Large book-tax differences: Alternative Perspectives

Abstract

In this dissertation I investigate the information properties of the accounting construct known as book tax difference (BTD). I relate BTDs to various features and characteristics of firms, such as earnings persistence,[ Earnings persistence measures the extent to which current earnings persist or recur in the future. High persistence indicates a sustainable earnings generation process that is particularly valued by investors.] sales changes, business expenditures, and corporate asset structure. I specifically examine large temporary differences between book income and taxable income that have not been studied in full regarding their available information content. Because BTDs represent differences between two reporting systems, my emphasis is on the information obtained because of the unique properties of the two measures of earnings obtained by applying U.S. GAAP and the U.S. tax code. Understanding the relationship between current and future earnings and the pricing of future earnings is an ongoing question of interest in financial accounting. The inability of users to rely on current financial reports to forecast the sustainability of income has pushed investors and policymakers to pursue alternative sources of information to assess the quality level of earnings. For this reason, I examine the time series relation between current and one-year-ahead earnings, also referred to as earnings persistence, which is a primary measure of earnings quality. I apply an autoregressive model of PTBI in period t+1 on PTBI in period t and examine how sales change, managerial ability, and long-term investment in physical and intangible assets interact with large BTDs and pre-tax book income (PTBI). The foundation of my thesis is based on work by Hanlon (2005) and Blaylock et al. (2012) in the tax and earnings persistence literature and Kothari, Laguerre, and Leone (2002) and Amir, Guan, and Livne (2007) papers in the earnings growth literature. Following Hanlon’s (2005) specification as the basis for my main empirical model, I built upon this line of accounting research. The first chapter is a comprehensive introduction to the literature and the dominant theory of the thesis. In the second chapter, I investigate how downturns in sales are associated with lower earnings persistence when there are large negative or large positive BTDs. Earnings persistence following a sales decline is incrementally lower when firms exhibit large negative or positive book tax differences. I argue that lower earnings persistence occurs when sales declines are accompanied by disruptive changes in operations, reflected in asset write-downs, goodwill impairments, and other special items resulting in large negative book tax differences. Thus, I help to explain the previously unexplained finding in Hanlon (2005) that firm-years with large negative book tax differences have lower earnings persistence. In the third chapter, I adopt another approach to examining why large positive and large negative book tax differences (LPBTDs and LNBTDs) are associated with lower earnings persistence (Hanlon, 2005). Specifically, I investigate whether lower earnings persistence of LPBTDs and LNBTDs, relative to smaller BTDs, is due to earnings management or to uncertainty about investment outcomes. I first show that high CAPX (HiCAPX) firms and high R&D (HiR&D) firms are disproportionately represented in the LPBTD and LNBTD sub-samples, respectively. In contrast, high discretionary accrual (HiDA) firms are uniformly distributed across the LPBTD, LNBTD and smaller BTD sub-samples. Next, I find that HiDA firms are associated with lower earnings persistence in all BTD sub-samples uniformly, but HiCAPX and HiR&D firms are only associated with lower earnings persistence in the LPBTD and LNBTD sub-samples, respectively. Finally, extending the Hanlon (2005) model, I demonstrate that the differential earnings persistence associated with LPBTDs and LNBTDs is concentrated in HiCAPX and HiR&D firms, respectively. In the fourth chapter, I look at multiple characteristics of LNBTD firm-year observations to examine different explanations why LNBTDs occur. First, I show the data does not support common misconceptions about large negative BTDs representing financially distressed companies or post-maturity companies in shake-out or decline stages. Second, I develop an alternative hypothesis that firms enter LNBTD status temporarily as part of innovation cycles. Third, I provide contextual evidence that supports this alternative hypothesis. I demonstrate that LNBTD firms are more likely to be firms with high levels of managerial ability and strategic investment.