Research

Central banks across the globe introduced large-scale asset purchase programs to address the unprecedented circumstances experienced during the pandemic. Many of these programs were announced as open-ended in order to shock-and-awe market participants and restore confidence in financial markets. This paper examines whether these whatever-it-takes announcements had larger effects than traditional policy announcements with explicit limits on scale. We use a narrative approach to categorize the announcements made by twenty-two central banks in advanced and emerging economies during the pandemic. Using event study, propensity-score-matching, and local projection methods, we measure the short-term effects of policy announcements on exchange rates and sovereign bond yields. We find that on average a central bank's first whatever-it-takes announcement lowers 10-year bond yields by an additional 25 basis points relative to size-limited announcements, suggesting that communication of potential policy scale matters. Announcement impacts on both exchange rates and yields differ across advanced and emerging economies.

We estimate the responsiveness of net labor migration to regional differences in unemployment rates across the United States since the mid-1970s. Our baseline estimate suggests an elasticity of roughly -0.3. For typical labor force participation ratios, an increase of 100 unemployed workers in an area is associated with net out-migration of roughly 47 workers. Instrumenting for regional unemployment produces even higher estimates. Our estimates are stable over time, inclusive of the Great Recession. The estimates depend crucially on accurate data and accounting for long-term trends in migration and unemployment.

Safe assets are a cornerstone of the modern international financial system, but the set of global safe assets can change over time. I present a model that helps explain the sources of time-varying sovereign bond safety and provides a definition of safety that can be taken to the data. Following the takeaways from the model, I construct a text-based index from newspaper mentions, the FLY, to measure global demand for safe assets. Compared to other measures of uncertainty and volatility, the FLY picks up relevant flight-to-safety episodes and the global savings glut, and is priced in the cross-section of sovereign bond returns. Additionally, innovations in the FLY foreshadow declines in the natural interest rate. I then estimate time-varying loadings on the FLY for a large set of sovereign bonds. I identify switches in a bond’s safety through changes in the sign and significance of its loading on the FLY. Safety switches are associated with sizable movements in macroeconomic variables: positive switches (i.e. becoming safe) are associated with expansions, increases in government spending, and higher debt; conversely, negative switches (i.e. becoming risky) are associated with contractions, decreases in government spending, and lower debt with a shorter maturity structure. The results are driven by advanced economies in the post-global- savings-glut period, they are not fully accounted for by credit ratings, and they hold up even when excluding the global financial crisis, the euro debt crisis, and the Covid-19 period. 

We examine the responsiveness of labor participation, unemployment and labor migration to exogenous variations in labor demand. Our empirical approach considers four instruments for regional labor demand commonly used in the literature. Empirically, we find that labor migration is a significant margin of adjustment for all our instruments. Following an increase in regional labor demand, the initial increase in employment is accounted for mainly through a reduction in unemployment. Over time however, net labor in-migration becomes the dominant factor contributing to increased regional employment. After 5 years, roughly 60 percent of the increase in employment is explained by the change in population.  Responses of labor migration are strongest for individuals aged 20-35. Based on historical data back to the 1950s, we find no evidence of a decline in the elasticity of migration to changes in employment.