The purpose of this paper is to address this increasing risk of Treasuries due to greater government involvement in the financial markets involving taxpayer money. A common measure of default risk is to use interest rate spreads on securities with comparable characteristics, such as, maturity, liquidity, and taxation.1 This problem is not easy to address for Treasuries because they are usually used as the benchmark for default risk. They are rated as AAA by the credit rating agencies. For example, if Treasuries are becoming riskier, other AAA-rated and lower-rated securities may also become riskier. Another problem is a “flight to quality” situation where the prices for Treasuries are bid up and the interest rate on them decreases relative to other comparable securities. As is generally known2 this problem is more likely to affect shorter-term securities than longer-term securities. During the financial crisis other countries, particularly the United Kingdom and European countries, were also being affected.3 Therefore, Treasuries could become riskier but still look favorable when compared to other countries where risk may be increasing. The domestic and foreign demand for an increasing supply of US Treasuries may affect the Treasury rates. Credit default swaps (CDS) on Treasuries are also a direct measure of the default risk but may be limited due to the size of the CDS market relative to the Treasuries market. It is normally assumed that Treasuries would never default.
According to the
We choose the spread between the 10-year swap and the 10-year US Treasury security (constant maturity) as the measure of the risk associated with the Treasuries.4 Haubrich (2001) discusses how LIBOR rates include a default risk component but remain a good alternative benchmark. The spread is defined as the rate on the 10-year swap minus the rate on a 10-year Treasury constant maturity and will hereafter be referred to as SWLTR. See Fig. 1 for the behavior of the daily spread, SWLTR. One can see that the SWLTR is well behaved in January 2006 to June 2007. Then it becomes more volatile and starts a deep decline around August 2008. If the Treasuries are becoming more risky then the normally positive SWLTR will decrease. In other words, as the Treasuries become more risky relative to the 10-year swaps, the change in SWLTR should be negative. The necessary assumptions to make this statement are in the Section 3, methodology, where we attempt to control for other factors that may also affect the spread change.
Other factors that may affect SWLTR are the increasing supply of Treasuries to finance the various programs. If the supply is greater than demand then the Treasury rate will increase to attract investors. If the Treasury rate increases more than the swap rate increases, the SWLTR will decrease in a manner consistent with an increase in Treasury risk. As the capital positions for banks improve the systemic risk component in the swap will decline and the swap rate will decrease. If the Treasury rate is held constant and the swap rate decreases, the SWLTR will decrease in a manner consistent with an increase in Treasury risk. An increase in the systemic risk of the LIBOR swap relative to Treasury rates will lead to an increase in SWLTR which is consistent with a decrease in Treasury risk relative to the other benchmark, the swap rate. We examine this spread for the period January 3, 2006 -March 20, 2009. We analyze the percentage changes in SWLTR, SWLTRPCH, for 29 events, which due to closeness in time are reduced to 20 test events. Thirteen of the tested 20 events are significantly negative, six events are significantly positive, and the one remaining event is not significant. The overall picture is one of a declining SWLTR and riskier Treasuries. For example, the SWLTR declines from 0.53 on January 3, 2006 to 0.28 on March 20, 2009.
The remainder of the paper is structured in the following way. Section 2 discusses the literature related to the default risk of US Treasury securities. Section 3 outlines our methodology. Section 4 provides the empirical results and Section 5 provides the summary and conclusions.
While it is normally assumed that US Treasury securities never default, history has proved that this could be wrong on occasion. While Treasury securities issued by a sovereign are generally considered to be the closest to being “risk-free” several sovereigns have defaulted over the past few decades. As mentioned in Dungey et al. (2008), Russia announced its default on Treasury bonds on August 17, 1998. In the case of the United States, where finance textbooks routinely mention that Treasury securities are “risk-free”, the same assumption is used. This misconception persists because of the general belief that the government can “print” money to pay the securities. It is also believed that a government can tax its citizens and pay off its debt. However, on closer examination it is clearly evident that neither of these is a viable option in the long run for a democratic government. The first option, if used unwisely can lead to hyperinflation making paper money worthless while the use of the second option is limited by the checks and balances that exist in a democracy to counteract high taxes. Due to these reasons among others and the fact that the US Treasury defaulted earlier (Zivney and Marcus, 1989) and came very close to defaulting a second time (Nippani et al., 2001) the question of default-free Treasury securities has come into question.
Zivney and Marcus (1989) show that there was a 60 basis point increase in Treasury bill rates at the initial occurrence of a default by the United States Treasury in May 1979. As they mention, the Treasury blamed the delay on several factors including the failure of Congress to act in a timely fashion on the debt ceiling legislation and technical difficulties in processing checks. On another occasion, between mid-October 1995 and March 1996, the Treasury nearly defaulted. This was due to a disagreement between the White House and Congress over the increase of the federal debt limit and as shown by Nippani et al. (2001), the market charged a default risk premium on Treasury bills at this time. In both studies the workings of the government have at least in part led to the default or potential default. The potential of Treasury default is also the subject of a more recent study by Liu et al. (2009). They examine whether the financial market charged a default risk premium to the US Treasury securities when the US Federal government had approached the debt limits between 2002 and 2006. They show that for the first two occurrences, the market charged a small default risk premium and did not for the last two because the market anticipated that an agreement would be reached ensuring that there would be no long-term standoff. There are also other empirical studies and discussions that question Treasury debt being considered risk free. Kotlikoff (2006) asks the question “Is the United States Bankrupt?” He cites a study conducted by Gokhale and Smetters (2005) which shows that the fiscal gap for the United States is $65.9trillion. He notes that this is five times the GDP of the country and twice the size of the national wealth. In his discussion of Kotlikoff’s (2006) study, Thakor (2006) offers a measured assessment of the burgeoning fiscal gap: “Although I do not agree with the implicit assertion that the US fiscal gap puts the country in the same position as a bereft and destitute firm that is bankrupt and on the verge of liquidation, I do agree that the current state of affairs is alarming and the problem needs to be tackled sooner rather than later”. He predicts that this economic reality “will become so painful at some point that the political will to renegotiate these extravagant promises and diminish the nation’s contingent liabilities is likely to emerge” (Thakor, 2006; p. 257).
To sum up, the US government has already defaulted once (Zivney and Marcus, 1989), came close to defaulting again (Nippani et al., 2001), and had to increase the debt limit several times in the recent past so that the market now takes these increases for granted (Liu et al., 2009). In addition, there are questions being raised whether the country is already bankrupt (Kotlikoff, 2006) to which commentators now think that the US government might renegotiate its commitments (Thakor, 2006). In addition, we have information that suggests that LIBOR swaps are the most likely replacement for Treasury bonds as a financial benchmark (Haubrich, 2001). All of these suggest that not only is Treasury debt no longer default risk-free, but it is also not considered so by the market due to increased debt levels.
At the present time, there are several efforts being undertaken by both the legislative and executive branches of the US government to help the economy in the current financial crisis. As has been shown recently, there have been occasions when the legislature and the executive branch disagreed,
This situation has also been discussed both in the national and international press over the past couple of years. Moody’s forewarns that “The US is at risk of losing its top-notch triple-A credit rating within a decade unless it takes radical action to curb soaring healthcare and social security spending.... The warning over the future of the triple-A rating-granted to US government debt since it was first assessed in 1917 - reflects growing concerns over the country’s ability to retain its financial and economic supremacy” (Guerrera et al., 2008). In a more recent announcement,5 Moody’s confirms this gloomy outlook:
“The United States, which posted a record deficit in the last fiscal year, may lose its Aaa-rating if it does not reduce the gap to manageable levels in the next 3-4years, Moody’s Investors Service said on Thursday.
The US government posted a deficit of $1.417trillion in the year ended September 30 as the deep recession and a series of bank rescues cut a gaping hole in its public finances.
The White House has forecast deficits of more than $1trillion through fiscal 2011".
“The Aaa-rating of the US is not guaranteed”, said Steven Hess, Moody’s lead analyst for the United States said in an interview with Reuters Television. “So if they do not get the deficit down in the next 3-4years to a sustainable level, then the rating will be in jeopardy” (Desai and Nam, 2009).
Our motivation for this paper stems from the aforementioned reasons. The financial disturbances studied in the earlier works of Zivney and Marcus (1989) and Nippani et al., 2001 were smaller than the current financial crisis by comparison. If those two events did impact Treasury yields, it is highly likely that these events which are potentially more serious would impact them as well.
For examining the impact of the current economic crisis on Treasury default possibility, we examine the spread between 10-year swap minus the rate on a 10-year Treasuries constant maturity.6 Haubrich (2001) observes that the spreads between swap rates and Treasury bonds are becoming a closely watched indicator of the market’s view of macroeconomic risk. He also states that there is a significant body of opinion that is looking at swaps as the most likely replacement for Treasury as a financial benchmark, should budget surpluses continue to grow. He further mentions that they have already become the standard for pricing many corporate bonds. Based on his work, it appears that the spread between the swap rates, which is being looked at as a long-term substitute for Treasury yields, would be a good benchmark to compare Treasury yields. We, therefore, use the percentage change in the spread during the event window, SWLTRPCH, to examine the impact of the major events over the study period on Treasury yields.
Credit default swaps (CDS) on Treasury bonds would also be a measure of default risk. A CDS price is the annual cost of insuring a bond against default if the US failed to adhere to its debt agreements. The CDS prices on US Treasury bonds and the debt of other industrialized countries have increased greatly, according to various reports over time. For example, a basis point on a CDS contract protecting $10million of debt from default is equivalent to $1,000 per year. Shanahan and Moses (2008) mention that a 10-year CDS on Treasuries was .02% or two basis points in July 2007 before the credit crisis. Mandaro (2009) reports the price has increased consistently during the crisis reaching a new record of 97 basis points on March 10, 2009. Fidler and Shah (2009), on May 22, 2009, report that the five-year US Treasury CDS price had decreased to 37.5 basis points, even though the financial markets were worried about the default risk of Treasuries as related to S&P’s expected rise in debt to GDP from 44% in 2008 to 77% in 2013. These changes in CDS prices demonstrate that all levels of default risk, including Treasury securities, have gone through the repricing of credit risk we have observed in all of our debt markets over the past few years. The continuous repricing of credit risk is a major function of our capital markets. These CDS price changes, particularly the increases up to March 2009, are consistent with our measure, the declining SWLTR, in reflecting an increase in the default risk of Treasuries during the sample period. McDonald (2008) reports the quantity of these CDS have been given as relatively small with a $4billion notional gross and a little over $1billion net. Given the relatively small size of the CDS market, we chose to use the SWLTR measure since it is based on two widely accepted measures, 10-year Treasury yields and the 10-year LIBOR swap rate.
The purpose of this paper is to examine whether the yield of the Treasury securities has risen because of an increase in default risk. Many events that have occurred starting December 12, 2007 to March 18, 2009, have increased the commitment of public funds. All of these events have potentially increased long-term Treasury risk and thereby could increase Treasury yields. A list of the events is given in Table 1.
As mentioned earlier, we use the spread between 10-year swap rate and 10-year constant maturity Treasury bond rate to examine the impact of increased Treasury risk. We look for changes in the spread to measure the impact of increased Treasury risk. There have been studies in the literature that examined the potential short-term impact on Treasury default. Nippani et al. (2001) and Liu et al. (2009) use the spread between Treasury bills and commercial paper to examine the impact on Treasury default risk. Both studies look at short-term potential Treasury default. In our study, we are not looking at short-term imminent default but at long-term potential7 increases in default risk. Hence, we chose to use the 10-year swap - 10-year Treasury spread.
There are several points made by Haubrich (2001) that made the swap-treasury spread the ideal choice for this study. The first is that the market is already looking at this spread as a major indicator of macroeconomic risk. The second is that swaps are at least in the opinion of some the most likely replacement for Treasuries, which implies that they are possibly the future standards as far as low default risk rates are concerned. Finally, they already seem to be on their way to replacing Treasuries as a standard for pricing corporate bonds. Unlike the commercial paper-Treasury bill spread based on short-term rates, this paper examines the potential default in long-term rates. At the same time, it is also consistent with the two studies mentioned above, in that the use of the spread based on a 10-year maturity controls for term to maturity.
Fig. 1 shows the behavior of SWLTR over the sample period of January 3, 2006 to March 20, 2009. To assist the examination of SWLTR during the sample period the SWLTR at the beginning of the event on the trading day before the event, SWLTRBeg, and the SWLTR at the end of the event(s), SWLTREnd, and the changes during the event periods, SWLTRCH, are provided in Table 2.
Consistent with the graph of SWLTR the general trend is down during the event periods. The data illustrate some problems in specifying an empirical model. The mean and median for SWLTR are both 0.55, with a minimum of 0.04 and a maximum of 0.90 and a standard deviation of 0.14 for 803 observations. If we use SWLTR as the dependent variable the dummy for the event will reflect the difference between the SWLTR during the event and the mean value. For example, the average value of SWLTR during the two-day event period for Event 1 would be 71.5. The difference between 71.5 and 55 is 16.5, a positive coefficient; however, the change for the event is 0.19. In a regression not reported here for brevity that uses SWLTR as the dependent variable with no other explanatory variables other than the events, all the coefficients for Events 1 to 13 are positive and significant and all the coefficients for Events 14 to 29 are negative and most are significant. In other words, the pattern of these coefficients demonstrates that SWLTR as a dependent variable would just measure how different the level of SWLTR is from the mean of SWLTR on that date. The problem with this approach is that it does not allow us to examine the impact of individual events. In other words, did the specific event cause the SWLTR to change?
The change in SWLTR, SWLTRCH, is a better measure of the effect on SWLTR during the event period. However, it has a problem, too. For example, in Table 2 the SWLTRCHs for Events 13 and 28 are both 0.03 but these changes are from different levels. In Event 13, the 0.03 represents a decrease of 4.41% of the 0.68 at the beginning of the event window while in Event 28 the same 0.03 represents a decrease of 9.09% of the 0.33 at the beginning of the event window. To solve this problem the percentage change in SWLTR, SWLTRPCH, is used as the proxy for the relative change in the risk spread, SWLTR. This measure is similar to using the relative return in event studies for stock prices. If an event has an effect on SWLTRPCH, then the coefficient for that event should be statistically significantly different from zero.
Our measure known as SWLTRPCH
For testing the impact of increased risk of Treasury securities, we use a regression model with dummy variables representing the events and other non-dummy variables to control for other factors related to a changing spread. The equation used in this study is defined asIn the above equation the dependent variable, SWLTRPCH
The independent variables are the event dummies.9 For example, Event 1, the announcement of the Term Auction Facility took place on December 12, 2007. Thus Event 1 takes a value of 1 for the day of the announcement and the trading day immediately afterwards. On other days it takes a value of zero. The period covered for examining the macroeconomic impact is from January 1, 2006 to March 20, 2009, a period of over three years including 803 observations. The regression results are adjusted for autocorrelation and heteroscedasticity by using the Newey and West (1987) method. We need a long pre-event period since the series of event periods are extensive. Event 1 occurs on December 12, 2007, and Event 29, the last event occurs on March 18, 2009. In case an event occurs on a non-trading day, the first two trading days following the announcement are taken as event days. In the case of several events, the trading days overlapped. Since it would be difficult to distinguish the impact of overlapping events, we conflated them. As all of the events increased macroeconomic risk and were expected to increase Treasury risk, it is possible to do this. Thus, Events 3 and 4 are included as one dummy variable, Events 7 to 12, Events 17 to 19 and Events 24 to 25 are all included as one variable each since the days of the events tend to overlap. We hypothesize that all twenty-nine events will increase Treasury risk and therefore cause the SWLTRPCH
In addition there are four control variables included in the regression. The first three controls are similar to the variables used in the Liu et al. (2009) study. For SWLTRPCH
The second control variable is LNTCH, defined as the change in the natural log of the primary dealer transactions by security - US government securities coupon securities due in more than 6years but less than or equal to 11years outright transactions. This variable controls for the liquidity effects in the 10-year Treasury securities. The transaction data include all Treasuries in the more than 6years but less than or equal to 11years maturities, not just the 10-year Treasury securities.11 The data are reported in millions as daily averages for the week ended each Wednesday; therefore, we backfill the weekly average value for each business day of the preceding week. This is consistent with the study of Liu et al. (2009) where they use a similar variable based on short-term Treasuries. We expect higher Treasury transactions to indicate more liquidity, depressing Treasury yields and increasing SWLTR; therefore, we expect the coefficient on LNTCH to be positive.
The third control variable is SWSPRCH, the change in the spread, (ask-bid), for the 10-year LIBOR swap at the end of the trading day.12 This spread variable is a proxy for the liquidity of the 10-year swap. As SWSPR decreases it indicates that the liquidity is increasing. As the liquidity increases we expect the swap rate to decrease. The lower swap rate would lead to a smaller SWLTR; therefore, we expect SWSPRCH to have a positive coefficient.
The fourth control variable is BLSCH, the change in the foreign purchases or buys of US Treasury securities minus the foreign sales.13 The major impact by foreign buyers and sellers should be reflected in the difference in purchases and sales; therefore, we introduce a variable based on the amount, Buys Less Sales, or BLS. We use the change in BLS, or BLSCH.14 This information is provided on a monthly basis; therefore, we backfill the monthly average value for each business day of the appropriate month. If BLSCH is positive we expect the greater net foreign demand for Treasuries to increase the price and lower the yield on Treasuries. Assuming no effect on the swap rate, the lower Treasury yield would lead to a greater SWLTR; therefore, we expect BLSCH to have a positive coefficient.
Descriptive statistics for the dependent variable, SWLTRPCH, the components used to calculate SWLTRPCH, and the control independent variables are provided in Table 3.
To summarize the model, we hypothesize the event dummy variables will have negative coefficients, SWSPRCH
The results are presented in Table 4. The expected coefficients on all the event independent variables are hypothesized to be negative. Thirteen of the 20 tested events are negative and significant. Event 1 (Term Auction Facility), Event 2 (Term Securities Lending Facility), Event 3 to 4 (Primary Dealer Credit Facility and JP Morgan purchase of Bear Stearns), Event 13 (Troubled Asset Relief Program is enacted), Event 14 (six central banks announce rate cuts and FED agrees to provide more funds for AIG), Event 17 to 19 (Citigroup is in midst of crisis and Ford, General Motors, and Chrysler ask Congress for a bailout), Event 21 (the Federal Reserve slashes rates to near zero), Event 22 (Treasury unveils aid for GMAC), Event 24 to 25 (Citigroup and Bank of America crises continue and remaining $350billion of TARP funds are released), Event 27 (banking shares hit lowest level of stock prices since 1992), and Event 28 (AIG reveals $61.7billion quarterly loss) are negative and significant at the .01 level. Event 5 (IndyMac files for bankruptcy) and Event 15 (US expects to commit to recapitalization of largest bank holding companies and announces details of Commercial Paper Funding Facility) are negative and significant at the 0.10 level. Event 6 (Conservatorships of Fannie Mae and Freddie Mac are announced) is negative but not significant. Six events are positive and significant at the 0.01 level. Event 7 to 12 (Lehman Brothers files for bankruptcy, Merrill Lynch agrees to be taken over by Bank of America, Federal Reserve announces loan of up to $85billion to AIG, Federal Reserve authorizes to extend credit to Goldman Sachs and Morgan Stanley, and WaMu closes and is sold to JP Morgan), Event 16 (economy shrinking at an annual rate of 0.3%, a cut in the Federal Funds target, and AIG decreasing the loan amount), Event 20 (US government agrees to bailout of Citigroup), Event 23 (bad economic statistical results are released), Event 26 (US House of Representatives passes economic stimulus package), and Event 29 (Federal Reserve announces quantitative easing program) are positive and the market appears to have viewed these events as positive in resolving uncertainty at the time of the event when viewed as part of the continuing crisis.
If we refer back to Table 2 and examine the event that had the largest absolute negative impact on SWLTR, it is Event 1 (the Federal Reserve announced the Term Auction Facility) with an impact of 0.19%. This was the first step in developing new monetary policy tools because the Federal Reserve’s traditional tools, particularly the discount window, were not working. Event 14 (six central banks announce rate cuts and FED agrees to provide more funds for AIG) is close with the second highest impact of 0.18%. The coordination of the six central banks illustrated the recognition that there was a global problem, not just a US problem. In addition, the AIG situation appeared to be worsening and one has to remember all the banking counterparties on the other side of the credit default swaps that AIG wrote. Event 24 to 25 (the Citigroup and Bank of America crises and the release of the remaining $350billion in the TARP) is the third highest negative impact with 0.16%. Event 21, Event 3 to 4, and Event 17 to 19 are negative impacts of greater than 0.10%. Two events, Event 7 to 12 and Event 16, are positive absolute impacts on SWLTR of 0.13%.
Only one of the control variables, SPRET, is statistically significant at the .05 level and it has the hypothesized negative coefficient. As SPRET decreases, the default risk of the swap should increase and the swap rate should increase; therefore, SWLTRPCH should increase. As SWSPRCH decreases, the swap becomes more liquid and the swap rate should decrease; therefore, SWLTRPCH should decrease. The sign for SWSPRCH is positive as hypothesized; however, it is not significant. As LNTCH increases, the Treasury security becomes more liquid and the Treasury rate should decrease; therefore, SWLTRPCH should increase. The sign is negative instead of the hypothesized positive but it is not significant. The lack of significance of these two liquidity variables may not be surprising because the Treasury and swap markets would generally be considered very liquid markets. The BLSCH is positive as hypothesized but it is not significant. The monthly backfilling may have affected the usefulness of this variable.
We discuss the importance of our results and their implications in the next section.
This paper analyzes the increasing default risk related of long-term US Treasury securities as major events have caused investors to question the credit worthiness of these securities. We use the spread between the 10-year USD LIBOR swap and the 10-year US Treasury security as a measure of the default risk. The 10-year swap is regarded as the other benchmark by the
The results demonstrate that 13 of the tested 20 events have a negative and statistically significant impact on SWLTR and one event is negative but not significant. Six of the events have an absolute negative impact on SWLTR of 0.12% to 0.19%. Six events have a positive impact on SWLTR and two events have an absolute positive impact on SWLTR of 0.13%. Controlling for the liquidity risks, the swap default risk, and the net foreign purchases of Treasuries, the lower spread is consistent with a greater default risk for US Treasury securities.
This study adds to the existing evidence in many ways. Consistent with earlier studies by Zivney and Marcus (1989), Nippani et al. (2001), and Liu et al. (2009) we show that there is enough evidence to suggest that Treasury yields are not viewed by the market as “default risk-free” especially during a financial crisis. The study also supports the use of the spread between the 10-year Treasury rate and the 10-year LIBOR swap rate as a measure of the default risk with the recognition that other factors/variables may affect that spread. Our study is also the first to show that the market charges a default risk premium on long-term Treasury securities in severe economic conditions. This lends support to the contention of Haubrich (2001) that swaps are more likely to replace or substitute Treasuries as a financial benchmark. This is also supported by the announcements of Moody’s credit ratings agency over the past two years that the AAA ratings for US government debt could be reassessed in 10years, if the situation does not improve. We also provide evidence to support the discussion by Kotlikoff (2006) that the level of debt may be too high and steps need to be taken to control increasing levels of government debt.
As of June 18, 2009, the total public-debt outstanding was just under 11.4 trillion dollars.15 According to Liu et al.16: “The US Treasury debt exceeded $10trillion on September 30, 2008, so each additional basis point in the risk premium translates into $1billion additional interest expense to the Treasury” (2009, p. 1465). There is also some evidence of the increasing debt levels are of public concern and are often discussed in the popular press. For example, in an article in the
The solution though, is no mystery. It will involve some combination of tax increases and spending cuts. And it would not be limited to pay-as-you-go rules, tax increases on somebody else, or a crackdown on waste, fraud and abuse. Your taxes will probably go up, and some government programs you favor will become less generous.
This is a legacy of our trillion-dollar deficits. Erasing them will be one of the great political issues of the coming decade.
This analysis is consistent with the views of Thakor (2006) who talks about renegotiating these extravagant promises and diminishing the nation’s contingent liabilities. Our paper provides empirical evidence to these views by showing that the extra burden which the government is undertaking to combat the current financial crisis, is coming at an increased cost to the taxpayers. While prior studies have shown evidence of a default risk premium being embedded in short-term Treasury securities, we show that even longer-term Treasury debt is now vulnerable to this situation. Our results add to the view expressed by Liu et al. (2009) that increased servicing costs could prove to be terrible strain for the Treasury in the long run. Our evidence that even long-term Treasury debt is susceptible to default risk is an emphatic statement by the market that it now considers Treasury debt of all maturities to be potentially risky.
We thank an anonymous referee, Yoon Choi, Richard Hofler, Vladimir Kotomin, Asli Ogunc, Drew Winters, and William Weaver for helpful comments, and Praveen K. Murugesan and Chibunna Obuzor for research assistance. Any remaining errors, if any, are solely ours.
1
Recent examples of studies that use yield spreads and changes in yields are Van Landschoot (2008), Chen (2009), and Kotomin et al. (2008).
2
For example,
3
See Claessens et al. (2009) for a discussion of a special issue of Journal of Banking and Finance on the global financial crisis and risk.
4
These rates are found in the Federal Reserve Statistical Release H.15 Selected Interest Rates.
5
See “Reducing Deficit Key to US Rating: Moody’s” published by Reuters on October 22, 2009. The above cite is from the website: .
6
Although thirty-year swaps are available during the sample period, thirty-year Treasury rate information was not. Although there are twenty-year treasury rates available, there are no twenty-year swaps.
7
There is definite evidence that the market is also focusing on potential long term rather than short-term default. In the Guerrera et al. (2008) article it is reported that Steven Hess, Moody’s lead analyst for the US, told the Financial Times that in order to protect the country’s top rating, future administrations will have to rein in healthcare and social security costs. “If no policy changes are made, in 10years from now we would have to look very seriously at whether the US is still a triple-A-credit” he said.
8
The daily rates were obtained from the website: .
9
See Cornett and Tehranian (1990) for a similar approach.
10
We wish to thank an anonymous referee for also suggesting this factor. The daily return on the KBW Index, a capitalization-weighted index composed of 24 geographically diverse stocks representing national money center banks and leading regional institutions, was also tried as a measure of default risk in the USD LIBOR market. It had a 0.80 correlation coefficient with SPRET
11
The historical primary dealer data can be found at: .
12
This information was provided by Xignite: .
13
See .
14
We wish to thank an anonymous referee for suggesting this variable.
15
The exact figure was $11,399,258,796,766.10 as per the website: .
16
See paragraph 1, p. 1465 of Liu et al (2009) “Did the repeated debt ceiling controversies embed default risk in US Treasury Securities?” Journal of Banking and Finance, 33, 1464-1471.
