The effect of investor sentiment on non-ferrous metals contracts at LME and optimizing a metals commodity portfolio

Abstract

The history of the financial markets is full of occurrences contradicting the assumptions of the traditional theories. Recent studies on Behavioral Finance argue that some of these observed financial phenomena can be explained by the probability that some of the traders at markets are not fully rational. These studies argue that the heuristics, biases, and sentiments affect the decisions of investors, who possess scarce cognitive resources and limited information. Behavioral finance is an approach that arose as an alternative approach to the criticized aspects of the traditional theories. Behavioral Finance discusses that non-financial drives influence the investors and can explain their investment preferences. Many empirical literatures discuss the effect of investor sentiment in finance. As investor sentiment is an influencer in the price formation of assets, it should be used to model the prices, returns, and volatility in the market. As studies on behavioral finance accumulate, empirical evidence on the influence of investor sentiment on the dynamics of the market increases. This study aims to assess the effect of sentiment in commodity markets. In the commodities markets, the participants are a mix of both investors and direct purchasers of the commodity itself, for its industrial use. The changing nature of the commodity markets can be observed through the increasing interest of financial investors, especially in the futures market, who do not have any physical exposure to the commodities in question. The literature argues that cointegration is one important concept in empirical behavioral finance. The study argues that investor sentiment is contagious among various markets and the conveyors of the sentiments are the financial investors, who diversify their portfolio with multiple asset classes traded in different markets, including commodities. Because of their significant role in industrial manufacturing, non-ferrous metals are an important area of concern for economic activity worldwide. The studies on the price patterns of industrial metals should also focus on the behavior of investors in both physical and derivatives markets. This study argues the existence of interconnectedness of the London Metal Exchange (LME) metals by other global financial markets through the way of investor sentiment. As for the sentiment measure, this study chooses an indirect approach to use a market proxy and assumes the Chicago Board Options Exchange's Volatility Index (VIX), which is a recognized measure of volatility and investor sentiment. As the literature states that the most successful and practical model for multivariate modeling is the Vector Autoregression (VAR) this study employs VAR as the analyzing methodology. Harry Markowitz introduced an analytical construct of a connection between the investment's return and the risk, which is called the Mean Variance Portfolio optimization. In this methodology, either the risk should be minimum for a specified return or the return should be maximum for an accepted value of risk. Portfolio diversification reduces the exposure of investors to the individual risk of the assets. It is all about deciding on the best possible portfolio diversification strategy. In time, many sophisticated estimation techniques were developed, in hopes of providing a better way than the Mean Variance Portfolio optimization method. However, it still remains as the milestone of the Modern Portfolio Theory. The study employs Mean Variance Portfolio optimization methodology to create a metal commodity portfolio from the assets traded at the LME. The biggest disadvantage of a regular Mean Variance Portfolio is actually a long-position-only (long-only) portfolio, which ignores a whole bunch of securities with negative expected returns. Investors overcome this disadvantage by the method of short selling. With short selling, investors are able to borrow and sell an asset without actually having its ownership. Investors, who short sell, earn their profits when the prices decline. Therefore, the study reconstructs the metal commodity portfolio by allowing short selling. This way, the study increases the flexibility of its portfolio and it benefits from the losing assets as well as the increasing assets. A portfolio with short positions results in a better performance if compared to a portfolio with long positions only. The first aim of this study is to investigate the idea that the investor fear gauge is universal for all exchanges and markets, and after blossoming in one driver market, such as the equity market, it will have effects on all other markets in degrees, such as the commodity markets. One of the other aims of this study is to emphasize the signifcance of non-ferrous industrial metals for the global financial markets by validating that the metals market is not independent of the developments of other markets with different kinds of assets classes, on the contrary, they express interrelationships among each other in various ways. The final aim of this study is to use the LME non-ferrous metals to create an optimum portfolio of metals commodity by using the Markowitz's optimization methodology, withouth and with short selling. The study gathers data for Nickel, Tin, Aluminum, Copper, Lead, and Zinc, which are the six non-ferrous metals of the LME. The study employs the VAR methodology to analyze the cointegration between industrial metals market prices and VIX. Test results verify a long-run relationship among the prices of non-ferrous metals and the fear factor in the global financial markets. Furthermore, the study constructs three different metals commodity portfolios: i) Portfolio 1 – Equal Weight Portfolio: To calculate the return series from the prices of each non-ferrous metal series, logarithmic expression is used. The results show positive returns for Copper, Lead and Tin, whereas they show negative returns for Aluminum, Nickel and Zinc for the defined period. As each of six assets are given equal weight, the resulting expected return for this portfolio is negative due to the losing assets. ii) Portfolio 2 – Optimized Mean Variance Portfolio: This is a metals commodity portfolio constructed by using the optimization methodology. The study calculates the covariance matrix for metal series. The highest covariance is among Nickel and Tin and the lowest covariance is among Copper and Lead. Moreover, all the relationships between the two metals are positive, which indicates that all metals tend to behave in the same direction with other metals. Then the study uses the Sharpe Ratio to approximate the market portfolio. It employs standard deviation as a measurement of total risk. The portfolio which maximizes the Sharpe Ratio is the optimum portfolio among the six non-ferrous metals. It can be observed that the highest weight in the optimum portfolio belongs to Lead. Copper and Tin also enter into the equation due to their positive expected return; however, the optimum portfolio does not include any Aluminum, Nickel and Zinc. The expected return is slightly lower than the individual asset with highest daily return. But it is higher than the portfolio with equal weights. The standard deviation is lower than both the portfolio with equal weight and the weighted average of the individual standard deviations of the metals, due to the fact that there are covariances among all two sets of assets. iii) Portfolio 3 – Long-Short Portfolio: Finally, the study constructs a second portfolio with short selling. This time the portfolio has long positions for Copper, Lead and Tin, whereas it has short positions for Aluminum, Nickel and Zinc. The study shows that the highest long position in the optimum portfolio belongs to Copper and it is limited with the threshold value. Whereas the biggest short position belongs to Zinc and it is also limited with the threshold value. All industrial metals enter into the equation. The resulting long-short portfolio has a larger number of assets, which results in a more diverse structure than the long-only portfolio, a daily return, which is significantly higher than both the portfolio with long positions only and any/each one of the individual assets, a portfolio variance, which is significantly lower than both the long-only portfolio and any/each one of the individual assets, and a higher Sharpe Ratio than the portfolio with long positions only, which indicates a higher portfolio performance. The study employs the London Metal Exchange Metals Index (LMEX) as the market benchmark and compares all three portfolios with LMEX in the same period. Although LMEX has a negative expected return due to the losing three metals out of six, it still has a better expected return than the equal weight portfolio. However, both optimized Mean Variance Portfolio and the long-short portfolio perform much better than the market index. When the study compares the three portfolios among each other, it utilizes Sharpe Ratio (SR) and Treynor Ratio (TR). a) SR of Portfolio 2 is higher than Portfolio 1. SR of Portfolio 3 is significantly larger than Portfolio 2. This indicates that Portfolio 2 performs better than Portfolio 1 and Portfolio 3 is the best performing among the three. b) TR is negative for Portfolio 1, which indicates a worse performance than the market benchmark. By assessing TR, the study concludes that Portfolio 2 outperforms Portfolio 1. However, the TR is negative for Portfolio 3, which is not very surprising since Portfolio 3 transforms negative performing assets into positive. It is an expected outcome as the benchmark return is negative, it leads to a negative Treynor ratio, especially as the portfolio's return is positive but the covariance between the portfolio and the benchmark index is also negative.