Do Incentives for Quality Matter?

Journal of Agricultural and Applied Economics, Apr 2007 by Alexander, Corinne, Goodhue, Rachael E, Rausser, Gordon C

We use an unusual dataset involving 14 tomato growers over 4 years to analyze the effect of incentive contracts on behavior in a fixed effects econometric model. We find that growers respond to incentive contracts by improving tomato quality, as predicted by economic theory. The comparison is not confounded by the usual contract endogeneity and simultaneity problems because of characteristics of the processing tomato industry and our dataset. We discuss the implications of our findings for the design of agricultural contracts.

Key Words: contracts, panel data, processing tomato industry

JEL Classifications: D86, C23, Q13

People respond to financial incentives in contracts-or so economists believe. Although such a response is obvious theoretically, it is difficult to obtain conclusive support empirically. Contract provisions and agent performance under contracts are often proprietary information, so obtaining data is difficult. Furthermore, observed data are often subject to simultaneity problems between contract determination and performance under the contract (i.e. because agents' contract choices are endogenous relative to incentives in the contracts, any data solely on performance under a contract are incomplete). In most cases, agents select only one contract so that it is impossible from selected contracts to eliminate the possibility that a hidden factor influences both contract choice and ensuing performance.

To analyze the effect of incentive contracts on behavior, we compare the quality of tomatoes delivered under an incentive contract that specifies premium schedules for multiple tomato attributes (denoted contract tomatoes) to the quality of tomatoes delivered under a fixed price per ton contract (denoted no-contract tomatoes). We also compare the quality of contract tomatoes delivered during the regular season to the quality of contract tomatoes delivered late in the season when an extra premium is paid. We use an unusual dataset of 14 tomato growers who deliver processing tomatoes under both the incentive contract and the fixed price per ton contract, over a 4-year period. The sample is multidimensional: processors value a number of tomato attributes, some of which are less costly than others for growers to deliver. We examine two direct measures of quality and two financial measures of quality to assess growers' responses to price incentives, rather than using a single measure. All quality attributes are graded by an independent third party, the state of California, so that neither party to a contract can deliberately misstate or mismeasure quality.

With a fixed effects model, we test whether the quality of the no-contract tomatoes is lower than the quality of the contract tomatoes and whether the quality of contract tomatoes delivered in the late season, when there is a base price premium, is lower than the quality of the contract tomatoes during the regular season. Our comparison supports the fundamental economic hypothesis that people respond to financial incentives in contracts; our analysis shows that growers do respond to incentive contracts by improving tomato quality.

Although a well-developed theoretical literature predicts how performance measures such as quality and output respond to contract incentives, the empirical literature is much less developed (Prendergast; Salanié). Our analysis contributes to the empirical literature regarding agents' responses to contract incentives.1 Existing empirical studies primarily address nonagricultural examples of incentive contracts, such as managerial compensation (e.g., Lemmon, Schallheim, and Zender; Murphy), worker compensation (Lazear), and franchising (e.g., LaFontaine; LaFontaine and Shaw). In part, this emphasis is due to the difficulties of obtaining data on contract terms and outcomes. In the case of managerial compensation, the reporting requirements for publicly held companies provide a data source. Similarly, franchising studies compile information from a number of published sources.

Broadly speaking, the empirical literature on agricultural contracts is concerned with two questions: what are the determinants of contract choice and how does contract design affect agents' responses and the principal's profit or utility. Examples of the empirical literature regarding the determinants of contract choice include Alien and Lueck and Goodhue et al. Our study addresses the second question. It is distinguished from other empirical studies that address this question in that the structure of the processing tomato industry, combined with the nature of our dataset, insulates our data from three common incentive endogeneity problems: the continuous evolution of contract terms, the simultaneity of contract choice and contract terms, and endogenous matching between heterogeneous principals and agents. With the exception of Ackerberg and Botticini, the few existing studies that address the effect of financial incentives on real behavior under agricultural contracts use originally proprietary data on the basis only of outcomes under a contract. They do not observe outcomes in the absence of the contract, so their conclusions are subject to concerns regarding sample selection (Goodhue, Rausser, and Simon; Knoeber and Thurman; Leegomonchai and Vukina). Hueth and Ligon, and Wu, who like us use data from the processing tomato industry, suffer from another selection problem; their data reports contract outcomes for multiple processors (principals), each contracting with multiple growers. Given their analytical techniques, the endogeneity of the processor-grower pairings might confound any analysis of responsiveness to contractual incentives. Ackerberg and Botticini introduce a means of controlling for this endogenous matching problem econometrically. Their analysis does not control for the evolution of contract terms over time; it relies on cross-sectional data.