Earnings Repatriation Tax Cost Risks and Bank Loan Contracting

Abstract

Unlike purely domestic firms, multinational companies have distinctive opportunities to engage in sophisticated international tax planning strategies. This study investigates whether banks perceive potential earnings repatriation taxes as a significant source of risk when designing loan agreements for these firms. Our findings reveal that U.S. multinationals facing higher potential repatriation tax burdens are subject to wider loan spreads, indicating increased risk premiums. Moreover, this effect is especially pronounced among firms with low profitability or limited financial flexibility, highlighting the risk-sensitive nature of these loans. We also observe that lenders are more likely to demand collateral and impose stricter financial covenants in loans to firms with substantial repatriation tax exposure, further underscoring that banks regard these taxes as a firm-specific risk factor. By exploring the intersection of international tax considerations, potential earnings repatriation taxes here, and debt contracting, our research makes a valuable contribution to the literature, shedding light on how global tax issues influence credit markets and lending behavior.

1. Introduction

A central challenge in international tax is to navigate the relationship between one country’s tax system and the rest of the world, and thus, how firms deal with international taxes becomes an important question in international business. When firms globalize, two or more jurisdictions assert their right to tax the firm or its transactions. Countries negotiate between themselves in order to deal with the concern of double taxation. Because each country designs its own tax regime, given its own policy goals, and countries have incentives not to fully cooperate under the current international tax regime, firms can take advantage of conflicts or differences between countries’ tax rules to commit tax arbitrage, structuring transactions to minimize tax payments. For example, in order to successfully engage in international tax planning, U.S. multinationals (MNCs) have engaged in many activities such as locating operations strategically, shifting income between locations (see, e.g., Rego, 2003; Klassen & Laplante, 2012), and leaving large amounts of foreign profits, held as cash abroad (see, e.g., Foley et al., 2007; Hanlon et al., 2015; Edwards et al., 2016). In this study, we investigate whether, and to what extent, banks consider borrowers’ potential earnings repatriation tax cost risks when designing loan contracts.1

Understanding the effects of earnings repatriation taxes on loan contracts is timely and relevant because its consequences are still under debate. President Biden in the U.S. considered further changes to earnings repatriation tax policy, but President Trump’s 2017 Tax Cuts and Jobs Act (TCJA) repatriation tax implications were upheld by the U.S. Supreme Court in the Moore vs. United States ruling in June 2024.2 Further debate on MNCs’ tax planning activities cost the governments of the European Union between 50 and 70 billion euros and the U.S. between $77 to $111 billion each year (Dover et al., 2015; Clausing, 2016). In response to concerns related to international tax planning, the OECD launched an action plan to combat base erosion and profit shifting (BEPS, see OECD, 2013). Additionally, a major component of the Tax Cuts and Job Act of 2017 introduces a one-time repatriation tax for U.S. MNCs and attempts to dissuade future transfer pricing by MNCs by taxing foreign profits going forward in the year earned, regardless of the timing of repatriation. However, the ongoing public debate in the U.S. about the corporate taxation of foreign profits has continued after the 2020 and 2024 presidential elections, with Presidents Biden and Trump both pushing for (different) changes during their presidencies that call into question the future of international tax planning for U.S. MNCs and the associated risks of current and future tax planning.

In this study, we empirically investigate the response of banks to U.S. MNCs’ earnings repatriation taxes when designing debt contracts. The U.S. is an ideal setting in which to study this research question. First, the U.S. has the largest number of MNCs. Approximately 89 percent of S&P 500 firms operate abroad and, on average, report material subsidiaries in 19 countries with 49 percent of their pre-tax earnings in foreign subsidiaries (J. L. Blouin et al., 2014). Second, for MNCs involving multiple countries, U.S. tax on profit generated in foreign subsidiaries is deferred until those profits are repatriated over our sample period, which provides MNCs an opportunity and incentive to engage in tax avoidance planning (Kim et al., 2011a; Lisowsky, 2010; Ma et al., 2020). Third, the benefits and cost of repatriation taxes can be significant in size, meaning they are likely to be considered by firms (borrowers) and banks.

We calculate potential earnings repatriation taxes as the “statutory-minus-effective tax rate difference” to measure the degree of a borrower’s international tax planning activities in our U.S. setting because this measure captures the savings of U.S. MNCs that engage in one significant international tax planning activity. Even though a high degree of international tax planning might result in a firm saving more cash, leading to a lower level of default risk, it decreases the liquidation value of the firm’s assets because the cash is trapped abroad. In addition, a high degree of international tax planning may also be indicative of high agency costs and low financial reporting quality, issues also relevant to debtholders. We thus expect a positive relation between potential earnings repatriation taxes and loan spreads because debtholders will require a premium to compensate for the additional risk they bear.

Using U.S. MNCs loan data, we find that banks charge higher interest spreads to borrowers with higher potential repatriation taxes, and this effect is more pronounced for borrowers with low profitability or facing financial constraint problems. In 2017, the United States implemented the Tax Cuts and Jobs Act (TCJA). As part of the Act, U.S. MNCs were allowed to repatriate their foreign earnings at a reduced tax rate, leading to a larger benefit for firms with high foreign profits “permanently re-invested” overseas. We use the TCJA as a quasi-natural experiment and find that interest spreads of firms with high potential repatriation taxes decrease more over the period covered by the Act.

We also investigate the effect of potential repatriation taxes on the use of other loan terms in debt contracts. We find that the debt contracts of firms with high potential repatriation taxes contain more financial covenants and are more likely to require collateral from borrowers, consistent with international tax planning issues extending beyond pure possible future cash outflows. Additionally, we find that MNCs with high potential repatriation taxes will be more likely to borrow from foreign banks, consistent with U.S. borrowers preferring to use “trapped” foreign cash to pay foreign interest expense.

Our results hold after controlling for measures of firm-level overall tax avoidance previously found to affect debt contracting (Hasan et al., 2014), and interestingly, we find that our effect is concentrated in firms with low levels of general tax avoidance. This result suggests that our measure of borrowers’ potential repatriation taxes contains information that is incrementally useful to banks beyond that in overall firm-level tax avoidance. In addition to the negative effect that repatriation taxes may have on cash flow (see e.g., Nessa, 2017), we also identify agency risk and information risk as two plausible channels through which potential repatriation taxes will affect bank loan spreads, even though the specific issues are different from domestic or overall tax avoidance activities, and we find that our results are driven by borrowers with (1) fewer independent directors, (2) lower levels of institutional ownership, (3) lower analyst coverage, (4) no issuance of management forecasts, (5) lower share turnover, and (6) higher discretionary accruals. Our results are also robust to different measurement specifications and to controlling for geographic concentration, investment opportunities, liquidity needs, and to including deferred tax assets and liabilities.

Our study makes several contributions to the literature. First, we add to the literature investigating how cross-country differences in institutional factors, in our case the tax rate, tax rate uncertainty related to the potential for regulatory change, and political changes to the tax system, affect the design of bank loan contracts. Prior international business studies, such as Li et al. (2011), find that increased globalization leads to a reduction in interest rate spreads but unfavorable non-price terms, such as short maturities or restrictive collateral requirements, for a sample of U.S. MNCs. Cross-country studies have shown that bank loan contracts are affected by institutional factors like judicial protection of property rights (Laeven & Majnoni, 2005) and creditor rights protection (e.g., Esty & Megginson, 2003; Qian & Strahan, 2007; Bae & Goyal, 2009). Despite some evidence on the effects of how tax avoidance affects debt pricing in the U.S. and China (Isin, 2018; Beladi et al., 2018), how banks set loan contracts specifically for multinational firms with different foreign affiliates in the context of international tax planning is less studied. While prior studies document some of the benefits that arise from tax planning and tax avoidance activities, our results indicate that banks (negatively) consider potential repatriation taxes when designing debt contracts for U.S. MNCs. Our study thus contributes to the international business literature investigating how operation of MNCs and differences in tax laws affect bank loan contracting.

Second, tax avoidance is an important consideration in capital markets, and we contribute to the understanding of how tax incentive and tax avoidance activities affect firms and capital providers’ behavior (Kim et al., 2011a; Lisowsky, 2010). Ma et al. (2020) show that the credit rating agencies assign a lower credit rating to firms with a higher degree of international tax planning activities, Hasan et al. (2014) find that overall tax avoidance affects bank loan deals, and Blaylock et al. (2022) show that the public debt market assigns a greater credit spread to foreign earnings when firms face greater repatriation costs.3 We show that repatriation tax costs have an additional effect on bank loan contracts, both price and non-price terms, beyond overall tax avoidance. Providing this evidence using a sample of syndicated loans, where contracts are tailor-made based on borrower and lender characteristics and preferences (e.g., Ma et al., 2022), it provides more convincing evidence of the role that potential repatriation taxes play in debt contracting. We view these results as providing a partial answer to J. R. Graham et al. (2012)’s call for further research demonstrating how non-equity market participants use the tax information contained in the financial statements. Our finding that U.S. MNCs are more likely to seek foreign financing when international tax planning activities are high is consistent with MNCs attempting to access the foreign cash that has accumulated as part of a tax avoidance strategy by borrowing from (and paying interest to) foreign banks using foreign cash holdings. Our study thus extends the literature of how tax incentives influence where multinationals locate their debt (e.g., Newberry & Dhaliwal, 2001; Mills & Newberry, 2004) and how financial accounting affects firms’ real decisions (e.g., J. R. Graham et al., 2011).

In the next section we develop our hypotheses. We describe the sample selection procedures and variables in Section 3. Section 4 presents the main empirical results and Section 5 the results of additional analyses. A summary and conclusions are provided in Section 6.

2. Hypothesis Development

2.1. U.S. Tax System and Potential Earnings Repatriation Taxes

Prior research and organizations such as the Organization for Economic Cooperation and Development (OECD) generally classify tax systems of countries as either worldwide or territorial. Among OECD countries, the United States has the third highest combined marginal corporate income tax rate among the 188 countries surveyed, behind the United Arab Emirates and Puerto Rico (Tax Foundation, 2016). The U.S. also had a worldwide tax system during our sample period which means that American companies are taxed on a global basis. After considering the foreign income taxes already paid, U.S. MNCs, in general, still owe tax to the United States government.

For MNCs involving multiple countries, U.S. tax on profit generated in foreign subsidiaries is deferred until those profits are repatriated.4 In other words, if the MNCs leave the money abroad, they can avoid paying the tax. As a result, U.S. MNCs actively engage in international tax planning activities such as shifting income out of the United States to foreign low tax rate jurisdictions (Klassen & Laplante, 2012), leaving a large amount of money abroad.

Repatriation taxes can be substantial.5 For example, in April of 2015, GE announced the repatriation of $36 billion in foreign cash, triggering $6 billion in taxes (McKinnon & Hoffman, 2015). The exact calculation of U.S. tax owed is further complicated when foreign operations are located in multiple jurisdictions with different statutory tax rates. Generally, the residual tax due is equal to any excess of the U.S. tax rate over the weighted average tax rate of the relevant foreign jurisdictions.6 Additionally, U.S. domestic tax is reduced by foreign tax credits associated with foreign income taxes paid on foreign earnings.

U.S. Generally Accepted Accounting Principles (GAAP) allows firms to include the current period earnings of foreign subsidiaries in parents’ net income, even when the foreign income has been designated as permanently reinvested. So, MNCs retain the financial statement benefit of reporting foreign subsidiaries’ earnings without needing to pay the associated U.S. tax on these earnings (APB Opinion 23).7 The economic significance of the un-repatriated foreign earnings caused by financial reporting and tax reporting incentives of U.S. MNCs is enormous. Furthermore, N. Chen et al. (2023) provide evidence that investors misprice foreign cash holdings. According to Capital Economics (2016), U.S. MNCs hold approximately $2.5 trillion in cash overseas, an increase of 20% over the last two years, and the largest amount ever recorded. The incentive of U.S. MNCs to keep foreign earnings abroad has increased with the rise in the foreign income of MNCs and the decline in foreign statutory tax rates relative to the U.S. statutory tax rate over time (Bates et al., 2009; J. R. Graham et al., 2010, 2011; J. L. Blouin et al., 2012, 2014).

The recent U.S. 2018 Tax Cuts and Jobs Act includes several features that change the incentives of profit shifting, such as lowering the statutory tax rate from 35 to 21 percent, and new international provisions. Theoretically, firms will have incentives to shift income into lower tax jurisdictions if the foreign tax rates fall below U.S. rates. As shown in Clausing (2020), however, the issue of repatriation tax costs will continue to loom large following the passage of the Tax Cuts and Jobs Act Tax Reform. For example, a reduction in the tax rate to 21 percent might not be enough to encourage inbound profit shifting because most profit shifting activity (95 percent in 2015) occurs in countries with tax rates below the global minimum tax rate. Moreover, given the statements that President Trump has made about further modifying the tax code related to earnings repatriation, we expect our research question to be relevant, and perhaps even more important, in the coming years.

2.2. Effect of Potential Repatriation Taxes on Loan Deals

Bank loans are a dominant source of external financing for both public and private firms in the U.S. (Faulkender & Petersen, 2006; Sufi, 2007; Bharath et al., 2008; J. Graham et al., 2008) and around the world (Qian & Strahan, 2007; Kim et al., 2011b). Additionally, banks maintain privileged access to borrowers’ private information and hold concentrated positions, leading to a greater ability to evaluate the cash flow consequences of tax considerations and a stronger incentive to monitor borrowers’ credit quality than holders of “arm’s length” debt such as bonds (Diamond, 1984; Rajan, 1992; Santos & Winton, 2008). Given the unique position that banks hold as sophisticated insiders, it is not clear how (and to what extent) they will consider potential earnings repatriation tax costs when designing debt contracts.

The main tool that lenders employ to address a borrower’s risk is interest spread. There are several reasons to expect that repatriation tax costs will affect borrowers’ risk. First, engaging in international tax planning activities by choosing to keep foreign profits overseas increases current period after-tax earnings/cash flows, thus reducing risk of default; but it also increases the uncertainty of future cash flows. Under U.S. international tax laws and financial reporting rules in our sample period, there will be a cash payment for repatriation taxes and a reduction in reported accounting earnings for firms when they repatriate foreign earnings (J. L. Blouin et al., 2014). Also, the elevated tax risk caused by the increased probability of tax authority audits and penalties also increase the uncertainty of future cash flow (Mills, 1998; Hasan et al., 2014). In addition, Nessa (2017) finds that higher repatriation taxes are associated with decreased dividend payments, consistent with potential repatriation taxes putting downward pressure on cash flows, a key consideration for lenders.

Second, high potential earnings repatriation tax costs may be indicative of the agency risk between management and firm stakeholders. Recent research argues that tax planning activities can facilitate managerial opportunism (e.g., Desai et al., 2003; Desai & Dharmapala, 2006; Dhaliwal et al., 2011) and the issue is unique and more serious in an international setting, given the rent extraction opportunities by managers that arise from the complex tax structures of U.S. MNCs (Klassen & Laplante, 2012; Hanlon et al., 2015; Edwards et al., 2016; Harford et al., 2017). For example, Hanlon et al. (2015) provide evidence that MNCs with greater repatriation tax costs make foreign acquisitions worse.

Additionally, to the extent that large foreign cash balances have been built up by MNCs seeking primarily to avoid corporate tax (as opposed to other non-tax purposes), then the disclosure level as well as financial statement quality may be lower. Prior research provides mixed evidence on the overall tax avoidance and disclosure level and reporting quality. For example, Desai et al. (2004) and Desai and Dharmapala (2006) argue that tax avoidance is often associated with opaque reporting and a lack of financial statement transparency. Hanlon (2005) finds that tax avoidance is negatively associated with earnings persistence, and Frank et al. (2009) show that tax avoidance is positively associated with discretionary accruals. On the other hand, Gallemore and Labro (2015) find that the quality of the internal information environment is better for firms with greater tax avoidance, and Lennox et al. (2013) shows that tax aggressive U.S. public firms are less likely to commit accounting fraud. Balakrishnan et al. (2017) show that tax aggressive firms have greater total disclosure and tax-related disclosure in their financial statements and conference calls. The relationship is much clearer in the international tax planning case. For example, firms are required to disclose cumulative foreign earnings designated as permanently reinvested and the associated tax liability. However, Ayers et al. (2015) find that 77% of firms either do not disclose this tax liability or state that it is impractical to calculate. The Post-Implementation Review Report on SFAS No. 109 concluded that the information provided in the financial statements is not detailed enough for investors to fully understand the tax consequences of repatriating foreign earnings held abroad (FAF, 2013). As a consequence, the FASB reached a “Tentative Board Decision” to require additional disclosure related to foreign earnings (FASB, 2015). These factors are all found to be important considerations in bank’s lending decision (Bharath et al., 2008).

Last, different from domestic tax avoidance activities, engaging in international tax planning activities that leaves foreign earnings abroad, with high potential repatriation tax costs, decreases a firm’s asset liquidation value. Creditors cannot obtain access to the assets when firms default on promised payments and, thus, will require an additional payment (e.g., Benmelech et al., 2005). The trapped earnings also create friction in domestic capital markets and increase the domestic financing need (Altshuler & Grubert, 2003; De Simone & Lester, 2017). Overall, there is reason to believe that higher potential earnings repatriation taxes will be perceived by lenders to increase a borrower’s risk.

All these analyses lead to our first empirical hypothesis, stated as follows:

H1. 

Loan spreads will increase with the level of a borrower’s potential earnings repatriation tax costs.

Providing further complexity and tension for our research question, U.S. MNCs may be able to access foreign earnings and cash via complex tax planning strategies or wait for a tax policy change or holiday (see, e.g., Altshuler & Grubert, 2003; Grubert, 2004; Kleinbard, 2011). For example, Martin et al. (2015) find that firms brought home approximately $12 billion dollars a year tax-free (from 1990 to 2004) by taking advantage of complex tax-advantaged reorganizations. If MNCs are able to access foreign cash without paying onerous taxes or just save the money for precautionary reasons (Bates et al., 2009), then it is unclear to what extent banks will discount borrowers’ foreign earnings held overseas and how this will map into interest spread.

To assess the total effect on the contract design of potential repatriation taxes, it is important to consider the many different contract provisions from which lenders can choose (Gigler et al., 2009). In additional to interest spreads, banks can also change other terms to enhance their monitoring effect. Aside from changing the interest rate, requiring collateral and including financial covenants are the tools most commonly employed by private lenders when designing debt contracts (Stice, 2018). For example, Bharath et al. (2008) find that both the interest spread and collateral requirements are more stringent for borrowers with poor accounting quality. J. Graham et al. (2008) and P. F. Chen et al. (2016) provide evidence that loans initiated after restatements and modified audit opinions are more likely to be collateralized and contain more covenants. Therefore, we predict that banks will be more likely to require collateral and include more financial covenants in debt contracts in order to monitor borrowers facing higher levels of potential repatriation tax costs. Formally, we predict:

H2. 

Banks are more likely to require collateral and financial covenants in debt contracts for firms with high potential earnings repatriation tax costs.

Last, we consider whether firms with high-level international tax planning activities attempt to access the foreign debt markets in order to utilize the foreign earnings held overseas. Newberry and Dhaliwal (2001) find that U.S. MNCs issue bonds through their foreign subsidiaries in order to deduct the interest expense from foreign income, and Mills and Newberry (2004) find that foreign MNCs facing low foreign tax rates issue more debt through their U.S. subsidiaries. These papers both provide evidence that firms incorporate tax considerations when choosing where to recognize interest deductions.

However, borrowers that issue debt in a foreign country, as opposed to domestically, also face additional costs. Information asymmetry is higher when a firm borrows abroad and the presence of multiple legal and tax environments makes monitoring more difficult and expensive (Doukas & Pantzalis, 2003). Empirically, it is unclear whether the tax benefits of borrowing abroad are sufficient to overcome the non-tax costs. Stated in the null form, this leads to our final hypothesis:

H3. 

Potential earnings repatriation tax costs of U.S. multinationals will not influence the likelihood that a firm borrows from foreign banks.

3. Research Design and Sample Selection

3.1. Research Design

To empirically test whether potential earnings repatriation taxes are associated with interest spreads (H1), we estimate the following model:

Interest Spread = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε

(1)

where Repatriation Taxes is a measure of a borrower’s potential earnings repatriation taxes, and is calculated as the maximum of 0 and pre-tax foreign income (PIFO) multiplied by the U.S. statutory corporate tax rate (35% after 1993, 34% otherwise) less any foreign taxes paid (TXFO), and scaled by total assets (AT) (see e.g., Foley et al., 2007; Hanlon et al., 2015; Bird et al., 2015).8 This variable is calculated using publicly available data from firms’ financial statements and captures the incremental U.S. tax due when earnings (cash) is repatriated from foreign subsidiaries during our sample period.9 We calculate repatriation taxes using the most recently available financial statement data released before the loan issuance date of the debt contracts we examine to ensure lenders would have the data available to them.10

We select this measure for several compelling reasons. First, intuitively, it effectively captures the tax savings U.S. multinationals accrue by deferring or potentially permanently avoiding U.S. tax payments through the retention of foreign earnings abroad during our sample period. Second, as a firm-year level metric, it aggregates the effects of various international tax planning strategies, providing a comprehensive indicator of firms’ tax deferral behavior. Prior research supports its validity, demonstrating that this proxy is positively correlated with overseas cash holdings and aligns closely with measures derived from Bureau of Economic Analysis data (Foley et al., 2007; Hanlon et al., 2015). Third, even under the revised tax regime, the “statutory-minus-effective tax rate difference” remains a relevant and reliable measure for capturing the benefits associated with the permanent reinvestment of foreign profits, despite potential changes in the specific calculation details, such as statutory tax rates. This robustness underscores the measure’s utility as an indicator of firms’ tax deferral and reinvestment incentives.

Interest Spread is the all-in-drawn-spread reported by Dealscan, which is the interest rate on the loan and reported as the number of basis points over LIBOR. In order to eliminate the possibility that our results are driven by firms whose foreign effective tax rates are below the US tax rate, we conduct all of our empirical tests using the full sample and also the subsample where firms’ repatriation costs are greater than zero.

We control for other company-level and facility-level factors related to interest spreads and our other variables of interest. We control for firm size, because small firms have greater information asymmetry and higher default risk (Fortin & Pittman, 2007; Bharath et al., 2008). We include a number of controls related to financial distress found in the prior literature: leverage, current ratio, market-to-book, return-on-assets, Z-score, tangibility, and sales growth (J. Graham et al., 2008). We include a borrower’s book-tax difference to control for the effect of general tax avoidance (we further differentiate this effect in Section 5.2). BT is the book-tax difference for a firm in a given year (Manzon & Plesko, 2002). We also include the percentage of foreign operations in all regressions because banks may consider a high proportion of foreign operations to be risky.11 Additionally, we control for revolvers because these loans have a lower loan spread than term loans (Zhang, 2008), and we control for whether or not a loan is an institutional loan; these loan types have a higher interest spread because of the higher information symmetry with the borrower. We control for the existence of performance pricing provisions, because they reduce adverse selection and moral hazard costs for lenders (Asquith et al., 2005). Last, we control for other contracting devices available to lenders: loan maturity, collateral, number of financial covenants, loan purpose, and year fixed effects (Booth, 1992; Beatty et al., 2002; P. F. Chen et al., 2016). Detailed variable definitions are included in the Appendix A.

To test H2, we estimate the effect of potential earnings repatriation taxes on the likelihood of requiring collateral and the number of financial covenants included in the debt contract. For the collateral test, we estimate the following logit model:

Prob(Collateral = 1) = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε

(2)

where Collateral is an indicator variable equal to one if the loan is backed by collateral, and zero otherwise. We employ an OLS regression to estimate the financial covenant model:

Financial Covenants = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε

(3)

where Financial Covenants is the number of financial covenants included in each loan facility. Our final hypothesis (H3) examines whether potential earnings repatriation taxes are related to borrowers’ use of foreign banks. To test H4, we estimate the following logit model:

Prob(Foreign Bank = 1) = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε

(4)

where Foreign Bank is an indicator variable equal to one if a member of the loan syndicate is a foreign bank, and zero otherwise (Ma et al., 2019a).

3.2. Data Sources and Sample Selection

We obtain loan facilities data from Thomson Reuters LPC Dealscan database, and financial accounting data from Standard and Poor’s Compustat North America Fundamental database. We use the Dealscan-Compustat link constructed and maintained by Michael Roberts and Wharton Research Data Services (WRDS) to merge loans from Dealscan with the fiscal years in which they were issued on Compustat (see Chava et al., 2008). We delete firms without pre-tax foreign income (pifo) or foreign income tax (txfo). We also delete firm years with negative pre-tax income because their potential earnings repatriation taxes are more difficult to reliably estimate (Klassen & Laplante, 2012). Firms in the financial services and utilities industries are also eliminated. After eliminating observations with missing values of our control variables, our final sample includes 7485 loan facilities issued to 1618 borrowers during the period 1988 to 2012.12

Table 1. Descriptive statistics.

Panel A: Sample Distribution by Year
Year	Frequency	Percent	Year	Frequency		Percent
1988	16	0.21	2001	440		5.88
1989	153	2.04	2002	337		4.5
1990	161	2.15	2003	364		4.86
1991	128	1.71	2004	436		5.82
1992	126	1.68	2005	497		6.64
1993	163	2.18	2006	464		6.2
1994	289	3.86	2007	501		6.69
1995	282	3.77	2008	252		3.37
1996	349	4.66	2009	111		1.48
1997	372	4.97	2010	255		3.41
1998	354	4.73	2011	533		7.12
1999	300	4.01	2012	229		3.06
2000	373	4.98	Total	7485		100
Panel B: Descriptive Statistics
Variable	N	Mean	Std.	Q1	Median	Q3
SPREAD	7485	143.29	111.54	50.00	125.00	200.00
FOREIGN	7485	0.73	0.44	0.00	1.00	1.00
Repatriation Taxes	7485	0.39	0.66	0.00	0.08	0.50
FCOVENANT	7485	1.20	1.31	0.00	1.00	2.00
SIZE	7485	7.37	1.71	6.19	7.41	8.59
LEVERAGE	7485	0.21	0.16	0.09	0.19	0.29
CURRENT	7485	1.97	0.99	1.30	1.74	2.38
MB	7485	1.86	0.97	1.24	1.56	2.15
ROA	7485	0.07	0.05	0.03	0.06	0.09
ZSCORE	7485	2.08	0.88	1.49	2.03	2.60
TANGIBILITY	7485	0.27	0.19	0.13	0.23	0.36
SalesGrowth	7485	0.16	0.26	0.02	0.10	0.22
ForeignOp	7485	0.57	0.78	0.13	0.38	0.77
BT	7485	0.01	0.04	−0.01	0.01	0.03
LOG(AMOUNT)	7485	18.98	1.52	18.13	19.11	20.03
LOG(MATURITY)	7485	3.65	0.70	3.22	4.09	4.09
INS_IN	7485	0.06	0.24	0.00	0.00	0.00
REVOLVER	7485	0.59	0.49	0.00	1.00	1.00
COLLATERAL	7485	0.38	0.48	0.00	0.00	1.00
PPINDEX	7485	0.47	0.50	0.00	0.00	1.00

Table 1 presents the descriptive statistics for the variables used in the analyses. See the Appendix A for variable definitions.

reports the annual distribution of loans over our sample period. As shown in the table, the number of loans generally increases over the time period from 1988 until 2012. A notable decrease in the number of loans takes place starting in 2007, at the beginning of the financial crisis, though the number of loans increases again beginning in 2010.

3.3. Descriptive Statistics

Table 1 Panel B presents the descriptive statistics of loans in our sample, and

Table 2. Correlation matrix.

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)	(17)	(18)	(19)
(1)	SPREAD
(2)	FOREIGN	−0.178 ***
(3)	Repatriation Taxes	0.009	0.078 ***
(4)	FCOVENANT	0.251 ***	−0.024 **	0.005
(5)	SIZE	−0.371 ***	0.440 ***	0.086 ***	−0.256 ***
(6)	LEVERAGE	0.134 ***	0.122 ***	−0.097 ***	0.056 ***	0.085 ***
(7)	CURRENT	0.158 ***	−0.215 ***	0.068 ***	0.094 ***	−0.400 ***	−0.168 ***
(8)	MB	−0.197 ***	0.009	0.197 ***	−0.056 ***	0.025 **	−0.180 ***	0.040 ***
(9)	ROA	−0.155 ***	−0.017	0.270 ***	−0.076 ***	−0.060 ***	−0.286 ***	0.150 ***	0.584 ***
(10)	ZSCORE	−0.163 ***	−0.102 ***	0.026 **	−0.008	−0.196 ***	−0.354 ***	0.225 ***	0.238 ***	0.427 ***
(11)	TANGIBILITY	−0.157 ***	0.091 ***	−0.008	−0.151 ***	0.187 ***	0.202 ***	−0.228 ***	−0.126 ***	−0.053 ***	−0.184 ***
(12)	SalesGrowth	0.046 ***	−0.039 ***	0.051 ***	0.043 ***	−0.141 ***	0.018	0.060 ***	0.155 ***	0.082 ***	0.015	−0.027 **
(13)	ForeignOp	0.062 ***	0.056 ***	0.103 ***	0.007	0.052 ***	0.020 *	−0.053 ***	−0.016	−0.099 ***	−0.165 ***	0.005	0.060 ***
(14)	BT	0.019 *	−0.057 ***	−0.196 ***	0.020 *	−0.048 ***	0.021 *	0.045 ***	0.022 *	0.203 ***	−0.073 ***	0.118 ***	−0.018	0.020 *
(15)	LOG(AMOUNT)	−0.345 ***	0.452 ***	0.094 ***	−0.120 ***	0.764 ***	0.077 ***	−0.304 ***	0.078 ***	0.026**	−0.100 ***	0.141 ***	−0.116 ***	0.018	−0.047 ***
(16)	LOG(MATURITY)	0.172 ***	0.079 ***	0.013	0.200 ***	−0.104 ***	0.140 ***	0.071 ***	−0.073 ***	−0.026**	−0.059 ***	−0.022 *	0.014	−0.003	0.005	−0.002
(17)	INS_IN	0.327 ***	−0.013	−0.011	0.165 ***	−0.023 **	0.158 ***	0.003	−0.044 ***	−0.073 ***	−0.117 ***	−0.062 ***	0.015	0.032 ***	0.011	0.028 **	0.222 ***
(18)	REVOLVER	−0.125 ***	−0.030 ***	0.003	0.042 ***	−0.132 ***	−0.081 ***	0.082 ***	−0.039 ***	0.024 **	0.079 ***	−0.01	0.008	0	−0.009	−0.027 **	0.316 ***	−0.312 ***
(19)	COLLATERAL	0.548 ***	−0.114 ***	−0.018	0.406 ***	−0.349 ***	0.148 ***	0.160 ***	−0.128 ***	−0.116 ***	−0.095 ***	−0.141 ***	0.097 ***	0.033 ***	0.032 ***	−0.275 ***	0.245 ***	0.307 ***	−0.049 ***
(20)	PPINDEX	−0.066 ***	0.109 ***	−0.002	0.533 ***	−0.035 ***	0.016	0.002	0.007	−0.017	0.024 **	−0.050 ***	−0.01	−0.011	−0.006	0.108 ***	0.170 ***	−0.077 ***	0.164 ***	0.124 ***

Table 2 presents the Pearson correlation matrix. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively. See the Appendix A for variable definitions.

provides a correlation matrix. The loans in our sample have a mean spread above LIBOR of 143.29 basis points and the average Repatriation Taxes is 0.0039 (it is multiplied by 100 for ease of presentation and interpretation). The spread is lower than the roughly 200 basis point average spread of all loans included in the Dealscan database over our sample period and the Repatriation Taxes is larger than the average of Compustat firms (Foley et al., 2007), consistent with U.S. MNCs in Dealscan being larger than the typical firms in Dealscan and Compustat. The mean (median) loan size is $174.95M ($199.24M) and the average loan maturity is 38.47 months. In total, 73% of the loan facilities include a foreign bank. The average number of financial covenants included in a debt contract is 1.20. Approximately half (47%) of the loans include a performance pricing provision, 38% require collateral, and 59% are a revolver.

Table 3. The effect of potential earnings repatriation taxes on loan interest spreads.

Dependent Variable=	Full Sample	Repatriation Taxes > 0 Sample
	SPREAD	SPREAD
	(1)	(2)
Repatriation Taxes	5.61 ***	5.13 **
	(2.68)	(2.12)
FCOVENANT	3.63 **	3.76 **
	(2.48)	(2.05)
SIZE	−16.34 ***	−15.08 ***
	(−10.27)	(−7.77)
LEVERAGE	42.79 ***	47.43 ***
	(3.91)	(3.76)
CURRENT	−4.25 ***	−4.53 **
	(−2.84)	(−2.42)
MB	−7.47 ***	−8.99 ***
	(−4.47)	(−4.45)
ROA	−95.63 **	−66.58
	(−2.55)	(−1.36)
ZSCORE	−11.61 ***	−10.25 ***
	(−6.59)	(−4.89)
TANGIBILITY	−24.75 ***	−18.96 **
	(−3.01)	(−2.06)
SalesGrowth	8.79 *	5.64
	(1.83)	(0.87)
ForeignOp	1.83	1.17
	(1.06)	(0.58)
BT	8.90	−22.22
	(0.30)	(−0.56)
LOG(AMOUNT)	−9.37 ***	−10.42 ***
	(−6.16)	(−5.32)
LOG(MATURITY)	−2.70	−1.07
	(−1.24)	(−0.35)
INS_IN	62.81 ***	67.91 ***
	(10.57)	(9.53)
REVOLVER	−12.00 ***	−11.23 ***
	(−4.82)	(−3.52)
COLLATERAL	59.28 ***	54.18 ***
	(17.50)	(13.59)
PPINDEX	−26.18 ***	−23.80 ***
	(−8.54)	(−6.37)
Loan Purpose FE	Included	Included
Year FE	Included	Included
Constant	547.77 ***	555.91 ***
	(17.20)	(11.01)
Observations	7485	4542
R-squared	0.595	0.607

Table 3 presents the results from the estimation of the following model: Interest Spread = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. All variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

presents the correlation matrix of variables in our sample. As expected, many of the contracting terms are significantly correlated with each other. Interest spread is positively correlated with the number of financial covenants, leverage, current ratio, sales growth, loan maturity, whether the loan is institutional, and with the requirement of collateral, and it is negatively correlated with firm size, market-to-book, profitability, Z-score, tangibility, loan amount, whether the loan is a revolver, and whether the loan contract includes performance pricing provisions. These univariate results are consistent with lenders having many different loan terms to negotiate with borrowers (Melnik & Plaut, 1986), not just interest spreads.

4. Empirical Results

4.1. International Tax Planning and Loan Interest Spreads

Table 3 presents the OLS regression results of Equation (1). We regress interest spread on potential earnings repatriation taxes and a set of control variables. We estimate the regression using the full sample and also using the subsample of borrowers that have positive potential earnings repatriation taxes. Our first hypothesis predicts that repatriation taxes will be viewed by lenders as increasing a borrower’s credit risk and, in turn, positively associated with interest spreads. In Column 1 (2), the coefficient of 5.61 (5.13) on Repatriation Taxes is statistically significant. The economic significance of the coefficients indicates that one standard deviation of change in Repatriation Taxes will lead to a change in spread of about 3.70 basis points (based on the coefficient in Column 1), translating into a 2.6% increase in borrowing cost.

Many of the included control variables are statistically significant. Interest spreads are negatively associated with firm size, current ratio, market-to-book, ROA, Z-score, tangibility, loan size, loan maturity, whether or not the loan is a revolver, and the inclusion of a performance pricing provision. Interest spreads are positively associated with the inclusion of financial covenants, leverage, whether or not the loan is institutional, and whether or not the borrower provides collateral.

4.2. Moderating Effect of Financial Distress

Firms with higher profitability and less financial constraint will be less likely to use foreign earnings held abroad, thus incurring the tax cost. If a U.S. MNC can generate enough cash or raise capital with ease, then it will not need to use foreign earnings to fund domestic operations and investments. For example, Apple Inc. has famously borrowed extensively to fund share repurchases and dividend payments, while having a large amount of cash and short-term securities overseas ($216 billion in 2016 according to Apple Inc.’s 10-K). However, firms with low profitability or that are financially constrained will be unable to raise capital at low cost, and they will be more likely to repatriate earnings from foreign subsidiaries in order to fund domestic investment opportunities or normal operations. We thus expect the relation tested in H1 to be stronger when a borrower’s profitability is lower or when they are financially constrained. We partition our sample at the annual medians of profitability and the Kaplan and Zingales (1997) financial constraint index and re-estimate Equation (1) in each subsample.

Table 4. Potential earnings repatriation taxes and loan spreads: cross-sectional analysis on profitability.

	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	Low Profitability	High Profitability	Low Profitability	High Profitability
	(1)	(2)	(3)	(4)
Repatriation Taxes	9.43 **	2.12	12.32 ***	0.95
	(2.40)	(0.93)	(2.81)	(0.34)
FCOVENANT	4.22 **	1.86	4.30 *	2.92
	(2.14)	(0.96)	(1.70)	(1.18)
SIZE	−18.72 ***	−16.23 ***	−14.75 ***	−16.85 ***
	(−8.09)	(−7.95)	(−5.50)	(−6.39)
LEVERAGE	61.03 ***	41.73 ***	56.15 ***	41.24 ***
	(3.27)	(3.43)	(2.68)	(2.59)
CURRENT	−6.89 ***	−3.05	−4.41	−5.44 **
	(−2.96)	(−1.51)	(−1.55)	(−2.07)
MB	−10.89 **	−6.33 ***	−15.89 ***	−6.67 ***
	(−2.47)	(−3.69)	(−3.02)	(−3.08)
ROA	−203.33 ***	53.80	−223.25 ***	74.38
	(−2.97)	(1.26)	(−2.93)	(1.19)
ZSCORE	−13.37 ***	−8.82 ***	−12.40 ***	−9.80 ***
	(−5.12)	(−3.88)	(−3.92)	(−3.44)
TANGIBILITY	−34.68 ***	−6.75	−30.01 **	−4.21
	(−2.70)	(−0.73)	(−2.10)	(−0.36)
SalesGrowth	7.84	8.82	6.37	5.59
	(1.25)	(1.26)	(0.75)	(0.58)
ForeignOp	0.75	3.80	0.45	3.28
	(0.36)	(1.37)	(0.20)	(0.84)
BT	93.30 *	−22.81	83.77	−53.98
	(1.83)	(−0.61)	(1.26)	(−1.13)
LOG(AMOUNT)	−8.07 ***	−9.65 ***	−9.88 ***	−10.11 ***
	(−3.74)	(−4.82)	(−3.63)	(−3.71)
LOG(MATURITY)	−2.34	−2.02	0.55	−1.00
	(−0.69)	(−0.73)	(0.12)	(−0.26)
INS_IN	51.94 ***	71.94 ***	58.21 ***	73.17 ***
	(6.31)	(9.12)	(6.38)	(6.71)
REVOLVER	−13.33 ***	−10.85 ***	−10.05**	−12.63 ***
	(−3.54)	(−3.52)	(−2.19)	(−2.98)
COLLATERAL	59.03 ***	55.96 ***	56.46 ***	48.07 ***
	(12.24)	(12.61)	(10.53)	(8.64)
PPINDEX	−32.65 ***	−18.32 ***	−29.21 ***	−17.71 ***
	(−6.84)	(−5.19)	(−5.03)	(−3.97)
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	560.14 ***	515.10 ***	487.35 ***	613.09 ***
	(11.90)	(12.91)	(9.52)	(15.21)
Observations	3757	3728	2285	2257
R-squared	0.564	0.632	0.567	0.649

Table 4 presents the results from the estimation of the following model: Interest Spread = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. And regressions are partitioned on median of Profitability (EBITDA/AT). All variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

and

Table 5. Potential earnings repatriation taxes and loan spreads: cross-sectional analysis on borrowers’ financial constraint.

	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	Unconstrained	Constrained	Unconstrained	Constrained
	(1)	(2)	(3)	(4)
Repatriation Taxes	2.00	8.66 ***	3.15	7.01 **
	(0.74)	(2.69)	(1.19)	(2.04)
FCOVENANT	1.75	4.33 **	1.22	5.40 *
	(0.79)	(2.07)	(0.43)	(1.93)
SIZE	−13.92 ***	−19.40 ***	−11.81 ***	−18.44 ***
	(−5.79)	(−9.58)	(−4.11)	(−6.81)
LEVERAGE	37.18 **	59.06 ***	25.01	77.80 ***
	(2.09)	(4.19)	(1.55)	(3.97)
CURRENT	−3.68 **	−7.37 ***	−3.94 *	−8.54 **
	(−2.15)	(−2.76)	(−1.91)	(−2.44)
MB	−7.19 ***	−12.55 ***	−9.11 ***	−12.74 ***
	(−3.43)	(−4.84)	(−3.69)	(−3.72)
ROA	29.42	−158.66 ***	9.78	−103.38
	(0.61)	(−2.80)	(0.16)	(−1.49)
ZSCORE	−11.07 ***	−11.52 ***	−9.99 ***	−9.41 ***
	(−4.54)	(−4.48)	(−3.55)	(−3.00)
TANGIBILITY	−79.35 ***	−17.25 *	−64.44 ***	−12.15
	(−4.34)	(−1.74)	(−3.75)	(−0.95)
SalesGrowth	10.65	6.18	8.21	9.05
	(1.59)	(1.04)	(0.90)	(1.04)
ForeignOp	3.57	0.43	0.03	0.13 **
	(1.11)	(0.22)	(0.83)	(2.04)
BT	5.31	−3.52	−7.91 ***	−5.69
	(0.13)	(−0.08)	(−3.22)	(−1.29)
LOG(AMOUNT)	−11.98 ***	−5.99 ***	−12.87 ***	−6.48 **
	(−5.06)	(−3.20)	(−4.31)	(−2.51)
LOG(MATURITY)	0.42	−7.84**	2.52	−6.96
	(0.13)	(−2.34)	(0.59)	(−1.41)
INS_IN	73.89 ***	58.46 ***	82.87 ***	59.65 ***
	(8.86)	(7.27)	(7.65)	(6.38)
REVOLVER	−10.53 ***	−12.21 ***	−11.27 **	−10.25 **
	(−3.05)	(−3.67)	(−2.45)	(−2.30)
COLLATERAL	65.22 ***	52.50 ***	62.70 ***	47.30 ***
	(12.52)	(12.70)	(9.74)	(8.81)
PPINDEX	−22.18 ***	−24.54 ***	−18.19 ***	−25.09 ***
	(−5.34)	(−5.38)	(−3.48)	(−4.10)
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	558.65 ***	565.88 ***	525.90 ***	573.16 ***
	(12.21)	(14.48)	(10.09)	(13.69)
Observations	3433	3402	2063	2039
R-squared	0.629	0.573	0.653	0.567

Table 5 presents the results from the estimation of the following model: Interest Spread = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. And regressions are partitioned on the median of financial constraint status (Kaplan & Zingales, 1997 index). All variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

present cross-sectional tests of the effect of potential earnings repatriation taxes on interest spreads when considering the likelihood that a firm will need to use foreign earnings held abroad. Table 4 presents the OLS regression results from Table 3 after partitioning the samples on the median of Profitability (EBITDA over total assets). In both the full sample and subsample of borrowers with positive potential earnings repatriation taxes (Columns 1 and 3), the coefficient on Repatriation Taxes is only significant in the group whose profitability is below the annual median.13 Unsurprisingly, the coefficients of Repatriation Taxes are larger for borrowers with profitability below the median. The coefficients of 9.43 and 12.32 in Columns 1 and 3 indicate that borrowers with Profitability below the median will be issued loans with an interest spread increase about 6.22 and 8.13 basis points for a one standard deviation of change in Repatriation Taxes.

Table 5 presents similar findings to those in Table 4. The sample is partitioned at the median based on the Kaplan and Zingales (1997) financial constraint index. Columns 2(1) and 4(3) present the results of estimating the effect of potential earnings repatriation taxes on interest spreads for borrowers that are more(less) financially constrained. The coefficients on Repatriation Taxes are positive and significant only for firms that are relatively more financially constrained.14 In terms of economic significance, the results in Table 5 indicate that borrowers above the median of the Kaplan and Zingales index will be issued loans with an interest spread increase of 5.72 and 4.63 basis points for a one standard deviation of change in Repatriation Taxes, for the full and subsample, respectively. Overall, the results in Table 4 and Table 5 provide evidence that banks consider the likelihood that a borrower will need to repatriate earnings held abroad in order to satisfy domestic obligations, potentially including interest payments; and the relation between potential earnings repatriation taxes and interest spread is driven by firms with high financial constraints.

4.3. Repatriation Taxes and the Requirement of Collateral and Inclusion of Financial Covenants

Table 6. The effect of potential earnings repatriation taxes on the requirement of collateral and inclusion of financial covenants.

	Full Sample	Repatriation Taxes > 0 Sample	Full Sample	Repatriation Taxes > 0 Sample
Dependent Variable=	COLLATERAL	COLLATERAL	FCOVENANT	FCOVENANT
	(1)	(2)	(3)	(4)
Repatriation Taxes	0.16 **	0.21 ***	0.07 ***	0.10 ***
	(2.39)	(2.62)	(2.60)	(3.11)
SIZE	−0.52 ***	−0.53 ***	−0.14 ***	−0.15 ***
	(−10.26)	(−8.44)	(−8.04)	(−6.78)
LEVERAGE	2.37 ***	2.48 ***	0.04	0.23
	(7.04)	(5.67)	(0.27)	(1.26)
CURRENT	−0.01	−0.01	−0.04 **	−0.03
	(−0.12)	(−0.08)	(−2.02)	(−1.31)
MB	−0.24 ***	−0.22 ***	−0.03	−0.04
	(−3.97)	(−2.88)	(−1.35)	(−1.48)
ROA	−3.80 ***	−4.34 **	−1.37 ***	−1.49 **
	(−2.80)	(−2.46)	(−2.92)	(−2.34)
ZSCORE	−0.16 ***	−0.15 **	0.03	0.04
	(−2.65)	(−1.98)	(1.25)	(1.52)
TANGIBILITY	−1.05 ***	−1.23 ***	−0.33 ***	−0.36 ***
	(−3.87)	(−3.65)	(−3.35)	(−3.11)
SalesGrowth	0.83 ***	0.92 ***	0.04	−0.00
	(4.87)	(4.27)	(0.49)	(−0.01)
ForeignOp	0.04	0.03	−0.01	−0.04
	(0.69)	(0.45)	(−0.62)	(−1.34)
BT	1.98 *	3.06 **	0.56	0.68
	(1.75)	(2.13)	(1.45)	(1.42)
LOG(AMOUNT)	−0.24 ***	−0.23 ***	0.00	0.00
	(−5.27)	(−3.83)	(0.14)	(0.19)
LOG(MATURITY)	0.30 ***	0.33 ***	0.06 ***	0.06 **
	(4.43)	(3.57)	(2.80)	(2.06)
INS_IN	3.22 ***	3.31 ***	0.40 ***	0.34 ***
	(10.41)	(8.47)	(5.67)	(4.35)
REVOLVER	−0.02	−0.03	−0.02	−0.04
	(−0.33)	(−0.30)	(−0.82)	(−1.25)
COLLATERAL			0.53 ***	0.54 ***
			(13.01)	(10.78)
PPINDEX	0.71 ***	0.69 ***	1.08 ***	1.08 ***
	(7.84)	(6.13)	(29.97)	(24.70)
SPREAD			0.00 **	0.00 **
			(2.45)	(2.04)
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	6.91 ***	5.96 ***	0.33	0.02
	(6.38)	(4.27)	(0.87)	(0.04)
Observations	7485	4542	7485	4542
R-squared			0.518	0.515

Table 6 presents the results from the estimation of the following models: Prob(COLLATEAL = 1) = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. FCOVENANT = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. All variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. z (t)-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

reports the empirical results of estimating Equations (2) and (3), the test of H2. In Columns 1 and 2 of Table 6, we observe that the likelihood of being asked to provide collateral increases with the potential earnings repatriation taxes for borrowers in the full sample and subsample. The coefficients on Repatriation Taxes are 0.16 and 0.21 in Columns 1 and 2 and are statistically significant. This represents an increase in the likelihood of requiring collateral of 2.28% (3.32%) if Repatriation Taxes increases by one standard deviation. Given that Table 1 shows that the unconditional probability of requiring collateral is 38%, a one standard deviation increase in Repatriation Taxes increases the unconditional probability of requiring collateral by approximately 6.0% (=2.28%/38%). Columns 3 and 4 provide evidence that banks impose more financial covenants on borrowers with higher potential earnings repatriation taxes. The coefficients on Repatriation Taxes are 0.07 and 0.10 in Columns 3 and 4 and are also statistically significant. These coefficients indicate that potential earnings repatriation taxes will increase the use of financial covenants by 3.9% (5.5%) when Repatriation Taxes increases by one standard deviation. Overall, these results are consistent with H2 and indicate potential earnings repatriation taxes affect the broad design of debt contracts by banks, not only the interest rate a borrower obtains.

4.4. International Tax Planning and the Inclusion of a Foreign Bank in the Loan Syndicate

Our last formal hypothesis (H3) predicts that potential earnings repatriation taxes will increase the likelihood that a foreign bank is included in a borrower’s loan syndicate. This prediction arises from the desire of borrowers to pay interest, and more importantly deduct interest expense, outside of the U.S. in order to access trapped foreign earnings held abroad.

Table 7. The effect of potential earnings repatriation taxes on the inclusion of foreign banks in the loan syndicate.

	Full Sample	Repatriation Taxes > 0 Sample	Full Sample	Repatriation Taxes > 0 Sample
Dependent Variable=	FOREIGN	FOREIGN	FOREIGN	FOREIGN
	(1)	(2)	(3)	(4)
Repatriation Taxes	0.17 **	0.23 **	0.16 **	0.20 **
	(2.22)	(2.44)	(2.04)	(2.25)
FCOVENANT	0.06	0.02
	(1.35)	(0.42)
SIZE	0.40 ***	0.41 ***	0.80 ***	0.79 ***
	(7.97)	(6.41)	(18.88)	(14.68)
LEVERAGE	1.05 ***	0.70 *	1.23 ***	1.01 **
	(3.39)	(1.72)	(3.98)	(2.44)
CURRENT	−0.10 **	−0.08	−0.13 ***	−0.13 **
	(−2.18)	(−1.49)	(−3.07)	(−2.38)
MB	−0.07	−0.12	−0.08	−0.07
	(−1.17)	(−1.62)	(−1.45)	(−1.00)
ROA	1.34	3.53 **	1.30	0.25
	(1.10)	(2.08)	(1.20)	(0.22)
ZSCORE	−0.02	−0.04	−0.01	0.04
	(−0.31)	(−0.44)	(−0.17)	(0.54)
TANGIBILITY	−0.04	0.03	−0.36	−0.19
	(−0.15)	(0.09)	(−1.29)	(−0.52)
SalesGrowth	0.45 **	0.24	0.43 **	0.34
	(2.40)	(1.04)	(2.45)	(1.56)
ForeignOp	0.16 ***	0.10	0.12 **	0.08
	(2.79)	(1.48)	(2.18)	(1.21)
BT	−1.88 *	−3.34 **	−2.43 **	−2.09
	(−1.73)	(−2.42)	(−2.38)	(−1.58)
LOG(AMOUNT)	0.43 ***	0.40 ***
	(9.58)	(7.41)
LOG(MATURITY)	0.46 ***	0.46 ***
	(7.02)	(5.19)
INS_IN	−0.77 ***	−0.76 ***
	(−4.64)	(−3.76)
REVOLVER	−0.14 *	−0.06
	(−1.83)	(−0.59)
COLLATERAL	0.06	−0.04
	(0.57)	(−0.30)
PPINDEX	0.41 ***	0.50 ***
	(3.89)	(3.90)
SPREAD	−0.00 *	−0.00
	(−1.91)	(−0.91)
Loan Purpose FE	Included	Included
Year FE	Included	Included	Included	Included
Constant	−11.05 ***	−11.09 ***	−3.91 ***	−4.41 ***
	(−9.09)	(−9.29)	(−3.61)	(−7.89)
Observations	7485	4540	4737	2884

Table 7 presents the results from the estimation of the following model: Prob(FOREIGN = 1) = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. All variables are defined in the Appendix A. Columns 1 and 2 show results at facility level while columns 3 and 4 present results at firm-year level. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects at facility level and year fixed effect at firm level, Standard errors are heteroskedasticity robust and clustered at firm level. z-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

presents the logit regression results of Equation (4) which tests this prediction.

We first regress our variable of interest and all control variables on Foreign Bank using the full sample and subsample in which Repatriation Taxes is greater than 0. We find that for both full sample and subsample, the coefficient on Repatriation Taxes is positive and statistically significant, consistent with potential earnings repatriation taxes leading to the inclusion of foreign banks in the loan deal. The coefficients on Repatriation Taxes in Columns 1 and 2 indicate that potential earnings repatriation taxes increase the likelihood that a foreign bank is included in the syndicate by 1.8% (2.9%). Given that Table 1 shows that the unconditional probability of including a foreign bank is 73%, a one standard deviation increase in Repatriation Taxes increases the unconditional probability of including a foreign bank in the loan by approximately 2.5% (=1.8%/73%).

One potential concern with including all of the control variables from the previous tests is that the loan terms of the debt contract may be unknown when the lending syndicate if formed. Therefore, to mitigate concerns that this could be driving our results, we re-run our test at the firm-year level (instead of facility level), including only the control variables related to firm characteristics, and report the results in Columns 3 and 4 of Table 7. Omitting the loan term control variables does not change our inferences.

5. Additional Analyses

5.1. Difference-in-Differences Design: Using the American Jobs Creation Act of 2004 as a Shock

In 2004, the United States implemented the American Jobs Creation Act (AJCA). As part of the Act, U.S. MNCs were granted the option to repatriate their foreign earnings at a reduced tax rate over a two-year period. Thus, AJCA created a temporary tax holiday that reduced the U.S. tax rate on repatriations from foreign subsidiaries from 35% to 5.25%. S. Albring et al. (2005) estimated that 282 firms that repatriated under AJCA saved $46 billion, and Browning (2008) claims that 843 U.S. MNCs saved $265 billion in tax savings.15 Several studies have examined the effect of the Act and the factors that influence MNCs’ decision to repatriate foreign earnings held abroad (see e.g., J. Blouin & Krull, 2009; S. M. Albring et al., 2011; Faulkender & Petersen, 2012). J. Blouin and Krull (2009) find that firms with lower investment opportunities abroad were more likely to repatriate during the tax holiday, and S. M. Albring et al. (2011) find that firms with fewer borrowing constraints repatriated more foreign earnings held overseas. We use the Act as a quasi-natural experiment and predict that banks will view the Act as increasing borrowers’ credit quality because of the decreased cost of repatriating foreign earnings, and firms with high potential earnings repatriation taxes will benefit the most.

To test this prediction related to AJCA, we employ a difference-in-differences research design:

Interest Spread = β₀ + β₁ ∗ Matched Repatriation Taxes + β₂ ∗ AJCA + β₃ ∗ AJCA ∗ Matched Repatriation Taxes + ∑ β_i ∗ Controls_i + ε

(5)

where AJCA is an indicator variable equal to one in the year 2004 and 2005, and zero otherwise. We estimate this regression using the sample of loans issued between 2001 and 2008 in order to limit the scope of the test to the time period around the Act.16 Because AJCA might affect Repatriation Taxes of that year, we use Matched Repatriation Taxes, which is the nearest Repatriation Taxes in the non-AJCA period. We use a balanced sample for firms in this regression where we require at least one loan in the AJCA period and at least one loan in the non-AJCA period. A difference-in-differences framework allows us to capture the extent to which lenders change the calculation of interest spread in debt contracts for borrowers with high versus low potential earnings repatriation taxes from the non-AJCA to AJCA period. Because AJCA is a short window event (two years) and the period is known when announced, we would expect long maturity loans will benefit less while short-term loans will benefit more. To test this, we partition our sample by loan maturity and re-run the regressions on each subsample.

Table 8. Difference-in-differences design: the American Jobs Creation Act of 2004.

Dependent Variable=	SPREAD
	Full Sample		Short Maturity	Long Maturity
	(1)	(2)	(3)	(4)
Matched Repatriation Taxes	6.76	12.99 *	25.15 **	−1.39
	(1.45)	(1.92)	(2.56)	(−0.24)
AJCA	−28.08 ***	−22.40 ***	−21.22 *	−15.91 **
	(−6.06)	(−4.01)	(−1.92)	(−2.41)
Matched Repatriation Taxes * AJCA		−13.11 **	−34.55 ***	1.52
		(−1.97)	(−2.76)	(0.23)
FCOVENANT	4.15	4.05	5.69	1.41
	(1.31)	(1.28)	(1.22)	(0.41)
SIZE	−8.50 ***	−8.67 ***	−4.86	−10.17 ***
	(−2.97)	(−3.06)	(−1.10)	(−2.93)
LEVERAGE	30.59	31.52	44.26	25.20
	(1.32)	(1.36)	(1.35)	(1.03)
CURRENT	−3.46	−3.54	−3.17	−3.27
	(−1.17)	(−1.20)	(−0.70)	(−1.05)
MB	−12.35 ***	−12.35 ***	−17.16 ***	−6.42 *
	(−3.16)	(−3.14)	(−3.18)	(−1.66)
ROA	−183.70 **	−195.35 **	−248.82 **	−125.76
	(−1.98)	(−2.09)	(−2.17)	(−1.25)
ZSCORE	−8.71 **	−8.61 **	−6.02	−11.11 **
	(−2.31)	(−2.30)	(−1.07)	(−2.29)
TANGIBILITY	−15.14	−15.22	−10.75	−13.79
	(−1.03)	(−1.03)	(−0.50)	(−0.95)
SalesGrowth	1.73	1.59	4.69	8.74
	(0.14)	(0.13)	(0.21)	(0.61)
ForeignOp	5.43	5.81	−0.33	8.46 *
	(1.36)	(1.51)	(−0.06)	(1.83)
BT	0.57	9.26	−125.44 *	114.43
	(0.01)	(0.18)	(−1.81)	(1.64)
LOG(AMOUNT)	−15.24 ***	−15.12 ***	−13.44 ***	−12.34 ***
	(−5.14)	(−5.11)	(−2.81)	(−3.54)
LOG(MATURITY)	2.19	2.17	27.88 ***	−29.16
	(0.48)	(0.48)	(3.03)	(−0.75)
INS_IN	27.70 **	27.43 **	21.88	26.14 *
	(2.40)	(2.38)	(0.90)	(1.81)
REVOLVER	−7.75	−7.48	1.44	−25.32 ***
	(−1.30)	(−1.26)	(0.15)	(−3.36)
COLLATERAL	67.68 ***	67.64 ***	64.11 ***	66.53 ***
	(8.46)	(8.48)	(4.99)	(8.75)
PPINDEX	−23.13 ***	−23.12 ***	−12.26	−27.86 ***
	(−3.26)	(−3.25)	(−1.29)	(−3.60)
Loan Purpose FE	Included	Included	Included	Included
Constant	574.55 ***	570.33 ***	434.76 ***	658.33 ***
	(10.46)	(10.47)	(4.78)	(3.76)
Observations	1843	1843	805	1038
R-squared	0.522	0.524	0.500	0.624

Table 8 presents the results from the estimation of the following model around the American Jobs Creation Act (AJCA): Interest Spread = β₀ + β₁ ∗ Matched Repatriation Taxes + β₂ ∗ AJCA + β₃ ∗ AJCA ∗ Matched Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. All variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

presents the regression results of the difference-in-differences model. Table 8 provides evidence that in specifications with (Column 1) and without (Column 2) the interaction term, the coefficient of AJCA is significantly negative. More important to us, and consistent with our prediction, we find that the interaction term between Matched Repatriation Taxes and AJCA is negative and statistically significant. These results are consistent with banks imposing a smaller repatriation tax penalty during the AJCA period when the costing would be lower. In Columns 3 and 4 we partition the sample of borrowers by loan maturity. Consistent with our prediction that the benefits of AJCA should be largest for borrowers that have shorter loan maturities, the results in Columns 3 and 4 demonstrate that the interaction term between Matched Repatriation Taxes and AJCA is negative and statistically significant for short-term loans but positive and insignificant for long-term loans.

5.2. Differentiate from General Tax Avoidance

Hasan et al. (2014) shows that greater overall tax avoidance is associated with higher borrowing costs. U.S. MNCs keep large earnings (cash) positions overseas because of the ability to avoid paying tax until repatriation. However, the effect of earnings repatriation taxes can be different from overall tax avoidance for several reasons. First, the effect is unique to MNCs and, therefore, banks need to pay special attention to this factor when lending to multinational firms. Second, firms may have several strategies for avoiding tax and trade-off between different methods. A firm with high potential earnings repatriation taxes need not have high overall tax avoidance level. Third, as we discussed in the hypotheses development section, the level of foreign earnings held abroad may deliver unique information related to governance or information risk beyond a firm’s overall tax aggressiveness. Therefore, we attempt to differentiate our previously documented effect from overall firm-level tax avoidance as documented by Hasan et al. (2014).

First, we add two additional tax avoidance measures to our main regressions and report the results in Panel A of

Table 9. Potential earnings repatriation taxes incremental to general tax avoidance.

Panel A: Controlling for Tax Avoidance Measures from Prior Studies
		Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=		SPREAD	SPREAD	SPREAD	SPREAD
		(1)	(2)	(3)	(4)
Repatriation Taxes		6.08 ***	5.50 **	6.11 ***	5.72 **
		(3.06)	(2.25)	(2.80)	(2.05)
TA_CETR		−15.76 *		−12.55
		(−1.83)		(−1.21)
DTAX			24.09		23.57
			(1.04)		(0.92)
Control Vars		Included	Included	Included	Included
Loan Purpose FE		Included	Included	Included	Included
Year FE		Included	Included	Included	Included
Constant		533.01 ***	542.70 ***	563.32 ***	607.25 ***
		(17.39)	(15.19)	(11.25)	(16.95)
Observations		7193	4790	4414	2866
R-squared		0.596	0.587	0.609	0.606
Panel B: Partitioning on Tax Avoidance Measures From Prior Studies
	Full Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD	SPREAD	SPREAD
	BT < Median	BT > Median	TA_CETR < Median	TA_CETR > Median	DTAX < Median	DTAX > Median
	(1)	(2)	(3)	(4)	(5)	(6)
Repatriation Taxes	6.69 ***	1.76	8.23 **	4.45 *	7.37 *	4.79
	(2.68)	(0.53)	(2.39)	(1.85)	(1.89)	(1.59)
Control Vars	Included	Included	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included	Included	Included
Year FE	Included	Included	Included	Included	Included	Included
Constant	523.70 ***	560.92 ***	477.15 ***	590.96 ***	518.68 ***	553.73 ***
	(11.76)	(15.18)	(12.01)	(15.58)	(10.93)	(12.16)
Observations	3757	3728	3617	3576	2415	2375
R-squared	0.575	0.625	0.615	0.588	0.575	0.613

Table 9 presents the results from the estimation of the following model: Interest Spread = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. Panel A reports results with tax avoidance measures in the regressions. Regressions are partitioned on median of tax avoidance measures in Panel B. BT is book-tax difference for a firm in a certain year (Manzon & Plesko, 2002). TA_CETR is minus one times the cash effective tax rate of a firm in a certain year. DTAX the permanent book-tax difference for a firm in a certain year (Frank et al., 2009). All variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

. We already include BT, a commonly used overall tax avoidance measure, the book-tax difference for a borrower in a given year (Manzon & Plesko, 2002), in our main regressions. TA_CETR is minus one times the cash effective tax rate of a firm in a given year. DTAX is the permanent book-tax difference for a firm in a given year (Frank et al., 2009). The results demonstrate that our previous findings continue to hold, even after including these firm-level tax avoidance measures, suggesting that the effect of potential repatriation taxes is incremental to the overall tax avoidance effect.

Second, we partition our sample by the annual median of these overall tax avoidance measures and re-run the regressions within each subsample. We present the results in Panel B of Table 9 and provide evidence that the effect earnings repatriation taxes is primarily located in the low overall tax avoidance subsamples, indicating that banks care more about repatriation taxes when the level of overall firm-level tax avoidance is low. These results are consistent with our conjecture that the information in potential earnings repatriation taxes is incremental to the overall tax avoidance effect.

5.3. Potential Channels: Agency Risk and Information Risk

Nessa (2017) also shows that repatriation tax costs are associated with decreased dividend payments. As we argue above, however, potential earnings repatriation taxes may be indicative of additional agency risk as well as information risk beyond the direct cash effect. Because agency risk and information risk are priced factors in bank loan spreads, we would expect the positive relation between potential earnings repatriation taxes and borrowing cost to be exacerbated in poorly governed and opaque firms. If potential earnings repatriation taxes merely reflects deferred tax assets or liabilities and the effect of potential earnings repatriation taxes is just a shifting of current cash and future cash, then we would not observe a difference between these groups. Following prior literature, we use the variable Independent Directors, which is the percentage of outside directors on the board of directors in a firm, Institutional Investors, which is the proportion of shares held by institutional investors, and Analyst Coverage, which is the number of analysts following the firm, to capture corporate governance quality (Jensen, 1993; Hartzell & Starks, 2003; Yu, 2008; Coyne & Stice, 2018; Stice et al., 2022).17 We use the variable Forecast Frequency, which is management forecast frequency, Turnover, which is monthly volume divided by shares outstanding, and Discretionary Accruals, which is the residual from a modified Jones (1991) model, in order to capture the level of firm disclosure (Beyer et al., 2010; Chae, 2005; Hasan et al., 2014). We partition our sample based on these measures and re-run our main tests on each subsample.

We present the results in

Table 10. Investigating potential channels.

Panel A: Independent Directors
	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	%Independent Directors < Median	%Independent Directors > Median	%Independent Directors < Median	%Independent Directors > Median
	(1)	(2)	(3)	(4)
Repatriation Taxes	13.79 ***	−3.67	14.15 ***	−3.75
	(3.38)	(−1.11)	(3.04)	(−0.98)
Control Vars	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	474.77 ***	299.97 ***	524.76 ***	347.74 ***
	(8.61)	(5.83)	(7.51)	(5.78)
Observations	1838	1575	1211	1052
R-squared	0.623	0.663	0.643	0.665
Chi-Statistics	10.92		8.80
p-value	0.001		0.003
Panel B: Institutional Investors
	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	Institutional Investors < Median	Institutional Investors > Median	Institutional Investors < Median	Institutional Investors > Median
	(1)	(2)	(3)	(4)
Repatriation Taxes	9.89 ***	1.69	8.80 ***	2.70
	(3.20)	(0.64)	(2.65)	(0.82)
All Control Vars	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	536.12 ***	458.40 ***	529.01 ***	589.59 ***
	(14.45)	(8.73)	(9.44)	(7.56)
Observations	3744	3741	2271	2271
R-squared	0.586	0.627	0.591	0.648
Chi-Statistics	7.74		3.02
p-value	0.005		0.082
Panel C: Analyst Coverage
	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	Analyst Coverage < Median	Analyst Coverage > Median	Analyst Coverage < Median	Analyst Coverage > Median
	(1)	(2)	(3)	(4)
Repatriation Taxes	8.24 **	3.00	7.56 **	3.85
	(2.49)	(1.22)	(2.40)	(1.17)
All Control Vars	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	581.08 ***	476.44 ***	594.93 ***	501.01 ***
	(13.61)	(10.32)	(15.16)	(9.04)
Observations	3901	3584	2407	2135
R-squared	0.522	0.602	0.543	0.627
Chi-Statistics	3.03		1.20
p-value	0.081		0.272
Panel D: Forecast Frequency
	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	Forecast Frequency = 0	Forecast Frequency > 0	Forecast Frequency = 0	Forecast Frequency > 0
	(1)	(2)	(3)	(4)
Repatriation Taxes	11.13 ***	1.48	9.76 **	2.12
	(2.79)	(0.56)	(2.13)	(0.68)
All Control Vars	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	441.89 ***	461.57 ***	440.13 ***	485.12 ***
	(11.86)	(13.42)	(8.59)	(11.27)
Observations	2180	3987	1289	2654
R-squared	0.610	0.640	0.626	0.642
Chi-Statistics	8.12		3.70
p-value	0.004		0.054
Panel E: Share Turnover
	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	Turnover < Median	Turnover > Median	Turnover < Median	Turnover > Median
	(1)	(2)	(3)	(4)
Repatriation Taxes	12.26 ***	0.74	9.23 **	1.01
	(3.59)	(0.30)	(2.45)	(0.34)
All Control Vars	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	581.49 ***	524.80 ***	619.00 ***	517.97 ***
	(14.32)	(10.76)	(15.05)	(10.34)
Observations	3712	3680	2263	2228
R-squared	0.643	0.553	0.656	0.573
Chi-Statistics	15.27		5.61
p-value	0.000		0.018
Panel F: Discretionary Accruals
	Full Sample		Repatriation Taxes > 0 Sample
Dependent Variable=	SPREAD	SPREAD	SPREAD	SPREAD
	DA < Median	DA > Median	DA < Median	DA > Median
	(1)	(2)	(3)	(4)
Repatriation Taxes	4.29	6.03 **	1.77	7.87 **
	(1.60)	(1.99)	(0.53)	(2.29)
All Control Vars	Included	Included	Included	Included
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	554.39 ***	576.48 ***	584.12 ***	546.26 ***
	(14.41)	(12.31)	(10.91)	(8.20)
Observations	3740	3739	2269	2269
R-squared	0.613	0.591	0.624	0.609
Chi-Statistics	0.35		2.93
p-value	0.554		0.087

Table 10 presents the results from the estimation of the following model: Interest Spread = β₀ + β₁ ∗ Repatriation Taxes + ∑ β_i ∗ Controls_i + ε. Regressions are partitioned on median of partition variables. Independent Directors is the percentage of outside directors on the board of directors in a firm. Institutional Investors is the proportion of shares held by institutional investors. Analyst Coverage is the number of analysts following the firm. Forecast Frequency is the management forecast issue frequency. Turnover is the monthly volume divided by shares outstanding. Discretionary Accruals is the residual from a modified Jones model. All other variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

. Across all specifications, we find that the coefficients on Repatriation Taxes remain positive and significant for poorly governed and opaque firms, while the coefficients on Repatriation Taxes are insignificant when firm disclose is high or the firm is well governed. The difference between groups is statistically significant in most cases. These results confirm our arguments that agency risk and information risk are two plausible channels through which bank loan spread is affected beyond the direct cash flow effect.

5.4. Additional Robustness Tests

We also conduct several additional tests to confirm that our results are robust to different specifications. First, while our main Repatriation Taxes proxy is measured annually, the relationship holds for other measures of international tax planning. Following Edwards et al. (2016), we use a three-year cumulative cost of repatriation called Repatriation Taxes_3 Year to measure the difference in U.S. taxes owed on foreign earnings and foreign taxes paid summed over the prior three years (assuming a 35 percent US tax rate). The variable is calculated as the sum of pre-tax foreign income multiplied by 35 percent minus the sum of foreign income taxes paid, scaled by total assets and multiplied by 100 to be presented as a percentage. We also follow Collins et al. (1998) and Mills and Newberry (2004) to calculate the statutory tax rate difference between US and other countries to proxy the incentive of income shifting, one activity of international tax planning. We use this new measure as our independent variable of interest and re-run our main test. The results are reported in

Table 11. Alternative measure of potential earnings repatriation taxes: three-year measure and foreign tax rate.

Dependent Variable=	Full Sample	Repatriation Taxes > 0 Sample	Full Sample	Repatriation Taxes > 0 Sample
	SPREAD	SPREAD	SPREAD	SPREAD
	(1)	(2)	(1)	(2)
Repatriation Taxes_3 Year	1.73 **	1.91 **
	(1.97)	(1.96)
FTR			39.49 ***	53.58 ***
			(3.13)	(3.25)
FCOVENANT	4.78 ***	4.25 **	3.60 **	3.67 **
	(3.11)	(2.13)	(2.48)	(2.03)
SIZE	−14.41 ***	−14.73 ***	−16.36 ***	−15.07 ***
	(−8.06)	(−6.54)	(−10.36)	(−7.82)
LEVERAGE	43.42 ***	48.69 ***	42.63 ***	47.86 ***
	(3.98)	(3.27)	(3.91)	(3.83)
CURRENT	−3.77 **	−5.99 ***	−4.33 ***	−4.73 **
	(−2.20)	(−3.02)	(−2.91)	(−2.54)
MB	−8.67 ***	−9.10 ***	−7.31 ***	−8.99 ***
	(−4.88)	(−3.99)	(−4.42)	(−4.53)
ROA	−77.51 *	−38.87	−80.01 **	−42.21
	(−1.78)	(−0.66)	(−2.24)	(−0.95)
ZSCORE	−10.23 ***	−10.29 ***	−11.61 ***	−10.03 ***
	(−5.26)	(−4.21)	(−6.63)	(−4.86)
TANGIBILITY	−18.46 **	−13.04	−25.26 ***	−20.10 **
	(−2.29)	(−1.23)	(−3.11)	(−2.24)
SalesGrowth	5.12	7.14	8.62 *	5.35
	(0.88)	(0.85)	(1.79)	(0.83)
ForeignOp	3.26	0.12 *	2.84 *	2.78
	(1.62)	(1.68)	(1.67)	(1.38)
BT	−11.68	−40.11	−4.28	−42.40
	(−0.33)	(−0.88)	(−0.15)	(−1.18)
LOG(AMOUNT)	−9.60 ***	−9.61 ***	−9.28 ***	−10.19 ***
	(−5.66)	(−4.46)	(−6.12)	(−5.21)
LOG(MATURITY)	−0.14	−1.27	−2.70	−0.69
	(−0.06)	(−0.43)	(−1.24)	(−0.23)
INS_IN	59.67 ***	70.05 ***	62.56 ***	67.32 ***
	(9.10)	(8.80)	(10.57)	(9.54)
REVOLVER	−10.87 ***	−8.50 **	−12.07 ***	−11.50 ***
	(−3.90)	(−2.46)	(−4.86)	(−3.63)
COLLATERAL	62.87 ***	61.22 ***	59.18 ***	53.62 ***
	(16.64)	(13.63)	(17.46)	(13.43)
PPINDEX	−25.36 ***	−23.04 ***	−26.20 ***	−23.72 ***
	(−7.77)	(−5.77)	(−8.55)	(−6.37)
Loan Purpose FE	Included	Included	Included	Included
Year FE	Included	Included	Included	Included
Constant	460.81 ***	523.94 ***	544.02 ***	525.25 ***
	(14.93)	(12.31)	(17.04)	(13.37)
Observations	6112	3563	7485	4551
R-squared	0.619	0.627	0.595	0.608

Table 11 presents the results from the estimation of the following model: Interest Spread = β₀ + β₁ ∗ Repatriation Taxes_3 Year/FTR + ∑ β_i ∗ Controls_i + ε. Repatriation Taxes_3 Year is three-year cumulative measure of the difference in U.S. taxes owed on foreign earnings and foreign taxes paid summed over the prior three years (assuming a 35 percent U.S. tax rate). The variable is calculated as the sum of pre-tax foreign income multiplied by 35 percent minus the sum of foreign income taxes paid, scaled by total assets and multiplied by 100 to be presented as a percentage. FTR is the statutory tax rate difference between U.S. and other countries (Collins et al., 1998). We replace FTR with 0 if FTR is less than 0. All other variables are defined in the Appendix A. All continuous variables are winsorized at 1% and 99% level. Regressions include loan purpose and year fixed effects and standard errors are heteroskedasticity robust and clustered at firm level. t-statistics are reported in parentheses. ***, **, * denote significance at the 1%, 5%, and 10% levels, respectively.

. The coefficients on Repatriation Taxes_3 Year and FTR are positive and statistically significant for both the full sample and subsample.

Second, our results are also robust to using different measures of firm performance and financial constraint, specifically return on equity (ROE) and Altman’s Z-score. Additionally, in our cross-sectional tests we partition our sample by terciles instead of at the median and find similar results. Third, prior literature has shown that diversified firms hold less cash because of their greater ability to generate cash internally. Therefore, we expect that the effect of Repatriation Taxes on debt will be smaller for diversified firms because diversified firms will be less likely to use foreign earnings (cash). Even though the firms in our sample are all multinationals, there is still significant variation in the degree of diversification across firms. We partition the sample by firms’ number of segments and re-estimate Equation 1 in each subsample. The results (untabulated) show that the coefficient on international tax planning is positive and significant when the number of segments is smaller than the annual median and insignificant when firms are more diversified.

Fourth, we also partition the sample by firms’ cash holdings and re-estimate Equation (1) in each subsample. The intuition is that when cash holdings are low, firms are more likely to need and, therefore, repatriate cash/earnings from abroad. The results (untabulated) show that the coefficient on Repatriation Taxes is positive and significant when the cash holdings are lower than the annual median and insignificant when the cash holdings are higher.

Fifth, our inferences are unchanged after controlling for geographic concentration, investment opportunities, and liquidity needs (Stice et al., 2017; Campbell et al., 2023). We use the HHI index of sales (relying on Compustat geographic segment data) to measure geographic concentration, weighted foreign industry growth as a proxy of foreign investment opportunities following Huang (2015), and cash scaled by total assets as a proxy for liquidity needs. Also, our results are not affected when we include deferred tax assets and deferred tax liabilities. Our results are also not sensitive to adding other control variables like R&D intensity, which is defined as the R&D expenses over total assets, deferred taxes scaled by total assets, or using the number of geographic segments as a proxy for foreign operations.

Finally, we perform our analyses at the facility-level, following prior literature, but our results are also robust to performing the analyses (estimations of all equations) at the package-level (untabulated).

6. Conclusions

In this study, we examine the effect of potential earnings repatriation taxes on bank loan contracting. We predict that loans issued to borrowers with relatively higher potential earnings repatriation taxes are associated with higher loan spreads. Moreover, we predict that the interest spread effect will vary with the likelihood that a borrower will need to access foreign earnings held abroad. We find empirical results that support these predictions. Specifically, we find that a one standard deviation change in potential earnings repatriation taxes leads to a change in interest spread of approximately 3.70 basis points, which represents a 2.6% increase in borrowing costs. Given the large size of the syndicated loan market, the economic significance is substantial. We also find that the effect of potential earnings repatriation taxes is more pronounced for firms with recent poor accounting performance (profitability) and for firms that are more financially constrained. Our results are robust to a difference-in-differences research design using The American Jobs Creation Act of 2004 as an exogenous shock.

We also predict and provide evidence that higher potential earnings repatriation taxes are associated with an increase in the likelihood that lenders require collateral and an increase in the number of financial covenants included in loan contracts, providing evidence that potential earnings repatriation taxes affect both the price and non-price contract terms of debt contracts. Our last prediction investigates whether borrowers are more likely to include a foreign bank in the loan syndicate when the level of international tax planning is high. We provide evidence that this is the case, which is consistent with borrowers negotiating for the inclusion of a foreign bank among the lenders in order to use foreign profits trapped abroad to make interest payments on the loan.

Overall, our empirical results contribute to our understanding of the interaction between the U.S. tax system and the cost of debt for U.S. multinational firms. We provide evidence that U.S. MNCs that engage in international tax planning activities that lead to higher potential earnings repatriation taxes are subject to additional costs that have not previously been documented. Given the increasing amount of cash that U.S. MNCs keep abroad and the increasing size of the private lending market, our results will be of interest to capital market participants and regulators. These findings will be of interest to regulators and market participants as the public debate in the U.S. about the future of corporate taxation continues after the 2024 presidential election. We acknowledge that our tests are necessarily U.S.-specific—we believe that investigating this issue in a broader international setting is a promising avenue for future research.