Ensaios em ciclos econômicos e preços financeiros

• Ano de defesa: 2022
• Autor(a) principal: Caruso, Marco Antonio Jacob
• Orientador(a): Teles, Vladimir Kuhl
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Ciclos econômicos; Recessão; Previsão; Curvas ROC; Alocação de ativos; Política monetária; Não linearidade
• Palavras-chave em Inglês: Business Cycle; Recession; Forecasting; ROC curves; Asset allocation; Monetary policy; Machine Learning; Probit; Nonlinearity; Markov switching; Markowitz; Machine learning
• Link de acesso: https://hdl.handle.net/10438/32273

This thesis consists of three essays on the interrelationships between business cycles and financial prices. Through different empirical methodologies, they access nonlinear aspects of this mutual relationship. In the first, the predictability of recessions in the US is reassessed since the 1960s through the main variables used in the literature and at different lags. The performance of alternative models is evaluated and compared – inside and outside the sample – to the widely cited benchmark specification, which contains the slope of the yield curve The contribution of this article is in the sum of the use of the ROC – AUC analysis as a classification structure for the models and the higher predictive power of the suggested specifications 6, 12 and 24 months ahead. It is concluded that the ratio between true and false positives for recession forecasts is statistically higher when we include both variables linked to the corporate credit spread and to the housing cycle. The second essay examines (i) whether the behavior of a set of financial prices are capable of identifying future regime changes in Brazilian economic growth and inflation and, based on these forecasts, (ii) whether optimized portfolios that recognize these changes improve the risk-return compared to a buy-and-hold single portfolio optimized for the entire sample. Regarding the first objective, the main innovation was the due anticipation of four states for the Brazilian economy in the sample period. The two Markov regimes estimated for Central Bank Economic Activity Index (IBC-Br, a monthly GDP growth proxy) keep good adherence with the recessions and expansions suggested Brazilian business cycle dating committee (CODACE), while the two estimated regimes for inflation expectations anticipate periods of IPCA acceleration and deceleration. In view of the asset allocation literature with regime changes in Brazil, the main contribution was the incorporation of the two new states based on forecasts on inflation and the increase in the Sharpe Ratio in portfolios sensitive to the proposed regime changes. The last essay assesses whether the ability of monetary policy to act on the Brazilian business cycle is attenuated in times of high economic uncertainty. We consider four measures of uncertainty and estimate their interaction effects with monetary policy shocks: one based on the incidence of related terms in the news, one that adds to a newspaper-based proxy some measures of macroeconomic forecasts dispersion and two based on the volatility of financial prices. The base model suggests that innovations in the Selic rate have their effects on GDP and IPCA reduced in times of greater domestic economic noise, vis-à-vis states of lesser uncertainty. The conclusions are similar for the four uncertainty proxies and the results are robust to different specifications. In this environment, policymakers would rediscover the well-known trade-off between “gradualism” and “shock” therapy, with policy having to be more aggressive to bring inflation back to the target and stabilize the economy. The main contribution to the research with data for Brazil was to make the proposed uncertainty measures endogenous.