Disclaimer: I haven't thought about these ideas carefully, so they may not be good ideas. Or they may have been done (note the date next to each idea). Or they may not be doable. Or boring, or fundamentally mistaken, or any of the other things that tend to derail research projects.

Don't take the ideas as given. Use them to get started thinking about a set of questions.

Some general tips for getting started with a dissertation are here.

Mainstream Macro

Cohort size and asset prices (10/2001)

In the data, price-earnings ratio are closely related to the fraction of young households in the population. Geneakoplos, Magill and Quinzii (2001) ask whether this is consistent with rational expectations in spite of the fact that it implies substantial predictability of stock returns. Their model is a 3 period olg endowment economy. They find that the model is consistent with the data.

It would be interesting to ask whether this finding survives in a world with capital accumulation and shorter period lengths. When old agents realize that young cohorts are getting smaller, they should decumulate capital (stock buybacks; higher dividends), which might substantially weaken stock market predictability. Does it matter that a small fraction of households hold nearly all financial wealth?

Does income inequality lead to consumption inequality? (10/2001)

Krueger and Perri (2001) point out a striking observation: since the 1980s income inequality has increased a lot, but consumption inequality has not. They propose an explanation along the lines of Kehoe and Levine: more income risk losens the IR constraints in credit markets, borrowing constraints are relaxed, and consumption inequality remains low relative to income inequality.

It would be interesting to look at the data more closely. What happens to the retired who should not be affected by changing income inequality? What happens to the wealth rich for whom borrowing constraints should not bind? To what extent did consumption inequality change within groups (education/age classes, for example)? Could one figure out from panel data how households finance consumption in low income states? How did that change since 1980?

It would also be interesting to think of alternative explanations or extensions of the theory. An OLG model should generate implications for consumption inequality of young versus old households. Are these consistent with data?

The Welfare Costs of Public Debt (10/2000)

As far as I can see, little is known about the welfare costs of public debt. Is it very costly to deviate from the optimal debt level? (Aiyagari and McGrattan look at that question in their JME paper). This seems like a question of first-order importance that could be answered with fairly standard tools. Of course, I might simply not be aware of existing results (if so, please send me references).

A paper related to the question is Bullard and Russel's working paper on the welfare cost of inflation. They find large welfare costs of inflation, but the same would seem to hold for other policies involving changes in public debt.

Fiscal Policies in an Open Economy (4/2000)

Mendoza (1998 AER) studies fiscal policies in an open economy. The approach is straightforward: Write down a standard growth model and compute the solution.

What happens to these results if there is human capital and the economy is populated by overlapping generations instead of infinitely lived dynasties?

Welfare costs of business cycles (6/99)

Theory suggests that people should not care about recessions -- the welfare costs are tiny. Some observations suggest that this is not true empirically. For example, elections are sometimes lost because of recent recessions.

How can we get people to care enough? What are our models missing?

Household Behavior

Data on intergenerational transfers (6/99)

Intergenerational transfers are poorly understood. One reason: existing studies look at one type of transfer in isolation (bequests, inter vivos transfers). The trouble is: that is only the tip of the iceberg. Financial transfers are probably much smaller than non-financial ones.

It may be possible to develop a dataset from the PSID that is more comprehensive. The 1988 supplement contains detailed data on financial transfers. By matching children with their parents' records, one could measure education expenditures and coresidence.

Question: What determines the total level of transfers? What determines the composition?

This would be a messy project, but with a potentially very large payoff.

A related question: Do children who work hard or run businesses receive larger transfers? This could be evidence for paternalistic behavior.

Human capital

Why did schooling stop rising in 1950? (10/08)

U.S. educational attainment rose smoothly from at least 1900 to 1950. Then it stopped rising. Why?

Taxation with risky human capital (6/01)

A large literature studies how human capital responds to taxation. Almost all of this literature abstracts from most uncertainty (about earnings etc.). It seems obvious that this abstraction should lead to overstated tax effects (use the intuition from portfolio choice). It would be useful to quantify how large that bias is. This would require embedding realistic earnings uncertainty into a standard life-cycle model with schooling and job-training. It would also be nice to have some analytical results about the effect of uncertainty on tax elasticities in simpler models.

A word of caution: There are a couple of papers looking at human capital under uncertainty, though it seems nobody has looked at this particular question. This needs to be checked, though.

Why is schooling lumpy? (12/07)

In virtually all theories of human capital, years of schooling are a continuous choice. In the data, the choice is lumpy (high school graduate or dropout). Why is schooling lumpy? Why is there a big (presumably permanent) earnings penalty for dropping out of college after 3 years with a good GPA? What can be learned about how human capital is produced and valued in the market?

Does this have to do with the fact that grades are a noisy signal of ability? Why is grade completion a better signal?

Macro-Labor

Measuring the return to experience (8/08)

Experience as it is typically measured (age - school - 6) is collinear with birth year and date. This makes it impossible to identify the returns to experience. As wages grow over the life-cycle, it is impossible to tell whether this is due to rising cohort quality, wages, or experience.

One could use a dataset that has actual measures of experience to solve that problem. The idea would be to observe people of the same birth cohort at the same date with different experience. But the variation in experience needs to be instrumented. One possibility: use birth month. Those who are born too late in the year may enter school a year later than those who are born a bit earlier in the year. Perhaps one could use the NLSY for this.

Rethinking labor supply (6/99)

It seems that lots of people want to work (why else would you be a grad student?). There is plenty of informal evidence. What implications does that have?

Growth/Development

Why do democracies make bad policy choices? (2/09)

It seems puzzling that voters persistently approve bad policy proposals (farm subsidies, industry protection, etc). Could one consistently tell the following story: Voters are rationally uninformed. They know about proposals that affect them a lot (why?) but not about others. Then the winning policy plan loads up on pork barrel spending and other goodies that lower welfare for the majority of voters.

This raises the interesting (unexplored?) question: how does information get provided to voters? Since they are not interested in acquiring it, information must be free. What kinds of biases does this introduce into the political process. Are institutions for info dissemination important for outcomes?

Which industries are productive in LDCs? (7/99)

The variance of productivities is very high in LDCs. Perhaps we could learn something about the obstacles to development by studying which industries are highly productive in LDCs. How are the findings related to predictions of standard models?

Is a standard human capital model consistent with large industry TFP differentials?

It looks like Herrendorf and Valentiny may have an answer to this question.

Informal firms (2/09)

La Porta and Shleifer (2008) present some interesting evidence on the differences between informal and formal sector firms (that are registered with a central government authority). They push for an interpretation of their findings. But the connection between data (correlations) and interpretation is tenuous.

It could be interesting to write down a model that nests the competing hypotheses about why informal firms exist. Could one distinguish the hypotheses with the available data? What are the welfare implications of proposals to promote or discourage informal sector firms?

International Migration

What Do People Embody? (8/00)

Chris Carroll (1999 Econ Dev & Cult Change) asks: are differences in saving rates across countries due to "culture"? To answer the question, he compares saving rates of U.S. immigrants with source country saving rates. There is no correlation. He concludes that culture is not the reason for cross-country differences.

The argument about "culture" seems to be a bit of a misnomer. But the idea is sensible: immigrant data might allow us to identify which characteristics (propensity to save, to work hard, ...) are embodied in people and which characteristics arise as a response of fundamentally identical people to different environments.

One example: Women's labor force participation differs a lot across countries and over time. Comparing immigrants and natives might shed light on the extent to which, for example, the increase in U.S. labor force participation reflects a change in social norms as opposed to a change in relative prices. See my paper on hours worked on this.