Research

The global innovation race is heating up, with advances in artificial intelligence, block chain, biotechnology, data storage and other cutting-edge technologies transforming sectors and global markets. This is also evidenced in the fact that the ratio of tangible to intangible assets of industry leaders around the world has been seeing a drastic change over the past four decades. The market capitalization of the S&P 500 is currently made up of 84% intangible assets, up from 20% in 1975 (according to MSCI).

With our research, we demonstrate that analysing intangible assets of a company in great detail allows building models to accurately forecast a company’s ability to convert its intellectual property into stock value in the following quarters. We also demonstrate that a company’s AI-IP impact correlates significantly with its future stock returns.

Our research offers a new innovation-related measure, IPI, which is contemporaneously associated with market valuation and predicts future operating performance and stock returns. Existing studies relating to innovation and market performance focus on the effects of either the input (R&D) or the output (patents) of innovation separately. Our study differs in focusing on Artificial Intelligence based innovative efficiency as a ratio of innovative output to input, based on the idea that efficiency should be highly value relevant. We find that the predictive power of AI innovation is incremental to that of other innovation related variables such as R&D intensity, significant R&D growth, patent counts and citations.

We demonstrate that portfolio analysis confirms our hypothesis and show that a portfolio of top companies generated a return of 237% (+159% to baseline), which is significantly higher than the market (+78%).

Introduction

The global innovation race is heating up, with advances in artificial intelligence, block chain, biotechnology, data storage and other cutting-edge technologies transforming sectors and global markets. To compete, companies can innovate in-house, or they can acquire others’ innovations. This is also evidenced in the fact that the ratio of tangible to intangible assets of industry leaders around the world has been seeing a drastic change over the past four decades. As these technological advances began driving innovation, the size and diversity within company asset portfolios began increasing consistently.

This dramatic shift is evident in the market capitalization of S&P 500 in 2015, which was made up of 84% intangible assets, as noted by MSCI. In 1975, intangible assets only made up less than 20% of the total asset value.

However, investors devote considerable attention to analyst reports, fundamental data, earning calls or news articles, while intangible assets receive less attention. Evidence exists that individuals pay less attention to, and place less weight upon, information that is harder to process. Information about intangible assets and innovation is hard to process, because it requires developing and applying a theory of how the economic fundamentals of a firm or its industry are changing. It also requires an analysis of the quality and path from intellectual property & inventions to final products on the market, the profit of which can be highly uncertain and long deferred.

In such a situation, where a majority of the S&P 500 is in intangible assets, it is only prudent to be employing the best in class techniques for research and benchmarking, to ensure that decisions are based on data of the highest quality.

Lauren Cohen et. al have demonstrated in a Harvard research book that the stock market is unable to distinguish between “good” and “bad” R&D investment, despite the fact that successful innovation is in fact predictable. They found that the market consistently misvalues information in an ex- ante, predictable way. Specifically, the market does not take into account the information in firms’ past R&D abilities. Firms that have been successful in the past and that invest heavily in R&D as percentage of sales earn substantially higher future stock returns than firms that invest identical amounts in R&D, but have poor past track records.

Over the last decades, the analysis of company patent portfolios has become an increasingly important part of competitive intelligence activities, as well as a key tool in analysing national, regional, and company technology strengths. Implicit in these analyses is the idea that identifying a company’s intellectual assets, specifically those intangible assets that patents protect, is tantamount to identifying areas of strength within a company. Given the rise and exponential growth of Artificial Intelligence technology, the focus on Artificial Intelligence related intellectual property is key to a company’s future growth and success. Companies with focus on Artificial Intelligence are two times more likely to grow exponentially than companies without.

In addition, the environment for innovation and technology has become more and more complex and difficult to analyse, especially with technologies emerging in the field of Artificial Intelligence. According to the market research firm Tractica, the global artificial intelligence software market is expected to experience massive growth in the coming years, with revenues increasing from around 9.5 billion U.S. dollars in 2018 to an expected 118.6 billion by 2025. The overall AI market includes a wide array of applications such as natural language processing, robotic process automation, and machine learning.

We can also assert that the rise of Artificial Intelligence is a global phenomenon. 83 percent of 2017 AI books on Scopus originate outside the U.S.. 28 percent of these books originate in Europe — the largest percentage of any region. University course enrolment in artificial intelligence (AI) and machine learning (ML) is increasing all over the world, most notably at Tsinghua in China, whose combined AI + ML 2017 course enrolment was 16x that of 2010. And there is progress beyond just the United States, China, and Europe. South Korea and Japan were the 2nd and 3rd largest producers of AI patents in 2014, after the U.S. Additionally, South Africa hosted the second Deep Learning Indaba conference, one of the world’s largest ML teaching events, which drew over 500 participants from 20+ African countries.

There is a growing awareness that the Artificial Intelligence intellectual property owned by companies can be an important factor in their commercial success. Intellectual property, particularly in the form of patents, provides the technological foundation upon which new products and services are built. Background research provides a strong rationale for the expectation that companies with strong IP portfolios, especially with intellectual in the Artificial Intelligence segment, will perform better in the stock market. A method devised to accurately measure the quality of company technology and innovation should therefore have a significant predictive effect on company stock performance.

Furthermore, information of this type should be particularly valuable because it is not currently available to market analysts, leading to the strong assumption that the quality of company technology might not be properly valued in the market. Deng et al. showed that companies with high-quality patent portfolios had market-to-book valuations that were 25 percent higher than other companies in the same industries with lesser-quality portfolios, both contemporaneously and for a number of years in the future.

The activity at the heart of our research is investment in research and development (R&D) combined with the strength and impact of its intellectual property, especially its Artificial Intelligence intellectual property (AI-IP). Given that R&D stimulates innovation and technology change, which can in turn lead to improvements in productivity, living standards and economic output, the quality of R&D investment and the ability to turn AI-IP into revenues and profitability is a critical task of the market.

With our research, we demonstrate that analysing intangible assets of a company in great detail allows building a model to accurately forecast a company’s ability to convert its intellectual property into stock value in the following quarter. We also demonstrate that a company’s AI-IP impact correlates significantly with its future stock returns.

Our findings add to a growing amount of literature highlighting the market’s inability to properly value investments in R&D. An alternative argument for why innovative efficiency would predict higher future returns derives from the q-theory (Cochrane, 1991, 1996; Liu Whited and Zhang, 2009). Firms with higher innovative efficiency tend to be more profitable and have higher returns on assets. All else equal, the q-theory implies that higher profitability predicts higher returns because a high return on assets indicates that these assets were purchased by the firm at a discount (i.e. that they carry a high-risk premium). Specifically, suppose that the market for capital being purchased by a firm is competitive and efficient. When a firm makes an R&D expenditure to purchase innovative capital, the price it pays is appropriately discounted for risk. For concreteness, we can think, for example, of a firm that acquires a high-tech target at a competitive market price. In this scenario, a firm on average achieves higher return (large number of patents, resulting in high cash flows) on its innovative expenditures as fair compensation if its purchased innovative capital is highly risky, and it receives low return if capital is relatively low-risk. Past innovation efficiency is, therefore, a proxy for risk, so firms that have high past innovative efficiency should subsequently be productive in patenting (Dierickx and Cool, 1989) and earn higher profits and stock returns. In other words, q-theory also predicts a positive innovative efficiency return relation.

Research Paper: "Using Artificial Intelligence to exploit IP related stock market inefficiencies"

Request a copy (PDF)

Authors Daniel Mattes, Jivka Ovtcharova, Gabriele Zedlmayer, Alexander Welzl
Summary

In this paper we demonstrate that analysing intangible assets of a company in great detail allows building a model to accurately forecast a company’s ability to convert its intellectual property into stock value in the following quarter. We also demonstrate the a company’s AI-IP impact correlates significantly with its future stock returns.

Our paper offers a new innovation-related measure, IPI, which is contemporaneously associated with market valuation and predicts future operating performance and stock returns. Existing studies relating to innovation and market performance focus on the effects of either the input (R&D) or the output (patents) of innovation separately. Our study differs in focusing on Artificial Intelligence based innovative efficiency as a ratio of innovative output to input, based on the idea that efficiency should be highly value relevant. We find that the predictive power of AI innovation is incremental to that of other innovation related variables such as R&D intensity, significant R&D growth, patent counts and citations.

We demonstrate that portfolio analysis confirms our hypothesis and show that a portfolio of top companies generated a return of 237% (+159% to baseline), which is significantly higher than the market (+78%).

Indicator Approach

Current Impact Index (CII) (0-1: below sector; >1: above sector)

The Current Impact Index (CII) is calculated with the mean value of all forward citations of identified AI-IP publications of a company, published within the previous 365 days, scaled by the mean value of all forward citations of identified AI-IP publications of all companies from the same industry, published within the previous 365 days. This measure expresses the technological impact of a company’s AI-IP compared to its industry peers. For example, a value of 1.10 would mean that the company’s current AI-IP impact is 10% higher than its peers at the day of the calculation. Calculated monthly.

IP to Market Value Indicator (IPI)

To create a robust AI-IP to market value model, we incorporate multiple parameters like R&D expenses, AI-IP, CII or revenue into the model. In this paper we applying a Long Short-Term Memory (LSTM) recurrent neural network. We frame the supervised learning problem as predicting the stock prices at the current month (t) given the CII values at the prior 24 time steps (2 years), the R&D expenses at the prior 4 time steps and the revenues at the prior 4 time steps (4 months). As the OLS regression above showed, lag times for up to 2 years before can be an important measurement. The input values have been normalized and chosen because of the evidently strong correlation between R&D expenses, intellectual property and resulting revenues. For each of the companies, a dedicated Long Short-Term Memory (LSTM) recurrent neural network is being trained. Calculated monthly.

Results Panel A, all companies, generated an accumulated profit of +78%, which is representing the baseline. Panel C, the bad companies, could only generate +55% (-23% to baseline), significantly less than the baseline. Panel B, the top companies, generated a return of 237% (+159% to baseline), significantly more than the baseline.

DataAI-42 Database Collection
Sample PeriodJanuary 1st 2017 up until July 1st 2019.
Conclusion

In this paper we demonstrate that analysing intangible assets of a company in great detail allows building a model to accurately forecast a company’s ability to convert its intellectual property into stock value in the following quarter. We also demonstrate the a company’s AI-IP impact correlates significantly with its future stock returns.

Our paper offers a new innovation-related measure, IPI, which is contemporaneously associated with market valuation and predicts future operating performance and stock returns. Existing studies relating to innovation and market performance focus on the effects of either the input (R&D) or the output (patents) of innovation separately. Our study differs in focusing on Artificial Intelligence based innovative efficiency as a ratio of innovative output to input, based on the idea that efficiency should be highly value relevant. We find that the predictive power of AI innovation is incremental to that of other innovation related variables such as R&D intensity, significant R&D growth, patent counts and citations.

It can be shown that lagged correlations of the introduced CII (Current Impact Index) for almost all of the industries are significant, as well as the correlation between the introduced IPI (AI-IP to Market Value Indicator) and the time lagged (t+1) stock price is significant in most of the industries.

We find that a long only portfolio strategy that takes advantage of the indicators presented in this paper yields abnormal positive returns that portfolio analysis confirms our hypothesis and show that a portfolio of top companies generated a return of 237% (+159% to baseline), which is significantly higher than the market (+78%).

Our findings suggest that it is possible to identify potential winners and losers based on IP related information and that such information can add additional value to every asset management strategy.

Competence in Artificial Intelligence

Rigourous systematic.


Daniel Mattes Daniel Mattes Daniel Mattes