WINTER 2007
LABOR ECONOMICS FIELD EXAM
Be as formal and precise as you can in all of your answers, using equations
(and well-labeled graphs, where appropriate). Your will be graded on the extent
to which you demonstrate your understanding of the problem, and the clarity
and precision of your answers.
ANSWER ALL THE QUESTIONS. YOU HAVE 3 HOURS
PART I
1. Many income support programs for low-income families can be described
as a guaranteed income grant of Gdollars, where earned income is “taxed
away” at rate τ, so that the total subsidy from the government is given
by max (0, G −τ Y ), where Yis earned income. Suppose a government
changes its existing program by increasing the tax rate τsignificantly. By
doing so, it expects that it will lower the “effective”/after-tax wage, and
therefore discourage work, causing many to drop out of the labor force.
Therefore, it expects that the total number of people participating in the
program to rise. Contrary to this expectation, they discover that as a
result of this change, the number of families who receive any subsidy falls
sharply. Explain carefully and precisely the leading explanations for this
“surprising” result.
2. Suppose you are hired as a consultant by a government that currently has
no income support program. They are, however, considering exactly the
kind of subsidy described in question 1: with guaranteed income level G0,
with earned income taxed away at rateτ0. They would like to know how
much the program will cost.
(a) Describe how you would produce this number if you were only given
CPS-type data on income, earnings, etc., and specify the assumptions
that need to hold (however potentially unrealistic) for the estimate
to be valid.
(b) Now suppose the government gives you the authority to run a so-
cial experiment. Given this ability, describe how you would run the
experiment, what data you would collect, and how you would ana-
lyze the data to better answer the question. In particular, describe
how you might try to separate costs due to “mechanical” effects from
those due to behavioral responses to the program, and compare the
magnitudes of these two components.
(c) Consider an alternative scenario: the government does not let you run
an experiment, because they have done so already... sort of. They tell
1
