Estimating Treatment Effects of Monetary Policies and Macro-prudential Policies: From the Perspectives of Macro-economic Policy Evaluation

Since the global financial crisis in 2008, an increasing number of economists, central banks and regulators across the world has realized the occurrence of fundamental changes in the dynamics of the economy. The breakout of the global financial crisis highlights the importance of financial shocks. Aiming to maintaining financial stability, Bank for International Settlements (BIS) initialized macro-prudential policies in the early of 2009. China, as one of important countries pioneering the practice of macro-prudential policies, adopted a so called two-pillar regulatory framework of monetary policies and macro-prudential policies to safeguard the macroeconomic and financial stability. However, due to the coincidence of policy targets and the interdependence in transmission mechanisms between monetary policies and macro-prudential policies, the practice of the two-pillar regulatory framework raises some important coordination issues (Beau et al. 2012). The aim of this paper is to discuss theoretically the coordination mechanisms between monetary policies and macro-prudential policies, and then evaluate empirically the effects of the practice of the two-pillar regulatory framework on policy targets, such as economic growth, inflation, and financial stability in China. One of main contributions of this paper is to estimate the causal effects of China's two-pillar regulatory framework from 2007 to 2017 by adopting new macroeconomic policy evaluation methods proposed by Angrist and Kuersteiner (2011) and Angrist et al. (2018). Compared to mainstream methods such as dynamic stochastic general equilibrium (DSGE) models, the macroeconomic policy evaluation methods based on Rubin's causal model alleviate the risk of model misspecification by avoiding to specify how an economy works and how outcome variables are determined. Moreover, the concept of dynamic treatment effect developed in the framework of macroeconomic policy evaluation coincides with the nonlinear impulse function induced by structural models. In other words, the new method can complement the DSGE models by providing parallel estimates insensitive with structural model setup. Another major contribution of this paper is to extend the existing macroeconomic policy evaluation methods by adopting statistical learning methods to estimating policy propensity score functions using macroeconomic big data and proposing a new test statistic for testing the conditional unconfoundedness assumption. It is well known that a major challenge in empirical macroeconomic research is how to capture exogenous policy shocks to identify causal effects. We address this issue in two aspects. First, in order to fully use all information available at the current period, we propose to model policy-making process based on macroeconomic big data and adopt statistical learning methods to solve the high dimensional problem. Moreover, we propose a new test statistic to test the exogeneity of the residuals estimated from the policy propensity scores using macroeconomic big data, which is a conditional unconfoundedness in the context of time series data. The latter actually provides a testable method of evaluating the validity of the use of the macroeconomic policy evaluation method. Finally, our empirical findings can be summarized as follows. First, when macro-prudential policies remaining neutral, monetary policies can effectively manage the aggregate demand and fulfill the output target by adjusting money supply and credit growth, while the transmission channel through interest rates does not work effectively. Second, when monetary policies remaining neutral, macro-prudential policies can maintain financial stability as expected, and at the same time, there are little effects on real economy targets. Last, when monetary policies and macro-prudential policies are jointly implemented, a same direction policy combination can further strengthen the effect on the output target and accelerate the process towards the target. However, the same direction combination has no significant exaggerating impact on financial stability variables. In addition, we find that the same direction combination may cause counteracting effects on some target outcome variables, such as the growth rate of capital adequacy ratio and the risk-weighted asset ratio. We ascribe the counteracting effect to the argument that the same direction policy combination weakens the negative correlations between monetary policies and banks' risk-taking level.