INDUSTRIAL CONCENTRATION AND FRINGE BENEFITS by John S. Heywood*
ABSTRACT This paper uses a major micro data set to examine the association between market structure and the provision of fringe benefits. By focusing on the probability of fringe benefit provision and by using individual data, this study departs from those which precede it. The analysis reveals that industrial concentration is an independent correlate with a large variety of fringe benefits, a finding which contrasts with earlier conclusions. 1. Introduction
Few issues have generated as much empirical research as the question whether and how wages and industrial concentration are associated. Dickens and Katz (1986) identify twenty-two separate studies designed to examine this association, and even this underestimates the total, as new studies continue to appear [Belman (1986), Kawashima and Tachibanaki (1986) and Blanchflower (1986)]. While this large literature lacks unanimity of results, a reasonably clear pattern emerges from the studies using individual (micro) data. The majority of these studies estimate industrial concentration to be a significant positive regressor in wage equations. 1 This contrasts with the general failure of industry level studies to confirm a role for market structure. The majority of those studies estimate no significant partial correlation between wages and industrial concentration. The stark contrast between the individual and industry results suggests a deficiency in the still fledgling literature examining the association of fringe
Assistant Professor of Economics at the University of Wisconsin-Milwaukee.
120
-
Industrial
Concentration
benefits with industrial concentration. Investigation of this second association has been conducted exclusively at the aggregate level, with results which might have been predicted from the industry wage equations: industrial concentration plays no direct role in determining the level of fringe benefits. 2 This exclusive reliance on aggregate studies should end with micro data, and its superior controls, brought to the examination. This paper uses micro data to present new results on the association between industrial concentration and fringe benefits. Specifically, it demonstrates that even after controlling for personal and industry characteristics, the probability of receiving any of eight fringe benefits is positively influenced by concentration. Indeed, even after controlling for current earnings, concentration continues to influence fringe benefit provision. Further, a union-concentration interaction emerges with a negative coefficient, suggesting a smaller influence for concentration among union members. These results closely mirror the behavior of concentration in micro data wage equations but directly oppose the findings of current aggregate studies of fringe benefits. In sum, they indicate that typical price-to-cost ratios may underestimate the inefficiency associated with concentrated industrial strucure.
2. Background and Previous Study Although the association of market structure and benefits has received nowhere near the attention given to that of market structure and wages, it has recently received significant examination. The rationales presented for a benefit-structure relationship nearly duplicate those long given for the wage-structure association. Early rationales include the purchase of political favor by monopolistic firms, the special power of unions to extract favorable terms in industries with market power, and the willingness of managers without market discipline to over-.pay workers. Of these early theories, the most common and perhaps persuasive is that unions successfully demand a share of the rents which flow from market structure. Hence, when these rents are high because of noncompetitive structure, union
Heywood
- 121
compensation will also be high. Such a connection might be considered a theory of an indirect effect, as concentration influences wages through unionization. Yet, none of the micro wage studies confirm a stronger concentration effect for union members and five of the studies find that nonunion members actually benefit from concentration to a greater extent than do union members [Weiss (1966), Jenny (1978), Mellow (1981), Heywood (1986a) and Belman (1986)]. The aggregate benefit equations present a completely different pattern, with four of five studies yielding a positive unionconcentration interaction [Alpert (1982), Long and Link (1983), Cymrot (1985) and Brush and Crane (1986)]. Thus, aggregate benefit studies suggest no direct relationship between benefit levels and concentration, and claim the importance of the interaction demonstrates that unions capture market structure rents. On the other hand, the micro wage equations typically do not support such a claim for unions, yet do demonstrate a direct influence of concentration. The pattern presented by the aggregate benefit equations need not be in conflict with the micro wage equations. Unions may be successful in capturing monopoly rents from firms in the form of fringe benefits but not in the form of additional wages. This may occur either because union members have a stronger desire for benefits than for wages [Freeman (1981)] or because firms resist wage increases and tolerate benefit increases, the costs of which are largely deferred. On the other hand, the conflict may be real. Just as the aggregate wage equations differed from the micro wage equations, so aggregate benefit equations may differ from their micro equivalents. In such a case, the micro benefit equations would also fail to confirm the rent capturing hypothesis. Other, more recent, theories continue to argue that compensation and industrial concentration are directly related. These include efficiency wage models in which workers' notions of fairness are directly related to the firm's ability to pay [see Akerlof (1984) and the Dickens and Katz (1986) discussion]. Alternatively, the capital associated with concentrated markets and the unobserved skills which command higher wages and benefits may be
122
-
Industrial
Concentration
complementary. Finally, expense-preference behavior by managers in noncompetitive product markets may result in higher wages designed to reduce managerial effort associated with monitoring [Heywood (1986b)]. While none of these more recent theories have received direct empirical scrutiny, they seem indirectly rebuffed by the aggregate benefit equations which show no association between fringe benefits and concentration. Again, if the benefit equations truly mirror the wage equations, micro data may provide the association with concentration which aggregate equations have not. Previous works using aggregate benefit equations suffer from a number of specific difficulties. Frequently, the high degree of aggregation results in small numbers of observations and the possibility of altered relative variances. Such a problem could bias concentration coefficients toward zero. For example, Long and Link (1983) use Chamber of Commerce data on 28 industries and Cymrot (1985) uses averages derived from 29 CPS industries of different degrees of aggregation. Alternatively, Brush and Crane (1986) use several hundred four digit SIC industries from the Census of Manufactures but, as a consequence, have access to poor controls. Alpert (1982, 1983) uses establishment data with numerous observations but is frequently forced to use ad hoc measures of crucial variables. For example, unionization must be proxied by whether or not the plant is 50 percent unionized. Stafford (1968) criticizes this proxy both for its imprecision and for the values of coefficients which emerge from its use. In general, previous aggregate studies have access to a small group of controls. The studies are unable to introduce measures of tenure or actual work experience, either of which may be correlated with benefits, and perhaps also with concentration. When available, age is used to proxy experience [Alpert (1982, 1983)]. Yet, industries with equally aged employees may have employees with radically different experience patterns. Employees in durable manufacturing industries characterized by high concentration suffer greater work experience disruption because of their industry's sensitivity to the business cycle. Consequently, true work experience, an omitted variable, could be positively
Heywood
- 123
associated with benefits and negatively associated with industrial concentration once one accounts for age. This biases the direct effect of concentration downward. In contrast, the Quality of Employment Survey used here allows for the introduction of both actual experience and tenure as independent variables. It is crucial to note that previous studies look only at the effect of concentration on the level of benefits being provided or on the share of benefits in total compensation. Both are discontinuous from zero. There exist a large number of observations with no benefits recorded. Studies which include plants with and without benefits run the risk of a serious selection problem. This risk has been well recognized in the case of the union e f f e c t on benefits. In fact, the original finding of Freeman (1981) that unions increase the level of pension benefits has been seriously questioned on just this ground. More recent studies show that nearly all of the union effect on pensions is felt in the probability of their provision rather than in the level of pension benefits. Hence, these studies reveal no significant e f f e c t of unionization on pension generosity. Other benefits have shown a more mixed pattern, with some of the union effect felt on levels and some on the probability of provision. 3 The same distinction between e f f e c t on levels and effect on provision pertains to examining the role of industrial concentration. Previous studies do not separate these two issues. Recent attempts to reconcile the individual and industry wage studies suggest a final rationale for an individual level study of benefits. Typical industry wage studies use variables collected from different sources measured over different populations. For example, the most common wage measure is the average wage of production workers in the industry, but the demographic variables included are usually averages for all workers in the industry. Belman and H e y w o o d (1988) demonstrate that correcting for this and other related incongruencies yields industry equations that. more closely resemble the individual ones. Industry benefit studies use a similar variety of sources. The measure of benefits used by Brush and Crane is only for production workers while other
124
-
Industrial
Concentrntion
variables are for all workers. Both Alpert and C y m r o t use a variety of sources from different years to construct their data sets. The advantage o f the individual approach is that, with the exception o f the concentration variable itself, all data are specific to the individual, eliminating problems of incongruence.
3. Empirical Approach and Results This study focuses exclusively on the probability of providing fringe benefits and uses micro data to do so. This choice results, in part, from the limitation that no micro data set includes detailed information on the level of fringe benefits. As a consequence, we can examine only one of the two effects outlined earlier. We ignore varying degrees of generosity in order to focus on whether or not benefits are provided. Yet, benefit provision may be related to benefit levels. Freeman (1981) and Fosu (1983, 1984) argue that it may not pay to administer low levels of fringe benefits because the fixed costs of administration will comprise too large a share of total benefit expenditures. Benefits will be provided only when levels are high enough to j u s t i f y these fixed costs. Thus, if unions capture rents b y increasing benefits levels, union members would be more likely to receive fringe benefits, because higher b e n e f i t levels would be more likely to justify the fixed costs. In this way we might expect that the factors which increase benefit levels also make benefit provision more likely. The 1977 Quality of Employment Survey contains information on the provision of a number of important fringe benefits. In each case workers are asked w h e t h e r their current j o b provides the particular benefit. These responses become the dependent variable in probit specifications. The use of individual data allows superior explanatory variables to be introduced. Each equation includes the individual's years of education, actual years of experience, experience squared, two tenure dummies, whether the individual attended trade school, w h e t h e r the individual has a serious health disability, the individual's race and sex, whether residence is in the South, the size of the plant in which the individual works, and the worker's union status. 4 These are all important
Heywood
- 125
determinants of compensation, many of which were not included in previous examinations because of the limitations of aggregate data. We proxy industrial structure with the four-firm concentration ratio for the worker's industry. This standard measure of monopolization uniformly emerges as the variable used in previous studies. 5 The variables also include an interaction of union status and the concentration ratio. The survey provides data on provision of eight important fringe benefits: pensions, life insurance, health insurance, the availability of alternative job training programs, thrift plans, eye care programs, paid vacation, and paid sick days. Unfortunately, the survey makes no attempt to distinguish the characteristics of each class of benefits, thus lumping all pensions together or all insurance plans together. The sample is restricted to manufacturing because of the difficulties in creating and interpreting concentration ratios in other sectors of the economy. Table 1 demonstrates the association between the provision of benefits and the standard wage equation variables. The association highlights the role of these variables as determinants of benefit provision and suggests a close connection between the different types of compensation. Generally, the human capital and demographic variables have the expected sign, and many, but not all, take on statistical significance. Those which fail to take on significance include race, health disability, and often, experience. These variables routinely emerge as significant from wage equations, and their failure to do so here may indicate a slightly less close relationship between individual characteristics and benefit provision than that demonstrated between characteristics and wages. Union status and plant size play a fairly consistent role, increasing the probability of benefit provision. Given the known associations between human capital, size, unionization and concentration, the inability of some earlier studies to include, all of these variables is a potential source of serious misspecification. The concentration and interaction variables demonstrate surprising strength. The inclusion of plant size and
126 - Industrial Concentration Table 1 Effect of Market Structure on Benefit Provision
Dependent Variable
Eye Pension
Life
Health
Constant
2.045" (3.126)
0.7633 (1.316)
1.652" (2.196)
2.202" (3.495)
Exp
0.0247 (0.8103)
0.0167 (0.6188)
0.0489 (1.384)
0.0373 (1.317)
Exp squared
-0.0003 (0.4433)
-0.0005 (0.8689)
-0.0015" (1.993)
-0.0006 (1.077)
Tenure1
0.5431" (2.188)
0.3973+ (1.737)
0.6228* (2.077)
-0.2956 (0.1211)
Tenure2
1.093" (3.180)
0,9917" (3.190)
1,580" (3.102)
0.0289 (0.0970)
Zd
0.0676 (1.568)
0.0308 (0.8232)
0.1237" (2.453)
0.0306 (0.8127)
Female
-0.2856 (1.238)
-0.4265* (2.035)
-0.5185+ (1.821)
-0.7262* (3.394)
Race
-0.0106 (0.3151)
-0.9216 (0.3247)
-0.0142 (0.0365)
-0.3024 (1.087)
0.0256
-O.OLO5
0.3250
-o.o957
Trade
(0.7781)
(0.0371)
(0.6150)
(0.4026)
Health
-0.3794 (1.181)
-0.1129 (0.4108)
0.3773 (0.8878)
-0.5095+ (1.928)
Plant
0.0008 (3.990)
0.0002 (1.367)
0.0002 (0.8476)
0.0004* (3.736)
C are
Heywood - 127 Table 1 (cont.) Effect of Market Structure on Benefit Provision Dependent Variable
Pension 1.881"
Union
(3.154) 0.0156+
Life 1.289"
(2.746) 0.0173"
Health 2.560*
(3.114) 0.0204+
Eye Care 1.192'
(2.676) 0.0195"
Cone
(1.677)
(1.978)
(1.647)
(2.340)
Cone x Union
-0.0230
-0.0242*
-0.390*
-0.0105
(1.661)
(2.271)
(2.188)
(1.020)
N
291
288
296
291
Chi-Squared
112.6
46.3
64.0
129.5
Portion getting benefit
0.781
0.804
0.909
0.421
6Prob/6Conc
0.00104
0.00176
0.00026
0.00518
*Significant at the 5 ~ level. +Significant at the 10~ level.
128 -
IndustrialConcentration Table 1 (cont.) Effect of Market Structure on Benefit Provision
Dependent Variable
Thrift
Train
Vacation
Sick
Constant
2.095 (3.651)
2.798* (4.428)
1.182 (1.343)
1.393' (2.462)
Exp
-0.0017 (0.0684)
0.0902* (3.278)
0.1076' (2.790)
(1.262)
-0.0003
-0.0019'
-0.0024*
-0.0008
(0.5709)
(3.340)
(2.891)
(1.521)
-0.3585+ (1.532)
0.5919+ (1.916)
-0.0753 (0.3305)
Exp squared
Tenure1
0.3547*
(1.602) Tenure2
0.5976*
(2.167) Ed
0.0854*
(2.448)
Female
0.0675
0.0325
-0.2826"
1.642"
0.2966
(0.9652)
(2.473)
(1.047)
0.1381"
0.0464
0.0885*
(3.468)
(0.8979)
(2.465)
-0.4894*
-0.1074
-0.7181"
(0.3526)
(2.460)
(0.3455)
(3.739)
0.2994
-0.0166
-0.4520
(1.194)
(0.0632)
(1.253)
-0.1386 (0.5421)
Trade
0,1485 (0.6608)
0.5025* (2.015)
0.8092 (1.205)
0,1039 (0.4424)
Health
-0.0024
0.0158
Race
Plant
South
-0.3958
-0.6318"
(0.0101)
(0.0609)
(0.7334)
(2.497)
0.0002* (2.118)
0.0006* (4.654)
0.0002 (0.6860)
0.0001 (0.7246)
-0.2300
-0.3786
(0,9128)
(1.017)
0.3357
(1.403)
0.5874*
(2.308)
Heywood - 129
Table 1 (cont.) Effect of Market Structure on Benefit Provision
Dependent Variable
Thrift
Train
Vacation
Sick
-0.2268 (0.5255)
1.384+ (1.854)
0.0618 (0.1527)
Union
0.6464 (1.596)
Cone
(2.089)
(1.653)
(1.891)
(2.005)
Conc x
-0.0129 (1.355)
-0.0043
-0.348+
(0.4246)
(1.771)
-0.0149 (1.562)
N
291
292
295
291
Chi-Squared
45.4
106.6
41.5
69.6
Portion getting benefit
0.498
0.487
0.933
0.543
~Prob/~Conc
0.00427
0.00450
0.00151
0.00292
0.0162
Union
0.0131+
0.0306+
0.0163"
*Significant at the 5% level. +Significant at the 10~ level.
human capital measures does not cause the concentration effect on wages to vanish. In all eight specifications the concentration variable is positive and statistically significant. In all eight specifications the interaction is negative, and in four it is significant. These results appear particularly strong when one realizes that union status, well recognized as a determinant of benefit provision, is significant in only five cases and plant size, considered crucial because o f - t h e supposed fixed costs of benefit administration, is significant in only four cases. 6 The market structure results contrast with previous examinations of fringe benefits. T h e y support the direct role of concentration and fail to confirm its indirect
130
-
Industrial
Concentration
effect (i.e., the ability of unions to capture product market rents). In none of the cases is the interaction even positive. Indeed, the probability of receiving benefits due to concentration increases less for union members than for nonunion members. This suggests that nonunion members actually capture more of the additional monopoly profits associated with concentration, the exact opposite of the evidence from aggregate studies of fringe benefits. 7 In fact, in four of the eight estimations, the interaction coefficient dominates the concentration c o e f f icient. This yields a negative net concentration c o e f f icient for union members in these cases. 8 While these results suggest that higher concentration does little to increase the chance of benefit provision for union members, such a suggestion must be tempered by the realization that high concentration may give rise to unions in the first place. To the extent this is true, the increased probability of benefit provision associated with union status cannot be convincingly separated from the influence of concentration. A multiple equation system would be required to detect the full effects of concentration. On balance, the current results and those o f the micro wage studies are strikingly similar. Like the current results, earlier studies found evidence of both a direct effect and a negative interaction. By evaluating the coefficients at the mean values for the explanatory variables, one can appreciate the magnitudes involved. The final row in Table 1 presents the change in the probability of receiving the particular benefit caused by a one point increase in industrial concentration. This measure includes both the direct effect and the indirect e f f e c t through the union interaction. 9 For example, a 20 point increase in industrial concentration, which is approximately one standard deviation (18.6), increases the probability of receiving eye care by approximately 0.104. Given that the mean value of the eye care variable is 0.421, this represents nearly a 25 percent increase. As suggested, holding u n i o n , status constant is potentially misleading, as it may be causally related to concentration. Despite this possible confounding effect, the size of the increases in probability are illustrative of a substantial influence for concentration.
Heywood - 131
Most past studies of both union and market structure effects on benefits use the proportion of total compensation that benefits represent as the dependent variable. Alternatively, they include earnings as an additional independent variable. While the former specification cannot be duplicated using the probability of provision, an effort can be made to hold the level of other compensation constant when focusing on the role of market structure. For most respondents in the Survey, a variable exists for total annual wages and salary earned from their current job. 10 Table 2 presents the evidence which results from including this measure of income as an additional explanatory variable in the eight probit estimations.
Table 2 The Effect of Market Structure with an Income Measure Dependent Variable
Union
Cone
Eye Pension
Life
Health
Care
1.945"
1.341'
2.851'
1.163"
(3.161)
(2.797)
(3.163)
(2.557)
0.0147
0.0161+
0.0203
0.0172"
(1.549)
(1.831)
(1.557)
(2.018)
Cone x
-0.0247+
-0.0260"
-0.0460"
-0.0099
Union
(1.726)
(2.312)
(2.358)
(0.9469)
Income
(1.990)
(2.269)
(2.305)
(2.219)
N
282
280
287
282
Chi-Squared
111.1
49.6
70.5
102.0
0.00004*
0.00005*
0.00009*
Note all specifications included all the variables from Table I.
0.0004*
132 - Industrial Concentration Table 2
(cont.)
The Effect o£ Market Structure with an Income Measure
Dependent Variable Union
Thrift 0.6110"
Train
Vacation
Sick
-0.2116
1.270
0.2715
(1.499)
(0.4856
(1.619)
(0.6460)
Conc
0.0158" (2.034)
0.0132+ (1.653)
0.0277+ (1.647)
0.0168" (2.082)
Concx
Union
-0.0126 (1.323)
-0.0054' (0.5267)
-0.0358+ (1.696)
-0.0212" (2.145)
Income
(0.2387
(0.4403)
(2.338)
N
283
284
286
282
Chi-Squared
41.5
103.1
46.0
89.6
0.00001
0.00001
0.00009*
0.0008* (4.000)
Note all specifications included all the variables from Table i.
The inclusion of the income variable does not alter the pattern established in the earlier estimates. The concentration variable remains a positive sign in all eight cases and remains significant in six. The interaction remains negative in all eight cases and retains significance in four. Again, these results are convincing because the union status variable retains significance in only five cases despite its known role in benefit provision. In general, these results indicate an association between concentration and benefit prov~_sion even after accounting for earnings. Such evidence takes on particular importance given previous studies confirming market structure as a determinant of earnings. Hence, not only do otherwise equal workers in concentrated industries receive higher wages, they are also more likely to receive a large variety of other benefits. They
Heywood
- 133
remain more likely to receive these benefits not only because of the direct effect of concentration, but also because of the less obvious effect through higher wages which reflect, in part, higher concentration. 3. Conclusion This paper began by establishing a fundamental tension created by earlier research. Current aggregate evidence on the relationship between concentration, unionization and fringe benefits contradicts micro evidence on the relationship between concentration, unionization and wages. While the former denies any direct influence of concentration and supports the notion that unions capture product market rents, the latter supports the direct influence of concentration while denying that unions disproportionately capture rents. This tension may reflect fundamental differences in the behavior of firms and unions toward wages and fringe benefits, or it may reflect profound measurement differences between aggregate and micro data. While no full micro data set includes accurate measurements of fringe benefit levels, this paper examines the probability of providing fringe benefits at all. The results strongly suggest that the tension exists between the use of aggregate and micro data, and not between the patterns of wages and fringe benefits. Concentration correlates directly with benefit provision, while the interaction emerges as negative and often significant. The latter indicates that unions capture a smaller portion of additional profits, and contradicts the claim that unions are particularly effective in obtaining benefits in the presence of monopolistic industries. Each of these results mirrors the micro wage equations, but contradicts the results of aggregate benefit equations. The concentration variable performs at least as well as union and plant size variables across the set of estimations. Even when worker incomes are held constant, those workers in more concentrated industries remain more likely to receive fringe benefits. These results would also not be expected on the basis of previous benefits research based on plant and industry level data.
134
-
Industrial C o n c e n t r a t i o n
Endnotes
The author thanks William Shepherd, Gary Solon, Frank and Robert Drago for fine related work. The paper from the suggestions of the seminar in labor economics.
J. Adams, William G. Stafford, Dale Belman comments on this and also benefited greatly University of Michigan
.
Among those who confirm this role for concentration, Dalton and Ford (1977) use the Census, Mellow (1981) and Belman (1986) use the Current Population Survey, Long and Link (1983) use the National Longitudinal Survey, Kwoka (1983) uses the Quality of Employment Survey, Heywood (1986) uses the Panel Study of Income Dynamics, Kawashima and Tachibanaki (1986) use micro data from Japan, and Jenny (1978) uses micro data from the French Census.
.
See Alpert (1982 and 1983), Long and Link (1983), Brush and Crane (1986) and Cymrot (1985). Some of these studies use various fringes and population subsamples. The comments in the text are based on the majority of those estimations.
.
See Fosu (1983, 1984) on the selection problem in estimating the union effect on fringe benefits.
.
Freeman and Medoff (1981) claim that union status inadequately captures union power. They suggest including a second variable measuring the percent of the industry unionized. This measure has generally not been included in studies examining the e f f e c t of concentration and is excluded here. To include this variable properly requires separately estimating the influence of concentration on the degree of unionization in a sample selection framework. While such a technique can be applied easily to wage equations [see Belman (1986)], it is far less clear how such a technique could be applied to the dichotomous benefit provision variable.
Heywood
- 135
.
The concentration ratio is created as the sales weighted average of the individual concentration ratios in the four digit industries that make up the approximately eighty industries defined in the manufacturing sector of the Quality of Employment Survey. An attempt was made to correct the ratio for the degree of import penetration. This correction fails to alter any of the results fundamentally and the associated regressions are not reported.
.
Analogous comparisons could be made using any degree of statistical significance. For example, three of the concentration coefficients remain significant at the 5% level while four of the union coefficients remain significant at that level.
.
Note again that only provision of fringe benefits is examined. It remains possible that although the interaction is negative in the probability equation, it would be positive in an appropriate micro benefit level equation. As mentioned, such data does not exist.
.
Indeed, in only three cases does a significantly positive net coefficient emerge for union members (eye care, thrift and sick days).
.
These measures are computed by first evaluating the standard normal distribution at the mean values of the explanatory variables times their coefficients, and then multiplying that evaluation by the sum of the concentration coefficient plus the interaction coefficient times the mean union level.
10.
Approximately 250 respondents willingly gave an exact estimate of their annual earnings on the. job. Another approximately 50 respondents refused to give an exact estimate, but were willing to identify their job income as being in one of ten categories. To retain these respondents in the sample they were assigned the midpoint of the relevant range as their income.
136
-
Industrial
Concentration
References
Akerlof, George. "Gift Exchange and Efficiency Wages: Four Views," American Economic Review, Vol. 74, No. 2 (May 1984):79-83. Alpert, William. "Unions and Private Wage Supplements," Journal of Labor Research, Vol. 3, No. 2 (Spring 1983): 179-199. "Manufacturing Workers Private Wage Supplements: A Simultaneous Equations Approach," Applied Economics, Vol. 15 (1983):363-378. "Unionism, Ashenfelter, Orley and George Johnson. in U.S. Relative Wages and Labor Quality Economic Manufacturing Industries," International Review, Vol. 13 (October 1972):488-508. Belman, Dale. "Firm Resistance to Unionization: The Direct and Indirect Effects of Industry Structure on Union Membership and Wages," Unpublished Ph.D. dissertation, University of Wisconsin - Madison, (1986). and John Heywood. "The ConcentrationEarnings Hypothesis: Reconciling Individual and Industry Studies," Working paper, University of Wisconsin-Milwaukee (1988). Blanchflower, David. " W a g e s and Concentration in British Manufacturing," Applied Economics, Vol. 18 (1986):1025-1038. Brush, Brian and Steve Crane. "The Effect of Market Power on the Fringe Share of Labor Compensation," Quarterly Journal of Business and Economics, Vol. 24, No. 4 (Autumn 1985):70-84. Cymrot, Donald. "Private Pension Coverage and Market Structure," Working paper No. 85-05, Miami University, March 1986.
Heywood
- 137
Dalton, James and E. Ford. "Concentration and Labor Earnings in Manufacturing," Industrial and Labor Relations Review, Vol. 31 (October 1977):45-60. Dickens, William and Lawrence Katz. "Interindustry Wage Differences and Industry Characteristics," National Bureau of Economic Research, Working paper No. 2014 (September 1986). Fosu,
Augustin. "Impact of Unionism on Pension Fringes," Industrial Relations, Vol. 22 (Fall 1983):419-425. "Unions and Fringe Benefits: Additional Evidence," Journal of Labor Research, Vol. 5 (Summer 1984):247-53.
Freeman, Richard. "The Effect of Unionism on Fringe Benefits," Industrial and Labor Relations Review, Vol. 34 (July 1981):419-509. and James Medoff. "The Impact of Percentage Organized on Union and Nonunion Wages," Review of Economics and Statistics, Vol. 63, No. 4 (November 1981):561-572. Heywood, John. "Labor Quality and the ConcentrationEarnings Hypothesis," Review of Economics and Statistics, Vol. 68, No. 2 (May 1986):342-349. "Product Market Structure and the Labor Market," Unpublished Ph.D. dissertation, University of Michigan (1986). Jenny, F. "Wage Rates, Concentration, and Unionization in French Manufacturing Industries," Journal of Industrial Economics, Vol. 26 (June 1978):315-327. Kawashima, Yoko and Toshiaki Tachibanaki. "The Effect of Discrimination and of Industry Segmentation on Japanese Wage Differentials in Relation to Education," International Journal of Industrial Organization, Vol. 4, No. 1 (March 1986):43-68.
138
-
Industrial Concentration
Kwoka, John. "Monopoly, Plant and Union Effects on Worker Wages," Industrial and Labor Relations Review, Vol. 36, No. 2 (January 1983):251-258. Long, James and Albert Link. "The Impact of Market Structure on Wages, Fringe Benefits and Turnover," Industrial and Labor Relations Review, Vol. 36, No. 2 (January 1983):239-250. Mellow, Wesley. "Employer Size, Unionism and Wages," Research in Labor Economics, Vol. I (1977):185. "Employer Size and Economics and Statistics, Vol. 1982):494-501.
Wages," Review o f 64, No. 3 (August
Miller, Edward. "Large Firms are Good for their Workers," Antitrust Bulletin, Vol. 26, No. 1 (Spring 1981):145-154. Podgursky, Michael. "Unions, Establishment Size, and Intra-Industry Threat Effects," Industrial and Labor Relations Review, Vol. 39, No. 2 (January 1986):277284. Stafford, Frank. "Concentration and Labor Earnings: Comment," American Economic Review, Vol. 58, No. 1 (March 1968):174-178. Weiss, Leonard. "Concentration and Labor Earnings," American Economic Review, Vol. 56, No. 1 (March 1966):96-117.