Measuring the Effect of Structural Change on Long-Term Public Expenditure Growth: the United States, 1929 to 1981

Abstract

This article presents a methodology for analyzing public expenditure growth to test for structural change. It subjects a version of Peacock and Wiseman's displacement hypothesis to a test based on the idea that the hypothesis involves a structural break. The method is based on a conventional model of the determination of government expenditures as applied to the United States for the period 1929 to 1981. First, the authors fit the equation for 1929 to 1981 (the entire period), for 1929 to 1945 (the early period), and for 1946 to 1981 (the later period). Second, they employ a test of structural stability between the early and later periods. Because this test is signifi cant, the authors then re-estimate the model for the entire period using an intercept shift dummy and coefficient shift dummies, one for each independent variable, for the later period Because the result is ambiguous, the authors take the fourth step and use the early period regression model to first forecast later period values of the dependent variable and then regress the actual values on those forecasted. The results of this exercise point unambiguously to net downward displacement. The authors also use the later period regression to forecast (or "backcast") early period values and again regress actual values on the forecasted ones. This analysis provides no evidence of either upward or downward displacement. They conclude that if a structural break did occur at the time of World War II, it took a form inconsistent with the displacement hypothesis of Peacock and Wiseman.