The Treasury

Global Navigation

Personal tools


Government and economic growth: Does size matter?

3  Empirical evidence

This section explores the empirical evidence around whether the level of government expenditure and revenue matters for economic growth. It covers three types of evidence: cross-country correlations between government size and economic performance, multi-variable or regression analysis of the impact on economic growth of government size and structure and the potential drivers of differences in productivity between the public and private sector.

3.1  Correlations between government size and economic performance

While it tells us nothing about causation, some studies have identified a negative correlation between levels of government expenditure or taxation and economic growth across countries (see, for example, Gwartney et al, 1998), as highlighted in Figure 2. Each diamond in Figure 2 represents one country in one of the four decades between 1969 to 2008, with New Zealand and Australia's most recent positions shown.[3] Furthermore, Tanzi and Schuknecht (1997) argue that countries with ‘small governments' have tended to do better on economic indicators and are not significantly different on social indicators.[4]

Figure 2 – Correlation between government size and economic growth
Figure 2 - Correlation between government size and economic growth.

As another example of correlation analysis, countries with smaller governments tend to have longer-lasting periods of higher growth. Figure 3 shows the percent of high-growth episodes in countries with total government spending of less than 40% of GDP over the last 50 years. The growth episodes are varied by the rate of average annual growth and by the number of years for which average growth was sustained. As the target growth rate or time period is increased, the share of growth episodes increases for countries with ‘smaller' governments. For example, over 70% of growth episodes where average annual growth was at least 4% for a 10 year period happened in countries with governments of less than 40% of GDP. This increases to around 90% if growth episodes are defined as an average of 5% annual average growth over the same period.

Figure 3 – Percent of high-growth episodes for governments less than 40% of GDP, 1961-2008
Figure 3 - Percent of high-growth episodes for governments less than 40% of GDP, 1961-2008.

However, we would advise against putting too much emphasis on these apparent correlations. It is risky to infer causation from correlation analysis; that is to assume that increases in government expenditure are undermining economic performance. As there are a number of other factors affecting economic growth, the correlation between economic performance and size of government may be a coincidence or reflect other factors that influence both variables. In particular, the tendency for both growth to slow and demand for government services to increase as countries grow richer (see section 2.3) may create a misleading correlation (Grimes, 2003). Given that the growth in public services in developed countries in the post-war period also coincided with a general decline in productivity growth, there is a significant risk that the negative correlation between government expenditure and economic growth rates is not causal (Nijkamp and Poot, 2004).


  • [3]The level of expenditure is measured at the start of the decade, which helps to reduce the risk that the correlation is capturing an increase in government size driven by economic growth, rather than an impact on economic growth from the size of government (see section 2.3). Note also that, due to issues of data availability, not all countries have a data point for each of the four decades with New Zealand, for example, having a data point for only the two most recent decades.
  • [4]Tanzi and Schuknecht (1997) define big governments as those with spending in excess of 50% of GDP, medium-sized governments with spending between 40 and 50 percent of GDP and small government with spending below 40 per cent of GDP.
Page top