EFFICIENCY IN LARGE DYNAMIC PANEL MODELS WITH COMMON FACTORS
研究了含共同不可观测因子的可交换非线性动态面板模型的渐近有效估计，推导了效率边界并提出了计算简单的有效估计量，适用于信用、公司债券或寿险合同的大规模组合风险预测和资本计算。
This paper deals with asymptotically efficient estimation in exchangeable nonlinear dynamic panel models with common unobservable factors. These models are relevant for applications to large portfolios of credits, corporate bonds, or life insurance contracts. For instance, the Asymptotic Risk Factor (ARF) model is recommended in the current regulation in Finance (Basel II and Basel III) and Insurance (Solvency II) for risk prediction and computation of the required capital. The specification accounts for both micro- and macrodynamics, induced by the lagged individual observations and the common stochastic factors, respectively. For large cross-sectional and time dimensions n and T , we derive the efficiency bound and introduce computationally simple efficient estimators for both the micro- and macroparameters. The results are based on an asymptotic expansion of the log-likelihood function in powers of 1/ n , and are linked to granularity theory. The results are illustrated with the stochastic migration model for credit risk analysis.