Do credit ratings really affect capital structure

Kisgen 2006, 2009 finds the credit rating-capital structure CR-CS model i.e., firm capital structure policy is influenced by credit ratings to be generally descriptive of how firms beha

1

Do Credit Ratings Really Affect Capital Structure?

Kristopher J Kemper* University of Wisconsin-Eau Claire

Ramesh P Rao Oklahoma State University

Abstract

This paper revisits recent investigations into the role credit ratings play in the marginal financing behavior of firms While it has long been documented that credit ratings may be an important determinant of firm capital structure policy, academics have only recently subjected this

motivation to empirical scrutiny We add to the brief existing literature by investigating the sensitivity of marginal financing behavior of firms to a number of attributes deemed to capture firms’ affinity to emphasize credit ratings in their financing behavior Our results suggest that credit ratings are not a first order concern in capital structure decisions

1 Introduction

Until recently, the effect of credit ratings on capital structure had not been formally investigated The motivation for the recent interest in this topic can be traced to Graham and Harvey’s (2001) survey paper that lists maintaining a credit rating as the second most important objective in a firm’s credit policy Kisgen (2006, 2009) finds the credit rating-capital structure

(CR-CS) model (i.e., firm capital structure policy is influenced by credit ratings) to be generally

descriptive of how firms behave We argue that the CR-CS motivation is more applicable to a subset of firms than to all firms generally At a minimum, even assuming CR-CS applies to the average firm, we expect that its appeal will vary systematically across firms classified by certain firm-level attributes We examine several attributes First, we test the sensitivity to firms that are active, or likely to be active, capital market participants versus firms that are less active in capital

*Department of Accounting and Finance, Schneider 300E, College of Business, University of Wisconsin – Eau Claire, Eau Claire, WI 54702 Ph: 715-836-3137 Email: kemperkj@uwec.edu

markets Second, we examine the sensitivity of the CR-CS model to the bond rating of the firm That is, is an AA rated firm more motivated to take capital structure actions to maintain its rating than an A rated firm? Also, is a firm that is on the cusp of the investment/non-investment grade rating more motivated by CR-CS considerations than other firms? Third, we test the sensitivity

of the CR-CS motivation to firms that are active in the commercial paper market compared to firms that are not Lastly, we examine the capital structure behavior of firms as it relates to the investment opportunities available to these firms We argue that firms with greater growth opportunities would be more likely to be concerned with maintaining or achieving a long-term rating The foundation for this argument is that these firms are likely to be raising capital in the near future and would be interested in doing so at the lowest possible cost

We reconfirm Kisgen’s (2006) findings that firms at the edge of a ratings change have a propensity to use less debt at the margin, thus supporting the CR-CS model However, we are unable to document that the CR-CS motivation is systematically related to any of the attributes listed above, which we argued should proxy for management’s inclination to adopt the CR-CS model Especially damaging is the fact that the CR-CS model does not appear to hold across all rating classes In fact, our analysis indicates that with the exception of B rated firms, firms in no other rating group seem to curtail debt financing when faced with the prospect of losing its rating Thus, Kisgen’s original findings appear to be driven by the subsample of firms with extremely low ratings This is weak evidence in support of the CR-CS model, especially given that B rated firms are generally associated with financial distress Therefore,

their marginal financing behavior to avoid debt may be more an indication of lack of access to the debt market than an indication of a conscious attempt to decrease debt financing

Additionally, the CR-CS model implies that firms on the cusp of the investment and

non-investment grade rating should be especially sensitive to the impact of their marginal financing behavior on their credit ratings However, our results do not find this to be the case With regard

to the other attributes, our results are just as puzzling We do not find that the CR-CS model is more applicable to firms that have external financing needs and firms that regularly access the capital markets, firms that access the commercial paper market, and high growth firms These results lead us to conclude that the CR-CS model is not a good descriptor of how firms determine their marginal financing decision We are careful to point out that this does not necessarily mean that firms and CFOs do not consider credit ratings to be an important determinant of their capital structure policy Such a conclusion is hard to justify given that survey evidence indicates

managers consider credit ratings to be one of the most important determinants of target capital structure It is conceivable that there is too much noise in marginal financing data to obtain significant findings It is also possible that firms on the verge of losing or improving their current ratings may use other tools at their disposal to maintain or improve their rating such as asset restructuring (e.g., asset sales, spin-offs) and operating cost changes (e.g., layoffs, outsourcing, offshoring) In this context it is important to note that other factors besides leverage also impact ratings including profitability, quality of assets, etc

policy (second only to “financial flexibility”) Using this survey as a motivation, Kisgen (2006) conducts one of the first formal tests of the CR-CS model Kisgen argues that for credit ratings to have an independent effect on capital structure, there must be discrete shifts in the costs

experienced by the firm across the various rating categories He argues that an implication of the model is that “firms near a credit rating upgrade or downgrade issue less debt relative to equity than firms not near a change in rating.” Operationally this means that a firm with a Plus or Minus rating will be reluctant to issue debt at the margin.1 The Plus and Minus designations serve as a signal that a firm is on the verge of a ratings change The result, according to the CR-

CS theory, is that firms on the edge should be reluctant to issue debt A firm with a Plus rating would not want to sacrifice an opportunity to move into a higher credit rating by issuing debt; and a firm with a Minus rating would not want the credit rating agency to consider lowering its rating as a result of any new debt issues In a follow-up study, Kisgen (2009) examines the financing behavior of firms that experience a ratings change If firms do make capital structure decisions with credit ratings in mind, Kisgen says we should expect firms to take action in the form of capital structure adjustments following a downgrade This adjustment would be a

reasonable response to a credit rating change if the CR-CS hypothesis is correct Consistent with the CR-CS hypothesis, Kisgen finds that firms issue less debt relative to equity in the year

following a downgrade in rating, further supporting the findings in Kisgen (2006) However, no effect on firms with ratings upgrades is revealed

motivation to firm-level attributes It will be interesting to know if the CR-CS hypothesis is valid across the broad spectrum of firms or if its appeal is limited to or conditional on certain

attributes Our basic thesis is that while all firms may have some desire to maintain or achieve a certain credit rating, that desire is likely more pronounced for certain types of firms

We identify several attributes which are deemed to be correlated with a firm’s

likelihood to consider credit ratings in their capital structure decisions We then explore if the validity of the CR-CS model as revealed by a firm’s marginal financing behavior is

systematically related to these attributes If a systematic relationship is found, this would

further support previous findings suggesting credit ratings play a role in capital structure We expect firms with significant external financing needs, and thus likely to be active capital market participants, to be concerned with their credit rating more than other firms Also, firms that currently have an investment grade rating might be more interested in activities that would prevent it from becoming a "fallen angel" than firms that have already lost that status Discrete costs associated with ratings changes are expected to be especially significant for firms on either side of the investment/non-investment grade threshold (i.e., BBB- and BB+ ratings) Firms that issue commercial paper are expected to be more interested in maintaining a credit rating in order to allow it to continue to finance parts of its operations with this type of

security Alternatively, firms that do not rely on commercial paper as a financing tool might be less likely to follow this theory Finally, a firm with investment opportunities might be more interested in its credit rating than a firm with less investment opportunities for the simple

reason that an unfavorable credit rating might inhibit its ability to fund these investments

3.1 Hypothesis one: external financing needs and capital market participation

We anticipate that firms with external financing needs, and therefore likely to be

actively engaged in the capital markets, are more sensitive to the CR-CS motivation These firms are going to be more concerned with the discrete cost jumps associated with credit

ratings in view of their need to tap the capital markets for new financing The higher discrete costs associated with having to issue a bond at a higher than anticipated rate over time will be greater for a firm that has a greater need to visit the capital markets than one that relies on

internal funding sources for its financing needs In alternate form, our hypothesis is expressed

as follows:

H1: Firms that have greater external financing needs will be more likely to consider credit ratings

in their capital structure decisions than firms that do not rely as much on external capital markets for their financing needs

3.2 Hypothesis two: effect across bond ratings

The CR-CS model implicitly assumes that firms care about their broad rating

category regardless of the rating This assumes that an AA- rated firm at the risk of losing its broad rating (i.e., AA) is as likely to modify its financing behavior to protect its broad rating

as an A- rated firm However, do some ratings drive firms to react more aggressively? This may especially be the case for firms on the cusp of the investment/non-investment grade

rating We expect that a firm threatened with the loss of its investment grade designation

(i.e., BBB-) will more likely attempt to preserve its status Likewise, a firm just below the investment grade rating (i.e., BB+) will likely make changes to its financing activity at the margin to enhance the likelihood of being “bumped” to the investment grade category The discrete costs associated with a change in broad rating for these firms are likely to be more significant than for firms in other rating categories As mentioned in Cantor and Packer

(1997), regulators use credit ratings as a threshold to determine whether an institutional

investor may hold the debt of a certain company Therefore, a firm that loses investment grade status will no longer attract institutional investors, as this is a critical regulatory

hurdle The loss of institutional investors can be quite costly to a firm due to discrete costs associated with the loss of this market Kisgen (2006) cites the restrictions faced by banks in terms of their ability to hold the debt of non-investment grade firms Due to concerns about bank stability and its role in a healthy and efficient market, federal regulations do not allow banks to take these speculative positions Insurance companies, as noted by Kisgen, also face similar regulatory restrictions Once again, a portfolio with speculative bonds is not deemed appropriate for this type of business, nor for pension funds Kisgen also mentions the role credit ratings play in determining the capital requirements for broker-dealers, as dictated by the Securities and Exchange Commission

Simply put, regulatory restrictions favoring investment grade bonds imply that

bonds that risk losing their investment grade status will have to offer a higher yield beyond that associated with an increase in default and liquidity risk The higher discrete costs

associated with the loss of an investment grade status suggests that the CR-CS model is especially relevant for firms on the cusp of losing their investment grade status.2

We test the following related hypotheses with respect to the sensitivity of the CR-CS motivation to bond ratings:

H2a: CR-CS motivated financing behavior varies systematically across the various bond rating categories

H2b: CR-CS motivated financing behavior applies more strongly to firms at the

investment/non-investment grade threshold than to other ratings classes

2 For more information on regulations and credit ratings, see Kisgen (2006, p 1037-8)

3.3 Hypothesis three: commercial paper issuers

We examine the sensitivity of the CR-CS model to commercial paper issuers Regular participants in the commercial paper market are expected to be especially sensitive to

maintaining their bond ratings

Commercial paper3 is an unsecured note issued by corporations for short-term funding The maturity for this type of security is less than 270 days and the resulting funds have a variety

of uses, such as payroll and financing inventory These issues are not typically backed by any specific collateral Therefore, lenders rely on the financial strength and financial quality of the firm to signal an ability to repay this obligation Firms are interested in this type of financing because it typically costs the firm less than a bank loan Investors like commercial paper because the return is higher than the return on a U.S Treasury bill with only a marginal increase in default risk Financing through commercial paper is so popular that it exceeds Treasury bills in terms of issuance Using information provided by the Securities Industry and Financial Markets Association, Kacperczyk and Schnabl (2010) note that short-term debt financing in the U.S was approximately $5 trillion in 2007 Of this, $1.97 trillion was commercial paper while $940 billion was U.S Treasury bills.4

A firm that is financially distressed would be unable to attract investors and, in

turn, unable to borrow short-term using this low-cost type of security In addition, the rating of a firm's commercial paper has an effect on the type of investors who will supply funds For example,

a financial intermediary such as a mutual fund might be handcuffed by regulations detailing the quality of commercial paper that is suitable for investment (Cantor and Packer, 1997)

Therefore, firms that borrow short-term through the use of commercial paper should be very interested in maintaining their credit rating If the long-term rating is compromised, a firm will either lose its ability to borrow using commercial paper, may suffer liquidity issues, or may have to pay higher rates in the commercial paper market For example, in 2009

Prudential Financial Inc lost its eligibility for a US commercial paper program after Fitch downgraded their short-term debt This occurred shortly after other industry members

(Hartford Financial Services Group Inc and Genworth Financial Inc.) experienced the same fate Fitch noted that the downgrade was due to investment losses and questioned the firm’s immediate financial flexibility.5

We expect to find that firms that issue commercial paper will be more likely to make capital structure decisions with credit ratings in mind than an otherwise equivalent firm In alternate form, our testable hypothesis may be stated as:

H3: CR-CS motivated financing behavior applies more strongly to firms that issue commercial

paper compared to those that do not

3.4 Hypothesis four: growth opportunities

A firm with investment opportunities may be more interested in maintaining a credit rating than a firm that does not have the same opportunities Growth firms are more likely to have positive net present value (NPV) projects available Therefore, funding becomes a core concern for a growth company making this subset of firms a more likely visitor to the capital markets As a result, the cost of new funds will be a function of the firm’s current rating and will affect the ability to grow

We state the final hypothesis in the following alternate form:

5 http://www.bloomberg.com/apps/news?pid=newsarchive&sid=amgmS0NBYhIc&refer=home

H4: CR-CS motivated financing behavior applies more strongly to firms with greater investment growth opportunities

4 Methodology

Our test methodology is an adaptation of Kisgen (2006) Kisgen tests the basic hypothesis that firms at the edge of a broad rating category (Plus or Minus rating within a broad rating, e.g., A- or A+) will be reluctant to issue additional long-term debt The Minus rated firms are

reluctant to issue additional debt at the margin because of the potential for risking their broad rating, while the Plus rated firms do so because it would enhance their potential to move up to the next broad rating class The test is conducted by way of estimating a regression model with

debt issuance as the dependent variable (NetDIss) and a credit rating dummy variable to capture

how close the firm is to losing its current broad rating designation The dummy variable captures whether the firm has a Plus or Minus in its bond rating Kisgen runs estimates using a combined

dummy variable (POM) and one that is decomposed into Plus and Minus categories separately (Plus, Minus) In addition to these variables of interest, Kisgen also includes several control

variables The regression models along with variable definitions are as follows:

(1)

(2)

Where:

NetDIss it = ( Di,t - Ei,t)/Ai,t

D it = book long-term debt plus book short-term debt for firm at time (Compustat

data item 9 plus data item 34)

D it = long-term debt issuance minus long-term debt reduction plus changes in current

debt for firm from time to (Compustat data item 111 minus data item

114 plus data item 301)

LTD it = long-term debt issuance minus long-term debt reduction for firm from time to

(Compustat data item 111 minus data item 114)

E it = book value of shareholders’ equity for firm at time (Compustat data item 216)

E it = sale of common and preferred stock minus purchases of common and preferred

stock for firm from time to (Compustat data item 108 minus data item 115)

A it = beginning-of-year total assets for firm at time (Compustat data item 6)

Plus = dummy variable (equal to 1) for firms that have a Plus credit rating at the

beginning of the period

Minus= dummy variable for firms that have a Minus credit rating at the beginning of the

period

POM= Plus + Minus = dummy variable for firms that have a Minus or Plus credit rating

at the beginning of the period

K i,t = set of control variables, including leverage: Di,t/(Di,t + Ei,t), profitability:

EBITDAi,t/Ai,t (EBITDA is Compustat data item 13), and size: ln(Salesi,t) (Sales is Compustat data item 12).6

We adapt the above two regressions to test the sensitivity of the CR-CS model to the four attributes that comprise our four hypotheses The first hypothesis tests the sensitivity of the CR variables in the above regressions to a firm’s access and desire to access the capital markets This requires that we construct a variable to capture the firm’s need to access the capital markets We

use two proxies The first is a measure of the firm’s external financing needs (EFN) We argue

that a firm with financing needs that are unmet with available in-house capital will be concerned with their credit rating as they are likely to be seeking new financing in the capital markets

EFN, however, could be satisfied by raising funds in the debt and equity capital markets As an

alternative, we measure the firm’s reliance on the external debt market in the preceding five

years, AverageDebt This variable proxies for the firm’s reliance on the debt market for its

capital.7 We then interact these figures with the credit rating dummy variables (POM, Plus and

6 In addition to the control variables used in Kisgen (2006) we included a short term and long term treasury rate variable (3 month and 10 year US treasury rates) to control for the economically turbulent financial crisis period of 2008-2009 that is part of our extended sample period compared to Kisgen’s study period, which ended in 2001 As

it turns out, the results are qualitatively similar with and without the two treasury rate variables Consequently, in the interest of maintaining comparability with Kisgen’s (2006) study we omit the two treasury interest rates The results with the two treasury rates as additional control variables are available from the authors upon request

7 As noted by an anonymous reviewer, AverageDebt does not necessarily measure frequency and regularity of

accessing the debt markets Our measure does not differentiate between a firm that raises the same cumulative

Minus) to create new independent variables

The first set of tests is performed using the variable, EFN

We measure EFN as commonly defined in many standard corporate finance textbooks

(e.g., Brigham, Gapenski and Ehrhardt, 1999), scaled by total assets, and remove all observations

with EFN > 1, as it is unlikely that a firm’s financing needs are greater than its assets8:

b = Addition to retained earnings/Net income

The second set of regressions replaces EFN with AverageDebt As mentioned earlier,

rather than just focusing on current external financing needs, we also consider the propensity to

borrow based on recent history AverageDebt is the average annual debt issued in the previous

five years scaled by total assets

amount of debt in one year as compared to a firm that raises some debt in each of the five years At a minimum,

our measure tells us if debt capital was raised in the preceding 5 years or not So, to the extent that entering the capital markets every 5 years is an indication of “regularly” accessing capital markets, then our proxy is reasonable

8 Note that EFN can assume negative values, which simply means that the firm has an excess of funds available Our results are qualitatively the same when negative values of EFN are eliminated from the sample or replaced with

a zero value

To test if the CR-CS model is valid across the spectrum of ratings (hypothesis 2), we estimate equations (1) and (2) after separating firms into categories based on their current broad rating (AA, A, BBB,…) We also examine sensitivity to investment/non-investment rating by grouping the firms into these two classes based on their bond ratings

Hypothesis three suggests that the CR-CS model is more significant for commercial paper issuers compared to non-commercial paper issuers Empirical assessment of this

hypothesis requires that we identify regular commercial paper issuers Unfortunately, Compustat does not track commercial paper as a separate balance sheet liability account (except in the case

of utilities and financials) Therefore, we use a proxy to indicate whether a firm is actively

engaged in the commercial paper market Fortunately, Compustat reports a variable called

Standard & Poor’s Short-Term Domestic Issuer Credit Rating.9

This variable is “used for issues that have maturities of one year or less, such as commercial paper."10

To test the sensitivity of the CR-CS model to commercial paper issuers, we estimate equations (1) and (2) separately for commercial paper and for non-commercial paper issuers The former (latter) is comprised of the set of firms that have (do not have) a Short-Term Domestic Issuer Credit Rating on Compustat

As noted previously in the case of utilities and financials, Compustat provides

outstanding commercial paper balances For this set of firms we conduct an additional test to see

if the CR-CS model is affected by the relative size of the firm’s participation in the commercial

paper market The test involves estimating equations (3) and (4) but instead of EFN we substitute

9 According to Compustat this variable replaced the commercial paper rating variable after 1998:

http://emi.compustat.com/docs-mi/help/issuer_rating.htm

10 http://www.standardandpoors.com/aboutcreditratings/

CommPaper which is the amount of commercial paper outstanding at the beginning of the year

of interest, scaled by total assets If reliance on the commercial paper market is an important

determinant of CR-CS motivation as we expect it should, then the CommPaper interaction terms

with the credit rating variables should be significantly negative

The final hypothesis tests the sensitivity of the CR-CS model to firm growth

opportunities To assess the role growth opportunities play in this theory, Tobin’s Q is used The

Q ratio is a commonly used proxy for growth opportunities (e.g., Lang, Ofek and Stulz, 1996)

Tobin’s Q is calculated using the simple approximation method of Chung and Pruitt (1994)

More specifically:

where = product of a firm’s share price at the beginning of a particular year and number of shares outstanding, = the liquidating value of the firm’s outstanding preferred stock, = value of the firm’s short-term liabilities net of its short-term assets, plus the book value of the firm’s long-term debt and = book value of the total assets of the firm

We then estimate equations (3) and (4) with the growth opportunity proxy, Q, in place of

EFN

To see if there are any non-linearities, rather than interaction effects in equations (3) and

(4) we estimate equations (1) and (2) for firms separated into Q values below and above 1

4.1 Sample

The sample contains all firms listed in Compustat from 1986 until 2009 The credit rating used from Compustat is Standard & Poor’s Long-Term Domestic Issuer Credit Rating, which is the same credit rating used in Kisgen (2006) All of the data needed to estimate regression

equations (1)-(4) are available in Compustat For all data provided by Compustat, we remove any

observations that do not have a CUSIP or have a computed debt ratio that is not between 0 and 1

Table 1 provides sample descriptive statistics for the variables used in the study The average firm sales are $6.5 billion, the mean leverage is 45%, the mean profitability ratio is 14%,

and the mean Q ratio for the sample is 1.1.11

5 Results

We begin our results section with a replication of Kisgen (2006) using our updated

sample of firms covering the period 1986-2009.12 Columns (1) and (2) in Table 2 present

estimated coefficients for equation (1), the latter includes control variables Similarly, columns (3) and (4) show estimates for equation (2) The results are qualitatively similar to Kisgen

(2006).13 More specifically, POM is statistically significantly negative suggesting that firms on

the edge of their broad rating class are likely to hold back issuing debt consistent with the

predictions of the CR-CS model However, the significance of POM appears largely to be driven

by the Minus segment of the ratings categories These firms near a rating decrease issue 0.3% less debt relative to equity (column 4) The firms near a rating upgrade do not appear to

significantly adjust their debt issuing patterns, which is in contrast to Kisgen’s findings

5.1 Hypothesis one: external financing needs and capital market participation

Table 3 reports results for the proposition that firms with greater external capital needs

11 To ensure that the four hypotheses we test are independent, that is, they are all not picking up the same underlying factor, we estimated correlations between the attributes used to test the four hypotheses Specifically, we ran

pairwise correlations between EFN, AverageDebt, Investment grade dummy, Commercial paper dummy, and Q

ratio With the exception of the correlation between the commercial paper dummy and the investment grade

dummy, all remaining correlations were less than or equal to |0.13|, suggesting that the attributes are more or less independent In the case of the commercial paper dummy and the investment grade dummy, the correlation was 0.54 This is not surprising in as much as the commercial paper market is confined to high quality firms, i.e.,

investment grade rated firms The high correlation suggests that hypotheses 2 (effect across bond ratings) and 3 (effect on commercial papers issuers) may not be totally independent However, this potential lack of independence should not be significant especially when we consider the narrower set of ratings that are on either side of the investment grade threshold (BBB- and BB+)

12 Kisgen’s sample covers the period 1986-2001

13 When we restrict the sample to the same time period as in Kisgen (2006) we are unable to replicate his results We

are unable to get a significant negative coefficient for POM, though the sign was correct We attempted to contact

the author to reconcile our differential findings but were unsuccessful