Gerard Hoberg

Industry Choice and Product Language

We analyze the words that firms use to describe their products to examine multiple-industry firm operations. Our central finding is that multiple-industry firms operate across industries with higher product language overlap. Multiple-industry firms also avoid industries with more distinct language boundaries and those with more specialized within-industry language. We also find evidence linking these results to specific synergies such as potential entry into new markets, and realized synergies in the form of higher 10-K product description growth. These findings are consistent with multiple-product firms primarily operating in industries that lack language specialization. Firms that do operate across multiple industries choose industries with high language overlap and potential synergies. Our findings support the Cremer, Garicano, and Prat (2007) theory of firm organization and organizational language.

Optimized vs. Sort-Based Portfolios

Factors and test portfolios can be formed by optimizing objective functions instead of by sorting. Optimizing is more parsimonious and flexible, and the portfolio returns can be easier to find. Our approach effectively marries some advantages of the Fama and MacBeth (1973) cross-sectional approach with those of the time-series approach in Black, Jensen, and Scholes (1971). Our paper shows that optimized portfolios can make a difference: they reverse the inference in Daniel and Titman (1997) and Davis, Fama, and French (2000).

Mutual Fund Competition, Managerial Skill, and Alpha Persistence

What economic forces limit mutual fund managers from generating consistent outperformance? We propose and test the hypothesis that buy-side competition from other funds matters. We make three contributions. First, we propose new style-based spatial methods to identify the customized rivals of each fund. Second, we construct dynamic, fund-specific measures of competition and generate measures of skill as a fund’s outperformance relative to its customized peers. Third, and finally, we show that funds outperforming their customized rivals generate future alpha when they face less competition. These results are economically significant and last for over four quarters. Received September 11, 2016; editorial decision September 26, 2017 by Editor Itay Goldstein. Authors have furnished an Internet Appendix, which is available on the Oxford University PressWeb site next to the link to the final published paper online.

Offshore Activities and Financial vs Operational Hedging

A key question is why many multinational firms forgo foreign exchange derivative (FX) hedging and instead use operational hedging. We propose an explanation based on illiquidity and the unique advantages of operational hedges. We use 10-K filings to construct dynamically updated text-based measures of the offshore sale of output, purchase of input, and ownership of assets. We find that firms use FX derivatives when they are liquid and generally available. Otherwise, they often favor purchasing input from the same nations they sell output to, an operational hedge. Quasi-natural experiments based on new derivative product launches suggest a likely causal relation.

Product Market Uniqueness, Organizational Form and Stock Market Valuations

We introduce a new framework for forming peer firm portfolios that can account for firm uniqueness and organizational form. Our new vocabulary-based peer firm portfolios explain much cross sectional dispersion in firm valuations and generate a direct measure of firm product market uniqueness. We find that firms have higher stock market valuations than their peers when their products are more unique. This result holds for conglomerate and focused single-segment organizational forms. Increased success in patenting, increased branding, and less venture capital financed entry into the firms product space all contribute to the long-term maintenance of uniqueness and thus higher valuations.Using text-based computational analysis, we examine whether rm stock market valuations reect product uniqueness and how rms covary with peers in the stock market. We nd that rms have higher stock market valuations than matched peer rms when their product portfolios are more unique relative to matched peer rms. We also nd that the returns of text-based peers better explain own-stock returns than traditional industry peers. These results hold for both conglomerate rms and single segment rms using best-matched single-segment peer rms. Overall, our ndings show that the stock market values product uniqueness and recognizes peer groups based on fundamental product characteristics that are not reected in standard peer groupings.

Web Text-based Network Industry Classifications: Preliminary Results

Studies of market structure and product market competition are important in many disciplines, such as economics, finance, accounting and management. Reliable data for such studies is easily available for public firms (e.g., 10-K filings), but no reliable data exists for private firms. In this work we propose to mine the Internet Archive Wayback Machine, a digital archive of the World Wide Web, to build a database of 300,000 companies to support analyses of market structure, product market competition, and innovation. The goal of the WTNIC project is to download pages from the archive to build a profile for each company, and to use machine learning techniques to define similarity between companies based on similarity of their product and service offerings. This paper describes the challenges that must be overcome, our approach to overcome these challenges, and some preliminary results.

Dynamic Interpretation of Emerging Systemic Risks

We use computational linguistics to develop a dynamic, interpretable methodology that can detect emerging risks in the financial sector. Our model can predict heightened risk exposures as early as mid-2005, well in advance of the 2008 financial crisis. Risks related to real estate, prepayment, and commercial paper are elevated. Individual bank exposure strongly predicts returns, bank failures, and return volatility. We also document a rise in market instability since 2014 related to sources of funding and mergers and acquisitions. Overall, our model predicts the buildup of emerging risk in the financial system and bank-specific exposures in a timely fashion.Received March 1, 2018; editorial decision November 18, 2018 by Editor Itay Goldstein.

Does Product Market Competition Drive CVC Investment? Evidence from the U.S. IT Industry

We study the effect of product market competition on the propensity to use corporate venture capital (CVC) as a part of an information technology (IT) firm’s innovation strategy. Using novel measures of product market competition based on product descriptions from firm 10-K statements and accounting for potential endogeneity, we investigate how product market competition between 1997 and 2007 relates to the magnitude of CVC spending. We first find that firms in competitive markets make higher research and development (R&D) and CVC investments. In addition, we find that increasing product market competition leads to a shift away from internal R&D spending and into CVC. These movements are significantly stronger for technology leaders, i.e., firms with deep patent stocks, in the IT industry. We also find that CVC appears to be an effective way of exploiting external knowledge for technology leaders in the IT-producing industry, but not for technology slow starters. CVC investments lead to significantly more patent applications for technology leaders but no appreciable difference for slow starters. Our results provide new insights for theories of innovation in competitive, dynamic markets, potentially as part of a portfolio that includes internal R&D as well as open innovation models.

Feature Selection Methods For Understanding Business Competitor Relationships

Understanding competition between businesses is essential for assessing the likely success of new ventures or products, for making decisions before investing capital in new businesses, and understanding the impacts of regulatory policy. One important resource for analyzing competitor relationships are business webpages, which can capture the mission, products, services, and key markets associated with a company. However, webpages also contain irrelevant, extraneous, or misleading text, hampering prediction. To address this challenge, predictive models use a process known as feature selection to identify only relevant terms. The diversity and specificity of business domains pose a challenge for automated approaches for feature selection. In this paper, we compare two approaches to feature selection: manually-curated lists of terms provided by experts and automated approaches to feature selection. We evaluate several approaches to feature selection and their impact on predicting competitor relationships, demonstrating that carefully designed automated feature selection approaches can surpass the performance of manually-curated word lists by 10%.

Buy-Side Competition and Momentum Profits

We show that a new measure of buy-side competition explains momentum profits. The momentum quintile spread is 1.11% when competition is low and negligible when competition is high. Better alphas are attained with superior Sharpe and Sortino ratios, no negative skewness and in more investible strategies featuring value-weighted portfolios and large capitalization stocks. Stock characteristics traditionally related to momentum do not explain our results. Tests based on long-term reversals, the trading patterns of funds, their style peers, distant funds, and retail investors suggest that slow information diffusion explains the large momentum spreads and momentum reversals in low competition markets.