Thoughts on US Monetary Policy.
Mar 2024 - Jun 2024
This is a group project on US policy, supervised by Professor Zhiyuan Li at Fudan University..
We proposed possible improvements for the sufficient statistics approach by Barnichon and Mesters (2023), and evaluated alternative predictive data, revealing more accurate forecasts during high inflation periods in the U.S.. We also analyzed the relationship between sufficient statistics methods and real-world monetary policy, examining the balance between policy rules and situational flexibility, and identifying practical limitations in policy application.
I delivered a ninety-minute presentation and was responsible for all theoretical sections.
Introduction: Our group presented the article "A Sufficient Statistics Approach for Macro Policy" by Barnichon and Mesters (2023), published in the American Economic Review. This paper proposes a practical new paradigm for evaluating the effectiveness of policies, specifically within the realm of macroeconomic policy and, more precisely, monetary policy. In our presentation, we did not delve into the detailed model specifications or derivations of the formulas but instead highlighted the core ideas, analyzed the interaction of this paper with previous literature, and pointed out potential shortcomings. Generally, the sufficient statistics approach simplifies the evaluation process of monetary policy by relying on limited statistics and bypassing possible misspecification issues in econometric methods, thereby enhancing the credibility of the results. However, the simplification makes the model more sensitive to its assumptions, and the selection and estimation of statistics significantly impact the outcomes. Furthermore, the model only allows for single-period policy evaluation, neglecting the impact of policy changes on future periods and thus failing to assess policy over time. In other words, policy adjustments between periods are treated as independent. We aim to address this issue in our working paper.
Our analysis began with the model's assumptions. Specifically, we questioned the use of a forward-looking loss function in the sufficient statistics model, which assigns equal weights to the dual objectives of inflation and unemployment. We conducted a thorough literature review on how the Federal Reserve sets the weights in its loss function. Also, the original paper used data from the Federal Open Market Committee's (FOMC) Summary of Economic Projections to construct the forecast path of key statistics. We questioned whether better predictive data might exist that could yield improved policy adjustment values, thereby revealing the extent of bias in the Federal Reserve's official forecasts. We collected data on inflation and unemployment from the Michigan Survey of Consumers (MSC) and the Survey of Professional Forecasters (SPF) by the Federal Reserve Bank of Philadelphia and recalculated the estimates. Our findings indicate that individual forecasts of inflation were more accurate during the high inflation period in the U.S. starting in 2021, leading to better policy adjustments.
Building on the literature review, data collection, and conceptual development, we modified the sufficient statistics model proposed by Barnichon and Mesters (2023) and used the revised model to study the extent to which the high inflation in the U.S. from 2021 to 2023 was caused by Federal Reserve policy lags and forecast deviations. Specifically, we found that approximately 19.1% of the inflation gap could be attributed to policy errors by the Federal Reserve, and about 7.6% to forecast biases. Moreover, the later the optimal policy is adopted, the harder it is to control inflation. The original version of our working paper is attached at the end of this report.
Finally, we transitioned from theoretical research to the real world, focusing on the process of formulating monetary policy and further analyzing the relationship between the sufficient statistics method and real-world decision-making. First, we reviewed the historical development of Federal Reserve monetary policy, exploring the balance between policy rules and timely adjustments. The real economic world is complex, and scientifically reasonable rules and frameworks can enhance decision-making reliability while maintaining some flexibility. Additionally, we conducted a preliminary exploration of the sufficient statistics method's applicability, considering the trade-offs between dual objectives and the differences between large and small country models, pointing out practical limitations in its application.
Click here for the report. (in Chinese)