A Brief Review of Statistical Models in Budgetary Studies

Abstract

Budgetary studies include more than quantitative data analysis. Actually quantitative analysis was at the periphery of budgetary research till a decade ago among political scientists, if not among economists. Therefore, a focus on the technical problems of data analysis in assessing the methodological quality of budgetary studies may be misleading. It is nevertheless important to summariz major deficiencies of data-analysis techniques used in the budgetary studies. And this is what this small paper tries to do. Among those problems, I picked up the problems in stacistical models which were most frequently used: the violation of basic assumptions in those models which were natively applied by researchers. Quantitative analysis of budgetary data can be traced back to German Cameralists. But statistical approach (that is inferential statistics) began in 1950's when Fabricant's famous study on the determinants of state expenditure first appeared. Another seminal study by H. Brazer in the same period triggered tremendous amount of cross-sectional analysis by economists. Political scientists who usually treated budgetary studies as one part of Executive-Congress relationship to focus on formal budgetary process began to imitate economists' approach from 1960's. But this trend of cross-sectional analysis was set back by Wildawsky's incrementalism which was operationalized. with some other's help, into autoregressive time series analysis. Crecine went beyond this by originating simulation model which digs out the internal process of budgeting and connecting this to the time series data.