The impact of the bankruptcy abuse prevention and consumer protection act on debt and equity holders

(1)

1

MASTER THESIS

MSc FINANCE: QUANTITATIVE FINANCE

The impact of the bankruptcy abuse prevention and consumer

protection act on debt and equity holders

By: Brijan Baboeram

(2)

1

Statement of Originality

This document is written by Student [Brijan Baboeram] who declares to take full responsibility for the contents of this

document.

I declare that the text and the work presented in this document are original and that no sources other than those

mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the

(3)

2

The impact of the bankruptcy abuse prevention and consumer

protection act on debt and equity holders

Brijan Baboeram

ABSTRACT

I examine the capital market consequences of the Bankruptcy Abuse Prevention and Consumer Protection Act of 2005 (hereafter: BAPCPA). Using a sample of 11 U.S. companies, I find that CDS spreads and betas increase after the announcement of the BAPCPA, which suggests a higher probability of a firm default and more risk for equity holders. I find that the debt capacity increase for cash-constrained firms. Using a triple-differences analysis, I find that the CDS spreads increase for distressed constrained firms. Last, I find that firms tend to hold more cash and increase capital expenditures after the BAPCPA Constrained firms also hold more cash but do not increase capital expenditures.

Introduction

The BAPCPA includes provisions that are meant to provide adequate protection to investors. One of these provisions is an amendment to the Automatic Stay injunction. After a debtor files for bankruptcy, the Automatic Stay protects the debtor’s assets against actions of the creditor. To avoid debtors abusing the protective measures provided by the automatic stay, United States Congress added some exceptions to the injunction. The BAPCPA of 2005 added the exception that the Automatic Stay could expire under certain conditions (11 USC 110). Another significant inclusion in the BAPCPA, is that firms

(4)

3 that file for chapter 11 bankruptcy must reorganize much quicker or risk losing control of the reorganization. The BAPCPA also makes it easier for a creditor to force a liquidation or remove the debtor in possession in chapter 11. (Benderson & Whitaker, 2006). The amendment to the Automatic Stay and the other inclusions could increase the bargaining power of creditors in-court. A reason to believe this is the case, is the substantial increase in bankruptcy filings after the law was adopted. In October 2005, over 600,000 bankruptcy cases were filed. This is a substantial increase compared to the bankruptcy filings in 2004, which were roughly 130,000 (Benderson & Whitaker, 2006). The act also caused a shift in the supply of ‘hedging’. For example, it structurally changed the supply of financial derivatives (Giambona and Wang, 2014). The act was implemented to provide guidelines in consumer bankruptcy cases. However, some provisions in the BAPCPA also apply to business bankruptcy cases (Benderson & Whitaker, 2006). For example, the time firms were given to reorganize in a chapter 11 bankruptcy was limited after the BAPCPA. But more importantly, the act increased the security of creditors in bankruptcy court. This resulted in a shift of bargaining power from the debtor to the creditor in bankruptcy court. Moreover, debtors are more strictly monitored by creditors in the bankruptcy process. Also, it is prohibited for debtors to pay out huge sums of cash to key employees in the bankruptcy process. Therefore, I want to examine the effect of the BAPCPA on both debt holders as well as equity holders.

To examine the consequences for debt holders, I examine the effect of the BAPCPA on the Credit Default Swap spreads of U.S companies. I expect that the CDS spreads will increase because the probability that firms file for chapter 7 bankruptcy

(5)

4 increases after the BAPCPA. I find that CDS spreads increase for all firms. The implications of this finding are that firms are more likely to formally default and that the price of bargaining power has increased for creditors.

Next, I investigate the implications for the equity holders by examining the effect of the BAPCPA on the beta of U.S companies. I expect that the beta of the companies will increase after the implementation of the BAPCPA. The first reason is that BAPCPA increases the risk of holding equity. The second reason is because of the new amendments place limitations on executive compensation if a firm has filed for bankruptcy. This is because debtors often respond to the expected loss of key executives by offering them stock options guaranties, salary, and pension. After the BAPCPA, this is only allowed under some circumstances and by meeting certain criteria (Benderson & Whitaker, 2006). This added provision might lead to less risk shifting and less asset substitution, hence a lower value for equity holders relative to debt holders (Modigliani and Miller, 1958). This is another limitation, that gives me reason to believe that the risk for shareholders will increase and lead to a higher beta.

Giambona and Wang (2014) examine the effect of the BAPCPA on the derivatives market. Using a sample of airlines, they find that distressed airlines hedge more after the BAPCPA. These findings suggest that distressed airlines engage more in hedging activity if there is limited counterparty risk. The BAPCPA increases the transparency in a bankruptcy process such that parties are more likely to engage in transactions. My study is closely related to that of Giambona and Wang (2014). A key difference between their study and mine is that I examine the effect of the BAPCPA on debt and equity holders.

(6)

5 Another difference between their study and mine is that they only take into consideration the change to the Automatic Stay but the BAPCPA includes several other important changes.

The remainder of my paper is organized as follows. In Section 1, I discuss the relevant literature and develop my propositions. In section 2, I will elaborate on the methodology and results. In section 3 of this paper, I will discuss the results and give concluding remarks. Section 4 discusses further research.

1. Background information

In the early 1990s, many practitioners in the bankruptcy community believed that the current Bankruptcy Code worked well but had become too complex. For example, each special interest had its own exception or carve-out under title 11. In response, Congress passed the Bankruptcy Reform Act of 1994. This act was able to solve most of the issues experienced by practitioners. (Benderson & Whitaker, 2006) One area, however, that was extremely controversial from the beginning and created a long path for legislative change were concerned consumer of individual bankruptcy. More specific, the bankruptcy worried that debtors were abusing the bankruptcy system by discharging debts which debtors could afford. (Benderson, 2006). For that issue, the BAPCPA was drafted. The president signed the BAPCPA of 2005 on April 20 of that year.

The amendments that BAPCPA made to the Bankruptcy Code were directed to consumer rather than business bankruptcies (Benderson & Whitaker, 2006). But a

(7)

6 substantial portion of the 500-page bill affects businesses. For example, BAPCPA forces business debtors that filed for chapter 11 bankruptcy to reorganize faster or risk losing control over the proceedings. Moreover, the new amendments place limitations on pay-outs to key employees. As mentioned before, BAPCPA makes it easier for creditors to force a liquidation or remove the debtor in possession in chapter 11 (Benderson & Whitaker, 2006).

2. Prior studies

In a related study conducted by Campello et al. (2011), the authors examine the effect of a change in tax code, called TD9599, on CDS spreads. Before 2012, the Internal Revenue Service levied huge taxes on loans renegotiated out-of-court. The reform in the tax code implicitly lowered the taxes on these out-of-court loans. By reducing the taxes owed out of court while leaving in-court costs the same, the creditors could experience a shift in incentives Campello et al. (2011). To measure the change in bankruptcy likelihood, the authors look at the market price reaction of an instrument directly tied to corporate default: Credit Default Swap.

This study differs from theirs for the following reasons. Firstly, the BAPCPA makes creditors better off in-court while the change in tax code makes creditors better off out-of-court while keeping in-court costs the same. Put differently, this study examines a shock in the opposite direction. Secondly, the BAPCPA also changes the position of the debtors in-court and therefore I look at potential changes to equity holders.

(8)

7 They find that firms hold more cash after there a CDS starts trading on their debt. To verify their hypotheses, they construct the following empirical framework. In their model, an entrepreneur has a certain amount of debt on which she must pay a fixed amount on a predetermined date. If the cash flow on that day is insufficient, she has the option to renegotiate with her creditors. To avoid these renegotiations and pay the creditors, the entrepreneur could also hoard cash. If there is also a CDS trading on the company debt, the creditors’ bargaining power increases. What’s more, the CDS also forces the creditor to pay out more cash to debt holders. Therefore, an entrepreneur is even less incentivized to renegotiate with debt holders and will hoard more cash to avoid missing a payment. The price of this increased bargaining power is the price paid for the CDS.

This study builds upon theirs by viewing the CDS also a way of the creditor to increase their bargaining power. The spread is then the price paid by the creditor to increase his bargaining power.

In a paper by Asquith et al. (1994), the authors examine ways in which firms try to avoid going to bankruptcy court through public and private debt restructurings. They find that a firm’s debt structure affects the way financially distressed firms restructure. Their empirical framework is constructed as follows. They look at U.S firms that issued junk bonds during the 1980s and subsequently became financially distressed. In their study a firm is considered financially distressed if it fails to meet its coverage ratio. They authors argue that there are two reasons why a bank would not forgive principal. First, if a bank is highly protected in bankruptcy court, there is no incentive to forgive principal. Second,

(9)

8 if the bank is not well protected in bankruptcy court, but still has seniority to other creditors, it would be willing to ‘’loosen’’ financial constraints and increase the debt capacity.

My study builds upon their study by taking into consideration the fact that the BAPCPA increases the protection of creditors in bankruptcy court. Hence, creditors could be more forgiving and ‘’loosen’’ financial constraints.

3. Hypotheses

I examine the effect of BAPCPA on CDS spreads. The payoff from this instrument depends on the creditworthiness of the company on which the contract is written. There are two sides to the contract: the buyer and seller of protection. There is a payoff from the seller of protection to the buyer of protection if the specified entity defaults on its obligations (Hull, 2010). The spread is the difference between the rate on a government issued bond and the rate bond issued by the company expressed in basis-points (bp). The spread is determined by multiple factors. One important factor is the probability of default, which is positively correlated with the spread. If a company is more likely to default, then the cost to insure against this event increases. (Hull, 2010) A second factor is the creditworthiness of the company, which is negatively correlated with the spread. It

(10)

9 is costly to insure against a company with a high credit rating. Lastly, the supply and demand of the derivative is another important factor for determining the spread. (Veronesi, 2013). If there is a CDS trading on the company debt, the creditors’ bargaining power increases. The reason for this increase in bargaining power is that creditors become ‘’tougher’’. For the same reason, a CDS also forces the creditor to pay out more cash to debt holders. (Subrahmanyam et al, 2015). Therefore, an entrepreneur is even less incentivized to renegotiate with debt holders and will hoard more cash to avoid missing a payment. The price of this increased bargaining power is the price paid for the CDS. Because BAPCPA makes it easier for creditors to force a firm to file for chapter 7 bankruptcy, I expect CDS spreads to increase. The underlying reason is that if the probability of a formal default increases the issuers of the CDSs must charge a higher spread.

Hypothesis 1: CDS spreads increases after the BAPCPA.

I expect that the effect of BAPCPA on CDS spreads is greater for distressed firms. The reason is that because distressed firms are more likely to enter renegotiations with creditors than non-distressed firms. Distressed firms are more likely to file for bankruptcy. In response, issuers of CDS must charge higher spreads.

Hypothesis 2: increase in CDS spreads will be larger for distressed firms

In addition to examining the effect of BAPCPA on CDS spreads, I examine its effect on debt capacity. Debt capacity is the amount of debt a company can take on before

(11)

10 becoming financially distressed (Modigliani & Miller, 1985). I expect firms’ debt capacity to increase because creditors are better protected in bankruptcy court. This added security could mean that they will demand lower rates of return or give higher loans.

Hypothesis 3: the debt capacity will increase after BAPCPA

My fourth prediction is that the firm’s beta will increase after the BAPCPA. This prediction is based on the Modigliani Miller theory. According to this theory, the risk for equity holders becomes larger with the amount of debt the firm takes on. This is because equity holders are residual claimants. Because I predict that a firm’s debt capacity will increase after the BAPCPA, its capital structure will have more debt to equity. This increase in debt could lead to riskier equity.

Hypothesis 4: firms’ equity beta will increase after BAPCPA. I expect this increase to be larger for distressed firms. The reason is that holding equity in a distresses firm is riskier than holding equity in a non-distressed firm.

I expect firms to hold more cash after the BAPCPA. The reason for this is that creditors can easily force a firm to file for chapter 7 bankruptcy. To prevent this scenario, equity holders will hold more cash.

Hypothesis 5: cash ratio will increase after the BAPCPA

As hypothesised above, I expect firms’ debt capacity to increase and firms’ will hold more cash to prevent missing a payment to debt holders. But with firms with sufficient cash

(12)

11 can also use this increase in debt capacity to increase capital expenditures. (Almeida et. al., 2011). That is, firms that are financially healthy will increase capital expenditures.

Hypothesis 6: the capital expenditures of firms will increase after the BAPCPA.

4. Research design

4.1 Method

4.1.1 Test of hypothesis 1

To test hypothesis 1, I estimate the following model:

𝐶𝐷𝑆𝑆𝑃𝑅𝐸𝐴𝐷𝑆 = 𝛽₀+ 𝛽₁𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽₂𝑃𝑅𝑂𝐹𝐼𝑇𝐴𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽₃𝑇𝐴𝑁𝐺𝐼𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽₄𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽₅𝐿𝑂𝐺𝐴𝑆𝑆𝐸𝑇𝑆 + 𝜀 (1)

Where: CDSSPREADS = the spread between the firm’s borrowing rate and the U.S treasury 5-year bond; POSTBAPCPA = a dummy variable that is equal to 1 after April 13th of 2005, a week before the implementation of the act; PROFITABILITY = a firm’s

EBIT divided by total assets; TANGIBILITY = the amount of property plant and equipment divided by total assets; LEVERAGE = the sum of long-term debt plus short-term debt plus preferred stock divided by this number plus market capitalization. POSTBAPCPA is the variable of interest. The coefficient on this variable is expected to

(13)

12 be positive as it measures the effect of the announcement of the BAPCPA on the level of CDSSPREADS. BAPCPA is equal to 1 a week before the implementation because after a revised bill has passed both houses, it is almost certain that it will be signed by the president 7 days later. PROFTIABILITY is a control variable. I include this variable because it is aligned with the study of Giambona and Wang (2014). These authors examine the same shock. TANGIBILITY is a control variable that is also used by Giambona and Wang (2014). The last control variable used in this equation is LEVERAGE. All these control variables are recommended to use in risk management studies (Giambona and wang, 2014). I predict that CDSSPREADS will increase after BAPCPA because firms can be more easily forced to file for chapter 7 bankruptcy.

To test hypothesis 2, I estimate the following model: 𝐶𝐷𝑆𝑆𝑃𝑅𝐸𝐴𝐷𝑆 = 𝛽₁𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽₂𝐷𝐼𝑆𝑇𝑅𝐸𝑆𝑆𝐸𝐷 + 𝛽₃𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 𝑥 𝐷𝐼𝑆𝑇𝑅𝐸𝑆𝑆𝐸𝐷 + 𝛽₄𝑇𝐴𝑁𝐺𝐼𝐵𝐼𝐿𝐼𝑇𝑌 +

𝛽₅𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽₆𝐿𝑂𝐺𝐴𝑆𝑆𝐸𝑇𝑆 + 𝜀 (2)

Where: CDSSPREADS = Where: CDSSPREADS = the spread between the firm’s borrowing rate and the U.S treasury 5-year bond; POSTBAPCPA = a dummy variable that is equal to 1 after April 13th_{of 2005, a week before the implementation of the act;}

DISTRESSED = a dummy variable that is equal to 1 if a firm has an Altman Z-score below 1.81. This cut-off is aligned with the study by Campello et al. (2011);

(14)

13 distressed after April 13th of 2005; LOGASSETS is the log of the amount of firm’s assets

measured on an annual basis. Notice that PROFITABILITY is dropped because this variable is directly used to estimate the Altman Z-score (Giambona and wang, 2014).

To test hypothesis 3, I estimate the following model: 𝐷𝐸𝐵𝑇𝑇𝑂𝐴𝑆𝑆𝐸𝑇𝑆 = 𝛽1𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽2𝐶𝑂𝑁𝑆𝑇𝑅𝐴𝐼𝑁𝐸𝐷 + 𝛽3𝐶𝑂𝑁𝑆𝑇𝑅𝐴𝐼𝑁𝐸𝐷 𝑥 𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽4𝐿𝑂𝐺𝐴𝑆𝑆𝐸𝑇𝑆 + 𝛽5𝑇𝐴𝑁𝐺𝐼𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽6𝑃𝑅𝑂𝐹𝐼𝑇𝐴𝐵𝐼𝐿𝐼𝑇𝑌 + 𝜀 (3)

Where: DEBTTTOASSETS = a firm’s long-term debt plus short-term debt over total assets; CONSTRAINED = a dummy variable that is equal to 1 if a firm cannot pay any dividends. This is aligned with the study by Fazzari et al. (1988). CONSTRAINED x POSTBAPCPA = a dummy variable that is equal to 1 if a firm is constrained after April 13th_{, 2005. I predict that DEBTTOASSETS will increase after the BAPCPA}

because debt holders will increase a firm’s debt capacity. Therefore, a firm’s debt to assets ratio will increase. According to Rampini & Viswanathan (2010) more constrained firms engage in less risk management and may exhaust their debt capacity. Therefore, constrained firms are likely to exploit this possibility to increase their debt capacity than non-constrained firms.

(15)

14 𝐵𝐸𝑇𝐴 = 𝛽₀+ 𝛽₁𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽₂𝑃𝑅𝑂𝐹𝐼𝑇𝐴𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽₃𝑇𝐴𝑁𝐺𝐼𝐵𝐼𝐿𝐼𝑇𝑌 +

𝛽₄𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽₅𝐿𝑂𝐺𝐴𝑆𝑆𝐸𝑇𝑆 + 𝜀 (4)

Where: BETA = a firm’s stock covariance with the S&P500 index divided by the variance of the S&P500 on an annual basis. I predict that the firm’s beta will increase because holding equity will become riskier after the BAPCPA.

To test hypothesis 5, I estimate the following model:

𝐶𝐴𝑆𝐻𝑇𝑂𝐴𝑆𝑆𝐸𝑇𝑆 = 𝛽₁𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽₂𝑁𝑊𝐶 + 𝛽₃𝐶𝐴𝑃𝐼𝑇𝐴𝐿𝐸𝑋𝑃𝐸𝑁𝐷𝐼𝑇𝑈𝑅𝐸𝑆 + 𝛽4𝑇𝐴𝑁𝐺𝐼𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽5𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽6𝐷𝐼𝑉𝐼𝐷𝐸𝑁𝐷 + 𝜀 (5)

Where: CASHTOASSETS = the amount of cash a firm has on its balance sheet in a certain year divided by total assets; NWC is the net working capital divided by total sales. CAPITALEXPENDITURES are the amount of capital expenditures. DIVIDEND is a dummy variable that is equal to 1 if a firm does not pay-out any dividends. These control variables are aligned with the study conducted by Bates et al. (2009) on cash holdings. I predict that cash holdings will increase because the consequences of missing a debt payment are more severe after the BAPCPA. I predict that constrained firms will hold more cash after the BAPCPA.

(16)

15 To test hypothesis 6, I estimate the following model:

𝐶𝐴𝑃𝐼𝑇𝐴𝐿𝐸𝑋𝑃𝐸𝑁𝐷𝐼𝑇𝑈𝑅𝐸𝑆 = 𝛽₁𝑃𝑂𝑆𝑇𝐵𝐴𝑃𝐶𝑃𝐴 + 𝛽₂𝑃𝑅𝑂𝐹𝐼𝑇𝐴𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽₃𝑇𝐴𝑁𝐺𝐼𝐵𝐼𝐿𝐼𝑇𝑌 + 𝛽₄𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸 + 𝛽₅𝐿𝑂𝐺𝐴𝑆𝑆𝐸𝑇𝑆 + 𝛽₆𝐿𝑂𝐺𝐶𝐴𝑆𝐻 + 𝜀 (6)

I predict that capital expenditures will increase because firms have a greater debt capacity after BAPCPA. I predict that capital expenditures will decrease for constrained firms.

4.2 Sample selection choices

To conduct my analysis, I collect data from COMPUSTAT capital IQ, CRSP and Campello et al. CDS spreads database. Although the CDS spreads database cover UK US and Japanese firms, I focus on a sample of US firms that have active CDS trading on their debt in the period 2003 to 2006 and a GVKEY code. I do so because the BAPCPA only applies to US based firms and I can find the matching COMPUSTAT fundamentals. After collecting the requisite data, I drop UK and Japanese firms. By applying these filters, I lose 95% of my observations in the database, leaving a sample of roughly 5,200 observations. Table 1 presents statistics for my sample. These results suggest that my sample consists of relatively large firms measured by assets. I also find that the CDS spread standard deviation is very large with the smallest observation being 1 basis point up to 2,330 basis points. Lastly, I find that 20% of the firms in my sample are in distress. After I apply the filters, still have a large enough sample to get statistical significant results.

(17)

16 Table 1. Summary Statistics

N Mean Std.dev Min Max

CDS 5,189 80.67 203.5 1 2,330 Log Assets 5,197 9.031 0.548 7.572 9.808 Distress 5,197 0.189 0.392 0 1 Tangibility 5,197 0.405 0.221 0.058 0.794 Z-score 4,761 2.237 0.668 1.191 4.045 Constrained 5,197 0.019 0.018 0 0.071 5.0 Results

5.1 correlations and VIF-scores

Table 1. Correlation matrix Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (1) CDS 1.000 (2) Capital Expenditures 0.014 1.000 (3) Distress -0.399 -0.171 1.000 (4) log Assets 0.012 0.198 -0.363 1.000 (5) Debt-to-Assets -0.032 0.525 -0.097 -0.311 1.000 (6) Leverage 0.566 0.454 -0.617 0.057 0.366 1.000 (7) Tangibility 0.014 0.432 0.040 -0.605 0.390 0.085 1.000 (8) Profitability -0.532 -0.520 0.481 -0.089 -0.023 -0.794 -0.334 1.000 (9) Beta 0.438 0.361 -0.076 -0.172 -0.064 0.381 0.587 -0.732 1.000 (10) Constrained -0.204 -0.696 0.083 0.086 -0.388 -0.436 -0.429 0.523 -0.511 1.000 (11) Cash 0.264 -0.286 -0.222 0.706 -0.640 0.090 -0.685 -0.140 -0.027 0.349 1.000 (12) BAPCPA 0.101 -0.003 0.252 -0.125 0.197 -0.019 0.139 0.080 0.253 0.084 -0.031 1.000

I derive from the table above that certain variables have a strong correlation. To find multicollinearity in my models, I compute the Variance Inflation Factor for correlations higher than 0.5 or lower than -0.5. I find 15 strong relations between variables.

(18)

17

Table 2. Variance Inflation Factor scores

Variable VIF 1/VIF

Tangibility 5.18 0.193 Cash 4.92 0.203 Debt-to-assets 3.32 0.300 Beta 2.90 0.344 Leverage 2.86 0.350 Constrained 2.35 0.425 Capital Expenditures 2.32 0.431 Mean VIF 3.41

(19)

18 5.2 Regression results

Table 3 presents the results of estimating model (1) and model (2).

Table 3. OLS-regressions

(1) (2) (3)

Dependent variable CDSSPREADS CDSSPREADS CDSSPREADS

POSTBAPCPA 62.47*** 39.16*** 9.525*** (3.790) (2.841) (1.503) PROFITABILITY -2,024*** (93.04) TANGIBILITY -303.4*** -139.3*** -43.42*** (11.14) (6.460) (2.307) LEVERAGE 172.4*** 449.1*** (16.94) (17.83) LOG ASSETS -75.64*** -9.543*** 3.434*** (3.625) (0.649) (0.138) DISTRESSED 69.56*** 126.5*** (5.884) (2.581) DISTRESSED x POSTBAPCPA 104.1*** -95.68*** (13.52) (2.957) CONSTRAINED 194.7*** (13.38) CONSTRAINED x POSTBAPCPA 497.1*** (31.70) POST x CONSTRAINED x DISTRESSED 296.9*** (24.69) CONSTANT 931.4*** (43.63) Observations 5,189 5,189 5,189 R-squared 0.463 0.556 0.639

Robust standard errors in parentheses. *** p0.01, ** p0.05, * p0.1

Column 1 presents the regression of BAPCPA on CDSSPREADS. This is the baseline regression. Column 2 presents the results of distressed firms after the BAPCPA on CDSSPREADS. Column 3 presents the results of distressed and constrained firms after the BAPCPA on CDDSPREADS. In each model, I find a positive and significant coefficient on POSTBABCPA. The magnitude of the coefficient in column 1 suggests that, on average, after the BAPCPA, the spread on the CDSs increased by 62.47 basis

(20)

19 points. This finding is consistent with hypothesis 1. The sign is positive, and this is aligned with my hypothesis 1.

In column 2, I find a positive and significant coefficient on DISTRESSED x POSTBAPCPA. The magnitude of this coefficient suggests that, on average, after the BAPCPA distressed firms experience an increase in CDS spreads by an additional 104.1 basis points. The magnitude of the BAPCPA drops to 39.16 but it still significant and positive. This finding is consistent with hypothesis 2.

In column 3 I find a positive and significant coefficient on POST x

CONSTRAINED x DISTRESSED. The magnitude of this coefficient suggests that, on average after the BAPCPA the spreads on CDSs increase by 269.9 basis points. The coefficient indicates that CDS spreads for firms increase the most in total because these firms have the highest probability of going bankrupt.

(21)

20

Table 4. OLS-regression

(1)

Dependent variable DEBT TO ASSETS

POSTBAPCPA 0.019*** (0.003) CONSTRAINED 0.054*** (0.002) CONSTRAINED x POSTBAPCPA 0.014*** (0.001) LOG ASSETS 0.0141*** (0.000) TANGIBILITY 0.251*** (0.0054) PROFITABILITY 0.473*** (0.015) Observations 5,197 R-squared 0.914

Robust standard errors in parentheses *** p0.01, ** p0.05, * p0.1

I find a positive and significant coefficient on POSTBAPCPA. The magnitude of the coefficient suggests that, on average, after the BAPCPA the debt to assets of firms increase by 0.019. This finding in consistent with hypothesis 3. The positive sign indicates that firms indeed increase their debt to asset ratio after the BAPCPA. Moreover, I find a positive and significant coefficient for CONSTRAINED x

POSTBAPCPA indicating that constrained firms have a higher their debt to asset ratios than unconstrained firms after BAPCPA. The magnitude of this coefficient is 0.014.

(22)

21 Table 5 presents the results of estimating model 4.

(1) (2)

Dependent variable BETA BETA

POSTBAPCPA 0.346*** 0.091*** (0.009) (0.012) PROFITABILITY -17.02*** (0.194) TANGIBILITY -0.751*** 0.764*** (0.043) (0.034) LEVERAGE -1.967*** 0.644*** (0.043) (0.049) LOG ASSETS -0.527*** 0.035*** (0.013) (0.002) DISTRESSED 0.166*** (0.013) DISTRESSED x POSTBAPCPA 0.803*** (0.021) CONSTANT 7.925*** (0.145) Observations 5,193 5,193 R-squared 0.670 0.851

Column 1 presents the results of the regression of BAPCPA on a firm’s beta. Column 2 distinguishes firms by whether they are distressed. I each model I find a significant and positive coefficient on POSTBAPCPA. The magnitude of the coefficient suggests that on average a firm’s beta increases by 0.346 after BAPCPA. This finding is consistent with hypothesis 4. In column 2 I find that the coefficient on DISTRESSED x POSTBAPCPA is significant and positive. This finding is consistent with hypothesis 4 that suggests that Betas of distressed firms increase more.

(23)

22 Table 6 presents the results of estimating model 5 and 6.

(1) (2) (3) (4)

Dependent variable CASH TO ASSETS CASH TO ASSETS CAPITAL

EXPENDITURES CAPITAL EXPENDITURES POSTBAPCPA 0.000898 0.00240* 19.21*** 47.55*** (0.000889) (0.00126) (4.091) (5.599) NET WORKING CAPITAL 1.44e-05*** (5.78e-07) CAPITAL EXPENDITURES -5.86e-05*** (1.23e-06) TANGIBILITY -0.0329*** -0.0352*** 895.1*** 446.0*** (0.00188) (0.00220) (16.97) (14.25) LEVERAGE 0.0859*** 0.114*** 538.9*** 846.3*** (0.00357) (0.00245) (16.03) (10.67) DIVIDEND -0.0334*** (0.00196) CONSTRAINED 0.0507*** -220.8*** (0.00210) (9.205) CONSTRAINED x POSTBAPCPA 0.0133*** 69.30*** (0.000966) (3.138) PROFITABILITY 0.326*** -65.00 (0.00934) (63.13) LOG ASSETS 337.1*** 17.45*** (4.729) (1.189) CASH TO ASSETS -2,468*** -2,946*** (52.51) (114.9) CONSTANT 0.101*** -2,998*** (0.00315) (51.42) Observations 5,197 5,197 5,197 5,197 R-squared 0.579 0.759 0.800 0.869

(24)

23 Column 1 presents the results for the regression of POSTBAPCPA on CASHTOASSETS. I find that a positive but no significant coefficient on POSTBAPCPA. The magnitude of this coefficient suggests that firms hold a bit more cash to assets than before the BAPCPA, but this coefficient is not significant. Column 2 presents the results for the regression of CONSTRAINED x POSTBAPCPA on CASHTOASSETS. I find a positive and significant coefficient on this interaction variable. This suggests that constrained firms indeed hold more cash after BAPCPA. This finding is consistent with hypothesis 5. Moreover, I find that the significance of POSTBAPCPA increased and is now significant at the 10% level.

Column 3 presents the results for the regression of POSTBAPCPA on CAPITAL EXPENDITURES. I find a positive and significant coefficient on POSTBAPCPA. The magnitude of this coefficient indicates that firms increase capital expenditures after the BAPCPA. Column 4 presents the results for the regression of CONSTRAINED x POSTBAPCPA on CAPITAL EXPENDITURES. I find a positive and significant sign on the coefficient. The magnitude of the coefficient indicates that constrained firms increase CAPITAL EXPENDITURES by 69.30. This finding is not consistent with my hypothesis 6. I do however find a negative and significant coefficient for CONSTRAINED. The magnitude of this coefficient is -220.8. This finding is consistent with my hypothesis 6. The magnitude of CONSTRAINED is larger than of the CONSTRAINED x POSTBAPCPA making the total effect negative.

(25)

24 5.3 Robustness checks

To asses the validity of my models, I constructed the following robustness checks.

Table 7. Robustness checks

(1) (2) (3)

Dependent variable LOG

CDSSPREADS CDSSPREADS LOG CDSSPREADLOG S POSTBAPCPA 0.235*** 0.357*** 0.135*** (0.0324) (0.0325) (0.0281) PROFITABILITY -20.15*** (0.597) TANGIBILITY -1.720*** -0.335*** 0.158** (0.0844) (0.0674) (0.0654) LEVERAGE 0.358** 3.036*** (0.146) (0.0954) LOG ASSETS -0.103*** 0.180*** 0.271*** (0.0225) (0.00486) (0.00353) DISTRESSED 1.913*** 2.315*** (0.0479) (0.0298) DISTRESSED x POSTBAPCPA -0.500*** -1.473*** (0.0651) (0.0457) CONSTRAINED 1.815*** (0.0792) CONSTRAINED x POSTBAPCPA 2.152*** (0.0905) POST x CONSTRAINED x DISTRESSED 0.895*** (0.0963) CONSTANT 6.064*** (0.280) Observations 5,189 5,189 5,189 R-squared 0.461 0.924 0.936

Table 7 presents the robustness tests for my first models. I find that taking the natural logarithm of the CDS spreads does not change my conclusion for hypothesis 1. In column

(26)

25 1, I find that BAPCPA is significant and positive. In column 2, I find that the interaction variable DISTRESSED x POSTBAPCPA becomes negative and this indicates failure of the robustness check. In column 3, I find that the POST x CONSTRAINED x DISTRESSED variable is still positive and significant.

Table 7. Robustness checks

(1) (2)

Dependent variable LONG TERM

DEBT TO ASSETS DEBT TO ASSETS SHORT TERM

POSTBAPCPA -0.00130* 0.0209*** (0.000718) (0.00304) CONSTRAINED 0.0192*** 0.0356*** (0.000871) (0.00258) CONSTRAINED x POSTBAPCPA 0.0256*** -0.0113*** (0.00125) (0.00176) LOG ASSETS 0.00356*** 0.0105*** (0.000122) (0.000328) TANGIBILITY -0.0155*** 0.266*** (0.00141) (0.00501) PROFITABILITY -0.0682*** 0.541*** (0.00731) (0.0181) Observations 5,197 5,197 R-squared 0.502 0.912

I find a negative but insignificant coefficient for POSTBAPCPA in column 1. This indicates that firms have lowered their long-term debt after BAPCPA. I find a positive and significant coefficient for POSTBAPCPA in column 2. This indicates that firms take on more short-term debt after BAPCPA.

(27)

26 6.0 Conclusion

In this paper, I examine the effect of BAPCPA on debt and equity holders. I predict that the risk of holding equity will increase and that debt holders will have more bargaining power in court. To test these predictions, I examine a sample of U.S firms and their CDS spreads and betas. I find that the CDS spreads indeed increase after BAPCPA and I also find that the firms’ beta increase. The increase in CDS spreads are stronger for distressed and constrained firms. The increase in beta is stronger for distressed firms. Lastly, I find that constrained firms hold more cash and reduce capital expenditures after the BAPCPA. My findings suggest that BAPCPA made debt holders better off at the expense of equity holders in bankruptcy. More importantly, the probability of a firm filing for bankruptcy has increased after BAPCPA. This explains the surge in CDS spreads and increase in firms’ beta. This also implies that the act is effective because BAPCPA was implemented to prevent abuse by debtors and give creditors a stronger position in bankruptcy court.

There are a few limitations. First, the sample size is not very large. I focus on a subset of 11 U.S firms and their CDS in 2004 to 2006. Because CDS grew exponentially in popularity from 2000 to 2008, this meant that the market was still relative small in 2004. Secondly, this study only uses one measure of distress, which is the Altman Z-score. There are many good other measures of distress such as the Merton’s distance-to-default model or the Merton-Vasicek model. Including these models of distress could increase the robustness of the study. However, the Altman Z-score is a widely accepted measure of distress by the literature. Also, this study uses the inability to pay-out dividends as an

(28)

27 indication that a firm is financially constrained. This is the measure used in the study of Fazzari et al. (1988). There are other measures of financially constrain such as cash flow volatility, firm size, firm age and credit rating. However, the Fazzari et al. (1988) study explains why using the pay-out ratio is a good measure of financial constrain.

This study contributes to the literature of policy effects on financing. The BAPCPA dramatically changed bankruptcy procedures. The positive effect for firms is that they can increase their debt capacity however equity holders lose bargaining power in court to debt holders. However, the act was put in place to protect lenders against consumers that file for personal bankruptcy or otherwise had small debts outstanding. But the act inadvertently changed the way businesses filed for bankruptcy. Furthermore, the act significantly reduced counterparty risk in a derivatives transaction (Giambona and Wang, 2014) while that was not its primary goal. I therefore want to emphasize that policy changes that are meant to change business dealings could cause a ripple effect in financial markets. There are still other facets of the BAPCPA that have not been studied yet. These could be the topic for further research. For example, it could be interesting to look at the change to chapter 15 included in the BAPCPA. This is the chapter that deals with cross-border bankruptcies and foreign assets.

(29)

28

Appendix A.

Variable Name Definition Data Source

1. Dependent Variables

CDSSPREADS The spread between the rate on the bond of the company

and a treasury bond measured in basis points. The maturity of this instrument is 1 year. I used Python to filter the firms that are (1) are not U.S firms. (2) are missing CDS spreads in my time window (2004-2006). (3) are not sorted by GVKEY

Campello Ladika & Matta

Database for CDS spreads.

Thomson Reuters Datastream

DEBT TO ASSETS The ratio between debt to assets for US firms between 2004

and 2006.

COMPUSTAT IQ Capital

BETA The beta of the companies measured as the covariance

between returns divided by the variance of the S&P500.

COMPUSTAT, CRSP

CASH TO ASSETS Cash and marketable securities divided by total assets. COMPUSTAT IQ, Capital

CAPITAL EXPENDITURES Compustat item Capx. This variable is winsorized at the

1-99% level

COMPUSTAT IQ, Capital 2. Explanatory Variables

POST BAPCPA A dummy variable that is equal to 1 on and after April 13th

of 2005. This is a week before the adoption of the BAPCPA Justice.gov

Z-score The sum of 1.2 _ (Current Assets { Current Liabilities)/Total

Assets +1.4 * Retained Earnings/Total Assets + 3.3 * EBIT/Total Assets + 0.6 _Market Capitalization/Total Liabilities + Total Sales/Total Assets. CurrentAssets is Compustat data item ACT, Current Liabilities is LCT, Total Assets is AT, Retained Earnings is RE, EBIT is EBIT, Total Liabilities is LT, and Total Sales is SALE. Market capitalization is the _rm's stock price (Compustat

data item prcc_f) multiplied by shares outstanding (csho). All Z-scores are winsorized at the (1-99%) level.

COMPUSTAT IQ capital

DISTRESS A dummy indicator variable that is equal to 1 if a firm has

an Altman Z-score lower than 1.81

CONSTRAINED An indicator variable that is equal to 1 if a firm cannot pay

any dividends. That is, if dvt/at = 0. This is aligned with the paper by Fazzari et al. (1988).

LOG ASSETS The natural logarithm of 1 plus total assets. This variable

is winsorized at the 1-99% level COMPUSTAT IQ Capital

TANGIBILITY This variable is measured as total property plant and

equipment over total assets. This variable is winsorized at the 1-99% level.

LEVERAGE The sum of long-term debt plus short-term debt plus

preferred stock divided by this number plus market capitalization. This variable is winsorized at the 1-99% level

PROFITABILITY This variable is constructed by dividing EBIT over total

assets. This variable is winsorized at the 1-99% level.

CAPITAL EXPENDITURES The amount of capital expenditures divided by total assets.

This variable is winsorized at the 1-99% level.

(30)

29

NET WORKING CAPITAL Working capital minus cash and cash equivalents. This

variable is winsorized at the 1-99% level

COMPUSTAT IQ, Capital

(31)

30 Reference list

Almeida, H., Campello, M., & Weisbach, M. (2011) Corporate financial and investment policies when future financing is not frictionless. The Journal of Corporate Finance, 17 675-693.

Arora, N., Gandhi, P., Longstaff, F. (2012) Counterparty credit risk and credit default swap market. Journal of Financial Economics, 103 280-293.

Asquith, P., Gertner, R., & Scharfstein., D. (1994) Anatomy of Financial Distress: An Examination of Junk-Bond Issuers. The Quarterly Journal of Economics, 109 625-658. Bates, T., Kahle, K., Stulz, R. (2009) Why Do U.S. Firms Hold So Much More Cash than They Used To? The Journal of Finance.

Berk, J. B., & DeMarzo, P. M. (2007). Corporate finance. Boston: Pearson Addison Wesley.

Benderson, J., & Whitaker, T. (2006) Bankruptcy Abuse Prevention and Consumer Protection Act of 2005 I.

Bolton, P., Oehmke, M. (2011). Credit Default Swaps and the Empty Creditor Problem. Review of Financial studies, 24 2617-2655.

Campello, M., & Matta, R. (2012). Credit Default Swaps and risk-shifting. Economics Letters, 117, 639-641.

Campello, M., & Matta, R. (2016). Investment Risk, CDS insurance and Firm Financing. 1-47. Working Paper.

Campello, M., Ladika, T., & Matta, R. (2018) Renegotiation Frictions and Financial Distress. Resolution: Evidence from CDS spreads 1-67. Working Paper.

Fazzari, S., Hubbard, G., & Petersen, B. (1988). Financing Constraints and Corporate Investments. Brookings Papers on Economic Activity, 1 141-206.

Giambona, E., Wang, Y. (2017) Derivatives supply and corporate hedging: Evidence from the Safe Harbor Reform of 2005. Working paper.

(32)

31 Hull, J. (2006). Options, futures, and other derivatives. Upper Saddle River, N.J: Pearson/Prentice Hall.

Li, J., Tang, Dragon Yongjun. (2016) The leverage externalities of credit default swaps. Journal of Financial Economics, 120 491-513.

Messick, D.M. & Van de Geer, J.P. (1981) "A reversal paradox." Psychological Bulletin 90.3 582.

Parlour, C., Winton, A. (2013). Laying off credit risk: Loan sales versus credit default swaps. Journal of Financial Economics, 107 25-45.

Purnanandam, A. (2008) Financial distress and corporate risk management: Theory and evidence. Journal of Financial Economics, 87 706-739.

Rampini, A., Viswanathan, S. (2010) Collateral, Risk Management, and the Distribution of Debt Capacity. Journal of Finance, 65.

Subrahmanyam, M., Tang, Dragon Yongjun., & Wang, Sarah. (2015) Credit Default Swaps, exacting creditors and corporate liquidity management. Journal of Financial Economics, 124 395-414.

The Bankruptcy Abuse and Consumer Protection Act of 2005.