Aumentar Tamaño del texto Disminuir Tamaño del texto

José Manuel Cueto, Universidad Carlos III de Madrid


Essays on multifactor models for stocks’ expected returns in European markets: A block-bootstrap validation proposal


In this thesis we study the efficiency of multifactor models, based on statistical and fi- nancial factors, for the pan-European Equity market. Since classical inferential techniques rely on several assumptions that are hardly fulfilled in real datasets, here we develop a block-bootstrap methodology to assess the validity of these models and the significance of the parameters involved.

Our research starts by introducing new factors that are built from statistical measure- ments on stock prices. Specifically, these statistical factors are the coefficient of variation, skewness, and kurtosis. Portfolios are built by means of the statistical factors and using data coming from Reuters, corresponding to nearly 2000 EU companies, and spanning from January 2008 to February 2018. Methods under assessment are time-series regres- sion, cross-sectional regression, and the Fama–MacBeth procedure. For each of the mod- els considered, we compare our bootstrap-based inferential results with those based on classical proposals (based on F-statistics). We found that a multifactor model including factors such as skewness and the coefficient of variation improves CAPM with regard to the adjusted R2 in the time-series regressions. Thus, a model including these two factors together with the market is thoroughly studied.

Next, we propose a multifactor model based on a combination of statistical and finan- cial factors, like Market Capitalization and Total Assets (as measures of size), Price to Book ratio (as a measure of cheapness), Return on Assets and Return on Equity (as mea- sures of profitability), and Momentum. In order to select the factors to be included in the model, we use a dimension-reduction technique designed to work with several groups of data called Common Principal Components. We decided to select the first four principal components (CPCs) since the average percentage of variability explained by them was 90%. Portfolios are built by means of these CPCs using data from Reuters, correspond- ing to nearly 1250 EU companies, and spanning from October 2009 to October 2019. Methods under assessment are time-series regression and cross-sectional regression and, as before, our bootstrap-based inferential results are compared with those obtained via classical testing proposals. The main findings are that these multifactor models improve the Capital Asset Pricing Model with regard to the adjusted-R2 in the time-series regres- sions, that Market and a factor related to Momentum and mean of stocks’ returns have positive risk premia for the analyzed period concerning Cross-section regression results, and that the tests based on block-bootstrap statistics are more conservative with the null than classical procedures.