Swapping the Nested Fixed Point Algorithm: A Class of Estimators for Discrete Markov Decision Models
Swapping the Nested Fixed Point Algorithm: A Class of Estimators for Discrete Markov Decision Models
V Aguirregabiria, P Mira - Econometrica, 2002
å
容ã®åã«ãã²ã¨ã¤ã
ã¨ã«ããNotationãåãããããï¼æ¸ãæ¹ãåããããããã¾ã Dynamic Optimizationã®æ¨å®ç³»ã®è«æãèªãã§ãªã人ã¯ããããèªã¿å§ããã®ãããã¨æããç°¡åã«Rustã¨ãHMã¨ãã®è§£èª¬ãè¼ã£ã¦ãããããã³ãã«ããã¨æãã
Rustã®CIã®ä»®å®ãã©ããªæå³ã¨ããã¡ããã¨è§£èª¬ãã¦ããã¦ãã
ã§ããããã°çã«ã¯Rustã¨ãHMã®è¨äºãèªãã§ããèªãã ã»ããåãããããã§ãã
Introduction
Rustã§ã¯ModelãFullã«è¨ç®ãã¦ãï¼Choice Probabilityã®ï¼Likelihoodãæ大åãããããªæ¨å®æ¹æ³ãæ示ãã¦ãããããã«ãã£ã¦ãå¹ççãªæ¨å®å¤ãå¾ãããä¸æ¹ã§ãè¨ç®éãå¤ããªãã
ä¸æ¹ã§ãHotz and Millerã¯ããªã¥ã¼ãã¡ã³ã¯ã·ã§ã³ãå®éã«ã¯è§£ããªãã§ãChoice Probabilityã®Moment Conditionçãªãã®ãä½ã£ã¦æ¨å®ããæ¹æ³ãæ示ãããããã¯è¨ç®éãæ¸ãããã¨ã«ã¯æåããããFinite Sampleã§ã漸è¿çã«ãå¹çæ§ãç ç²ã«ãã¦ããã
ããã§ããã®è«æã§ã¯ãã®ä¸éã«ä½ç½®ãã¦ããã¤Rust並ã¿ã®å¹çæ§ãéæãããããªæ¹æ³ãæ示ãã¦ããã
Rust
Rustã§ã¯ã
ï¼ï¼ãã©ã¡ã¼ã¿ãä¸ãããããã¨ã§ããªã¥ã¼ãã¡ã³ã¯ã·ã§ã³ã解ã
ï¼ï¼ããªã¥ã¼ãã¡ã³ã¯ã·ã§ã³ããChice Probabilityãæ±ãã
ï¼ï¼Likelihoodãé«ãããããªãã©ã¡ã¼ã¿ã®å¤ãæ¢ã
ã£ã¦ããä»®å®ãçµãããã©ã¡ã¼ã¿æ¢ãã®ä¸ã«Fixed Pointã¨ãã¦ã®ããªã¥ã¼ãã¡ã³ã¯ã·ã§ã³ã解ãéç¨ãå
¥ã£ã¦ããã®ã§Nested Fixed Point Algorithm(NFXP)ã¨ããã
HM
ï¼ï¼ãã¼ã¿ããConditional Choice Probabilityãä¸ãããã
ï¼ï¼ï¼ã§å¾ã確çãå
ã«Value Functionã®å·®ãæ±ãã
ï¼ï¼ããããæé©ååé¡ã解ãã¦CCPãPredictãã
ï¼ï¼Predicted CCPã¨ï¼ã®CCPã®è·é¢ãæå°åãããããªãã©ã¡ã¼ã¿ãæ¢ã
ã£ã¦ããæããCCP Estimatorã£ã¦å¼ã°ãã¦ãã
AM
ã§ã¯ããã®è«æã¯ä½ããã¦ãã®ãï¼ãã£ãã°ããã«ããã¨ã両æ¹ã®ææ³ãçµã¿åããã¦ããã
HMçã«èããã¨ãé¸æ確çã¯ç¢ºçâVFâ確çã®ä¸åç¹ã«ãªã£ã¦ããã
ï¼ï¼ã¾ãé©å½ãªç¢ºçãGuessãã
ï¼ï¼ããããHMçãªç¢ºçã¸ã®Mappingãèãã
ï¼ï¼ä¸ã®Mappingã¯ãã©ã¡ã¿ã«ä¾åãã¦ããã®ã§ãï¼ä¸åã£ããã®Mappingã§å¾ãããï¼Likelihoodãæ大åãããããªãã©ã¡ã¿ãé¸ã¶
ã¤ã¾ããPseudo-likelihood
ãæ大ã«ãããããªÎãÎkã¨ãã¦ã¢ãããã¼ããã
ï¼ï¼ããã§å¾ãããæ°ããªCCPãPk+1ã¨ãã¦ã¢ãããã¼ããããã¤ã¾ããPkã¨ÎkããHMçãªMappingã§æ°ããªPãå¾ãã£ã¦ãã¨ã§ãã
ï¼ï¼ï¼ã«æ»ã£ã¦ãÎk+1ã¨Pk+1ãã¢ãããã¼ããã
ã£ã¦ããå·¥ç¨ãåæããã¾ã§ç¹°ãè¿ãã£ã¦ãããã®ã§ããå½¼ãã¯Nested Pseudo Likelihood Algorithm(NPL)ã£ã¦èªãã§ã¾ãã
ããã§æ±ããããæ¨å®å¤ãNFXPã®æ¨å®å¤ã¨åãã§ãããã¨ã示ãã¦ãã¾ããï¼With è¥å¹²ã®Conditionï¼
çµè«
ããæå³ãPolicy Function Iterationã¿ãããªæãã§ãããããæ®éã®åé¡ã§ãValue Function IterationããPolicy Function Iterationã®æ¹ãè¨ç®ãéãï¼ãçµæãåãï¼ã®ã§ãããã¨ä¼¼ããã®ãæãã¾ãã
èè
ãã¡ã¯ãã®å¾æ¼¸è¿æ£è¦æ§ã¨ãã示ãã¦ãã¾ãããã³ãã¯èªã¿ããã£ãããèªã¿ããããã§ã¯ãã£ãã®ã§ãããæéããªãã®ã§ã¹ããããã¡ãã£ã¨æãã¾ããã®ã§ãå¾æ¥èªãã§ããã«å ããããããã¾ããã