From at least 1893 economists have viewed income as an important determinant of government size and the hypothesis that government size increases with income is now enshrined in the literature as Wagner's Law. More recently, however, public choice economists and growth theorists have tended to reverse that causality by questioning whether government size is a constraint on (or promulgator of) economic growth. Typically, increases in government size arising from increased consumption are viewed as constraints on growth, while increases in size that arise from government investment are viewed as positive in their effect on growth. In this paper we are concerned with the two-way interrelationship between government size and income growth highlighted by these separate literatures and investigate this relationship in three distinct stages. In the first part of the paper we set out what has actually happened to the real size of government for twenty OECD countries over the period since 1970 and survey some of the newer factors and approaches used to explain its more recent evolution. The second part re-estimates the parameters of the demand curve for government allows us to speculate whether the changing pattern of government growth represents a break in the structure of the model determining government size or, more simply, represents a change in the variation of the underlying variables. We find that the same model works at least as well as it did in earlier periods with coefficients that are close to their earlier estimates. We follow this by estimating a simple growth model that highlights the size of government consumption in relation to income and output growth for the same countries over the same time period. Increases in size do appear to constrain economic growth. The third part of our paper recognizes that while each of the two causal relationships has received considerable attention in their own right, less attention has been given to effecting a separation of their co-mingled effects. To do so, we estimate the two relationships simultaneously in the context of our panel. This allows us assess whether ignoring the simultaneity of the two-way relationship seriously biases the measure of either the income effect (in determining government size) and/or the measure of government's effect on economic growth when each are estimated separately. While our discussion suggests that single equation estimates of the income elasticity in Wagner's Law may have been biased upwards (in absolute terms) and the constraining effect of government size on growth biased downwards, our three stage estimates finds only modest support in the data. The paper concludes by exploring the interrelationship between government size and government regulation. In particular, we test the hypothesis that the appearance of slower growth in government side is due to the increased substitution of indirect control of private production for direct governmental output. On cross sectional data, we find the opposite. In our sample, larger government size is associated with more rather than less regulation.
