TRADE ASSOCIATIONS: WHY NOT CARTELS?

• Author: Salvatore Modica,Andrea Mattozzi,David K. Levine
• Published date: 01 February 2021
• Date: 01 February 2021
• DOI: http://doi.org/10.1111/iere.12487

INTERNATIONALECONOMIC REVIEW

Vol. 62, No. 1, February 2021 DOI: 10.1111/iere.12487

TRADE ASSOCIATIONS: WHY NOT CARTELS?∗

By David K. Levine, Andrea Mattozzi, and Salvatore Modica1

EUI, Italy, WUSTL, USA; EUI, Italy; Università di Palermo, Italy

The relevance of special interests lobbying in modern democracies can hardly be questioned. But if large

trade associations can overcome the free riding problem and form effective lobbies, why do they not also

threaten market competition by forming equally effective cartels? We argue that the key to understanding the

difference lies in supply elasticity. The group discipline, which works in the case of lobbying, can be effective

in sustaining a cartel only if increasing output is suff‌iciently costly—otherwise the incentive to deviate is too

great. The theory helps organizing a number of stylized facts within a common framework.

1. introduction

The objective of this article is to contend with three stylized facts: (1) Large lobbying orga-

nizations exist and are effective; (2) large commercial cartels are less common and appear dif-

f‌icult to sustain; and (3) nevertheless, trade unions that both lobby and cartelize do form and

are effective. We reconcile this apparent puzzle with a simple common theory.

Our interest is in large organizations that face a free rider problem. In standard industrial

organization models of collusion through repetition, cooperation can be sustained using sim-

ple trigger strategies. Furthermore, as shown by Pecorino (1998) in a repeated tariff lobby-

ing game, the critical value of the discount factor above which cooperation can be sustained

does not necessarily rise with the number of f‌irms. This argument, however, hinges on assum-

ing perfect information: A threat to disband a cartel when its rules are not followed cannot

be effective with many f‌irms in the presence of small noise.2We provide an alternative model

in which organizations use peer punishments to overcome free riding. In both types of models

the possibility of successful collusion depends on the incentives to deviate. The greater this in-

centive is, the greater the punishments needed to induce compliance, and with imperfect mon-

itoring, greater punishments are more costly.

Our contention is that since the incentive to deviate in lobbying organizations is naturally

limited to the gain from failing to contribute, collusion is relatively easy. In a market setting,

however, large cartels are more diff‌icult to sustain because deviation is potentially far more

prof‌itable if production can be ramped up on a large scale to take advantage of a gap between

price and marginal cost. The ability to take advantage of this opportunity crucially depends on

the elasticity of supply. If marginal cost rises rapidly with output, then the gain from deviating

is limited and collusion can be sustained. This—we argue—is the case for trade unions but in

general not the case for manufacturing f‌irms or farms.

∗Manuscript received October 2019; revised July 2020.

1We would like to thank Giacomo Calzolari, Rohan Dutta, Daniel Garcia, and seminar participants at the Pam-

plona Jornadas de Economia Industrial, the Bocconi Theory Meeting in Capri, the SAET meeting in Faro, the Fudan

University Theory Workshop, the SHUFE Theory Workshop, the SJTU Theory Workshop, the LAMES meetings in

Buenos Aires, and the University of the Pacif‌ic in Lima. We gratefully acknowledge support from the EUI Research

Council and Italian MIUR PRIN 2017H5KPLL. Please address correspondence to: Andrea Mattozzi, Department of

Economics, European University Institute, VillaLa Fonte, Via delle Fontanelle 18, I-50014 San Domenico di Fiesole -

Italy. Phone: +39 055 4685914.E-mail: Andrea.Mattozzi@eui.eu.

2See Green (1982), Sabourian (1990), Levine and Pesendorfer (1995), Fudenberg et al. (1998), Al Najjar and

Smorodinsky (2001), and Pai et al. (2014).

47

© (2020) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of So-

cial and Economic Research Association

48 levine, mattozzi, and modica

We emphasize that in the case of small cartels, the situation is different. Small cartels are

similar to lobbying organizations in that there is a natural restriction on increasing output: In-

dividual demand is downward sloping and f‌irms are not so eager to produce large amounts.

Hence small cartels are easier to sustain. It should be clear that we do not reject the tradi-

tional “small cartel” theory in which collusive arrangements are enforced by threat of future

price retaliation. As we have indicated, it is for large cartels that these types of punishments

are not useful. Our key point is that small cartels—those traditionally studied in the indus-

trial organization literature—are different than large cartels. In particular we argue that sup-

ply elasticity plays a much less signif‌icant role in the case of small cartels than it does for

large cartels.

Going back to the stylized facts mentioned above: that large lobbying organizations exist

and are effective is the core of much of the political economy of farm subsidies and trade re-

strictions. Certainly we observe that special business interests such as farmers or the chamber

of commerce for example are effective at lobbying government for subsidies and for entry and

trade restrictions. These organizations, which we refer to as trade associations, are small as a

share of the economy but often quite large in absolute size. Because of their large absolute

size, they face a substantial free rider problem in raising resources for lobbying: This is well

documented by Olson (1965) and his successors. Still, they are able to overcome this free rider

problem to be effective at lobbying.3In the case of farming, for example, agriculture repre-

sents slightly more than 1% of U.S. GDP but there are more than 2 million farms, and they

command around 0.5% of GDP in subsidies. In Japan the GDP share is similar, there are over

3 million farms and subsidies exceed 1% of GDP.4Notice, however, that although f‌irms ben-

ef‌it from lobbying, they would also benef‌it from collusion in the form of an output-restricting

cartel, and the free riding problem appears similar: Produce more and reap extra prof‌its in the

cartel case, do not contribute to the lobbying effort in the lobbying case.

This raises the puzzle we want to address: If trade associations are so effective at overcom-

ing the free rider problem in order to lobby, why are they not equally effective at overcoming

the free rider problem of sustaining a cartel? But also, how come trade unions, which also are

output-restricting cartels, do typically exist? That large cartels are uncommon and diff‌icult to

form is well understood in the industrial organization literature where focus is on concentra-

tion ratios and industries with few f‌irms—but this does not explain either the ubiquity of large

lobbying organizations or the success of equally large trade unions.

In order to understand when trade associations are successful at lobbying and at carteliza-

tion, we need a theory of how they overcome free rider problems. We know from the work of

Ostrom (1990) and her successors how this can be achieved: Groups can self-organize to over-

come the free rider problem and provide public goods through peer monitoring and social

punishments such as ostracism. Formal theories of this type originate in the work of Kandori

(1992) on repeated games with many players and have been specialized to the study of orga-

nizations by Levine and Modica (2016) and Dutta et al. (2018). The basic idea is that groups

choose norms consisting of a target behavior for the group members and individual penalties

for failing to meet the target; these norms are endogenously chosen in order to advance group

interests. Specif‌ically the group designs a mechanism to promote group interests subject to in-

centive constraints for individual group members, and it provides incentives in the form of

punishments for group members who fail to adhere to the norm. Here, we build on this theory

to compare the public goods problem of lobbying to that of cartelization.

The theory presented in this article helps organizing a number of stylized facts. First, we

observe trade associations that lobby but do not cartelize, but rarely ones that cartelize but

3See, for example, Grossman and Helpman (2001).

4The share of agriculture in value added is from http://data.worldbank.org/indicator/NV.AGR.TOTL.ZS.

Total agricultural support as a percent of GDP is from http://stats.oecd.org/viewhtml.aspx?QueryId=

70971&vh=0000&vf=0&l&il=&lang=en. Data on the number of farms are from Lowder et al. (2016). In gen-

eral, farmers are very effective at overcoming the free rider problem to lobby for farm subsidies, see, for example,

Acemoglu and Robinson (2001).