Notes on the size of government
I’m working on a summary of the evidence on economic effects of the size of government. I’ll probably cite the usual suspects such as Barro and DeLong in due course, but I’d be interested in comments, particularly pointing to evidence outside the usual pattern of growth regressions and so on.
Size of government and economic growth
The emergence of the debate over the optimal size of government roughly coincided with the development of large data sets giving broadly comparable measures of national output for a large number of countries over time, and with the development of the computing power and econometric techniques required to undertake statistical analysis of such data sets. The standard data set in this context is derived from the International Comparisons Project undertaken by the OECD, and is currently managed by the World Bank (it was originally developed by researchers at Penn State University the University of Pennsylvania, and is commonly referred to as the Penn World Tables). Another source of evidence, suitable for statistical analysis, is derived from data sets in which the 50 US states are compared.
The most common form of statistical analysis is the â€˜growth regressionâ€™ in which estimates of the rate of economic growth are modelled as functions of a range of explanatory variables, commonly including measures of government expenditure. Vast numbers of these regressions have been estimated and relatively few consistent results have been obtained. Correlations that appear significant in one formulation of the model frequently disappear when the model is respecified or the data set is varied.
Bearing this qualification in mind, the general consensus of the literature is as follows. There is no well-established relationship between aggregate levels of government expenditure (expressed as a sure and subsequent economic growth. However, there is a strong causal relationship running in the opposite direction. Countries with high income levels resulting from past economic growth tend to have higher levels of government expenditure, reflecting greater demand for the services (health, education and income security among others) commonly provided by governments.
Although attempts to relate aggregate government expenditure to economic growth have been unsuccessful, there has been more success in the examination of particular components of government expenditure. The general pattern is unsurprising. Government investment expenditure is found to increase subsequent economic growth, while consumption expenditure is found to reduce subsequent growth, exactly as would be expected for investment and consumption in general.
It appears, however, that the benefits of government investment in both physical infrastructure and human capital (mainly education, but also childrens services and some components of health care) are greater than for private investment. Conversely, the negative growth effects of government consumption appear to be greater than those of private consumption, perhaps because government consumption tends to â€˜crowd outâ€™ government investment.
It should be noted that the fact that investment leads to future growth, does not mean that investment should always be favoured over consumption. Consumption is, after all, the ultimate purpose of economic activity, and investment is simply a means of generating higher consumption in the future. Thus, for example, while committing public or private resources to improving the health of elderly people is unlikely to yield any contribution to future economic growth, but that does not mean that expenditure in this area should necessarily be reduced.