Gold investment: An analysis of stocks, bonds and gold under non normal market conditions

Abstract

The aim of this thesis is to investigate how well gold can act in diversifying a portfolio consisting of bonds, stocks and gold. The thesis starts off with an investigation of the supply and demand side of gold. Next we investigate how the gold price is influenced by the past and present day gold price and stock price, as measured by the S&P 500, the inflation (CPI) and the interest rate measured by a 3 month t-bill. The test period is from January 1. 1986 till April 1. 2013. Using monthly data. The results show that the gold price today is influenced by the gold price yesterday, and the inflation. The stock price and the T- bill have no effect on the price of gold, meaning that gold is uncorrelated with these. In the next part of the thesis we investigate in detail the findings of the linear regression. For this period we use daily log returns, and instead of a three month t-bill, we use a one year t-bill, assuming these represent the interest rate equally. The three sub periods cover the years 2000 until 2003, 2004 until 2007 and 2008 until 2011. We start of by investigating the volatility, showing that a Student-t distributed GARCH (1, 1) model is the best fit for the data. The results show that the bond is the less volatile of the three assets. Comparing gold and S&P 500, gold is the less volatile of the two assets in the period from 2000 until 2003 and from 2008 until 2011. The period with high market growth, stocks has been less volatile then gold, but only slightly. The results indicate, as shown in the linear regression, that gold can prevent some extreme losses under highly volatile market conditions. Next we simulate a Student-t copula, to generate a joint distribution of the return series, taking into account that the distributions are non-normal. The correlation matrix generated shows some dependence between stocks and bonds. Gold on the other hand is uncorrelated with both stocks and bonds (and only slightly in the first period). This implies that the results from the linear regression, holds even at the non-normal and daily level. Based on the above analysis, the thesis concludes with a value at risk analysis, showing that the optimal portfolio that minimizes value at risk/ETL in all the sub periods contains gold. It also shows that an investor during the financial crisis could have substituted a portion of his or hers portfolio from bonds to gold, and realized an excess return, in contrast to a negative return in a portfolio consisting of only stocks and bonds. The thesis thereby concludes that gold can work as a portfolio diversifier during stress on the financial markets.