Created 6/18/1998
Go to DeLong's Home Page

Notes on Edward Wolff's book proposal: Productivity Convergence: Theory and Evidence


J. Bradford DeLong, U.C. Berkeley

1. Scope

I have a great deal of respect for Edward Wolff. His work in the past has been highly original and very valuable. I think he will do a good job analyzing new growth theory--what in it is worthwhile, and what in it is worthless. I highly recommend publishing this book. I highly look forward to seeing what he has to say.

I am not sure that the theories--factors--empirical results threefold division of the book will turn out to be the best way to organize it. How can you talk about theories unless you already have some strong opinions about the factors that have been salient in economic growth at the back of your mind? If I were writing the book I would probably organize it somewhat differently. I would first describe the empirical shape of the subject: how fast growth has been, and how much the different countries of the world have diverged. I would then discuss models--interweaving the factors that different schools of thought and studies and models have taken to be salient. And I would then consider the empirical evidence.

But these are matters of taste, and Wolff's judgment in matters of organization has been good in the past. I only ask that he think about how best to draw potential readers in.

I also find myself wondering whether his three-fold division of the empirical studies makes the most sense. Wolff proposes three chapters: one on the long-run sources of growth among the nations that are now in the OECD, a second on post-World War II growth within the OECD, and a third on post-World War II growth in the rest of the world.

One problem with this (as I have perhaps made myself boring by repeatedly pointing out) is that the OECD is a set of countries that have been, in large part, selected for their economic growth success. The OECD was originally made up of the set of Marshall Plan recipients, plus the U.S. and Canada. During the 1960s the then-rich countries of Asia--Australia, New Zealand, and Japan--were added as the focus of the group shifted to become a group of rich countries. And recently South Korea has been added (because it is now, or is about to become, a key industrial country), along with Mexico (as part of the U.S.'s attempt to accelerate the development of its southern neighbor).

This means that the OECD is an odd group to attempt to do inference on. The temptation to do so is irresistible, because of the excellent information compiled at and provided by the OECD secretariat in Paris. But it seems to me that there are strong reasons for being wary.

I usually try to correct for the biases implicit in the set of the member of the OECD by adding the once-rich countries of the southern cone of South America, excluding Japan (and, now, Korea and Mexico) and saying that analyses of this OECD-prime are analyses of those countries that were thoroughly integrated into the first global economy of the late nineteenth century. Alternatively, you can stick with the *original* OECD--without Japan, Australia, and New Zealand (and, now, Korea and Mexico)--and say that you are analyzing the behavior of Marshall Plan recipients. (And, at least in Barry Eichengreen's and my view, analyzing the development trajectory of a set of nations that as a result of Marshall Plan aid adopted very similar and very successful politico-economic institutions in the post-World War II world).

It seems to me similarly important to include Japan in the post-World War II set of developing countries. There is a temptation to count Japan as an industrial economy by virtue of its role as an aggressor in World War II. But in World War II Japan punched far above its weight. Post-World War II Japan marks the upper limit of how well an economy can do at catching up to the industrial leaders.


2. Theoretical Underpinnings

I think the look at production functions and growth accounting will be very valuable to readers. However, I found it hard to determine what Wolff was going to do. Where is the value-added beyond Nadiri (1970) in this part of chapter 2? And what does Wolff think of the Denison approach to understanding economic growth? It certainly is very different from cross-country growth regressions, or from growth models.

I also found myself wishing for more of a hint to what Wolff was going to say about variations of the Solow growth model. How credible is it to view empirical dynamics as convergence to steady states? What--exactly--is meant by "different production functions in different countries"? Last, I have always been puzzled by Mankiw, Romer, and Weil's specification of human capital. It seems to imply that a primary-school teacher in the U.S.--with 20 times the real wage of China--imparts 20 times as much "human capital" to his or her students as does a primary-school teacher in China. I can believe that there is a significant gap in quality-of-education--that a year is not just a year, but needs to be weighted somewhat by the resources available to each teacher. But Mankiw, Romer, and Weil's assumption always seemed to me to be very extreme.

I would be very interested to know what Edward Wolff thinks about it.

I think that a very valuable contribution of this book would be to focus more of economists' attention on the so-called "technology gap" literature. New growth theory appears to have ignored it (or almost ignored it), and Brian Arthur's feelings about the rest of the economics profession have recently been very publicly aired in the New Yorker. This general line of argument has been subject to a hard-nosed critique by Liebowitz and Margolis (with which I do not agree). My suspicion is that Liebowitz and Margolis have had more of an impact than they should have had, and that Wolff's book may be useful in redressing the balance.

With respect to endogenous growth theory, I have never been sure whether endogenous growth theory formalizes relationships between the rest of the economy and productivity growth, or parodies them. Is there any reason to think that any particular one of the channels proposed by endogenous growth theories are in fact important in real life? Are the implications of a number of important externality-generating factors different from the implications of one, key factor? How useful, really, is endogenous growth theory in understanding patterns of economic growth?


3. The laundry list

Wolff proposes a laundry list of factors that affect economic growth: technology transfer via catch-up, high investment, vintage effects, public infrastructure, computers, science, education, foreign trade, foreign direct investment, population growth, socio-political variables, and structural change. I suspect that I would be most interested in Wolff's analysis of economic growth and structural change: does our default assumption of a one-sector growth model lead us seriously astray?

There is also the issue of the *level* of causation. Investment is important, yes; but are we analyzing the direct causes of growth (in which case we want to focus on investment) or the indirect steps taken by governments that set the stage for economic growth (in which case investment is endogenous)? Population growth is important, yes; but isn't population growth overwhelmingly an endogenous variable? Isn't your population growth rate largely determined by where you are in the demographic transition--which is largely determined by how rich you are?

There is also the issue of whether there is one path or many paths to economic growth. People have talked about "developmental states" and "neoliberal states." People have talked about export-promotion and about import-substitution. People have talked about state-led megaproject capital-intensive industrialization, and also about market-led grassroots development. All of these strategies can be carried out very badly, leading to economic catastrophe. Each has advocates who claim that it also can be carried out well--but only if a bunch of things go nearly exactly right.

It seems to me that linear grwoth models are ill-equipped to handle and analyze such a situation, but I don't have a much better way of proceeding to propose.


4. OECD long-term growth

Chapter 4 looks very straightforward. I have little to suggest: it seems that Wolff has this chapter well in hand.


5. The post-WWII OECD

The big problem with making sense of the post-WWII OECD is that the dimensionality of the set of variables of interest is vastly greater than the dimensionality of the data. Economists are not very good at dealing with situations in which the dimensionality of the data is much lower than that of the set of variables of interest. I think that Wolff should devote a lot of attention to trying to figure out how to make progress on this issue (perhaps along the lines of Xavier Sala-i-Martin, "I Just Ran Four Million Regresssion"). It would be very helpful.

Let me also urge Wolff to upgrade the productivity slowdown to a full chapter, and to try to integrate it into growth theory. So far the two literatures have been almost totally distinct. Yet they should be very closely related, and should have a lot to say to each other.

Let me also urge Wolff to upgrade the sector-by-sector discussion to a full chapter. Dollar and Wolff poses a substantial challenge to the one-sector model through their industry-level results. It seems to me--because some factors operate at the industry level and others at the aggregate level--that this is an area in which something genuinely new and interesting can be accomplished.


6. NICs and LDCs

I have always been very suspicious of Alwyn Young's claims about Singapore. It struck me when I first heard of his paper that if TFP in Singapore today is the same as it was in 1965, then Singaporean factor prices must be on the same factor price-frontier today as they were in 1965. This seems to me to be extremely, extremely unlikely--so I concluded that Young was probably placing much too much weight on official statistics than they can bear. Wolff's view of this issue--and the related issue, from Abramovitz and David (1973), of whether the first-stage of industrialization is usually very capital-intensive--would be very useful.


7. Future prospects

It is a brave economist who ventures into futurology, but let me urge Wolff to do so--with special attention to theories that microelectronics is not just another "leading sector" but is a genuinely new thing, generating a "new economy."

The answer to the question of how many countries can specialize in producing light manufactures for the industrial core is an interesting one. The answer is not obvious to me. I am somewhat surprised that the answer seems obvious to Wolff, and I would like to know why...

I am also surprised at Wolff's judgment that GDP per capita in sub-Saharan Africa today is considerably less than under colonialism. It is not clear to me that that is so. Once again I am curious...

Last, let me urge Wolff to drop OPEC and the former CPEs from the analysis. Their problems and opportunities would require another entire book or two, because they are very different from the rest of the world...


8. Final reactions...

All in all, my final reaction upon finishing the book prospectus is one of annoyance: I am annoyed that I cannot read the resulting book now, because I very much want to.

Professor of Economics J. Bradford DeLong, 601 Evans Hall, #3880
University of California at Berkeley
Berkeley, CA 94720-3880
(510) 643-4027 phone (510) 642-6615 fax

This document:

Search This Website