Commodity prices are of interest to investors, central banks and policymakers since they are believed to influence general price levels. Therefore, in this thesis I study whether it is possible to forecast commodities returns using economic indicators over different horizons and economic cycles. I establish an out-of-sample (OOS) predictability using different economic variables such as: inflation, unemployment rate, dividend price ratio, industrial production, among others. The time span of the analysis is from 1951 to 2014, over a monthly, quarterly an annual horizon. I observe that inflation is consistently a good predictor for in-sample (IS) and OOS univariate models. Multivariate OOS estimations tend to be more accurate when predicting commodity returns than univariate regressions. Furthermore, the unemployment rate and the commodity currencies are strong statistically significant predictors during economic recessions.
Date of Award | 21 Oct 2016 |
---|
Original language | English |
---|
Awarding Institution | - Universidade Católica Portuguesa
|
---|
Supervisor | Paul Ehling (Supervisor) |
---|
- Commodity price predictability
- Out-of-sample forecast
- Economic cycle
Economic variables today, returns tomorrow
Pimentel, M. L. C. D. S. (Student). 21 Oct 2016
Student thesis: Master's Thesis