The federal government is arguably the most influential actor in the United States economy. Its agenda includes a wide range of activities, such as, rule drafting, enforcement actions, and tariff determinations. These activities are carried out daily by hundreds of agencies. In our new paper, we develop a measure of the entire agenda of the federal government, using machine learning techniques. Utilizing this measure, we study the industrial organization aspects of regulation and the associated economic impact on companies. Specifically, we focus on the effects of regulatory dispersion, that is, the effects of companies’ regulatory burdens being spread across a greater number of government agencies.
What is the government’s agenda?
To quantify the government agenda, we take advantage of the comprehensive information contained within the Federal Register (FR). The FR is the official daily publication of the federal government which details the activities of all the major federal agencies. Surprisingly, nearly half of the FR relates to activities other than formal rulemaking. These include lesser-known government actions such as hearings, public meetings, transactions approval, and grant applications, all grouped under the inconspicuous title “notices.” This is the first quick takeaway: ignoring this component of the government’s agenda, as academic studies tend to do, would miss a significant portion of the government’s activities.
We use machine learning techniques to categorize FR activities into 100 topics. While many past studies find an increase in government activity over time, we highlight how this trend varies across topics. For example, “Environment: data & studies” has increased steadily over time, whereas “Aviation safety: inspection” has declined. Our measure also provides a unique perspective on two of the biggest shocks to the government agenda over the past 25 years: the 2008-09 financial crisis and the 2016 presidential election. Around the financial crisis, the topics with the largest increases included “Consumer protection” and “Securities: futures.” Then, the Trump administration oversaw a stark decrease in overall government activity, with the fall being greatest in “Energy conservation” and “Endangered species.”
Our data highlights the extent to which many individual topics are regulated by multiple federal agencies. We present this mathematically with a measure of concentration that is analogous to the Herfindahl-Hirschman Index (HHI). We find that the most concentrated topic is “Securities: investment companies,” a highly specialized topic that is regulated by only a handful of financial agencies. In contrast, “Government Procurement: Small Businesses” is one of the most dispersed topics, handled by several dozen agencies.
Real effects of the government’s agenda
We use our unique measure of the government agenda to analyze the economic impact of the U.S. government activities on U.S. companies. Additionally, we focus on the industrial organization of government activities and how this relates to the regulatory burden of companies. Specifically, we examine the effects of regulatory dispersion, in particular the extent to which each company’s regulatory burden is spread across multiple government agencies. We find that regulatory dispersion may be beneficial or costly. On one hand, companies could benefit from being overseen by multiple agencies if this enables them to follow the least restrictive set of regulations. This scenario is analogous to firms benefiting from choice of court venue. We refer to it as the Benefits of Regulatory Dispersion hypothesis. Alternatively, regulatory dispersion might be detrimental to companies, as it potentially causes increased regulatory uncertainty and a lower ability to shape the regulatory agenda. We refer to this as the Benefits of Regulatory Concentration hypothesis.
To empirically test these hypotheses, we develop a measure of the regulatory dispersion of each company, in each year. Regulatory dispersion is a function of the topics relevant to a company’s business, combined with the set of government agencies that regulate each of these topics. To uncover the regulatory topics that each company must handle, we apply the same machine learning algorithm, which was trained on Federal Register data, to the financial disclosures of firms (10-Ks). This exercise shows that firms typically deal with many distinct topics. We combine this company-specific measure with our previously described HHI measure of government activity (the dispersion of government activity across agencies) to quantify the regulatory dispersion of each company-year.
Our findings highlight the extent to which many firms are accountable to numerous regulatory bodies: firms’ businesses relate to multiple regulatory topics, and each topic is spread across multiple agencies. However, there is considerable variation in companies’ regulatory dispersion, with some companies being beholden to far fewer government agencies than others. Interestingly, the regulatory dispersion of each company is not strongly correlated with other company characteristics, for example firm size, firm growth, and the firm’s total regulatory burden.
Our main empirical tests provide strong support for the Benefits of Regulatory Concentration hypothesis. First, firms that face greater regulatory dispersion have higher SG&A costs. Since these firms must answer to a greater number of government agencies, they devote more resources to government relations and to regulatory compliance. Second, firms with greater regulatory dispersion also have lower productivity, lower profitability, lower sales growth, and lower asset growth. This evidence is consistent with the idea that an increased regulatory burden diverts resources away from more value-increasing efforts.
Finally, we examine how regulatory dispersion affects capital investment and hiring policies. One possibility is that the regulatory burden reduces the net present value of potential projects, forcing companies to hire fewer people and to reduce capital expenditures. Alternatively, companies may need to hire more people to satisfy regulatory demands and perhaps even add physical capital to comply with the increased regulation. Our findings provide suggestive evidence that the latter channel represents the dominant effect: firms with higher regulatory dispersion have significantly higher employment and significantly higher capital expenditures.
A growing academic literature studies how regulation affects economic activity. Relative to existing studies, our approach has unique statistical and technical advantages. For example, it varies within firm and over time, utilizes a little-known administrative data set, and takes advantage of advanced linguistic tools to extract relevant information. The main takeaways are twofold. First, we capture the intensity of the entire government’s agenda and not only of final rules or subsets of the regulatory code. Second, we link the industrial organization of the federal government to firms’ regulatory burden. The burden borne by companies depends not only on the quantity of regulations, but also on the number of agencies who are involved in those regulations.
With our new measure, we shed new light on the real economic impacts of regulation. Classic theories on this topic range from public interest to public choice. We emphasize that the economic impact depends not only on the number of rules, but also on broader constructs of government activity and on how those activities are organized across different agencies. Greater regulatory dispersion leads to significant increases in costs and significant decreases in growth and profitability.
Joseph Kalmenovitz is an Assistant Professor of Finance at Drexel University’s LeBow College of Business.
Michelle Lowry is a TD Bank Professor of Finance at Drexel University’s LeBow College of Business and the Academic Director of the Gupta Governance Institute.
Ekaterina Volkova is an Assistant Professor of Finance at the University of Melbourne.
This post is adapted from their paper, “The Government Agenda and the Effects of Regulatory Dispersion”, available on SSRN.