Our aim was to construct some simple tests of whether ‘activist’ policies (of government spending on R&D support or on investment subsidies) or ‘incentivist’ policies (reducing the barriers to business entry/exit and also taxes on business people’s incomes) are the cause of growth. We followed the standard practice of gathering data on growth rates of many countries (approximately 100) and many periods (three decades, the 70s, 80s, and 90s); and similarly gathering data on these candidate variables. We then asked whether growth was related to one or to the other, or to neither, using standard tests of statistical significance.
In this sort of study the data you gather is the key ingredient. We treated the gap between home real interest rates and world real interest rates as the measure of investment subsidy: the government can by a variety of interventions, especially exchange controls, keep down the cost of capital (basically because home savers have Hobson’s choice and foreign lenders’ money can be subsidised to keep it low). For R&D subsidies we had measures of how much governments spent on R&D. In some separate work we also looked at education spending and infrastructure, as other activist policies.
For the business tax rate we used two main elements: the rate of general tax (which we set equal to the share of government spending in GDP – this can be thought of as the underlying tax rate that must be levied, whether today or later when compounded by interest payments) and also the loading of costs onto business through restrictions on entry and exit. The first came from the Penn Tables, the second from the World Bank.
The resulting ‘business tax rate’ is an amalgamation of the two.
When you relate growth to this business tax variable you get a strong negative relationship. When you relate it to the activist measures, in no case do you find any relationship at all. These statements remain true however you ‘control’ for other possible factors driving growth.
So far, so good for those who think incentives are important for growth. This evidence is not the only sort available. There are many similar studies of growth and taxes, from the OECD and elsewhere – detailed in our chapter. We also have (mainly due to Parente and Prescott) studies of episodes such as ‘growth miracles’ (Hong Kong, China, India) and ‘growth disasters’ (the Congo, Sudan, Zimbabwe). They show pretty clearly that if you create barriers to business, you destroy growth and vice versa. This is a pure incentive story. You can strengthen it by reflecting that India grew pathetically under the Nehrus when all the rage was building infrastructure and expanding education (so that India had a record rate of unemployment among PhDs); when it was liberalised, growth broke out.
However, there is still work to be done to settle the matter. The problem is that these correlations can be found between growth and many different variables, if you are willing to work hard enough at gathering ‘helpful’ data. For example, take the case for education or ‘human capital’ as the cause of growth: human capital is hard to measure and with not too much effort one can find a measure that is correlated with growth. The same goes for such things as the level of infrastructure, or R&D. How then to distinguish between these causal theories? Here is an example of the difficulty: do education levels cause growth and then growth cause barriers to business to come down? Or do barriers coming down cause growth and growth then cause there to be more education spending?
In further work we hope to refine these tests. Our basic idea is that you can set out each theory of the economy with its different growth mechanism and explanation of the other ‘factors’. Then you simulate the behaviour of each type of economy and see which comes closest in its simulated behaviour to the behaviour in the data. We are at a fairly early stage in this work. So we must conclude in the time-honoured academic way: ‘more work is needed’!