关键词: 经济周期, 混频数据, MS-MIDAS模型, 区制监测, 二元选择, 动态模型, 线性趋势, 偏误纠正, 条件Logit估计

Abstract: This article proposes bias correction estimators of MLE and approximate conditional logit estimators for dynamic binary choice models with heterogeneous linear trends. We obtain the analytical bias form of ordinary MLE and show how to estimate the bias. Small sample experiments show that approximate conditional likelihood function can eliminate the affects of heterogeneous parameters greatly, and the bias correction estimators can significantly reduce the bias of MLE. Finally these methods are applied to estimate the model of cash dividends payment.

Key words: business cycle, mixed-frequency data, MS-MIDAS model, regime monitoring, Binary Choice, Dynamic Models, Linear Trends, Bias Correction, Conditional Logit Estimation