Printer Friendly

Industry income and congressional regulatory legislation: interest groups vs. median voter.


Interest group theories of regulation suggest that industries will be able to gain

political benefits at the expense of the larger, but disorganized and disinterested, general

public. However, casual observation indicates that industries are often the targets of

costly legislation. We examine how the political influence and vulnerability of

industries are affected by industry income. Results show wealthy industries are more likely

to be subject to costly legislation, whereas no relationship was found between industry

income and the granting of political benefits. We interpret these results as supportive

of both interest group and median voter models of the political process.


The work of many political economists, including Olson [1965], Stigler [1971], Peltzman [1976], and Becker [1983], suggests industries can be powerful players in the political process since small, well-organized special interests, such as industries, do well in political competitions over income redistribution. One reason for their success is their comparative advantage in controlling free riding among their members that outweighs the relatively small numbers of votes that these groups might deliver in an election. This outcome is a specific prediction of the Stigler-Peltzman interest group model which differs, however, from those of the median voter models in which small special interests have little political influence relative to more general, mass interests. Director's Law, for instance, maintains that politically engineered income redistributions are from the tails of the income distribution to the middle.

We provide evidence which complements the Stigler-Peltzman theories with results more akin to Director's Law. In the Stigler-Peltzman approach, wealth-enhancing political benefits are often conferred on well-organized special interests which gain at the expense of a large, ill-informed and disinterested general public. The lobbying of industry associations on behalf of specific industries supports this view. However, as evidenced by industry lobbying against certain regulatory actions, legislatures sometimes pass regulations aimed at generating an income transfer from industry to the general public. Director's Law suggests that wealthy industries will be the most likely targets for such adverse legislation. This is because wealthy interest groups, even though organized, may present a target of sufficient size to generate an attempt by the general public to claim a portion of that wealth.

In this study we examine the political influence and vulnerability of industries on the basis of their comparative wealth. We examine the lobbying activity before Congress of a diverse sample of industries and classify the position of industry lobbyists as either advocates or opponents of specific legislation. The lobbyists' positions allow us to identify whether an industry is benefited or harmed by a piece of legislation. Our results show that, consistent with Director's Law, wealthy industries are more likely to be the targets of costly legislation, but industry wealth is not an important determinant of legislation benefiting industries.(1)


The basic assumption underlying the Stigler-Peltzman model is that elected representatives make political decisions which can transfer wealth from one group to another in the economy. Since these income transfers affect how their constituents will vote, legislators will support such legislation if they expect it will increase the number of votes they receive in the next election. In general, a legislator will support a set of transfers that maximizes the probability of re-election, as determined by the size and effectiveness of the groups affected and the amount of the transfers.

The transfer between the targeted and the benefiting groups can be pecuniary, such as a tax break or subsidy, but often will be a non-pecuniary regulatory action affecting entry barriers to an occupation or industry or imposing environmental, worker safety, product safety, or other standards. Whether pecuniary or not, however, a transfer imposes costs upon the targeted group that are wealth reducing.

The legislator whose self interest lies in retaining office is primarily concerned with the relative size and political effectiveness of the benefited and targeted groups when deciding on the optimal set of political transfers to support. Group size affects the number of votes the politician can potentially gain or lose by supporting a particular transfer. However, size is tempered by the ability of the affected groups to deliver the votes of their members--in other words, the ability of the interest group to control free riding by its members on the political activities of the group. The ability to control free riding depends on the effectiveness of the interest group's political organization. The size and political effectiveness of interest groups enter into the legislator's decision symmetrically. The fundamental insight of the Stigler-Peltzman approach is that a large, poorly organized group, such as the general public, may be unable to deliver enough votes to protect itself politically from a transfer to a small, but well-organized, interest group such as an industry.

Since industries are not always the beneficiaries of politically induced wealth transfers, we conclude that politicians rationally impose costly regulation on industries whenever it is politically advantageous to do so. However, this will occur only if the transfer is large enough for the public to overcome its free-riding problem and deliver sufficient votes to their legislator. This requirement--that politicians impose large costs on the targeted industry--introduces an important constraint on the process which is not considered in the Stigler-Peltzman model. In particular, legislated transfers are constrained by the wealth of the industry from which the costly transfer arises because the affected industry must be able to "afford" the costs imposed on it. Otherwise, the politician may end up losing the support of the affected industry while not gaining votes from a disappointed, disinterested public.

However, wealth enters into the political decision in an asymmetric fashion. The legislative transfer is constrained by the wealth of the targeted industry, but the wealth of a benefiting group is important only to the extent that it interacts with the size and organization variables. The introduction of this wealth constraint yields a corollary to the Stigler-Peltzman result: Wealthier groups are more likely targets for adverse political action than less-wealthy groups of the same size and political effectiveness. Thus, applied to industries, this corollary can be formally stated as:

HYPOTHESIS 1: Industries targeted for adverse political action are wealthier than industries not targeted for adverse political action.(2)

To test this hypothesis some further specifications are made. First, we examine legislative, and not regulatory, decisions. Because only Congress has the ability to enact wealth transfers which can affect virtually any industry, we can test the hypothesis over a wider cross-section of industries by focusing on legislation. In addition, the behavior of legislators is more consistent with the vote-maximizing assumption of the Stigler-Peltzman model. Regulators may be motivated by different interests, such as promotion or future employment. Second, our tests assume that Congress uses income, specifically accounting net income, as a proxy for wealth. While there are well-known problems with accounting measures of profits, income data is widely publicized and available to Congress through tax return information and corporate annual reports. Thus, we think it is a reasonable data source for the purposes of our study. More importantly, because members of Congress have limited election horizons, we assume that income, the flow return on wealth, is more immediately useful as an indicator of an industry's ability to pay for legislation than wealth itself.(3)

In our corollary, as in the Stigler-Peltzman model itself, the wealth of industry beneficiaries of political transfers does not constrain the behavior of legislators. Any industry can lobby for wealth-enhancing favors from the political system. Thus, assuming no interaction between industry wealth and size or ability to control free riding, wealthy groups should be as likely to benefit from legislation as groups with less wealth. This allows us to present and test an additional hypothesis relating to the wealth of political beneficiaries:

HYPOTHESIS 2: Industries receiving political benefits are no more wealthy than industries receiving no political benefits.

Overall, these hypotheses imply a pattern of industry-related political actions which differs somewhat from that predicted by the Stigler-Peltzman model. In the Stigler-Peltzman model, ceteris paribus, wealth is unimportant to the politician deciding to confer benefits or impose costs. To the contrary, the present analysis predicts wealth is important in choosing targets for costly legislation but is less important in choosing beneficiaries. This pattern of outcomes is more consistent with Director's Law.



The hypotheses are tested using a sample of forty-seven four-digit SIC code industries which were chosen randomly from the COMPUSTAT industry aggregate file.(4) The representativeness of the sample was determined using a chi-squared goodness-of-fit test. Based on two-digit SIC codes, the sample distribution and the breakdown of all industries on the COMPUSTAT industry aggregate file are not statistically different.

Each of the industries included in the sample was then matched with an industry trade association. Congressional testimonies by these trade associations were examined to determine whether industries were targets or beneficiaries of Congressional regulatory action. The trade associations were identified using the four-digit SIC codes and the Encyclopedia of Associations [1988]. For those instances where more than one trade group could be identified with an industry, the largest trade group was selected. Of the industries represented in the sample, only one industry (wholesale--metals and minerals, SIC 5050) did not have an industry trade group.(5)

The legislative and financial data are taken from two years, 1975 and 1981, and the data were combined and used in one test. Because our hypotheses make predictions about political decision making in general, one period may not be representative of legislators' decisions over time. The combined period is preferred because it dampens the effects of unusual decisions which are not likely to reoccur. The combined period is also preferred because using two time periods yields a larger sample of industry-related legislation, which increases the statistical power of the tests.

To increase the representativeness of the test period, we combine two periods which were different in terms of legislative composition and actual decisions. For example, the proportion of anti-industry legislation was higher in 1975 when the Senate was controlled by the Democratic party, whereas the proportion of legislation favorable to industry was higher in 1981 when the Senate was controlled by the Republican party.


The study uses testimony from Congressional hearings to classify industries in two ways. To test Hypothesis 1, industries were classified by whether they were targeted by unfavorable legislation. To test Hypothesis 2, industries were classified by whether they benefited from favorable legislation or received other political favors.

The industries were first classified as being "politically affected" or "unaffected" by costly Congressional legislation. Those industries which had trade groups that testified against a legislative bill which was pending at the end of either 1975 or 1981 and which was later passed were classified as being politically affected. The trade group's written and/or oral testimony was used to establish the trade group's position on a bill.

Typically, those testifying were straight-forward regarding their industries' position. For example, one lobbyist told Congress,

For reasons that I will mention here

before the committee today, the

American Mining Congress respectfully

would like to register its objection to

the enactment of land use legislation

and particularly HR 3510. (Howard L.

Edwards, representing the American

Mining Congress, May 2, 1975.)

In a few cases (less than 5 percent of those considered), the industry's position was not made clear. Most of these cases involved an industry which supported a bill's intent, but asked for changes such as rewording, amendments or deletions. Based on the amount of the trade group's testimony spent discussing the change, we judged whether the change was "major" or "minor." Major changes were those where this discussion exceeded 50 percent of that specific trade group's testimony, when measured in pages. The industries requesting major changes were also classified as politically affected, while industries not meeting any of these criteria were classified as unaffected industries.(6)

Two additional screens were then applied. First, because the testimonies represent views on intermediate bills, the final legislation could contain changes satisfying the concerns of the industry. Thus, the provisions of the passed legislation were also examined in the Congressional Quarterly Almanac [1976; 1977; 1982] and/or the published public law. For an industry to remain classified as politically affected, the areas of objection had to be included in the passed bill. This assures that the final legislation was indeed costly to the affected industry.

Second, because adverse industry-related legislation could also hurt the public (e.g., a tax on the industry which is passed on to consumers through higher prices), the testimonies were re-examined to determine whether any broad-based consumer, environmental, or public interest group had testified in favor of the legislation. These latter groups would be expected to testify in favor of anti-industry bills which would be beneficial to the public. This step assures that the legislation did cause a transfer of wealth from the industry to the public.

Likewise, industries were classified as "politically benefited" or "non-benefited" by favorable legislation using a similar classification procedure. The benefited industries were those that testified in favor of, rather than against, proposed legislation that was pending at the end of 1975 or 1981 and was later passed. The remaining industries were classified as non-benefited.(7)


Based on data from COMPUSTAT's industry aggregate file, the forty-seven industries were ranked by industry net income for 1974 and 1980 in 1980 constant dollars.(8) The income figures for 1974 and 1980 were the most recent available in the two years of interest, 1975 and 1981. The mean industry income was $2.196 billion with a standard deviation of $7.087 billion. The largest industry income, $48.967 billion, was more than 1700 times greater than the lowest industry income, $6.5 million, which indicates the sample is diverse with respect to this particular attribute. Table I lists industries by income and classifications.

Our first hypothesis predicts that industries with high incomes are more likely to be affected by unfavorable legislation than industries with lower incomes. The second hypothesis predicts that whether an industry receives political benefits or not is not related to the level of industry net income. The Stigler-Peltzman model also suggests that two key factors affecting a group's political success are group size and organizing costs. Therefore, we test the hypotheses using a multivariate procedure which controls for both of these factors.(9)

To control for group size, we use the number of individuals employed in the industry.(10) Employees who see their wages or benefits as being tied to industry income may have an incentive to oppose (support) legislators who vote for wealth transfers from (to) the industry.(11) Ceteris paribus, the Stigler-Peltzman model predicts that large pro-industry voter groups should be able to deliver more votes. This implies that larger groups should be less vulnerable to costly legislation and more likely to receive political benefits. The number of industry employees is obtained from data issued by the U.S. Department of Commerce's Bureau of the Census.

Likewise, it is important to account for differences in organizing costs across industries. Concentrated industries should be able to organize more easily, as the Stigler-Peltzman model suggests. This is also consistent with Olson [1965], who shows that as groups become diffused, organizing costs increase because incentives to free ride and communication and coordination costs increase. Being better organized, concentrated industries should be able to avoid political costs and gain political benefits more easily than less-concentrated industries. Consequently, the four-firm industry concentration ratio, also available from the Bureau of the Census, is used as a second control variable.(12)

The hypotheses are tested by examining the estimated coefficients for the subsequent model:

Y = [b.sub.0] + [b.sub.1] INCOME + [b.sub.2] EMPLOY + [b.sub.3] CONC where

INCOME = combined industry income
 for 1974 and 1980 in 1980
 constant dollars (millions);

EMPLOY = average number of
 employees for combined
 CONC = average four-firm
 concentration ratio for
 combined period.

The model uses a dichotomous dependent variable, Y. To test Hypothesis 1, Y is coded "1" for politically affected industries and "0" for politically unaffected industries. Thus, the coefficient for INCOME should be positive since high-income industries are expected to be better targets for adverse political action. The coefficient for EMPLOY should be negative because larger pro-industry groups should make adverse industry action less likely, and the coefficient for CONC should be negative because high-concentration industries can organize more easily and should be better able to avoid costly legislation.

To test Hypothesis 2, Y is coded "1" for politically benefited industries and "0" for non-benefited industries. The coefficient for INCOME in this model is expected to be insignificant because there should be no relationship between industry income and political benefits. The coefficients for EMPLOY and CONC should both be positive because larger pro-industry groups and industries with low organizing costs should be able to gain political benefits more easily.

Since the dependent variable is dichotomous, rather than continuous, the two models are estimated using probit regression. The probit model is based on the assumption that the dependent variable is an estimate of the probability that an observation belongs in one of two groups. It assumes that the probability function has a cumulative normal distribution and that the estimated coefficients will be asymptotically unbiased, efficient, and normal.(13)

Table II shows the pairwise correlations between the three independent variables. In particular, the correlation between income and employees is of interest because it shows that these variables are not significantly related as might be expected.

Table III contains the outcomes for the two probit regressions. The first two columns, which report the results for Hypothesis 1, show INCOME is correctly signed and is significant at the 0.05 level based on a one-tailed test (t = 1.873). This outcome, which supports Hypothesis 1, is consistent with the view that industries with high incomes are more likely to be targeted by adverse legislation than industries with lower incomes. The second two columns, which report the results for Hypothesis 2, show INCOME is not significant based on a two- tailed test (t = 0.679, p > 0.10).(14) Thus, as predicted by Hypothesis 2, industries with high incomes and industries with low incomes are equally likely to receive political benefits.

The only control variable which is significant is EMPLOY in the model testing Hypothesis 2 (t = 1.484, p < 0.10). Two of the remaining three coefficients on these variables are correctly signed but are not significant. These variables might be insignificant because the differences in group size and organizing costs, while large between industries and the public, as the Stigler-Peltzman model suggests, are too small on an inter-industry basis to explain variations in political effectiveness between different industries. For example, compared to the general public, industries can organize more easily, but between industries the ability to organize may be about the same. This is likely since forty-six of the forty-seven industries in our sample were organized.(15)


The traditional public choice theories, such as Stigler [1971] and Peltzman [1976], conclude that small, well-organized groups, such as industries, are likely to receive political benefits. Because industries often lobby against certain acts of Congress, these traditional theories are incomplete. We argue that politicians seeking to impose costs on an industry will be constrained by the industry's wealth, but the wealth of the beneficiary group does not affect the politician's decision. Our paper provides evidence which supports these two predictions. We interpret these results as being consistent with traditional public choice theories when modified to include the income distribution message of Director's Law.

While our research sheds light on a relatively unexplored area, as with any empirical study, some caveats are in order. From a practical standpoint, the results could be an artifact of the particular industries or years used in the study. Certainly, a larger sample of industries or an extended investigation period would increase the power of the tests. A useful extension would be to replicate the study in an earlier time period. For example, in the 1930s and 1940s, when public interest groups were less common, we would expect an even stronger relationship between income and costly legislation. Lacking public interest groups, which help to organize the public, the need for larger wealth transfers would become even more acute if the public is to be aroused to support anti-industry action.

From a theoretical standpoint, we have focused on the flow of political costs and benefits rather than the stock position at any point in time. While it is not clear whether the stock position has a positive, negative, or no effect on the current flow, it is likely that past political decisions will bear directly on present political decisions. For example, if an industry has incurred considerable costs in the previous periods, legislators may refrain from imposing new costs, regardless of the industry's current income. Thus, future research should extend the Stigler-Peltzman model to a multi-period setting and develop quantitative measures for industries' absolute political costs and benefits.


Pearson Correlations between Independent Variables
INCOME 1.000
EMPLOY -0.083 1.000
CONC -0.273 0.047 1.000

(a)Independent variables:

INCOME = combined industry income for 1974 and 1980 in 1980 constant dollars (millions);

EMPLOY = average number of employees for combined period;

CONC = average four-firm concentration ratio for combined period. [Tabular Data I to III Omitted]

(1)Industries have sufficient wealth to not be representative of the bottom tail of the income distribution. Thus, we focus on just a portion of Director's Law, i.e. transfers from wealthy groups. (2)This hypothesis is based on legislative decision making at the macro level and, therefore, differs from typical voting studies. Most voting studies look at each legislator's vote on a specific issue, whereas we examine the collective vote on many issues. One important advantage of this approach is that by examining a large sample of legislative decisions, the study's external validity is increased producing more generalizable results. (3)Because politicians periodically seek re-election, the politician needs to make good on promised wealth transfers prior to the next election. This requires that the industry on which costs are imposed can transfer wealth to the public immediately. Consequently, an industry which has good long-term earning power but has little current income would not be a good target for costly legislation. (4)The sample size of forty-seven industries resulted from a compromise between maximizing statistical power and minimizing data collection costs. Regarding the latter, we classified the industries as target or beneficiaries by reading Congressional hearing testimonies; to classify the industries used, over 1000 pages of testimony were read. The forty-seven industries represent over 30 percent of all industries on the COMPUSTAT file. (5)Theoretically, industries facing high organizing costs, such as geographically fragmented industries, or industries facing low benefits from organizing, such as those that do not face adverse regulation, may not choose to organize. (6)While the cutoff level chosen is arbitrary, this 50 percent level is also reasonable. We had a subsample of these Congressional testimonies read and a classified by a second independent rater. Interrater correlation was 95.3 percent. (7)The reader can obtain a complete list of the industry trade groups and their testimonies from the authors. (8)Constant dollar amounts were computed using the U.S. Department of Labor's Producer Price Indexes except for retail and service industries where Consumer Price Indexes were used. (9)The hypotheses were also examined using a Mann-Whitney test, a non-parametric procedure which compared the mean ranks of industry income for two groups. This test found the mean rank for the politically affected group was significantly higher than the mean rank for the unaffected group at the 0.001 level. The Mann-Whitney statistic for the difference between income ranks of the benefited and non-benefited groups was not significant at the 0.10 level. These tests were repeated for a reduced sample of industries, first where non-testifying industries were excluded (since these industries may have voiced their political views in other ways), then where industries which were classified as both affected and benefited were excluded (since we could not evaluate whether these industries were affected or benefited on a net basis). These alternate tests yielded the same results. That is, as expected, the affected industries had significantly higher income than the unaffected industries, but there was no statistical difference between the incomes of the benefited and non-benefited groups. (10)Shareholders could also be included in the pro-industry group. However, because many shareholders, and certainly the largest shareholders (e.g., financial institutions), maintain diversified holdings within the economy, the effect of any legislation on their portfolio may be zero. In such a case, the shareholder would be indifferent to the legislator's decision. (11)This is especially true for employees with industry-specific job-skills. (12)The tests were also run using the eight-firm concentration ratio with no change in the results. (13)See Aldrich and Nelson [1984] for additional background on probit models. (14)The two-tailed test for Hypothesis 2 was used since we had no prior prediction about the direction of this relationship. (15)The probit model was also run first omitting the non-testifying industries and then omitting the benefited and affected industries for reasons described in note 9. The results were similar.


Aldrich, John H. and Forrest D. Nelson. Linear Probability, Logit, and Probit. Beverly Hills: Sage Publications, 1984. Becker, Gary S. "A Theory of Competition Among Pressure Groups for Political Influence." The Quarterly Journal of Economics, August 1983, 371-400. Congressional Quarterly, Inc. Congressional Quarterly Almanac, Washington, D.C., 1976, 1977. Gale Research Co. Encyclopedia of Associations, 22nd ed. Detroit, 1988. Olson, Mancur. The Logic of Collective Action. Cambridge: Harvard University Press, 1965. Peltzman, Sam. "Toward a More General Theory of Regulation." Journal of Law and Economics, August 1976, 109-48. Stigler, George J. "The Theory of Economic Regulation." Bell Journal of Economics and Management Science, Spring 1971, 3-21.

STEVEN F. CAHAN and WILLIAM H. KAEMPFER, Assistant Professor, University of Wyoming, and Associate Professor, University of Colorado, Boulder. We wish to acknowledge the helpful comments from Tony Lowenberg, Mike McKee, Frank Selto, and Tom Willett.
COPYRIGHT 1992 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Cahan, Steven F.; Kaempfer, William H.
Publication:Economic Inquiry
Date:Jan 1, 1992
Previous Article:Stabilization policy can lead to chaos.
Next Article:"What does she see in him?:" the effect of sharing on the choice of spouse.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters