RT Dissertation/Thesis
T1 Dynamic quantile causal inference and forecasting
A1 Ruzicka, Josef
AB Standard impulse response functions measure the average effect of a shock on a responsevariable. However, different parts of the distribution of the response variable may reactto the shock differently.The first chapter, “Quantile Structural Vector Autoregression”, introduces a frameworkto measure the dynamic causal effects of shocks on the entire distribution of responsevariables, not just on the mean. Various identification schemes are considered: shortrunand long-run restrictions, external instruments, and their combinations. Asymptoticdistribution of the estimators is established. Simulations show our method is robust toheavy tails. Empirical applications reveal causal effects that cannot be captured by thestandard approach. For example, the effect of oil price shock on GDP growth is statisticallysignificant only in the left part of GDP growth distribution, so a spike in oil price maycause a recession, but there is no evidence that a drop in oil price may cause an expansion.Another application reveals that real activity shocks reduce stock market volatility.The second chapter, “Quantile Local Projections: Identification, Smooth Estimation,and Inference”, is devoted to an increasingly popular method to capture heterogeneity ofimpulse response functions, namely to local projections estimated by quantile regression.We study their identification by short-run restrictions, long-run restrictions, and externalinstruments. To overcome their excessive volatility, we introduce two novel estimators:Smooth Quantile Projections (SQP) and Smooth Quantile Projections with Instruments(SQPI). The SQPI inference is valid under weak instruments. We propose information criteriafor optimal smoothing and apply the estimators to shocks in financial conditions andmonetary policy. We demonstrate that financial conditions affect the entire distributionof future GDP growth and not just its lower part as previously thought.The third chapter, “Smooth Quantile Projections in a Data-Rich Environment”, modifiesthe estimator from the second chapter to construct distribution forecasting in a settingwith potentially many variables. To this end we introduce a novel estimator, Smooth QuantileProjections with Lasso. The estimator involves two penalties, one controlling roughnessof the forecasts over forecast horizons, while the other penalty selects the most informativeset of predictors. We also introduce information criteria to guide the optimal choice ofthe two penalties and represent the problem as a linear program in standard form.
YR 2021
FD 2021-04
LK https://hdl.handle.net/10016/33112
UL https://hdl.handle.net/10016/33112
LA eng
NO I gratefully acknowledge funding from the Ministerio de Educación, Cultura y Deporte through its grant Formación de Profesorado Universitario (FPU).
DS e-Archivo
RD 10 may. 2024