Structural Estimation

1
Structural Estimation

• Course Applied Econometrics
• Lecturer Zhigang Li

2
Econometric Policy Evaluation

• The goals of econometric policy evaluation are
• To consider the causal impact of policy
  interventions on the economy
• To compute their consequences for economic
  welfare
• To forecast the effects of new policies
• Fit or Interpret?

3
Evolution of Causal Inference Fit or Interpret?
(Heckman, 2000)

• The Cowles Commission Tradition (1940s)
• Variant 1 (1970s) Still structural but smaller
• Variant 2 (1970s) VAR Time Series
• Variant 3 (1980s) Calibration Movement
• Variant 4 (1980s) Nonparametric approach and
  sensitivity analysis
• Variant 5 (1980s) Natural experiment

4
The Cowles Commission Tradition

• Correlation does not imply causation, so
  knowing correlation is not enough for future
  policy design.
• Intellectual success
• The concepts of exogenous (external) and
  endogenous (internal) variables
• The notions of structural parameters
  (policy-invariant parameters suggested by
  economic theory)
• The problem of identification (many theories may
  be consistent with the same data).
• Empirical failure
• Large scale structural models, which is perceived
  as an empirical failure due to unacceptable
  identifying (or simplifying) assumptions.

5
New Structural Models Simpler, More Precise, and
Dynamic

• Based on dynamic general equilibrium theory under
  uncertainty with more precisely defined
  structural (or causal parameters).
• E.g. Mincers female labor supply model
• Euler equation estimation method
• Exploit new sources of micro (panel) data (It is
  argued that time series variation was too limited
  to recover structural parameters)
• Findings not well accepted empirically

6
The VAR Time Series ApproachLess Theory and
More Data Description

• Advocate loosely specified economic time-series
  models to describe more accurately macro data
• Findings hard to interpret and models too
  arbitrary

7
Calibration

• Unlike the VAR econometricians, the calibrators
  focus on a few main correlations and means and
  build explicit models to account for them.
• More sophisticated economic models and and less
  rigorous econometric estimates of parameters
• Reject fit as a goal of empirical economic
  models and emphasize interpretability

8
Nonparametric and Sensitivity Analysis

• View that functional forms and distributional
  assumptions in conventional structural approaches
  as a major source of their lack of credibility.
• Identify structural parameters not using explicit
  functional forms, or examine the sensitivity of
  estimates to different identifying assumptions.

9
Natural Experiment

• Do not impose arbitrary structure onto the data
• Intuitive economic theory and less emphasis on
  structural parameters (in contrast to Structural
  School)
• Simple data summary (in contrast to VAR School)
• Low computation costs (in contrast to Structural
  and Nonparametric School)
• Stand Alone Feature The absence of structural
  frameworks makes it difficult to cumulate
  knowledge across studies. Hard to make
  counterfactual policy predictions.

10
Dynamics of Automobile Purchases(Eberly 1994)

• Question Why households are slow to adjust their
  (durable goods) consumption bundles.
• Two theories
• Transaction costs
• Liquidity constraint

11
How to test the theories I?

• If the transaction costs theory is right, the
  households should
• Purchase their durable stock with its value equal
  to a fixed fraction (or the target share) of its
  wealth
• Allow the durable stock to depreciate until it
  reaches a critical share of wealth
• Repurchase a new durable good so that the stock
  again equals the target share wealth
• If the optimal consumption rule is followed,
• ?lnKiß0ß1?Demanderror

12
How to test the theories II?

• If the liquidity constraint theory is right, then
  the model should be
• ?lnKiß0ß1?Demandß2Yierror
• Here Y measures liquidity constraints.

13
Findings

• Liquidity or transaction costs?
• For the constrained group, liquidity has a
  substantial and statistically significant
  negative effect on the consumption profile.
• For the liquid group, liquidity has a small
  negative effect.
• Empirical findings indicates that at adjustment
  about 13 percent of total wealth should be held
  in durables. This would be allowed to drift down
  to about 12 percent before adjusting again.

14
Price Discrimination in Broadway Theater (Leslie,
2004)

• Broadway theater refers to all plays and
  musicals performed in theaters in the Times
  Square region of Manhattan, New York City, with
  seating capacities in excess of 499.
• The theoretical framework is a utility-based
  model of consumer behavior incorporating viewer
  characteristics and institutional details of the
  Broadway theater industry.
• Data consist of price and quantity sold for all
  17 ticket categories for all 199 performances of
  Seven Guitars.

15
Behavioral Model

• Consumers are presented with a menu of different
  ticket options for seeing a play, or not.
• Each individual is characterized by income yi and
  taste ?i.
• The net utility of individual i from product j is

16
Findings I Structural Parameters

• Perceived quality difference in seats
• Median-quality seat is 70 more valuable than
  low-quality seat
• High-quality seat is over three times more
  valuable than low-quality seat
• ql1, ql1.69, ql3.33

17
Findings II Counterfactual Experiments and
Welfare Analysis

• Revenue Utility Attendance
• Benchmark 6.27 3.59 907
• Uniform 8.02 3.60 810
• No Booth 6.73 3.58 873