Research

This paper introduces  the family of Dynamic Kernel models. These models approximate the predictive density function of a time series through a weighted average  of kernel densities possessing a dynamic bandwidth. A general specification is presented and several particular models are  studied in details.   We propose an M-estimator for model parameters and derive its asymptotic properties  under a misspecified setting. A consistent density estimator  also introduced. Monte Carlo results show that the new models effectively track the time-varying distribution of several data generating processes. Dynamic Kernel models outperform extant kernel-based  approaches  in tracking the predictive distribution of GDP growth.

A novel dynamic model for joint estimation of multiple quantiles of a time series conditionally on a set of covariates is presented. The model preserves quantile monotonicity and allows for a clear interpretation of covariate effects across quantiles. Model parameters are estimated using a two-step M-estimator. The resulting estimator is consistent, and its finite sample properties are analysed through simulations. The new model is used to study the impact of different levels of stress in the financial system  on GDP growth rate. The analysis shows that worsened financial conditions imply a more pessimistic economic outlook when the financial scenario is already severely distressed, and an overall increased macroeconomic uncertainty. Additionally, past information on GDP growth is found to be critical in studying and predicting economic vulnerability.  These findings hold true even when alternative measures of real economic activity are considered.

We propose a new rank-based test for the number of common primitive shocks, q, in large panel data. After estimating a VAR(1) model on r static factors extracted by principal component analysis, we estimate the number of common primitive shocks by testing the rank of the VAR residuals' covariance matrix. The new test is based on the asymptotic distribution of the sum of the smallest r-q eigenvalues of the residuals' covariance matrix. We develop both plug-in and bootstrap versions of this eigenvalue-based test. The eigenvectors associated to the q largest eigenvalues allow us to construct an easy-to-implement estimator of the common primitive shocks. We illustrate our testing and estimation procedures with applications to panels of macroeconomic variables and individual stocks' volatilities. 

We introduce a randomized testing procedure to test the null hypothesis that there are no pricing errors in a panel of asset returns - that is, the null of zero alpha. The distinct features of the proposed methodology are that it does not require the estimation of any covariance matrix and it allows for both N and T to grow large, with the former possibly faster than the latter. Further, unlike extant approaches, the procedure can accommodate conditional heteroskedasticity, non-Gaussianity, and even strong cross-sectional dependence among asset returns. We also propose a de-randomized decision rule to choose in favor or against the correct specification of a linear factor pricing model. Monte Carlo simulations show that the test has satisfactory properties and it compares favorably to several existing tests. The usefulness of the testing procedure is illustrated through an application of linear factor pricing models to price the constituents of the S&P 500.

We introduce a new class of order-preserving linear transformations. These operators map vectors whose entries are sorted in ascending order into equally ordered ones. Starting from this result, we develop a new model to track the quantiles of a time series given all past information on itself and on a set of explanatory variables. This model preserves monotonicity of the estimated quantiles over time. An empirical analysis shows that the new method effectively captures the relation between macroeconomic and financial tail risk. The model also delivers competitive density and tail risk predictions.