Nonparametric identification of dynamic models with unobserved state variables
- Creators
- Hu, Yingyao
-
Shum, Matthew
Abstract
We consider the identification of a Markov process {W_t, X^(*)_t} when only {W_t} is observed. In structural dynamic models, Wt includes the choice variables and observed state variables of an optimizing agent, while X^(*)_t denotes time-varying serially correlated unobserved state variables (or agent-specific unobserved heterogeneity). In the non-stationary case, we show that the Markov law of motion View the MathML source is identified from five periods of data W_(t+1), W_t,W_(t−1), W_(t−2), W_(t−3). In the stationary case, only four observations W_(t+1), W_t,W_(t−1), W_(t−2) are required. Identification of f_(W_t, X^(*)_t|W_(t-1), X^(*)_(t-1)) is a crucial input in methodologies for estimating Markovian dynamic models based on the "conditional-choice-probability (CCP)" approach pioneered by Hotz and Miller.
Additional Information
© 2012 Elsevier B. V. Received 27 November 2010; Received in revised form 29 September 2011; Accepted 30 May 2012; Available online 15 June 2012. We thank Xiaohong Chen, Jeremy Fox, Han Hong, Ariel Pakes, and Susanne Schennach for their suggestions. Seminar participants at Berkeley-Haas, BU, Clark, UC-Davis, Harvard, LSE, MIT, NYU, Penn, Penn State, Toulouse, UCL, UCLA, USC, the Cowles 2008 Summer Conference at Yale, the 2008 ERID Conference at Duke, the 2008 Greater New York Econometrics Colloquium at Princeton, the "Econometrics of Industrial Organization" workshop at Toulouse, SITE 2009, and the 2010 "Five Star Forum" at Renmin University's Hanqing Institute provided useful comments.Additional details
- Eprint ID
- 35634
- Resolver ID
- CaltechAUTHORS:20121126-103400456
- Created
-
2012-11-26Created from EPrint's datestamp field
- Updated
-
2021-11-09Created from EPrint's last_modified field