Stress Testing Corporate Credit Risk - Economy.com

economic & COnsumer credit Analy tics

October, 2007

Moody’s analytics

Stress Testing Corporate Credit Risk

Prepared by Tony Hughes Senior Director, Credit Analytics +610.235.5000

Juan Licari Senior Director, Credit Analytics +610.235.5000

Stress Testing Corporate Credit Risk BY Tony Hughes

U

ntil recently, burgeoning liquidity was a feature of world financial markets. The rise of pension funds, hedge funds and petrodollars and increasingly aggressive foreign forays by Asian central banks meant that the globe was awash with cheap money. The inexorable search for yield meant that cash was flowing in all directions and risk, defined broadly, became incredibly cheap. The subprime mortgage mess was just one consequence of this liquidity bubble. Other excesses are almost certainly out there, cloaked by the depth and magnitude of the subprime sector’s considerable current woes and the threat this poses to the economic expansion. What set the recent liquidity bubble apart from past bubbles was that it was so widespread and affected so many distinct asset classes simultaneously. Chart 1 shows the spread on A-rated corporate debt over and above the daily observed federal funds rate; it demonstrates clearly that at the same time that billions of dollars of investor funds were pouring into securitizing subprime mortgages, the spread on investor grade corporate debt was also shrinking to record low levels, around a third of a percentage point. Following the recent financial turmoil, such spreads have widened to around 140 basis points though this is still rather low by historical standards. In large part, these low spreads can be attributed to generally sound credit performance in the corporate sector. As we have often

noted, despite the unease in financial markets, corporate balance sheets are in generally sound order and corporate failures—outside the mortgage industry—are few and far between. One consistent measure of corporate credit riskiness is the median expected default frequency (EDF) for the top 500 listed firms in the U.S., calculated and published by Moody’s KMV. This series is depicted in Chart 2. Note that the overall EDF of large listed firms tends to rise during periods of recession (the early 1980s for instance) or stock market regression (like 1987). Periods where recession coincides with significant stock market decline, like in the early part of the decade, show up as more pronounced and sustained rises in average EDF. Further, for the past 35 years, overall corporate credit risk demonstrates a sustained downward trend and

currently resides close to an all-time recordlow level given the history of the series. The circumstances described—low spreads married to low EDFs—carry a number of important potential consequences. First and foremost, there may exist a bubble in the corporate sector and it may eventually blow up in already-strained investors’ faces. This observation may seem to fly in the face of the fact that corporate credit quality is currently in excellent shape but, it must be remembered, actual delinquencies and defaults tend to lag behind the occurrence of poor origination standards. That corporate defaults are currently low may hide the fact that standards slipped in 2006; if these defaults come to fruition, it is possible that credit quality in the corporate world could crash just like the

Chart 1: A Bubble in Corporate Credit?

Chart 2: Corporate Credit Risk, Median EDF

Spread, A-rated corporate bonds and federal funds, 1997-2007

U.S. largest 500 firms, 1970-2007

7

1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

6 5 4 3 2 1 0 -1 13-Jan-97

11/1/99

8/19/02

6/6/05

1974Q3 1982Q2 1987Q4 1990Q3

2000Q2

70

Sources: Redbook, Federal Reserve Board

Source: Moody's KMV 1

MOODY’S ANALYTICS / Copyright© 2007

2

1

ANALYSIS �� Stress Testing Corporate Credit Risk

subprime mortgages that were originated at the same time and which have already shown significant signs of weakness. Second, the subprime mortgage problems have already sparked a credit crunch, which will have severe consequences should corporate credit actually crumble. Companies facing short-term cash flow problems—even those with sound longterm prospects—may, in the present environment, fall into illiquidity and insolvency if funding cannot be sourced through the usual channels. This implies that any stumble in the corporate world would be magnified by the tight credit conditions facing everyone, even those with sound current credit histories. Third, the subprime mess brings with it the very real possibility of a recession in the broader global economy, meaning that corporate profitability may be in jeopardy, particularly for those firms reliant on the humble U.S. consumer for their ongoing sustenance. All of these factors suggest that, though conditions are currently benign, risks to corporate credit are manifest. Active participants should be cognizant of potential economic problems on the horizon and be ever prepared to act. This article describes a methodology that can be used to evaluate corporate credit risk under a variety of future economic scenarios and thus addresses this issue directly. This method brings together cutting-edge credit risk assessment tools developed by Moody’s KMV with state-of-the-art econometric techniques implemented by Moody’s Analytics. This integrated approach delivers a robust and consistent framework that is shown to be very useful as a stress testing device. Moreover, the proposed technique is flexible enough that it can be used to quantify corporate credit risk exposure of a particular portfolio under a given down or upside scenario for the macroeconomy. The results of this article are twofold: First, we identify the economic drivers that have shaped past corporate credit outcomes (as measured by EDF). Because our aim is stress testing, we develop a novel approach to identify the impact of shifts in the economic landscape when stress is applied as opposed MOODY’S ANALYTICS / Copyright© 2007

to merely measuring the size of likely baseline responses. We then quantify the effects of alternative macroeconomic scenarios, drawn from our structural models, on the future performance of corporate credit risk. Regarding the linkage between macroeconomic factors and credit risk, several indicators are found to play a key role in explaining changes in corporate default probability. In short, good economic prospects are associated with lower default, but the marginal effects of some alternative macro series vary with the position of the economy in the business cycle. Once the relevant drivers are identified, the estimated econometric models are used to quantify the effects of two alternative economic scenarios on corporate credit risk. An optimistic outlook—typified by the U.S. economy sailing through the choppy waters of the current environment and avoiding recession—enforces the current downward trend of default probabilities. In other words, should the economy avoid recession, the credit quality erosion taking place today on the retail side of the credit market would not transmit into a substantive hike in corporate defaults. The fact that a recessionary outcome is avoided gives companies enough room to maneuver away from debt payment interruptions in the wake of the credit crunch. On the other hand, once a recession scenario is brought to the table, the corporate risk picture changes significantly. Weak performance of the labor and output markets would have a measurable effect on asset markets and this will, in turn, filter into the financial soundness of many or most corporations. That is, if the real side of the U.S. economy was to face critical pressure, corporate credit markets would not be immune to a contagion effect and would be expected to suffer measurable deterioration in their underlying credit quality. Drivers of corporate credit risk. The Moody’s KMV Expected Default Frequency is used as a measure of default risk or probability of default. This EDF is a forward-looking construct that takes account of financial market performance and accounting relations to predict the probability that a firm will default within

www.economy.com

a given time horizon by failing to make an interest or principal payment. The measure reflects the true or physical probability of default and is calibrated to historical defaults of publicly-traded companies. As an example, a firm with an EDF of 10% has an estimated one-in-ten chance of defaulting over the next year. The underlying principal in deriving an EDF measure is that a firm defaults when the value of its business or the market value of its assets falls below what it owes; this is known as the “default point.” In other words, as long as the value of the business is greater than the amount of obligations due, equity holders have the ability and the incentive to pay, in full, its current and future debt obligations. If the value falls to zero or below, the incentive is nonexistent and the likelihood of default rises accordingly. The required inputs for the model include a firm’s equity value as well as its financial statements. The following are the main components of an EDF credit measure: »» Market Value of Assets: The market’s view of the enterprise value of the firm as determined by the firm’s equity value, equity volatility, and liability structure. Moody’s KMV employs a proprietary option theoretic model to compute the market value of assets since it is not directly observable. »» Default Point or the level of the firm’s obligation: The default point represents the amount of liabilities that are expected to be due in the event that the firm is in distress. In general, the default point is close to a value equal to short-term liabilities plus half of long-term liabilities. »» Asset Volatility: This represents a measure of the business risk of any given firm. Technically, it is the standard deviation of the annual change in the market value of the assets expressed in percentage terms. The model computes a measure of “distance-to-default” that is subsequently mapped to an EDF level. Distance-todefault is the number of standard deviations that the market value of assets is above the 2


Chart 3: EDF Changes and Asset Returns

U.S. largest 500 firms, 1970-2006 Year-to-year change in median EDF

Year-to-year change in median EDF

www.economy.com

Chart 4: EDF Changes &Changes in Asset Volatility

U.S. largest 500 firms, 1970-2006, tim 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.2

y = -1.1418x + 0.1276

-0.1 0.0 0.1 0.2 Median year-to-year asset return

0.3

0.4

Source: Moody's KMV

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -0.2 -0.3 -0.4 -0.5 -0.2

-0.1 0.0 0.1 Median year-to-year change in asset volatility

0.2

Source: Moody's KMV 3

1 The empirical mapping between “distance-to-default” and EDF measure improves the realism of the structural model by relaxing the assumption that assets follow a Brownian motion and the assumption of a fixed default point.

Modeling EDF drivers under stress. Beasset returns are surprisingly poor given funcause we aim to measure the behavior of damentals or if asset volatility is surprisingly EDF under non-standard, stressed condihigh, we can imply that, over and above the tions, we employ a non-standard, semistress imparted by the macroeconomic drivparametric modeling approach to the issue ers, an additional source of unattributable at hand. A standard regression framework stress is also present. Only by combining would estimate the impact of economic these direct and indirect stresses can the full variables on EDF under a baseline or hisimpact of a stressed environment be cortorically average amount of felt stress; this rectly captured. approach would be entirely appropriate for The technique we use to model the forecasting the future performance of corpoperformance of the factors under stress is rate EDFs given our baseline macro forecast. known as quantile regression, which is well When undue stress is applied, however, known in the econometrics literature. One’s these baseline responses might not, in realfirst thought might be that this technique ity, be entirely appropriate. For example, involves segmenting the data into quartiles the marginal impact of an extra semi-trailer or deciles and then estimating separate reon the probability of a bridge collapsing may gressions for each, but this characterization be significantly higher during a blizzard than is not accurate. All the data are used to esit is in calm, warm conditions. The analogy timate the quantile regression, it is just that extends to corporate credit perfectly. One data in the relevant part of the distribution way of identifying stressed circumstances are given more weight in the calculation of is by considering tail locations in the conthe regression coefficients that observations ditional distribuChart 5: EDF Changes and Changes in Default Point tion. If the variU.S. largest 500 firms, 1970-2006 able of interest 0.8 is performing y = 2.4684x - 0.2547 0.7 “surprisingly 0.6 0.5 badly” given the 0.4 observed inde0.3 0.2 pendent variables, 0.1 0.0 we can conclude -0.1 that the unob-0.2 -0.3 served factors are -0.4 detracting from -0.5 0.05 0.10 0.15 0.20 the performance Median year-to-year change in default point (liability) of the asset in Source: Moody's KMV question. When Year-to-year change in median EDF

calculated default point. Distance-to-default is empirically mapped to EDF by tracking the default experience of thousands of firms from Moody’s KMV’s extensive default database. Moody’s KMV uses actual default rates for companies in similar risk ranges to determine a one-to-one relationship between these two series.1 Periods of generally negative or low asset returns, large positive changes in default point, or positive changes in asset volatility are strongly associated with periods of sharp increase in EDF measures. This is illustrated by comparing the median year-to-year changes in each of the three main EDF drivers to the year-to-year change in the median EDF measure in the sample of 500 largest U.S. firms between 1970 to 2006 (see Charts 3, 4, and 5). Each point in these charts represents one quarter in the sample period. Given these patterns, it is possible and desirable to estimate statistical relationships between macroeconomic conditions and changes in EDF measures through changes in the three main drivers: change in the market value of assets, change in the default point, and change in asset volatility. The following sections outline an econometric approach that is able to link these indicators of corporate credit risk with some key economic factors to create an integrated framework that is suited to the performance of stress testing exercises.

4

5


3


far removed from the most pertinent region. Our aim here is not to provide a detailed technical outline of the methodology, references for which are available upon request. When using quantile regression for stress testing, we assume that stressed circumstances will instantly occur at the beginning of the forecast period. We thus “kick” the data out into the tail of the distribution (the 95th or 5th quantile in each case) and then let the stressed circumstance gradually recede until baseline conditions are reestablished. In this instance, we constructed this mechanism manually but, in reality, the duration of stressed circumstances should be estimable and reproducible. We then compare these stressed results with a baseline forecast consisting of relatively benign economic conditions coupled with a process that remains at the conditional mean through the entire forecast period. Interface between structural macro and corporate credit risk models. When modeling the drivers behind corporate credit risk, two alternative approaches are considered. One possibility is to start by modeling corporate risk at the aggregate level, and then use these estimates as a second step to benchmark the individual behavior of each company over time. An alternative is to develop a firm-specific model directly and then aggregate its outputs later (if one was interested in the overall performance of the corporate credit industry as well.) For both cases, we describe the estimation outputs and show some key results obtained from the stress testing exercises. Top-to-bottom approach. The top-tobottom approach develops an estimation of the aggregate credit risk measures first, and calculates firm-specific outputs as part of a second stage, where each company is benchmarked against the results obtained from the aggregate estimates. For the purposes of the aggregate estimations, the results of modeling each EDF driver at its median level are provided.2 The estimation was made over the 19722007 time period, generating very robust 2 It is worth mentioning that similar conclusions were found with mean EDF drivers as well.


www.economy.com

Table 1:

Estimation Results, Top-to-bottom Approach, Asset Returns Coefficient

t-statistic

Asset Returns, Lagged 1 Quarter

0.547

8.71

Asset Returns, Lagged 4 Quarters

-0.334

-5.59

Asset Returns, Lagged 8 Quarters

-0.140

-2.38

Default Point

0.530

3.10

Dividend Yield

-0.039

-5.82

Unemployment Rate

0.023

3.37

Real GDP, Annual Growth Rate, Lagged 4 Quarters

0.465

1.77

Real Investment (Non-residential), Annual Growth Rate

0.175

2.43

Lag Distribution of 10-year bond rate

0.001

3.19

-0.075

-1.89

Constant Term Adjusted R-squared

0.62

Standard Error of Regression

0.05

F-statistic Durbin-Watson Statistic

26.34 1.65

Notes: Sample (adjusted): 72Q1 - 07Q1; Included observations: 141 Sources: Moody’s KMV and Moody’s Economy.com

Table 2:

Estimation Results, Top-to-bottom Approach, Change in Asset Volatility Coefficient

t-statistic

Change in Asset Volatility, Lagged 1 Quarter

1.019

26.58

Change in Asset Volatility, Lagged 4 Quarters

-0.146

-3.32

Change in Asset Volatility, Lagged 8 Quarters

-0.024

-0.73

Inflation Volatility

60.331

3.24

Lag Distribution of Real GDP, Annual Growth Rate

0.013

2.70

Lag Distribution of Asset Returns

0.010

2.56

-0.031

-4.15

Constant Term Adjusted R-squared

0.91


0.02

F-statistic Durbin-Watson Statistic

238.35 1.71


4


and intuitive results (see Tables 1, 2 and 3). Although estimation outputs are shown for each EDF driver, the modeling framework we applied solves all three drivers simultaneously, thus accounting for any interdependence between the three factors. This simultaneous system is flexible enough to fully exploit and reflect the natural connections that exist across all three drivers. All three factors were found to be persistent over time.3 In terms of the economic forces that help explain asset returns over time, the model suggests that a strong economy is broadly associated with higher asset returns, as might be reasonably expected. As GDP and nonresidential investment growth strengthens, confidence in the overall economy helps to boost the aggregate value of assets. Similarly, higher long-term interest rates are associated with stronger returns on assets, due to the substitutability of long-term bonds and stocks as investment vehicles. On the other hand, dividend yields tend to be negatively related to asset returns. A result that is also worth mentioning is the negative relationship between unemployment and returns. After controlling for economic growth, firms might find it easier and cheaper to hire new employees in an environment where a higher unemployment rate pertains. In other words, labor becomes less scarce and firms are able to reduce costs of new hirings and matches offered for preexisting employees. The shrinking of labor costs can thus help companies boost their current and prospective profits, raising the corresponding asset return. As for the second risk driver—change in asset volatility—it was also found that an increase in underlying inflation volatility generates a commensurate hike in asset volatility. Moreover, an overheated economy influences this driver positively while the well-known positive correlation between asset returns and asset volatility is confirmed by the estimation output. The third factor that drives default probabilities—change in the default 3 This is captured by the strong statistical significance of the lagged dependent variables shown in Table 1.


www.economy.com

Table 2:

Estimation Results, Top-to-bottom Approach, Change in Default Point Coefficient

t-statistic

Change in Default Point, Lagged 1 Quarter

0.742

11.91

Change in Default Point, Lagged 4 Quarters

-0.104

-1.73

Change in Default Point, Lagged 8 Quarters

-0.021

-0.49

Asset Volatility

0.041

2.14

Inflation Rate

0.195

3.75

Lag Distribution of Real GDP

0.020

3.98

Constant Term

0.019

2.70

Adjusted R-squared

0.81


0.01

F-statistic

98.83

Durbin-Watson Statistic

2.02


point—was found to be positively associated with inflation and factors that suggest an economy in rapid expansion mode. In other words, as economic conditions improve and new business opportunities develop, companies are more likely to borrow money to expand their operations, leading to a corresponding increase in the amount of liabilities on hand. With the econometric estimations in place, stress testing exercises are performed in order to capture the effects of alternative macroeconomic scenarios on corporate defaults. The two economic outlook scenarios that are considered in this article are obtained from the Moody’s Analytics U.S. macro model. They represent the outcomes of a structural model that considers the connections between alternative sectors of the U.S. economy. The advantage of using such a framework relies on the consistency behind the various components of the model: those mitigating and propagating effects that govern the behavior of the economy are explicitly captured within a unified system. The first macroeconomic scenario assumes a broadly weak performance for the

U.S. economy in accord with our baseline expectations, but with a recession narrowly avoided. Labor and output markets are assumed to hold up despite the current turmoil in financial markets and, with the assistance of the external sector, are able to keep the economy from contracting enough to warrant a recession classification. On the other hand, the alternative downside scenario considers a 1-in-25-year recession that is consistent with a significant spillover of the mortgage crisis into other consumer credit products, creating a measurable hit on the performance of the labor and output markets. To illustrate this procedure, Chart 6 represents the predicted evolution of the median change in asset volatility under the two economic scenarios under consideration. Once both outlooks are run through the model and the corresponding future paths for all three drivers quantified, the EDF for each individual firm can be obtained, benchmarked against the computed aggregates. This procedure delivers simulated paths for each company’s asset returns, changes in asset volatility, and changes in default point. The final step in this top-to-bottom approach consists of map5


Chart 6: Change in Asset Volatility

Chart 7: Top-to-Bottom Approach, Alternative Scenarios EDF values, alternative scenarios, 2006 vs. 2007

Alternative scenarios 0.25

0.25 Actual

0.20

Recession scenario

Baseline scenario

0.20

0.15 EDF 2007Q4

0.10 0.05 0.00 -0.05 -0.10 -0.15

0.15 0.10

Baseline scenario

0.05 0.00 0.00

-0.20

Recession scenario 0.05

0.10 0.15 EDF 2006Q4

72 75 78 81 84 87 90 93 96 99 02 05 08 11 14 Sources: Moody's KMV, Moody's Economy.com

0.35

t=2006Q4

0.30 EDF 2007M12

Base scenario (t+1=2007Q4)

200

Severe scenario (t+1=2007Q4)

150 100 50 0 0.02 0.03 0.1

0.2 0.5 1.5 3 EDF Bucket, %

6

12

17


Bon-Ton Bakers

0.25 0.20 0.15 0.10

Cost Plus

Sears

0.05 0.00 0.00

20

Sources: Moody's KMV, Moody's Analytics

ping these drivers of corporate credit risk into an EDF measure for each individual company using the empirical mapping developed by Moody’s KMV that was explained earlier. Under the assumption of the U.S. economy sailing away from recession, the model predicts a continuing decrease in overall corporate defaults reflecting the long-established downhill trend in the aggregate series. Once the currently elevated asset volatility settles down and asset returns come back to a bullish trajectory, default probabilities for most of the largest 500 U.S. firms resume their descending path toward new record-low levels. If, on the other hand, a recession outcome is considered, the corporate credit risk picture changes significantly. Asset volatility picks up due to the high levels of uncertainty, while company liabilities tend to increase as firms face challenging financial times. The asset market suffers as well, pushing stock returns to negative ter-

7

EDF values, severe recession scenario, 2006 vs. 2007

EDF values, alternative scenarios, 2006 vs. 2007 250

0.25

Chart 9: Bottom-Up Approach, Retail-General Stores

6

300

0.20

Sources: Moody's KMV, Moody's Analytics

Chart 8: Top-to-Bottom Approach, EDF Distribution

Frequency

www.economy.com

0.01

0.02 0.03 EDF 2006M12

0.04

0.05

Sources: Moody's KMV, Moody's Analytics 8

ritory. All this gets translated into higher EDF measures across the board and results in a significant erosion in corporate credit risk going forward. Charts 7 and 8 illustrate these findings by comparing forecasted default probabilities (at the end of 2007) with the corresponding EDFs observed at the end of 2006. Bottom-up approach. This alternative methodology consists of modeling individual firms directly, without the support of a benchmarking intermediate step. This bottom-up approach requires the development of a panel data model with each individual firm acting as a cross-section unit. The advantage of this methodology over the previous one is reflected by its flexibility in terms of modeling specification: under the bottomup approach each company can be modeled to react differently to alternative economic drivers. But this comes at a price in terms of the complexity and dimensionality of the model once very heterogeneous firms are

considered as part of the same pool. Modeling at the firm level is better suited for analyzing a group of companies that share some common characteristics, such as firms that belong to the same industry, and where the modeler is interested in running industry or group-specific scenarios. If the user is interested in the response of a given set of firms to an overall macroeconomic scenario, the top-to-bottom approach, illustrated earlier, is recommended. As an application of the bottom-up alternative, a total of 26 firms are considered that are within the Retail-General Stores industry.4 With this sample in mind, a dynamic, fixed-effects, panel data estimation is run to analyze the impact of a retail-specific recession scenario (see Tables 4, 5 and 6). We construct this scenario using our well-established industry forecasting models. 4 The list of the companies can be found in Tables 4, 5 or 6.

6

9


www.economy.com

Table 4: The first driver of credit risk, asset returns, continues to be positively related to the strength of the economy, as measured by sustained GDP growth and low unemployment rate. Similarly, strong disposable income and retail sales growth both unsurprisingly lead to a good performance for retail companies with a corresponding increase on asset values. Solid profit margins are found to have a positive impact on asset returns as well. Asset volatility, on the other hand, as before responds to changes in the volatility of inflation and net profits. Moreover, an overheated economy, characterized by excessive sales markups and high interest rates, pushes, at the margin, for higher levels of asset volatility. A weak labor market was found to be an additional source of uncertainty with rising unemployment rates generating generally higher levels of underlying asset volatility. The third risk factor, the change in the default point, increases as operating expenses in the retailing industry grow. Similarly, higher disposable income and stronger retail sales go hand-in-hand with increases in firms’ liabilities. The stronger the retail market, the more likely that firms will borrow to expand their businesses. Finally, higher interest rates put financial pressure on the debt side of a company’s balance sheet with a corresponding increase of its default point. With the estimated dynamic panel data model in hand, a recession scenario is considered that severely hits the retail sector. This is just a variation on our macroeconomic recession scenario with a few linkages into the retail world given a little bit of extra sauce. The effects of the recession on future probabilities of default are then quantified for the 26 firms that are part of this exercise. As expected, most companies are severely affected by the downside stressed circumstances. Some firms in particular, however, are found to react more acutely than others (see Chart 9). For most of the companies, their future EDF remains below 5%, but there are four companies whose probabilities of default are shifted well beyond this limit. MOODY’S ANALYTICS / Copyright© 2007

Estimation Results, Bottom-up Approach, Asset Returns Coefficient

t-statistic

Asset Returns, Lagged 4 Months

0.610

39.50


-0.106

-6.77

0.274

23.49

-0.227

-3.32

Change in Default Point Change in Asset Volatility, Lagged 2 Months Change in Asset Volatility, Lagged 4 Months

0.302

4.51

FED Funds Rate

-1.168

-3.77

Unemployment Rate, Lagged 3 Months

-1.358

-2.57

Real GDP, Annual Growth Rate

3.390

4.79

Median Income, Annual Growth Rate

1.371

4.65

Change in Net Profit Margins

1.025

1.89

Retail Sales, Annual Growth Rate

0.510

1.51

Fixed Effects TUESDAY

12.83

2.81

99 CENTS

2.53

0.52

BAKERS

-2.23

-0.22 2.05

BIG LOTS

8.03

BJS

5.53

1.48

BON TON

9.97

2.43 0.97

COST PLUS

4.30

COST U LESS

9.68

2.72

COSTCO

9.09

2.60 0.69

DILLARDS

2.41

DOLLAR GENERAL

6.55

1.65

DOLLAR TREE

6.47

1.54 2.04

DUCKWALL

6.97

FAMILY

7.52

1.95

FREDS

8.63

1.96 1.46

GOTTSCHALKS

5.07

JCPENNEY

5.86

1.65

KOHLS

9.92

2.69 2.23

MACYS

8.27

MAUI

11.17

0.61

NORDSTORM

8.65

2.33 1.80

SAKS

8.53

SEARS

5.72

1.31

TARGET

8.54

2.45

TJX

9.78

2.56

WALMART

4.97

1.40

Adjusted R-squared

0.54

Standard Error of Regression F-statistic Durbin-Watson Statistic

27.02 121.72 2.52

Notes: Sample (adjusted): June 94 - August 07; Included observations: 158 after adjustments; Cross-sections included: 26; Total pool (balanced) observations: 3,456 Sources: Moody’s KMV and Moody’s Economy.com

7


www.economy.com

Table 5: Bakers jumps from a 0.05 probability to a 0.35 chance of default—the maximum allowed within this model. This should not be surprising since Baker’s performance was the worst of the sample at the end of 2006. Cost Plus, the second worst performer at the end of last year, jumps to a 10% chance of default. Bon-Ton, which performed poorly at the end of 2006, gets severely hit and moves to the upper limit for its calculated future probability of default. Finally, Sears shows up as a bad performer going forward, but what makes this an interesting case is the fact that its performance in 2006 was among the best of the group. Relatively speaking, this company is the one that is affected the most by a severe retail-specific recession. The reason behind this remarkable decline in fortunes is related to the extraordinarily negative effect that the recession outlook has on Sears’ asset returns. Summary. Our methodology proposes a transparent connection between key economic factors and corporate credit quality. Both top-down and bottom-up approaches delivered very robust results; they coincided in predicting a measurable deterioration in corporate credit quality should the U.S. economy enter into a recession but a continuation of sound conditions should the recession scenario be narrowly avoided. With the likelihood of recession growing daily, however, threats to corporate credit soundness should in no way be discounted. Problems in subprime, it must be remembered, were not apparent at the dawn of 2007 even though, by this stage, the damage caused by loose standards had already been wrought. Dismissing the potential for problems in the corporate sector because of current soundness is equally shortsighted. From a methodological standpoint, the outcome obtained from linking leading corporate risk measures with sound econometric techniques is highly encouraging. Given the robustness of the results, the integrated framework seems well-suited for stress testing purposes. With the advent of Basel II and its requisite requirement to stress test one’s entire portfolio, and with investors seeking ever MOODY’S ANALYTICS / Copyright© 2007

Estimation Results, Bottom-up Approach, Change in Asset Volatility Coefficient

t-statistic

Change in Asset Volatility, Lagged 4 Months

0.829

68.17


-0.186

-15.26


0.042

13.90

-0.034

-9.64

FED Funds Rate

1.358

14.77

Unemployment Rate

0.428

2.38

Change in Default Point

Inflation Volatility Median Income, Annual Growth Rate

0.588

3.24

-0.836

-7.66


0.201

1.98

Sales Markup

0.831

7.43

Net Profits Volatility

0.269

2.51

Fixed Effects TUESDAY

-27.76

-9.17

99 CENTS

-25.77

-8.57

BAKERS

-25.31

-8.43

BIG LOTS

-26.34

-8.54

BJS

-26.77

-8.87

BON TON

-27.02

-8.41

COST PLUS

-26.88

-8.84

COST U LESS

-27.20

-8.98

COSTCO

-26.57

-8.81

DILLARDS

-26.45

-8.72

DOLLAR GENERAL

-27.28

-9.03

DOLLAR TREE

-27.34

-9.06

DUCKWALL

-26.49

-8.76

FAMILY

-26.51

-8.78

FREDS

-26.19

-8.69 -8.78

GOTTSCHALKS

-26.57

JCPENNEY

-25.50

-8.51

KOHLS

-27.04

-8.98 -8.53

MACYS

-25.72

MAUI

-24.78

-7.70

NORDSTORM

-26.35

-8.71 -8.21

SAKS

-25.67

SEARS

-24.81

-7.66

TARGET

-25.36

-8.38

TJX WALMART Adjusted R-squared Standard Error of Regression F-statistic Durbin-Watson Statistic

-27.14

-8.98

-26.90

-8.89

0.73 7.50 259.80 2.47


8


www.economy.com

Table 6: more informed risk management practices, the ability to identify these transparent linkages represents a very valuable tool for quantifying the impact of alternative scenarios on corporate default probabilities. The methodology described here can thus help industry analysts to better manage their portfolios in response to changing economic conditions.

Estimation Results, Bottom-up Approach, Change in Default Point Coefficient

t-statistic

Change in Default Point, Lagged 2 Months

0.730

61.75

Change in Default Point, Lagged 8 Months

-0.042

-3.59

Asset Returns

0.188

16.58


-0.096

-8.35

Change in Asset Volatility

-0.351

-8.59


0.302

7.88

FED Funds Rate

0.324

2.08

Disposable Income, Annual Growth Rate

0.326

1.68

Operating Expenses, Annual Growth Rate

0.029

1.96


0.447

3.67

Net Profits Volatility

0.269

2.51

TUESDAY

-3.45

-1.28

99 CENTS

2.79

1.05

BAKERS

11.63

2.88

Fixed Effects

BIG LOTS

-2.26

-1.17

BJS

-2.21

-1.44

BON TON

3.09

0.87

0.75

0.41

COST U LESS

COST PLUS

-4.24

-2.88

COSTCO

-2.55

-2.09

DILLARDS

-3.73

-2.56

DOLLAR GENERAL DOLLAR TREE

1.95

0.68

-0.90

-0.48

DUCKWALL

-3.78

-2.74

FAMILY

-2.03

-1.47

FREDS

-2.40

-1.31

GOTTSCHALKS

-4.12

-2.79

JCPENNEY

-5.11

-4.19

-0.34

-0.18

KOHLS MACYS

-1.44

-0.92

MAUI

-9.52

-0.81

NORDSTORM

-4.19

-3.31

SAKS

6.58

1.64

SEARS

0.01

0.00

-3.37

-2.81

TJX

-2.96

-2.04

WALMART

-0.95

-0.76

TARGET

Adjusted R-squared

0.64


23.96

F-statistic

187.75

Durbin-Watson Statistic

1.48



9

AUTHOR BIO ��

www.economy.com

About the Authors Tony Hughes Tony Hughes is senior director of Credit Analytics at Moody’s Analytics, where he manages the company’s credit analysis consulting projects for global lending institutions. An expert applied econometrician, Dr. Hughes also oversees the Moody’s CreditCycle and manages CreditForecast.com. His varied research interests have lately focused on problems associated with loss forecasting and stress-testing credit portfolios. Now based in the U.S., Dr. Hughes previously headed the Moody’s Analytics Sydney office, where he was editor of the Asia-Pacific edition of the Dismal Scientist web site and was the company’s lead economist in the region. He retains a keen interest in emerging markets and in AsiaPacific economies. A former academic, Dr. Hughes held positions at the University of Adelaide, the University of New South Wales, and Vanderbilt University and has published a number of articles in leading statistics and economics journals. He received his PhD in econometrics from Monash University in Melbourne, Australia.

Juan Licari Juan M. Licari is a director at Moody’s Analytics. Dr. Licari is a member of the Credit Analytics group and specializes in financial economics. Juan leads consulting projects with major industry players, builds econometric tools to model credit phenomena, and has implemented several stress-testing platforms to quantify portfolio risk exposure. He has a leading role in the development and implementation of credit solutions and is actively involved in communicating these to the market. Dr. Licari holds a PhD and an MA in economics from the University of Pennsylvania and graduated summa cum laude from the National University of Cordoba in Argentina.

More on Consumer Credit Analytics from Moody’s Analytics CREDITFORECAST.COM Exclusive forecasts and analysis of household finances based on credit bureau data from Equifax. www.creditforecast.com

MOODY’S CREDITCYCLETM Integrating regional economics with industry-leading consumer credit forecasting & stress testing. www.economy.com/mcc

ADVISORY SERVICES Cutting-edge economic research, modeling capabilities & data analytics at your disposal. www.economy.com/consulting U.S./CANADA +1.866.275.3266

EMEA +44.(0) 20.7772.1646

ASIA/PACIFIC +61.2.9270 8111

OTHER LOCATIONS +1.610.235.5299

Email: [email protected]


10

About Moody’s Analytics Economic & Consumer Credit Analytics Moody’s Analytics helps capital markets and credit risk management professionals worldwide respond to an evolving marketplace with confidence. Through its team of economists, Moody’s Analytics is a leading independent provider of data, analysis, modeling and forecasts on national and regional economies, financial markets, and credit risk. Moody’s Analytics tracks and analyzes trends in consumer credit and spending, output and income, mortgage activity, population, central bank behavior, and prices. Our customized models, concise and timely reports, and one of the largest assembled financial, economic and demographic databases support firms and policymakers in strategic planning, product and sales forecasting, credit risk and sensitivity management, and investment research. Our customers include multinational corporations, governments at all levels, central banks and financial regulators, retailers, mutual funds, financial institutions, utilities, residential and commercial real estate firms, insurance companies, and professional investors. Our web and print periodicals and special publications cover every U.S. state and metropolitan area; countries throughout Europe, Asia and the Americas; and the world’s major cities, plus the U.S. housing market and other industries. From our offices in the U.S., the United Kingdom, and Australia, we provide up-to-the-minute reporting and analysis on the world’s major economies. Moody’s Analytics added Economy.com to its portfolio in 2005. Its economics and consumer credit analytics arm is based in West Chester PA, a suburb of Philadelphia, with offices in London and Sydney. More information is available at www.economy.com.

© 2010, Moody’s Analytics, Inc. and/or its licensors and affiliates (together, “Moody’s”). All rights reserved. ALL INFORMATION CONTAINED HEREIN IS PROTECTED BY COPYRIGHT LAW AND NONE OF SUCH INFORMATION MAY BE COPIED OR OTHERWISE REPRODUCED, REPACKAGED, FURTHER TRANSMITTED, TRANSFERRED, DISSEMINATED, REDISTRIBUTED OR RESOLD, OR STORED FOR SUBSEQUENT USE FOR ANY PURPOSE, IN WHOLE OR IN PART, IN ANY FORM OR MANNER OR BY ANY MEANS WHATSOEVER, BY ANY PERSON WITHOUT MOODY’S PRIOR WRITTEN CONSENT. All information contained herein is obtained by Moody’s from sources believed by it to be accurate and reliable. Because of the possibility of human and mechanical error as well as other factors, however, all information contained herein is provided “AS IS” without warranty of any kind. Under no circumstances shall Moody’s have any liability to any person or entity for (a) any loss or damage in whole or in part caused by, resulting from, or relating to, any error (negligent or otherwise) or other circumstance or contingency within or outside the control of Moody’s or any of its directors, officers, employees or agents in connection with the procurement, collection, compilation, analysis, interpretation, communication, publication or delivery of any such information, or (b) any direct, indirect, special, consequential, compensatory or incidental damages whatsoever (including without limitation, lost profits), even if Moody’s is advised in advance of the possibility of such damages, resulting from the use of or inability to use, any such information. The financial reporting, analysis, projections, observations, and other information contained herein are, and must be construed solely as, statements of opinion and not statements of fact or recommendations to purchase, sell, or hold any securities. NO WARRANTY, EXPRESS OR IMPLIED, AS TO THE ACCURACY, TIMELINESS, COMPLETENESS, MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OF ANY SUCH OPINION OR INFORMATION IS GIVEN OR MADE BY MOODY’S IN ANY FORM OR MANNER WHATSOEVER. Each opinion must be weighed solely as one factor in any investment decision made by or on behalf of any user of the information contained herein, and each such user must accordingly make its own study and evaluation prior to investing.


11

CONTACT US For further information contact us at a location below: U.S./CANADA +1.866.275.3266

EMEA +44.20.7772.5454

Email us: [email protected] Or visit us: www.economy.com Copyright © 2010, Moody’s Analytics, Inc. All Rights Reserved.

ASIA/PACIFIC +852.3551.3077

OTHER LOCATIONS +1.610.235.5299