Measuring credit risk: does complexity matter?

Jesswein, Kurt

ABSTRACT

This paper examines the issue of how to measure a company's ability to meet its external financing commitments. Success in this area is critical for businesses extending credit and is a topic of considerable interest yet with varied coverage in classrooms and business training rooms. A major component is finding an appropriate metric with which to evaluate the creditworthiness of borrowers. Given the level of importance placed on the topic in the world of credit and in academia, one finds a plethora of approaches with little consensus among them.

We review a variety of approaches to measure current and future liquidity and creditworthiness. Initial tests determine the relative strengths and weaknesses of the approaches with recommendations provided on how to best evaluate the topic, both in the classroom and in the practitioners' world. Of particular interest is the calculation of the various coverage ratios designed to measure the borrower's ability to meet its current and future financial obligations.

Results indicate that many of the key financial ratios used seem to follow very similar tracks. Extensive variations among the different formulations of the ratios appear to offer little additional insights. We are left with a call to solidify or refine the most straightforward approaches to evaluating credit as it appears that a simpler set of information to work with would allow for better analysis of the underlying reasons for any deviations or any volatility in said numbers.

INTRODUCTION

Measuring a company's ability to meet any external financing commitments is a topic that is critical to the success of businesses extending credit, such as commercial banks, and one of considerable interest in university classrooms and business training rooms. A major component of this topic is finding an appropriate metric with which to evaluate the creditworthiness of borrowers. Given the level of importance placed on the topic in the world of credit and in academia, one finds a plethora of approaches with little consensus among them.

This paper examines the variety of approaches used to measure a company's current and future liquidity and creditworthiness. Initial tests are performed to determine the relative strengths and weaknesses of some of the approaches with recommendations provided on how to best approach the topic, both in the classroom and in the "real-world." Of particular interest is the calculation of the various coverage ratios designed to measure the borrower's ability to meet its current and future financial obligations.

OVERVIEW OF LIQUIDITY MEASUREMENT TOOLS

In lending and various other debt contracts, the providers of funds tend to place various conditions on the borrowers in order to protect their investment. Many of the conditions amount to debt covenants requiring the borrowers either to provide specific information or actions (positive covenants) or more likely restrict the borrowers' activities (negative covenants). Examples of positive covenants would include the requirement that a borrower provide timely, audited financial statements or ensure proper insurance of the assets being financed. Negative covenants may include restrictions on the payment of dividends or thresholds placed on particular financial ratios that the borrower may not exceed (or fall below). In fact, accounting-based covenants are quite common, although they are often instituted on a case-by-case basis. In one broad-based study, Dichev & Skinner (2002) examine the variety of lending terms most often found in the loan facilities. After their exhaustive study, they found the most common items found in debt covenants were debt-to-cash flow measures and both interest coverage and fixed charge coverage ratios

Unfortunately, there are difficulties in examining such measures in that they can be defined in a variety of ways. This variation of definitions is argued to be necessary given that lenders must often uniquely define financial statement variables as they customize loans to suit specific borrower characteristics (Leftwich, 1983). This does not preclude, however, the topic of addressing which approaches may be most meaningful. For example, many (most) finance and accounting textbooks include various credit ratios in their coverage of financial statement analysis with little consistency among the approaches. This is particularly the case for the fixed charge coverage ratio.

In most cases the fixed-charge coverage ratio is seen as an extension of the more standard interest coverage ratio (often referred to as times interest earned), a ratio itself defined along the lines of a comparison of a company's operating earnings to its interest expenses. Extending the ratio to the broader "fixed charge" measure often includes adding components such as the required principal payments on debt and capital leases (not just the interest) and other fixed finance charges such as the required payments (the implicit interest and/or principal payment) of off-balance sheet operating leases and of preferred stock dividends. A review of many of the differing approaches follows.

Many financial management textbooks, such as Block & Hirt (2005), define the fixed charge coverage ratio as the ratio of income before fixed charges and taxes to fixed charges, with fixed charges defined as the sum of interest expenses and operating lease payments (sometimes also referred to as rental expenses). We will generally use the operating lease terminology because it is the accounting conventions associated with operating leases that are the key factors in causing the expenses to be of such concern.

Brigham & Daves (2007) and others define fixed charge coverage as the ratio of earnings before interest, taxes, depreciation and amortization plus operating lease payments to the sum of payments for interest, debt principal, and leases. Similarly, Koller, Goedhart and Wessels (2005) and others define it as the ratio of earnings before interest, taxes, depreciation, amortization, and rental (lease) expenses to fixed charges, defined as the sum of interest and lease payments.

Another view shifts the focus from one that examines fixed financing charges to one that encompasses the entire debt servicing that a company faces. Gitman (2006) and others expand the basic fixed charge coverage definition to one measuring the ratio of earnings before interest and taxes plus lease payments to the sum of interest expense, lease payments, as well as debt principal and preferred stock payments, with both of the latter divided by one minus the tax rate because the payments are made with after-tax dollars.

The fixed charge coverage ratio can also be defined along the lines of the earnings to fixed charges ratio required by the Securities and Exchange Commission (SEC) in securities offerings (Regulation S-K [section]229.503). In this regard, Gibson (2007) and others define the fixed charge coverage ratio as the ratio of recurring operating earnings plus one-third of the total operating lease payments to the sum of interest expense (including capitalized interest) and one-third of operating lease payments. The one-third factor applied to the operating lease payments is a widely-used "rule-of-thumb" measure of estimating the finance charges implicit in the use of operating leases. This type of approach most impressively (depressingly?) materializes itself in to what is very likely the most complex fixed charge ratio (or any other financial ratio) found in any standard college textbook, in this case in Wild, Subramanyam, & Halsey (2007). The authors attempt to operationalize the SEC's earnings to fixed charges ratio (from Regulation S-K Paragraph 503d) as follows [and remember this is a condensed version]:

 Pre-tax income before discontinued operations, extraordinary items,
 and cumulative effects of accounting changes plus interest incurred
 less interest capitalized, amortization of interest expense and
 discount or premium, interest portion of operating rental expenses,
 preferred stock payments (on a pre-tax basis) and the amortization
 of previously capitalized interest, all divided by the sum of total
 interest expense, the amortization of interest expense and discount
 or premium, the interest portion of operating rental expenses, and
 any preferred stock payments (on a pre-tax basis).

Others, most notably Damodaran (2001), attempt to introduce a more "scientific" approach to estimating the finance charges associated with operating lease payments. Rather than blindly following the "one-third" rule, they argue for a more "economically-sound" argument in which the present value of future operating lease obligations are capitalized with the capitalized amount then multiplied by the company's cost of debt to determine the "true" cost of financing implicit in the operating leases.

Shifting from the academic to the practitioner's world of finance, we find just as wide a variety of approaches. For example, Worldscope (Thomson Financial, 2003) equates the fixed charge coverage ratio to its definition of the interest coverage ratio, defining it as the ratio of earnings before interest and taxes to the sum of interest expense and preferred dividends divided by one minus the tax rate. Moody's defines it as the sum of net income, non-cash adjustments and changes in working capital, interest expense, and lease expense, all divided by the sum of interest and lease expense (Neuhaus, 2001). Standard & Poor's defines it as the ratio of earnings before interest and taxes and rent to total interest plus rents, taking into account any preferred stock dividends when material (Standard & Poor's, 2005). And if that is not confusing enough, Standard & Poor's, in its Compustat database, defines its close cousin, the debt service ratio, in at least two distinct ways. First, it is defined as the ratio of free cash flows (cash flow from operations less capital expenditures and common stock dividend payments) plus interest expense to the sum of interest expense and current maturities of long-term debt, Second, it is defined as the ratio of net income plus depreciation and amortization to the total amount of debt due within one year. Standard & Poor's also provides two other closely related measures--EBITDA interest coverage and pretax fixed coverage. These are specifically defined as the ratio of earnings before interest, taxes, depreciation and amortization to interest expense and as pretax income plus interest and rental expenses divided by the sum of interest and rental expenses.

Another major participant in the credit analysis game, the Risk Management Association (RMA), focuses on what it refers to as "debt service principal and interest coverage." It defines this ratio as net cash after operations divided by current debt obligations and is essentially the ratio of the company's cash flow from operating activities (referred to as cash income) to the sum of its cash interest obligations and current portion of its long-term debt and capital leases.

Last but not least, whenever examining the topic of a company's ability to meet financial obligations, discussions often flow toward the topic of bankruptcy, which may be seen as the ultimate in inability to meet such obligations. Measuring the risk of bankruptcy has its own variety of approaches, which tend to be broader in scope but not any less confusing that the topic of coverage ratios.

Led by relatively simple models proposed by Beaver (1966), Altman (1968), and Ohlson (1980) and later developed into significantly more complex offerings such as by Hillegeist, Keating, Cram, & Lundstedt (2004) and others, researchers have for many years looked for the holy grail of accounting ratios that could be most useful in predicting bankruptcy. The best known of these models is the Altman Z-score. Using multiple discriminant analysis on a variety of financial ratios, the final model is a simple weighted average of five accounting ratios (working capital, retained earnings, earnings before interest and taxes, and sales, each in relation to total assets plus the ratio of market value of equity to book value of liabilities). The result is compared with arbitrary cutoff points indicating either a high or low probability of bankruptcy.

Altman's model remains the standard against which all others are compared and tends to be the one most embraced by practitioners (IOMA, 2003), even though it is some 40 years old and has faced a constant barrage of criticism. Altman (Altman, Haldeman & Narayanan, 1977) and Beaver (Beaver, McNichols & Rhie, 2005) themselves have looked to improve upon the more basic models proposed earlier in their careers (one important extension has been in dealing with the problems associated with operating leases). Yet despite the continued work in this area, simplicity may be the key. In a recent working paper examining the history of bankruptcy prediction models, Bellovary, Giacomino, & Akers (2006) conclude that given the already high predictive ability of even the simplest models (they cite Beaver's 92 percent accuracy with one ratio to a more recent model that considers 57 factors yet yields only an 86 percent accuracy rate!), efforts should be shifted from developing new and improved models to refining the existing ones and making them more understandable and useful for practitioners. Thus, more may be less as too many ratios or ratios with too much complexity can actually make a model less useful.

Against this broad and often conflicting set of approaches to assessing credit risk of borrowers, we focus on an empirical examination of the key tools proposed. Realizing that financial ratio calculations are complicated by the complexity of accounting and reporting standards and the differing needs of users of financial statements, we focus on two items.

First and foremost we examine whether the level of complexity really matters. That is, is there any additional information contained in the more-broadly defined ratios and more complex models that would lead one to prefer one over the others or should simplicity be the overriding concern? For example, there is the divergence in the fixed charge coverage ratio's handling of lease expenses. Many models include these charges, some do not. And for those that do incorporate the expenses, there are those who incorporate the entire amount of lease payments as a charge and those who only use the implied financing component of the leases. And for those incorporating only the financing component of the leases, there is the split between those who favor a practical approach of using the "rule of thumb" estimate (one-third of the total lease expense) and those who favor a more mathematically-sound present value approach.

And second, we examine how well the varied approaches correlate with the more grandiose concept of assessing overall default or credit risk. For example, how do the various measures relate to a company's given risk rating and how well do they correlate with broader bankruptcy metrics (e.g., Altman's Z-score) that tend to examine a wider array of factors than simply short-term liquidity? The results of these inquiries may lead us to make better judgments about how to approach this matter, both in the classroom and in applying it in "real-world" situations.

DATA AND METHODOLOGY

Data for this study was gathered from Compustat (Research Insight). We focus primarily on one set of companies, namely large companies (defined as those with revenues in excess of $1 billion in their most recent reporting year) that do not primarily operate in the financial services area. To be included in the study, the company also had to have financial data for each variable examined plus a credit rating assigned to them by Standard and Poor's for each of the past five years (for most companies this includes data from 2001 through 2005, but for others with non-December year-ends, the time period may extend in to 2006). Given the strict criteria employed, only 259 companies remained in the primary sample group. A broader set of companies, in which the final requirement of having five consecutive years of data for all variables examined is relaxed, provided us with a much larger group of 1,320 companies. Extending our analysis to this more broadly-defined group of companies allows us to make broader generalizations of the results of the study.

Given the wide range of approaches to this topic, a virtually endless array of potential variables could be examined. However, for the purposes of this study, the following ones were selected:

(1) the credit rating (SPDRC) assigned by Standard & Poor's as its opinion of an issuer's overall creditworthiness at each time period. Each letter rating has an associated numerical rating within Compustat (i.e., AAA = 2, A = 8, BBB- = 12, etc.). Although Standard & Poor's is by no means the only agency capable of providing credit ratings, the availability of their ratings within the Compustat database facilitated their use in the study;

(2) the Z-score (ZSCORE) calculated by Compustat for each period (the sum of 1.2 times the working capital divided by total assets, 1.4 times the retained earnings divided by total assets, 3.3 times the earnings before interest and taxes divided by total assets, 0.6 times the difference between the market value of equity [year-end stock price times common shares outstanding plus the par value of any preferred stock] and the book value of the liabilities, and 0.999 times sales divided by total assets) for each period;

(3) the interest coverage ratio (IntCov) calculated by Compustat (total of operating and non-operating income before taxes and minority interest plus interest expense, all divided by interest expense) for each period;

(4) the EBITDA coverage ratio (EBITCov) calculated by Compustat (operating income before depreciation and amortization expenses divided by interest expense) for each period;

(5) the debt service coverage ratio (DbtServ) as calculated by Compustat (net cash flow from operations less cash dividends and capital expenditures plus interest expense, all divided by the sum of interest expense and current portion of long-term debt due; and

(6) three separate fixed charge coverage ratios. The first one (FxdChg) is defined as total pretax income plus interest expense and lease expenses, all divided by the sum of interest and lease expenses. The second (FxdFin) uses only the financing cost (estimated as one-third of the total lease expense) rather than the total lease expenses; and the third (FxdLeas) uses the present value approach to estimating the financing cost of the lease expense. This is measured as the sum of the most recent year's lease payment plus the discounted values for each of the successive five years of obligations, all multiplied by the company's effective interest rate as estimated by dividing the company's interest expense for the year by its total amount of debt for the year.

The focal point of the study is the relationship among the various measures as tools used to analyze creditworthiness and liquidity. Initially, Pearson and Kendall-tau correlations were reviewed to look at the relative strength of association between alternative liquidity measures and the perceived creditworthiness of the company as estimated by Standard & Poor's credit ratings. Despite the large sample sizes, both parametric and non-parametric methods were used due to considerable concerns about the homogeneity of the variances within the data. Kendall-tau was chosen over the more often used Spearman-rank-correlation test because its correlations reflect the strength of the relationships between the variables and it copes with ties much better than the Spearman method. It is also superior as a test of independence because it is sensitive to some types of dependence which can not be detected using the Spearman method. A straightforward discussion on the preference for Kendall-tau can be found in Noether (2007).

These relationships among the variables were examined first for the narrow sample group and then extended to the more broadly-defined sample. In addition, an analysis was made on the relationships among the variables. Since many of them are based on similar criteria, it can be assumed that they are likely closely related to one another. However, it is the extent of that relationship that is of interest. If a particular conclusion or relationship can be found by using a more concise tool, the argument can be made to favor further refinements to the more efficient method to make it a more effective tool rather than trying to develop a better tool.

RESULTS

For the initial tests, a total of 259 companies met all of the criteria for inclusion in the sample. That is, there was sufficient data available such as size of company (over $1 billion in sales in the most recent annual period), five years of credit rating data, and five years of financial ratio data, to be included in the study. On the other hand, relaxing all but the size criterion increased the sample size to 1,320 companies, a larger sample in which greater generalizations about the results might be made when appropriate. Relaxing all of the criteria (except for the requirement of being a non-financial company) resulted in a sample of 5,802 companies. However, due to missing data, the "available" sample size would have been much smaller and given the large number of extreme and nonsensical data outliers, this larger sample was of very limited value and was therefore not examined further.

The two samples proved to be quite similar in terms of the breadth of credit ratings within each sample. The smaller sample of 259 companies was spread out among the 16 highest ratings, from AAA through B-, with 4.2 percent of the companies landing in the first group of four ratings, 30.7 percent in the next group, 45.7 percent in the third group, and 19.3 percent in the final group. The broader sample had 3.3 percent of the companies with one of the top four ratings, 20.1 percent in the second group of four, 43.7 percent in the third group, and 23.8 percent in the fourth. (An additional 1.6 percent had rankings lower than B-.)

Examining the relationships themselves, it is probably not too surprising to find that each of the key variables, from the Z-score to the myriad of coverage and debt service measures, proved to be significantly correlated (generally beyond the 99th percentile in significance) with the company's credit rating for each of the past five years. (See Table 1 below). In addition, we note that the levels of correlation with the credit rating also remained fairly consistent over the five years. Only the Kendall-Tau statistics are produced due to the concerns over normality of the data, although the Pearson correlations had many of the same results with the exception of some extremely unusual results attributed to the non-normal distributions.

Also seen in the table, the three fixed charge coverage ratios typically had much higher levels of correlation with the credit ratings than the other ratios, much more so than even the more encompassing Z-score. On the other hand, the debt-service ratio is significantly less likely to be related to a company's current credit rating.

If we examine the more broadly defined sample (see Table 2 below), we find similar results across the board although there are some major differences in the levels of correlation between the various variables. For example, the correlation coefficients are generally lower in each case, with the biggest drop-off in the Z-score correlation itself.

In addition, the debt service ratio, already significantly lower than the others within the narrow sample, essentially becomes a non-factor with the broader sample. Another noteworthy observation is that it appears that the more complex fixed charge ratio (the one explicitly dependent on the valuation of operating leases as debt-equivalents) is slightly better at evaluating credit ratings than the simpler one-third approach.

Having found generally high levels of correlation among the variables, we next briefly examine the relationships among the variables. As they tend to measure very similar items, it is not surprising to find very close relationships among them. Although the tables are not reproduced here, the relationships tended to be in the 0.50 to 0.70 range among the majority of the variables, with a couple of notable exceptions. First, the debt service ratio has very low correlations with the other variables, at approximately 0.20 across the board. This is not totally unexpected given its lower level of relationship with the credit rating to begin with. On the other hand, the two competing fixed charge coverage ratios (one-third versus present value rules) are very closely related, with a relationship that has become more significant over time. Based on the earliest data, the correlation was 0.76 between the two measures for the smaller sample group with this figure steadily rising to 0.92 in the most recent period. The broader sample had similar results: a coefficient of 0.73 initially, rising to 0.91 in the most recent period.

Making sense of this close relationship can come from different directions. From the practitioner's viewpoint, there appears to be little to gain from adding complexity in evaluating the financing charges associated with operating leases beyond using the basic one-third rule favored in industry. From an academic viewpoint, we are left to ponder why this association has become consistently stronger over the past few years. For example, interest rates generally increased between 2001 and 2005. Whether this positively (or negatively) impacted the use of operating leases or simply effected the present value calculations is left for a future study.

CONCLUSIONS AND DISCUSSION OF FUTURE RESEARCH

Although much more can certainly be examined regarding the various measures of assessing credit and liquidity risk, this first small step has provided some insights. For example, despite the overwhelming number of formulations, it appears that most of the key ratios seem to follow very similar tracks. Thus, rather than confusing potential users of financial information with so many variations, solidifying or refining the more straightforward approaches to evaluating credit may be called for. Notwithstanding any specific requirements associated with unusual or extraordinary credit situations, it would appear that focusing on a simpler set of ratios would allow for better focus (and hence better analysis) of the information contained in the ratios and underlying reasons for any deviations or any volatility in the said numbers.

There is much additional research waiting to be pursued in this area. For example, all of the data used, including the credit ratings, were from a single source, Standard and Poor's. If S&P largely uses its own data and formulations to make credit rating decisions, then there may be an undue amount of overlapping results. Evaluating other sources of credit ratings (e.g., Moody's or specific private lenders' evaluations), could provide different insights.

This analysis is rather static. A key reason for analyzing liquidity and credit risks is to look forward and predict where problems may arise. For example, can shifts in credit ratings be predicted or forecasted based on projecting specific financial variables into the future; that is, are there specific financial ratios that can be used as leading or lagging variables in determining trends in credit ratings?

Furthermore, one major criticism of many of the earlier bankruptcy studies was that their so-called results were specific to the time frame and/or types of companies evaluated. Extending this analysis to examine differences in results based on NAICS (North American Industry Classification System) codes or other industry designations or expanding it to cover other time periods could make the results stronger.

Our results are but a start in a process that needs to meet the challenge of overcoming the lack of connection between academic researchers, striving to build better mousetraps, and practitioners, looking for efficient mousetraps to improve their successes (and minimize their failures) in assessing credit risk. If academics can do a proper job of exposing users of financial information (managers, lenders, analysts, investors, etc) to the proper tools to conduct their analysis, then further studies in this area may prove to be quite useful.

REFERENCES

Altman, E. (1968). Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. Journal of Finance, 23:4, 589-609.

Altman, E. I, R.G. Haldeman, & P. Narayanan (1977). Zeta analysis[TM]: A new model to identify bankruptcy risk of corporations. Journal of Banking and Finance, 1:1, 29-54.

Beaver, W. 1966. Financial ratios as predictors of failure. Journal of Accounting Research 5, 71-111.

Beaver, W, M. McNichols, & J.W. Rhie (2005). Have financial statements become less informative? Evidence from the ability of financial ratios to predict bankruptcy. Review of Accounting Studies, 10:1, 93-122.

Bellovary, J., D. Giacomino, & M. Akers (2006). A review of bankruptcy prediction studies: 1930 to present. Working.

Block, S.B. & G.A. Hirt (2005). Foundations of financial management (Eleventh Edition). New York: McGrawHill/Irwin.

Brigham, E. & P. Daves (2007). Intermediate financial management (Ninth Edition). Mason: Thomson/South-Western.

Damodaran, A (2001). Corporate finance: Theory and practice (Second Edition). New York: John Wiley.

Dichev. I.D. & D.J. Skinner (2002). Large-sample evidence on the debt covenant hypothesis. Journal of Accounting Research, 40:4, 1091-1123.

Gibson, C. (2007). Financial reporting & analysis: Using financial accounting information. Mason: Thomson/SWestern.

Gitman, L.J. (2006). Principles of managerial finance (Eleventh Edition). Boston: Pearson/Addison-Wesley.

Hillegeist, S.A., E.K. Keating, D.P. Cram, & K.G. Lundstedt (2004). Assessing the Probability of Bankruptcy. Review of Accounting Studies, 9, 5-34.

IOMA-The Institute of Management & Administration (2003). Z-score: the old reliable method for predicting bankruptcy still works. Managing Credit, Receivables & Collections, 3:10, 6-7.

Koller, T., M. Goedhart, & D. Wessels (2005). Valuation: measuring and managing the value of companies (Fourth Edition). New York: John Wiley.

Leftwich, R. (1983). Accounting information in private markets: Evidence from private lending agreements. The Accounting Review, 58:1, 23-42.

Neuhaus, D. (2001). Global chemicals industry: financial ratio analysis for chemical companies. New York: Moody's, Inc.

Noether, G.E. (2007). Why Kendall tau? Accessed on March 1, 2007 at http://www.rsscse.org.uk/ts/bts/noether/text.html.

Ohlson, J. (1980). Financial ratios and probabilistic prediction of bankruptcy. Joumal of Accounting Research, 18:1, 109131.

Standard & Poor's (2005). Corporate ratings criteria: 2006. New York: McGraw-Hill.

Thomson Financial (2003). Worldscope database definitions guide. Accessed on March 1, 2007 at http://bib.kuleuven.be/etew/data/handleidingen/ worldscope_datatype_definitions_guide.pdf.

Wild, J.J., K.R. Subramanyam, & R.F. Halsey (2007). Financial statement analysis (Ninth Edition). New York: McGraw-Hill/Irwin.

Kurt Jesswein, Sam Houston State University

Table 1: Kendall Tau b Correlation Coefficients
Correlation with Credit Rating (Primary Sample of 259 Companies)

 Year 0 Year -1 Year -2 Year -3 Year -4

Zscore 0.2736 0.3030 0.3078 0.2947 0.2728
IntCov 0.5075 0.4823 0.4902 0.4433 0.4398
EBITCov 0.4972 0.4790 0.4134 0.3870 0.3669
FxdChg 0.5362 0.5084 0.5333 0.4832 0.4469
FxdFin 0.5507 0.5183 0.5299 0.4837 0.4520
FxdLeas 0.5485 0.5303 0.5519 0.5111 0.4891
DbtServ * 0.1018 0.1506 0.1557 * 0.1023 * 0.1065

Note: All correlations significant beyond 99% with the exception of
those cells marked * which are significant beyond 95%.

Table 2: Kendall Tau b Correlation Coefficients
Correlation with Credit Rating (Broader Sample of 1,320 Companies)

 Year 0 Year -1 Year -2 Year -3 Year -4

Zscore 0.1771 0.1949 0.2117 0.2234 0.2045
IntCov 0.3919 0.4004 0.4156 0.3939 0.3866
EBITCov 0.3983 0.3790 0.3646 0.3396 0.2976
FxdChg 0.4511 0.4610 0.4682 0.4531 0.3985
FxdFin 0.4422 0.4520 0.4562 0.4439 0.4021
FxdLeas 0.5215 0.5247 0.5038 0.4718 0.4282
DbtServ * 0.0481 0.0782 0.0851 ** 0.0424 ** 0.0297

Note: All correlations significant beyond 99% with the exception of
those cells marked * which are significant beyond 95%,
** significance below 95%.