估計總體經濟動態效果的三篇論文

Abstract

本論文包含三個關於估計總體經濟動態效果的章節。第一章延伸Angrist et al. (2018) 提出的半母數 (semiparametric) 時間序列估計方法，應用於估計政策轉變對經濟結果分配特定分量的影響效果，定義分量政策效果 (quantile policy effects, QPE) 為在政策施行後的各期間，兩個潛在結果 (potential outcomes) 在特定分量的差距。應用QPE評估美國於1948至1993年實施之總體審慎政策效果，研究發現總體審慎政策對銀行信用成長存在異質的 (heterogeneous) 影響效果。\n\n第二章在動態追蹤資料的架構下，應用局部投影 (local projections) 方法估計制度轉換的動態影響，研究發現Nickell偏誤會影響局部投影的衝擊反應 (impulse response) 估計值，若將追蹤資料拆分成兩個不相重疊的樣本，再分別估計一階差分的局部投影模型，可大幅降低衝擊反應估計值的偏誤。應用上述方法進行實證研究發現，若國家自非民主政體轉變為民主政體，在20年後平均約使人均GDP提高24% - 25%。\n\n第三章檢視在景氣循環的不同時點，臺灣貨幣政策對實體經濟的影響效果是否有所不同。應用Romer and Romer (2004) 的認定方法，建構新的臺灣貨幣政策衝擊的時間序列，衝擊反應的估計結果顯示，據此方法認定之貨幣政策衝擊對實質產出、失業率、實質銀行放款有較大的影響效果；研究進而發現，在景氣熱絡時期，貨幣政策對實質產出、失業率、實質銀行放款的影響效果較景氣低迷時大。

This dissertation consists of three independent chapters on estimating the dynamic effects in macroeconomics. In the first chapter, I extend Angrist`s et al. (2018) flexible semiparametric time series methods for evaluating the quantile responses of outcomes to a policy change. The differences between certain quantiles of the potential outcomes at different horizons are defined as quantile policy effects (QPE). I apply the QPE estimator to examine the effectiveness of cyclical macroprudential policies on bank credit growth in the U.S. from 1948 to 1993. The estimated results of QPE show that the responses of bank credit growth to macroprudential actions at different quantiles are heterogeneous.\n\nIn the second chapter, I employ local projections method to estimate the effect of regime transitions in the dynamic panel data framework. I show that splitting the panel data into two non-overlapped subsamples and respectively estimating the first-difference local projections estimation can reduce the bias in the regime effect estimates contaminated by the Nickell bias and capture asymmetric and nonlinear effects of regime transitions. Applying this approach, I find that a transition to democracy causes the GDP per capita to increase by 24% - 25% over 20 years.\n\nIn the third chapter, I investigate the asymmetric impact of monetary policy over the course of Taiwan’s business cycle. Employing Romer and Romer’s (2004) identification approach, I construct a new measure of the monetary policy shock for Taiwan’s economy. The finding suggests find that the impact of monetary policy estimated by the new measure is larger and faster than short-term restrictions approach. Moreover, the impacts of monetary policy on real output, employment and real bank loans are less effective in recessions than in expansions.