State antitrust enforcement in the US and implications for small business entry and relocation

Abstract

This article examines the impact of the important, yet little studied, state-level antitrust enforcement activity on entry and relocation behaviour by small US firms. Feinberg and Husted (2011) have shown that this enforcement, especially nonhorizontal cases, may be viewed by potential entrants as a negative aspect of the state business climate. However, they did not pursue a more disaggregate analysis of small firm entry behaviour; nor did they investigate different responses between manufacturing, wholesaling and retailing firms. Another related issue is the extent to which state cases filed in tandem with federal investigations have the same impact on establishment entry as do purely ‘independent’ cases. These considerations are dealt with in this article. The author uses annual state-level data from the Statistics of US Business to examine entry and relocation reactions to state antitrust enforcement by firms within three small-business categories: 1–19 employees; 20–99 employees; 100–499 employees. Generally speaking, the smallest retail and wholesale firms seem to favour vigorous antitrust activity, especially enforcement targeted against cartel behaviour by suppliers. The largest small-firm retailers and wholesalers (those with 100–499 employees) seem somewhat threatened by such activity, especially the more controversial nonhorizontal enforcement. However, it must be acknowledged that the effects on entry or relocation of small firms – both positive and negative – are quite small.

Keywords:

• antitrust
• small business
• entry
• states

JEL Classification:

• L10
• L40

Acknowledgements

The author thank Brian Headd and two anonymous referees for helpful suggestions on earlier drafts. This article is an expansion of a research report written for SBA’s Office of Advocacy. Views expressed are those of the author alone and not of the US Small Business Administration.

Funding

The author acknowledges funding [under contract SBAHQ-11-M-0208] from the US Small Business Administration.

Notes

¹ A recent working paper by Costa and Barros (2012) does provide a theoretical framework for studying this issue, with somewhat ambiguous predictions.

² The employment size of entry is determined by size of firm, not size of establishment; e.g., the opening of a new Starbucks in a state would be classified as large-firm entry. Only firms with at least one salaried employee are included in the analysis.

³ Of course, this may also relate to a higher tax burden but that is controlled separately.

⁴ In about 15% of cases filed there are multiple ‘lead plaintiffs,’ and in a smaller percentage of cases other states joined in at a later stage, including at the settlement stage.

⁵ Alternatively, one might normalize antitrust cases by the number of establishments in the state, but as the correlation between GSP and the number of establishments is +0.9899 this would not have any major impact on our results.

⁶ Given that most states elect governors and attorneys general separately, the correlation coefficient between dummy variables for each being a Republican is not especially high: +0.25.

⁷ While various indexes of state regulatory burden exist, these are not available annually for the time period considered in this study.

⁸ Fixed effect models allow each state’s entry rate to differ by a constant amount compared to others, while random effect models incorporate more subtle differences across states in their unexplained variation (or error terms). For the variables of interest, the two types of models produce similar results in Table 2.

⁹ While this latter effect seems counterintuitive at first, it is consistent with a weaker economy leading to more entrepreneurial efforts by small firms.

¹⁰ Essentially, this approach estimates all nine industry-sector/firm-size regressions simultaneously, allowing for common influences among these to be reflected in the statistical findings.

¹¹ The full regression results are available on request from the author.

¹² Where no coefficients are reported, results were statistically insignificant.