Does Aggregate Technological Innovation Predict Stock Returns?

Does a surge in innovation predict a surge in equity value, or instead creative destruction of equity value, over the next few years? To explore this question, assuming patent applications need not be approved to be exploited, we examine relationships between the growth rates of U.S. patent applications/patents as simple measures of innovation and U.S. stock market returns. Using U.S. patent activity (numbers of applications and patents) by calendar year for 1866-2009 as available and contemporaneous annual levels of  Shiller’s S&P Composite Index for 1871-2009  (139 years), we find that:

Per the U.S. Patent and Trademark Office:

“There are three types of patents:

• Utility patents may be granted to anyone who invents or discovers any new and useful process, machine, article of manufacture, or composition of matter, or any new and useful improvement thereof
• Design patents may be granted to anyone who invents a new, original, and ornamental design for an article of manufacture; and
• Plant patents may be granted to anyone who invents or discovers and asexually reproduces any distinct and new variety of plant.”

We consider only utility and design applications and patents.

For illustration, the following graph compares the behaviors of the S&P 500 Composite Index and utility patent counts (log scales) over the sample period. Both series tend to increase over time, stocks more rapidly than patents, but visual inspection indicates no other obvious relationship.

For greater precision, we compare some statistics.

The following table compares the average annual changes and the standard deviations of annual changes for various patent activities and the S&P Composite Index during 1871-2009. It also shows the same-year Pearson correlations between changes in patent activities and changes in the stock index.

Results confirm that patent activities have on average grown more slowly than stock valuations, with lower or comparable volatilities. Correlations are small, indicating R-squared statistics of 0.01 or less and therefore explanatory powers of 1% or less between patent activity and stock market returns.

Might changes in patent activities lead or lag stock market returns?

The next chart shows the Pearson correlations for various lead-lag scenarios between annual changes in patent activities and annual changes in the S&P Composite Index, ranging from stocks lead patent activities by five years (-5) to patent activities lead stocks by five years (5) over the entire sample period. Results suggest that:

• Lead-lag relationships with stock market returns are more systematic for applications than patents.
• While correlations are mostly small, stocks tend to lead patent activities positively by one to three years, and patent activities tend to lead stocks negatively by one to three years.

Potential explanations are that (1) more (less) funding tends to be available for innovation during strong (weak) equity markets and (2) innovation tends to produce some destruction of market value as non-adaptation temporarily outpaces adaptation.

Note that considering different measures of aggregate technological innovation and different lead-lag relationship intervals introduces some data snooping bias into results.

As a robustness check, we repeat the lead-lag calculations for several subperiods.

The next three charts show lead-lag relationships for two equal subperiods and for the last 20 years (with a little overlap due to methodology). Results for these subperiods suggest that the relationship between utility patent applications and the stock market is the most consistent.

To get a sense of how the negative leading relationship for utility patent applications might translate to investing, we look at the effect of ranking years based on annual change in these applications.

The final chart shows the average annual returns by the thirds (terciles) of the sample with the lowest, middle and highest growth rates in utility applications two years prior over the entire sample period (45-46 observations per third). For comparison, the chart also shows the average annual returns by terciles of the sample ranked by stock market returns two years prior.

Results confirm a negative relationship between past growth in utility patent applications and stock returns, and suggest a relationship somewhat different from (but not stronger than) reversion in returns. A double-sort that first halves the entire sample based on lagged change in utility applications and then halves each half based on lagged stock market returns (producing four groups of 34 observations each) indicates that mean reversion in stock returns is the stronger effect.

However, for the above data sets, applying a simple out-of-sample rule such as invest in SPY (cash) during a calendar year if the growth in utility patent applications two years prior is below (above) its inception-to-date median does not work well for 1994-2009. Nor does substitution of lagged S&P Composite Index returns for lagged utility patent applications work well.

In summary, evidence from simple tests suggest that relatively strong (weak) growth in utility patent applications may indicate stock market weakness (strength) the next few years, but the tendency appears not to be robust enough for investors to depend upon it.

You May Also Enjoy...

• The Efficient Innovation Premium
• Does Consumer Confidence Predict Stock Market Returns?
• Power of Skewness and Kurtosis to Predict Stock Returns
• Aggregate Earnings and Stock Market Returns
• Aggregate Investor Sentiment and Stock Returns