New Home Sales and Future Stock Returns
Each month the Census Bureau announces and the financial media report U.S. new home sales (seasonally adjusted and annualized) as a potential indicator of future stock market returns. Release date is near the end of the next month. Does this economic indicator convey useful information about future equity returns? To investigate, we relate future S&P 500 Index returns to new home sales over monthly, quarterly and annual intervals. Using monthly data for the S&P 500 Index and for seasonally adjusted annualized new homes sales over the period January 1963 through December 2011 (588 months or 49 years), we find that:
The following chart compares seasonally adjusted annualized new home sales and the S&P 500 Index (on a logarithmic scale) over the entire sample period. While the two series sometimes move together, visual inspection reveals no obvious systematic relationship.
For a closer look, we relate changes in the two series.
The following three scatter plots relate next-period S&P 500 Index return to change in new home sales at monthly, quarterly and annual intervals. Results suggest slight tendencies of stocks to advance (decline) after increases (decreases) in new home sales at monthly and quarterly intervals. However, with R-squared statistics of 0.01, variations in new home sales explain only about 1% of the variation in stock market returns. Also, since the Census Bureau releases new homes sales data with a nearly one-month delay, any tendency at a monthly interval is likely unexploitable.
Annual changes in new home sales exhibit practically no relationship to next-year stock market returns.
Might new home sales lead the stock market at some different interval?
The next chart summarizes Pearson correlations based on monthly data for scenarios ranging from S&P 500 Index returns lead changes in new home sales by 12 months (-12) to changes in new home sales lead stock market returns by 12 months (12) over the entire sample period. The strongest indication is that stocks lead new home sales by one month. Especially given the nearly one-month lag in announcements, new home sales appear not to be a useful leading indicator for stocks.
These regressions test for a linear relationship. Might there be a non-linear regularity between changes in new home sales and stock returns?
The next chart summarizes average next-period S&P 500 Index returns by quintile of changes in new home sales at monthly and quarterly intervals over the entire sample period. The number of observations per quintile is about 117 (39) for monthly (quarterly) intervals. While there is some indication that weak (strong) new home sales are bad (good) for stocks, lack of systematic progressions across quintiles undermines belief in reliable relationships.
Standard deviations of next-period stock returns also do not vary systematically by quintile for either monthly or quarterly intervals. For quarterly intervals, standard deviations are higher after weak new home sales than strong new home sales.
As a robustness test, we drill down by subperiods at quarterly intervals.
The final chart summarizes average next-quarter S&P 500 Index returns by quintile of quarterly change in new home sales over two equal subperiods. The number of observations per subperiod quintile is only 19. Lack of consistency across subperiods undermines belief in a reliable relationship, but there is a consistent indication that bad news on new home sales is bad for stocks over the next couple of months.
In summary, evidence from simple tests indicates that change in new home sales has little use as an indicator of future stock market returns. The most promising indication is that very weak new home sales indicate a weak stock market over the next couple of months.
Cautions regarding findings include:
- This analysis does not rule out the possibility that surprises in new home sales, relative to some measurable expectation, more usefully forecast stock market returns.
- Any trading of stocks based on new home sales data would involve trading frictions that work against successful exploitation.
- As noted, sample size in the subperiod quintile analysis is small.