Forecasting gold returns using principal component analysis from a large number of predictors
(2023) DABN01 20231Department of Economics
Department of Statistics
- Abstract
- Gold is known in the financial world to be an important asset in unstable periods, especially as a hedge against inflation. If the gold price can be forecasted, it will be possible to strategically invest in gold rather than acquire it as a last-minute hedge against economic downturns. Although there are many studies on forecasting, few focus directly on gold returns.
This study investigates the role of economic variables in predicting gold returns, followed by a twelve-month forecast to determine its future returns. By principal component analysis, a large number of predictors are extracted down to seven factors, these are; the business cycle, nominal, interest rate, commodity, exchange rate, stock price, and government bond yield... (More) - Gold is known in the financial world to be an important asset in unstable periods, especially as a hedge against inflation. If the gold price can be forecasted, it will be possible to strategically invest in gold rather than acquire it as a last-minute hedge against economic downturns. Although there are many studies on forecasting, few focus directly on gold returns.
This study investigates the role of economic variables in predicting gold returns, followed by a twelve-month forecast to determine its future returns. By principal component analysis, a large number of predictors are extracted down to seven factors, these are; the business cycle, nominal, interest rate, commodity, exchange rate, stock price, and government bond yield factor. The ARMA model is used to predict gold returns with these factors and two additional variables, the Kansas City Fed's financial stress index, and the U.S. economic policy uncertainty index. The available dataset contains data from January 2000 through December 2019. With an in-sample period from January 2000 to December 2009 and an out-of-sample from January 2010 to December 2019.
Three alternative predictive models are compared to evaluate the forecast performance. The original model included all variables, the AR (1) benchmark model, and the model excluding the government bond yield factor. The results from the mean squared forecasted errors showed that the forecasting models containing predictors extracted from the principal component analysis outperformed the benchmark model in forecasting. The forecast for the twelve months preceding the data period showed solely positive returns. Regarding predictive power, the interest rate factor contributed to increasing gold returns. On the contrary, the business cycle factor, nominal factor, and stock market factor all tend to have a negative effect on the return of gold. However, the other variables showed insignificant results; therefore, the evidence was not strong enough to draw additional conclusions. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9117812
- author
- Allgén, Fredrik LU
- supervisor
- organization
- course
- DABN01 20231
- year
- 2023
- type
- H1 - Master's Degree (One Year)
- subject
- keywords
- Forecasting, PCA, Gold, ARMA
- language
- English
- id
- 9117812
- date added to LUP
- 2023-11-21 12:53:17
- date last changed
- 2023-11-21 12:53:17
@misc{9117812,
abstract = {{Gold is known in the financial world to be an important asset in unstable periods, especially as a hedge against inflation. If the gold price can be forecasted, it will be possible to strategically invest in gold rather than acquire it as a last-minute hedge against economic downturns. Although there are many studies on forecasting, few focus directly on gold returns.
This study investigates the role of economic variables in predicting gold returns, followed by a twelve-month forecast to determine its future returns. By principal component analysis, a large number of predictors are extracted down to seven factors, these are; the business cycle, nominal, interest rate, commodity, exchange rate, stock price, and government bond yield factor. The ARMA model is used to predict gold returns with these factors and two additional variables, the Kansas City Fed's financial stress index, and the U.S. economic policy uncertainty index. The available dataset contains data from January 2000 through December 2019. With an in-sample period from January 2000 to December 2009 and an out-of-sample from January 2010 to December 2019.
Three alternative predictive models are compared to evaluate the forecast performance. The original model included all variables, the AR (1) benchmark model, and the model excluding the government bond yield factor. The results from the mean squared forecasted errors showed that the forecasting models containing predictors extracted from the principal component analysis outperformed the benchmark model in forecasting. The forecast for the twelve months preceding the data period showed solely positive returns. Regarding predictive power, the interest rate factor contributed to increasing gold returns. On the contrary, the business cycle factor, nominal factor, and stock market factor all tend to have a negative effect on the return of gold. However, the other variables showed insignificant results; therefore, the evidence was not strong enough to draw additional conclusions.}},
author = {{Allgén, Fredrik}},
language = {{eng}},
note = {{Student Paper}},
title = {{Forecasting gold returns using principal component analysis from a large number of predictors}},
year = {{2023}},
}