Government Spending and Growth Rates: a Methodological Debate

The following is a sequence of letters in “The Independent” in early 2002. I have not got the letters being criticised, but the readers should be able to infer much of what was in them. I have tried to bring my text in line with “The Independent’s” (sensitive) subbing.
Keywords: Social Policy; Growth & Innovation;

Govt spending and growth rates: A matter of values

The Independent January 23, 2002.

Roger Kerr uses some extraordinary logic in his letter on growth and government spending. ( 19 December) John Stuart Mill for one, would be bemused.

But let us follow the ‘no country has sustained per-capita income growth of 4 percent [per annum – sic] or more with a government sector share of the economy as high a New Zealand’s level of around 40 percent.’

As it happens in the 1990s some OECD economies have had a ‘government sector share’ of below that 40 percent and have not sustained that growth rate either: United States, Japan, Australia, and Iceland. Two have had low government spending and high per capita GDP growth: Ireland and South Korea. However those two economies have had vigorous industries assistance policies. On the Kerr logic, I take it he joins with Jim Anderton and Peter Hodgson in advocating a more active industries assistance regime.

Of course one can torture the data until it confesses. I leave that task to the sadists, and report that un-tortured data shows no meaningful correlation between the level of government spending and the economic growth rate. This seems to suggest that either the theory which predicts the relationship is wrong or that the effect is so small it is drowned out by other more significant effects.

The policy import appears to be that within quite wide limits a country can chose the size of public sector without affecting aggregate economic performance (providing it is fiscally responsible and manages the sector efficiently, and so on).

Thus the best balance between the public and private sectors is a matter of political values, of which I, Kerr, Susan St John, the editor of The Independent, and each of the paper’s readers may have a view. We may disagree, but it would be torturous misuse of economics to try to buttress our political views by alleging that the balance somehow affects the overall growth rate of the economy.

Ideology selects the facts in GDP analysis

The Independent 13 February, 2002.

Roger Kerr is struggling with his logic. (The Independent 30 Jan)

He said a certain set of countries with high growth rates had “property A” (low government spending) and this demonstrated that property A was necessary for high growth.

I pointed out that those countries also had property B (industry policy), so Kerr’s logic was that property B was needed for high growth.

Kerr responded that property B was not needed because he disapproved of it.

It seems to be that the countries with high growth rates happened to have property A which Kerr approves of, at which point he draws an illogical inference to strengthen his approval. But he is not willing to do this in regard to property B, if he disapproves of it.

Bryce Wilkinson tries to defend Kerr by improving his logic. (The Independent 30 Jan) I am sure all readers will wish him well in this challenging task. Wilkinson’s problem is an empirical one. Certainly there is a problem of government failure, so not all its spending is as effective as it can be. But some of the spending is very effective, so the problem is in practice (not logic) the two effects trade off.

Is the net effect positive or negative? The empirical investigations suggest they seem to cancel out as far as the impact on output and growth is concerned. The policy implication is that carefully designed government spending should not damage output, and can enhance community welfare in ways that GDP does not measure.

Wilkinson cites Winston Bates’ study for the Roundtable. May I say I was disappointed by its quality, but no more than I am frequently so by Roundtable commissioned work.

It is a pity that Bates has not even tried to take in to consideration some of Susan St John’s points, which would tighten up its empirical quality. Sadly much of Bates’ contribution is written in the language of ideological rhetoric rather than independent research, and his bibliography indicates how ideologically selective he has been. Still, unlike Wilkinson, he does not descend into anecdote.

Ambiguous, conflicting research on GDP

The Independent 27 February, 2002.

Bryce Wilkinson has still not grasped the point. (The Independent 20 February)

The entirety of the research on the impact of government spending on economic performance is equivocal.

Anybody can select some of the research to give the conclusion their ideology prefers, but they have to be carefully selective by ignoring research that contradicts it.

Readers will be aware that an article published in December by a Treasury official comes to a quite different conclusion from Wilkinson, for instance.

On the basis of the totality of that ambiguous and conflicting research I conclude that levels of government spending, with in reason, do not affect economic performance. Wilkinson concludes otherwise. From reading her letters and other writings I believe Susan St John is closer to my view, and of course Roger Kerr is closer to Wilkinson’s.

However Wilkinson is misleading himself if he thinks I have no account of the New Zealand growth process, or how we can improve it. I have written general articles, learned articles, and books on the topic. I can only take it that since they do not fit Wilkinson’s preconceptions, he has not turned his mind to them yet.

He also makes a grave analytical error when he says the government’s declared goal is ‘lifting New Zealander’s incomes back into the top half of the OECD’ The government’s goal is in terms of production (GDP per capita) not income (Say, GNI per capita). Every first year economics student knows the difference between the definitions of ‘domestic’ and ‘national’, important because some of the income from production goes to non-national factor owners offshore.

This is an important distinction where a country is running a large current account deficit or planning to rely heavily on foreign direct investment.

For instance Irish National Income has risen somewhat less than Irish GDP.

Desperately seeking definition of standards

The Independent 27 March, 2002.

Roger Kerr The Independent has asked Susan St John and myself to ‘cite any body of reputable research that suggests New Zealand can achieve 4 percent income growth on a sustained basis with a government spending ratio as high as 40%.’

Perhaps Kerr could cite some which demonstrates the opposite first. Then we would know what his standards of ‘reputable research’ are.

The work commissioned by the Business Roundtable is certainly not.

Eminent economist omitted from government spending argument

The Independent 8 May, 2002.

Roger Kerr’s whiskey winning letter ( 10 April) mentions the Bates’ report ‘has six-and-a half pages of references’. He did not mention that most of them are not learned research articles, and even fewer are on the topic which is in dispute: the extent to which there is overwhelming evidence that moderately high government spending inhibits economic performance. Nor does he mention that the most cited author in the list is R. Kerr.

What alerted me to the inadequacy of the Bates’ report is there is not a single reference to Tony (A.B.) Atkinson, an economist so eminent that he does not need to be introduced to the profession. Writing on the welfare state and economic performance Atkinson said ‘Countries with higher spending, it is alleged, have poorer economic performance. However, not only is the evidence mixed, but also such an argument is more difficult to establish than may first appear.’

Atkinson looks at nine research studies and concludes ‘two find an insignificant effect of the welfare state variable on annual growth rates, four find that transfers are negatively associated with average growth, and three studies find a positive sign to the coefficient of the welfare state.’ Not a single one of these studies is cited in the Bates’ report. I suggest Kerr drink his whiskey before readers demand he return it.

Bryce Wilkinson accuses me of having a mote in my eye. (The Independent10 April) I accept the finding of a speck of dust clouding my vision. Every honest scientist knows that he or she makes mistakes, which will be corrected by future research, and so thereby there will be progress. That is the difference from ideology.

Wilkinson may recall that he is quoting from the biblical injunction ‘Why beholdest thou the mote in thy brother’s eye, but consider not the beam that is in thine own eye?’ He may contemplate whether a bottle of whiskey is enough to float a beam.

This letter won “The Independent’s” weekly prize for the best letter of a bottle of Dewar’s whiskey.

‘Theology ‘not science, dominates debate

The Independent 29 May, 2002.

Bryce Wilkinson (The Independent 15 May) seems unable to recall the content of some of my letters to The Independent . I acknowledged that there would be behavioural responses to tax, government spending and other interventions, so why does he repeat the argument?

Wilkinson should address the evidence that their effects on the GDP growth rate do not seem to be great, perhaps because some are negative and some are positive (as Tony Atkinson discusses).

Wilkinson changes his argument. I said that the Bates report omitted a number of studies, especially ones which contradicted his conclusion. I was asked to nominate one. I nominated a group of studies, plus the judgement on them by an eminent economist who comes to quite a different conclusion to the Bates report, arguing that the evidence for an impact on GDP is ambiguous, and the channels through which government taxation and spending impacted on economic performance were complicated, but both positive and negative.

Wilkinson acknowledges that much of the evidence is inconclusive, and now begins to count the number of studies in the Bates report in comparison to that which I mentioned.

Recall my point was that the Bates report was not comprehensive and the totality of the evidence was inconclusive, which suggested that insofar as there was an effect it could be in either direction and was probably small.

I am also astonished about Wilkinson’s claim that ‘test of statistical significance are not tests of economic plausibility.’

What is a test of economic plausibility if economics is a science which aims to describe an empirical world? And in the light of this stunning new methodology, why does he waste time trying to defend the unsatisfactory empirical conclusions that the Bates report and Roger Kerr derive?

Having apparently satisfied himself of the smallness of the mote which he said was in my eye, Wilkinson changes his criticism to that I am concerned with ideology. Since one of my interests is in scientific methodology, again I plead guilty.

‘Ideological’ is a useful shorthand to describe the approach which Chicago professor Melvin Reder called the ‘tight prior’, a Bayesian expression to describe the situation where the theory is invulnerable to evidence which contradicts it.

One can observe such ideologists relying on anecdote, shifting the tests for their theory when the evidence is unfavourable, deploring well respected methodologies when they undermine the argument, stumbling over their logic, using extravagant expressions when standard language would suffice, attacking the critic rather than the argument, and so on.

In a recent review on a book on Science and Religion, the well known physicist Freeman Dyson wrote ‘[Polkinghorne’s] argument makes sense if you accept the rules of theological argument, rules which are different from the rules of scientific argument.

‘The way scientific argument goes is typically as follows: We have a number of theories to explain what we have observed.

‘Most of the theories are probably wrong or irrelevant. Then somebody does a new experiment or a new calculation that proves Theory A is wrong. As a result Theory B now has a better chance of being right. The way a theological argument goes is the other way around. We have a number of theories to explain what we believe. Different theologians have different theories. Then, somebody, in this case Polkinghorne, declares Theory A is right. As a result, Theory B now has a better chance of being wrong.’

Too often, I am afraid, economic debates in New Zealand are dominated by theological rather than scientific approaches.

So I would be happy to accept Bryce Wilkinson’s accolade of my being a serious methodologist, and promise that any bottle of whiskey that goes with it will not turn me into a tight prior.

This letter won “The Independent’s” weekly prize for the best letter of a bottle of Dewar’s whiskey.