Essays on financial and housing wealth effects on consumption and the role of consumption-wealth ratio on stock returns predictability

Abstract

This thesis consists of three essays. The first and the second essays are related to the study of the wealth effects on consumption, while the third one studies how a proxy of the consumption-wealth ratio is able to predict excess stock returns. The first essay, investigated in the second chapter, studies the long-run effects on consumption of financial and housing wealth in Italy and the UK, using quarterly data over the period 1972q4-2012q4, and two different methods of estimation. It also attempts to evaluate how financial and housing wealth effects evolved over the sample period via rolling exercises. The empirical results show that: i) total wealth effect on consumption is larger in the UK than Italy; ii) housing wealth plays no role in Italy, while it is significant in the UK; and iii) in both countries, financial wealth exerts a positive and significant impact on consumption of about the same magnitude. As for the dynamics of wealth effects, the related results show that while in Italy the housing wealth effect is insignificant over time, in the UK this kind of effect is relatively increasing over large part of the sample. Further, financial wealth effects in the two countries feature opposite trends over time: slightly increasing in Italy and declining in the UK. The second essay, investigated in the third chapter, examines the long-run financial and housing wealth effects on consumption using panel annual data over the period 1970-2012 for 14 OECD countries. It applies recently developed nonstationary panel methodologies that assume cross-section dependence through common factor models. The analysis is repeated for two groups of bank-based and market-based economies. This essay offers three main results. First, both housing and financial wealth exert a positive and significant impact on aggregate consumption. Second, the housing wealth effect is shown to be larger in magnitude than the financial wealth effect for the sample of all countries as well as for the two groups of bank-based and market-based economies. Third, wealth effects tend to be higher in market-based economies than bank-based ones. The third essay, investigated in the fourth chapter, examines the predictive ability of a macroeconomic indicator, denoted “This thesis consists of three essays. The first and the second essays are related to the study of the wealth effects on consumption, while the third one studies how a proxy of the consumption-wealth ratio is able to predict excess stock returns. The first essay, investigated in the second chapter, studies the long-run effects on consumption of financial and housing wealth in Italy and the UK, using quarterly data over the period 1972q4-2012q4, and two different methods of estimation. It also attempts to evaluate how financial and housing wealth effects evolved over the sample period via rolling exercises. The empirical results show that: i) total wealth effect on consumption is larger in the UK than Italy; ii) housing wealth plays no role in Italy, while it is significant in the UK; and iii) in both countries, financial wealth exerts a positive and significant impact on consumption of about the same magnitude. As for the dynamics of wealth effects, the related results show that while in Italy the housing wealth effect is insignificant over time, in the UK this kind of effect is relatively increasing over large part of the sample. Further, financial wealth effects in the two countries feature opposite trends over time: slightly increasing in Italy and declining in the UK. The second essay, investigated in the third chapter, examines the long-run financial and housing wealth effects on consumption using panel annual data over the period 1970-2012 for 14 OECD countries. It applies recently developed nonstationary panel methodologies that assume cross-section dependence through common factor models. The analysis is repeated for two groups of bank-based and market-based economies. This essay offers three main results. First, both housing and financial wealth exert a positive and significant impact on aggregate consumption. Second, the housing wealth effect is shown to be larger in magnitude than the financial wealth effect for the sample of all countries as well as for the two groups of bank-based and market-based economies. Third, wealth effects tend to be higher in market-based economies than bank-based ones. The third essay, investigated in the fourth chapter, examines the predictive ability of a macroeconomic indicator, denoted “cayit”, for excess stock returns in a panel setting of 9 Euro countries, using quarterly data over the period 1988q1-2014q4. This indicator, regarded as a proxy for the logarithm of the consumption-wealth ratio, is the series of the residuals from an estimated long-run relationship between consumption, asset wealth and disposable income. The empirical analysis first focuses on the estimation of the "cayit" series using a panel cointegration approach, which controls for cross-sectional dependence via a common factor structure. Afterwards, the analysis aims to estimate panel regressions to forecast excess stock returns using "cayit" as a sole predictor, and along with other predictors. The empirical results point to predictability of future excess stock returns for the panel of 9 Euro countries, both in-sample and out-of-sample. Notably, in-sample results reveal that: i) "cayit" affects positively and significantly future excess returns over each horizon ranging from 1 to 8 quarters; ii) its forecasting power increases over horizons. As for the out-of-sample predictions, results highlight that a model with "cayit" outperforms two benchmark models: the constant expected returns benchmark and the autoregressive benchmark. Moreover, in line with in-sample results, the model that includes "cayit" improves over horizons compared to the two benchmarks. ”, for excess stock returns in a panel setting of 9 Euro countries, using quarterly data over the period 1988q1-2014q4. This indicator, regarded as a proxy for the logarithm of the consumption-wealth ratio, is the series of the residuals from an estimated long-run relationship between consumption, asset wealth and disposable income. The empirical analysis first focuses on the estimation of the series using a panel cointegration approach, which controls for cross-sectional dependence via a common factor structure. Afterwards, the analysis aims to estimate panel regressions to forecast excess stock returns using as a sole predictor, and along with other predictors. The empirical results point to predictability of future excess stock returns for the panel of 9 Euro countries, both in-sample and out-of-sample. Notably, in-sample results reveal that: i) affects positively and significantly future excess returns over each horizon ranging from 1 to 8 quarters; ii) its forecasting power increases over horizons. As for the out-of-sample predictions, results highlight that a model with outperforms two benchmark models: the constant expected returns benchmark and the autoregressive benchmark. Moreover, in line with in-sample results, the model that includes improves over horizons compared to the two benchmarks.