The determinants and extent of UAW pattern bargaining.
Under the UAW's bargaining tradition, once the union selects a target negotiators work to achieve an agreement with that company that can serve as a pattern for a contract with other employers. (Solidarity, September 1-15, 1987, p.8)
Although pattern bargaining has been recognized as a strategy of many unions, the UAW, with 917,000 members in 1989, is perhaps its best-known and most influential proponent. The UAW's policy of pattern bargaining directly affects workers in a wide range of industries and occupations.
To gain a better understanding of UAW pattern bargaining, in this paper I estimate the degree of pattern-following exhibited in UAW contractual wage outcomes in the 1955-79 period and in the 1987 bargaining round. More important, I evaluate the roles of various factors in determining how closely the pattern is followed, and how these roles have changed in the 1980s, to increase our knowledge of wage determination in unionized firms. To the extent that the forces underlying pattern bargaining before 1980 are still present today, a deeper understanding of the earlier era may aid our comprehension of contemporary events.
The importance of wage patterns and spillovers in the determination of wages in unionized firms, especially before 1980, is the subject of much debate. Two notable studies of UAW pattern bargaining, Levinson (1960) and Alexander (1961), have been subject to differing interpretations, with some authors drawing attention to the deviations from pattern settlement that were found (Flanagan 1976; Burton and Addison 1977; Mitchell 1980) and others emphasizing the uniformity (Katz 1987). Interindustry spillovers are even more hotly debated. For example, McKersie (1961), Seeber (1987), and Bourdon (1979) emphasize the influence of auto settlements on negotiations in agricultural implements and aerospace in the 1970s, whereas Hirsch and Addison (1986) and Mitchell (1982) strongly question the importance of interindustry spillovers. This paper helps resolve these debates by adding to our understanding of the role of patterns in UAW wage determination, especially across industries.
In addition, the role of pattern bargaining before 1980 is a focal point of the debate over whether the observed behavior of the 1980s marks a fundamental change in collective bargaining. For example, Freedman and Fulmer (1982) argue that pattern bargaining was a crucial characteristic of collective bargaining prior to 1980 but that the 1980s were significantly different because of a lack of patterns, whereas Ready (1990) argues that patterns were still important in the 1980s and Mitchell (1983) argues that patterns were never an integral feature of unionized wage determination.
Finally, evaluating the role of pattern bargaining is important for understanding trade unionism. Contemporary union modeling focuses on a single union maximizing some type of wage-based objective function in negotiations with a single firm. If, however, within-union spillovers are important in UAW wage determination, some rethinking of union goals will be needed.
In this paper, two new microeconomic-level data sets of UAW contract wage settlements are assembled to estimate the effects of the UAW's target wage settlement on subsequent UAW negotiations. This study improves on Levinson (1960) and Alexander (1961) in several important ways. The data used here span a broader range of industries, allowing a fuller analysis of interindustry wage spillovers. Firm-specific data are also collected so that the effect of firm-specific conditions on the wage bargain can be estimated. Methodologically, the larger data sets developed here allow a statistical evaluation of the role of determinants of pattern-following in a regression framework, as opposed to reliance on cross-tabulations. Finally, whereas Levinson and Alexander examined only wage increases, both wage increases and wage levels are analyzed here.
UAW Bargaining Behavior
By North American standards, the UAW is a very large, broad-based industrial union. Membership is not confined to the automobile assembly industry, but rather jurisdiction is claimed
over all employees of plants and shops engaged in the manufacture of parts (including tools, dies, etc.), and the assembly of these parts into farm, automobile, automotive propelled products, aircraft and agricultural implements. (1957 Constitution, Article 5)
In addition, the UAW Constitution provides for jurisdiction in "such other branches of industry" as the union decides in conjunction with the AFL-CIO (Article 5). The vast scope of the union is underscored by the range of goods produced by UAW members-refrigerators, clocks, mattresses, floor wax, model trains, butter, and trumpets, to name but a few consumer products (UAW 1989).
Except during the concessions of the early 1980s, the UAW has begun each of its three-year bargaining rounds by establishing a target settlement with one of the Big Three automobile assembly companies.(1) For example, in 1979 and 1990 the target was General Motors and in 1976 and 1987 it was Ford. The UAW then bargains for this same settlement at the other auto producers. More important, given the UAW's broad industry coverage, the outcome of the key bargain has historically set the pattern for all UAW negotiations - with automobile manufacturers, farm implement producers, aerospace equipment firms, and producers, in other industries - especially before 1980.(2) Although the pattern is not always achieved, these interindustry pressures make the UAW case especially interesting and important.
To better understand the sequential nature of the UAW bargaining process, consider a typical negotiation round - 1973.(3) Negotiations with the Big Three auto firms to replace the contracts that expired on September 14, 1973, began in July of that year. With the announcement on August 21 of Chrysler as the target, negotiations finally began in earnest at Chrysler. With no settlement by September 15, UAW Chrysler workers went on strike while Ford and General Motors workers continued working under temporary contract extensions. Only after Chrysler workers completed ratification of the new agreement and returned to work could negotiations begin at Ford; and not until the UAW Ford council approved the Ford settlement did serious negotiations at General Motors begins.
Contracts with the major agricultural implement firms, John Deere, International Harvester, and Caterpillar, expired 15 days after the auto contracts in 1973, allowing negotiations in those firms to follow the settlements in the auto industry. Similarly, contracts at many auto parts firms expired shortly after the auto contracts. The UAW's procedure therefore consisted of two stages: one, selection of the target firm and settlement of the target contract; and two, subsequent negotiations with other firms. This analysis concentrates on the second stage-the attempt to achieve the terms of the pattern settlement at other firms.
It was once believed that pattern bargaining characterized collective bargaining in many industries: for example, steel (Seltzer 1951), flat-glass manufacturing (Bain 1969), and rubber and meat-packing (Carpenter and Handler 1961). Since at least the late 1970s, however, many scholars have sharply questioned the importance of patterns (for example, Shulenburger 1978; Flanagan 1978).(4) Despite the explicitness of the UAW's pattern bargaining agenda, there appear to be only three studies of the degree of pattern-following in wages achieved by the UAW outside of the Big Three. Levinson (1960) examined settlements at 85 firms in the Detroit area during 1946-57 and classified each settlement as above, below, or exactly equal to the Big Three settlement pattern. Similarly , Alexander (1961) analyzed 37 automobile parts firms in the Detroit area in 1959. Levinson and Alexander both found that the closeness of a firm's relationship to the automobile industry, its size, and its ability to pay were important determinants of the amount of pattern-following observed. Douglas (1958) undertook an ambitious analysis of grievance procedures, union status provisions, and fringe benefits, in addition to wages, among UAW contracts with eight companies. He found a high degree of uniformity in a
variety of provisions - wages, pensions, supplemental unemployment benefits (SUB), and union bulletin boards - but variability in other (primarily non-wage) provisions.
Thus, these few previous studies confirm the existence of pattern-following within UAW contracts. Firm size, relation to the auto industry, and ability to pay are found to be important influences.
There is also a body of work that examines a more general wage spillover hypothesis. This literature examines the economy-wide transmission of wage settlements that are first negotiated in key bargains (Burton and Addison 1977). The empirical implementation in such studies, however, tends to be ad hoc, especially in identification of aggregate wage rounds and key bargains: Eckstein and Wilson (1962) defined the key group to be durable manufacturing, Vroman (1984) used the steel and auto settlements, and Christofides, Swidinsky, and Wilton (1980) tried a variety of specifications. Note carefully, however, that there is no such identification problem in this study: the UAW has well-defined key bargains and bargaining rounds.
Critics of wage spillover theories have questioned the existence of wage rounds (Burton and Addison 1977). Yet, the numerous UAW settlements that shortly follow the target settlement clearly support the existence of wage rounds within the UAW. Figure 1 shows the fraction of UAW settlements (reported in Current Wage Developments) in each round that are settled in each quarter of the bargaining round. A very high proportion of settlements closely follow target settlements (which are indicated by vertical lines). In fact, on average before 1979, 72% of the non-target UAW workers in the data set I assembled from Current Wage Developments (described below) were covered by contracts that settled within three months after the target. Figure 1 also confirms the popular belief that wage rounds deteriorated in the 1980s.
Data on UAW Contacts
To analyze the importance of pattern bargaining in UAW contract settlements, two new contract-based data sets are constructed. The unit of observation in each data set is a UAW contact with a firm. One data set focuses on wage rates for two narrowly defined occupations, and the other focuses on negotiated wage increases. More details are available in a data appendix (not printed here, but available from the author on request). Appendix Table A1 summarizes the sources for the variables used in the analysis.
The negotiated wage increases are collected from the Bureau of Labor Statistics publication Current Wage Developments, which reports the annual improvement factors and cost of living increases for many contracts. The unit of observation is the wage increase over the life of a UAW contract at one firm. Collection was based simply on completeness of data for contracts settled after the Ford contract in June 1955 and before the General Motors contract in September 1979. The resulting sample contains 636 contracts in 95 UAW firm bargaining pairs and covers approximately 60% of all UAW workers. The data do not include the 1979 round because major 1979 contracts were reopened early in 1982 in a round of concessionary bargaining.(5) To test whether the UAW's behavior has changed, 41 contracts from the 1987 bargaining round are also analyzed. A number of contracts can be matched to firm-level balance sheet and stock return data, allowing an analysis of the importance of firm-specific conditions.
Most of the contracts are multi-year agreements, which provide for multiple wage increases over the life of a single contract. In addition, many contracts follow the two-part wage contract pioneered in the 1948 UAW-GM agreement in stipulating contingent and noncontingent increases (Harbison 1950; Garbarino 1962). The noncontingent annual improvement factors are specified in the contract and normally occur once each year. Typically, there are also quarterly cost of living adjustments (COLAs). The annual improvement factors and COLAs are combined into a compound increase measure for each contract and converted to annual terms.
The second data set is based on wage levels at the effective dates of contracts negotiated by the UAW. Many union contracts contain the resulting base wage rates as an exhibit or appendix to the agreement - a valuable, yet seemingly overlooked, data source. UAW wage rates were collected primarily from written agreements (that is, contracts) between the UAW and numerous firms. In addition, wages were assembled from Bureau of Labor Statistics publications and personal communications with companies.
In this data set, the wage variable is the wage rate from the contract. Within each labor grade, most contracts have a range of rates or a fixed progression from a starting rate to the job rate. The job rate - the highest rate except for a leader rate - is used in this analysis. The analysis concentrates on the wages of two occupations: janitors (who typically have the lowest rate in the structure) and tool and die makers (who often have the highest).(6) The resulting sample consists of 321 janitors and 294 toolmakers with effective dates ranging from 1948 to 1981. More recent data are not as comprehensive, but some evidence from the 1987 round is discussed below.
Because the UAW explicitly identifies a target firm, in this study there is no ambiguity regarding the key bargain (unlike in studies of economywide interindustry wage spillovers). For each round, we can thus define the target wage and increase as the wage and increase resulting from the contract settlement with the target firm.(7) A straightforward gauge of the amount of pattern-following is the mean of the ratio of the contract wage to the target wage, in increase and level terms. Within the Big Three automakers, the mean of this ratio is 0.981 for janitors' wages and 1.00 for annualized percent increases in contract wages.
As Table 1 illustrates, the degree of uniformity is quite striking outside the Big Three, too. At lest before 1980, wage levels at the agricultural implement firms and aerospace companies were very close to those at the Big Three. Only Maytag had wage rates near the lower aggregate level for manufacturing in the Midwest. The percent increase settlements exhibit the same pattern.
[TABULAR DATA OMITTED]
Although the simple mean of the ratio of contract wages to target wages is useful for obtaining a quick impression of wage patterns, a stronger analysis call for the use of controls. In fact, a prominent criticism of empirical estimates of wage spillovers is that the results may reflect aggregate wage movements rather than union wage patterns. Thus, in this study I use multivariate regression analysis to judge the statistical significance of the variables of interest while controlling for a variety of factors. Most notably, aggregate wage movements can be controlled for so that the estimated spillovers are in addition to a market-driven component. Equations of the following type are estimated: (1) [W.sub.ijt] = [delta]w.sup.T.sub.t] + [beta.sub.jw.sup.T.sub.t] + GAMMA'[CHI.sub.ijt] where [W.sub.ijt] is the log contract wage or percent increase in the contract wage at firm i in industry j at time or settlement round t, [w.sub.t.sup.T] is the target settlement, and [X.sub.ijt] contains firm, industry, and aggregate controls.
The extent of pattern-following in both wage increases and wage levels is analyzed. The primary parameter of interest is delta, the estimated target spillover effect over all industries. Estimates of [beta.sub.j] illuminate interindustry differences in the amount of the target spillover into subsequent UAW negotiations. Additional understanding of UAW wage determination is gained by estimating how importance of the target, [delta + beta.sub.j], varies across firm sizes and financial conditions. The regressions contain industry and aggregate indicators of labor and product market conditions to analyze the degree to which the UAW is able to pursue its pattern bargaining strategy after accounting for market-based constraints.
Column 2-5 of the top panel of Table 2 present the ordinary least squares (OLS) regressions of the percent increase in the contract wage on the target wage and product and labor market indicators. These results indicate that the target wage is statistically very important. The coefficient is estimated very precisely, with only 652 observations.(8) In addition, the aggregate variables, especially the national average hourly earnings, unemployment rate, and consumer price index, appear to be determinants of the wage increase (column 2), but the inclusion of the target wage significantly diminishes their importance (column 3). To better control for local labor market conditions, state unemployment rates were used as well, but the results were unchanged. The target effect is still significant after the inclusion of industry and region effects.(9) This data set has a panel aspect because the same bargaining pairs are often present in several years. The results in column 4 and the importance of the target are unaltered, however, after allowing for bargaining-pair fixed effects.
Whereas the pattern settlement is normally phrased in terms of a wage increase, UAW rhetoric has historically focused on uniformity of wage levels. Furthermore, if workers actually care about equity, or firms care about parity, one would expect them to care about wage levels. Therefore, an understanding of UAW wage determination requires an investigation of spillover effects using UAW wage levels data.
The bottom panel of Table 2 examines the statistical importance of the influence of the target wage level on the subsequent UAW settlements. The amount of uniformity found is extremely high-quite close to unity for both janitors and toolmakers. Note that the aggregate wage level is important when the target wage is not controlled for, but the effect of the target eclipses the role of product and labor market forces.
[TABULAR DATA OMITTED]
Implicit Table 2 is the assumption that the importance of the target is uniform across industries. Both Alexander (1961) and Levinson (1960), however, using narrower data sets, found industry to be an important determinant of the degree to which the pattern is followed. More specifically, UAW contracts with firms that were "closer" to the automobile industry, in terms of final products and therefore cost and demand conditions, were influenced more strongly by the target settlement than were contracts with firms "farther away" from the auto industry. The UAW represents workers in many industries, and understanding the role of pattern bargaining in those industries is crucial for understanding the UAW's behavior.
The data sets used here allow an exploration of a much wider range of industries that in the earlier works. Column 1, 4, and 6 of the Table 3 present the industry distributions of the UAW contract samples. Only 22% of the contracts are in the automobile assembly and parts industry (SIC 371); 34% are in aerospace (SIC 372) and agricultural and construction machinery (SIC 3523 AND 3531). Thus, 44% of the contracts are outside the "core" industries of the UAW.(10) On the other hand, over 70% of the workers are in the auto industry (SIC 371), and aerospace and agricultural implements each have approximately 10% . Thus, although 44% of the contracts are in other industries, only 10% of the workers are.(11) The janitor and toolmaker industry distributions exhibit a similar pattern.
[TABULAR DATA OMITTED]
To test for significant interindustry deviations from pattern-following, Table 3 allows the effect of the target to vary by industry. (Unfortunately, this table is limited to the pre-1980 period, because industry-cell sizes for data from the 1980s are quite small. Differences in the 1980s are discussed below.) In the percent increase data (columns 2 and 3), controlling for industry and region effect and allowing the importance of the target to vary across industries yields an estimate for the target wage of 0.912, even with controls for product and labor market forces. The negative sign on all of the interaction terms in columns 2 and 3 reaffirms that the pattern is followed most closely within the Big Three. Few coefficients of the interacted terms are statistically significant at conventional levels, however, and an F-test fails to reject the null hypothesis that all are zero at conventional levels of significance.(12) Especially striking is the uniformity within auto parts, agricultural implements, and airplane assembly. There are, however, deviations from the pattern as well - in fabricated metals, other aerospace, and non-electrical machinery except agricultural implements and construction equipment. Allowing for individual bargaining-unit fixed effects does not alter the results. In summary, although some statistically significant industry deviations are present, the importance of the target is large across a wide range of industries.
Columns 5 and 7 of Table 3 show the results of a test for interindustry spillovers in wage levels. As in the wage change data, the Big Three wages move quite closely with wages in other firms in the auto and agricultural implement industries. Those industries "farther away" from the auto industry, in terms of UAW emphasis and industry cost and demand conditions, are below pattern; they are electrical equipment, machinery except electrical, and miscellaneous manufacturing. At the 5% level, F-tests reject the hypothesis that the interacted coefficients are all zero, so interindustry differences are found. Given that little difference was found in percent increases, this result suggests that the UAW is more successful in achieving uniformity in increases than in levels. Table 3 also identifies the limit of the target's influence. The magnitude of the coefficient for other manufacturing implies that UAW wages are determined in these industries without regard for the target settlement. This important result suggests a boundary on the extent of transmission of an intra-union target spillover.
[TABULAR DATA OMITTED]
Yet, the importance of the target wage in the other industries is quite strong and should not be overlooked. The results of Table 3 support the finding of previous studies that industry is an important determinant of pattern-following, but the previous studies have stressed the link to the auto industry via similar cost and demand conditions. I agree that industry is important, but using broader data sets, a high degree of uniformity is found in those industries traditionally emphasized by the UAW, but not necessarily closely linked economically to the auto industry - agricultural implements and aerospace.
Bargaining Unit Size and Firm Size
In addition to a finding of industry effects, previous studies have found that firms with more employees followed the pattern settlement more readily. Levinson (1960), for example, found firm size to be an important determinant of pattern-following, but was unsure whether this result reflected the greater profitability of larger firms or a greater effort by the UAW to make the pattern hold at larger firms. If smaller units are allowed to deviate from the pattern due to the perceived financial weakness associated with small manufacturing operations, the important determinant is is firm size. I believe, however, that the relevant consideration is the size of the UAW bargaining unit, because the UAW sees large bargaining units as leaders and therefore puts great effort into ensuring that they follow the pattern. The UAW's insistence on pattern-matching settlements is much harder to maintain if management negotiators can point to a major deviation earlier in the round.
Table 4 tests the importance of bargaining unit size and firm size on pattern-following in a framework similar to that of Table 3. Note that columns 1 and 2 are the results of a single regression that allows the target spillover not only to vary by bargaining unit size and firm size, but also to have different effects before and after 1980. Before the 1980s, the size of the bargaining unit, not firm size, appears to have been an important determinant of pattern bargaining. The smallest units felt the influence of the target the least - and the difference is statistically significant. The coefficients for the number of employees are insignificant and reveal no meaningful pattern. For the wage levels data, bargaining unit size is not available, but no meaningful differences across firm size are found in the wage regressions. This finding supports my hypothesis that firm size is not important.
[TABULAR DATA OMITTED]
An alternative explanation, however, is suggested of the portion of a firm's production that can be halted by a strike is viewed as key to the bargaining power of larger bargaining units (Ozanne 1967). This alternative is tested in columns 3 and 4 of Table 4 by directly assessing the effect of the percentage of a firm's employees represented by the UAW. Before the 1980s, this measure of bargaining power has a statistically significant effect, but only at those firms in which less than 5% of the work force was UAW-represented. No other meaningful differences are found. Rather, the results suggest that bargaining unit size per se is an important determinant of pattern bargaining.
Further support for the idea that initial pattern-following precedents are important is found in the timing of settlements within each UAW bargaining round. The union may feel extra pressure to enforce the target settlement at negotiations early in the bargaining round so that firms in subsequent negotiations cannot point to previous deviations. Later settlements, however, will not be as important for establishing the applicability of the pattern. Further, if pattern-following is important to the UAW, one would expect that the union would first try to settle with firms in which achieving the pattern is thought to be less difficult. Thus, a negative coefficient on a variable for the number of months since the target settlement, as interacted with the target wage, is predicted. A negative and significant coefficient is indeed found, but only for the pre-1980 period: -0.006 with a standard error of 0.001. A similar results is found for wage levels: for janitors, the estimated coefficient of months since the target settlement interacted with the target wage is - 0.001 with a standard error of 0.0003.
The regressions in Table 4 also highlight some differences between the UAW's behavior in the 1980s and its behavior in earlier years. Whereas firm size was not important before 1980, there was a distinct tendency for smaller firms to deviate from the pattern in the 1987 UAW bargaining round. Moreover, bargaining unit size was less important in the 1987 round, as was the timing of settlements (not reported). These results support the popular notion that financial conditions and ability to pay were more significant in unionized wage determination in the 1980s than before.
Ability to Pay Before 1980
Most studies of pattern bargaining have posited that the target is not an inflexible standard to be met in all instances (Seltzer 1951; Levinson 1960; Bourdon 1979). In particular, outcomes at firms that are in a weak financial state are thought to deviate from the pattern. This view has been expressed most pointedly in regard to the 1980s, but it has also been asserted in discussions of the earlier period. On the other hand, standard UAW rhetoric is that differences in ability to pay arise due to differences in management, not workers, and "inefficiency in management should [not] be subsidized out of workers' living standards" (MacDonald 1963:215). The academic view predicts that firm-specific measures of profitability will influence the amount of pattern bargaining, whereas the prediction based on UAW rhetoric is that industry measures should matter. A way to empirically test these propositions is to interact a measure of firm or industry financial health with the target wage and evaluate the level of statistical significance that results.
In the firm-specific case, various different financial health measures can be constructed. Alternative specifications include equity returns for various periods prior to settlement, change in return on equity, operating income and net sales, changes in net income and sales, and income and sales per assets. None of these indicators are statistically significant before the 1980s. Table 5 presents two measures - operating income per total assets (columns 2 and 3) and average return on equity for the twelve months prior to the settlement date (columns 4 and 5). The lack of importance of these profit indicators is unaltered by the inclusion of industry ability to pay measures (columns 3 and 5).
[TABULAR DATA OMITTED]
The lack of statistical evidence that firm-specific conditions affect the amount of pattern-following contradicts a commonly held belief. One problem complicating the interpretation of these results, however,is the difficulty of defining the appropriate scope of a firm's profitability. Many of the UAW's bargaining units are in large, diversified companies, and it is not clear whether the most compelling definition of profits is on a plant,division, or corporation level. Although the UAW may represent workers in only one plant of a firm, the measures available are corporate level variables. It id conceivable that the statistical influence of firm-specific conditions is more important when the UAW bargaining units are involved in a larger fraction of the corporation's business.
As a proxy for that source of influence, a measure of the fraction of employees in the entire corporation represented by the UAW was constructed for the firms in the wage increase data. The results show that the lack of statistical importance of firm-specific conditions is invariant to inclusion of this measure. Although further investigation of the appropriate scope of profits in union bargaining may be useful, the observed insignificance of firm-specific conditions is consistent with UAW rhetoric.
UAW rhetoric regarding industry conditions can also be tested in this framework. As a measure of industry ability to pay, the four-digit SIC industry ability value added per production man-hour was calculated from the Census of Manufactures. In the percent increase data before 1980, this measure in not an important influence on pattern-following except in the specification that also controls for a firm's equity return. On a firm level, the resulting data set is too small for meaningful analysis after firm-specific profits are incorporated, but column 6 of Table 5 shows that firms in an industry with greater ability to pay follow the target wage more closely, at least for janitors. Overall, before 1980, ability to pay did not substantially affect UAW collective bargaining outcomes, contrary to a widely held assumption.
Evidence for the 1980s
It is widely believed that in many industries once characterized by a high degree of uniformity in wage settlements, ability to pay became a more significant factor in wage negotiations during the 1980s (Lipsky and Donn 1987). This purported shift, in fact, is what Freedman and Fulmer (1982) referred to as a fundamental change in collective bargaining. The below-pattern settlements negotiated at Chrysler beginning in 1979, with the UAW gaining a seat on Chrysler's board, are often cited as evidence. But in light of the fact that Chrysler workers have now regained wage rate parity with Ford and General Motors, it is important to understand whether the events of the early 1980s truly changed collective bargaining in a fundamental and lasting fashion.
Table 5 illustrates a striking difference between the 1955-79 experience and the 1987 UAW bargaining round: ability to pay, on both industry and firm levels, was much more important in UAW wage determination in the 1987 round than in the earlier years. As noted above, Table 4 shows significant differences in the role of bargaining-unit size and firm size between the pre- and post-1980 samples, differences that are likely related to ability to pay. Figure 1 illustrates the decline, after about 1979, of bargaining rounds in which many settlements were concluded in the immediate aftermath of the target settlement. The time between expiration of the auto contracts and expiration of the agricultural implement contracts with Deere and Caterpillar is no longer 15 days, but rather an entire year. Still, do these changes mean that pattern bargaining is dead?
For a comparison of wage settlements before 1980 with settlements later in the 1980s, turn again to Table 1.Wage rates at the Big Three are closer than ever, as illustrated by the General Motors comparison, but differences arise from differing payments under profit sharing plans (Katz and Meltz, 1991). Similarly, wages at the major agricultural implement firms have not significantly eroded relative to those in the automobile industry. Evidence from the aerospace industry suggests that previous levels of pattern-following have not been maintained, but it is not obvious that the differences are fundamental. One should also note, however, that Bell and Martin Marietta have two-tier wage plans.
Further industry comparisons are presented in the lower portion of Table 1. The results here are more striking. Settlements in the agricultural equipment industry matched those in automobile assembly, and the two are still quite closely linked. On the other hand, in those industries not traditionally associated with the UAW - metal products, electrical equipment, and miscellaneous manufacturing - the pattern has clearly not been maintained. In auto parts, too, the ability to achieve the pattern has eroded. Together, these findings suggest that pattern bargaining within the UAW has certainly declined in strength, but that it is not necessarily defunct.
This study has empirically examined the bargaining behavior of the UAW, with a specific focus on pattern bargaining - the linkage of wage settlements to an initial target settlement. An analysis of both wage levels and percentage wage increases across different contracts for the years 1955-1979 and the 1987 bargaining round has shown statistically significant spillover effects from target settlements. Depending on the specification, the coefficient on the appropriate target wage is generally in the 0.80 to 0.90 range - and always highly significant - across a diverse range of industries.
An analysis of the determinants of how closely the pattern is followed shows that larger bargaining units and (for the period before 1980) units that settled shortly after the target settlement was concluded have tended to follow the pattern more closely than other units. Earlier studies emphasized the influence on pattern-following of industry product demand and cost linkages to the auto industry, but I have found a high degree of pattern-following in agricultural implements and aerospace - two industries with few economic ties to the auto industry, but with political and historical UAW links to it. Most striking is the finding that, contrary to commonly held beliefs, pattern-following does not seem to have been greatly influenced by firm-specific financial health indicators before 190. On the other hand, industry and firm-specific ability to pay was of much greater importance in the 1987 bargaining round.
In summary, the evidence supports the idea that the importance of pattern bargaining has declined and that ability to pay was more important in the late 1980s than in earlier years. Yet, the present is still characterized by some of the same pressures that were apparent in the time when pattern bargaining and wage spillovers played a primary role in UAW wage determination. After settlements with General Motors and Ford, UAW Chrysler negotiators began negotiations in 1990 wearing "Pattern or Picket" buttons,and the pattern was achieved. The 1990 settlement with J. I. Case is explicitly linked to others in the the industry by guarantees of wage and pension increases that equal those negotiated at Deere and Caterpillar in the fall of 1991. Thus, if the UAW's bargaining experience is at all representative of other unions in the 1980s, pattern bargaining has diminished in scope and scale but is nevertheless still a significant force. Still needed, however, is an understanding of why unions placed such great emphasis on wage patterns before 1980. With such an understanding, we might better comprehend trade union behavior and decipher the continuing influence of the past in current collective bargaining.
(1) The Big Three automakers are General Motors, Ford, and Chrysler. A bargaining round is defined as starting with the settlement at the target firm and encompassing all settlements up to, but not including, the next target settlement. (2) For example, consider a statement in a consultant's report to a manufacturer of school furniture: "American Seating Company's interest in making early preparations for 1958 negotiations stems from its need to escape the UAW's automobile industry pattern of wages and fringe benefits." (Recommendations Concerning Preparation for 1958 Negotiations," report by Fred Rudge Associates, Inc., to American Seating Company, July 31, 1957. From the files of Fred Rudge Associates, Inc., Industrial Relations Library, Princeton University.) (3) For a detailed description of negotiations in the agricultural implement industry from 1937 to 1958, see Melcher (1964), especially Chaps. 6-7. For more on the General Motors negotiations of 1970, see Serrin (1973), and for UAW negotiations in Canada in the early 1980s, see White (1987). (4) Additionally, a considerable literature before 1970 investigated the role of unions in erasing wage differentials at the occupational, interplant (including interfirm), interindustry, and geographical levels. Prominent examples are McCage (1912), Lester and Robie (1946), Reynolds (1957), Kerr (1957), and MacDonald (1963). These studies found that unions did indeed strive for wage standardization across firms as well as within firms, and found strong evidence that many unions drove the wage structure toward greater uniformly within industries. (5) Furthermore, after 1979 the data coverage in Current Wage Developments deteriorates.
The events of the early 1980s have been well documented and need not be repeated here. See Katz (1985, 1987) and Seeber (1987) for more on the UAW; for a more general look, see Lipsky and Donn (1987) and Kochan, Katz, and McKersie (1986). (6) A potential problem is that some contracts simply enumerate the labor grades and wage rates without indicating which job pertains to which grade. Not all of the janitor observations are known to be the wages of janitors; for 27% of these observations, all that is known is that this wage rate is the lowest in the rate schedule. Since there rarely are any jobs classified below janitors, however, and all plants need janitors, I do not see this uncertainty as serious. The tool and die maker case, on the other hand, may present more of a problem. Not all plants have toolmakers, so the data will be biased downward if, for example, the highest rate is actually that for electricians. (7) Because wage data were not available for every quarter for Chrysler or GM, the Ford wage is always used as the target wage level in the wage levels data set. Since the wages of the Big Three were nearly identical in this period, any resulting bias should be very small. (8) A strong form of the spillover hypothesis would predict a coefficient of one. This hypothesis is rejected at conventional levels of significance, but a spillover of 0.70 is meaningful nontheless.
The reported standard error do not account for the possible presence of heteroskedasticity. The results do not change, however, if standard errors calculated from White's (1980) heteroskedasticity-consistent covariance matrix estimator are used. (9) The F-value (p-value) for the null hypothesis that all the industry coefficient are zero is 2.89 (0.001); for the hypothesis that all region coefficients are zero, it is 1.03 (0.407). (10) Agricultural implements and aerospace are "core" UAW industries because of historical and political UAW factors, not necessarily because of an industrial link to the auto industry. A wide range of industries are represented in the remaining 44%, including makers of school furniture, household appliances, firearms, drapery hardware, pens, rulers, typewriters, and cigar-making machinery. (11) This membership distribution roughly matches that reported in UAW (1986). (12) The F-value is 1.28 with a p-value of 0.233.
Alexander, Kenneth. 1961. "Market Practices and Collective Bargaining in Automotive Parts. "Journal of Political Economy, Vol.69, No.1 (February), pp.15-29. Bain, Trevor. 1969. "Flat Glass: Industrial Peace Revisited. "Industrial Relations, Vol.8, No.3 (May), pp.259-68. Bourdon, Clint. 1979. "Pattern Bargaining, Wage Determination, and Inflation: Some Preliminary Observations on the 1976-78 Wage Round." In Michael J. Piore, ed., Unemployment and Inflation: Institutionalist and Structuralist Views. White Plains, N.Y.: M. E. Sharpe, pp.115-33. Bureau of National Affairs. Various issues. Daily Labor Report. Washington, D.C.: BNA. Burton, John, and John Addison. 1977. "The Institutional Analysis of Wage Inflation: A Critical Appraisal." In Ronald G. Ehrenberg, ed., Research in Labor Economics. Greenwich, Conn.: JAI Press, pp.333-76. Carpenter, Walter H., Jr., and Edward Handler. 1961. Small business and Pattern Bargaining. Babson Park, Mass.: Babson Institute Press. Christofides, Louis N., Robert Swidinsky, and David A. Wilton.1980. "A Microeconometric Analysis of Spillovers Within the Canadian Wage Determination Process." Review of Economics and Statistics, Vol. 62, No. 2 (May) pp. 213-21. Douglas, Roy M. 1958. "Uniformity Among Labor Agreements: A Comparison of UAW Labor Agreements in the Automobile and Agricultural Implement Industries, 1947-1956." M.A. diss., University of Illinois. Eckstein, Otto, and Thomas A. Wilson. 1962. "The Determination of Money Wages in American Industry." Quarterly Journal of Economics, Vol. 76, No. 3 (August), pp. 379-414. Flanagan Robert J. 1976. "Wage Interdependence in Unionized Labor Markets." Brookings Papers on Economic Activity, No. 3, pp. 635-73. _______. 1978. "Discussion." Proceedings of the Thirtieth Annual Meetings (New York, Dec. 28-30, 1977). Madison, Wis.: Industrial Relations Research Association, pp. 211-13. Freedman, Audrey, and William E. Fulmer. 1982. "Last Rites for Pattern Bargaining." Harvard Business Review, Vol. 60, No. 2 (March/April), pp. 30-48. Garbarino, Joseph W. 1962. Wage Policy and Long Term Contracts. Washington, D.C.: The Brookings Institution. Harbison, Frederick H. 1950. " The General Motors-United Auto Workers Agreement of 1950." Journal of Political Economy, Vol. 58, No. 5 (October), pp. 397-411. Hirsch, Barry T., and John T. Addison. 1986. The Economic Analysis if Unions. Boston: Allen & Unwin. Katz, Harry C. 1985. Shifting Gears: Changing Labor Relations in the U.S. Automobile Industry. Cambridge: MIT Press. ______. 1987. "Automobiles. "In David B. Lipsky and Clifford B. Donn, eds., Collective Bargaining in American Industry. Lexington, Mass: D.C. Health, Chap. 2. Katz, Harry C., and Noah M. Meltz. 1991. "Profit Sharing and Auto Workers' Earnings: The United States vs.Canada." Relations Industrielles, Vol. 46, No. 3 (Summer), pp. 515-29. Kerr, Clark. 1957. "Wage Relationships - The Comparative Impact of Market and Power Forces." In John T. Dunlop, ed., The Theory of Wage Determination. London: MacMillan, pp. 173-93. Kochan, Thomas A., Harry C. Kaltz, and Robert B. McKersie, 1986. The Transformation of American Industrial Relations. New York: Basic Books. Lester, Richard A., and Edward A. Robie. 1946 Wages Under National and Regional Collective Bargaining Research Report Series No. 73, Industrial Relations Section, Princeton University. Levinson, Harold M. 1960. "Pattern Bargaining: A Case Study of the Automobile Industry." Quarterly Journal of Economics, Vol. 74. No. 2 (May), pp. 296-317. Lipsky, David B., and Clifford B. Donn. eds. 1987. Collective Bargaining in American Industry. Lexington, Mass.: D.C. Health. MacDonald, Robert M. 1963. Collective Bargaining in the Automobile Industry. New Haven, Conn.: Yale University Press. McCabe, David A. 1919. The Standard Rate in American Trade Unions. Baltimore: The Johns Hopkins Press. McKersie, Robert B. 1961. "Negotiations in International Harvester Co. "In Arnold R. Weber, ed., The Structure of Collective Bargaining. Chicago: University of Chicago Press, pp. 279-303. Melcher, Arlyn J.1964. "Collective Bargaining in the Agricultural Implement Industry: The Impact of Company and Union Structure in Three Firms." Ph.D. diss., University of Chicago. Mitchell, Daniel J.B. 1980. Unions, Wages, and Inflation. Washington, D.C.: The Brookings Institution. _______. 1982. "How to find Wage Spillovers (Where None Exist)." Industrial Relations, Vol. 21, No.3 (Fall), pp. 392-97. _______. 1983. "The 1982 Union Wage Concessions: A Turning Point for Collective Bargaining?" California Management Review, Vol. 25, No. 4 (Summer), pp. 78-92. Ozanne, Robert. A Century of Labor Management Relations at McCormick and International Harvester. Madison, Wis.: University of Wisconsin press. Ready, Kathryn J. 1990. "Is Pattern Bargaining Dead?" Industrial and Labor Relations Review, Vol. 43, No. 2 ( January), pp. 272-79. Reynolds, Lloyd G. 1957. "The Impact of Collective Bargaining in the Wage Structure in the United States." In John T. Dunlop, e.d., The Theory of Wage Determination. London: Macmillan, pp. 194-221. Seeber, Ronald L. 1987. "Agricultural Machinery." In David B. Lipsky and Clifford B. Donn., eds., Collective Bargaining in American Industry. Lexington, Mass.: D.C. Heath, Chap. 3. Seltzer, George. 1951. "Pattern Bargaining and the United Steelworkers. "Journal of Political Economy, Vol. 59, No. 4 (August), pp. 319-31. Serrin, William. 1973. The Company and the Union: The "Civilized Relationship" of the General Motors Corporation and the United Automobile Workers. New York: Alfred A. Knopf. Shulenburger, David E. 1978. "Wage Leadership and Patterns of Settlement in Construction." Proceedings of the Thirtieth Annual Meetings (New York, Dec: 28-30, 1977). Madison, Wis.: Industrial Relations Research Association. pp. 185-92. UAW. 1986. We Make Our Own History: A Portrait of the UAW. Detroit. _______. 1989. Ammo, Vol. 26, No. 7 (June) U.S. Department of Labor, Various issues. Current Wage Developments. Washington, D.C.: Bureau of Labor Statistics. Vroman, Wayne. 1984. "Wage Contract Settlements in U.S. Manufacturing." Review of Economics and Statistics, Vol. 66, No. 4 (November), pp. 661-65. White, Bob. 1987. Hard Bargains: My Life on the Line. Toronto: McClelland & Stewart. White, Halbert. 1980. "A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. "Econometrica, Vol. 48, No. 4. (May), pp. 817-38.
|Printer friendly Cite/link Email Feedback|
|Author:||Budd, John W.|
|Date:||Apr 1, 1992|
|Previous Article:||Wage rule formation in the aerospace industry.|
|Next Article:||The economic progress of black women, 1940-1980: occupational distribution and relative wages.|