables to be used in the study were developed as follows. A total of 6j5 manufacturing and mining industries were selected for the study. They were classified according to the three-digit industry code utilized by the population census. Growth and concentration figures were taken directly from Weiss' study since he reported them according to the three-digit code. The remainder of the industry data was available only on a major industry basis as the Statistics of Incorne does not report the required figures on a detailed industry level. Major industry figures from the Statistics of Income were matched by name to the major groups of the population census. For each group these aggregate figures were assigned to the detailed industry classifications. The Census of Manufacturers reports industry data on a major two-digit and minor four-digit basis. However, the establishment and employment figures from this source must be chosen on the same basis as the assets data from the Statistics of Inc ome. The major industry classifications for the population and manufacturing censuses correspond exactly, but there are two cases for which the Statistics of Income classification differ. These are in Transportation Equipment and Food and Kindred Products where Motor Vehicles and Equipment and Beverages are reported separately by the Statistics of' Inc orne. Since it is desired to have as detailed an industry breakdown •as is possible, this classification is used. Figures for Beverages are developed from the Census of Manufacturers by summing figures for the 208X four-digit industries. These are Malt Liquors (20-82), Malt (2083), Wine and Brandy (2084), Distilled Liquor except Brandy (2085), bottled and canned soft drinks (2086) and flavorings (2087). In order to match the Food and Kindred' Products figures to those from the Statistics of Income, these Beverages figures were subtracted from Food and Kindred' Products (20). Similarly, the figures for Motor Vehicles and Equipment were developed by summing those for Truck and Bus Bodies (37-13), Truck Trailers (3715), and Motor Vehicles and Parts (3717). These totals were subtracted from those for Transportation Equipment (37) to obtain the proper figures for matching. The results of this industry.matching are shown in Table V and the calculated variables in Table I. This .industry data was entered on punched cards so as to be in a form for computer input. It was then matched with the population census data which is stored on magnetic tape. Each individual employee selected from the population-census was assigned the industry variables corresponding to the three-digit industry in which he was employed. The resulting record, again on magnetic tape, constituted an observation for regression analysis. I NT) NAME 126 Metal Mining 136 Coal Mining 146. Crude Pet. & Nat. Gas 156 Non-Metalic Mining Lumber, and Wood Products 206 Logging 207 Sawmills, Planing Mills, Millwork 208 Misc. Wood Products 209 Furniture and Fixtures Stone, Clay and Glass Products 216 Glass and Glass Products 217 Cement, Concrete, Gypsum and Plaster 218 Structural Clay Products 219 Pottery and Related Products 23o Misc. Non-Metalic Minerals TABLE V BASIC INDUSTRY DATA UNION- Acc^rpo qflTpq ADVER- NO. NO. IZATION OD ° an-jaa TISING EST. EMF. $' 000 • $'000 ' $! 000 94 99 13 30 4211305 2324810 7146456 2356819 1791980 1884238 4450769 1769976 518 1716 9086 6769 2233 8178 18501 7306 92501 210519 312800 116812 50 5333283 7193236 28558 37789 518302 43 47 50 5333283 5333283 2449833 7193236 7193236 4945371 28558 28558 60459 37789 37789 10160 518302 518302 347599 95 8713029 10581391 74279 15022 554042 69 65 60 85 • 8713029 8713029 8713029 8713029 10581391 10581391 10581391 10581391 74279 74279 74279 74279 15022 15022 15022 15022 554042 554042 554042 554042 IND NAME Primary Metal Industries 237 Blast Furnaces, Steel Works, etc. 238 Other Primary Iron & Steel 239 Primary, Non-Ferrous Metals Fabricated Metal Products 246 Cutlery, Hand Tools, Hardware 247 Fabricated Structural Metal Products 248 • Misc. Fab. Metal Products Machinery Except Electrical 256 Farm Machinery 257 Office, Computing & Acctg Mach. 258 Misc. Machinery 259 Electrical Machinery 267 Motor Vehicles and Equipment TABLE V (Continued) UNION IZATION ASSET'S $'000 SALES $' 000 ADVER TISING $'000 NO. EST. NO. EMP. 99 75 83 26282013 26282013 26282013 27194223 27194223 27194223 118575 118575 118575 6446 6446 6446 1096359 1096359 1096359 60 11772597 18712082 167205 24782 1057986 73 11772597 18712082 167205 24782 1057986 73 11772597 18712082 167205 24782 1057986 90 20137931 24938517 262160 29839 1348245 42 20137931 24938517 262160 29839 1348245 69 20137931 24938517 262160 29839 1348245 73 '14300144 22500260 389261 8091 1122284 98 16561217 24273354 196745 2283 210519 TABLE V (Continued) I NT) NAME i^I0ILT ASSETS IZATION Transportation Eqpt Except Auto $'000 268 Aircraft and Parts ' 82 9694529 269 Ship & Boat Building & Repair 70 9694529 276 Railroad & Misc. Transp. Eqpt 50 9694529 Instruments and Related Products 286 Professional Eqpt and Supplies 53 4545251 287 Photographic Eqpt and Supplies 43 4545251 289 Watches, Clocks, Etc. " 90 4545251 296 Miscellaneous Mfgrs 55 4518551 Food and Kindred Products 306 " Meat Products 83 18010406 307 Dairy Products 6l 18010406 308 Canning and Preserving 69 l8010406 309 Grain Mill Products ' 74 l8oio4o6 SALES $'000 ADVER TISING $'000 NO. EST. NO. EMP. 16465594 16465594 16465594 46287 46287 46287 4324 4324 4324 100571 100571 100571 6200826 6200826 6200826 139904 139994-139994 3526 3526 3526 296558 296558 296558 7249222 133726 14273 571432 50112468 50112468 50112468 50112468 963744 963744 963744 963744 36061 36061 36061 36061 1492617 1492617 1492617 1492617 INT) NAME 316 Bakery Product's 317 Confectionery & Related Prodticts 319 Misc. Food Preparations 318 Beverages 329 Tobacco Products Textile Mill Products 346 Knitting Mills 3^7 Dying and Finishing 348 Floor Coverings Except Hard Surfaces 349 Yarn, Thread and Fabric 356 Misc. Textiles 359 Apparel and Accessories 367 Misc. Fan. Textiles Paper and Allied Products 386 Pulp, Paper & Paperboard 387 Paperboard Containers 389 Misc. Paper & Pulp Products ABLE V (Continued) UNION IZATION ASSETS $' 000 SALES $'000 ADVER TISING $'000 NO. EST. NO. EMP. 50 51 72 18010406 i8oio4o6 i8oio4o6 50112468 • 50112468 50112468 963744 963744 963744 36061 36061 36061 1492617 1492617 1492617 70 5248563 8215474 401319 5558 206197 60 3306120 4817872 263388 504 84467 31 50 8929065 8929065 14196673 14196673 88344 88344 7675 7675 901677 901677 62 . 24 39 8929065 8929065 8929065 14196673 14196673 14196673 88344 '88344 88344 7675 7675 7675 901677 901677 901677 64 32 • 4507669 4507669 11843834-11843834 115827 115827 29297 29297 1180517 1180517 90 62 71 9977017 9977017 9977017 11579374 11579374 11579374 97458 97458 97453 5271 5271 5271 555398 555398 . 555398 IND NAME Printing and Publishing 396 Newspapers 39° Printing & Publishing Except Newspapers Chemicals and Allied Products 406 Synthetic Fibres 407 Drugs and Medicines 408 Paints, Varnish and Related Products 409 Misc. Chemicals and Related Products Petroleum and Coal Products 416 • Petroleum Refining 419 Misc. Pet. & Coal Products Rubber and Plastics Products 426 42°/ Rubber Products Misc. Plastic Products TABLE V (Continued) UNION IZATION ASSETS £'000 SALES $1 000 ADVER TISING $'000 NO. EST NO. EMP. 90 56 8146482 12238036 92601 35368 864101 8146482 12238036 92601 35368 864101 75 33 67 70 23202242 23202242 23202242 23202242 26065022' 26065022 26065022 26065022 986856 986856 986856 986856 11309 11309 11309 11309 699166 699166 699166 699166 OA 90 39748483 39748483 36004854 36004854 164859 164859 1608 1608 179166 179166 81 50 4906503 4906503 7737607 7737607 127152 127152 4462 4462 349050 349050 TABLE V (Continued) IND NAME IZATION ASSETS Leather and Leather Products $'000 436 Leather Tanning, etc. 67 1843319 437 Footwear Except Rubber 48 1843319 438 Leather Products Exc. Footwear 48 1843319 $'000 $'000 3907359 43585 4534 349050 3907359 43585 4534 349050 3907359 43585 4534 349050 APPENDIX II DETAILED REGRESSION RESULTS Table VI shows in detail the results of the multiple •regressions. For each occupation the number of observations, the mean salary, the constant term and the coefficients of the four industry variables and four personal variables are shown. In brackets immediately below each coefficient is shown the standard error of the coefficient. Immediately below this is the F~ probability, where F is a statistic calculated by dividing the coefficient by its standard error and squaring the resulting term. This is the probability of obtaining a value of F greater than or equal to the one calculated, given that the coefficient is actually zero. At the far right hand side of the table, the value 2 of R , the coefficient of multiple determination, is given and underneath that is the standard error of the estimate. As explained in Chapter IV, the high'collinearity between the capital requirements and capital labour variables - 65- -66 necessitated the dropping of one of them from the regression runs. Thus, the capital requirements variable does not appear in Table VI. Regressions run with this variable in place of capital-labour produced almost identical results. Table VII gives the regression results for the runs made with the unionization variable. The coefficients are shown for unionization when it was the only industry variable run. Then coefficients for unionization and concentration are shown when both were run in the same regression. The presence of one or two asterisks Indicates significance at the 5^ or l°/o level on a, 2-tailed T-test. TABLE VI MULTIPLE REGRESSION RESULTS SALARY RELATED TO 4 INDUSTRY AND 4 PERSONAL VARIABLES OCCUPATION N,M CONSTANT CONCEN TRATION ADVER TISING CAP/LAB GROWTH SOUTH AGE EDUCA TION RURAL RSQ Accountants (Std. Err.) (F Prob) 100, + 977.4 7208 (2126.1) - 1.80 (15.33) (.8733) - 10470. (19610.) (.6013) 4.20 (3.65) (.1513) - 6.49 (5.04) (.1980) + 317.4 (469.5) (.5079) ^ 93.0 (17.58) (.0000) + 362.0 (154.7) (.0204) - 61.1 (459.7) ( .8640) .2854 (1846.) Chemists 1,3, - 2097. 7503 (3081.) + 7.91 (32.27) (.7945) + 978.2 (21630.) (.9165) 18.56 (7.77) (.0203) - 3.36 (11.29) (.7602) +1060. (648.2) (.1023) ^ 57.33 (29.29) (.0538) + 630.4 (168.7) (.0006) - 698.8 (722.9) (.3413) .4360 (1991.) Draftsmen 114, + 576.8 6389 (1807.) + 24.69 (12.33) (.0452) + 18330. (20510.) (.3773) .4221 (5.29) (.8957) - 3.93 (4.30) (.3664) - 145.3 (410.2) (.7222) i- 90.53 (15.75) (.0000) 1- 203.1 (136.9) (.1376) - 92.34 (354.4) (.7840) .3244 (1476.) Industrial Engineers 58, + 50.29 8145 (3756.) + 15.74 (41.43) (.7039) - 99260. (43040.) (.0241) <- 18.76 (30.56) (.5494) - 4.84 (10.15) (.6402) - 984.8 (1397.) ( .4910) +103.1 (33.52) (.0035) + 479.1 (309.8) (.1245) - 973.4 (990.5) (.3325) .2969 (2781.) Mechanical Engineers 110, - 1063. 9494 (3988.) + 19.24 (31.83) (.5543) + 26330. (54360.) (.6345) + 2.62 (11.75) (.8083) • .491 (8.53) ( .9089) + 354.7 (1078.8) (.7392) +102.57 (42.60) (.0171) ^ 523.0 (280.2) (.0616) -2180. (1272.) (.0858) .1107 (3972.) Sales Engineers 25, 11950 -15750. (26100.) + 87.12 (292.5) (.7619) -129500. (164000.) (.4466) -120.8 (176.5) (.5100) -29.03 (54.31) (.5944) -2561. (3546.) (.4867) +491.4 (242.7) (.0574) +1232. (1746.) (.4967) +1600. (6398.) (.7927) .3877 (6949.) Engineers (NEC) 33, 9520 -15540. (9715.8) +117.7 (87.4) (.1879) -130100. (104300.) (.2223) - 95.51 (95.06) (.3267) -21.19 (21.90) (.3451) -2589. (3442.) (.4652) +381.7 (112.1) (.0024) +1216. (625.2) (.0609) -5104. (3212.) (.1215) .4577 (5310.) TABLE VI MULTIPLE REGRESSION RESULTS SALARY RELATED TO 4 INDUSTRY AND 4 PERSONAL VARIABLES OCCUPATION N,M CONSTANT CONCEN TRATION ADVER TISING CAP/LAB GROWTH SOUTH AGE EDUCA TION RURAL RSQ Accountants (Std. Err.) (F Prob) 100, + 977.4 7208 (2126.1) - 1.80 (15.33) (.8733) - 10470. (19610.) (.6013) 4.20 (3.65) (.1513) - 6.49 (5.04) (.1980) + 317.4 (469.5) (.5079) + 93.0 (17.58) (.0000) + 362.0 (154.7) (.0204) • 61.1 (459.7) (.8G40) .2854 (1846.) Chemists 53, - 2097. 7503 (3081.) + 7.91 (32.27) (.7945) + 978.2 (21630.) (.9165) 18.56 (7.77) (.0203) - 3.36 (11.29) (.7602) +1060. (648.2) (.1023) + 57.33 (29.29) (.0538) + 630.4 (168.7) (.0006) - 698.8 (722.9) (.3413) .4360 (1991.) Draftsmen 114, + 576.8 6389 (1807.) + 24.69 (12.33) (.0452) + 18330. (20510.) (.3773) .4221 (5.29) (.8957) - 3.93 (4.30) (.3664) - 145.3 (410.2) (.7222) + 90.53 (15.75) (.0000) + 203.1 (136.9) (.1376) - 92.34 (354.4) (.7840) .3244 (1476.) Industrial Engineers 58, + 50.29 8145 (3756.) • 15.74 (41.43) (.7039) - 99260. (43040.) (.0241) + 18.76 (30.56) (.5494) - 4.84 (10.15) (.6402) - 984.8 (1397.) (.4910) +103.1 (33.52) (.0035) + 479.1 (309.8) (.1245) - 973.4 (990.5) ( .3325) .2969 (2781.) Mechanical Engineers 110, - 1063. 9494 (3988.) + 19.24 (31.83) (.5543) + 26330. (54360.) ( .6345) + 2.62 (11.75) (.8083) - .491 (8.53) ( .9089) + 354.7 (1078.8) (.7392) +102.57 (42.00) (.0171) + 523.0 (280.2) (.0616) -2180. (1272.) (.0858) .1107 (3972.) Sales Engineers 25, -15750. 11950 (26100.) + 87.12 (292.5) (.7619) -129500. (164000.) ( .4466) -120.8 (176.5) (.5100) -29.03 (54.31) (.5944) -2561. (3546.) ( .4867) +491.4 (242.7) (.0574) +1232. (1746.) (.4967) +1600. (6398.) (.7927) .3877 (6949.) Engineers (NEC) 33, -15540. 9520 (9715.8) +117.7 (87.4) (.1879) -130100. (104300.) (.2223) - 95.51 (95.06) ( .3267) -21.19 (21.90) (.3451) -2569. (3442.) (.4652) +381.7 (112.1) (.0024) +1216. (625.2) (.0609) -5104. (3212.) (.1215) .4577 (5310.) - 69 -TABLE VII REGRESSION RESULTS FOR UNIONIZATION OCCUPATION Accountants Chemists Draftsman Industrial Engineers Mechanical Engineers Sales Engineers Engine e rs (NEC) Personnel and Labour Relations Workers Prof., Tech. and Kindred Workers (NEC) •Managers, Officials and Proprietors Bookkeepers Secretaries Stenographers Typists Salesmen and Salesclerks UNIONIZATION ALONE -10.97 +13.82" +24.03* +30.14 +17.10 -61.65 +66.23 -29.48 - 3.98 + 9-99 + 2.50 +12.35** + 1.92 +30.39** - 4.08 UNIONIZATION CONCENTRATION - 9.09 • 31 + 20.02 + 13.45 + 20.08 - 99.31 - 25.06 -105.9 -. 5.36 + 16.20 + .91 + .04 - 4.39 + 29.28 + 14.17 + . 23 - 1.16 + 15.17 - 15.37 + 2.28 +243.3 + 80.18 +180.22 + 14.74 - 19.1c + 6.68 + 23.04** + 21.83 + 11.75 - 71.25** 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items