Análise de Cluster da variabilidade orçamentária na administração pública federal

The public budget is a tool that quantifies the political decisions, reflecting the agreements and interests of various stakeholders. The law provides that the budget is a planning tool, and as such implies predictability of the implementation of the agreements indicated in its adoption. To verify the relevance of the public budget as a planning tool, the research examines the percentage of budgetary variability throughout the budgetary process and from various classifications of public spending. The methodology used was the cluster analysis, considering the types single, complete, average and Ward. The context of this evaluation included the budgetary models, the planning and the theory of Principal-Agent, democratic governance, highlighting the coalition presidentialism and budgetary election cycles or Political Budget Cycle. As an unexpected result, budgetary variability revealed a marked swing of the expenses agendas at the time of payment of the amounts paid with authorizations from others exercises ("restos a pagar") compared with the project of budgetary law, law, authorized, executed and paid with budgetary authorizations of the exercise. Besides that, there was no significant variation between years over the period covered (2002-2012). As expected, individual parliamentary amendments presented the largest variability and entities of indirect administration are more susceptible to variations. The results indicate the existence of hegemonic patterns in the budget. The presence of the assumptions of coalition presidentialism can be infered from the annual fluctuations in individual parliamentary amendments. The Principal-Agent theory is useful for understanding the variability, with a dominance of the executive branch in the budgetary process. There was no evidence of Political Budget Cycle in Brazil. The model that best describes the Brazilian budget is incrementalism, with space for specific explanations through the serial judgement and the punctuated equilibrium. The most amazing finding was the outstanding change of agenda in the payment of amounts to be paid ("restos a pagar"), suggesting that some of the agreements signed during legislative consideration may change in the final moment of his execution. Thus, it is possible to say that this variability of payments compromise the budget as an effective planning tool in the Brazilian federal budget.