How to find a topic for your paper.

I do not expect a publishable paper or a Ph.D. dissertation. But I do expect a research paper, one that adds new knowledge to the existing stock, not merely a summary of the existing literature. In any case the process of finding topics is much the same whatever the goal.

The search for a topic can follow many different paths. Often it begins with a policy question. Are you interested in whether fixed exchange rates are a good idea?  Then you might want to ask how did things really work under the classical gold standard. Are you concerned about the effects of immigration on real wages? Then you might want to findout what happened to real wages when immigration was cut in the 1920s.  Are you concerned that production and prices might be affected by industrial concentration? Then you might want to look at what happened to the price level when there was a movenment toward concentration and monopoly at the turn of the century. Are you concerned about the economic consequences of a war in Iraq? Then you might want to look at what happened to prices and production when the United States got into a "small" war, such as the Spanish American War. Sometimes, history can provide "natural experiments" that provide more information than can recent data. A recent paper in the American Economic Review, for example, looked at the effect of strategic bombing of Japanese cities during World War II on there long run size.

Whatever your interest, you will typically begin by reading the frontier literature in the field. But at some point you will have to find a way to add to what is already known.  There are a number of typical ways that the existing literature in a field is expanded.

(1) Repeating a study with data from a different country. When asked why he was touring New Zealand, the balladeer Tom Lehrer said that it was easier to find a new country than a new song. Milton Friedman and Anna J. Schwartz in their book Monetary Trends repeated all of the regressions they ran for the United States on on data from the United Kingdom. The reason was that they had already reached numerous conclusions about the United States from previous studies; the United Kingdom provided an important new test of their ideas. Nearly every country has had experiences with money, railroads, land tenure, etc. that can be compared and contrasted with the United States.

(2) Explaining an "anomaly" in an existing study. In a famous study of hyperinflation Phillip Cagan showed that a monetarist model emphasizing the rate of inflation explained the data fairly well. But a number of points toward the end of the hyperinflations were "outliers." A number of papers have been written to explain these points, as the result for example, of anticipated monetary reforms.

Of course, it would be better (for your grade, for your career, and for the future of civilization) to come up with an entirely new theory that explains hyperinflation better than the monetary theory: Inflation rates, let us say, are determined by the level of "profit-dissonance," which can be measured by the coefficient of variation of profit rates. When the random walk in profit dissonance carries it above 2.05, hyperinflation results. This sort of thing is what Thomas Kuhn (The Structure of Scientific Revolutions) called a paradigm shift. But normal research doesn't work this way. Most of us are destined to be the economists who gradually extend and refine Cagan's model.

(3) Refuting an existing study by adding additional information. Much of the classical debate between the monetarists and the Keynesians took this form. A Keynesian might claim that changes in the government deficit explained changes in GNP. A monetarist would answer by claiming that when money was added to the expanatory equation the impact of fiscal policy was reduced.

(4) Extending an existing study backwards or forwards in time. Howard Bodenhorn (a successful Rutgers Ph.D from a few years past) and I did a paper in which we extended Lance Davis's study of regional interest rates backwards in time. Davis's study covered the period from the Civil War to the turn of the century. Howard and I looked at the period before the Civil War. We had a reason for doing so: we suspected that the Civil War had significantly disrupted the capital market. And it wasn't easy to find the data. If it had been, Lance Davis and many others who have studied the postbellum period, would have extended the data backwards themselves. But it was worth doing: we found considerable evidence of capital market integration in the antebellum period.

(5) Repeating a study done for one particular market with data from another market. For example, Robert Fogel, a winner of the Nobel prize in economics, estimated the "social savings" created by U.S. railroads in the  interregional transportation of agricultural products. Some economic historians extended Fogel's work to other countries: Britain, Mexico, etc. Others extended it to different markets. Gary Walton later extended Fogel's study to passenger traffic on the railroads.  Stanley Engerman later performed a similar calculation for the Second Bank of the United State. Similar calculations could be made for other forms of transportation and for other technological innovations. Of course, Fogel's study was more significant than most repetitions are likely to be. The railroad was considered by historians to be one of the most important technological innovations of the nineteenth century, and Fogel showed that its impact was rather small.

(6) Repeating a study done for one government regulation for another regulation. Recently, Charles Calomiris and Eugene White studied the political economy of the adoption of Deposit Insurance in the 1930s. There choice was motivated by current policy considerations. But I would guess that almost any piece of federal legislation would repay careful study. At the very least a reading of the Congressional Record and of the Congressional hearings would reveal which economic interest groups promoted and opposed a particular piece of legislation.

(7) Discovering a body of unexploited data. Frequently, the order of things is reversed -- a researcher finds the data and then looks for interesting questions the data can answer. Recently, Lance Davis and Robert Gallman completed a massive study of the U.S. whaling industry in the nineteenth century. This project has yielded a large number of research papers and has drawn considerable attention. It began when Gallman stumbled on the remarkably complete records of the American whaling industry while searching for something else in the library. (Considerable detail is available on every whale caught by an American vessel including length, weight, oil content, marital status, and previous arrest record.)

As an experiment I tried opening the Historical Statistics of the United States, the most important compendium of historical data, randomly. I landed on page 690 which gives annual data on cigar and cigarette production from 1870 to 1970. It naturally brought to mind certain questions -- Is more detailed price data available in the primary sources cited in the Historical Statistics?; did cigarette consumption rise at the expense of cigar consumption, or did total consumption of tobacco products increase?; When have there been changes in the federal tax on cigarettes?; Can Gary Becker's theory of rational addiction be tested with this data; etc. etc. And, of course, the sobering thought occurred that if the data is this good, someone has probably already done the study.

(8) Reexamining older studies using new statistical or theoretical models. Econometrics is constantly evolving -- some would say progressing. Often new statistical techniques can show that conclusions drawn previously by using more primitive techniques are in error. New theories can also provide a basis for reexamining old data. Cagan's study of hyperinflations is again a good example. As new theories were developed in macroeconomics, such as rational expectations, they were tested on Cagan's data.

(9) Following up published suggestions for future research. Many books and journal articles contain suggestions for future research. These are sometimes worth considering. Naturally, you have to ask yourself why the author is willing to share a valuable idea with strangers. But there are circumstances when it appears that useful suggestions have been made.

(10) Reexamining one of the "myths" of economics. The classic example in this genre is Ronald Coase, "The Lighthouse in Economics," Journal of Law and Economics, 1974, 357-76. Economic theorists since John Stuart Mill have used the lighthouse as an example of a public good: a valuable service, but one the private sector would not produce since there is no way to charge the ships that benefit from the lighthouse. Coase examined the actual history of lighthouse finance, and showed that the true story is far more complicated.

(11) Surveying the literature on a particular question. This is usually what students think of as research, and I would like you to stay away from survey articles if you can. I would prefer that you get some empirical data and try to analyze it. Most survey articles don't add to knowledge. But some do, by generalizing from various studies, or by developing a framework that brings various pieces of work together. So I don't want to make a blanket statement that survey articles are unacceptable, but you will have to persuade me that your survey will add to knowledge, and won't simply be a summary of what other people have said.

(12) Asking someone for a topic. Its perfectly acceptable to ask people to suggest a topic.

You can ask me. I have a list of topics you can choose from, although I would like you to at least make an effort to find a topic on your own. I can also help you turn a general question or interest into a paper topic.

You can ask one of the other professors in the Department. Of course, you need to be polite. Make an appointment and realize that the professor is mostly interested in his or her own research. Something like this might work: "Oh mighty one, is there some small corner of your research on X that I could work on? Perhaps the data I collect may be of some small help in your great work."