Does empirical analysis ever settle a policy question?

Economics has gotten intensely empirical over the last decade, which makes many of the debates on theoretical methodology and on heterodoxy versus orthodoxy frankly irrelevant. When I received my PhD in 1985, anyone who was academically ambitious and whose fields were international trade or economic development (as mine were) would not think of doing an empirical dissertation. Only labor economics was somewhat different.

It is a very different world now. You can hardly publish an applied theory piece in these areas if you do not have at least a few regressions to go with it--and you better make sure you pay serious attention to specification and endogeneity. In the (bad) old days, I remember that it was often enough to say that you had instrumented for a potentially endogenous variable, while hiding the (typically inadequate) instrument list in a footnote.

But does empirical analysis ever help resolve important policy questions? This is the question that a post from my colleague George Borjas got me thinking about. I asked myself: is there something about the world that I believe in that I may have not have, absent some empirical evidence that convinced me?

I must confess answers did not come very easily. But here are three important examples.

1.  Many development economists have been traditionally concerned that the "structural" features of low-income economies make them unresponsive to the typical market-based, price signals. If there were not ample evidence that poor country exports respond robustly to real exchange rates, or that poor farmers respond strongly to prices of their crops, I may have become a structuralist too.

2.  Another strand in development thinking--going back to Sir Arthur Lewis and Simon Kuznets at least, contends that inequality is not only inevitable in the earlier stages of growth, but that it is actually good for growth as it enables the rich (who are the only ones assumed to save and invest) to have the resources with which they can stimulate economic growth. But by now we have enough accumulated evidence to discount this story. The least that can be said is that more inequality is not conducive to higher economic growth. It may even go in the reverse direction, but the evidence is not terribly strong on that either (despite one of my own early papers).

3. I might also have believed that democracy, whatever its other merits, is not very good for growth, and that the kind of reforms required to generate development require some form of authoritarianism. One again, the empirical evidence soundly refutes this view. Democracies do not pay a growth penalty, and on top have other redeeming economic features (such as better distribution, lower instability, and greater predictability).

Can readers come up with other examples?

UPDATE: Gallagher asks whether I am backing off from my reading of the empirical literature that there is no systematic relationship between trade liberalization and economic growth. No, I am not, but I did not want to seem like a broken record, so I left this off.   While on this topic, I see much of the profession--at least those who are empirically oriented--from having moved away from an unconditional linkage between trade liberalization and growth. The reigning conventional view is that trade liberalization produces growth only if a number of other conditions are present--if the right institutions are in place, if the macro environment is stable, and so on. The "so on" piece is important, because it leaves us off the hook, in case things still do not work out...

Comments

[unemployed observer]

One thing I believe there is no strong empirical evidence, but I firmly think is true is about the enforcement of Intellectual Property Right in developing/emerging economy.

To me, developing countries should be able to use many technologies such as window vista without paying the monopoly price. I can't image Microsoft will stop producing window software because of some poor countries' violation of IPR. But I can foresee there will be a productivity shock if they have to pay a high price.

In most incidents, when the countries' development reaches some threshold, their production will move from duplicating products to creating new products. they then will naturally ask for IPR enforcement.

[unemployed observer]

I mean no strong empirical evidence is whether strict enforcement of IPR on developing countries is good for global growth (R&D increases impact on developed countries vs the neg shock impacts on developing countries)

[faruk]

Kuznet also believed (argued?) that the relationship between capitalist development and inequality would be an n-shaped curve in the long run. But the increasing inequalities in the post-keynesian era showed that this is more about policies than about any inherent feature of capitalism. It does not look like an 'n' anymore.

[gallagher]

Similar to the Kuznets curve regarding inequality and income, since the early 1990s a literature on an "environmental kuznets curve" (EKC) has emerged. Here the initial theory was that in the early stages of economic growth a nation would withstand significant economic degradation but at a certain "turning point" nations' per capita environmental degradation would begin to decrease. This was based on some early regression analyses with rather limited datasets. The DC policy community and IFIs ran with this claiming that developing nations should grow now and worry about the environment later. Given that initial turning points were seen to be around mean global income, a green future was right over the hump.

There have now been close to 100 peer reviewed articles on the EKC and economists are very wary of such crude policy prescriptions (although the policy advice cited above is rife in Washington, most recently in arguments against adding environmental components to trade deals). Indeed, the EKC relationship only holds for a small handful of pollutants and mostly in developed countries. In some cases pollution increases linearly with income (CO2 and solid waste for example). Finally, when EKCs are found they are estimated to occur at much higher turning points (15,000 to 25,000 GDP per capita and higher).

Given that it may take decades for a low-income country to cross from the upward to the downward sloping part of the curve, the accumulated damages in the meanwhile may far exceed the present value of high future growth. Therefore active environmental policy to mitigate emissions and resource depletion in the early stages of development may be justified on purely economic grounds.

[gallagher]

Dr. Rodrik,

In your NBER "Skeptics Guide" to the trade and growth lit and other work showing that "institutions rule" you argue that there is not a robust correlation between trade liberalization and economic growth. Have newer studies come out to make you back off such an assessment?

[Thorsten]

Dear Prof. Rodrik,

while I am not a structuralist, because I find it hard to identify completely with a school of thought, I think that there are actually features of the structuralist school of thought, the way I understand it, that seem to me highly relevant for the debate not only in development economics.

Again, this issue has been covered also recently by "mainstream" economists. I am thinking of the importance of social structures for the working of the economy. For me the basic message of structuralism is that to understand how an economy works you have to take into account the social setting. One example might be the wage setting mechanism or the structure of financial markets (e.g. the Korean & Taiwanese financial systems differ(ed) from the US-American one).

Another issue is the savings issue. While less related to the development problem, Aghion and others have applied the post-Keynesian notion that more saving is bad (I am exaggerating somewhat) and being a net exporter is good and showed that the effect of savings depends on the level of development.

A third and last issue that came to my mind is the New Economic Geography: Now it is completely fine to allow for polarization or convergence in models of (spatial) development, whereas in the old days neoclassicals would preach convergence and heterodox economists like Kaldor predict divergence.

[notsneaky]

In response to gallagher and the Update, I think this paper by Romain Wacziarg pushes a bit in the other direction (opposite to R&R):

http://wber.oxfordjournals.org/cgi/content/abstract/15/3/393

(it's available for free).

Anyway, nothing wrong with a broken record if it's playing a good song.

[Steve Waldman]

DR -- Even some of us without a laptop to recommend are eager to mention how grateful we are that you have joined this crazy blogosphere.

Thanks to gallagher, I've found your skeptic's guide, and look forward to reading it. But can you drop some breadcrumbs on some of your other empirical lessons? To the policy-interested layperson, (1) seems to comport with what we might read in the newspapers, but (2) and especially (3) seem (delightfully!) surprising. Could you drop a few references for each of your bullet points where we might learn what you have learned?

[Bill Sundstrom]

In ECON 1 we always used to say (and show that) "the minimum wage causes higher unemployment." Card and Krueger's work convinced many of us that even if this is true, the effect is probably very small. This has made it possible for many mainstream economists to support at least modest minimum wage hikes on distributive grounds.

[Nicholas Gruen]

The emerging consensus in development in the late 70s and 80s that export orientation mattered is an example of the kind you're looking for I think, though perhaps it is comprehended in your first point.

[Kramladen]

I think, as Bill Sundstrom, that empirical analysis can settle policy questions. The other quesion is whether other people accept it.

The complicated econometric methods employed in many empirical studies are not always well understood - that makes it harder to settle questions most of the poeple have firm beliefs about.

Best,

Kramladen