Blog - Investing Notes
November 18, 2008 - Update: PPI and the Stock Market
In our Real Earnings Yield Model, we argue that inflation at the consumer level is fundamentally a wealth discount rate important in determining the value of equities to investors. Inflation at the producer level (derived from the Producer Price Index - PPI) is logically an advance indicator for inflation downstream at the consumer level (derived from the Consumer Price Index - CPI). Do investors therefore reliably react to changes in PPI as an indicator of the future wealth discount rate? Using monthly historical PPI data (for finished goods, seasonally adjusted) from the Bureau of Labor Statistics (BLS) and contemporaneous S&P 500 index data for the period January 1950 through September 2008 (705 months), we find that:
The following chart shows PPI and the S&P 500 index over the entire sample period. Obviously, both indexes tend to rise over time. Visual inspection reveals no systematic relationship between the two series.
For greater precision, we compare monthly changes in PPI to monthly S&P 500 index returns.
The following scatter plot relates the monthly S&P 500 index return to the monthly change in PPI over the entire sample period. Because BLS releases PPI data about the middle of the following month (for example, data for March in the middle of April), we offset the two data series to avoid look-ahead bias. In other words, the plot relates PPI for a given month with stock returns for the next month. There is a weak negative relationship between the the two series, with a Pearson correlation of -0.08, suggesting that stocks tend to advance (decline) as PPI falls (rises). However, the R-squared statistic is just 0.01, indicating that monthly PPI releases explain just 1% of contemporaneous monthly stock market returns. There is not much of a signal for monthly trading.
Might changes in PPI lead changes in stock prices?
The next chart shows the Pearson correlations for various lead-lag scenarios for monthly changes in PPI and monthly S&P 500 index returns over the entire sample period. Offsetting the two series such that changes in PPI lead and lag monthly S&P 500 index returns by 1-6 months produces mostly weak negative correlations. The strongest correlation (still weak at -0.10) occurs for zero lead-lag, meaning the change in PPI and S&P 500 index return for the same month. For this case, the stock market leads release of PPI by about two weeks.
What if the relationship is cumulative, significant not for monthly changes but perhaps for quarterly changes?
The next scatter plot compares the quarterly S&P 500 index return to the quarterly change in PPI for the entire sample period (235 quarters). There is a noticeable negative relationship between the the two series, with Pearson correlation is -0.22. The greater (more positive) the change in PPI, the lower the contemporaneous stock returns. The R-squared statistic is 0.05, indicating that quarterly PPI changes explain about 5% of contemporaneous quarterly stock market returns. The relationship is still too dispersed to motivate trading.
What about trading in the few days around releases of new PPI data?
The final chart shows the mean daily returns for the S&P 500 index in the three days before and three days after PPI release dates for 1994-present (176 events, excluding September 2001 due to interruption of stock trading), with one standard deviation error bars. BLS announces new PPI data before the stock market opens, so trading day 1 is the day of release. The mean daily return for all days during the sample period is 0.03%. The event pattern suggests that investors may have a tiny bit of excess worry just before release dates and a little bit more relief after release dates. However, the deviations from the mean return for all days are very small compared to daily variability.
Would the results be different if we selected months for which the change in PPI is very high or very low (negative)? No. The Pearson correlation between the three-day return after PPI release and the change in PPI is close to zero (-0.04), so any relief in evidence seems not to depend systematically on whether the change in PPI is up or down, or large or small. In any case, as noted, noise on average overwhelms signal during the event. Moreover, CPI releases often follow PPI releases within one or two days, perturbing any PPI signal.
In summary, PPI is not a reliable signal for either short-term or intermediate-term trading of the broad U.S. stock market.
Note that this analysis may contain a small amount of revision bias because BLS occasionally adjusts old data.
See Blog Synthesis: The Economy and the Stock Market for other research on relationships between macroeconomic indicators and stock market behavior.
To discuss this research, go to the Economic Indicators Forum.
