The cost of U.S. sovereign default on equity market : based on credit default swap spreads

(1)

1

The Cost of U.S. Sovereign Default on Equity Market:

Based on Credit Default Swap Spreads

Master Thesis

Program: Master in International Finance

Student: Qizhi Hu

Student Number: 10430164

Thesis Supervisor: Rafael Matta

(2)

2

Abstract

The sovereign credit default swap contracts insure creditors against country default and the spreads reflect the probability of sovereign default. In this thesis, we monitor the behavior of U.S. sovereign CDS spreads to estimate changes in the probability of sovereign default induced by the debt ceiling crisis in 2011 and 2013. Based on this, we assess the cost of sovereign default on the U.S. equity market value. As expected, the U.S. CDS spreads and implied probability of sovereign default increased in response to the debt ceiling crisis. In particular, we estimated a 4% and 5% increase in the 5-year sovereign default probability triggered by the 2011 and 2013 debt ceiling debacles, respectively. Further, using S&P 500 index as the proxy, we estimate the cost of sovereign default on equity market to be $ 8.7 billion.

(3)

3

The U.S. Congress is entitled to borrow money on the credit of the country and is therefore mandated to exercise control over federal debt.The enactment of Bipartisan Budget Act of 2015 suspended the debt limit till 16 March 2017, which was thereafter reset upward to $19,809 billion [1]. It was then expected that the U.S. Treasury could again meet federal obligations in early October 2017 [2] and the Congress was urged to take actions before extraordinary measures run out on September 29, 2017 [3].

Looking back, the 2011 debt ceiling crisis attracted far more attention than the other debt limit episodes. On May 16, 2011, U.S. Treasury Secretary Timothy Geithner announced that the federal debt had reached its statutory limit and declared a debt issuance suspension period, which would allow certain extraordinary measures to extend Treasury’s borrowing capacity till around August 2, 2011 [4]. After the introduction of Budget Control Act, on 2 August 2011, Obamasigned into law this measure, eventually allowing a series of increases in the debt limit up to $2.4 trillion. This Act also eliminated the need for further increases of debt limit until early 2013[5].

As anticipated, federal debt reached its limit onDecember 31 2012, although the debt limit was suspended until May 19. Thereafter, a new debt issuance suspension period was declared, allowing extraordinary measures to be taken for meeting federal obligations until 2 August 2013 [6]. With the passing of the Continuing Appropriations Act, the debt ceiling crisis ended on 17 October 2013, allowing a suspension of the debt limit through 7 February 2014. As revealed by the debt limit episodes, the debt limit provisions enacted in 2011 and October 2013 are resembling.

Researchers in this field have attempted to measure the cost of sovereign default, although in many cases the occurrence of the event itself is also accompanied by worsened economic

(5)

5

environments. Thus, a sound approach involves the identification of an exogenous shock that impacts the probability of default yet not influences the fundamentals of the economics.

The event of debt ceiling crisis serves as an exogenous shock to that allows us to identify the causal effect of sovereign default. This is because the debt ceiling crisis is often deemed as an outcome of a political decision making process and not a as a result of changes in economic fundamentals. Thus, this unique event becomes an exogenous shock to the market, and therefore can be exploited for financial analyses. While announcements worsening the debt ceiling crisis raise the probability of US sovereign default, episodes that keep the sovereign from hitting the debt ceiling (e.g. through the raise of debt limit) lower this probability. Of note, this assessment is valid based on the assumption that debt ceiling-related announcements revealed to market participants affects the U.S. equity market value only through the effect on the US sovereign’s probability of default. We consider this assumption plausible as the equity value of U.S. firms are not directly affected by the fiscal and political decisions on debt limit raise, and that the U.S. firms are legally separate from the federal government, without asset attachment.

In the present study, we measure the changes in the probability of default through Credit Default Swap (CDS) spreads. Single-name CDS contracts are written agreements between a buyer and a seller, where the buyer pays the seller a periodic fee, i.e. the CDS spread, and the seller makes a lump-sum payment if the underlying reference experiences a credit event. In case of this study, such payments from CDS contracts are only triggered by the event of sovereign default. In other words, the hitting of debt ceiling could trigger sovereign default and thereby a payment on U.S. CDS contracts. Accordingly, the elicited changes in sovereign default probability during the debt ceiling crisis can be reflected by the CDS spreads. In contrast to corporate CDS, the sovereign CDS are denominated in Euros to separate the sovereign risk that the contract ensures from the payments made on this contract. It was estimated that the average gross (net) notional amount of outstanding US CDS is $17 ($3.2) billion [7].

(6)

6

In this thesis, we compile key events in congressional debt ceiling debacles of 2011 and 2013 and isolate the events that alter the probability of sovereign default. A similar approach was previously adopted by Hébert et al., who used Argentinian corporate CDS to measure the effect of changes in the default probability on equity returns of Argentinian firms [8]. Moreover, we use the S&P 500 index to reflect the U.S. equity markets. The index includes 500 leading companies in leading industries. Although being a single gauge for large-cap U.S. equities, S&P 500 is deemed representative as it captures approximately 80% coverage of available market capitalization.

In the present study, we traced the changes in both 1-year and 5-year U.S. CDS spreads before, during and after the debt ceiling crisis to estimate the effect of 2011 and 2013 debt ceiling on the probability of sovereign default, and the on the stock market value using S&P500 index. By understanding these values, we further calculate the cost of U.S. sovereign default on the equity market.

The rest of the thesis is structured as follows. Chapter 2 describes the testing methodology. Chapter 3 addresses the effect of 2011 debt ceiling crisis on the change of sovereign default probability, as well as our estimation for the costs of sovereign default on the equity market. In Chapter 4, we re-evaluate this model using the 2013 debt ceiling crisis as an example. Discussions are provided in Chapter 5 and Chapter 6 concludes.

(7)

7

Chapter 2. Methodology

In this section, we present model-based methodology for how debt ceiling crisis affects the CDS spreads, the probability of sovereign default as well as the equity market value.

2.1. Hypotheses

CDS spreads reflect the price creditors are willing to pay to insure against default risk, and thereby can be exploited to estimate the probability of default. We hypothesize that

 Hypothesis 1: the CDS spreads of particularly the 1-year contact and the corresponding

implied probability of sovereign default should arise when the likelihood of hitting debt ceiling increases; yet should decline when the debt ceiling crisis is resolved. Moreover, the magnitude of such changes should be more prominent when it comes to a game-changing announcement or event.

 Hypothesis 2: the total value of the U.S. equity market, taking S&P500 indices as

proxies, should decline when the probability of sovereign default elevates.

2.2. Assessing the effect of debt ceiling crisis on sovereign default probability

2.2.1. Data sources

We collect a time series of 1-year and 5-year U.S. CDS spreads in the year of 2011 and 2013 from Thomson Reuters's DataStream. We obtain the daily yield from the U.S. Department of the Treasury [9] and set the yield on 5-Year US Treasury bond as annual interest rate. The index file on S&P 500 was downloaded from The Center for Research in Security Prices (CRSP).

(8)

8

2.2.2. Estimating the probability of default based on observed CDS spreads

We test Hypothesis 1 by tracing the 1-year and 5-year U.S. CDS spreads prior to, during and after the debt ceiling period. We follow the model initially reported by Duffie [10] to

calculate the probability of default implied by the CDS spreads. Using the reduced-form model (that assumes the default follows a Poisson process), we follow the procedures reported by Campello et al. [11] and obtain the equation (1):

Eq 1

where h is the default hazard rate (per time period), y is the interest rate, and t is the payment period. This model is solved for 5-year CDS spreads as well as 1-year CDS spreads with quarterly payouts. The annual interest rate was set based on the yield of a zero-coupon 5-year U.S on each day. The recovery rate on CDS-referenced debt is set to 60% based on the average recovery rates on defaulted sovereign bond issuers [12].

To estimate the probability of default, we set the CDS spread equal to the observed daily value and solve for the hazard rate [11]. We then calculate the implied 1-year and 5-year probability of default using the following equation:

𝑃𝐷 = 1 − exp⁡(−ℎ ∗ 𝑡) Eq 2

Where PD is the implied probability of default, h is the determined default hazard rate (per time period) and t is the CDS spread payment period.

(9)

9

2.2.3. Estimating CDS spreads based on regression

Next, we examine how debt ceiling crisis affected US CDS spreads using the following regression model (Eq 3):

y i,t = α +β1× treatment i,t +β2 × treated i,t +β3 × treatment i,t × treated i,t + control + error Eq3

where y i,t is the CDS spread of type i (1-year or 5-year) at day t; treatment equals 1 if day t

is part of the debt-ceiling crisis period, otherwise 0; treated equals 1 if the CDS spread is for the 1-year contract, and equals 0 in case of 5-year contract.

The coefficient β3 would indicate the effect of debt ceiling crisis on the 1-year CDS spreads while controlling for the changes in the 5-year CDS spreads.

2.3. Estimating the cost of sovereign default on equity market

Assume the market value of U.S. equity is M0 before an event, such as sovereign default, and M1 is the market value after the event takes place, we obtain the equation:

M0 = (1- P0) × V + P0 × (V-C) Eq 4

In the event of sovereign default, M1 = (1- P1) × V + P1 × (V-C) Eq 5

where P0 and P1 are the probability of default before and after the default event; V is the value

of equity market in the absence of sovereign default; and C is the cost of sovereign default for the equity market.

Combining Eq 4 and Eq 5, we obtain Eq 6:

(10)

10

Thus, by calculating the changes in the probability of default through the CDS spreads, as well as the changes in the equity market value using S&P 500 indices as the proxies, we can estimate the cost of sovereign default.

Chapter 3. Empirical results

3.1. Changes in U.S. CDS spreads around 2011 debt ceiling crisis

We first demonstrate the effects of 2011 debt ceiling on the changes in U.S. CDS spreads graphically. The US CDS spreads of the 1-year contract around the year of 2011 was displayed in Figure 1, using the most liquid 5-year CDS contract as control. Intuitively, the US CDS spreads will increase when the probability of sovereign default is anticipated to go up due to debt ceiling crisis. Indeed, the 1-year CDS spread increased from the pre-crisis level 9 bps to the crisis peak level 85 bps. After passage of the debt ceiling deal, the 1-year CDS spread declined and eventually arrived at the pre-crisis level by 19 October 2011.

(11)

11

Figure 1. US CDS spreads of 1-year and 5-year contracts in 2011(top) and zoomed in (bottom)

The CDS spreads did not return to the pre-crisis level immediately due to downgrading of the U.S. government's credit rating following the debt-ceiling crisis, as well as fears of contagion from the Eurozone crisis. These concerns resulted in the largest S&P 500 monthly loss since the Financial Crisis and tripling of the VIX volatility index.

(12)

12

Expectedly, while the 5-year CDS spread responded moderately to the debt ceiling crisis, a noticeable spike in the spread of 1-year CDS contract was observed around 28 July 2011, indicating a substantial change in sentiment about sovereign default in the short-term. In addition, the longer-term treasury bills essentially remained the same, partially suggesting that the Treasury market is only concerned about repayment risk over the next month(s).

3.2. Changes in probability of default around 2011 debt ceiling crisis

We continue to assess the changes of implied probability of sovereign default during 2011 debt ceiling period through CDS spreads.

Table 1 highlights the most important events steering the development of 2011 and 2013 debt ceiling crisis, giving rise to the changes of default probabilities, manifested as variations in the CDS spreads. Using Eq 1, we calculate the daily implied probability of default during the 2011 debt ceiling period using CDS spreads of either 1-year or 5-year contracts. The changes in implied probability of default are depicted it in Figure 2.

(13)

13

Table 1.

Key events occurred during the debt ceiling crisis of 2011 and 2013.

Date Events

16 May 2011 The United States hit the $14.3 trillion debt limit

25 May 2011 The Senate rejected both the Republican House budget proposal and the Obama budget proposal

14 July2011 The credit rating agency Standard and Poor's unveiled it is placing the U.S. sovereign rating on CreditWatch with negative implications 22 July 2011 The Senate voted along party lines to table the “Cut, Cap and Balance” Act; Obama had promised to veto the bill even had it passed

Congress

26 July 2011 Boehner's plan ran into trouble as the Congressional Budget Office failed to reduce spending and deficits as much as advertised 30 July 2011 Two sources familiar with negotiations between the White House and congressional leaders said the framework of a deal is being put

together

31 July 2011 Obama announced a deal between his administration and congressional leaders had been reached, pending congressional approval 1 August 2011 The U.S. House passed the debt ceiling

2 August 2011 The Senate approved the measure, which was signed by Obama

25 September 2013 Treasury announced that extraordinary measures is expected to be exhausted latest on 17 October 2011 01 October 2013 Shutdown began. The House requested negotiations with the Senate, but was rejected

(14)

14

Figure 2. Implied 5-year probability of U.S. sovereign default during 2011 debt crisis period based on

CDS spreads of either 1-year or 5-year contracts

Intuitively, the patterns of changes in CDS spread and in the implied probability of default are similar. Same as for the CDS spreads, after the 2011 debt crisis is resolved, the implied probability of default also reduced to pre-crisis level. Moreover, the magnitude of changes in CDS spreads (Figure 1) or default probability (Figure 2) is in good agreement with the announcement of key events listed in Table 1.

3.3. Estimating the cost of sovereign default on equity market based on observed CDS spreads

In Chapter 2, we hypothesized that the event of debt crisis results in a plunge in the market value of U.S. equity. Next, we examine the casual effect of sovereign default on the U.S. equity market, using the total market value of S&P 500 indexes as proxies for the latter. Empirically, we this hypothesis by estimating the decrease of total equity market value around event days

(15)

15

(see Table 1 for details). Further, we also calculate the cost of sovereign default using Eq 6 in section 2.2.3.

We summarize the analysis results in Table 2, using either a 2-day time window (two-sided) or 4-day time window (two sided) around the event day. To mention, in case two event days are too adjacent and accordingly the time-windows heavily overlap, we either slightly adjust the time window, or exclude the event from analysis.

As shown in Table 2, the calculated costs of sovereign default varied a great deal between different time-windows, likely owing to the volatility of the swap market during the selected time periods. Thus, it is important to collect sufficient data points, perform regression analysis on the CDS spreads, and then determine the cost of sovereign default using the estimated CDS spreads before and after the debt ceiling crisis.

(16)

16

Table 2.

Calculated cost of sovereign default on equity market value based on observed 1-year CDS spreads during 2011 debt ceiling crisis period.

Event date

S&P500 total market value

($B) Implied probability of default

2 Δ S&P500 total

market value ($B) ΔD

3 Cost of sovereign default

on equity market ($B) - 4 d + 4 d - 2 d + 2 d - 4 d + 4 d - 2 d + 2 d ± 4 d ± 2 d ± 4 d ± 2 d ± 4 d ± 2 d 16 May 2011 12,66 12,44 12,59 12,51 1,26% 3,00% 1,26% 1,52% -0,2205 -0,0771 1,74% 0,25% 12,71 30,44 25 May 2011 12,54 12,26 12,29 12,42 1,76% 4,62% 3,06% 4,73% -0,2778 0,1253 2,85% 1,67% 9,73 -7,48 31 May 2011 12,32 12,00 12,42 12,24 3,90% 4,44% 4,73% 4,62% -0,3165 -0,1736 0,54% -0,11% 58,32 -151,35 14 July 2011 12,57 12,4 12,29 12,21 5,98% 6,56% 5,74% 7,08% -0,1695 -0,0786 0,58% 1,33% 29,32 5,89 19 July 20111_12,21 _12,4 _12,31 _12,57 _7,08% _6,56% _7,08% _6,73% _0,1908 _0,2566 _-0,52% _-0,35% _36,67 _74,2 22 July 20111_12,57 _12,51 _12,4 _12,46 _6,73% _7,65% _6,56% _9,19% _-0,0588 _0,059 _0,92% _2,63% _6,39 _-2,24 25 July 20111_12,58 _12,46 _12,57 _12,21 _6,62% _9,19% _6,73% _9,75% _-0,1194 _-0,3605 _2,57% _3,01% _4,65 _11,97 26 July 20111_12,51 _12,21 _12,58 _12,17 _7,65% _9,75% _6,62% _10,08% _-0,3016 _-0,4114 _2,09% _3,46% _14,42 _11,89 1

for these days, a 1-day window was taken for analysis instead of a 4-day window as adjacent events are too close

2_{based on 1Y CDS data}

3 _{ΔD = change of implied 5-year probability of US sovereign default}

Note: the S&P 500 data in August 2011 was not collected for calculation considering that i) the debt ceiling crisis was already resolved at the end of July 2011 and that ii) the stock market in August 2011 fell due to the downgraded rating of the U.S. and the concerns for the contagion of European sovereign debt crisis

(17)

17

3.4. Regression analysis of CDS spreads

We identified two important parameters that influence the CDS spreads, namely i) the

contract duration, i.e. 1-year CDS spread or 5-year CDS spread; ii) if the CDS spread belongs to the debt crisis period. We regressed the CDS spreads that cover the three time periods, i.e. before, during after the debt ceiling crisis on these identified parameters using the equation y

i,t = α +β1× treatment i,t +β2 × treated i,t +β3 × treatment i,t × treated i,t + control + error (Eq 3).

As already described in section 2.2.2, y i,t is the CDS spread of type i at day t where

treatment equals 1 if day t is part of the debt-ceiling crisis period, otherwise 0 and treated equals 1 if the CDS spread is for the 1-year contract, otherwise 0.

As shown in Panel A of Table 3, 152 observations were included in this regression analysis. The coefficient β1 is positive, in line with real-time observations that the CDS spread is higher during the debt crisis period. However, this coefficient is small and insignificant, suggesting that the 5-year CDS spread serves as a good control as it does not react strongly to the debt ceiling crisis. The coefficient β2 is negative, indicating that in the absence of debt crisis, the 1-year CDS spread is lower than the 5-year CDS spread, again in line with our observations as shown in Figure 1. As shown in Panel C, the double interaction coefficient for treatment * treated (i.e. CDS type * in/out of debt ceiling crisis period) is positive and

significant, suggesting that the 1-year CDS spread is much more responsive to the debt-ceiling crisis than the 5-year CDS control arm.

(18)

18

Table 3.

Regression analysis on CDS spreads in 2011

Panel A: regression statistics

Multiple R R Square _{Adjusted R Square Standard Error Observations}

0,6504 0,4231 0,4114 10,157 152 Panel B: ANOVA df SS MS F Significance F Regression 3 11195,61 3731,87 36,17 _{1,35023107090366E-17} Residual 148 15268,23 103,16 Total 151 26463,85 Panel C: Coefficients

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90,0% Upper 90,0%

Intercept 47,94 3,39 14,16 - 41,25 54,63 42,34 53,54

β1 4,35 3,61 1,21 0,230 -2,77 11,48 -1,62 10,32

β2 -31,13 4,79 -6,5 < 0,005 -40,59 -21,66 -39,05 -23,2

(19)

19

3.5. Estimating the cost of sovereign default on equity market based on CDS spread regression

Using the regression equation obtained in section 3.4., we calculate the 1-year CDS spread during the 2011 debt ceiling crisis to be 42.4 bps (equals 47.94 +4.35 – 31.13+21.23 = 42.39). We also calculated the average CDS spread and total market value of S&P 500 index

observed during the pre-crisis period (10-day window).

Using Eq 1, Eq 2 and Eq 6, we performed the following calculation (see Table 4). As indicated in the table, the 2011 debt ceiling crisis gave rise to an approx. 4% increase in the probability of sovereign default. The decrease in the market value of S&P 500 index can be translated into 3% of the total equity value. Further, based on 2011 U.S. CDS spreads, we estimate the cost of sovereign default on equity market to be $9.88 billion.

Table 4.

Estimated cost of sovereign default on equity market value based on 2011 debt CDS spreads.

Pre-crisis level Level during crisis

Calculated 1-year CDS Spread 9,80 42,39

Hazard Rate (per time period) 0,061% 0,265%

Implied probability of sovereign default 1,217% 5,155%

Total market value S&P 500 ($B) 12,56 12.211

Cost of sovereign default ($B) 8.73

1

(20)

20

Chapter 4. Re-evaluate methodology using 2013 debt ceiling crisis as an

example

Using the same methodology we estimated the change of sovereign default probability induced by the 2013 debt ceiling crisis. As shown in Table 5, we first calculated the cost of sovereign default on equity market value based on observed 1-year CDS spreads around event days during 2013 debt ceiling crisis period (Table 1). Next, we regressed the CDS spreads before, during and after the 2013 debt ceiling crisis using equation 3.

As shown in Panel A of Table 6, 66 observations were included in this regression analysis.

Similar to our observations with observed 2011 CDS spread data, the coefficients β1and β2 are positive and negative, respectively. As shown in Panel C, the double interaction

coefficient for treatment * treated (i.e. CDS type * in/out of debt ceiling crisis period) is positive and significant, suggesting that the 1-year CDS spread is much more responsive to the debt-ceiling crisis than the 5-year CDS control arm. More importantly, this double interaction coefficient is comparable to that derived from the year of 2011 (24.26 vs. 21.23), supporting the hypotheses of our study.

(21)

21

Table 5.

Calculated cost of sovereign default on equity market value based on observed 1-year CDS spreads during 2013 debt ceiling crisis period.

Event date

S&P500 total market value

($B) Implied probability of default

1 Δ S&P500 total

market value ($B) ΔD

2 Cost of sovereign default

on equity market ($B) - 4 d + 4 d - 2 d + 2 d - 4 d + 4 d - 2 d + 2 d ± 4 d ± 2 d ± 4 d ± 2 d ± 4 d ± 2 d 24 Sept 2013 15,80 15,55 15,64 15,53 0,78% 4,27% 1,03% 4,09% -0,2515 -0,1038 3,49% 3,06% 7,21 3,39 01 Oct 2013 15,55 15,38 15,53 15,40 3,97% 6,92% 4,09% 6,16% -0,1762 -0,1308 2,94% 2,07% 5,99 6,33 17 Oct 2013 15,63 16,02 15,58 16,00 7,50% 4,81% 7,44% 4,93% 0,3933 0,4249 -2,69% -2,51% 14,62 16,91 1_{based on 1Y CDS data}

(22)

22

Table 6.

Regression analysis on CDS spreads in 2013

Panel A: regression statistics

Multiple R R Square _{Adjusted R Square Standard Error Observations}

0,7793 0,6073 0,5883 10,265 66 Panel B: ANOVA df SS MS F _{Significance F} Regression 3 10102,13 3367,375 31,95621 1,31E-12 Residual 62 6533,229 105,3747 Total 65 16635,35 Panel C: Coefficients

Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Lower 90,0% Upper 90,0%

Intercept 25,02153 2,650467 9,440422 1,32E-13 19,72332 30,31974 19,72332 30,31974

β1 9,222905 3,58875 2,569949 0,013 2,049097 16,39671 2,049097 16,39671

β2 -9,47253 3,748327 -2,52714 0,014 -16,9653 -1,97974 -16,9653 -1,97974

(23)

23

Same as in Chapter 3, we calculate the 1-year CDS spread during the 2013 debt ceiling crisis to be 49.0 bps following the regression equation. We also calculated the average CDS spread before the crisis, and then implied 5-year probability of sovereign default using Eq 1 and Eq 2. As indicated in Table 7, the 2013 debt ceiling crisis gave rise to an approx. 5% increase in the probability of sovereign default, similar to the calculated 4% increase for 2011 debt ceiling crisis.

Table 7.

Implied probability of sovereign default based on 2013 U.S. CDS spreads.

Pre-crisis level Level during crisis

Calculated 1-year CDS Spread 5,99 49,03

Hazard Rate (per time period) 0,037% 0,306%

(24)

24

Chapter 5. Discussions

In the past decade, a great number of research papers have addressed the cost of sovereign default for various financial aspects [13-16] and many corresponding methodologies have been explored by researchers in this field [17]. Directly measuring the size of sovereign default cost could be biased, as governments almost inevitably default in response to worsened economic environments, making the measurement of sovereign default cost misleading.

For these reasons, exogenous shocks are preferred with which natural experiments [18] can be conducted to evaluate the casual effects of sovereign default. For example, in an earlier paper, Campello and co-authors used a change of U.S. tax code as the shock to measure the cost wedge between in- versus out-of-court distress resolution [11]. Another paper used the legal rulings of a law case between the Republic of Argentina and a hedge fund as the event to assess the costs of sovereign default.

In this study, we used the U.S. debt ceiling crisis as the exogenous shock to identify the effect of sovereign default on the market value of U.S. firms. The unique nature of this natural experiment is that the default risk is not contributed by deteriorated economic conditions, but simply driven by political and fiscal decisions of the U.S. government. Thus, by determining the change of sovereign default probability triggered by this event, the unbiased cost of sovereign default can be attained. Intuitively, CDS spread serves as an attractive tool to assess the former. Thereby in this study, we traced the behavior of the U.S. CDS spread around the debt ceiling crisis, based on which calculated the changes in the probability of default. Further, using the S&P 500 index as the proxy for equity market, we estimated the cost of U.S. sovereign default on the equity market.

In the paper of Hébert et al., the authors found that a 10% increase in the default probability of Argentina caused a 6% decline in the value of Argentine equities [8]. In this present study,

(25)

25

using 2011 U.S. debt ceiling as the exogenous trigger, we found that a 4% increase in the probability of sovereign default resulted in a 3% decrease in the U.S. equity value. Despite the intrinsic differences in these two economic entities, the cost of sovereign default on the equity market seems to be of the same magnitude.

Admittedly, experiment in the present study is also challenged by two major limitations. First, CDS may not always represent a robust measure of sovereign default risk. As indicated by the analysis of Badaoui et al. (via a factor model approach) [19], thesovereign CDS spreads are highly driven by liquidity (55.6% of default risk and 44.32% of liquidity), suggesting that CDS cannot accurately reflect the probability of default when abnormal liquidity is observed. In this study, the influence of CDS liquidity is not deemed a concern, and the most liquid 5-year U.S. CDS spread was used as control. Having said that, in the same paper, the authors pointed that sovereign bond spreads are less subject to liquidity frictions (73% of default risk and 26.86% of liquidity) and may represent an alternative proxy for sovereign default risk, particularly if the CDS tool is not available.

Another limitation of the present experiment lies in the measurement of changes in the equity market value. We hypothesized that the equity market value should decline in response to the debt ceiling crisis. Nonetheless, when the economic conditions are relatively dynamic, e.g. during economy regrowth as in the case of 2013 fall, the effect of increased sovereign default risk (e.g. triggered by debt ceiling crisis) is likely to be overshadowed. For example, during the period of debt ceiling crisis in 2013, the market value of S&P 500 still manifested a noticeable trend of increase. This observation suggests that the 2013 debt ceiling crisis may not be an appropriate exogenous shock, if the purpose is to gauge the casual effect of sovereign default on the equity market. Because of the complexity of equity market, in the future, sufficient controls should be applied to have a better insight into the effect of sovereign default on the equity market.

(26)

26

Chapter 6. Conclusions

In this thesis, we measured the cost of U.S. sovereign default on the equity market through the change of sovereign default probability estimated by CDS spreads. We observed that the U.S. CDS spreads and implied probability of sovereign default increased during the debt ceiling crisis in 2011 and 2013. We found that the 2011 and 2013 debt ceiling crisis caused a 4-5% increase in the 5-year implied probability of sovereign default. Using the S&P 500 index as the proxy for the U.S. equity market, we estimate the cost of sovereign default on U.S. equity market to be $ 8.7 billion.

References

1. U.S. Treasury, Daily Treasury Statement, March 16, 2017.

2. Phillips, A., The Debt Limit: More Unpredictable than Usual. Goldman Sachs U.S. Daily, August 8, 2017. 3. https://www.treasury.gov/initiatives/Documents/Mnuchin%20to%20Ryan%20on%20 DISP%20-%207-28-17%20(2).pdf. 4. http://www.treasury.gov/connect/blog/Documents/20110516Letter%20to%20Congres s.pdf.

5. CRS Report R41965, The Budget Control Act of 2011.

6.

http://www.treasury.gov/initiatives/Documents/Debt%20Limit%20Letter%202%20Bo ehner%20May%2020%202013.pdf.

7. Augustin, P., Sovereign credit default swap premia. Journal of Investment Management, 2014. 12: p. 65-102.

(27)

27

9. https://www.treasury.gov/resource-center/data-chart-center/interest-rates/Pages/default.aspx.

10. Darrell, D., Credit Swap Valuation. Financial Analysts Journal, 1999. 55(1): p. 73-87.

11. Campello, M., T. Ladika, and R. Matta, Debt Restructuring Costs and Firm Bankruptcy: Evidence from CDS Spreads. 2015.

12. 2017, M.s.i.s.J., Sovereign Default and Recovery Rates 1983-2016.

13. Borensztein Eduardo, U.P., The Costs of Sovereign Default. IMF Staff Papers, 2009. 56: p. 683-741.

14. Acharya, V., I. Drechsler, and P. Schnabl, A Pyrrhic Victory? Bank Bailouts and Sovereign Credit Risk. The Journal of Finance, 2014. 69(6): p. 2689-2739.

15. Arteta, C. and G. Hale, Sovereign debt crises and credit to the private sector. Journal of International Economics, 2008. 74(1): p. 53-69.

16. Gornemann, N., Private Investment, and Economic Growth. 2014.

17. Aguiar, M. and G. Gopinath, Defaultable debt, interest rates and the current account. Journal of International Economics, 2006. 69(1): p. 64-83.

18. Nicola Fuchs-Schuendeln, T.A.H., Natural Experiments in Macroeconomics. National Bureau of Economic Research Working Paper Series, 2015. 21228.

19. Badaoui, S., L. Cathcart, and L. El-Jahel, Do sovereign credit default swaps represent a clean measure of sovereign default risk? A factor model approach. Journal of Banking & Finance, 2013. 37(7): p. 2392-2407.

The cost of U.S. sovereign default on equity market : based on credit default swap spreads