PhD Research
Job Market Paper
Forecasting Macroeconomic Variables: A Systematic Comparison of Machine Learning Methods with Qian Li
Abstract:
This paper evaluates the performance of an extensive set of machine learning algorithms in forecasting macroeconomic variables relative to baseline econometric models. We conduct a pseudo-out-of-sample forecast for fifteen real, nominal, and financial variables. The findings can be summarized in three points. First, machine learning models perform better than the benchmark model in forecasting real variables but worse than the baseline models in forecasting nominal variables (price indices) and financial variables. Second, machine learning models forecast better than benchmark models during periods of high volatility, like recessions and the COVID-19 pandemic. Third, models that employ dimension reduction frequently appear in the top five most accurate models in forecasting real variables, especially at longer horizons.
Working Paper
The Role of Oil Price Shocks in Shaping Unemployment Dynamics
Abstract:
We utilize local projections to investigate the impact of structural oil price shocks on unemployment rates and spells across the United States, emphasizing both national and state-level variations. Oil supply shocks lead to long-run increases in the national unemployment rate, incidence, and short-term unemployment. In contrast, economic activity shocks reduce all unemployment rates and spells, especially in oil-producing states. Consumption demand shocks have minimal impact on unemployment rates and durations, while inventory demand shocks show only temporary effects on durations.
Work In Progress
Revisiting the Response of Monetary Policy to Oil Supply Shocks
Abstract:
The paper uses local projections to investigate the macroeconomic and monetary policy responses to adverse oil supply shocks. The Federal Reserve raises interest rates twice: on impact and ten months after the shock to counter ongoing high inflation. A net oil exporter, Canada raises interest rates sharply in response to the shock to counter inflation. Switzerland initially maintains steady interest rates to prevent Swiss Franc appreciation, followed by gradual rate increases to manage inflation as the exchange rate stabilizes. Despite these efforts, inflation remains high in Switzerland.
Pre-PhD Research
Working Paper
Firms of a Feather Merge Together: Cultural Proximity and M&A Outcomes with Manaswini Bhalla, Manisha Goel and Michelle Zemel
Abstract:
Using data from India, we show that shared caste identities between two firms’ directors increase the likelihood of entering a merger and acquisition (M&A) deal. This may indicate directors’ reliance on caste as an informal information channel. But it may also be driven by their agency or overestimating synergies, leading to suboptimal deals. Indeed, we find that caste-proximate M&A deals create less value than caste-distant deals for both acquirer and target. The negotiation process and long-run performance also do not improve. Evidence strongly supports the presence of agency costs but not information gains or overestimation of synergies.
Presented by co-authors at Early Career Women in Finance 2018, IIM Calcutta-NYU Stern India Research Conference 2018 at NYU Stern, at the NYU-NSE 2018 conference at the National Stock Exchange, Mumbai, at UC, Irvine, and University of Washington, Seattle. Presented by me at the Indian Institute of Management - Bangalore, and at the 14th Annual Conference on Economic Growth and Development, Indian Statistical Institute, Delhi Center.