Governança corporativa e desempenho : um estudo em cooperativas de crédito no Brasil

Abstract

This thesis aims to determine if the good corporate governance practices influence the performance of Credit Unions in Brazil, through the analysis of financial statements and a structured search. During the search, given the available data that limit ourselves, at first, to establish the determinants of performance of credit unions analyzed using as dependent variable the "leftovers" and the independent variables with greater degree of significance as equity the fixed assets, credit and leasing operations, operating revenues, revenues from services and administrative expenses. This segment has been characterized in recent years by growth and change in its profile and are among the least studied financial institutions. Moreover, it has undergone major regulatory changes by the Central Bank that aims to prevent the agency conflict and increase the participation of credit unions in the financial system. The credit union members are dispersed and have little benefit in closely monitor the manager's performance where managers have different interests of members, such as propensity to risk and expected return providing more information on the organization. The credit union has been an element driver of the economy, filling spaces not occupied by traditional financial institutions in response to the global concentration phenomenon promoting economic and social development of the communities where they operate ensuring the maintenance of jobs in the communities and the provision of goods and services tailored to local needs. To collect information on the size and the level of governance we used a questionnaire, whose questions aimed to verify the existence of structural items that make up a minimal basis for the existence of corporate governance. The inspection of the data indicated a dynamic panel data Arellano-Bond model estimated in two steps, using the independent variables as instruments, with a lag. We used a sample unbalanced panel of 2,480 observations from 2010 to 2014, a total of 742 cooperatives. Data from cooperatives were targeted for model robustness check considering the full sample, the co-operatives in PL less than R $ 1 million; between R $ 1 million and R $ 10 million; between R $ 10 million and R $ 100 million; above R $ 100 million and the balanced sample that referred only to cooperatives with information for each year. The work contributed to the literature by implementaring a specific methodology supported in dynamic panel for analysis and by identifying the factors determines the performance of cooperatives from the set of variables.