Objective research to aid investing decisions
Value Allocations for July 2019 (Final)
Momentum Allocations for July 2019 (Final)
1st ETF 2nd ETF 3rd ETF

Personal Consumption Expenditures and the Stock Market

Posted in Economic Indicators

A reader, citing the book Ahead of the Curve by Joseph Ellis, inquired about the hypothesis that consumer spending drives economic cycles and is therefore a leading indicator for the stock market. For example, Mr. Ellis presents a chart showing annual change in Personal Consumption Expenditures (PCE) and annual change in S&P 500 operating earnings based on quarterly data. The chart also shows bear markets for U.S. stocks. The chart discussion states: “Most bear markets begin…when the Y/Y [year-over-year] rate of growth in consumer spending is peaking… Most bear markets then proceed as (the rate of growth in) consumer spending continues to slow, and are largely over by the time recessions…are under way. Importantly, most bear markets end…when consumer spending and S&P 500 profits are at, or even prior to, their worst Y/Y comparisons.” Does PCE usefully predict stock market behavior? The Bureau of Economic Analysis (BEA) releases seasonally adjusted, annualized total PCE monthly with a lag of about one month (Summary Table 2.85, Line 1: “Personal Income and Its Disposition, Monthly”). Using this series and monthly S&P 500 Index levels for January 1959 through August 2015 (680 months), we find that…

The following chart shows on logarithmic scales the behaviors of seasonally adjusted total PCE (left axis) and the S&P 500 Index (right axis) over the sample period . Both series generally rise over time, but the large mismatch in volatilities makes it very difficult to discern any tradable relationship between them.

For a closer look, we relate monthly changes in the two variables for different lead-lag scenarios.


The next chart plots Pearson correlations for various lead-lag scenarios between monthly change in PCE and monthly S&P 500 Index return, ranging from the stock market leads consumption by 12 months (-12) to consumption leads the stock market by 12 months (12) over the entire sample period and since 1990. Month 0 represents a coincident relationship. Results suggest that:

  • The stock market may lead PCE positively by a few months, with an advancing (declining) stock market stimulating (depressing) future consumption. This effect is elevated in the recent subsample.
  • PCE may modestly lead the stock market by a month, but any such effect is evident only in the recent subsample and is not exploitable due to release delay. The first exploitable relationship is month 2.

For deeper perspective, we focus on the first exploitable relationship.


The following scatter plot relates monthly S&P 500 Index return to monthly change in PCE two months ago over the entire sample period. The two-month lag accounts for the delay an investor experiences in acting on PCE releases. For example, BEA releases PCE for January at the end of February, so an investor reacting to the release could invest accordingly in March. The Pearson correlation between the two series is -0.05 and the R-squared statistic 0.003, indicating that changes in PCE explain practically none of the variation in S&P 500 Index returns two months hence.

To investigate non-linearities in the relationship, we consider average stock market return by range of ranked fifths (quintiles) of PCE.


The next chart shows average S&P 500 Index monthly returns by quintile of monthly changes in PCE two months ago over the entire sample period (135 observations per quintile), before 1990 (74 observations per quintile) and since 1990 (61 observations per quintile). The average monthly return for all months in the sample is 0.6%.

Results suggest that the lowest (most negative) changes in PCE indicate a relatively strong stock market during the month following release. A possible interpretation is that the stock market is rebounding after poor returns that preceded the worst PCE announcements. However, lack of systematic progressions across quintiles and inconsistencies between subperiods undermine belief in a reliable relationship.

Might quarterly data be more useful than monthly due to a cumulative effect?


The following scatter plot relates quarterly S&P 500 Index return to quarterly change in PCE, lagged one month to account for release delay, over the entire sample period (225 non-overlapping quarters). The Pearson correlation between the two series is -0.05 and the R-squared statistic 0.001, indicating that changes in PCE explain practically none of the variation in S&P 500 Index returns over the next actionable quarter.

Might quarterly changes in PCE reliably predict stock market behavior at some other horizon?


The final chart plots Pearson correlations for various lead-lag scenarios between quarterly change in PCE and quarterly S&P 500 Index return (with release delay), ranging from the stock market leads consumption by six quarters (-6) to consumption leads the stock market by six quarters (6) over the entire sample period, before 1990 and since 1990. Results generally confirm those above for monthly data, with a strong (weak) stock market tending to stimulate (depress) future consumption. There is no indication that change in PCE usefully predicts stock market behavior.


In summary, evidence from simple tests offers little support for belief that changes in Personal Consumption Expenditures usefully predict stock market returns at monthly and quarterly horizons. The strongest indication is that stock market returns are above average during the month after announcement of weak PCE growth.

Cautions regarding findings include:

  • Tests are largely in-sample. An investor using real-time data may (as suggested by subsample tests) draw different conclusions at different times.
  • The subsample since 1990 is small in terms of number of economic cycles.
  • Given the fairly weak findings, analysis does not include backtesting of any market timing strategies.
  • Retroactive adjustments to PCE data by BEA may confound measurement of the power of PCE to predict stock market returns.
  • This analysis does not rule out the possibility that surprises in PCE, relative to some measurable expectation, more usefully forecast stock market returns.
Why not subscribe to our premium content?
It costs less than a single trading commission. Learn more here.
Daily Email Updates
Research Categories
Recent Research
Popular Posts