Robert William Fogel Interview: Conversations with History; Institute of International Studies, UC Berkeley
Page 5 of 6
I'm curious as I listen to you talk. Clearly, methodology is important, history is important, economic theory is important. Over the lifetime of your career, you've addressed big projects -- slavery, railroads, aging -- and have tried to locate them and understand what these phenomena were in a broader institutional [framework]. What drew you to these subjects? Is it something in your background? Is it the evidence that's available? Just what interests you? I would be interested in the origins of the topics that you've selected.
A lot of them are accidental. The general trend is not accidental, but when I was sitting in Simon Kuznet's class on economic development across countries and over time, at one point he said, "One of the most important factors in economic growth and development was the invention of the railroad, but no one has really studied it," and I said, "Oh, I like railroads." Actually, I had done my Master's thesis on the Union Pacific Railroads, so I already knew something about it. So I decided to write my Ph.D. dissertation on railroads and American economic growth.
I got into the study of slavery as an accident. The precipitating factor in this case was the editing of a book called The Reinterpretation of American Economic History, which was supposed to bring together about thirty of the best articles in cliometrics; that is, in works of economic history using pretty high-tech statistics and econometrics. We divided the book into nine sections, one of which was on the economics of slavery. Stan Engerman and I wrote a long introduction to it.
The work up until that day had focused on whether slavery was profitable or not, and whether the slave South was growing more or less rapidly than the North. In the introduction we asked, "What issues about slavery should the cliometricians take on next?" We said, "They ought to show how much less efficient slavery was than free agriculture." To emphasis that theme, we thought we would do a little back-of-the-envelope calculation. We thought that the back-of-the-envelope calculation would show that slavery was at least 50 percent less efficient, but when we did it, it turned out to be 6 percent more efficient. We were startled. We said, "Boy, we really messed that one up," and we started looking at it in a little bit more detail, and we did it more carefully. At that point, we just hoped it was less than 1 percent. We thought maybe people had been too sanguine in thinking that it was a 50 percent difference. But when we redid it, it turned out to be 36 percent more!
So at that stage, we applied for and received a large NSF grant to study slavery. One of the things we had to study was the demography of slavery, because the health and longevity of slaves was a factor in how you measured productivity. It was there that we got interested in differential mortality trends, and we began to realize that we knew very little about trends in mortality in the United States before the 1890s. We were aware that the Mormons had collected genealogical data, and that was available in Salt Lake City, with which you could reconstruct the pattern of change and mortality. So we began a project on the improvements in life expectancy from about 1710 on.
That lay the basis -- those experiences and the creation of those data sets, and the recognition that we now have high-speed computers that enable us to get data that we knew was there but nobody thought was usable. With the appearance of high-speed computers and then later on, laptops, so we could actually take them into the archives and input the data at very low cost, and process it at relatively low cost, we [were able to take on this task]. We received in 1991 a grant to begin the study of the aging process in the Union Army. It was a $3 million grant, and we calculated what it would have cost us to have done that project with the technology of 1978: the answer was about $300 million. We would have needed a special bill through Congress. It was way outside the parameters of even big grants at the National Science Foundation or the National Institutes of Health.
So the change in technology made it possible to go through massive data retrieval. In the case of the Union Army sample, in order to get a lifetime picture of the changes in health over the life from early childhood to death, we had to link together information from twenty different data sources. To describe that lifecycle pattern, including all of the health information, takes 15,000 variables. When I was using computers in the mid-to-late 1970s, 5,000 observations with maybe 50 variables was thought to be the outer limit of what you could handle. Here we had 40,000 observations, with 15,000 variables on each observation. And I won't say it was a snap, but it was doable.
So the long-term trend is that one problem leads you to another, and the technology coming onboard allows you to do things you could not have done in the first instance.
At every stage in our work on long-term changes in health and longevity, at every stage we were pushing the limits of technology, and our main problem was to convince our peers that it could be done. They were all convinced it was a good idea, but they had to be convinced it could actually be done, and within the expenditure limits that we were proposing.
I guess some of these results, like the results on slavery, were pretty controversial. So it's a combination of what the evidence shows, on the one hand, being controversial, but also the controversy aroused by saying, "We want to go in this direction."
It was very controversial within our own research group. There were times when I thought that Stan was nuts, and there were times that he thought that I was nuts, and our graduate research assistants thought we were both nuts, although sometimes they were ahead of us. But it was so contrary to received wisdom.
On slavery. When we got our first results, we were so dubious about it that we delayed publication for two years, trying to falsify the result. We only published it after two years when we couldn't break our own findings.
Next page: Conclusions
© Copyright 2004, Regents of the University of California