Introduction
This paper provides an insight into the statistical relationships between the LME base metal complex and CME copper. We will look at the relative performance and detailed correlation metrics between these contracts, using data covering the period from January 2000 until June 2021.
Industrial metals are often considered as a potential vehicle to express a global macroeconomic view from an investment standpoint. This analysis will show that non-copper LME industrial metal contracts such as aluminium, lead, nickel, tin and zinc offer different return profiles and imperfect correlation to the copper contracts on both LME and CME.
Quantitative analysis of LME base metals versus CME copper
For our analysis, we have used the following datasets:
- For the LME, we have used the 3-month (3M) rolling forward contract for aluminium, copper, lead, nickel, tin and zinc.
- For the CME copper, we have used the rolling second nearest main delivery month of the following expiration months:, March, May, July, September and December.
Table 1 presents the mapping between the Bloomberg tickers, labels and the contract descriptions of each market that we analyse in this document.
Historical performance
We indexed all prices to 100 on 1 January 2000 and graphed the indexed performance in Figure 1. In addition, we also show an equally weighted portfolio of six LME base metals (“LME Avg”) that is rebalanced annually at the beginning of each calendar year. It is evident from the data that the LME Avg captures the overall performance of the LME metal markets with reduced volatility 1 of returns.
To add a bit more granularity to the analysis, we examined the annual percentage returns of the individual metals and LME Avg, ranked them by performance and then presented these in Figure 2. The data set covers 21 full calendar years from 2000 to 2020 and YTD returns until the end of June 2021. Both the top and the bottom performers change almost every year. The most frequent “winner” was LME Lead, taking the top spot six times, followed by LME Nickel (5 times) and LME tin (4 times). LME Copper had the strongest price performance of all the base metals only once in 2020. We present the full breakdown in Table 2.
As one would expect, the LME Avg never takes the top or bottom spot due to the nature of its construction.
Nevertheless, in nine out of the 21 full years, the LME Avg outperformed both the LME Copper and CME copper contracts while at the same time offering some protection from the potentially poor idiosyncratic performance of any individual metal.
Correlation analysis of LME base metals and CME copper
Next, we look at the correlation relationship between LME base metals and CME copper.
We calculated the Pearson correlation coefficient between random variables X and Y (PCCx,y), ππ ,π, using the daily percentage change data for each market.
The PCCx,y is a measure of linear correlation between two sets of data. The result of the PCCx,y is always between -1 (perfect negative correlation) and +1 (perfect positive correlation). The correlation of 0 implies a lack of a linear relationship.
The PCCx,y formula is as follows: ππ ,π= πππ£(π,π)/ππππ
Where:
πππ£(π,π) is the covariance between random variables X and Y
ππ is the standard deviation of X
ππ¦ is the standard deviation of Y
The PCCx,y was calculated for the following market pairs and the outcomes are tabulated below in Figure 3:
1. CME copper vs. LME Zinc
2. CME copper vs. LME Aluminium
3. CME copper vs. LME Nickel
4. CME copper vs. LME Copper
5. CME copper vs. LME Tin
6. CME copper vs. LME Lead
As the starting point, we calculated the one-year (260 days) correlations between the LME base metals complex and CME copper as of 30 June 2021 presented in Figure 3. At that date, the correlations between LME base metals 2 and CME copper ranged from 0.42 for the LME Lead to 0.60 for LME Nickel. Looking at LME base metals alone, the lowest correlation exists between LME Lead and LME Tin, with a correlation value of 0.27 as of 30 June 2021, indicating a weak linear relationship.
To provide more historical context, we also calculated the one-year (260 days) rolling correlation for LME base metals and CME copper, along with quarterly (65 days) correlations to illustrate the short-term behaviour.
As expected, the historical correlation estimates vary over time. This variability can be related to the mixture of estimation uncertainty3, positive relationship between correlations and volatility, in addition to genuine changes in market fundamentals. To keep this publication concise, we will focus on general observations relating to the base metal market correlations.
Historically, LME Tin has been the base metal with the lowest correlation with LME Copper, even exhibiting periods of negative correlation, when examining 65-day rolling figures. However, LME Tin recently lost its status as the least correlated base metal. Currently, the lowest correlation exists between the CME copper contract and LME Lead.
The quarterly correlation figures exhibit more variability than the annual estimates. This phenomenon is due to higher sampling variation, more pronounced responses to periods of increased volatility and idiosyncratic price moves due to the unique fundamentals of each metal market. One specific example is the negative correlation estimates observed primarily for tin in late 2019/early 2020. More recently, the values of the correlations of LME base metals versus CME copper have converged to a relatively narrow range.
Overall, although the correlation values are not stable, they provide a high-level insight into the relationship between various base metals. Over the time scales considered, all correlations are comfortably below 1, which means that they can provide a diversification benefit to a copper-centric portfolio.
Summary
In this article, we have presented historical trends for the price and return correlations between CME copper and LME base metals. It is apparent that, although all metals share many price drivers, the performance of each base metal varies year on year, with the top performer changing almost every year. The correlations between CME copper and LME base metals can range from low to high depending on the lookback period examined, but they always stay well below one. LME base metal pairs such as lead and tin exhibit lower correlations than non-copper LME base metals and copper.
Market participants, who are currently active only in the copper market could further evaluate the benefit from a broader base metal market exposure due to the imperfect correlation between the CME copper market and non-copper LME base metals. In addition, the variability of performance and price drivers due to the individual market fundamentals may present an opportunity to express a bullish or a bearish view in any particular metal or a set of metals.
1 From a portfolio diversification standpoint, a low historical volatility metal such as aluminium has had lower volatility over the period examined than LME Avg but also a much lower mean return.
2 Ignoring the correlation between LME Copper and CME copper, which is expected to be the highest.
3 Since the rolling correlation estimates are based on finite samples of 65 or 260 observations, there will be some variation in the correlation values calculated from different samples taken over time even if the distributions are unchanged.