Volatility models and the multivariate links with applications to the economy of Mozambique

Abstract

Mozambique, like any other country, needs to know the behavior of the main macroeconomic variables to make better decisions. This study aims to investigate the behavior as well as the relationship between macroeconomic variables and financial time series related to Mozambique. For the study, we use univariate Autoregressive conditional heteroskedasticity models, Vector autoregressive models and multivariate Generalized autoregressive conditional heteroskedasticity models. Overall, the study concludes that: (i) Asymmetry of shocks, volatility and currency-specific behavior affect economic performance, particularly in an open economy, influencing international capital movement and goods and services transactions, (ii) Mozambique’s real Gross domestic product plays an important role from cointegration relationships, impulse response functions to forecast error variance decomposition. (iii) The analysis of co-volatility showed the existence of relationships in volatility between different markets which influences the systematic behavior presented by the variables over time. These results contribute and reinforce the existing literature in terms of the choice of the appropriate model, criteria and tests presented that considerably affect the type of results. With this analysis of macroeconomic series related to Mozambique’s economy, it somewhat helps the process of policy making that increases Mozambique’s sustainable economic growth rate. In this thesis the problem of the normality of the log-returns for stock prices is also addressed for different formulations of price returns, namely intra-day and inter-day log-returns, with and without data trimming and for a large set of companies stock prices.