An Analysis of Firm Related Communications using a Marketing Lens

The determinants of firm performance have elicited a lot of research attention, particularly in the economics and management literatures. A generally acknowledged perspective is that firm outcomes are a certain function of (i) internal factors like firm demographics, organizational structure, capital structure, (ii) of external factors like industry structure and characteristics, regulatory environment, macroeconomic conditions, and (iii) of firm strategy, which connects these internal and external factors. However, firm strategy is an intangible quantity whose measurement, modeling and analysis have proven to be a challenge. Managers, when asked to describe firm strategy, do so in words. Most external communication by firms, used for informing various stakeholders the firms’ strategy, is textual in form. Firms' compliance filings, made at the behest of the government and regulators, often contain strategic content of interest and are again primarily textual in form. At the same time, most individuals and groups communicate by writing (which is textual in nature) or speech (transcribe-able to text). Consequently, the analysis of text data has assumed significance in various fields, especially in the social sciences. Text data (e.g. speech or interview transcripts, complaint emails, offer documents and prospectuses, product reviews, etc.) often capture qualitative aspects of information on a subject that may otherwise be unavailable. My research questions are motivated by these communications originating from or are about a firm. I infer the latent information, which has been studied in the past as “voice” or “mindset”, using text analytic methods and study its impact on firm performance. I classify all firm related communications into two categories – one that originates from the firm and the other that originates from any of its stakeholders (e.g. consumers, analysts, expert reviewers). The other dimension to these communications is the context in which they exist, i.e. is it a statutory or solicited form of communication (e.g. SEC filings, solicited open-ended responses iii from consumers) or is it voluntary (e.g. press release, letter to the shareholders, expert reviews, blogs). This classification allows us to a priori classify the communications into groups based on their origin and their nature. Having classified the forms of communications into these four quadrants, I then draw upon extant text-analytic methods to mine and extract qualitative themes or topics of interest underlying a mass of raw text responses, and develop a scalable approach to test these subjective themes for explanatory and predictive validity of firm performance. In this dissertation I use a marketing lens to measure the firm related communications from unstructured data to better understand its implications for consumers, firms and regulators. I define firm related communications broadly to include all types of discussion captured in unstructured format- consumer opinions as captured in consumer reviews and blogs, intra firm discussions and, public communications by firms and regulatory institutions. I combine econometric methods and natural language technologies to examine the influences on and of firm related communications. The first of the three essays proposes and validates an extension to a popular latent topic modeling algorithm which enables better interpretation of consumer insight from these communications. The second essay uses this proposed extension derive a measure of marketing effectiveness from communications and uses it to explain brand performance. The third essay uses this proposed extension to classify firms as following a niche or popular product-market choices based on strategic information in their statutory filings. I find a statistically significant performance differential attributable to product-market choices.