This article features the work of Sonja O. Rego, the Chairperson and Professor of Accounting and the KPMG Professor at the Kelley School of Business.
There are numerous benefits of corporate tax avoidance, including generating cash tax savings and providing financial reporting benefits. However, tax avoidance can also impose significant costs on firms, including the costs of implementing tax strategies and future costs if the firm is audited by the IRS or other tax authorities.
As legislators and tax authorities increasingly scrutinize corporate tax practices, the relationship between tax avoidance and firm risk becomes increasingly important. This relationship is even more important given the significance of income tax expense for a firm’s after-tax profits.
As prior research has not provided direct evidence on the relationship between tax avoidance and idiosyncratic risk, this study offers a framework for documenting the types of firms where tax avoidance is positively, negatively, or not associated with both priced risk and idiosyncratic risk. The authors also offer evidence on the relative prevalence of each relationship.
Statement of Problem
The authors examine the relationship between corporate tax avoidance—defined as all transactions that reduce the amount of explicit taxes that a firm owes—and two components of firm risk: priced risk and idiosyncratic risk. They define “priced risk” as the risk resulting from a firm’s equity and “idiosyncratic risk” as firm-specific risk that should not require a risk premium.
Previous studies identify a negative association or no association between firm-level tax avoidance and priced risk and a positive association between tax avoidance and tax uncertainty, which the authors expect to be associated with higher idiosyncratic risk based on forecasted predictions. The authors also note that conventional wisdom suggests that firms that engage in greater amounts of tax avoidance have greater tax uncertainty.
The authors hypothesize that the relationship between tax avoidance and priced risk, as well as the relation between tax avoidance and idiosyncratic risk, differs across subsamples of firms.
Data Sources Used
The data sample includes all Compustat firms incorporated in the United States from fiscal years 1991 through 2016. The authors remove real estate investment trusts, financial institutions, and utilities because regulations in these industries likely affect both firm risk and tax avoidance opportunities. They measure tax avoidance using an industry and size-adjusted cash effective tax rate (ETR), subtracting a firm’s 5-year cash ETR from its mean 5-year cash ETR for the firms in the same group of total assets and industry.
Analytic Techniques
The authors develop predictions based on situations where they expect tax avoidance to be related to priced or idiosyncratic risk. They use a latent class mixture model to identify latent, or homogeneous, subsamples of observations that exhibit similar relations between the variables of interest within each subsample, while allowing the relations to differ across the subsamples. The model identifies homogenous classes in which tax avoidance is positively, negatively, or not related to priced or idiosyncratic risk, which allows the authors to determine the prevalence of each relation in their sample.
The authors then examine how firm attributes differ across the latent classes. The authors use the insights gained from the latent class models to develop parsimonious prediction models—models that enable out-of-sample predictions—to identify observations where tax avoidance is positively associated with both priced or idiosyncratic risk.
Results
Ordinary Least Square (OLS) regressions, which produce an “on average” coefficient for the entire data sample, suggest that tax avoidance is negatively related to both priced and idiosyncratic risk. The latent class mixture model reveals, however, that these on-average negative coefficients conceal a significant, positive relation that exists for a large sub-sample of observations. While 64.4% of the sample demonstrates a negative association between tax avoidance and priced risk, there is a not-trivial latent class — representing 35.6% of the sample — that exhibits a significant positive association between tax avoidance and priced risk. Tax avoidance is associated with increased exposure to priced risk among low-operating risk, mature firms that have more opportunities to engage in tax strategies with uncertain outcomes.
Additionally, while the OLS regression suggests a negative association between tax avoidance and idiosyncratic risk, the latent class mixture model reveals that 58% of sample observations exhibit a positive association between tax avoidance and idiosyncratic risk. The authors find that firms with a positive relation between tax avoidance and idiosyncratic risk have lower operating risk, but more incentives to increase firm profits and risk.
Finally, the parsimonious prediction models demonstrate that firms predicted to have a positive relation between tax avoidance and each measure of risk have negative future tax outcomes.
The negative future tax outcomes can include higher relative cash ETR volatility, greater additions to unrecognized tax benefits, greater future settlements with tax authorities, and a higher likelihood of future negative media coverage of the firm’s tax avoidance. The findings also suggest that it is important for financial statement users to evaluate the impact of corporate tax avoidance on priced and idiosyncratic risk within the context of firm type.
Additionally, the authors note that their findings have implications for research on corporate risk-taking, corporate social responsibility, and the reputational costs of tax avoidance. The prediction models may be useful in answering research questions that examine the contexts in which tax avoidance generates net costs or benefits for investors, or to examine differential responses to tax law changes for firms more exposed to priced or idiosyncratic risk.
Hutchens, M., Rego, S., & Williams, B. 2024. Tax Avoidance and Firm Risk: New Insights from a Latent Class Mixture Model. The Accounting Review 99 (1): 1-29.
Leave a Reply