Recent research has reported deciencies of structural models employed for quantify- ing credit risk. Eom et al. (2004) use ve structural models for bond pricing and conclude that these tend to underestimate spreads of safe bonds while overstating creditspreads for bond issues of rms with high asset volatility and leverage. Bharath and Shumway (2008) construct a naïve bankruptcy predictor as an alternative to the classical Mer- ton distance to default (DD) model which outperforms the original. They reason that if the predictive power of our naïve probability is comparable to that of [the original model], then presumably a more carefully constructed probability that captures the same information should have superior power. Campbell et al. (2008) construct the current state-of-the-art statistical model for bankruptcy prediction using simple market and ac- counting variables. They demonstrate a substantial underperformance of Merton's DD model relative to theirs in terms of Pseudo-R 2 s and conclude that summarizing default
risk premia, and replicate the set of macro fundamentals of Goyal and Welch (2008), who investigate the predictability of equity risk premia.
We find that our volatility factor is strongly linked to systematic forces driving both bond and equity risk premia. Across different ratings and graphs, the volatil- ity factor is closely related to the eight factors of Ludvigson and Ng (2009), with an average adjusted R-squared value of 30% (Table 1). The relation is even stronger when we use Goyal and Welch’s (2008) factors, where the average adjusted R-squared value is more than 60%. This result is meaningful because it validates results in the prior literature linking the equity premium to creditspreads (Jagannathan and Wang (1996), Chen, Collin-Dufresne, and Goldstein (2009)). It also suggests that the volatil- ity, rather than the level, of creditspreads may be the main channel through which these two assets’ risk premiums are linked. We report detailed results in Appendix J.
Mots clés : risque de crédit; écarts de crédit; modèles à forme réduite; volatilité
The paper investigates a two -factor affine model for the creditspreads on corporate bonds. The first factor can be interpreted as the level of the spread, and the second factor is the volatility of the spread. The riskless interest rate is modeled using a standard two-factor affine model, thus leading to a four- factor model for corporate yields. This approach allows us to model the volatility of corporate creditspreads as stochastic, and also allows us to capture higher moments of creditspreads. We use an extended Kalman filter approach to estimate our model on corporate bond prices for 108 firms. The model is found to be successful at fitting actual corporate bond creditspreads, resulting in a significantly lower root mean square error than a standard alternative model in both in-sample and out-of-sample analyses. In addition, key properties of actual creditspreads are better captured by the model.
Table (7) and Table (8) present empirical results using ex-post and real-time measures of busi- ness conditions, respectively. The results are quite similar whether we use ex-post or real-time measures of business conditions and can be summarized as follows: As expected, creditspreads generally widen during periods of economic slowdown. More importantly, unexpected changes in the Fed funds rate has a positive and significant effect on creditspreads during periods of economic slowdown although the effect is insignificant when we do not distinguish between different economic conditions. The effect is strongest for the credit spread between BAA and AAA rated bonds and decreases almost monotonically for the two other creditspreads. Further- more, the results are not as significant for the empirical specification with real-time recession probabilities from Chauvet and Piger (2008). This might be due to the fact that the relation between creditspreads and industrial production is stronger than the relation between creditspreads and other additional variables used by Chauvet and Piger (2008). These results are not only significant statistically but also economically. Consider the empirical specification with the NBER recession dummy variable for which the economic interpretation of the coefficient estimates is the easiest. A 10 basis point (approximately one standard deviation) unexpected increase in the Fed funds rate results in an additional increase of 2.8 basis points in the credit spread between BAA and AAA rated bonds during NBER recessions. These results suggest that firms with low credit ratings are more sensitive to unexpected changes in monetary policy during recessions in line with predictions of imperfect capital markets.
As with other forms of debt, the securities issued by an SPV have different levels of subordination: the equity tranche of a CDO will bear the first capital loss, while the mezzanine tranches and senior (or even super-senior) tranches are protected by their higher ranking in terms of subordination. Investors in these CDO tranches receive a higher return for the same rating given by rating agencies, mainly because they have a leveraged credit position. To demonstrate this, we will take the theoretical example of an asset portfolio consisting of 100 equally weighted names, with no recovery rate. There is no leverage in that portfolio: if ten names default, the portfolio maintains a value of nearly 90%. On the contrary, a tranche with a subordination level of less than 10% will be harmed, and possibly left with no value. Another way to measure leverage is to consider the mark-to-market change in the CDO tranche created by a parallel change in the combined creditspreads of the portfolio’s bonds. Many other structures, such as Constant Proportion Portfolio Insurance (CPPI) or Constant Proportion Debt Obligation (CPDO), exhibit similar leveraged positions that result from other mechanisms.
from the existing credit risk literature on how much of the observed corporate spreads over Treasury yields can be explained by default risk.
To address this puzzle, many parallel and subsequent studies investigate the ability of non default risk factors (such as market, liquidity and …rm-speci…c factors) to explain credit spread di¤erentials. These studies include those of Collin-Dufresne et al. (2001), Driessen (2003), Campbell and Taksler (2003), Huang and Kong (2003), Longsta¤ et al. (2005), and Han and Zhou (2006) among others. However, even after accounting for non default factors the puzzle remains unsolved because a large proportion of creditspreads remains unexplained. In particular, Collin-Dufresne et al. (2001) perform a regression that includes all potential explanatory variables predicted by theoretical models but fail to explain more than 25% of credit spread changes. They state that "variables that should in theory determine credit spread changes in fact have limited explanatory power". Collin-Dufresne et al. (2001) have also detected a common systematic factor that potentially could explain the large part of the unexplained changes. However, several macroeconomic and …nancial candidates fail to measure it. It appears, then, that their model is missing an important component which may not be captured by macroeconomic fundamentals. This paper focuses on the drivers of the missing component in credit spread determinants. Thus, it extends the Collin-Dufresne et al. (2001) model by allowing for a regime switching structure in the credit spread dynamics. 1 See also Delianedis and Geske (2001) and Amato and Remolona (2003) who reach the same results using
Mots clés : Spread de crédit en euro, modèle EGARCH, distribution de loi conditionnelle, effets d’asymétrie, prévision de volatilité.
Volatility forecasting is one of the major current issues in contemporary finance. Indeed, the accurate estimation of volatility, unobservable parameter in the market, is crucial for asset allocation decision making and risk management. This study proposes a modelling approach of conditional volatility of creditspreads in a non-Gaussian setting that accounts for the stylized facts observed in the markets. Asymmetry is controlled for in two ways: distinction between the effects of shocks on the variance according to their sign (through an EGARCH process) and the use of skewed distributions to characterize innovations. We find that, adding asymmetry effects to the conditional Student’s distribution and Generalized Error Distribution (GED) improves the quality of our estimations. The EGARCH model with a skewed GED distribution, which controls for both asymmetry and the presence of fat tails, seems to do a better job of characterizing the dynamic of creditspreads and offers better out-of-sample forecast of volatility.
of the increase of investors’ riskiness. On the other hand, the uncertainty can be solely considered as cause of the widening of creditspreads and hence, the investors likelihood to default. Only taking into consideration the frictions in the credit market and in the financial market allows to model these issues and investigate the impact of such shocks. Therefore, the representative agent paradigm does not hold anymore. In a frictionless economy, the wealth distribution does not matter and risk perception does not affect the equilibrium outcome. However, when there are financial frictions, the individual characteristics matter and micro-behavior determines final allocations. In this thesis, I tried to approach the subject using different tools and techniques. The first chapter was dedicated to the credit channel of monetary policy and the importance of credit market imperfections to identify different sub-channels. Using a tractable contractual device in the credit market assuming asymmetric information among contractors, I show that the balance-sheet sub-channel is the largest in terms of monetary policy transmission capacity. This finding was confirmed by an empirical evidence based on the outcome of the Senior Loan Officer Opinion Survey. In the second chapter, I tried to make the link with the labor market. The persistence of unemployment after the financial crisis raised many questions regarding the contribution of the adverse financial shock to this high persistent level. I study the impact of uncertainty and risk shocks on the unemployment via the job creation and the job destruction dynamics. The third chapter is dedicated to the impact of asymmetric information on the job creation, mainly. Vacancies were supposed to be financed by external funds which is, by definition, more expensive than the internal funding. An adverse shock that increases the risk premium in the credit market or deteriorates the entrepreneurs balance-sheet reduces significantly levered investors risk-bearing capacity. Hence, their borrowing capacity is reduced and consequently, job creation decreases.
3004 Credit Traps †
By Efraim Benmelech and Nittai K. Bergman*
Can the Federal Reserve stimulate lending when there are disruptions in the financial system? According to the credit channel literature, expansionary mone- tary policy alleviates financial frictions and increases the availability of credit (e.g., Bernanke and Gertler 1989, 1990, 1995 ). 1 Indeed, the severity of the recent finan- cial crisis has led central banks around the world to adopt traditional as well as unconventional policy measures to combat the crisis and boost lending (Gertler and Kiyotaki 2010 ). In the United States, the Federal Reserve experimented with new policies of quantitative and credit easing by lending directly to financial institu- tions, providing liquidity to key credit markets, and purchasing long term securities (Bernanke 2009). Other major central banks such as the European Central Bank and the Bank of England followed suit with similar “quantitative easing” policies. While there is some evidence suggesting that in the United States these polices have been effective in reducing creditspreads, lending by US banks did not return to its pre-crisis levels.
In a …rst step we estimate a term structure of empirical survival probabilities via credit rating transition matrices. These rating transitions are estimated from the generator of the Markov chain underlying the rating migration, as in Lando and Skodeberg (2002) and Christensen, Hansen and Lando (2004). These studies suggest estimating a Markov-process generator rather than a one-year transition matrix. Lando and Skodeberg (2002) have shown that this continuous-time analysis of rating transitions using generator matrices improves the estimates of rare transitions even when they are not observed in the data, a result that cannot be obtained with the discrete-time analysis of Carty and Fons (1993) and Carty (1997). A continuous-time analysis of defaults permits estimates of default probabilities even for cells that have no defaults. The rating transition histories used to estimate the generator are taken from the January, 09, 2002 version of Moody’s Corporate Bond Default Database. A precise description of the data used to obtain the transition estimates is given in Appendix E. Using default data from 1987 to 1996, a generator matrix G is estimated. The estimated generator matrix is presented in Table 6. With this generator, the transition matrix for a horizon of t periods and the corresponding default probability can be obtained by computing
Abstract: Although human-driven landscape modification is generally characterized by habitat destruction and
fragmentation, it may also result in the creation of new habitat patches, providing conditions conducive to spontaneous colonization. In this article, we propose the concept of "colonization credit" (i.e., the number of species yet to colonize a patch, following landscape changes) as a framework to evaluate the success of colonization, in terms of species richness, in new/restored habitats, taking into account the spatial structure of landscapes. The method mirrors similar approaches used to estimate extinction debt in the context of habitat fragmentation, that is, comparisons, between old and new habitat patches, of the relationships among spatial patch metrics and patch species richness. We applied our method to the case of spontaneous colonization of newly created habitat patches suitable for wet heathland plant communities in South Belgium. Colonization credit was estimated for the total species richness, the specialist species richness, and the species richness of three emergent groups (EGs) of specialist species, delineated on the basis of dispersal traits. No significant colonization credit was identified either in patches created 25-55 years ago or in those created within the past 25 years, with the exception of species from our first EG (mostly anemochorous species with long-term persistent seed bank). However, the differential response of species in that first EG could not be explained through their characteristic life history traits. The results of this study are encouraging and suggest that deliberate, directed restoration activities could yield positive developments in a relatively short period of time.
An important research area of the corporate yield spread literature seeks to measure the proportion of the spread that can be explained by factors such as the possibility of default, liquidity, tax differentials and market risk. We contribute to this literature by assessing the ability of observed macroeconomic factors and the possibility of changes in regime to explain the proportion of yield spreads caused by the risk of default in the context of a reduced form model. For this purpose, we extend the Markov Switching risk- free term structure model of Bansal and Zhou (2002) to the corporate bond setting and develop recursive formulas for default probabilities, risk-free and risky zero-coupon bond yields as well as credit default swap premia. The model is calibrated with consumption, inflation, risk-free yields and default data for Aa, A and Baa bonds from the 1987-2008 period. We find that our macroeconomic factors are linked with two out of three sharp increases in the spreads during this sample period, indicating that the variations can be related to macroeconomic undiversifiable risk. The estimated default spreads can explain almost half of the 10 years to maturity industrial Baa zero-coupon yields in some regime. Much smaller proportions are found for Aa and A bonds with numbers around 10%. The proportions of default estimated with credit default swaps are higher, in many cases doubling those found with corporate yield spreads.
and the variation of oil prices (Stock and Watson, (2003)). The NBER dating committee reported that movements in payroll employment were important in choosing March 2001 for the beginning of the recession and for the observation that the economy was in recession. When we analyze the Chicago Fed National Activity Index (CFNA), we see that within the full indicator index during this recession, there was signiﬁcant degrees of heterogeneity among the category indexes. For example, Employment, unemployment and labor hours, and the Production and income indexes fell as the overall index while the consumption category hardly registered any negative values. In fact, consumer spending continued to experience positive growth during much of the recession: “Simply using the consumer category as a proxy for the CFNA index would clearly result in diﬀerent inferences” (Evans et al., 2002, p. 32). It seems that some individual indicators provided false signals on the state of the aggregate economy while other individual indicators did pretty well. Another particularity of the 2001 recession is related to ﬁnancial indicators. The term spread (the ten-year Treasury bond rate minus the Fed funds rate) on government debt and stock returns provided advance warning on the 2001 recession but fall short of providing a signal of previous recessions (Stock and Watson, (2003)). These modiﬁcations can be attributed to changes in in- dustrial economies since the 1990s, including the development of the ﬁnancial markets. The U.S. recession of 2007-2009 period also reﬂects an important change in U.S. economy. This time, it was preceded by an important ﬁnancial crisis. But it is the household leverage growth and the dependence on credit card borrowing that drove the recession (Mian and Sapi, (2009)). Durable consumption was again among the serious signs of weakness in the economy and dramatic increase in household leverage from 2000 to 2007 was the primary driver of the last recession.
were committed to provide as sponsor to some Securitisation structures 9 . However,
if financial account (in their new presentation) are used to compile the total credit to non financial private sector, securitisation is not an issue credit from all sectors is taken into account, including all kind of conduits to which banks sell loans portfolios. • If financial accounts are not available for a given country or period, the cross-border component of credit to private sector has to be estimated using the BIS International Banking Statistics (IBS). Several caveats need to be made regarding the IBS regarding cross-border lending. First, while these series take into account cross-border lending by foreign banks, they don’t include loans provided by non-bank entities and in addition they are not always available on the same period than the domestic bank credit.
As in the EU case considered in (8), the second term is negative, as long as lenders are not overly optimistic (as spelled out in Proposition 2, so that ∂Y C (iD,ρ)
∂ρ < 0) so that Y C (iD) < Y C (iD, ρ ∗∗ ). 1 Credit rationing may therefore obtain. On the other hand, if borrowers are suﬃciently pessimistic so that dρ di ∗∗ 6 0:
It follows that Assumptions (A1)(i), (A1)(iii), (A2), (A3), and (A4)(ii) hold. However, since the commodity is perishable and asset j is segmented, the super-replication property (A4)(i) is not satisfied. Also, (A1)(ii) does not hold, because w B = (1, 0).
The failure of these assumptions implies that the economy does not have equilibria. Indeed, by monotonicity of preferences the asset price q is strictly positive. Hence, Inada’s condition and bud- get feasibility imply that the optimal portfolio of agent B belongs to (0, 1/q]. This is incompatible with credit segmentation, because only B can short-sale j. 10 2
But there also exist interactions between credit market regulation and labour market institutions. The literature on this issue is particularly interesting because it shows that the links between credit poli- cies and unemployment are more complex than exposed in the third section. However, it is limited to a few theoretical papers. A first category of stud- ies considers financial deregulation and labour mar- ket flexibilization as substitutes. In Rendon (2001), reducing firing and hiring costs boosts employment. Access to external finance curbs unemployment since it allows firms to finance labour adjustment costs. Therefore, if credit is easily available, removal of labour market adjustment costs becomes less effective since these costs can easily be financed by external finance. Symmetrically, if the labour mar- ket is made perfectly flexible, access to external finance has a weak impact on employment. In Belke and Fehn (2002), strong labour protection allows workers to partly capture the rent resulting from the entrepreneur’s project. This decreases the project’s rate of return below the minimum threshold required by fund providers. Hence, the firm cannot