Research note · 5 min read

Innovation as a Leading Indicator

How patent activity, R&D efficiency and specialised hiring measure a company's investment in its future, and why research finds markets slow to price it.

SharpeMetrix Research · November 16, 2026

Most of a company’s investment in its future does not appear on its balance sheet. A factory is capitalised and depreciated over years. Research and development is, in most cases, expensed in the year it is spent. A company that doubles its R&D budget reports lower earnings today, even if that spending produces the products that drive the next decade.

This note explains how SharpeMetrix measures innovation and why the academic evidence suggests markets price it slowly.

What the Innovation signal measures

SharpeMetrix’s Innovation family scores companies on how intensively and how productively they innovate. It uses two kinds of data:

  • Patents. Patent filings and grants, with attention to quality and impact, not just counts.
  • H1B visa filings. Applications to hire specialised foreign workers, which track a company’s investment in technical talent.

Both are measured at company level and industry level. SharpeMetrix’s research has found company-level innovation becoming a stronger driver than industry-level patterns in recent years, and visa data often more informative than patent counts alone. Hiring is an early step: engineers are hired before patents are filed, and patents are filed before products ship.

Scores are ranked 0–100 against comparable companies. This matters because patenting habits vary widely. A pharmaceutical company and a software company protect their work in very different ways, and comparing raw patent counts across them would mostly measure industry practice.

Spending versus efficiency

The simplest innovation measure is R&D intensity: R&D spending as a share of sales or market value. Chan, Lakonishok and Sougiannis (2001) found that, on average, R&D-intensive companies did not earn higher returns than others. However, R&D-intensive companies whose shares had performed poorly went on to earn significant excess returns, consistent with the market undervaluing R&D at companies whose reported earnings had been depressed by it.

Later research focused on what the spending produces. Hirshleifer, Hsu and Li (2013) measured innovative efficiency as patents and patent citations per dollar of R&D. Firms with high efficiency earned higher subsequent returns, and the effect was not explained by standard risk factors. Kogan, Papanikolaou, Seru and Stoffman (2017) used stock market reactions to patent grants to estimate the economic value of individual patents, and found that this value varies enormously: a small number of patents account for much of the total.

The practical lesson is that spending and patent counts are both crude. Quality, impact and the ratio of output to input carry more information.

Why markets underreact

If innovation is valuable and the data is public, why would prices not reflect it? Research suggests several reasons.

Accounting hides it. Expensing R&D lowers reported earnings and book value. Screens and valuation models built on those numbers penalise innovative companies.

The payoff is distant and uncertain. The return on a research programme may arrive years later, if at all. Investors facing uncertain, long-dated payoffs may discount them too heavily or ignore them.

It is hard to evaluate. Judging whether a patent portfolio is strong requires technical knowledge most investors do not have. Cohen, Diether and Malloy (2013) found that companies with a consistent track record of turning R&D into sales growth earned higher future returns, even though that track record was observable. They argued that investors did not distinguish between firms that were good at R&D and those that simply spent on it.

Attention is limited. Patent grants and visa filings are released through government databases, not earnings releases. Few analysts read them systematically.

A generic illustration: two medical-device makers each spend 12% of revenue on R&D. One has filed a steady stream of highly cited patents and hired engineers year after year; the other’s spending has produced little measurable output. Their R&D-to-sales ratios are identical. Their innovation scores are not.

How it fits the composite

Innovation tends to be a slow-moving signal. Patent portfolios and hiring patterns change gradually, so the scores carry relatively low turnover. It also measures something largely absent from the other families. Value and Quality look at reported financials, which understate innovative firms; Momentum and Low Volatility look at prices. Innovation looks at activity that has not yet reached either.

That is the reason to include it as one of seven families, not a reason to rely on it alone. Innovative companies can be expensive, volatile and concentrated in a few sectors. The composite balances innovation against valuation, quality and risk, and portfolio construction keeps sector weights within ±7% of the benchmark.

Limits

Patents are an imperfect proxy. Many companies protect their work through trade secrets, speed to market or software that is not patented. Visa filings reflect immigration policy as well as hiring demand. Patent data arrives with delays, since applications are typically published around 18 months after filing. Any innovation measure should be read as an indicator of activity, not a forecast of commercial success.

References

  • Chan, L. K. C., Lakonishok, J., & Sougiannis, T. (2001). The stock market valuation of research and development expenditures. Journal of Finance, 56(6), 2431–2456.
  • Cohen, L., Diether, K., & Malloy, C. (2013). Misvaluing innovation. Review of Financial Studies, 26(3), 635–666.
  • Hirshleifer, D., Hsu, P.-H., & Li, D. (2013). Innovative efficiency and stock returns. Journal of Financial Economics, 107(3), 632–654.
  • Kogan, L., Papanikolaou, D., Seru, A., & Stoffman, N. (2017). Technological innovation, resource allocation, and growth. Quarterly Journal of Economics, 132(2), 665–712.

More research

See it with your own book

Every model, every holding, every trade. Open for 30 days.

Full platform access with no credit card. Or book a 20-minute walkthrough and we'll map the models to your current lineup.