Seeking Shelter: Empirically Modeling Tax Shelters Using Financial Statement Information

Using confidential tax shelter and tax return data obtained from the Internal Revenue Service, this study develops and validates an expanded model for inferring the likelihood that a firm engages in a tax shelter. Results show that tax shelter likelihood is positively related to subsidiaries located in tax havens, foreign-source income, inconsistent book-tax treatment, litigation losses, use of promoters, profitability, and size, and negatively related to leverage. Supplemental tests show that total book-tax differences (BTDs) and the contingent tax liability reserve are significantly related to tax shelter usage, while discretionary permanent BTDs and long-run cash effective tax rates are not. Finally, the model is weaker, yet still significant, in the FIN 48 disclosure environment. This research provides investors and policymakers with an extended, validated measure to calculate the presence of extreme cases of corporate tax aggressiveness. Such information could also aid analysts and other tax and non-tax researchers in assessing the benefits and risks of firm behavior.