Corporate financing and anticipated credit rating changes

Abstract

Firm circumstances change but rating agencies may not make timely revisions to their ratings, thereby increasing information asymmetry between firms and the market. We examine whether firms time the securities market before a credit rating agency publicly reveals its decision to change a firm’s credit rating. Using quarterly data, we show that firms adjust their financing structures before credit rating downgrades are publicly revealed. Specifically, firms on average increase their debt financing by 1.29 % before the disclosure of a rating downgrade, and this increase is due to the issuance of debt rather than the repurchase of equity. In contrast, firms do not take significant financing actions before credit rating upgrades.

Appendix

1.1 Firm action variables

In our analysis we examine the effects on changes in debt, equity and net debt for firm i in quarter t defined as follows:

${\triangle det}_{i,t}=\frac{{\Delta D}_{i,t}}{A_{i,t-1}}$: debt change, where $\Delta$ ${D}_{i,t}$ is long-term debt increase (Compustat DLTISY)¹⁴ minus long-term debt reduction (Compustat DLTRY) plus the change in current debt (Compustat DLCCHY) for firm i in quarter t, and $A_{i,t-1}$ is total asset (Compustat ATQ) of firm i in quarter $t-1$.

${\triangle eqt}_{i,t}=\frac{{\Delta E}_{i,t}}{A_{i,t-1}}$: equity change, where $\Delta$ ${E}_{i,t}$ is the sale of common and preferred stock (Compustat SSTKY) minus purchases of common and preferred stock (Compustat PRSTKCY) for firm i in quarter t.

We also analyze net debt change [as in Kisgen (2006)] as the difference between ${\Delta }{det}_{i,t}$ and ${\Delta eqt}_{i,t}$, defined as $\Delta$ ${net}_{i,t}=\frac{{{\Delta D}_{i,t}-\Delta E}_{i,t}}{A_{i,t-1}}$ .

We further look into details of debt changes by examining the effects on short term and long-term debt, respectively.

${\triangle Sdet}_{i,t}=\frac{{\triangle SD}_{i,t}}{A_{i,t-1}}$, where ${\triangle SD}_{i,t}$ is the change in current debt (Compustat DLCCHY) for firm i in quarter t.

${\triangle Ldet}_{i,t}=\frac{{\triangle LD}_{i,t}}{A_{i,t-1}}$, where ${\triangle LD}_{i,t}$ is the long-term debt increase (Compustat DLTISY) minus long-term debt reduction (Compustat DLTRY) for firm i in quarter t.

1.2 State variables

We include the control variables (${{\mathbf {X}}}_{i,t}$) which are conventionally considered in capital structure studies including: Leverage, Size, Price, Liquidity, Profit, Earnings, Growth, Tangibility and Non-Debt Tax Shields (NDTS) to separate their influences from the role of information gap on firms’ financing activities.¹⁵

${Leverage}_{i,t}$: the ratio of the sum of short-term debt (Sdet) (Compustat DLCQ) and long-term debt (Ldet) (Compustat DLTTQ) to the sum of short-term debt, long-term debt and stockholders’ equity (Compust LSEQ minus LTQ) for firm i in quarter t.

$Size_{i,t}$: the logarithm of sales (Compustat SALEQ) for firm i in quarter t.

${Price}_{i,t}$: the logarithm of the stock’s quarterly closing price in the quarter (Compustat PRCCQ) for firm i in quarter t.

${Liquidity}_{i,t}$: the ratio of cash and cash equivalent (Compustat CHEQ) to total assets (Compustat ATQ) for firm i in quarter t.

${Profit}_{i,t}$: the ratio of EBITDA to total assets (Compustat ATQ) for firm i in quarter t.¹⁶

${Earnings}_{i,t}$: the ratio of retained earnings (Compustat REQ) to total assets (Compustat ATQ) for firm i in quarter t.

${Growth}_{i,t}$: the ratio of total book value of debt plus quarterly close price (Compustat PRCCQ) times the number of common stock shares outstanding (Compustat CSHOQ) to total asset (Compustat ATQ) for firm i in quarter t.

${Tangibility}_{i,t}$: the ratio of (net) property plant and equipment (Compustat PPENTQ) to total asset (Compustat ATQ) for firm i in quarter t.

${NDTS}_{i,t}$: the ratio of deferred taxes and investment tax credit (Compustat TXDITCQ) to total assets (Compustat ATQ) for firm i in quarter t.

1.3 Forecasting rating changes

We estimate a logit model by regressing two distinct categories: downgrades and ‘others’ (no rating change or upgrades) of S&P Long Term Rating (‘others’ is the reference category) on independent variables as re-written below:

$$\begin{aligned} &{LTD}_{i,t+1}^{D}=\,I^{D}({{\mathbf {X}}}_{i,t}) \nonumber \\& {LTD}_{i,t+1}^{D}=\left\{ \begin{array}{ll} 1, &\quad {SPLT}_{i,t+1}<{SPLT}_{i,t}\\ 0, &\quad {SPLT}_{i,t+1}\ge {SPLT}_{i,t} \end{array} \right. \end{aligned}$$

(14)

$t=1,2,\dots ,47,$ where ${LTD}_{i,t+1}^{D}$ is the response variable that indicates the rating change choice made by the rating agency. The state variables are conventionally considered in capital structure studies including: Leverage, Size, Price, Liquidity, Profit, Earnings, Growth, Tangibility and non-debt tax shields (NDTS) (see also, Ederington and Yawitz, 1986), including firm action variables: ${\triangle det}_{i,t}$ and ${\triangle eqt}_{i,t}$.

The predicted rating downgrade probability ${{\widehat{LTD}}_{i,t+1}^{D}}$ for firm i in quarter $t+1$ is given by:

$$\begin{aligned} {{\widehat{LTD}}_{i,t+1}^{D}}=Prob\left( {LTD}_{i,t+1}^{D}=1\right) =\frac{exp({{\mathbf {X}}}_{i,t}^{\prime }{{{\widehat{\beta }}}})}{1+{exp({{\mathbf {X}}}_{i,t}^{\prime }{{\widehat{\beta }}})}} \end{aligned}$$

(15)

The standard interpretation of the logit model is that for a one unit change in the predictor variables, the outcome relative to the reference group is expected to change by its respective parameter estimation given that other variables in the model are unchanged.

The estimation of (14) shows that the probability of downgrade ${{\widehat{LTD}}_{i,t+1}^{D}}$ is decreasing with profitability. The p-values from goodness of fit test shows that the model is a good fit for the data overall.

For long-term credit ratings, 1.17 % of the observations are downgrades. Thus, downgrades are rare events, and the predictors could suffer from small sample bias. Therefore we use the King and Zeng (2001) rare events small sample correction method for a binomial logistic model. This improves the predictability of the probabilities.