Monday, August 20, 2018

Productive and Unproductive Error


The list of new working papers this morning from nep-his contained one by Richard Langlois, “The Fisher Body Case and Organizational Economics”, which examines the case of Fisher Body. Mary and I also examined the case in our paper “The historian’s craft and economics,” This blog post is both a concurring and dissenting opinion on the Langlois paper.

Fisher Body was an early manufacturer of automobile bodies that merged with General Motors in the 1920s. Both our paper and Langlois paper are, however, less about the story of the merger than they are about the story about the story. The story about the story begins when Klein, Crawford and Alchian published a paper in 1978 that used Fisher Body as an example of vertical integration driven by the holdup problem arising from asset specificity. For non-economists the terms may be unfamiliar, but the intuition is straightforward. The relationship between the automobile manufacturer and the manufacturer of bodies will be most profitable if they make investments that are fitted to each other. For example, the assembly plants are close to the body plants, or the body plants are designed to produce the sort of bodies the automobile company uses. In other words, you have assets, factory and equipment that are specific to a particular use and much less valuable in alternative uses. Hold up is the idea that once such investments have been made one side can try to exploit the other. “We are going to lower the price we pay for your automobile bodies, and you will have to accept it because no one else is going to buy those bodies that are designed for our cars.” That is a really simplified version, but you should get the idea.

In 1988, Klein described this explanation for the merger
“Fisher effectively held up General Motors by adopting a relatively inefficient, highly labor-intensive technology and by refusing to locate body-producing plants adjacent to general Motors assembly plant” (Klein, 1988: 202).

In 1988, Coase began to question this explanation, and he and several other researchers presented considerable evidence that the sort of holdup described by Klein never took place and that the merger was motivated by other concerns. Klein and Coase argued back and forth, and several other people jumped in. Ultimately Klein acknowledged that the hold up he described never took place.
Both Coase and Hansen and Hansen present the case as an example of economists’ failure to give sufficient attention to empirical evidence. Langlois on the other hand sees the entire episode as an example of the process of developing economic knowledge. Here is the conclusion of his paper



I agree with the overall thrust of Langlois argument but disagree with his specific case. I think there are numerous examples of cases where economists have developed explanations that turned out to be incorrect but were important because they generated debates and stimulated research that led to better understanding.

For instance, Doug North’s explanation for economic development in the antebellum period prompted research into agricultural production in the South and industrial development in the North that overturned his explanation and led to a better understanding of antebellum development. I think a similar thing is going on with Pomeranz’s work on the Great Divergence. Academics should not be criticized just for being wrong, especially if it leads to research that improves our understanding of something. 

So what is the difference between North and Pomeranz, on the one hand, and Klein et al, on the other. First, Klein et al simply use Fisher Body to provide a little color to the theory. Their primary objective is not to understand the history of Fisher Body and GM. The work of North and Pomeranz on the other hand was driven by the desire to understand history. Second, North and Pomeranz based their theories on the available evidence, and some of their claims were later challenged by new and better evidence. The initial claim of Klein et al was not based upon the available evidence. There was already evidence in support of a contrary claim and no evidence in support of their claim that Fisher held-up GM. Finally, the responses to North and Pomeranz took us beyond where we were before they wrote. In contrast, the responses to Klein et al largely took us back to where we were before. According to Langlois



The use of Fisher by Klein et al is an example of trying to cram history into a theory rather than using theory to understand history. When I was studying economic history at the LSE in the early ‘80s I became increasingly interested in the use of economic theory in economic history. My advisor, Geoff Jones, told me that theory could be very useful but that I had to be careful not to try to make a historical story fit into a theory. Theory might help guide the search for evidence, but you have to go where the evidence leads you even if that doesn’t fit your theory. And you must tell the story honestly.  Although I went on to get a Ph. D. in economics rather than history, I have always tried to keep Geoff’s advice as a guiding principle. I think Klein et al would have been less likely to follow the path they did if they had been driven by a desire to understand Fisher Body.

Wednesday, August 15, 2018

That's Just Econ 101


The other day my wife and I were talking about how people like to use phrases like “that’s just Econ 101”. Unfortunately, statements that precede that phrase are almost never Econ 101. It’s just whatever the person happens to think. I said that fairy tales should be re-written to include the phrase. “If you don’t make the shoes, elves will do it at night. That’s just Econ 101.” "He spins straw into gold. That's just Econ 101."

Bill Gates provided support for the claim that people to tend to say things about basic economic theory that are not part of economic theory.





He says that it costs as much to build the 1,000th unit as it cost to produce the 10th. Economists call the additional cost of producing another unit of a good marginal cost. What he is describing is constant marginal cost. The problem is that he drew an upward sloping supply curve. An upward sloping supply curve means that higher and higher prices are required to get suppliers to provide additional units of the good. Why are higher and higher prices required? Because of increasing marginal cost. Gates description of constant marginal cost is thus inconsistent with his graph. This is actually a small problem because Gates could have just as well said that economists assume increasing marginal cost when drawing typical supply and demand curves.

His real point seems to be that there are now many goods for which there are large startup costs and near zero marginal costs. That is how he describes software production. He suggests that this is a new development that arises from the intangible nature of goods like software.

It is not actually a new situation, it does not arise from the product being intangible, and undergraduate economic textbooks have models of this situation.  There are many examples of products for which there are very large start up costs and relatively small (near zero) constant marginal cost. Railroads are not new, and they are not intangible, but they required large expenditures on construction and, once built, the additional cost of another passenger or another bushel of wheat was practically zero.  Insurance is intangible, but it is not obvious that it can be produced at zero marginal cost.

This is a graph from McCloskey’s Applied Theory of Price, which, by the way, is available to you at the amazingly low cost of your time to download it.


Gates concern that the rules have not kept up with the changes in the economy is also not new. People said the same thing in the late nineteenth century with the rise of railroads and other big businesses with high start up costs and low marginal costs. 


Gates does not address the demand side but these are generally firms that have some degree of market power. They face a downward sloping demand like the firm in the graph. Because other people do not regard other goods as perfect substitutes the firm won't lose all of its customers if it raises its price. The more the firm can convince people that other goods are not close substitutes for its good the greater its ability to raise its price above marginal cost. I some ways the more important thing is whether it can keep other companies from offering close substitutes. In other words, can it create what economists call barriers to entry. You can have a great idea, but if you can't keep other people from copying it you are not going to make great profits.

In short,

1. If you want to charge a price that is greater than marginal cost you need to convince people that other goods are not close substitutes for yours. Think of the old Porsche slogan: "Porsche. There is no substitute."

2. If you want to make more than an average rate of profit you need to keep other people from copying you, that is introducing substitutes for your good (or you need to keep coming up with new things that don't have close substitutes).

Those two ideas actually are Econ 101.

Tuesday, June 12, 2018

How I Became a Business Historian (a bit of shameless self-promotion)


Yesterday, the June issue of Enterprise & Society, the journal of the Business History Conference, arrived, and my paper Trust Company Failures and Institutional Change in New York, 1875–1925Enterprise & Society 19, no. 2 (2018): 241-271, is the lead article. And Larry Neal emailed me to tell me how much he liked the paper.

I was think about how I came to write that paper because I did not plan to be a business historian, but there is considerable evidence suggesting that I am one. In addition to Enterprise & Society I have published papers in Business History Review, Business History, and Essays in Economic and Business History.

I had a great opportunity to become a business historian when I was working on my M.Sc. in Economic History at the LSE. I got to work with Geoff Jones, who is one of the leading business historians in the World and currently the Isidor Strauss Professor of Business History at Harvard Business School, but I was interested in the political economy of the opium trade, not business history. Because I was interested in political economy and institutional change I went on to work with Doug North and earn a Ph.D. in economics at Washington University. Prompted by discussions with Andy Rutten, I decided to write my dissertation on the evolution of bankruptcy law in the United States. 

This story about institutional change, the origins of the first lasting bankruptcy law in the United States, turned out to be a story about business history ("Commercial Associations and the Creation of a National Economy: The Demand for Federal Bankruptcy Law," Business History Review 72 (Spring 1998): 86-113).  Because corporate reorganization was not included in the 1898 Bankruptcy Act, I began a separate research project on the origins of corporate reorganization. This story about institutional change also turned out to be a story about businesses trying to shape the law to fit their needs and also ended up in Business History Review ("The People's Welfare and the Origins of Corporate Reorganization: The Wabash Receivership Reconsidered," Business History Review 74 (Autumn 2000): 377-405). That paper won the Newcomen-Harvard Special Award from Harvard Business School and the Newcomen Society. I’ve published papers on a few other things (see my CV if you are interested), but most of my work has been along two paths that were opened by those papers.

The bankruptcy paper led to other papers about the use of bankruptcy law and the evolution of bankruptcy law (“The Role of Path Dependence in the Development of U.S. Bankruptcy Law, 1880-1938” (with Mary Eschelbach Hansen) Journal of Institutional Economics 3 (August 2007): 203 225;  “Crisis and Bankruptcy: The Mediating Role of State Law, 1922-1932,” (with Mary Eschelbach Hansen) Journal of Economic History 72 (2012):440-460; and  “Religion, Social Capital and Business Bankruptcy in the United States, 1921-1932” (with Mary Eschelbach Hansen) Business History 50 (November 2008): 714-727). My wife joined me in this research and we are currently working on a book about bankruptcy in the twentieth century. She also developed another research project that involved collecting and digitizing bankruptcy records, and she has published additional papers out of that project.

The paper on corporate reorganization led to the study of trust companies. Several important corporate reorganization cases involved the Farmers’ Loan and Trust Company. The name was familiar to me from teaching American Economic History because of the income tax case, Pollock v. Farmers’ Loan and Trust Co., and two important railroad regulation cases, Reagan v. Farmers’ Loan and Trust Co. and Stone v. Farmers’ Loan and Trust Co.  I was curious what this company did that left its fingerprints all over nineteenth century legal and economic history. So I wrote a book about the Farmers’ Loan and Trust company and its influence on the law. About the time I finished the book there was increased attention to the Panic of 1907. Descriptions of New York City trust companies as novel, unregulated and reckless did not fit with what I had been reading and writing about trust companies like the Farmers’ Loan and Trust Co.  So I ended up writing a paper that argued that trust companies were not as unregulated as some people suggested and that to understand the Panic one has to understand that not all trust companies were the same. Because I argued that regulation was not as inadequate as had been suggested, I wanted to know how the regulation of trust companies evolved over time, which led to the paper on trust company failures.

In all the cases of institutional change that I have studied there were underlying changes in the costs and benefits of different institutions as well as the costs of seeking institutional change, the sort of things that appear in theories of institutional change, but what most interested me is the creative ways in which people responded to those changes in costs and benefits. I’m most interested in the role of the creative response, to use Schumpeter’s phrase, in the evolution of institutions. In short, I still see the business history that I do as part of the new institutional economics that I set out to do.

Monday, June 11, 2018

Was Slavery Central to American Economic Development?




Antebellum Economic Growth 

Last week on Twitter Matt Yglesias raised a question about changes in how historians interpreted slavery. One historian, Joshua Rothman replied and Edward Baptist added his two cents.



I'm not sure what Rothman means by "in fact and as a matter of history." Perhaps it is reference to Baptist, who prefers to avoid mixing facts with his history. In any case, Rothman seems to believe that the centrality of slavery to American economic development is not something a reasonable person could dispute. I regard myself as a reasonable person. So, on the off chance that someone might be interested in why I would dispute the claim that slavery was central to American economic development I'm going to ask that we take a closer look at the antebellum economy.

The argument against the centrality of slavery is based on two things: the assumption that Rothman is using the word central as it is defined in the dictionary and used by most people, and the available evidence on the antebellum economy. Central means that something is not just important but that it is of primary importance. The central character in a movie is not just an important character, she is the main character, the primary character. This is clearly what Baptist has in mind when he claims that  “the returns from the cotton monopoly powered the modernization of the rest of the American economy” and that "more than $600 million, or almost half of the economic activity in the United States in 1836, derived directly or indirectly from cotton produced by the million odd slaves― 6 percent of the total US population―who in that year toiled in labor camps on slavery’s frontier.” By the wayBaptist's tweet about it being easier to make claims about alternate universes than to come to grips with the past of this one was particularly appropriate given his expertise in making false claims. I and others have shown that Baptist's estimate is nothing but smoke and mirrors, a combination of numbers he makes up and bad accounting; see also Pseudoerasmus posts on Baptist. In addition, Olmstead and Rhode also show that Baptist made up  things that he claimed to have found in the testimonies of enslaved people.  

By the way, if Rothman actually just means to say that slavery was important, I wish he would do that. I don't know anyone who disagrees with that. When, however, he claims that slavery was central to American economic development he is helping Ed Baptist to keep pedaling his snake oil.

To show why I believe that it is not accurate to state that slavery was central to American economic development I begin by reviewing the growth of total output, then I look at the composition of this output, and then I look at the regional distribution of income generated from this production. I argue that the fundamental problems with Rothman's claim are that no one thing was central, and claiming that "one big thing" was central gives a misleading view of economic development.

I should note that the argument is not new. It is essentially the same approach that has been used to counter exaggerated arguments about the role of cotton textiles in the industrial revolution (see McCloskey by way of Pseudoerasmus or the role of railroads in American economic development (see Fogel, but if you want the quick version McCloskey has a back of the envelope calculation of the impact of railroads in The Rhetoric of Economics). I've made essentially the same argument before. I'm hoping that by providing more detail about the antebellum economy that I might clarify the argument, make it more persuasive, and illustrate why most economic historians don't like "one big thing" theories of economic development.

I.                    Overview of Antebellum Economic Growth

Between 1790 and 1860 real GDP grew at a 4.4 percent annual rate, somewhat faster than the long run rate of growth (1790- 2000) of 3.87 percent. However, because population growth was also more rapid than the long run average, per capita real GDP increased at an average annual rate of 1.34 percent, somewhat less than the long run rate of growth of 1.77 percent (all growth rates are from measuringworth.com).

Real GDP in millions of 1996 dollars



Source: Historical Statistics Millennial Edition, Series Ca 9

Real GDP per capita



Source: Historical Statistics of the United States Millennial Edition, Series Ca 11


Keep in mind we are always talking about estimates.  Nevertheless, these are estimates; not just guesses. They are not simply made up numbers. If you want stuff that is just made up rather than based on actual historical research I suggest Ed Baptist’s work. Given these cautions we can look in more detail at antebellum growth, but it is important to keep in mind that when I say something accounted for about 3.9 percent of output in 1850, you should not ignore the about. 


II. What Were People Producing?
               
What were Americans producing in the antebellum period? The table below shows how output was divided between commodities and services. Over the course of the nineteenth century the share of output accounted for by services was increasing, but commodities still accounted for nearly 60 percent of output on the eve of the Civil War.



Source: Gallman, Robert E., and Thomas J. Weiss. "The service industries in the nineteenth century." In Production and productivity in the service industries, pp. 287-381. NBER, 1969.

The next table shows the division of commodity output between different sectors. Over the course of the nineteenth century the  relative importance of manufacturing was increasing, but in the decade preceding the Civil War agriculture still accounted for the majority of commodity output.


Source: Gallman, Robert E. "Commodity Output, 1839-1899." In Trends in the American economy in the nineteenth century, pp. 13-72. Princeton University Press, 1960.

The following table shows how agricultural output was divided between livestock and crops. Unlike the previous tables, these tables show the value of output rather than the share of output. It is, however, apparent that agricultural production was split relatively evenly between crops and livestock.


Source: Towne, Marvin, and Wayne Rasmussen. "Farm gross product and gross investment in the nineteenth century." In Trends in the American economy in the nineteenth century, pp. 255-316. Princeton University Press, 1960.

The next table also shows the value of output and provides a more detailed breakdown of agricultural production. 







Source: Towne, Marvin, and Wayne Rasmussen. "Farm gross product and gross investment in the nineteenth century." In Trends in the American economy in the nineteenth century, pp. 255-316. Princeton University Press, 1960.


On the eve of the Civil War, grain production was the largest source of income from crop production, but the most important single commodity was clearly cotton. Taken as a whole, however, the preceding tables illustrate why it is probably not useful to regard cotton, or any other single good, as the central to American economic development. On the eve of the Civil War, cotton accounted for about 35 percent of the value of crops produced. That is a large percentage, but because crop production was only about 51 percent of agricultural output, cotton only accounted for about 17.8 percent (.35 x .51 = .178) of agricultural production. That is still a large percentage, but if we are interested in cotton’s importance for the whole economy, we must keep going. Because agriculture accounted for 56 percent of commodity output in 1859 and cotton accounted for 17.8 percent of agricultural output, cotton accounted for about 9.9 (.178 x .56 = .099) percent of commodity output. Finally, because commodity output accounted for 59 percent of all output, cotton would have accounted for about 5.8 percent (.099 x .59 = .058) of all output. If you conduct the same exercise for 1850 you would get an estimate of about 4.8 percent. If you compare the cotton values from the above table with estimates of nominal GDP from measuringworth.com, cotton equals 4.95 percent of GDP in 1860 and 4.57 in 1850. Although the second method is more direct, I wanted to show why cotton is a relatively small share of the whole economy: people produced many different things. Even after cotton’s rapid expansion during the first sixty years of the nineteenth century, it accounted for less than 6 percent of GDP. 

I do not have as much information for manufacturing, but Joseph Davis research on industrial production gives us some idea of the growth of industrial output and the relative importance of different components of industrial production. Below is a table showing the weights that he used for different components of his index of industrial production, reflecting their relative importance.  



Source: Historical Statistics Millennial Edition, Series Ca19


Source: Davis, Joseph H. "An annual index of US industrial production, 1790–1915." The Quarterly Journal of Economics 119, no. 4 (2004): 1177-1215.

Employing the same sort of exercise as above, we find that cotton textiles, the largest component in manufacturing in the United States would have accounted for around 3.9 percent (.218 x.30 x .60 = .039) of GDP.

Livestock production accounted for 15.5 percent of output. Food grain production accounted for 3.7 percent. Total grain production accounted for 6.7 percent. 


Let me reiterate my use of the word about. All of these are estimates, and the further back in time we go the more the estimates tend to be based on smaller amounts of evidence. It is all possible that I made an error in here somewhere. New estimates, however, are unlikely to change the overall conclusion because cotton was only a fraction of all crops, crops were only a fraction of all agricultural production, agricultural production was only a fraction of all commodity production, and commodity production was only a fraction of all output. Multiplying fractions tends to generate small fractions relatively quickly. The math just reflects the underlying reality: the United States in the early nineteenth century already produced a wide array of goods and services.


III. Where Were People Producing?                 
One could argue that the focus on cotton does not give an accurate estimate of the impact of slavery. Not all cotton was produced by slaves and slaves produced other things. However, a look at the regional distribution of income also does not support the primacy of slavery in economic development.

Regional variations in personal income reinforce the argument that it is probably not useful to regard slavery and the cotton that enslaved people produced as central to American economic development. The following table shows estimates of personal income generated in different parts of the country.





Source Robert E. Gallman "Economic Growth and Structural Change in the Lang Nineteenth Century" in Cambridge Economic History of the United States. See also Global Prices and Incomes Database, United States in 1860, Personal Income Totals.

The North’s leading role in the economy was a consequence of both more people and higher per capita output.




Source: Lindert, Peter H., and Jeffrey G. Williamson. American Incomes 1774-1860. No. w18396. National Bureau of Economic Research, 2012.

There is one final argument about the centrality of slavery that I should address. Some people claim that because of the spillover effects of slavery. This for instance is the idea behind Baptist's attempt to add up imaginary estimates to calculate the importance of cotton. There are two problems with this approach. First, you can do it with any good. For wheat I could count land sales, and the cost of transportation, and the cost of equipment, etc. Using Baptistian income accounting I could easily show that the amount of national output accounted for by grains and cotton was greater than the total output. How is that for an alternate universe? The other problem is that the evidence does not support the claims of strong interregional linkages during the antebellum period (see, for instance, this post).    

Conclusion


Slavery was not central to American economic development in the sense that it did not power the modernization of the rest of the economy. The claim that slavery was central to American economic development is factually incorrect: slavery was important, but no one thing was central. The claim also promotes a misleading view of the process of economic growth. It suggests that economic growth is about one big thing. No one thing was big enough to drive economic growth, not railroads, not cotton, not cotton textiles. Explanations for economic development need to explain why people were investing and innovating in a lot of different things. Consequently, economic historians have tended to move away from one big thing theories of economic growth toward understanding the underlying causes of development, such as the emergence of a culture of growth (Mokyr), or the rise of bourgeois values (McCloskey)  or the evolution of growth promoting institutions (North, Wallis and Weingast). 

Slavery was not central to American economic development, but it was an important part of American economic development, and economic historians have devoted considerable attention to understanding slavery its continued impact on the American economy (see, of course, work by Fogel, and Wright, and for a small sample of recent work you can look at Logan, Cook, and ParmanLogan and PritchettNunn, Naidu, and Collins and Wanamaker.















Wednesday, June 6, 2018

Podcasts I Listen To


I saw Tim Harford’s list of best podcasts. I agree with some, but thought he missed some good ones. I don’t know if these are the best, but they are podcasts that I often listen to while working out or walking the dog.

Podcasts for Anyone Interested in Economics.

Podcasts for Economists. The discussions on these shows tend to be directed more toward an audience of economists.
EconTalk Econ Talk is usually directed toward a broad audience, but some episodes are directed more toward economists.

History Podcasts That I Like

A Podcast That I Do Not Like
Revisionist History (link intentionally omitted). I listened to the episode on country music, which I love. If you are interested in having someone that knows nothing about country music explain to you what they think country music is about you should listen to it.

Monday, June 4, 2018

Free and Unfree Labor: The Political Economy of Capitalism, Share-Cropping, and Slavery


The UCLA Center for Social Theory and Comparative History hosted an event on the topic of Free and Unfree Labor in March. Below is a link to the page that has an audio recording. Unfortunately, there does not appear to be video.

Speakers:
Gavin Wright is William Robertson Coe Professor of American Economic History, Stanford University and author of Sharing the Prize: The Economics of the Civil Rights Revolution in the American South (2018) and Slavery and American Economic Development (2006). Professor Wright will present on "Slavery and Anglo-American Capitalism.”
Suresh Naidu is Associate Professor of Economics and Public Affairs at Columbia University. Professor Naidu will present on "Labor Markets in the Shadow of American Slavery.”
John Clegg is a doctoral candidate in the Department of Sociology at NYU. His paper is entitled “The Real Wages of Whiteness.”

Wright tells why Eric Williams, Barbara Solow, and Joseph Inikori are right about the importance of British development during the Industrial Revolution and Ed Baptist is wrong about the importance of slavery for American economic development during the 19th century. He also has some positive things to say about the recent work of historians like Caitlin Rosenthal and economists like Trevon Logan.

Naidu talks about the importance of the overall repressiveness of the South as a prerequisite for repression on individual plantations.

Clegg talks about his work on wages of poor whites. It seemed the most preliminary and most difficult to assess without access to the actual paper, or at least the tables and graphs. He points to the recent work of Keri Leigh Merritt, but it was not clear to what extent he regards his work as either supporting or contradicting hers.


Wednesday, May 30, 2018

Economic & Business History Society

The annual conference of the Economic & Business History Society is going on now in Finland. Unfortunately I was not able to go to the conference this year. It looks like they put together a great program.

They are livestreaming the keynote by Deirdre McCloskey at 10:30 eastern time.

I should also note that the society's journal Essays in Economic & Business History is now listed in the Chartered Association of Business Schools Academic Journal Guide. Congratulations to the editor, Jason Taylor, the associate editors, and the editorial board.

Monday, May 14, 2018

Industrial Revelations


I think most people have a very vague notion of what the Industrial Revolution was, and descriptions and pictures are not particularly helpful. You really need to see a spinning wheel, a spinning jenny, and a water frame at work to appreciate what was happening in the 1700s. I have been fortunate enough to visit some great museums and see some of these things at work, but I don’t have the opportunity to do that with my students. That is where Industrial Revelations comes in handy. There are several seasons of Industrial Revelations, and I haven’t had time to watch them all, but the first season with Mark Williams (aka Arthur Weasley) is great for showing many important technological changes during the Industrial Revolution. Here is a link to the textile episode, which I think is one of the best.