Working Papers

• Intergenerational Income Mobility: A multivariate distribution regression approach
  with J.Meier, B.Melly & C.Naguib
  Abstract: We study intergenerational income mobility in Switzerland using multivariate distribution regression. We estimate the joint cumulative distribution of children's and parents' income controlling for different explanatory variables. Derivation of specific mobility measures from the joint cumulative distribution is straightforward, and taking covariates into account during the estimation of the joint distribution enables us to gain deeper insights into drivers of intergenerational income mobility. Further, this innovative approach allows us to decompose structural from compositional differences. Results focus on mobility differences between sons and daughters. Specifically, we find that a higher father's income share for sons is related to higher upward mobility. However, this relationship does not hold for daughters. We further analyze differences in mobility outcomes between sons and daughters and conduct a decomposition analysis. The decomposition shows that average hours worked are crucial in explaining the differences between sons and daughters. Nevertheless, a large part of these observed differences remains unexplained.
  
  
• A Synthetic Control Method for the Analysis of Effects across the Distribution
  Abstract: This paper extends the synthetic control method to evaluate distributional effects. Synthetic control methods are commonly employed for policy interventions on an aggregate unit level, where the treated and control units typically comprise a sizable number of individual entities. If interest lies in a variable measured at the individual level, there is the opportunity to analyze the effects across the distribution and uncover heterogeneous treatment effects. The proposed synthetic control method introduces a novel approach for deriving such distributional effects. Crucially, the weights of the synthetic unit depend on the position within the distribution, such that the weighted sum of control units is allowed to vary across the distribution. Furthermore, the proposed method modifies how individual values are aggregated, enabling the usage of well-established estimation procedures from the synthetic control literature. The proposed method is applied to analyze the impact of the introduction of the minimum wage in the canton of Neuchâtel in Switzerland. Results suggest a positive effect on labor income at the lower end of the distribution for individuals working in Neuchâtel. Furthermore, four distributional synthetic control methods are compared in the application as well as simulations. Results show an improved fit of the potential distributions if not treated between the synthetic and treated units if weights are allowed to vary across the distribution.
  
  
• Using Natural Language Processing to Identify Monetary Policy Shocks
  with A.Piller & L.Schwaller
  Abstract: High-frequency changes in federal funds futures prices around Federal Open Market Committee announcements have become an important tool for identifying the effects of monetary policy on macroeconomic variables. However, a number of recent studies have pointed out that these measures of monetary policy surprises suffer from endogeneity issues when employed to instrument monetary policy shocks. Using state-of-the-art text analysis methods, we create a new monetary policy surprise series that filters out the changes in federal funds futures (FFF) prices that are solely due to the corresponding FOMC statement and not due to other confounding factors. When identifying the impulse responses to a monetary policy shock based on our monetary policy surprises, we obtain two key results. First, the estimated effects on the macroeconomy are substantially larger than when using prevalent monetary policy surprises. Second, the persistence of the two-year Treasury yield response is considerably lower.