SIMPLE THEORETICAL REAL BUSINESS CYCLE (RBC) MODELS focus exclusively on productivity disturbances as the source of economic fluctuations and growth. These models imply that the ratios of consumption and investment to output, hours worked per capita and, therefore, the leisure-labor ratio are constant along steady-state paths. Since it is not widely believed that the data is consistent with this implication, the RBC literature has developed to allow for such factors as permanent changes in demographic composition, the world real interest rate, the depreciation rate, the world price of oil and government policies.(1)
In this paper, we empirically examine how stochastic fiscal trends--defined as permanent innovations in various categories of government purchases relative to GNP--impinge on the above-mentioned ratios. More specifically, using appropriate statistical tests, we first verify that at least the consumption-output and leisure-labor ratios do not appear to display stationarity in the U.S. economy. Next, we estimate the long-term relationships linking these two ratios to our fiscal trends by estimating cointegrating vectors, and present some inference results on these vectors. Finally, we interpret the results in the light of a theoretical neoclassical growth model extended to allow for various categories of government purchases of goods and services with different degrees of substitutability for private consumption and, possibly, different wealth effects.
Our paper is related to two previous lines of inquiry. The first is the literature on stochastic trends exemplified by King, Plosser, Stock, and Watson (1991). We add to this literature by asking to what extent the common trends embedded in real economic aggregates can be interpreted as fiscal trends. Our motivation for focusing on ratios in the empirical work is that this allows us to abstract from the stochastic general productivity trend of the sort highlighted by benchmark RBC models--such as King, Plosser, and Rebelo (1988)--which predict that the ratio of consumption and investment to output or the leisure-labor ratio are constant along steady-state paths. Consequently, if there are unit roots in these ratios, these must be attributed to some other factors. By focusing on the relationship of these ratios to fiscal variables, we are able to isolate the effects of fiscal trends in generating long-term movements in real economic aggregates.
The second line of inquiry to which our work is related is the empirical work that attempts to test the equilibrium approach to fiscal policy by using estimation methods, summarized in Aschauer (1988) and Barro (1989, 1993). We add to this literature by focusing on components of government spending [as do also some recent papers by Aschauer (1989a, b)] but, more importantly, by focusing on long-term interactions and by estimating effects on the leisure-labor choice as well.
Our results suggest that dynamic general-equilibrium models that incorporate government consumption and government investment in a manner that allows for nonseparable effects on utility and production--such as the models presented by Christiano and Eichenbaum (1992) and Baxter and King (1993)--are more consistent with the long-run properties of the data than simple RBC models.
The balance of the paper is organized as follows. In section 1, we present the theoretical framework and set out the competitive equilibrium. Section 2 sets out our empirical methodology and presents the results. Section 3 concludes.
1. ANALYTICAL FRAMEWORK
Our theoretical model is an extension that incorporates the government sector in the basic RBC model developed in McCallum (1989). Government purchases of goods and services substitute/complement private consumption and investment, and are also allowed to affect total factor productivity directly. We distinguish between three components of government purchases: government …