Is copper price a reliable leading indicator of economic activity and therefore of future corporate earnings and equity prices? To investigate, we employ the monthly price index for copper base scrap (not seasonally adjusted) from the U.S. Bureau of Labor Statistics, which spans multiple economic expansions and contractions. Using monthly levels of the copper scrap price index and the S&P 500 Index during January 1957 through November 2020 (nearly 64 years), *we find that:*

The following chart compares behaviors of the copper scrap price index and the S&P 500 Index on logarithmic scales over the full sample period. Visual inspection suggests the series sometimes track and sometimes do not. There is no obvious consistent relationship between them.

For a closer look, we compare monthly change in the copper scrap price index to next-month stock market return.

The following scatter plot relates next-month S&P 500 Index return to monthly change in copper scrap price index (released about the middle of the next month) over the full sample period. The Pearson correlation for the two series is 0.03 and the R-squared statistic 0.001, indicating practically no linear relationship between the two series.

Might the relationship between monthly change in the copper scrap price index and stock market return be stronger over a forecast interval longer than a month?

The next chart plots Pearson correlations for various lead-lag relationships between monthly change in the copper scrap price index and monthly S&P 500 Index return since 1957, since 1990 and since 2000, ranging from stock market return leads copper scrap price index change by 12 months (-12) to copper scrap price index change leads stock market return by 12 months (12). Notable points are:

- There is a consistent tendency for the stock market to lead copper price change by one month, with the tendency strengthening over time.
- There is some tendency in recent data for copper price change to lead stock market return by up to three months emerges, but this tendency is absent for the full sample.

To check for non-linearity in the relationship, we calculate average next-month stock market return by range of monthly changes in the copper scrap price index.

The next chart summarizes average next-month S&P 500 Index returns by ranked fifth (quintile) of monthly changes in the copper scrap price index over the full sample period and the subperiod since 2000. There are 153 (50) observation for each quintile during the full sample period (recent subperiod). Notable points are:

- The stock market is much stronger on average after the biggest jumps in copper price than after the biggest drops, especially for the recent subperiod.
- However, average stock market return variation across quintiles is not systematic. Stocks are on average about as strong for the middle quintile as for the highest quintile, and stocks are on average weak for the fourth quintile. These results undermine belief in a reliable relationship.

Based on the above lead-lag chart, does aggregating changes in the copper scrap price index over quarterly intervals enhance predictive power?

The following scatter plot relates next-quarter S&P 500 Index return to quarterly change in the copper scrap price index over the full sample period (255 quarters) and since the beginning of 2000 (83 quarters). Notable points are:

- Over the full sample period, the Pearson correlation for the two series is -0.01 and the R-squared statistic 0.00, indicating no relationship.
- Since the beginning of 2000, the Pearson correlation for the two series is 0.19 and the R-squared statistic 0.03, indicating that quarterly variation in copper scrap price index change explains about 3% of the variation in next-quarter stock return. However, excluding the outlier at the lower left (fourth quarter of 2008) and the outlier at the upper right (second quarter of 2009), the R-squared statistic is 0.00. In other words, the financial crisis appears to contain all the predictive power of copper price changes.

In summary, *evidence from simple tests offers little support for belief that changes in copper price exploitably predict U.S. stock market returns at monthly and quarterly horizons, except possibly during an extreme stock market crash/rebound.*

Cautions regarding findings include:

- The above analyses are in-sample. An investor operating in real time would not have data for the full sample period and may therefore have different views of what constitutes large jumps or drops in the price of copper.
- Return variability makes the 2000-2020 subsample small for reliable inference (especially given the large effect of the financial crisis).
- The above analyses do not account for delay in releases of copper scrap price index data after the end of the month. This delay would interfere with exploitation.
- The copper scrap price index data used above is as-revised (“subject to revision four months after original publication”) rather than as-released. Effects of past revisions on statistical relationships with market returns could be material, especially since economic/market crashes drive predictability.
- Copper futures prices, while available immediately, embed investor expectations and may exhibit different behaviors from spot copper.