A quantitative research on a firm’s change in profitability and
solvability around credit rating events
Active management of a company’s credit rating
By
Mats Poorthuis [S2969785]
Master Thesis MSc Finance Business Economics
University of Groningen
Department of Economics, Econometrics and Finance June 2017
Word count: 11395
Abstract
The gap between the reality of credit ratings and the negative image they created during the crisis triggers me to investigate the real financial effect of a credit rating change. This paper examines the effect of a change in credit rating and managers’ response to a credit rating change on profitability and solvability ratios of North-American non-financial firms between 1986 and 2015. After adjustments, a total of 421 rating changes over 270 unique firms are examined. I find that upgraded firms tend to have higher ROA, ROIC, and NPM than their downgraded counterparts over the full three-year ex-post period. ROE shows significant results up to and including quarter eight ex-post. Furthermore, upgraded firms have lower D/A ratios relative to their downgraded peer group in the medium-term and lower D/E up to and including quarter seven following the event. The variable CE/S is insignificant. I found evidence that active management drives the profitability and solvability variables of downgraded firms closer to the upgraded firms in the medium-term.
Research background and motivation
The ongoing discussion about the usefulness and credibility of credit ratings and the uncertainty of the effect of rating changes led me to investigate the impact of these ratings. The impact of a rating change on a firm’s financials interests me, especially after following the courses Corporate Valuation and Risk Management & Financial Institutions. It became clear to me that that credit ratings have impact on the value and reliability of a company.
Author: M.E. Poorthuis
Date: June 7, 2017
1. Introduction
In this paper, I examine the impact of a credit rating change on seven profitability, solvability and investment ratios. Additionally, managers’ medium-term reaction to the rating change is analysed using the same variables. A credit risk rating is a widely used tool for investors and for the entities looking for investors. Three major agency bureaus (Moody’s, Standard & Poor’s, and Fitch) earn their revenue by selling their view on creditworthiness of firms to investors. Credit rating agencies (CRAs) typically assign letter grades to indicate ratings, such as, AAA, AA, A, BBB, and so on. Triple A indicates the best creditworthiness compared to a companies’ peer group and C the worst. These ratings indicate the probability that a firm could pay back a loan, for example, bonds issued to investors. Credit ratings provide a useful risk measure for investors who lack the skills to assess creditworthiness (Opp et al., 2013).
Despite critical comments (Danielsson, 2011), the Basel Committee1 use credit ratings to
determine the default risk of a counterparty. A number of economists, financial regulators, and market operators (e.g., Moosa, 2010; Cannata and Quagliariello, 2009) criticised the role of Basel II regulations during the recent financial crisis. Although the credibility of credit ratings is questionable since 2007, these credit risk instruments remain essential to many participants in financial markets. A credit rating is still a useful tool for investors to determine extensive statistics on default rates for an entity they invest in. Several studies (see, e.g., Faulkender and Petersen, 2006; Sufi, 2009; Tang, 2009) highlight that the credit certificates facilitate companies’ access to credit markets. Furthermore, ratings affect the choice of payment method in mergers and acquisitions (Petmezas et al., 2014), and IPO pricing (An and Chan, 2008). In summary, credit risk ratings have real effect on the financial markets.
The gap between the reality of credit ratings and their reliability for market operators motivates me to investigate the real effect of rating changes. More specific, the relevance and financial effect of a credit rating change is examined in this paper; I investigate the effect of a credit rating change on a company’s profitability, solvability and capital expenditures (capex). In addition, I examine the corporate governmental decisions after a rating change by analysing financial ratios.
Several studies (see, e.g., Pettit et al., 2004; Cantor, 2004; Altman, 1968) found that credit ratings are primarily related to a company’s debt coverage ratio. The debt coverage ratio is a measure of the EBITA available to pay debt obligations. Managers do have the ability to change
1 Under Basel II, the default risk of a loan is determined by the credit rating of the company. The risk
the two components of the coverage ratio: debt and EBITA. For example, by paying off outstanding debt or increasing sales turnover. The important relation between credit ratings and the coverage ratio drives me to investigate both components. Firms with low credit ratings bear extra costs (see, e.g., Diamond, 1999; Li et al., 2004) compared to high rated companies. Therefore, a manager should aim for the most effective possible rating for the company and convince the rating agencies that the company can meet their debt and interest obligations.
Koller et al. (2015) described the debt coverage ratio as primarily indicator for a company’s credit rating. In line with his findings, the coverage ratio explains the credit rating in my data sample as well (see, graph B.1). I aim to focus on both components of the coverage ratio, rather than mimic existing studies (see, e.g., Cantor, 2004; Altman, 1968; Pettit et al., 2004) and examine the relation between the debt coverage ratio and a company’s credit rating. Following this argumentation, my paper aims to answer the following research questions:
[1] What is the medium-term effect of a credit rating change on the profitability, solvability and investment ratios of a North American company?
[2] What is the medium-term result of active management on the profitability, solvability and investment ratios of downgraded firms relative to their upgraded counterparts?
Using a dataset containing all available rating changes for non-financial North American firms between 1986 and 2015, I investigate the impact of a rating change on the medium-term profitability and solvability of these companies. I examine whether a change in a credit rating stimulates a firm’s management to maintain or repair the given rating by analysing the (1) return-on-assets, (2) return-on-equity, (3) return on invested capital, (4) capital expenditures-to-sales, (5) net profit margin, (6) debt-to-equity ratio, and (7) debt-to-assets ratio. The sample consists of 421 rating changes in 270 unique firms. 235 of these changes were downgrades and 186 upgrades. I found the required data—the rating information (S&P) and the financial ratios—on Wharton research data services (WRDS).
My paper contributes to the discussion on the usefulness of credit risk ratings and delivers evidence that managers actively aim to maintain or upgrade their credit rating using profitability and solvability components. The results suggest that CRAs credit ratings have a monitoring effect on the management of a company to increase their profitability and optimize their capital structure. CRAs facilitate firms’ access to credit markets and provide information to investors who lack the resources or skills to access credit risk.
The remainder of this paper is organized as follows. I review the literature and formulate the hypotheses in Chapter 2. In Chapter 3, I discuss the research design. The sample selection procedure and the data items used are given in Chapter 4. Chapter 5 discusses the empirical findings. I conclude and summarize the paper in Chapter 6.
2. Literature review and hypotheses
In Section 2.1, I discuss prior literature relevant to the hypotheses. The following topics are covered: the history of CRAs and their role during the recent financial crisis, the usability of credit ratings for investors, and credit ratings and their monitoring role in corporate debt financing. In Section 2.2, two studies that have similarities to this study are described. The first part (2.1) is an attempt to answer the research question from the literature, where the second section (2.2) assisted me setting up the research design. Section 2.3 describes the economic framework and Section 2.4 defines the hypotheses.
2.1 Background and prior research
Although the true value of credit ratings is questionable, the impact of credit ratings should not be underestimated. As proven by the recent financial crisis (2007), credit ratings could have
serious consequences for the financial sector (White, 2010). The first rating bureau was
introduced by John Moody in 1909 and was especially focussed on railroad bonds. After some time, several other agencies followed Moody’s example. In 1916, Poor’s Publishing Company was created, followed by Standard Statistics Company (1922), and the Fitch Publishing Company (1924). Nowadays, the three biggest agencies are Moody’s, Standard & Poor’s (S&P), and Fitch.
obligations (CDOs)2 were rated AAA, especially by the three credit rating agencies introduced above. Finding investors for CDOs which are rated AAA was not difficult. Returns were certain, the ratings indicate almost no chance of default. A portfolio of subprime mortgages could still attract investors due to AAA-ratings. Large public investors, like pension funds, invested huge amounts of money in these kinds of financial instruments. They seemed safe, but unfortunately the opposite was true. Due to conflicts of interest and the high level of competition between CRAs, several products where overrated, including CDO’s (Becker and Todd, 2011; Manso, 2013). Ultimately, when the US housing bubble collapsed, these products were not that safe after all, and investors lost large amounts of their invested capital.
Despite these views, agencies certification services remain essential to many investors. According to several studies (see, e.g., Avramov et. al., 2007; Li et al., 2004), credit ratings play an important role in the investment decisions of participants in financial markets. For example, Avramov, Chordia, Jostova and Philipov (2007) establish a strong link between momentum investing and a firms’ credit rating. Momentum investors short sell the shares of firms with a low credit rating and go long in their high rated counterparts. Their study provides evidence that momentum investors earn money with this strategy, mainly due to a decrease in net profit margins, sales growth and operating cash flows of low rated firms over the holding periods. These results suggest a positive relation between a firm’s credit rating and the feasibility of attracting share capital.
Although Avromov’s research illustrates that low credit rated firms are lucrative short positions, some investors are still willing to provide debt financing to such companies. Institutional investors invest more in share capital of low rated firms, while passive investors are more attracted to better-rated firms. Active investors are more likely to find cheap stocks
among these firms that require their financial results to progress (Farooqi et al., 2017). Passive
investors tend to invest in better performing stocks, which consist of mostly highly rated firms.
Both studies (Avromov et al., 2007; Farooqi et al., 2017) suggest that low credibility is
associated with a cheap stock, or rather, a low share price. Moreover, a rating upgrade announcement positively affects a firm’s share price (Li et al., 2004). In case of a downgrade, the long-term returns show a significant negative response. In summary, investors base their investment decisions on the creditability of the firms they invest in. Momentum investors
2 A collateralized debt obligation is an asset-backed security, where the underlying assets are fixed-income
associate low rated firms as perfect short positions, while other institutional investors see some of these low rated cheap stocks as valuable investment opportunities.
The above results suggest that a credit rating announcement plays a significant role in the possibility of attracting share capital. Less is known about the impact of a rating change on a corporate medium-term debt financing structure, focussing on the years following a rating change. There is limited guidance from prior research regarding the investment behaviour of firms that underwent a credit rating shift. Diamond (1991) states that borrowers with higher credit ratings have a lower cost of capital, and thus a higher present value of future profits. To keep this cost at a low level, the rating needs to be maintained. His findings suggest that a decrease (increase) in a firms’ credit rating results in higher (lower) cost of capital. Besides the relation between a credit rating and cost of capital, debt financing increases as the credit rating strengthens (Sufi, 2006). His empirical findings illustrate that after the introduction of bank loan ratings, there was an increase in the supply of available debt financing. Firms that obtain a bank loan rating experience an increase in their debt-to-asset ratio and leverage ratio. Sufi’s focus was on newly rated firms, whereas in this paper the effect of a rating change is examined. As discussed later, I will perform panel regressions on rating downgrades and upgrades.
In summary, rating agencies provide credit ratings that affect a company’s borrowing costs and provide certification for borrowers through rating changes. Despite its potential implications and relevance, the rating change effect on a company’s financials and active management to repair or maintain a rating is relatively unexplored.
2.2 Related literature
Two studies closely related to my study are Driss, Massoud and Roberts (2016) and Bannier,
Christian and Wiemann (2012). The former study describes the effect of a credit watch3 (with
direction downgrade) on debt financing and invested capital. They found that firms with confirmed ratings experience an increase in long-term debt financing and invest more capital. I find the research design Driss et al. (2016) used in their study useful for my paper. To investigate the effects of a rating change, they analyse the dynamic pattern of corporate outcomes on four quarters prior and following the rating watch period. The panel regressions they employed have some of the same characteristics of the panel regression I use in this study. A significant difference between this study and theirs is that the reference quarter differs. Driss
3 The credit watch is a period where credit rating agencies aim to monitor a company’s credit rating. A credit
(1) et al. set the reference quarter as the quarter ending immediately prior to the watch period, where the reference quarter in my study is the quarter where the rating change occurred.
Bannier et al. (2012) investigate whether a credit rating event affects a firms’ investment decisions. They found that rating downgrades negatively affect future investment decisions. The empirical model used by both studies is applicable for my study. The parameter of interest is the beta that measures the effect of a rating change on firms’ profitability variables. I imitate both studies in setting vector dummies equal to 1 if an upgrade occurs and 0 otherwise. I describe this in detail in Chapter 3.
My paper is different from both studies (Driss et al., 2016; Bannier et al., 2012) in several ways. Bannier et al. (2012) looked into firms’ investment decisions and Driss et al. (2016) focused mainly on five profitability factors in the short-run. I concentrate on a broader range of aspects: firms’ profitability, solvability and capital expenditures in the medium-term. Both studies found significant financial effects one year following a rating change, but the rating change effect may last longer for some variables. Furthermore, I use the book of Koller, Goedhart and Wessels (2015) as the economic framework, which described the math of value creation using most of the factors applied in my study. This is discussed below. The explanation of econometrics for finance (Brooks, 2014) is applied to set up the panel data models.
2.3 Economic framework
As discussed above, Koller’s (2015) theories about value creation are used as the economic framework in this study. To explain the ability of a company’s management to change a credit rating, it is important to understand what drives a company’s credit rating. CRAs base an important part of their credit rating analysis on financial ratios (Pettit et al., 2004). As discussed in detail below, the credit rating is primarily related to the debt coverage ratio. Coverage explains the companies’ ability to pay its debt out of its pre-tax operating cash flow if it is invested only in debt payments:
𝐷𝑒𝑏𝑡 𝐶𝑜𝑣𝑒𝑟𝑎𝑔𝑒 =𝑁𝑒𝑡 𝐷𝑒𝑏𝑡
𝐸𝐵𝐼𝑇𝐴
(2) The theoretical method to calculate profitability combines ROIC and size into a currency metric. The fundamental formula I use to explain the theoretical part of the five profitability variables in this paper is Kollers’ economic profit formula:
𝐸𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝑃𝑟𝑜𝑓𝑖𝑡 = 𝐼𝑛𝑣𝑒𝑠𝑡𝑒𝑑 𝐶𝑎𝑝𝑖𝑡𝑎𝑙×(𝑅𝑂𝐼𝐶 − 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑐𝑎𝑝𝑖𝑡𝑎𝑙)
where ROIC is the return on invested capital, which is the cumulative amount of money a company invested in its core operations. I refine the profitability analysis by including four explanatory variables: return-on-assets (ROA), return-on-equity (ROE), return on invested capital (ROIC), and net profit margin (NPM). In addition, I examine the effect of capex with the variable capex-to-sales (CE/S).
Second, to realize a required debt coverage, a manager can change the capital structure. Changing debt levels within a company will affect the debt coverage ratio (see formula (1)).
Additionally, the debt ratio partly determines the cost of capital4 in formula (2), and, therefore,
the economic profit. The two variables that explain the effect of debt financing are debt-to-equity (D/E) and debt-to-assets (D/A). The variables and their relations with formula (1) and (2) are discussed in the hypotheses section.
2.4 Hypotheses
Based on a combination between the related literature and the economic framework, I created a set of hypotheses. As mentioned above, to understand how several factors affect a credit rating, it is important to understand what a drives a company’s credit rating. Empirical evidence shows that credit ratings are primarily related to the debt coverage ratio (see, e.g., Cantor, 2004; Altman, 1968; Pettit et al., 2004). Graph B.1 (Appendix B) shows how the debt coverage ratio explains the rating differences within my sample. Furthermore, companies with high credit ratings have lower cost of capital and aim to maintain this rating (Diamond, 1991). Altogether, the empirical evidence suggests that a company that receives a downgrade attempts to increase their debt coverage ratio to restore the credit rating. A manager can increase the coverage ratio by increasing the company’s EBITA or reducing the amount of outstanding debt. The chosen variables and corresponding hypotheses are based on a firm’s ability to change the coverage
4According to Koller, the weighted average cost of capital can be calculated using the following formula:
𝑊𝐴𝐶𝐶 =𝐷
𝑉𝑘𝑑(1 − 𝑇𝑚) + 𝐸
𝑉𝑘𝑒, where 𝑘𝑑 and 𝑘𝑒 are the cost of debt and equity, respectively, 𝐷 𝑉 and
ratio and aim for a better rating. I discuss the hypotheses regarding profitability first, followed by the hypotheses about the solvability ratios.
2.5 Profitability and capital expenditures hypotheses
As mentioned above, high EBITA relative to the value of outstanding debt are related to a better credit rating. High EBITA decrease debt servicing and mean it would take fewer years to repay outstanding debt. This suggests that downgraded firms within my sample have failed to convince Standard & Poor’s that they can withstand financial stress matching the rating before the downgrade. Therefore, I expect that downgraded firms have low profitability relative to their upgraded counterparts, at the moment of the rating change.
According to formula (1), firms with low EBITA tend to have lower amounts of invested capital, higher cost of capital, or low returns on the capital invested, relative to firms with higher profitability. This is confirmed by several studies (Bannier et al., 2012; Diamond, 1991), which found that the amount of capital invested decreases and the cost of capital increases around negative rating events. At the moment of the rating change, I expect downgraded firms to have lower profitability and capex ratios than their upgraded counterparts. Besides EBITA, the profitability and capex factors are affected by other variables, such as the value of equity and assets, amount of capex, and net income. The relationship between these variables and the manager’s ability to change the profitability ratios are discussed next.
Downgraded firms want to restore their rating to reduce the cost of debt in the years following the rating change (Diamond, 1991). Therefore, companies that received a downgrade need active management to increase their credit rating. My hypotheses state that the gap between profitability and capex ratios of downgraded firms’ relative their upgraded counterparts shrinks in the medium-term due to active management. Managers have several abilities to increase their profitability and capex ratios. I follow formula (2) in my argumentation and focus on the ROIC variable. An increase in ROIC results in more economic profits. The ROIC is calculated by dividing the NOPLAT (adjusted EBITA) by the amount of invested capital. First, a company can increase their EBITA by increasing the net profit margin (reducing costs or increase the sale price). An increase in the net profit margin drives the ROIC, ROA and ROE upward. Furthermore, improving employees’ productivity can lower production costs per unit and increase a company’s EBITA.
the company’s assets. Second, reduce the amount of inventory that is needed to generate sales by improving inventory management. A just-in-time inventory strategy, for example, reduces the amount of inventory in stock and therefore decreases the amount of invested capital. Another option is improving the receivable collection. More sales turnover may not translate in higher EBITA if a company is unable to collect the account receivables. Therefore, improving the receivable collection increases EBITA and lowers the amount of invested capital.
Overall, I expect that a manager of a downgraded firm increases revenues and suspend invested capital to repair the credit rating. Besides the impact on the ROIC variable, the options discussed above also change the ROA, ROE, CE/S and NPM. Table 1 summarizes the expected variable relation with a credit rating change.
2.6 Solvability hypotheses
As mentioned above, high values of debt relative to a company’s EBITA result in a low coverage ratio. Firms with high debt ratios are at greater risk of failing to make interest and principle payments and therefore are riskier than firms with low debt ratios. Thus, I expect to find higher debt ratios for downgraded firms than for their upgraded counterparts at the moment of the rating change. Additionally, as discussed above, the value of equity declines after a rating downgrade (Li et al. 2004), driving up the D/E ratio even further.
A manager can improve the company’s leverage position after a downgrade by decreasing the amount of debt within a firm. The company could decrease their D/E ratio by increasing the value of equity throughout raising more equity or increase the company’s net profits. Issuing new equity also leads to a drop in share prices of around 3% on the announcement day (see, e.g., Eckbo et al., 1995; Smith, 1986). Although, a decline in the share price decreases the market value of equity, it leaves the book value of equity unaffected.
In summary, I expect high debt levels and low profitability factors around negative rating events and low debt levels and high profits around positive rating events. Additionally, in the medium-term I expect the gap between upgraded and downgraded companies to shrink. Upgraded firms want to maintain their rating, while downgraded firms aim for an upgrade. I summarize the variables and directional predictions in Table 1. Overall, the first hypothesis for my research is:
[1] At the moment of the rating change, there is a significant difference between the profitability, solvability and investment ratios of upgraded and downgraded firms.
In addition to the hypothesis [1], I examine the effect of active management in the mediu-term, testing the following hypothesis:
[2] The gap between profitability, solvability and investment ratios of downgraded firms’ relative to their upgraded counterparts shrinks in the medium-term due to active
management.
Table 1
Variable explanation and expected relation with the rating change Direct effect Medium-term effect
Variable Upgrade Downgrade Upgrade Downgrade Calculation
ROA + - - + Operating income before
depreciation as a percentage of firms’ total assets.
ROE + - - + Net income as a percentage of a
firms’ total book value of equity.
ROIC + - - + Net income as a fraction of
invested capital.
CE/S + - - + Property, plant and equipment
expenditures relative to total sales turnover.
NPM + - - + Net income as a percentage of
total revenue.
D/A - + + - Total liabilities as a fraction of
total assets.
D/E - + + - Total liabilities as a fraction of
shareholders equity. Direct effect indicates the expected rating change reaction of the corresponding variable in the short run (Q0-Q4). Medium-term effect shows the expected reaction in the medium-term, relative to the
short run (Q5-Q12). A positive sign means higher expected variable values in the medium-term
(3)
3. Research design
In this chapter, I focus on the general regression specification. In the regression model, I investigate the impact of a rating change on the three years following the event in comparison to the three years prior to the event. Figure 1 displays the time frame for each rating change observation between 1986 and 2015.
Figure 1
Time frame research design.
This figure displays the timeframe for the empirical research design. The ex-ante period starts twelve quarters before the event period. The twelve quarters after the event date are the ex-post period. Q0 is
the quarter in which the rating change occurred.
3.1 Regression
Using the above time frame, I examine the firm’s reaction to a rating change. To detect in which quarters the time effects are present, a pooled OLS regression with quarterly dummies is employed. I conduct separate cross-sectional regressions for each of the time periods (quarter dummies). The model is called the least squares dummy variable (LSDV) approach, which sets dummies equal to 1 for all observations in a certain quarter. For example, 4 quarters following the event data has a dummy equal to 1 for all observations in quarter 4, and 0 otherwise. Note that I removed quarter 0, the event quarter, to avoid perfect multicollinearity between the dummy variables (dummy trap). Next, to determine whether there is a negative (positive) effect following a rating downgrade (upgrade), dummy variables indicate whether a rating decreased (increased). A dummy equal to 1 indicates a rating upgrade and 0 a rating downgrade. Altogether, the basic panel regression model has the following specifications:
𝑦𝑖𝑡 = 𝛼 + ∑ 𝛽1 12 𝑞=1 𝐵𝑒𝑓𝑜𝑟𝑒𝑖𝑡𝑞×𝑈𝑝𝑔𝑟𝑎𝑑𝑒𝑖 + ∑ 𝛽2 12 𝑞=1 𝐴𝑓𝑡𝑒𝑟𝑖𝑡𝑞×𝑈𝑝𝑔𝑟𝑎𝑑𝑒𝑖 + 𝜀𝑖𝑡
where 𝑦𝑖𝑡 is a measure for each of the profitability or solvability variables, 𝐵𝑒𝑓𝑜𝑟𝑒𝑖𝑡𝑞 and
a rating upgrade if equal to 1. The regressors of interest are the interaction dummy variables
𝐵𝑒𝑓𝑜𝑟𝑒𝑖𝑡𝑞×𝑈𝑝𝑔𝑟𝑎𝑑𝑒𝑖 and 𝐴𝑓𝑡𝑒𝑟𝑖𝑡
𝑞
×𝑈𝑝𝑔𝑟𝑎𝑑𝑒𝑖. These variables disappear in case of a
downgrade, as the upgrade dummy is 0 in that case. Coefficients of these interaction dummy variables indicate the changes in explanatory variables in a relative quarter for upgraded firms relative to downgraded firms. In line with the hypotheses, I would expect a significant value for the regressor variables for the quarters following the event. Consequently, I test whether the regressor betas have non-zero values, indicating a significant effect between upgraded and
downgraded firms. The hypotheses for all seven descriptive variables of 𝑦𝑖𝑡 are:
𝐻0: 𝛽2 = 0
𝐻1: 𝛽2 ≠ 0
Following Hill, Griffiths and Lim (2012), I estimate standard errors clustered at firm and quarter level. This assumption relaxes the assumption of homoscedasticity and zero error correlation, and therefore corrects for the least squares error assumption that errors need to be homoscedastic and that all errors need to be uncorrelated (Brooks, 2014).
Side effects are removed in the regression using control variables. Following related literature (Driss et al., 2016; Bannier et al., 2012), the most important control variables are added to model (3). Appendix E shows a detailed description of all variables. First, I construct the firm size, measured as the value of total assets. According to Bannier (2012), firm size is highly significant along all explanatory variables in his outcomes. Larger firms have more financial leverage because they are expected to have lower costs of financial distress (see, e.g., Graham et al., 1998; Hovakimian et al., 2001).
Second, the market volatility variable controls for undiversifiable risk. The market volatility is measured by taking the standard deviation of the monthly stock returns on the Compustat North America value-weighted index. I used daily closing prices, over the corresponding regression quarter.
post-(5) (4) period (Li et al., 2004). Therefore, Tobin’s Q is an improper control variable in my regression and is excluded in the model.
Finally, the weighted value of property, plant and equipment (PPE) is added to model (3). Firms with high levels of PPE can provide more collateral for debts and, therefore, have better investment positions. Several studies (e.g., Titman et al., 1988; Rajan et al. 1995) find a positive relation between tangibility and financial leverage. The PPE is scaled by the companies’ total assets, and is found to be significant as a control variable in Driss’ (2016) paper.
The control variables increase the confidence level of the results, as they control for market risk, firm size, and tangibility. Hence, including these variables in the model controls for variability among rating changes. I model the control variables in combination with regression (3) with the following constructs:
𝑦𝑖𝑡 = 𝛼 + ∑ 𝛽1 12 𝑞=1 𝐵𝑒𝑓𝑜𝑟𝑒𝑖𝑡𝑞×𝑈𝑝𝑔𝑟𝑎𝑑𝑒𝑖+ ∑ 𝛽2 12 𝑞=1 𝐴𝑓𝑡𝑒𝑟𝑖𝑡𝑞×𝑈𝑝𝑔𝑟𝑎𝑑𝑒𝑖+ 𝛽3𝑉𝑜𝑙𝑎𝑡𝑖𝑙𝑖𝑡𝑦𝑖𝑡+ 𝛽4𝑆𝑖𝑧𝑒𝑖 + 𝛽5𝑇𝑎𝑛𝑔𝑖𝑏𝑖𝑙𝑖𝑡𝑦𝑖𝑡+ 𝜀𝑖𝑡
Following Brooks (2014), I need to control for time and industry effects. Therefore, model (4) is expanded by adding two dummy variables controlling for time and industry effects. The calendar effects control for variety between different quarters. It is important to remove
spurious calendar variation to increase the significance of the regression outcomes (Cleveland
et al., 1980). The calendar quarter dummy variable equals 1 if the research quarter corresponds to the calendar quarter, and 0 otherwise. Quarter 1 is excluded to avoid perfect multicollinearity.
Next, the industry fixed effects control for differences across industries. Schröder and Yim (2016) highlight the importance of adding industry effects to the model. Their paper find significant differences in profitability and financial performance between industries, rather than pooling all industries together in an economy-wide model. I use the global industry classification standards (GICS) as industry fixed effects sector groups. Eleven dummy variables are added to model (4) to control for industry fixed effects. Energy sectors are excluded to control for multicollinearity. The final regression model has the following characteristics:
Regression (5) is a widely used model in academic literature (e.g., Driss, 2016; Betreand et al., 2003; Schoar, 2002; Chemmanur et al., 2010) to examine the effect of an event on firms’ behaviour. Therefore, I expect the model to give robust results.
4. Data description
In this chapter I discuss the research area and time frame first, followed by the data and sample selection procedure (Section 4.1) and data specifications (Section 4.2).
To collect the financial information as inputs for model (5), several data sources are consulted. I use the rating information of Standard & Poor’s to test the hypotheses, for two reasons. First, S&P are the world’s leading provider of credit ratings (Sylla, 2002). Second, ratings between different rating agencies do not deviate much. As mentioned above, S&P rating agency have been in business for over 100 years.
Furthermore, the time frame (Figure 1) is based on the availability of financial data. The digitalisation of the credit ratings started in 1986. Financial ratio data is available until 2015. Thus, the timeframe for this research is 30 years (1986–2015).
The three major CRAs started their activities in North America and expanded from there. Additionally, the North American bond market is one of the largest and has the most issuers in the world. This is especially due to the long-term creditworthiness of North America. Countries that receive a downgrade in their creditability have negative externalities for the private sector and real economic activity (Almeida et al., 2014). This confounding factor is filtered out; the countries (USA and Canada) in my data sample have a high and constant rating over the event period. Based on the data availability and high sovereign credit rating, I find North America as a reliable research area.
4.1 Data and sample selection
Compustat collects and publishes quarterly reports on firms’ long-term credit ratings. I aggregated the quarterly credit ratings from all North American corporations available on the Compustat database, obtaining more than three million individual rating observations. As mentioned above, financial institutions are excluded from this dataset. To examine the
medium-term effect of a rating change, three ‘clean5’ years prior and following a rating change are
5 By clean I mean no rating upgrades or downgrades in the twelve quarters following the event date. If there
required. Therefore, I included only those firms that do not have any rating change activity in twelve quarters prior to and twelve quarters after the event date. Only firms that have data available for each quarter within the sample are included.
This method6 results in eliminating the observations that do not satisfy the requirements.
Each observation is linked to the explanatory variables. Observations with missing variables are deleted from the sample. Next, I follow Driss (2016) and mitigate the effect of outliers. All variables each quarter that fell outside the range of 0.1st and 99.9th percentile are excluded. Although I deleted some extreme outliers, variables have non-normal distributions. I employed an estimation method, which assumes normality. Therefore, I appeal to the central limit theorem, which states that the test statistics will asymptotically follow the appropriate distribution even in the absence of normality (Brooks, 2014). Finally, I delete firms with (near) default ratings (RD or worse), to control for extreme variable values. Appendix A shows the excluded variables, Table 2 represents the selection procedure.
In summary, the sample consists of 10525 rating observations, of which 421 were the rating changes and the remainder were the ex-ante and ex-post information. In the data sample, some events occurred in the same firm; 270 distinct firms were identified that meet the requirements.
4.2 Specifications
From WRDS, I exported relevant financial ratios. The database created financial ratios by combining different data sources. Accounting related data were obtained from Compustat, earnings related data from the EBIS database, and pricing related data from CRSP. Table 3 shows the descriptive statistics of the data sample described in the previous section. There are
6 I created a model that recognized a rating change in the history of each entity. In the sample that was left, the
model identified series with twelve of the same ratings prior and twelve identical ratings after the event date. Table 2
Sample selection procedure
Variable Obs. No. firms
Total non-financial North-American companies 3023585 Less: observations that did not fit the time frame (2995635)
Total rating changes that satisfied time frame assumptions 27950 1118 Less: incomplete information / missing variables (17200) (688) Total complete rating changes with full information available 10750 430
Less: 0.2% outliers (75) (3)
Total observations between 0.1st and 99.9th percentile 10675 427
Less: firms with RD rating or worse (150) (6)
different sample sizes per variable, indicating that not all the required data is available on WRDS. Therefore, I employed an unbalance panel regression, which is discussed in detail below.
Graphs displaying summary sample statistics can be found in Appendix B. The majority of the firms within the data sample are rated A or BBB. One firm scored CCC, and six firms rated AAA. Figure B.3 shows the number of observations over the sample period. As can be seen, the number of firms rated increased in the first years of issuing ratings and steadily decreased over time. It should be noted that the dot-com bubble and the recent financial crisis are included in the time frame. This may negatively affect some ratios within the periods of financial recession. During periods of financial crisis (1999 and 2008), the number of downgrades relative to upgrades is clearly noticeable.
The sample is divided by credit rating change: 186 upgrades, and 245 downgrades. On average, upgraded firms in my sample have total assets of around $26 billion. The value is relatively high compared to the North American average, caused by the data selection criteria
of WRDS.7 The average quarterly ROA is 16.9% for upgraded firms and 14.6% for downgraded
ones.
7 WRDS industry financial ratios is a collection of 70 most commonly used financial ratios by researchers. The
database classified ratios on a universe and industry selection and left irrelevant information out of the database.
Table 3
Descriptive statistics (mean, median, standard deviation, and number of observations) of firms included in the research sample, as well as the corresponding profitability and solvability ratios
Upgrade Downgrade
Variable N Mean Median Std. dev. N Mean Median Std.
dev. ROE 4500 0.147 0.136 0.150 5375 0.123 0.113 0.146 ROA 4625 0.169 0.164 0.075 5675 0.146 0.142 0.070 ROIC 4300 0.134 0.121 0.100 5375 0.114 0.103 0.082 CE/S 3345 0.508 0.375 0.510 4644 0.476 0.351 0.559 NPM 4400 0.074 0.067 0.078 5525 0.052 0.057 0.127 D/A 4300 0.274 0.260 0.143 5725 0.316 0.305 0.139 D/E 4625 1.674 1.284 1.615 5800 2.080 1.642 2.291 Size 4650 26274 6442 50073 5875 20656 5681 40794 Volatility 4650 0.001 0.000 0.009 5875 0.001 0.000 0.010 Tangibility 4650 0.410 0.374 0.253 5842 0.412 0.372 0.242
According to Koller et al. (2015), the variables in Table 3 are common ratios. Table 3 shows that the mean value of ROA is higher than ROE. This difference is due to the WRDS financial ratios calculations. ROA is calculated using the net operating profits, where the net income is used to determine ROE (Appendix E). Also according to Koller (2015), financial institutions consider ROE ratios between 0.15-0.20 as representing attractive investment projects. The ROE in both sub samples is slightly below the attractive range. As expected, downgraded companies have higher debt levels than upgraded companies. In summary, the data sample used in this paper corresponds to North American market averages and gives a clear representation of the real world.
5. Empirical results
In this chapter, the empirical results are discussed. First, the data is analysed based on graphs where quarterly changes in the profitability variables are plotted over time (Appendix C). I explain the effect of a rating change on corporate outcomes based on seven graphs. Second, the empirical results are explained in detail.
5.1 Rating change graphs
Appendix C suggests that the profitability and capex variables behave as expected in the hypotheses. Quarter zero indicates the quarter in which the rating change occurred. All profitability variables of upgraded firms outperform their downgraded counterparts in the post-watch period. ROIC and ROE show significant higher values over the full time frame, whereas ROA and NPM especially distinguish in the short-run. Additionally, graph C.4 shows the CE/S plot. The graph suggests that upgraded firms have higher CE/S ratio at the moment of the rating change, and from quarter 2 up to quarter 8. From the graphs, a clear outperforming pattern is detected; firms that received an upgrade performed better in the event quarter than downgraded companies. The profitability and capex ratio plots suggest that the gap between upgraded and downgraded firms shrinks in the medium-term for all variables. In the CE/S variable a seasonality pattern is detected. I control for this cyclic movement in the regression model by using quarter dummy variables (see, Chapter 3).
suggests that there is a interpretable relation between decreasing capex and lower debt levels.8 In addition, upgraded firms show an increase in the D/A ratio and at the same time an increase in CE/S. In contrast to the D/A ratio, the D/E variable suggest that a firm’s management do not aim for lower debt levels after a downgrade. Graph C.7 shows that D/E increases for downgraded firms in the medium-term.
In summary, the graphs show a clear difference between upgraded and downgraded firms in the event quarter. In the medium-term the gap between upgraded and downgraded firms shrinks (except for the D/E ratio), illustrating a manager’s attempt to increase a company’s credit rating after a downgrade. The graph structure of upgraded firms is flatter, perhaps because upgraded firms want to maintain their credit rating. In the next section, the regression results are discussed.
5.2 Quarterly firm outcomes
Table 4 shows the results of quarterly firm outcomes. Consistent with my hypotheses, firms that received a rating upgrade have higher values of their return-on-equity, assets and invested capital, as well as the capex and net profit margin in the ex-post period relative to downgraded firms in quarter 1. However, CE/S gives insignificant results. The solvability ratios show a similarly expected effect in my models. D/A and D/E both display a negative sign, where I predicted that downgraded firms would increase their debt investing in the ex-post period relative to upgraded firms. Firm size and tangibility are significant for all variables, while market volatility is only significant for D/A. In the next sections, the firm outcomes and investment decisions are discussed in detail.
5.3 Profitability of upgraded firms
Table 4 presents evidence that 𝛽2 is not equal to zero over the ex-post period for the profitability
variables ROA, NPM and ROIC. ROA and NPM are significant at a 1% level in the medium-term, while ROIC shows slightly less significant outcomes at a 10% level. The ROE variable is only significant in the short run. ROE shows significant results at a 5% level up to and including quarter 8. Table 4 shows insignificant results for CE/S for all quarters. Although Graph C.4 suggest that the CE/S decreases for downgraded firms relative to their upgraded
counterparts, the model results do not provide significant evidence that 𝛽2 is different from zero.
The results show a turning point for the patterns ROE, ROA, ROIC and NPM after the second post-event quarter. Hence, the gap between upgraded and downgraded firms increases in the first half a year following the rating change and shrinks from quarter three onward. These patterns suggest that managers intervene to repair their company’s downgraded rating.
The profitability variables appear to have explainable signs. They all have positive values, so that a rating upgrade leads to higher return on assets, equity and invested capital and an
Table 4
Tests of a rating change effect on individual quarters with the least dummy variables model
Variable ROE ROA CE/S ROIC NPM D/A D/E
Intercept 0.1103 *** 0.1074 *** 0.8455 *** 0.1103 *** 0.0422 *** 0.3503 *** 2.7391 *** Before12 x Upgrade -0.0048 0.0062 0.0046 0.0064 0.0052 -0.0142 -0.1152 Before11 x Upgrade -0.0063 0.0041 -0.0402 0.0041 0.0047 -0.0167 -0.1257 Before10 x Upgrade -0.0072 0.0046 -0.0141 0.0034 0.0048 -0.0200 * -0.1383 Before9 x Upgrade -0.0041 0.0060 -0.0497 0.0036 0.0076 -0.0222 * -0.1624 Before8 x Upgrade 0.0048 0.0091 * -0.0144 0.0057 0.0101 -0.0261 ** -0.2063 * Before7 x Upgrade 0.0125 0.0125 ** -0.0232 0.0131 * 0.0134 ** -0.0267 ** -0.1817 Before6 x Upgrade 0.0126 0.0128 ** -0.0335 0.0119 * 0.0158 *** -0.0339 *** -0.2484 * Before5 x Upgrade 0.0150 0.0140 *** -0.0432 0.0146 ** 0.0182 *** -0.0367 *** -0.2288 * Before4 x Upgrade 0.0065 0.0141 *** -0.0357 0.0105 0.0207 *** -0.0384 *** -0.2792 ** Before3 x Upgrade 0.0132 0.0170 *** -0.0349 0.0182 * 0.0249 *** -0.0394 *** -0.3292 *** Before2 x Upgrade 0.0151 0.0186 *** 0.0340 0.0186 ** 0.0259 *** -0.0437 *** -0.3562 *** Before1 x Upgrade 0.0190 * 0.0219 *** 0.0124 0.0200 *** 0.0283 *** -0.0419 *** -0.3469 *** After1 x Upgrade 0.0284 *** 0.0269 *** -0.0039 0.0233 *** 0.0284 *** -0.0405 *** -0.3452 *** After2 x Upgrade 0.0668 *** 0.0383 *** -0.0203 0.0429 *** 0.0384 *** -0.0597 *** -0.5509 *** After3 x Upgrade 0.0633 *** 0.0367 *** 0.0270 0.0416 *** 0.0362 *** -0.0682 *** -0.5531 *** After4 x Upgrade 0.0555 *** 0.0328 *** 0.0551 0.0391 *** 0.0303 *** -0.0664 *** -0.5756 *** After5 x Upgrade 0.0295 *** 0.0202 *** 0.0389 0.0225 *** 0.0192 *** -0.0362 *** -0.3597 *** After6 x Upgrade 0.0267 ** 0.0153 *** 0.0317 0.0199 ** 0.0183 *** -0.0352 *** -0.3527 *** After7 x Upgrade 0.0250 ** 0.0156 *** 0.0279 0.0208 *** 0.0187 *** -0.0332 *** -0.3411 *** After8 x Upgrade 0.0224 ** 0.0132 *** 0.0393 0.0190 *** 0.0191 *** -0.0304 *** -0.2365 After9 x Upgrade 0.0148 0.0142 *** -0.0039 0.0140 ** 0.0152 *** -0.0295 *** -0.2352 * After10 x Upgrade 0.0147 0.0137 *** -0.0113 0.0136 * 0.0148 *** -0.0295 *** -0.2045 After11 x Upgrade 0.0192 ** 0.0134 ** 0.0304 0.0124 * 0.0153 *** -0.0288 *** -0.1706 After12 x Upgrade 0.0219 ** 0.0150 *** 0.0284 0.0159 ** 0.0155 *** -0.0282 *** -0.1698 Market volatility 0.0420 0.0612 0.1385 0.0977 0.0924 -0.2911 ** -1.7532 Tangibility -0.0668 *** 0.0581 *** 0.3950 *** -0.0461 *** 0.0127 ** 0.0212 *** -1.1834 *** Firm size 0.0000 *** 0.0000 *** 0.0000 0.0000 *** 0.0000 *** 0.0000 *** 0.0000 *** Observations 9850 10275 7989 9650 9892 10017 10392 No. Of companies 394 411 323 386 396 401 416 Adjusted R2 0.1256 0.1676 0.1485 0.1565 0.0722 0.1157 0.0495
increase in the net profit margin, relative to a rating downgrade. The results suggest that upgraded firms increase their profitability ratios around positive rating events.
Variables ROE and ROA show higher values for upgraded firms relative to their downgraded counterparts. The gap between upgraded and downgraded firms for the variables ROE and ROA increased to 6.7% and 3.8% at the end of the second quarter. The high increase in ROE relative to ROA is probably due to a decreasing value of equity, where the value of assets decreased less or stayed constant. Next, I imitate Campbell and Shiller (1998) and plot regressions between the profitability ratios and each of their components determining the value of the ratio to explain which of the variables drives the ratio. According to the graphs in Appendix D, the ROE and ROA variable are especially driven by the profit growth ratios, rather than the growth rate of equity or assets. The scatterplots suggest that the results of both variables in Table 4 are driven by EBITA.
In the second quarter, the variables NPM and ROIC of upgraded firms outperform their downgraded counterparts by 3.8% and 4.3% respectively. The insignificant results of CE/S suggest that there is no significant difference in capex over the post-event period. These results in combination with the scatterplots suggest that the ROIC is determined by an increase in EBITA for upgraded firms relative to their downgraded counterparts, up to and including the second quarter. In the medium-term, the gap between both samples decreases, suggesting an attempt to increase the credit rating for downgraded companies.
5.4 Debt ratio models
The results in Table 4 show significant evidence to reject the null hypotheses of 𝛽2 equal to
zero for the D/A ratio at a 1% level. The regression output reports significant results for the D/E ratio up to and including quarter seven post-event. The debt-to-assets ratios for upgraded firms outperform the downgraded counterparts in the medium-term, where the debt-to-equity ratios for the same firms outperform until quarter eight.
Both debt ratios have negative signs. This means that the D/E and D/A ratio is higher for downgraded firms relative to their upgraded counterparts over the entire time frame. This confirms my hypotheses that the debt levels are higher for firms with a low rating than ones with high ratings. After analysing the scatter plots in Appendix D, it remains unclear which of the components the debt ratios are driven by. All components of the D/A and D/E ratios (debt, asset and equity growth) show flat structures when plotted against the debt ratios.
5.5 Rating watch period
My analyses so far focussed on the prevalent monitoring process on the point in time when the company receives a rating change. In contrast to Driss’ (2016) findings, there is no clear turning point for the patterns in quarter 0. Some results are already significant before a rating event occurred. The possible explanation for the significant results in the ex-ante period is that rating agencies’ supervision starts before a rating change. This supervision is called a rating watch period and illustrates S&P’s opinion regarding the potential direction of a credit rating.
The watch period is a possible explanation for the significant results prior to a rating event. Following the results of Driss (2016), who focussed especially on a rating watch period, the average watch period consists of 141 days for confirmed firms and 93 days for non-confirmed firms. Confirmed firms are firms that remain the same rating as before the watch period and non-confirmed firms did receive a rating downgrade following the watch period. Shifting the event quarter in Table 4 one period back in time, the major significant results are still found in the ex-ante period. The ROA, NPM and D/A ratios still found significant results prior to the new event quarter. Quantitative analysis is an important part of the evaluation of a credit (Pettit et al., 2004), and the results suggest that S&P utilize ROA, NPM and D/A in determining a company’s financial risk.
5.6 Robustness check I: extraordinary rating changes
In my panel data, two types of rating events occurred. Namely, ratings that changed one rating notch and ratings that changed more than one notch. External events are more extreme in nature and lead to larger rating changes. In the scenario that profitability and solvability variables are affected by active management the sub sample containing one notch dummies should show significant results.
I employ a panel regression with dummy variables for large rating changes. The dummy has a value of 0 for one-notch rating changes and 1 otherwise. Seven models are tested, with the profitability and solvability factors as explanatory variables, the dummies for rating changes as descriptive variables, and the volatility factor. If the variables are indeed determined by external events rather than monitoring, only the large rating change dummies should show significant effects.
Table A.5 shows the results of the panel regression. The one notch rating change dummies show significant results for the variables ROE, ROA, ROIC, NPM and D/A. The more than one notch rating changes show insignificant results for ROE, ROIC, and NPM. The results suggest that the ex-post regressions outcomes are a result of management actions rather than external events for these variables. ROA and D/A are significant for both samples, suggesting that severe external events appear to impact these variables, but they do not entirely explain the explanatory variable response. In contrast to foregoing, the results in Table A.5 suggest that D/E is driven by external events rather than active management. The results for this variable are significant for more than one notch changes and insignificant for the one notch sample. These outcomes are a possible explanation for the increasing D/E ratios for downgraded firms in graph C.7, compared to the decreasing D/A ratios for the same group. Active management results in declining D/A ratios for downgraded firms in the medium-term, whether external events explain the movements in the D/E variable.
Table A.5 further illustrates that managers aim to repair a one notch rating downgrade rather than a more than one notch downgrade. In the medium-term, the small rating changes show significant results for all variables, except CE/S and D/E. Hence, active management is detected to repair small downgrades, yet for larger drops caused by external events this is not the case.
5.7 Robustness check II: panel unit root test
A valid concern with my panel regression results is that the rating change differences in the time frame period are stationary. If the data is stationary, then the covariance between two values only depends on the length of time between those two values. To address this concern, I test the panel data for stationarity assumption underlying the panel regression framework (Brooks, 2014). The assumption implies that there are no periodic fluctuations, such as seasonality, and a that the data has a constant autocorrelation structure. Overall, the panel unit root test ensures that the data is not influenced by linear trends.
To test for stationarity, an augmented Dickey-Fuller test is employed. Table A.6 shows the results. The probabilities for Fischer tests are computed using an asymptotic Chi-square distribution. The null hypotheses of unit root is rejected at a 1% significance level for all explanatory variables. The test statistics are well above critical values in all cases, indicating that the series do not contain unit roots. Thus, all profitability and solvability variables I use in the model regressions are stationary and the assumption of stationary data is satisfied.
6. Conclusion and discussion
6.1 Conclusion
In this thesis, I examine the rating change effect on a firm’s profitability and solvability in the medium-term and the mangers response to a rating change. Using financial data from WRDS and Compustat, I employ seven panel data regressions. The hypotheses for my research state that at the moment of the rating change, there is a significant difference between the profitability, solvability and capex ratios of upgraded and downgraded firms, where the gap of downgraded firms’ relative their upgraded counterparts shrinks in the medium-term due to active management.
At the moment of the rating change, the variables ROA, ROE, ROIC, NPM, D/A and D/E show significant differences between upgraded and downgraded firms. For these six variables the null hypotheses that there is no significant difference between profitability and solvability is rejected. Hence, the profitability variables show significant higher results for upgraded firms than for their downgraded counterparts, and upgraded firms show lower solvability ratios than downgraded ones. The CE/S ratio is found insignificant at the moment of the rating change. Therefore, the hypothesis that the difference between the capex-to-sales for upgraded firms relative to downgraded firms at the moment of the rating change is rejected.
results for the one notch rating change dummy variables over the medium-term period, illustrating an active management effect. Furthermore, one notch rating change regression output shows significant differences for ROA, ROIC, NPM and D/A between upgraded and downgraded firms over the full medium-term period. The ROE variable is not significant over the whole ex-post period. Hence, the hypothesis that the gap between profitability and solvability ratios of downgraded firms relative to their upgraded counterparts shrinks in the medium-term due to active management is fully accepted for the variables ROA, ROIC, NPM and D/A. Besides, external events determine the D/E ratio and no evidence is found that this variable is affected by active management. Finally, the CE/S shows insignificant results over the whole ex-post period. Hence, I cannot accept the hypothesis for the variables ROE, D/E and CE/S.
In summary, upgraded firms have higher ROA, ROIC and NPM in the medium-term. The gap between these variables is decreased in the medium-term due to active management. In addition, the debt ratio variable D/A shows significant lower values for upgraded firms than for their downgraded counterparts. This variable is also confirmed to be affected by active management in the medium-term period. Overall, the results suggest that management of downgraded firms decrease their D/A ratio relative to upgraded firms and increase the profitability ratios ROA, ROIC and NPM. Both components of the debt coverage ratio are therefore affected with the intention to change a North American company’s S&P credit rating.
6.2 Discussion
The results contribute to the knowledge about policy implications. The regression results suggest that firms change their corporate governmental decisions around rating change events. The variables ROA, ROIC, NPM and D/A outcomes are expected, following Driss’ (2016) research. Where their study found that rating watch periods monitor mangers in changing their financing behaviour, I found evidence that managers aim to repair their rating after a rating change occurred. Hence, in the period following the recent financial crisis, credit ratings are still a useful tool to monitor managers. They facilitate firms’ access to credit markets, and provide information to investors who lack the resources or skills to access credit risk.
6.3 Limitations and future research
However, the results suggest that the profits are higher and debt financing lower for upgraded firms, there are several other factors that drive the ratios tested in my models. For example, the value of assets and equity are important components in determining the debt ratios. In addition, although I attempt to be comprehensive in collecting rating information from 1986 to 2015, it is still possible that some data was not available on Compustat. The amount of data over which I construct the test is further limited due to the time frame restrictions. The relative long time frame of six years and the requirement that no other rating changes occur during this period excludes a lot of financial data. Depending on the systematic characteristics of the excluded firms, this could affect my results. With the aim of improving a company’s credit rating, future research might attempt to explain the optimal type of corporate governance to increase a firms rating.
References
[1]
Almeida, H., Cunha, I., Ferreira, M., Restrepo, F., 2014. The real effects of sovereign credit rating downgrades. Unpublished working paper. University of Illinois at Urbana Champaign, Illinois.
[2] Altman, I., 1968. Financial ratios, discriminant analysis and the prediction of corporate
bankruptcy. The journal of finance 23.4, 589-609.
[3] An, H., Chan, K.C., 2008. Credit ratings and IPO pricing. Journal of Corporate Finance
14, 584–595.
[4] Avramov, D., Chordia, T., Jostova, G., Philipov, A., 2007. Momentum and credit
rating. The Journal of Finance 62.5, 2503-2520.
[5] Bannier, C., Christian W., and Wiemann, M., 2012. Do credit ratings affect firm
investments? The monitoring role of rating agencies.
[6] Becker, B., Todd, M., 2011. How did increased competition affect credit ratings?
Journal of Financial Economics 101.3, 493-514.
[7] Brooks, C., 2014. Introductory Econometrics for Finance. Cambridge University Press,
Glasgow.
[8] Campbell, Y., Shiller, J., 1998. Valuation ratios and the long-run stock market outlook.
The Journal of Portfolio Management 24.2, 11-26.
[9] Cannata, F., Quagliariello, M., 2009. The role of Basel II in the subprime financial
crisis: guilty or not guilty?
[10] Cantor, R., 2004. An introduction to recent research on credit ratings. Journal of
Banking & Finance 28.11, 2565-2573.
[11] Chemmanur, J., Hu, G., Huang, J., 2010. The role of institutional investors in initial
public offerings. Review of Financial Studies 23.12, 4496-4540.
[12]
Cleveland, S., Devlin, J., 1980. Calendar effects in monthly time series: detection by spectrum analysis and graphical methods. Journal of the American Statistical
Association 75.371, 487-496.
[13] Danielsson, J., 2011. An academic response to Basel II. Special Paper-LSE Financial
Markets Group, London.
[14] Diamond, D., 1991. Monitoring and reputation: The choice between bank loans and
directly placed debt. Journal of political Economy 99.4, 689-721.
[15] Driss, H., Massoud, N., Roberts, G., 2016. Are credit rating agencies still relevant?
Evidence on certification from Moody's credit watches. Journal of Corporate Finance.
[16] Eckbo, E., Masulis, W., 1995. Seasoned equity offerings: A survey. Handbooks in
operations research and management science 9, 1017-1072.
[17] Farooqi, J., Jory, S., Ngo, T., 2017. Institutional investors’ activism and credit ratings.
Journal of Economics and Finance 41.1, 51-77.
[18] Faulkender, M., Petersen, M.A., 2006. Does the source of capital affect capital
structure? Review of financial studies 19, 45–79.
[19] Graham, R., Harvey, R., 2001. The theory and practice of corporate finance: evidence
[20] Healy, M., Palepu, G., 1988. Earnings information conveyed by dividend initiations and
omissions. Journal of financial Economics 21.2, 149-175.
[21] Hill, G., Griffiths, W., Lim, G., 2012. Principles of Econometrics. Wiley, Hoboken.
[22] Hovakimian, A., Opler, T., Titman, S., 2001. The debt-equity choice. Journal of
Finance. Quant. Anal. 36, 1–24.
[23]
Kim, J., Song, B., Zhang, L., 2011. Internal control weakness and bank loan contracting: Evidence from SOX Section 404 disclosures. The Accounting Review 86.4, 1157-1188.
[24] Koller, T., Goedhard, M., Wessels, D., 2015. Valuation: measuring and managing the
value of companies. McKinsey & Company, Cornwall.
[25] Li, H., Visaltanachoti, N., Kesayan, P., 2004. Effects of credit rating announcements:
The Swedish Stock Market. International Journal of Finance 16.1.
[26] Manso, G., 2013. Feedback effects of credit ratings. Journal of Financial Economics
109.2, 535-548.
[27] Mikkelson, H., Partch, M., 1986. Valuation effects of security offerings and the
issuance process. Journal of Financial Economics 15.1, 31-60.
[28] Moosa, I, 2010. Basel II as a casualty of the global financial crisis. Journal of Banking
Regulation 11.2, 95-114.
[29] Opp, C., Marcus, M., Harris, M., 2013. Rating agencies in the face of regulation.
Journal of Financial Economics 108.1, 46-61.
[30]
Petmezas, D., Karampatsas, N., Travlos, N.G., 2014. Credit ratings and the choice of payment method in mergers and acquisitions. Journal of Corporate Finance 25, 474– 493.
[31] Pettit, J., Fitt, C., Orlov, S., Kalsekar, A., 2004.The New World of Credit Ratings. UBS
research report.
[32] Rajan, G., & Zingales, L., 1995. What do we know about capital structure? Some
evidence from international data. The journal of Finance 50.5, 1421-1460.
[33] Schoar, A., 2002. Effects of corporate diversification on productivity. The Journal of
Finance 57.6, 2379-2403.
[34]
Schröder, D., Yim, A., 2016. Industry Effects in Firm and Segment Profitability Forecasting: Corporate Diversification, Industry Classification, and Estimation Reliability. Unpublished working paper. University of Londen.
[35] Sylla, R., 2002. An historical primer on the business of credit rating. In Ratings, rating
agencies and the global financial system, 19-40. Springer US.
[36] Smith, W., 1986. Investment banking and the capital acquisition process. Journal of
Financial Economics 15.1, 3-29.
[37] Sufi, A., 2009. The real effects of debt certification: evidence from the introduction of
bank loan ratings. Review of financial studies 22, 1659–1691.
[38]
[39] Titman, S., Wessels, R., 1988. The determinants of capital structure choice. The Journal
of finance 43.1, 1-19.
[40] White, L., 2010. Markets: The credit rating agencies. The Journal of Economic
Appendix A: tables
Table A.1
S&P Ratings with corresponding numbers Number Rating 1 AAA 2 AA+ 3 AA 4 AA- 5 A+ 6 A 7 A- 8 BBB+ 9 BBB 10 BBB- 11 BB+ 12 BB 13 BB- 14 B+ 15 B 16 B- 17 CCC+ 18 CCC 19 CCC- 20 CC 21 C 22 RD 23 SD 24 D Table A.2
Variables excluded from sample
International firm code Name Out of sample variables
2448 Materion Corp. D/E
2807 Casey's General Stores Inc ROIC, D/E
2950 Chattem Inc D/E
3121 Clorox Co D/E
5639 Hill-Rom Holdings, Inc. D/E, D/A
9155 Rite Aid Comp. D/E
10757 UNS Energy Corp. D/E
10795 United Continental HLDGS Inc. D/E
10443 Tenneco Inc. all
12785 Pilgrim's Pride Corporation all
9401 - default 7980 - default 7007 - default 5187 - default 3041 - default 4313 - default
