ECONOMIC GROWTH CENTERYALE UNIVERSITYP.O. Box 208269New Haven, Connecticut 06520-8269http://www.econ.yale.edu/~egcenter/CENTER DISCUSSION PAPER NO. 814SUPERVISION AND TRANSACTION COSTS:EVIDENCE FROM RICE FARMS IN BICOL, THE PHILIPPINESRobert E. EvensonYale UniversityAyal KimhiHebrew UniversitySanjaya DeSilvaYale UniversityApril 2000Note: Center Discussion Papers are preliminary materials circulated to stimulate discussions and criticalcomments.ABSTRACTLabor markets in all economies are subject to transaction costs associated with recruiting,monitoring and supervising workers. Rural labor markets in developing economies, where institutions suchas labor and contract law and formal employment assistance mechanisms are not in place, are regarded tobe particularly sensitive to transaction cost conditions. The inherent difficulty of measuring transactioncosts has limited studies on this topic.In this paper, we analyze supervision activities reported in a cross-section survey of rice farmersin the Bicol region of the Philippines. This survey is unique because it provides supervision data at thefarm task level. We present a simple optimization model in which supervision intensity increases theproductivity of hired workers, which is assumed to be lower than that of family members due to thetransaction costs. The model predicts that supervision intensity will increase with transaction costs. Weuse different institutional conditions to proxy for transaction costs, and estimate the demand for supervisiontime for four different classes of rice production tasks. The estimation strategy controls for selectivity inboth hiring and supervising. The results show a positive effect of transaction costs on supervisionintensity.We then extend the analysis to a farm efficiency specification to test the proposition thatsupervision activities improve farm efficiency. This framework allows us to relate institutional conditionsto farm efficiency directly and indirectly through their effect on supervision activities. We find thattransaction costs have a negative direct effect on farm efficiency, but this is partially offset by the positiveefficiency effects of increased supervision intensity. The results enable us to associate institutionalconditions with transaction costs and to draw policy inferences regarding the value of improvedinstitutional conditions.KEYWORDS: Transaction Costs, Supervision, Labor Markets, PhilippinesJEL CLASSIFICATIONS: 013, D23, J43, Q122I. IntroductionLabor markets in all economies are subject to transaction costs associated with recruiting,monitoring and supervising workers. Transaction costs in the labor market typically arise due toinformation problems of two types: 1) moral hazard because the true work effort is not easily verifiableand enforceable, and 2) adverse selection because information on the attributes of heterogeneous workersis not readily available. Recruiting costs can also increase if communication and transportation networksare weak so that the labor markets are segmented. Transaction costs will be lower in environmentswhere contract are easily enforced, information on workers and employers are readily available and labormarkets are well connected. The level of transaction costs affects labor and land contract choices andfamily labor advantages. Rural labor markets in developing economies, where institutions such as laborand contract law and formal employment assistance mechanisms are not in place, are regarded to beparticularly sensitive to transaction cost conditions. A number of studies of contract choice support thiscontention. The inherent difficulty of measuring transaction costs, however, has limited studies on thistopic.In this paper, we report an analysis of supervision activities based on a cross-section survey ofrice farmers in the Bicol region in the Philippines. This survey is unique because it provides supervisiondata at the farm task level in addition to information on production activities and household characteristicsover a range of institutional conditions. It also provides barangay (village) level variables that help us toquantify the institutional conditions. Our primary concern is to analyze the demand for supervision time onsurvey farms. We develop estimates of the effect of different institutional conditions on supervision timefor four different types of rice production tasks.We also extend the analysis to a farm efficiency specification to test the proposition thatsupervision activities improve farm efficiency. This framework allows us to relate institutional conditionsto farm efficiency directly and indirectly through the effect on supervision activities. This enables us to3associate institutional conditions with transaction costs and to draw policy inferences regarding the valueof improved institutional conditions.Only a few studies have formally studied the demand for supervision. Empirical studies areespecially rare because most farm level surveys have not explicitly measured supervision intensities.Several studies have related the demand for supervision to wages and the size of work groups. Efficiencywage models suggest that supervision may be substituted by wage premiums when monitoring is costly(Bulow and Summers 1986). This justifies the finding of Groshen and Krueger (1990) and Kruse (1992)that wages and supervision intensity are negatively correlated. However, if variations in shirking costsamong firms are more important than variations in monitoring costs, supervision and wages may bepositively correlated (Neal 1993). That is, if wages are high, the cost of shirking is higher for employers.Supervision also depends on the size of the work group (Ewing and Payne 1999), but the sign of the effectis ambiguous: scale economies in supervision make monitoring more cost-efficient in larger work groups;on the other hand, large work groups are more difficult to supervise. To our knowledge, the literature hasnot addressed the relationship between institutional conditions (or transaction costs) and supervision. Wehypothesize that the demand for supervision will be greater in high transaction cost environments.In part II of this paper, we develop the specification utilized in this paper. In part III, wesummarize the data. Part IV reports our supervision demand estimates. Part V reports our farmefficiency estimates. Part VI concludes with policy implications.II. A Simple Model of Supervision