Abstracts Category : Other

Macroeconomic analysis and forecasting: An empirical investigation

by Caballero Carlos Cuerpo

This dissertation analyzes topics of special relevance for policy makers in the aftermath of the financial crisis. In a first step, the focus lies in the estimation of the slack of the economy in pseudo-real time according to the accumulation of selected macroeconomic imbalances. The analysis presents a novel approach putting the focus on the specification of the model rather than on the prior selection of the methodology itself. Multivariate methods, coupled with Kalman filtering are considered as an adequate modeling choice reaching a compromise between three criteria that are difficult to reconcile a priori; statistical fit, economic soundness and replicability. The approach is illustrated with its application to the Spanish economy, by selecting the best model amongst bivariate combinations of Gross Domestic Product (GDP) and 52 accompanying variables. In a second step, this essay assesses the out-of-sample forecasting performance of structural and non-structural models with quarterly data covering the last 37 years for seven macroeconomic aggregates: GDP, private consumption, private investment, employment or total hours worked, the GDP deflator, real wages and the nominal interest rate. The forecasting performance is assessed using a recursive procedure through four different dimensions: a time dimension (from one to eight quarters ahead), a contextual dimension (smooth growth period and recession phase), a country-specific dimension (results for Spain, USA and the euro area) and a model-specific dimension (comparison of structural general equilibrium models against traditional benchmarks such as Vector Autoregressive or VAR models and Bayesian VARs). Finally, the third step aims at calibrating the relative importance of the channels for the international transmission of shocks. To optimally weight the relative strength of the existing interlinkages between countries, the analysis first circumscribes the transmission of shocks to three encompassing channels; namely trade flows, banking exposures and contagion via agents' perception (reflected in the co-movement of sovereign yields). Then the optimal weighting scheme is obtained within a Global VAR framework, by minimizing the short-term GDP forecast error of the model. Once the relative weights of the channels are calibrated, they are used together with the actual bilateral flows to construct a weighted indicator reflecting the potential for spillovers between countries. Depending on the reference country, this indicator yieldsthe inward (which countries are relatively more important for a specific economy, and to what extent) as well as the outward (which countries are more dependent on the evolution of a selected economy) spillover potential for a countryAdvisors/Committee Members: Poncela Blanco, Pilar (dir.), UAM. Departamento de Anlisis Econmico: Economa Cuantitativa.