Does big government constrain economic growth?

Different studies come up with different findings on the effects of size of government on economic growth. It seems to me, however, that the weight of evidence favours the view that big government is detrimental to economic growth.

In a recent paper Morris Altman suggests that “moderate amounts of ... big government are not ... bad for the economy” (see here). The paper examines the relationships between the various components of economic freedom indexes and economic growth. The results show that some components of these indexes (e.g. secure property rights) have a strong relationships with economic growth, but there is no obvious relationship between the size of government and economic growth.

These findings about the size of government are contrary to the findings of other studies, including the recent paper "Big, not better", by Keith Marsden. The findings are also contrary to research that I have undertaken myself (initially for the New Zealand Business Roundtable).

The difference in findings seems to be associated with the use of different data sets. The data set used by Altman includes both high and low income countries whereas the studies that have found a negative relationship between big government and economic growth (the ones that I am most familiar with anyway) use an OECD data set (which excludes low income countries). I am not sure how it is possible to explain why some low income countries manage to have high economic growth rates despite big government. Perhaps big government in these countries is more closely linked to productive investment (e.g. high levels of public spending on infrastructure) rather than to patronage or income transfers, as in high-income countries with big government and low growth.

My study used data for 21 OECD countries, initially covering the period 1960 to 1997 (How much government?, New Zealand Business Roundtable, 2001, Appendix 3, pp 83-87). I have just updated the results to cover the period to 2006, with very little change occurring in estimated coefficients. In the most basic form of the model the rate of economic growth in each period (approximately a decade) is the dependent variable and size of government (general government total outlays as a percentage of GDP) at the beginning of each period is the explanatory variable. (In the updated version there are 5 observations for each country i.e. 105 observations in total.) The use of initial size of government means that we can be fairly sure that the direction of causation runs from size of government to GDP growth, rather than vice versa.

Results for a simplified version of the model are shown in the chart below.

(The regression result shown in the chart is:
r = 5.6 – 0.076G Adj R squared = 0.27
(0.5) (0.012)
where r = rate of growth in GDP (% per annum) and G = total government outlays as a percentage of GDP. )

The chart shows clearly that big government is associated with lower growth rates in OECD countries. The results suggest, for example, that were New Zealand to contract the size of its government to be about the same as for Australia (a reduction in total government outlays from 41 percent of GDP to 34 percent) this would result in an increase in economic growth rate of about 0.5% per annum. This might not appear to be a large increase in growth rate, but for a middle-income country like New Zealand it could amount to the difference between catching up to average income levels of high-income countries and struggling to avoid falling further behind.