Tax Losses and the Valuation of Cash

Transcription

1 Tax Losses and the Valuation of Cash Shane Heitzman USC Marshall School of Business Rebecca Lester Stanford Graduate School of Business May 12, 2017 Abstract We explore the economic consequences of tax loss carryforwards using hand-collected data for a large sample of firms between 2010 and Under a costly external finance channel, firms with tax losses are more likely to anticipate a binding financing constraint that arises because of lower tax benefits of debt, greater incentives for risk taking, and a higher marginal cost of issuing equity. Alternatively, under a managerial entrenchment channel, tax losses shield managers from monitoring by block holders and the market for corporate control. We explore these channels in the context of cash valuation. The costly external finance hypothesis predicts that the value of cash is increasing in tax loss carryforwards, while the managerial entrenchment hypothesis predicts the opposite. Our results are consistent with the costly external finance hypothesis: investors in firms with more tax loss carryforwards place a higher value on the marginal dollar of cash. The sensitivity of cash valuation to tax losses is strongest for domestic losses, for losses not subject to statutory limitations, and in firms that adopt tax loss-based poison pills. In anticipation of costly external finance, cash holdings appear to increase in tax loss benefits. Keywords: Cash valuation, cash holdings, taxes, losses We appreciate helpful research assistance from Allison Kays, Dang Le-My, and James Tse and comments from seminar participants at Arizona State University.

2 Tax Losses and the Valuation of Cash 1. Introduction The valuation of a firm s cash holdings has emerged as an intuitive approach to understanding the importance of capital market frictions and agency conflicts. The value of an additional dollar of cash is greater when firms face financing constraints (Faulkender and Wang, 2006), as internal funding allow managers to pursue positive net present value projects when external funding is costly. In contrast, the value of cash appears to fall when managers operate in weaker monitoring environments (Dittmar and Mahrt-Smith, 2007), suggesting that an additional dollar of cash is more likely to be spent on projects providing private benefits to entrenched managers (Jensen, 1986; Harford, 1999). These cash valuation models provide a simple and intuitive empirical framework in which to investigate the economic consequences of a variety of firm attributes. In this paper we study whether tax loss carryforwards tax attributes that shield future profits from taxation have implications for the cost of external finance and managerial entrenchment by examining whether the value of cash is sensitive to the firm s tax loss carryforward benefits. Our study is motivated by a number of empirical facts. First, over three-quarters of public firms in the U.S. report tax loss carryforwards and these losses are economically significant: aggregate Federal tax loss carryforwards for U.S. corporations in 2012 approached $2 trillion (Treasury Inspector General for Tax Administration, 2015). The potential tax benefits are even greater once state and foreign tax losses are considered. Second, tax loss carryforwards are a product of tax policy decisions that determine the government s risk sharing arrangement with firms (Domar and Musgrave, 1944; Scholes et al., 2014); these policies vary over time and across geographical boundaries in which the firm operates. Third, laws written to discourage trafficking in tax losses (i.e., IRC Section 382) have led to shareholder rights plans specifically designed to protect a firm s 1

3 tax losses from impairment under these laws (Erickson and Heitzman, 2010; Sikes et al., 2014). Prior academic work has focused largely on documenting the magnitude of tax loss carryforwards (Cooper and Knittel, 2010), studying how the losses affect investment decisions (Auerbach and Poterba, 1987; Edgerton, 2010; Dobridge, 2016; Langenmayr and Lester, 2017) and financing decisions (Graham, 1996), and showing how firms use accounting choices to maximize the tax benefits of losses (Maydew, 1997; Albring et al., 2011; Erickson et al., 2013). In this paper, we extend prior research by examining the relation between tax losses, corporate financial policy and governance within an empirical framework centered on investors valuation of corporate cash holdings. Based on the costly external finance channel documented by Faulkender and Wang (2006), there are at least three reasons to expect a positive relation between tax loss benefits and the valuation of cash. First, non-debt corporate tax shields such as tax loss carryforwards significantly reduce the marginal tax benefits of debt (DeAngelo and Masulis, 1980; Graham, 1996). While debt is a key source of external funding, tax loss carryforwards can increase the after-tax cost of debt financing by reducing the present value of interest deductions. 1 Consequently, investors may place a higher valuation on internal cash as the marginal source of financing. Second, lenders may price protect against a tax loss firm s enhanced incentives for risky investment. The asymmetric taxation of profits and losses in most corporate tax systems introduces a convexity into the firm s tax function that results in a positive association between income volatility and expected tax liabilities (Graham and Smith, 1999). However, firms with tax loss carryforwards are more likely than other firms to face a linear tax function in which expected tax 1 Specifically, firms that have existing tax losses may not be able to immediately deduct interest expense paid on borrowing; instead, the interest deductions will add to the existing tax loss and be carried forward. Thus, the interest deduction may not be used until a future period if/when the company reports taxable income. 2

4 liabilities are insensitive to income volatility. 2 If lenders anticipate the effect of tax loss carryforwards on the riskiness of a firm s marginal investment, these firms may face a higher cost of borrowing. Third, tax losses can also discourage firms from raising equity capital, as such issuances can trigger tax rules (IRC Section 382) that limit the firm s ability to monetize its existing tax loss benefits in future periods (Erickson and Heitzman, 2010; Sikes et al., 2014). 3 Consequently, the presence of tax loss carryforwards indirectly increases the cost of new equity financing (including stock issued to finance an acquisition) because of the increased probability of impairing existing tax loss benefits. Collectively, these three reasons tax benefits of debt, risky investment, and tax loss limitations suggest that firms with tax loss benefits face higher costs of external funds. Therefore, under a costly external finance hypothesis, the value investors place on cash holdings is increasing in tax loss carryforward benefits. In contrast, the value of cash can decrease in tax loss benefits if these tax attributes affect the capacity for managerial rent extraction. Dittmar and Mahrt-Smith (2007) show that investors place a lower value on cash in firms with weaker monitoring. Blockholders and potential acquirers serve as important external monitors of corporate managers (Holderness, 2003). However, trades by investors holding as low as 5 percent of the company s stock can trigger statutory limitations on the future realization of tax loss benefits under Section 382. As a result, tax loss carryforwards provide managers with a pretext to adopt plans such as the NOL Poison Pill, a shareholder rights 2 Firms with tax losses are then induced to increase their overall risk-taking to restore the expected after-tax return on investment projects (Langenmayr and Lester, 2017). 3 In short, if ownership by 5 percent shareholders changes by more than 50 percent within a three-year period, the amount of tax losses that can be used in future years is subject to a statutory limitation under IRC Section 382. While this U.S. rule was implemented to discourage firms from acquiring tax loss firms only for the expected future tax benefits, this limitation can affect any firm that crosses the 50 percent ownership change threshold. That is, the limitation can be triggered even without an acquisition of controlling interest by any single party. 3

5 plan designed to protect the value of the firm s tax loss carryforwards by discouraging trades by those investors most likely to provide external monitoring benefits. 4 The firm does not need to actually adopt such a device for monitoring to be affected; rather, as long as the existence of tax loss benefits affords the manager greater flexibility to minimize external monitoring mechanisms tied to ownership transfers, the value of cash should fall as potential tax loss benefits increase. To provide empirical evidence, we construct a comprehensive panel of tax loss carryforward data for a large sample of U.S. firms using hand-collected information from firms publiclyavailable financial statements between 2010 and With a few notable exceptions (Auerbach and Poterba, 1987; Graham and Mills, 2008; Cooper and Knittel, 2010), prior research relies almost exclusively on Compustat reporting to identify loss carryforwards (variable tlcf ). However, Compustat understates their frequency: approximately 89 percent of the non-financial and non-regulated firms in our sample report a tax loss asset, whereas Compustat reports only 59 percent for the same sample. Our data collection efforts extend beyond addressing the significant measurement error by also obtaining details on existing statutory limitations on the tax loss benefit, the allocation of loss carryforwards across geographical locations, and valuation allowances recorded by the firm that reduce the tax loss asset recognized on the firm s balance sheet. Consequently, our hand-collected data more accurately capture the population of firms with tax loss carryforward benefits and more precisely measure the total dollar amount and jurisdiction of 4 A commonly used instrument for a firm s governance structure is a shareholder rights plan ( poison pill ) designed to prevent unfriendly takeovers by forcing dilution of the buyer s interest once their ownership stake comprises 15 or 20 percent (Ryngaert, 1988; Brickley et al., 1994; Comment and Schwert, 1995; Coates, 2000; Fich et al., 2016). Since 2005, dozens of firms have adopted poison pills that are triggered at a lower 5 percent ownership threshold (based on the statutory tax rules of Internal Revenue Code Section 382) to protect the firm s tax loss asset from inadvertent impairment. Erickson and Heitzman (2010) and Sikes et al. (2014) study these tax loss poison pills. While the Delaware Court ruled that the poison pill was an appropriate action to preserve the value of the tax loss asset, Sikes et al. (2014) find that the market reaction to announcements of these poison pills is negative, which suggests that investors view these plans as a mechanism to insulate management from the threat of takeover in the market for corporate control. 4

6 these tax shields. Our primary empirical measure is a potential tax loss carryforward benefit (the tax loss benefit ), calculated as the maximum potential cash tax savings from the utilization of existing tax loss carryforwards. Tax loss carryforwards are generated by persistent tax losses. Thus, it is not surprising that accounting measures of profitability are key determinants of the balance of loss carryforwards. However, the data reveal significant heterogeneity in the type of firm reporting tax loss carryforwards; tax loss firms include both high growth firms as well as large, established multinationals. Firms with high tax loss benefits have significantly higher market-to-book ratios, conduct more R&D, and have higher stock returns, suggesting that such firms have substantial economic value. High tax loss benefit firms also have higher leverage, consistent with the endogeneity of tax status to capital structure decisions (Graham, Lemmon and Schallheim, 2008). A manager facing high tax loss benefits today can also expect those tax attributes to be relevant to the firm s decisions for several years, as tax loss carryforwards appear relatively persistent. To test whether cash valuation is sensitive to tax losses, we follow Faulkender and Wang (2006) and regress annual excess stock returns on the annual change in cash. The value of a marginal dollar of cash for the firms in our sample is $ To estimate the effect of tax loss benefits on this value, we include the level of the firms tax loss benefit at the beginning of the year, scaled by the market value of equity, and interact this term with the change in cash. We find a positive and significant coefficient on the interaction term of 0.84, implying that the value of a marginal dollar of cash is about $0.16 greater in the quartile of firms with the highest tax loss 5 This is lower than estimates in prior studies such as Faulkender and Wang (2006), which may be in part due to a smaller sample size, a shorter sample period, and selection on a larger size of firm. Furthermore, prior studies document a range of estimates based on the specific sample and model specification used. 5

7 assets. 6 These results support the costly external finance hypothesis and show that the tax loss benefits are associated with an increase in the value of corporate cash. To supplement this evidence, we test whether the sensitivity of cash valuation to tax loss benefits is conditional on the existence of Federal statutory limitations (under Section 382 of the Internal Revenue Code). Approximately 22% of the tax loss firms in our sample disclose the existence of a limitation on at least a portion of their loss carryforwards, potentially impairing their ability to monetize the tax loss benefits. We predict and find that the sensitivity of cash valuation to potential tax loss benefits is strongest among the firms not disclosing a limitation. Our next set of tests investigate the importance of existing financial constraints on the relationship between tax losses and cash valuation. If tax losses increase external financing costs due to limitations on the deductibility of interest or due to the incentives for risky investment, these factors should be most important when the firm is already constrained. Said differently, existing cash is likely to have more value when a firm needs external funding to continue operating, but for which the firm will incur significant tax-related costs to obtain it. To test these effects, we reestimate the cash valuation model separately for firms identified as constrained on four dimensions: dividend policy, the Whited and Wu (2006) index, long-term debt ratings, and a composite measure that considers a firm constrained based on any of these three individual measures. Across all measures except for long-term debt ratings, we find that the effect of the tax loss benefit on the valuation of cash is greater in the financially constrained subsamples, although only of marginal significance. 6 Interestingly, stock returns increase significantly in the tax loss asset. One explanation is that they are associated with greater volatility of after-tax cash flows. 6

8 If costly external financing explains why investors place a higher value on cash in high tax loss firms, then these firms should also hold higher levels of cash for precautionary reasons (Opler et al., 1999; Denis and Sibilkov, 2009). Consistent with this prediction, we find that a firm s tax loss benefits are positively related to the level of cash holdings. High tax loss benefit firms hold approximately 12 percent more of total assets in cash than low or no tax loss benefit firms. Consistent with our evidence on cash valuation, this estimate is strongest when the firm does not face statutory limitations on the future use of the tax losses. The effect of tax loss benefits on cash holdings holds after controlling for firm performance, a firm s repatriation tax costs, and a firm s unrecognized tax benefits (UTB), all of which prior literature has shown are associated with the level of cash holdings (Opler et al., 1999; Foley et al., 2007; Hanlon et al., 2017). Finally, we study whether the sensitivity of cash valuation to tax loss benefits depends the location (jurisdiction) of the tax losses. We first decompose the tax loss benefit measure into a domestic component and a foreign component for firms which provide disclosure of the location of the tax loss or the associated deferred tax asset (about 79 percent of the tax loss sample). We find that the sensitivity of cash valuation to tax loss benefits is driven by the domestic tax loss component. We also find, consistent with our earlier results, that this effect is strongest within the sub-sample of firms not subject to the U.S. statutory limitation. This study makes several contributions to the academic literature. First, we contribute to the literature studying loss firms generally (Denis and Mckeon, 2016) and tax loss firms in particular. Approximately 89 percent of the large publicly-traded firms in our sample report tax loss carryforwards, a much higher percentage than Compustat data indicate. While high tax loss benefit firms have performed poorly in the past and appear financially constrained, our data show that the sample also includes high growth firms that generate substantial accounting losses in their early 7

9 years as they create economic value. These firms persist in the sample despite, or possibly because of, the tax loss benefits that generate cash tax savings in future periods. Building on prior literature that focuses on the effects of tax losses on firm investment and income shifting, we demonstrate that these tax loss carryforwards can have important consequences for firm value through financing and governance channels. Second, this paper adds to the corporate finance literature studying the valuation of cash and cash holding decisions. The relative importance of tax loss carryforwards to the firm is likely to persist in subsequent years, suggesting that these losses should play an important role in short- and long-run financing decisions. Beyond repatriation taxes (Chen, 2016), the literature has not considered how taxes affect the valuation of cash despite the fact that income taxes can account for over 35 percent of income and that prior literature shows that this expense has direct valuation implications (Thomas and Zhang, 2013). We add to this literature by providing evidence that tax losses affect the value of a marginal dollar of cash due to its impact on the cost of external financing, and that tax loss carryforwards, despite their obvious association with past accounting performance, are associated with an increase in corporate liquidity. Finally, this paper informs policy makers about the mechanisms by which tax losses can affect other important firm decisions. In the past sixteen years, U.S. tax loss rules have changed three times to permit more generous tax loss offsets (Dobridge, 2016). These statutory extensions show that policy makers consider and alter these tax loss rules when attempting to achieve certain fiscal policy goals, such as providing fiscal stimulus. However, these policies also likely have nuanced or indirect effects on investor welfare, and we show that tax losses appear to be important through their impact on financing and governance attributes. Tax loss rules on carrybacks and 8

10 carryforwards vary across countries, are likely to shape how firms interact with their subsidiaries abroad, and represent an important avenue for future research. The paper proceeds as follows. Section 2 outlines the research design. Section 3 describes the sample and provides descriptive statistics on the sample of hand-collected firm data. Section 4 presents the results, and Section 5 concludes. 2. Research Design Our primary research question is whether and how the existence and amount of tax loss carryforwards benefits affect investor valuation of firm cash holdings. We test this relationship using the regression specification from Faulkender and Wang (2006) adapted to include the components of net financing as proposed by Halford et al. (2016) and used by Harford, Wang and Zhang (2017). Specifically, we estimate the following: r i,t R B C i,t TLBEN i,t 1 C i,t i,t = γ 0 + γ 1 + γ M 2 + γ i,t 1 M 3 TLBEN i,t 1 E i,t NA i,t + γ i,t 1 M i,t 1 M 4 + γ i,t 1 M 5 i,t 1 M i,t 1 RD i,t + γ 6 + γ M 7 i,t 1 I i,t M i,t 1 + γ 8 D i,t M i,t 1 + γ 9 C i,t 1 M i,t 1 + γ 10 L i,t + γ 11 STKIS i,t M i,t 1 (1) + γ 12 STKPR i,t M i,t 1 +γ 13 NETDT i,t M i,t 1 + γ 13 ACCUM i,t M i,t 1 + γ 14 C i,t 1 M i,t 1 C i,t M i,t 1 + γ 15 L i,t C i,t M i,t 1 + ε i, where denotes the change in the variable over the year. The dependent variable is the firm s excess stock return, where ri,t is the total return for firm i in year t, and R B i,t is firm i s benchmark return in year t. The benchmark return is constructed as the value-weighted return on a size and book-to-market matched portfolio (Fama and French, 1993; Faulkender and Wang, 2006). 9

11 C is the firm i s cash holdings and is defined as cash plus marketable securities. Both the change in cash holdings ( C i,t ) and cash at the beginning of the year C i,t 1 are included in Eq. (1). TLBEN is the potential tax savings from tax loss carryforwards and is estimated using financial statement disclosures. We measure tax loss benefits at the beginning of the year to mitigate the information content of new tax loss carryforwards for current performance. 7 Section 3.2 describes the construction of this measure in greater detail. The model also includes other firm-specific factors associated with stock returns: firm i s earnings (E), calculated as earnings before extraordinary items; net assets (NA), equal to total assets less cash; R&D (RD) and interest (I) expenses; total dividends (D) paid to common shareholders; and market leverage (L), calculated as total long-term debt plus debt in current liabilities, divided by the market value of assets. Following Halford et al. (2016), we decompose net financing into stock issues (STKIS), stock repurchases (STKPR), net debt issues (NETDT), and an indicator equal to one if cash holdings increased during the year. R&D is set equal to zero if missing. All variables (other than market leverage L) are scaled by the firm s equity value (MVE) at the beginning of the fiscal year such that the coefficients can be interpreted as the dollar change in a firm s equity value associated with a one-dollar change in the corresponding independent variables. The coefficient γ1 estimates the marginal value of a dollar of cash and theoretically should be equal to $1.00. To test the effect of the tax loss asset on the valuation of cash, we interact the change in cash with the measure of a firm s tax loss benefits (TLBEN). Under the costly external 7 The ending tax loss carryforward is comprised of the beginning carryforward plus the change in carryforwards during the year. A primary source of changes in carryforwards is firm profits, which reduce the carryforward, or firm losses, which increase it. In unreported tests, we include the change in the tax loss carryforward in Eq. (1) and find that increases in carryforwards are associated with lower stock returns, consistent with the change in carryforward being driven by an underlying tax loss. 10

12 financing hypothesis, the coefficient γ3 should be positive. Under the managerial entrenchment hypothesis, it should be negative. 3. Sample and Descriptive Statistics 3.1 Sample To test our research question, we rely on hand-collected data on tax losses disclosed in firms financial statements. We do not rely on the single Compustat variable for tax loss carryforwards ( tlcf ), as it reflects significant measurement error (Mills et al., 2003) and does not provide critical details about the magnitudes, sources, and limitations of the tax loss benefits. For example, the hand-collected data include data on worldwide tax losses (federal, state, and foreign losses) that are otherwise unavailable even using tax data from the IRS. Due to the necessary hand collection, we start by identifying a sample of the largest publicly traded firms. We first sort all U.S. listed firms on Compustat based on an annual composite ranking of assets, sales, and the market value of equity. We identify the largest 1,500 firms based on this ranking in any year between 2010 and 2015 for a sample of 1,958 distinct firms. We then handcollect data from the tax footnote in every available year of our sample period, yielding an initial sample of 9,910 firm-years. We drop all regulated and financial firms (2,302 observations) as these firms are subject to different rules that may affect firm valuation and the calculation of taxable income, and we retain observations with at least two years of accounting and market data. These steps result in a final sample of 6,855 firm-year observations. Table 1 provides details on the data obtained from the tax footnotes. Panel A compares the frequency of tax loss carryforwards in the hand collected sample to Compustat data. We show that 6,099 firms, or 89 percent of the full sample, report some amount of tax loss carryforward, either 11

13 through disclosure of a gross tax loss carryforward (the full amount of the loss available to offset future income) or through a deferred tax asset (the tax-effected amount of the loss carryforward) in the firm s income tax footnote. This figure is significantly higher than the 59 percent carryforward rate indicated by Compustat and reveals that a much larger proportion of firms report tax loss carryforwards than the Compustat data commonly used in academic research would otherwise indicate. Panel A also provides further details for the subsample of 6,099 firm-year observations that disclose a tax loss carryforward. We identify 66.6 percent (4,065 observations) based on disclosure of the total amount of tax net operating losses. Approximately 22.2 percent of the tax loss sample disclose statutory limitations on future utilization of the tax loss under Section 382 of the Internal Revenue Code, and nearly 90 percent of tax loss firms report an accounting valuation allowance that reduces the gross amount of total deferred tax assets recorded on the balance sheet. 8 Panel B provides further descriptive details on the losses reported. We first present statistics for the 4,065 observations disclosing the total tax loss carryforward. The average (median) tax loss carryforward is $823.2 ($207) million, meaning that an average firm could offset nearly $1 billion of future taxable income with existing tax losses. To better evaluate the relative significance of this amount given that the loss may apply to many jurisdictions, we report the carryforward amounts by location when disclosed. The average firm providing jurisdiction-specific details reported $475 million in carryforwards at the Federal level, $454 million at the state level, and $353 million in foreign jurisdictions. For comparison, the average loss carryforward reported in Compustat is $631 million; however, the Compustat data likely represent the simple sum of pretax 8 Firms generally report the valuation allowance in total, as opposed to reporting the amount of the valuation allowance specific to each deferred tax asset (such as the deferred tax asset related to tax loss carryforwards). While these allowances are often related to tax net operating losses, they can apply to any deferred tax asset. 12

14 losses carried forward and do not include jurisdiction-specific details, precluding us from identifying the source and relative value of the tax losses. We next show descriptive statistics for the subsample of firms reporting a deferred tax asset (5,877 observations). Recall that a deferred tax asset equals the firm s gross tax loss carryforwards times the applicable tax rate in the corresponding tax jurisdiction. The mean (median) gross deferred tax asset for tax loss carryforwards is $221.7 ($45.6) million. We note that firms either provide an uncontaminated amount by reporting the tax loss benefit amount on a distinct line in the deferred tax asset and liability section of the income tax footnote, or combine the tax loss asset with other tax attributes like credit carryforwards. For the 4,263 firm-years that report the deferred tax asset for the tax loss carryforward separately, the mean (median) asset is $170.3 ($36.3); for the 1,614 that bundle the tax loss carryforward asset with other items, the mean (median) is higher at $357.4 ($70.1) million as expected. A subset of firms also discloses the relevant jurisdiction of the tax asset. For these firms, the average gross deferred tax asset is $165.5 million at the Federal level, $42.7 million at the state level, and $127.4 million across foreign jurisdictions. The distribution is skewed, with reported medians at $28.5, $12.2, and $20.8 million respectively Construction of Tax Loss Benefit Measure To appropriately measure the economic significance of a firm s tax loss carryforwards, we account for variation in the economic value of a given dollar of tax losses carried forward. Our objective is to construct a tax rate-weighted measure of tax loss carryforwards that reflects the variation in the value of tax losses across Federal, state, and foreign jurisdictions. 9 In theory, a firm s deferred tax asset for the loss carryforward already reflects this methodology, but there are 9 The value of any given dollar of tax losses depends on the potential tax savings it generates; a one-dollar carryforward at the Federal level generates several times the cash tax savings as the same carryforward will at the state level. Thus, it would not be appropriate to use an unweighted measure because that would treat each dollar of tax losses (despite jurisdictional and tax rate differences) as the same. 13

15 several reasons why we do not rely primarily on deferred tax asset disclosures. First, a firm s deferred tax asset is shaped by accounting rules that could omit some tax loss carryforwards in determining the reported deferred tax asset. 10 Second, the deferred tax amount can include other tax attributes such as credit carryforwards, which can result in overestimation of the potential tax loss benefits. In our sample, approximately 27 percent of firms that disclose a deferred tax asset for tax loss carryforwards combine this amount with other deferred tax items. Third, the gross deferred tax asset usually does not include detail on the underlying jurisdiction. To overcome these issues with using the deferred tax asset amount, we calculate a rateweighted measure of tax loss benefits that prioritizes information on the tax loss carryforward disclosure. To construct the measure, we first use disclosed tax loss carryforward data at the jurisdictional level as disclosed by two-thirds of the sample (4,065 observations). We construct the tax loss benefit amount as TLCFj τj where TLCFj refers to the total losses carried forward in jurisdiction j, and τj is the associated tax rate. Based on the sample of firms that disclose both the deferred tax asset and the total amount of loss carryforwards by jurisdiction, we estimate the U.S. federal, state and foreign rates at 35 percent, 5 percent, and 27 percent respectively (i.e., the median tax rates as shown in Table 1, Panel B). If the firm does not provide jurisdictional data but does provide the gross tax losses carried forward, we apply a blended rate of 23 percent (also the median tax rate from Table 1, Panel B) to the total gross tax loss carryforward. This latter methodology is used to estimate tax loss benefits for approximately 8.9 percent of the sample (541 observations). For the remaining 24.4 percent of our tax loss sample firms, we use deferred tax asset disclosures. Although imperfect for the reasons outlined above, these deferred tax asset data permit 10 For example, the exercise of stock options generally creates a tax deduction. During most of our sample period, a portion of the stock option exercise deduction (the excess tax benefit) that increases tax loss carryforwards is reported off book ; the correct and full amount of the tax loss carryforward is disclosed (including this stock option deduction), but the relevant deferred tax asset ignores it. 14

16 measurement of tax losses that would otherwise be unavailable due to nondisclosure of total tax loss carryforward amounts. For these firms, we estimate tax loss benefits based first on jurisdiction specific disclosures (720 observations), then uncontaminated deferred tax assets (470 observations), and finally contaminated deferred tax assets (303 observations). Panel C of Table 2 provides the values for tax loss benefits constructed using this methodology. The average (median) firm has tax loss benefits of $190.7 ($39.4) million. Alternative methodologies, namely those that prioritize the deferred tax asset amount in the estimation of tax loss benefits, yield similar results and are highly correlated (ρ > 0.94). 3.3 Descriptive Statistics Table 2 provides descriptive statistics for the variables used in our empirical tests. Panel A presents statistics for the variables used in the multivariate specifications of tax loss and cash holdings determinants in addition to other characteristics of our sample firms. Panel B presents statistics for the variables used in the main cash valuation tests. We first present the mean values for the full sample, followed by five groups formed by sorting observations on the level of tax loss benefits as a percentage of total assets. The first group includes the 756 firm-years with no evidence of tax losses. The remaining observations are sorted into quartiles of tax loss benefits each year. By construction, the ratio of tax loss benefits to assets is increasing across the quartiles, from 0.2 percent of total assets in the bottom quartile to 10.8 percent of total assets in the top quartile. Even among firms in the top quartile of tax loss assets, Compustat only identifies 70 percent of these firms as having tax loss carryforwards. Approximately 77.3 percent of firms in the top quartile separately identify the portion related to the federal tax loss carryforward, followed by 64.0 percent and 46.6 percent for state and foreign losses. Among firms reporting tax losses, the percentage of firm-year observations subject to the statutory IRC Section 15

17 382 limitations on these losses increases monotonically, from 14.3 percent in the bottom quartile to 36.6 percent of firms in the top quartile. Consistent with the level of tax loss benefits insulating firms from current taxation, we find that cash taxes paid, current tax expense, total tax expense, three-year cash taxes paid, and the volatility of cash taxes all decline monotonically as tax loss benefits increase. For example, the average cash taxes paid by firms without tax loss benefits is about 4.4 percent of assets, decreasing to 0.7 percent of assets for firms in the top quartile of tax loss benefits. High tax loss firms hold the most cash (18.1 percent of book assets), report the lowest amount of pre-tax ROA (0.2 percent), and have the most debt (33.3 percent of book assets). While firms with high tax loss carryforwards are unsurprisingly poor-performing by historical accounting metrics, the data reveal a more nuanced picture. These are the smallest firms in the sample by book asset measures, but they are still large by conventional measures (in part due to sample construction), averaging $5.2 billion in assets. Furthermore, high tax loss firms report the greatest levels of R&D expenditures (9.1 percent of sales) and high levels of capital investment (5.7 percent of assets). Approximately 70.8 percent of firms with high tax losses have foreign activity, similar to the subsample of firms without tax losses and more than those reporting small losses. Foreign operations contribute more to firm profits than domestic operations, with average foreign pretax income equal to 1.2 percent of total assets versus -0.4 percent for domestic pretax income. In addition, high tax loss firms also report high sales growth during the year, averaging 16.2 percent. Because firms that report the largest tax loss carryforwards appear to have substantive growth 16

18 opportunities and are responsible for considerable investment activity, they should also be more sensitive to factors that affect their access to financing. 11 Panel B provides descriptive statistics for the variables used in the cash valuation model. The average change in cash (scaled by beginning market value of equity) of 0.5 percent is very similar to the value in Faulkender and Wang (2006) of 0.4 percent. The fourth quartile of firms based on tax loss benefits reports the highest average values for the change in cash (0.8 percent of the market value of equity). We construct four measures of financial constraints and find that high tax loss firms appear the most constrained based on the payout measure and the Whited and Wu (2006) index. Firms with high tax losses issue less stock, repurchase more shares, and surprisingly, issue more net debt during the year. This is consistent with managers being sensitive to the threat of triggering Section 382 limitations through equity issuances, but is at odds with the standard view that firms with the most non-debt tax shields issue less debt on the margin (Graham, 1996). Table 2, Panel C provides descriptive statistics on the persistence of tax losses. We partition the sample into five subgroups based on the ratio of tax loss benefits to assets in 2010 and track the average tax loss benefit across these groups over future periods. For all groups except the high-tax-loss firms, tax loss benefits as a fraction of assets increases slightly over the sample period. For example, firms in the third quartile of tax loss firms report a small increase from 2.1 to 2.5 percent of total assets. By comparison, firms reporting the highest level of tax losses 11 The descriptive statistics on the other quartiles reveal that there is not a monotonic relationship between the level of tax losses and several important firm characteristics. For example, the firms in the second quartile are the largest firms based on the book value of assets. The second and third quartiles contain the highest proportion of firms with a foreign presence, and they are the most profitable in terms of foreign and domestic ROA. In short, firms in the second and third quartiles of tax loss firms do not exhibit the common characteristics of poorly-performing, constrained firms and instead are large, profitable, multinational companies. 17

19 experience a decline over the sample period, from 11.9 percent of total assets in 2010 to 7.1 percent in This relative decline in tax loss benefits over the period could be attributable to different outcomes of sample firms. Panel D shows that high tax loss firms have higher rates of acquisition. While firms with the most tax losses are the most likely to experience an impairment of the tax loss value due to acquisition, their non-tax attributes (including growth opportunities and R&D activity) are associated with a higher probability of being acquired within a one-, two-, and sixyear horizon: by the end of calendar 2016, 14.2 percent of firms with the highest tax loss benefits in 2010 have been acquired compared to 6.0 percent of firms reporting no tax losses. Finally, Panel E presents the incidence and level of tax loss benefits by Fama and French 48 industry definitions. Tax losses are most prevalent in the Communications, Pharmaceutical, Electronic Equipment, and Computers industries all of which are R&D and investment-intensive. The tax losses for these industries average 5.1 to 7.1 percent of total assets. In contrast, firms reporting the lowest level of tax benefits are in retail and service industries. Because industry characteristics are important determinants of tax incentive and attributes, we include industry fixed effects in all regressions. 3.4 Determinants of Tax Losses Table 3 provides a multivariate analysis of the determinants of the tax loss benefit. In Columns (1) and (2), we estimate a logit regression explaining the probability of a having a tax loss carryforward based on Compustat (col. 1) and hand-collected (col. 2) data. The results in Columns (1) and (2) confirm that R&D expense, foreign presence, and the level of debt are positively and 12 Looking at dollar amounts, tax loss benefits are relatively flat over the five-year period, with the drop from 11.9 to 7.1 percent being driven by a 64% increase in book assets for the median firm. 18

20 significantly associated with having a tax loss carryforward, whereas the market-to-book ratio and dividend payments are negatively correlated with the loss carryforward. We note that the association with firm size is statistically significant in both columns, but with opposite signs, suggesting that Compustat is less likely to detect tax losses in large firms. Interestingly, profitability does not appear strongly correlated with the likelihood of having loss carryforwards. Turning to the 90% of observations that do report a tax loss carryforward, we examine the determinants of the amount of tax loss benefits scaled by total assets in Column (3). While the level of tax loss benefits is decreasing in size, firms with high growth opportunities (market-tobook and R&D intensity) have the largest tax loss benefits. As expected, weak performance measured by a lack of dividend payouts, as well as domestic and foreign losses, are strongly associated with greater tax loss benefits. Tax loss benefits are increasing in leverage, but decreasing in acquisitions, suggesting that acquired tax attributes do not drive the tax loss benefits observed in the data. In panel B, we extend the study of the determinants of tax loss benefits using jurisdictionspecific details by replacing total tax loss benefits with its Federal, state, or foreign components. We restrict the samples to those firms that disclose any jurisdictional information, and thus, the samples in each regression are the same. In general, we observe consistent determinants across the three specifications with a few key exceptions. First, domestic and foreign profitability is an important determinant of federal and state tax benefits, but only domestic profitability is significantly associated with foreign tax losses. Second, market-to-book and R&D activity are associated with greater domestic tax loss benefits, but lower foreign tax benefits. One interpretation is that multinational firms engage in R&D activities within the U.S., where there is a developed infrastructure, a high quality employment pool, and attractive tax incentives, followed 19

21 by moving the intellectual property offshore to facilitate income shifting (Hanlon et al., 2015). Leverage is also associated with greater domestic tax loss benefits, but not with foreign benefits. Again, this appears consistent with firm s motivation to locate deductions in the jurisdiction where they have the most value. 4. Empirical Results 4.1 The Sensitivity of Cash Valuation to Tax Loss Benefits Table 4 presents the results of our main tests examining the effect of a tax loss asset on the valuation of firm cash holdings. Because we condition the valuation on the tax loss benefit at the beginning of the year, and because we have hand-collected data on tax loss benefits for years ending in 2010 through 2015, we estimate the regressions using stock returns in years 2011 through In Column (1) we replicate the regression from Faulkender and Wang (2006), separating the net financing variable into four variables (Stock Issued, Stock Repurchased, Net Debt Issued, and Cash Accumulation) following Halford et al. (2016). We find that the marginal value of cash for our sample of firms is $ The control variables exhibit similar relationships with excess returns as in the prior literature, and consistent with Faulkender and Wang (2006), the coefficient on the interaction term that captures variation in the value of cash as the level of debt increases is negative and significant. Column (2) includes a variable for the tax loss benefit as of the beginning of the year scaled by the market value of equity, as well as the interaction of this measure with the change in cash holdings. The positive and significant coefficient on the interaction term means that the tax loss 13 It is closer to $0.85 if the four net financing variables are collapsed into a single variable as in Faulkender and Wang (2006). 20

22 asset is associated with a higher valuation of firm cash holdings. 14 We find a positive and significant coefficient on the interaction term of 0.84 (t = 3.83), implying that the value of a marginal dollar of cash is approximately $0.09 greater in the quartile of firms with the highest tax loss assets [0.84 x ( )]. This result is consistent with the prediction that tax loss benefits are associated with a higher cost of external financing, and this in turn increases the value that investors place on internal cash holdings. In Panel B, we partition the sample based on whether the firm discloses that future utilization of the tax loss benefit is limited under Section 382 of the U.S. tax code. Potential tax loss benefits are unlikely to reduce the tax benefits of debt, lead to higher incentives for risk taking, or provide a credible threat of a further Section 382 impairment when they are already impaired as a result of past events. Thus, we expect that the influence of tax loss benefits on the valuation of cash will be mitigated when the utilization of tax losses is already limited. To narrow our focus to cases in which the limitations have the greatest material impact on the realization of tax loss benefits, we require the firm to also have established a valuation allowance equal to at least 50% of the potential tax loss benefits. We estimate the regression allowing the coefficients on cash, tax loss benefits, and the interaction to vary across the partition on limitations. In Column (1), we find that the interaction between the change in cash and the tax loss benefit is positive for both sub-samples, but the coefficient on the interaction term is significant only for the subset of firms for which the benefit is not limited (coeff. = 0.89, t = 3.57). However, the difference between the coefficients is insignificant (p = 0.18). The evidence is generally consistent with tax loss benefits affecting the valuation of cash within a subsample for which tax loss benefits are more likely to matter. 14 For the same reasons that the tax benefits of interest deductibility fall with the presence of tax loss carryforwards, the tax costs of interest income from holding cash also fall. Thus, evidence that cash is more valuable in firms with tax loss carryforwards can also arise if cash is less expensive to hold in such firms. 21

23 In Column (2), we repeat the analysis, but instead partition the sample into whether or not the firm had, at any point in the sample period, adopted an NOL-based poison pill. We identify these firms using the Factset Shark Repellent dataset as in Sikes et al. (2014). If the adoption of a poison pill to protect the tax loss from impairment signals a manager s belief about the economic importance of the tax losses, the sensitivity of cash valuation to tax losses should be higher in firms that adopt these pills. We find that the coefficient on the interaction term is positive and significant for both subsamples. While the coefficient for the sub-sample of firms that have adopted a poison pill appears larger (3.66 as compared to 0.78), we note that the difference is not statistically significant (p = 0.12) The Sensitivity of Cash Valuation to Tax Losses in Financially Constrained Firms In Table 5, we provide further evidence on the positive association between tax loss benefits and cash valuation by partitioning firms into constrained and unconstrained subsamples. We expect that the positive effect of tax loss benefits on the value of cash under the costly external financing hypothesis is greatest among firms already experiencing financing constraints. For these firms, the statutory limitations on the deductibility of interest or the future utilization of the tax loss benefits are likely to bind precisely when the firm is most constrained when external debt or equity financing is necessary for the firm to continue operating but for which the firm will incur significant tax costs to obtain such financing. Following prior research, we construct four measures of financial constraints including, i) Payout policy, which considers a firm constrained if it pays no dividend in year t - 1; ii) the Whited- 15 Two points are worth noting here. First, the coefficient on the change in cash is significant only for firms without pills. While the NOL poison pill sample is small, one interpretation of this result is that the poison pill adoption is associated with weaker corporate governance, which results in a lower valuation of cash. This result would be consistent with investors negative reactions to announcements of these plans (Sikes et al., 2010). Furthermore, governance could explain the positive coefficient on the interaction between the change in cash and tax loss benefits for the poison pill sample, as cash valuation is higher in firms with more tax loss benefits to protect. 22

24 Wu index, which is a function of cash flow-to-assets, dividend payments, long term debt, size, sales growth, and industry sales growth and considers firms in the top 40% of the index as constrained; iii) LT Debt Rating, which considers the firm constrained if it has long-term debt outstanding but no long-term debt rating in year t-1, and iv) Composite, which is an indicator equal to one if the firm is identified as constrained under any of the prior measures, and zero otherwise. Partitioning our observations into constrained and unconstrained sub-samples based on these measures, we re-estimate Eq. (1). Across all partitions, the coefficients on the change in cash are higher for constrained firms, consistent with prior research. We next examine the interaction between the change in cash and tax loss benefits across the partitions. Using Payout Policy to partition the sample, the effect of tax loss benefits on the value of cash is positive and significant (coeff. = 0.88, t = 3.82) for constrained firms (those that do not pay dividends) and positive but insignificant (coeff. = 0.13, t = 0.26) for unconstrained firms, although the difference is not significant (p = 0.16). Based on the Whited-Wu partition, the coefficients on the interaction between tax loss benefits and the change in cash are positive and significant across both groups, but larger for the constrained group (2.02 vs. 0.66; p-value of difference 0.01). In contrast, the effect of tax loss benefits on the value of cash for firms constrained according to a long-term debt rating are the opposite, with tax loss benefits associated with a significantly lower valuation of cash in constrained firms (coeff. = -0.01, t = 0.48) and higher valuation in unconstrained firms (coeff. = 1.16, t = 3.31). This inconsistency appears to mirror the unexpected positive relation between tax loss benefits and leverage and the change in leverage, suggesting a more complicated relationship between tax loss benefits and leverage, worthy of future exploration. For the composite measure, the results again suggest that the impact of tax loss benefits on the value of cash are highest in constrained firms. 23