Small Government Spurs Growth? Economists Say No.

Small-government conservatives’ most powerful economic argument–which progressives have failed to counter effectively in the minds of Americans–is that making government smaller results in faster economic growth. So, by this theory, small government makes all boats rise–rich and poor alike. It follows that progressives who argue for higher taxes and government spending are either foolish or churlish or both: they’re actually hurting poor people by denying them the benefits of economic growth.

This argument is powerful because part of it is undeniably true, at least in the long term. Over the last century, half century, quarter century, economic growth has yielded just plain outright spectacular gains in prosperity and well-being throughout the economic spectrum, and throughout the world. (Yes–by some quite reasonable measures–even in the U.S., even for the poor and middle class.) Progressives who deny this have their heads just as deeply in the sand as conservatives who use it to say, essentially, “why worry? be happy.”

The actual flaw lies in the other part of the argument: that small government contributes to this growth and prosperity. It may be true, to some extent, in developing countries. (One possible explanation: because the governments in developing countries are more corrupt, more government just makes things worse.) But in developed countries like the U.S., the empirical data is quite clear: there’s little or no correlation between government size and growth. That conservative belief is rooted on faith and ideology, not facts.

Progressives often counter the conservative argument with their own spectrum of beliefs. Two failings:

1. They often attack the growth belief–denying those very real facts–instead of rebutting the small-government belief.

2. Many of those progressive beliefs are themselves faith-based (everyone’s opinions are equally valid, right?), or are founded on pretty shaky evidence (there’s a lot of about economies that we just don’t know).

So their arguments have not been as effective as they could be. This is unfortunate, because there’s a much stronger reality/fact-based argument lying on the ground, waiting for progressives to pick it up.

Economists have been studying national growth and prosperity–and what causes/correlates with growth–for decades. In particular, they’ve done dozens of cross-country studies, using an endless variety of statistical techniques and economic models to analyze an endless variety of possible growth factors. They’ve come to consensus on a number of things. See, for instance Greg Mankiw‘s list, reproduced in an earlier post.

What’s missing from that consensus list? Any correlation between government size and growth. (Again, at least in developed countries, within those countries’ range of taxation–23-49% of GDP.)

As I said, there have been dozens of studies. Some show a positive correlation between government size and growth, some show negative. In almost every case, the correlations are small to tiny. Only a handful of outliers showed any kind of statistically significant correlation.

One example, a 1996 study by Robert Barro (the world’s #2-ranked economist by number of citations), exemplifies how questionable the correlations are. He finds insignificant correlations between government size and growth for OECD countries  (positive) and “rich countries” (negative), but for all countries combined he finds a (barely) significant negative correlation. With these contradictory results in a single study–and across all the studies–what can you conclude?

Happily, we don’t need to review all those studies ourselves, because someone’s already done it. In 2003 Nijkamp and Poot delivered their “Meta-Analysis of the Impact of Fiscal Policies on Long-Run Growth.” Their stated goal is to examine the “conventional prior belief regarding the impact of fiscal policy on growth as the hypothesis that increases in government consumption, defence, or increases in tax rates, lower growth; while increases in government expenditure on education or infrastructure enhance growth.”

The abstract:

The issue of whether the public sector enhances or retards long-run economic growth has been debated passionately in recent years. We use meta-analysis to shed light on the issue. A sample of 93 published studies, yielding 123 meta-observations, is used to examine the robustness of the evidence regarding the impact of fiscal policy on growth. Five fiscal policy areas are considered: general government consumption, tax rates, education expenditure, defence, and public infrastructure. Several meta-analytical techniques are applied, including descriptive statistics, contingency table analysis and rough set analysis. On balance, the evidence for a positive impact of conventional fiscal policy on growth is rather weak, but the commonly identified importance of education and infrastructure is confirmed. The results are sensitive to several research design parameters, such as the type of data, model specification and econometric technique. The top two tiers of journals appear less supportive of the conventional priors with respect to government and growth than lesser-ranked journals.

The details on government size and growth:

They found 41 studies that analyzed the correlation between growth and “Government consumption or ‘size”. Here’s the breakdown for those 41 studies.

Bigger government correlates with slower growth: 29%
Bigger government correlates with faster growth: 17%
Bigger government has an inconclusive correlation with growth: 54%

They comment on the high percentage of inconclusive findings overall (35.8%), saying “our sample undoubtedly suffers from file drawer bias or publication bias in that significant findings are likely to be more prominent in our sample than in the excluded papers (Begg 1994).” In other words, there are a lot of inconclusive studies out there than never got published, which would push that already-high 54% number even higher.

Their conclusion based on these results:

It is then easily calculated that the 95 percent confidence interval for the proportion of studies that support the null hypothesis is (0.15, 0.43). This interval lies far away from unity. Consequently, we conclude by means of our sample that the relative distribution of economic activity between the private and public sectors across countries and regions appears to have no clear impact on long-run growth at the macro level.

In other words, what I said above: (at least in developed countries like the U.S.,) there is no correlation between government size and growth.

This is all correlation, of course. It can’t “prove” that X, Y, or Z causes growth. But it can disprove it. As Mankiw says (again, see previous post):

…correlations among endogenous variables can rule out theories that fail to produce the correlations, and they can thereby raise our confidence in theories that do produce them,…

Since there’s no correlation between government size and growth, we can rule out theories that claim that there is a correlation–much less causation.

Two more quotations that I think are especially important:

With respect to government consumption of resources, the composition of this claim on resources may be more important than the level. For example, public expenditure on education, R&D and health care are forms of capital accumulation rather than current consumption and therefore sources of growth, but current consumption expenditures that ensure the right institutional environment (in terms of property rights and safety) may also be growth enhancing (e.g. Barro 1997). Moreover, public funds allocated to environmental policy may also benefit growth in the long run (e.g. Bovenberg and Smulders 1996).

Short story, which makes all the sense in the world: money can be spent effectively or ineffectively, either enhancing or impeding long-term growth. Looking at the data can help us figure out which is which.

At the same time, it has been increasingly recognized in growth studies that the way in expenditures are financed matters too. For example, Kneller et al. (1999) define a range of taxation and expenditure variables and explicitly take account of the budget constraint. By means of a panel of 22 OECD countries, 1970-95, they find – firstly – that distortionary taxation reduces growth, while not-distortionary taxation does not, while – secondly – productive government expenditure enhances growth, but non-productive expenditure does not.

This supports the very well-reasoned post offered up by Matt Yglesias the other day  (seconded and enhanced over at the Reality Based Community). It’s the combination of policies that enhance prosperity and well-being, not any single factor in isolation. (i.e. “cut taxes.”)

I’m hoping this post gives progressives the kind of evidence they need to counter the conservatives’ most telling economic argument, on its own terms. It’s obvious to us that government can do a lot to promote both prosperity and widespread well-being; they’re not mutually exclusive, or a zero-sum game. (Though of course they can be with specific policies.) One big step to attaining those ends is effectively countering the faith-based belief that has dominated economic discussion since Reagan.

It’s time to stop playing on the conservatives’ home field by arguing about whether tax cuts pay for themselves by stimulating economic growth. Because the fact is–except in the very short term–they don’t stimulate economic growth

At risk of plagiarizing from Reagan, but in hopes of “changing the trajectory,” we need let Americans know that “small government is not the answer.”