CEA Policy Evaluation and Analysis 4 — Questions and Answers
Question 1: In difference-in-differences (DiD) estimation, the 'parallel trends' assumption requires that:
- Treatment and control groups have identical pre-treatment outcome levels
- In the absence of treatment, both groups would have followed the same trend (Correct answer)
- The treatment effect is constant across all subgroups in the sample
- Pre-treatment covariates are balanced between treated and control units
Correct answer: In the absence of treatment, both groups would have followed the same trend
Parallel trends assumes that the counterfactual change in outcomes for treated units equals the observed change in the control group over the same period.
Question 2: An analyst finds a statistically significant program effect with p < 0.05 but the confidence interval nearly spans zero. The most appropriate interpretation is:
- The program has a substantively large and reliable effect
- The effect may be statistically significant but is small or precisely zero (Correct answer)
- The result is spurious and the program has no effect
- Statistical significance alone confirms the policy should be expanded
Correct answer: The effect may be statistically significant but is small or precisely zero
With large samples, trivially small effects can be statistically significant; economic and practical significance must be assessed alongside p-values.
Question 3: Which of the following is an example of a 'deadweight loss' from a targeted subsidy program?
- Benefits paid to individuals who would have taken the subsidized action anyway (Correct answer)
- Program administrative costs that absorb part of the budget
- Taxes raised to fund the program distort labor supply decisions
- Transfer payments that redistribute income without affecting efficiency
Correct answer: Benefits paid to individuals who would have taken the subsidized action anyway
Infra-marginal beneficiaries who would have acted without the subsidy receive windfall transfers that represent deadweight loss relative to a perfectly targeted program.
Question 4: The 'Local Average Treatment Effect' (LATE) in instrumental variable estimation applies specifically to:
- The average effect across all units in the population
- The effect on units whose treatment status is changed by the instrument (compliers) (Correct answer)
- The effect on always-takers who receive treatment regardless of the instrument
- The effect estimated from a large representative national sample
Correct answer: The effect on units whose treatment status is changed by the instrument (compliers)
LATE identifies the treatment effect only for compliers — units induced to change treatment status by variation in the instrument — not for never-takers or always-takers.
Question 5: When a policy analyst applies a higher discount rate in a net present value calculation, what happens to the relative weight of future benefits?
- Future benefits are weighted more heavily relative to present costs
- Future benefits are discounted more steeply, reducing their present value (Correct answer)
- Future benefits and costs are affected equally, leaving NPV unchanged
- Future benefits increase in nominal terms to compensate for higher rates
Correct answer: Future benefits are discounted more steeply, reducing their present value
A higher discount rate reduces the present value of future cash flows more aggressively, making long-run benefits worth less in today's dollars.
Question 6: Which concept explains why government interventions may produce worse outcomes than markets even when market failures exist?
- Moral hazard from insurance-like program design
- Government failure, where political and bureaucratic incentives distort policy (Correct answer)
- The Coase theorem's prediction of efficient private bargaining
- Regulatory capture leading to under-enforcement of rules
Correct answer: Government failure, where political and bureaucratic incentives distort policy
Government failure theory holds that imperfect information, public choice dynamics, and bureaucratic incentives can cause policy interventions to create net social harm.
Question 7: A policy that imposes a $10 per-unit tax on a good with perfectly inelastic demand will result in:
- Equal sharing of the tax burden between producers and consumers
- The full tax burden falling on consumers with no change in quantity sold (Correct answer)
- Producers absorbing the full tax through lower prices they receive
- A reduction in equilibrium quantity proportional to the tax rate
Correct answer: The full tax burden falling on consumers with no change in quantity sold
With perfectly inelastic demand the quantity demanded does not change, so sellers can pass the entire tax to buyers as a higher price without losing sales.
CEA Policy Evaluation and Analysis 4
In difference-in-differences (DiD) estimation, the 'parallel trends' assumption requires that: