Sieve Estimation of Time-Varying Panel Data Models With Latent Structures

• DOI: 10.1080/07350015.2017.1340299
• Publication Date: 2017-06-12T19:28:20Z
• Authors: Liangjun Su; Xia Wang; Sainan Jin
• Keywords: Sieve (category theory)

We propose a heterogeneous time-varying panel data model with latent group structure that allows the coefficients to vary over both individuals and time. assume change smoothly time form different unobserved groups. When treated as smooth functions of time, individual functional are across groups but homogeneous within group. penalized-sieve-estimation-based classifier-Lasso (C-Lasso) procedure identify individuals’ membership estimate group-specific in single step. The classification exhibits desirable property uniform consistency. C-Lasso estimators their post-Lasso versions achieve oracle so can be estimated well if were known. Several extensions discussed. Simulations demonstrate excellent finite sample performance approach estimation. apply our method study trending behavior GDP per capita 91 countries for period 1960–2012 find four