Working papers:

Forecasting Bilateral Refugee Flows with High-dimensional Data and Machine Learning Techniques [ link ] with Andre Groeger, Tobias Heidland, Finja Krueger and Conghan Zheng (R&R at Journal of Economic Geography)

We develop monthly refugee flow forecasting models for 150 origin countries to the EU27, using machine learning and high-dimensional data, including digital trace data from Google Trends. Comparing different models and forecasting horizons and validating them out-of-sample, we find that an ensemble forecast combining Random Forest and Extreme Gradient Boosting algorithms consistently outperforms for forecast horizons between 3 to 12 months. For large refugee flow corridors, this holds in a parsimonious model exclusively based on Google Trends variables, which has the advantage of close-to-real-time availability. We provide practical recommendations about how our approach can enable ahead-of-period refugee forecasting applications. 

The Effects of Monetary Policy Shocks on Risk [ link ]

I study how monetary policy affects the predicted distributions of GDP growth and inflation in the US, in particular whether it can alleviate tail risks. These risks are captured as the time-varying conditional 5th (downside risk) and 95th (upside risk) quantiles of the forecasted distributions of the variables of interest. I find that contractionary shocks shift the expected distributions of both variables to the left, in line with established results, but exacerbate tail risks and create additional modes in the distribution which suggest increased probabilities of very bad growth outcomes and policy ineffectiveness vis-à-vis inflation. The bimodality is not present for expansionary shocks. In addition, in response to the shock, the distributions are skewed and inflation uncertainty actually increases in response to an attempt by the authority to control it. The results recall the idea of multiple equilibria resulting from weak policy responses to increases in inflation, but suggest that multiplicity may be related only to contractions. The findings are derived using a new model setup which combines quantile regression with a dynamic factor model.

Other work in progress:

Adaptation and time-varying effects of climate shocks on the United States with Alessandra Testa

We use a time-varying coefficient SVAR woth stochastic volatility to assess whether temperature shocks in the US have different effects over time. To do this we investigate if the shocks have become larger or if the transmission has changed, possibly due to adaptation. In particular, we construct a new time series of patents related to climate adaptation and mitigation to capture technological adaptation effects.