Does the order of claims to assets on the balance sheet reflect equity risk?

2.1 The association of financial leverage with common stockholder risk

Modigliani and Miller’s (1958, 1963) framework (hereinafter “MM”), shows that the financial risk to common stockholders increases with the debt-to-equity ratio (Appendix 1). In a frictionless market, common stockholders earn a residual return (rU−i)(1−τ)DLSL, that increases with the ratio of the market value of debt to common equity [2]. The intuition is that common shareholders earn a residual return after paying a fixed return to creditors. However, paying a fixed return to creditors increases the risk that common shareholders will not earn a positive return. Studies by Botosan and Plumlee (2005) and Dhaliwal et al. (2006) support MM’s prediction by providing evidence that the cost of equity capital is positively associated with the debt-to-equity ratio.

Rosenberg and McKibben (1973) test the MM framework and find that idiosyncratic risk is positively associated with the ratio of senior securities to total assets (LEV). Idiosyncratic risk is specific to a firm’s stock and is uncorrelated with the market return. It is defined as the variance of the residual (n_nt) for firm n in period t, from estimating the following market model:

Christie (1982) theoretically extends MM, by showing that in an MM world with a constant interest rate, no dividends, a single class of riskless debt and constant rate of return volatility:

However, the assumption that the right-hand side of the balance sheet consists of riskless bonds with a fixed rate of interest may not hold for at least two reasons. First, many liabilities do not have fixed interest rates, rather they may have variable rates, implied rates (e.g. trade payables) or none at all (e.g. deferred taxes and accruals for payroll and warranties).

Secondly, liabilities are not riskless because some firms default. Consequently, debt covenants and bankruptcy laws restrict how firms can invest, finance and pay creditors and investors. Debt covenants may require the firm to maintain a minimum working capital level and prohibit additional borrowing unless specific financial ratios are met (Barclay & Smith, 1995). If a firm defaults on payment, creditors generally have the right to force the firm into bankruptcy, which by a court-supervised proceeding either reorganizes the debt contracts or liquidates the assets (Barclay & Smith, 1995). When the assets are liquidated, secured creditors (including lessors) are granted settlement priority over unsecured creditors and preferred stockholders, respectively (Barclay & Smith, 1995). In short, all claims on the right-hand side of the balance sheet are not identical in risk. Following prior studies (Diamond, 1991; Kimmel & Warfield, 1995; Nissim & Penman, 2003), we examine three dimensions of claims (current vs. noncurrent liabilities, operating vs. financing liabilities and liability-like and equity-like features of preferred stock). To provide further granularity, we examine individual liabilities including controversial liabilities (deferred income taxes and capital leases).

Further, the MM framework does not consider the costs of bankruptcy that are associated with lower observed leverage levels (Megginson, 1997). We examine whether financially distressed firms allocate risk to the right-hand side of the balance sheet differently from nondistressed firms.

We next discuss the background to all three dimensions of the right-hand side of the balance sheet, including specific classes and the associated risks of each category. We follow with our predictions.

2.2 Current and noncurrent liabilities

We begin our discussion with the classification of current and noncurrent liabilities. The classification is required by the FASB codification (ASC 210-10-05-4) and Regulation S-X (17 CFR § 210.5-02 – Balance Sheets) [3]. The FASB codification requires separated classification of current liabilities and it provides examples such as current operating liabilities and short-term debt (ASC 210-10-45) [4]. Similar to the FASB codification, Regulation S-X lists specific current liabilities on the face of the balance sheet (e.g. amounts payable to trade creditors), while permitting note disclosure for short-term debt (borrowings from financial institutions and commercial paper), and it requires a total for current liabilities. Regulation S-X also lists categories of non-current liabilities reported on the face of the balance sheet or by note disclosure (bonds, mortgages and capitalized leases). Deferred taxes are shown on the face of the balance sheet.

Current liabilities consist of legally enforceable claims from operating creditors and financial creditors. Operating creditors are owed for amounts of trade payables, payroll, sales taxes, income taxes and warranties. Financial creditors are owed for short-term debt including commercial paper, short-term loans and the current portion of long-term debt. In bankruptcy proceedings, secured short-term debt has a higher settlement priority than unsecured short-term debt.

Noncurrent liabilities consist of amounts owing to operating creditors and financial creditors. Operating creditors are owed for deferred income taxes and pension liabilities. Financial creditors are owed for long-term borrowings from financial institutions, bondholders and capital leases. Pension liabilities, long-term borrowings and capital leases are legally enforceable while deferred income taxes are not suggesting that deferred income taxes are less associated with financial risk than other noncurrent liabilities. Compared to operating creditors, financial creditors have unique contractual rights. Long-term borrowings typically have debt covenants that restrict investment, payout and financing policies, such as meeting working capital balances and restricting borrowing if specific financial ratios are not met (Barclay & Smith, 1995). Capital leases typically grant repossession rights to the lessor in the event of default, and they have the highest priority of claims among secured creditors under bankruptcy laws, while creditors of secured long-term borrowings have priority over unsecured creditors (Barclay & Smith, 1995).

Current and noncurrent liabilities differ by three risks: liquidity, operating and interest-rate risk. Liquidity risk is “the risk that a solvent but illiquid borrower is unable to obtain refinancing” (Diamond, 1991). Short-term borrowings must be refinanced after they mature and before the investment pays off. If the lender is unwilling to refinance, the borrower must liquidate the assets, to settle the short-term debt, sometimes at less than their intrinsic value. Because current liabilities mature before noncurrent liabilities, we expect that they are associated with greater liquidity risk.

Arguably, operating risk is higher for current liabilities than noncurrent liabilities. Operating risk occurs when creditors have the power to suspend operations if they are not paid or if payments are too slow and uncertain, both of which can increase the risk of bankruptcy. Suppliers hold the threat to stop delivery, leaving few if any practical substitutes for unique materials, brand-name merchandise and utilities (Petersen & Rajan, 1997; Cunat, 2007). Unpaid employees may stop working and litigate, leaving few if any practical substitutes for skilled or unionized labor (Findlaw, 2021). Governing authorities, such as the U.S. federal government, have strong collection powers such as seizure of property without a court ruling when the taxpayer refuses to pay (Internal Revenue Code, 26 U.S.C. § 6331).

A counterargument is that noncurrent liabilities include secured creditors (financial institutions, bondholders and lessors) who in the event of default could stop operations by forcing liquidation or repossessing assets. However, in the event of default, the U.S. Bankruptcy Code constrains their powers to liquidate the firm’s assets and destroy firm value. Chapter 11 of the Code (11 U.S.C. § 1101-1195) puts in place “guardrails” for management to negotiate a plan to restructure the debt contracts with the creditors (Casey, 2020). Under Chapter 11, an automatic stay prevents creditors from foreclosing on their collateral until the firm exits bankruptcy (Gilson, John, & Lang, 1990). Consequently, management more often negotiates a restructuring plan to continue operations, either privately or by court-supervision, rather than liquidating the firm’s assets (Barclay & Smith, 1995; Warren & Westbrook, 2009) [5]. In short, because bankruptcy laws constrain repossession and liquidation rights for secured creditors, we expect that operating creditors of current liabilities impose more operating risk and bankruptcy risk.

Finally, we argue that interest-rate risk is higher for noncurrent liabilities than current liabilities, but the effect may be mitigated. Investors demand a premium for holding securities with a fixed rate of return over a long maturity term (e.g. fixed-rate notes payable and capital leases); thus, noncurrent liabilities potentially bear higher interest rates than current liabilities. However, firms mitigate interest-rate risk with variable interest rate terms, hedging with derivatives and frequent adjustments to discount rate estimates for defined benefit pension plan liabilities.

Liabilities also bear financial risk arising from fixed, variable and implied interest rates, but it is unclear whether current liabilities bear greater risk than noncurrent liabilities. Financial risk is arguably greater for variable rates than fixed rates because the interest charges vary more over time. However, short-term debt and long-term debt potentially use both variable and fixed rates, while accounts payable and capital leases have implied rates.

In short, we predict that common stockholders view current liabilities as riskier than non-current liabilities because operating risk and liquidity risk is greater while interest-rate risk is mitigated.

2.3 Operating and financial liabilities

Liabilities are also characterized by their cash flow activity – operating and financing types (Nissim & Penman, 2003). Operating activities “generally involve producing and delivering goods and providing services,” and generally are from “transactions and other events that enter into the determination of net income” (ASC 230-10-20). They are owed to operating creditors: suppliers, employees, customers and government. On the other hand, financing activities involve “borrowing money and repaying amounts borrowed, or otherwise settling the obligation; and obtaining and paying for other resources obtained from creditors on long-term credit” (ASC 230-10-20). They are owed to financial institutions (e.g. banks and leasing companies) and holders of debt securities (e.g. commercial paper and bonds) and they operate in well-functioning capital markets (Nissim & Penman, 2003). Both operating liabilities and financial liabilities have financial risk, but we do not predict which category is greater.

We argue that common stockholders view operating liabilities as riskier than financial liabilities because they bear more operating risk that leads to greater bankruptcy risk. Operating creditors hold the threat to prematurely destroy the going-concern value of the firm, and suppliers have an information advantage over financial creditors (Cunat, 2007; Giannetti, Burkart, & Ellingsen, 2011; Petersen and Rajan, 1997). As discussed, suppliers and employees can halt operations if not paid on time, or if payments are too slow and uncertain. Suppliers also have an information advantage over financial institutions to sooner stop or threaten to stop delivery (Giannetti et al., 2011; Petersen & Rajan, 1997) [6].

On the other hand, as mentioned, financial creditors’ rights to liquidate assets are constrained by the U.S. Bankruptcy Code. Chapter 11 of the Code provides “guardrails” for management to restructure debt contacts, continue operations and minimize parties from holding each other up (Casey, 2020). A counterargument is that suppliers are also parties to debt restructuring plans and their liquidation rights are also constrained. However, suppliers still have the “threat of stopping the supply of intermediate goods to their customers” that financial creditors lack (Cunat, 2007), suggesting that operating liabilities have greater operating and bankruptcy risk than financial liabilities.

In the accounting literature, Nissim and Penman (2003) find that operating leverage has a higher association with future profitability (and the price to book ratio) than financing leverage. However, they do not examine and compare the association of operating leverage and financing leverage with risk. In short, we predict that common stockholders view operating liabilities as riskier than financial liabilities.

2.4 Preferred stock

Preferred stock is a hybrid of liability and equity. Like traditional debt, it requires periodic payments in the form of perpetual fixed-rate dividends. However, like common stock, the dividend payments can be postponed or suspended until liquid resources are available (Heinkel & Zechner, 1990). Further, preferred stockholders cannot force the firm into bankruptcy (Barclay & Smith, 1995). Like equity, some classes of preferred stock have voting rights (Kimmel & Warfield, 1995). In bankruptcy proceedings, preferred stockholders rank lower in settlement priority over secured and unsecured creditors and higher than common stockholders (Barclay & Smith, 1995).

An example is Chicken Soup for the Soul Entertainment, Inc. (CSSE) Series A preferred stock, issued in 2018, with an annual cumulative dividend rate of 9.75% [7]. The shareholders have no voting rights except to elect two members to the board of directors if dividends are in arrears for eighteen or more consecutive or non-consecutive months. In the event of liquidation, the shareholders receive $25 per share and any accumulated and unpaid dividends before common stockholders are paid.

The accounting standards for classifying preferred stock are less straight forward because of several FASB standards (topics 480, 815 and 505) and two balance sheet categories under Regulation S-X (17 CFR § 210.5-02). These requirements define various classes of preferred stock and specify how they are classified on the balance sheet (liabilities, mezzanine equity and stockholders’ equity). The FASB defines equity as “the residual interest in the assets of an entity that remains after deducting its liabilities” (ASC 505-10-05-3), implying that unless it is classified as a liability, it is equity. In the FASB Master Glossary, preferred stock is a “security that has preferential rights compared to common stock.” SEC Regulation S-X, since 1979, classifies two types of preferred stock outside of liabilities. The first type, redeemable preferred stock, is classified as mezzanine equity, between liabilities and stockholders’ equity (17 CFR § 210.5-02.27). The regulation defines redeemable preferred stock as mandatorily redeemable by the holder or redeemable outside the control of the issuer. The second category, which we refer to as nonredeemable preferred stock, is classified as stockholders’ equity (17 CFR § 210.5-02.28). It includes preferred stock that is not redeemable or is only redeemable by the issuer (i.e. callable).

FASB topic 480 and subtopic 815-15 address classification of preferred stock as a liability and the bifurcation into liability and equity components, respectively. ASC 480, Distinguishing Liabilities from Equity, since 2003, requires preferred stock to be classified as a liability if it is certain to be redeemed (mandatorily redeemable preferred stock) or has an unconditional obligation to issue a variable number of equity shares based on a fixed monetary value, variations in something other than the fair value of equity shares or variations inversely related to changes in the fair value of equity shares (ASC 480-10-25) [8]. ASC 815-15, Derivatives and Hedging – Embedded Derivatives, since 1999, bifurcates the embedded derivative of preferred stock into a fair-valued liability and mezzanine equity or stockholders’ equity [9]. Embedded derivatives include redemption rights by the holder or the issuer, rights by the holder to convert to common stock and other features (ASC 815-15-25-17D). The subtopic applies when embedded derivatives in their entirety are not clearly and closely related to the host preferred stock instrument’s economic risks and characteristics (liability-like or equity-like), together with other requirements (ASC 815-15-25-1) [10].

While preferred stock has several features, our analysis is constrained by the four categories that Compustat provides: redeemable, nonredeemable, convertible and non-convertible. Other features (voting rights, callability, embedded derivatives and mandatory redemption features) require hand-collection which is not practicable for our sample size (102,928). One study that examines redeemable preferred stock with voting rights was limited to a sample of 239 observations (Kimmel & Warfield, 1995). Rather, we test the preferred stock classes that are recorded by Compustat, which includes either mezzanine equity (redeemable) or stockholders’ equity (nonredeemable, convertible and non-convertible) [11]. Liability classification of embedded derivatives (ASC 815-15) and mandatorily redeemable preferred stock (ASC 480-10-25) are not examined.

We predict that preferred stock is positively associated with risk because it bears financial risk. Perpetual fixed-rate dividends are like interest payments on debt and are senior in payment priority to common stock dividends, which increase return volatility to common shareholders. Some preferred stock securities have cumulative rights to pay amounts in arrears, similar to accrued and unpaid interest. Further, preferred stock may have voting rights if dividends are sufficiently in arrears, permitting preferred stockholder to influence payment (e.g. OME and CSSE) [12]. Preferred stock also enhances debt capacity so that the firm can borrow more (Heinkel & Zechner, 1990), which also bears financial risk. However, preferred stock has little or no liquidity or bankruptcy risk because postponement or suspension of dividends cannot force liquidation or trigger bankruptcy. A counterargument is that if a firm suspends dividends indefinitely, there is no financial risk associated with preferred stock. However, such firms are likely insolvent and few. For solvent firms, when dividends are declared, preferred stockholders are paid before common stockholders.

Further, since we are interested in whether the ordering of preferred stock after total liabilities matters, we predict that preferred stock is less associated with risk than total liabilities. Namely, preferred stock bears less liquidity risk and bankruptcy risk than liabilities because the dividends can be deferred and deferral cannot force liquidation or trigger bankruptcy.

Prior empirical studies (Cheng et al., 2003; Kimmel & Warfield, 1995; Linsmeier, Partridge, & Shakespeare, 2020) find mixed results from examining the association of preferred stock with systematic risk and common stock prices. Kimmel and Warfield (1995) find that redeemable preferred stock with voting rights or conversion rights is negatively associated with systematic risk, incremental to operating risk (1979–1989). Cheng et al. (2003) find that nonredeemable preferred stock is positively associated with systematic risk (1993–1997), incremental to operating risk. Our study differs from these two studies by using a stronger measure of risk specific to the firm (idiosyncratic risk) [13], a longer sample period (1986–2016), and we use a first difference model to control for time-invariant omitted variables. In a related working paper, Linsmeier et al. (2020) examine whether investors view preferred stock as equity-like (liability-like) by whether it is positively (negatively) associated with common stock price. They find that preferred stock is positively (negatively) associated with common stock price for financially distressed (non-distressed) firms. Our paper differs in that we compare preferred stock and other claims to idiosyncratic risk for distressed and nondistressed firms, and we use a first-difference design. In short, while prior empirical findings are mixed, preferred stock arguably bears financial risk, thus we predict that it has a positive association with risk. Further, we predict preferred stock is less associated with risk than total liabilities (current and noncurrent liabilities, and operating and financing liabilities) because it bears less liquidity risk and bankruptcy risk.

3.1 Data measurement

Our dependent variable is the firm-specific risk borne by common stockholders, measured as idiosyncratic risk. Idiosyncratic risk (IRISK_t) is the portion of stock return variability which is not attributed to overall market return variability. We measure IRISK_t as the standard deviation of the residual from regressing daily stock returns on Center for Research in Security Prices (CRSP) value-weighted market returns for the one-year period starting from the beginning of the fourth month of fiscal year t until the end of the third month of fiscal year t+1. We measure idiosyncratic risk for one year (t) (IRISK_t), three years (t−2 to t) (IRISK_t3) and five years (t−4 to t) (IRISK_t5) (e.g. Ashbaugh-Skaife, Collins, Kinney, & Lafond, 2009) [16]. The latter two measures capture the longer-term effects for noncurrent liabilities and preferred stock. We retrieve return data from CRSP and derive our risk measures based on daily returns starting at the beginning of the fourth month of fiscal year t, t-2 and t-4 through to the end of the third month of fiscal year t +1. The three-month lag permits us to measure the stock return during the approximate period that financial statement information is available to the stock market. IRISK_t, IRISK_t3 and IRISK_t5 are multiplied by 10 to better compare the variable coefficients in our regression analysis.

The independent variables in our analysis include the first two dimensions: current liabilities (CL_t) and noncurrent liabilities (NCL_t), and total operating liabilities (TOL_t) and total financial liabilities (TFL_t). We also examine the intersection of the two dimensions as current operating liabilities (COL_t), noncurrent operating liabilities (NCOL_t), current financial liabilities (STD_t) and noncurrent financial liabilities (NCFL_t).

We then decompose these categories as follows. Current operating liabilities decompose into current operating liabilities excluding income taxes (COLx_t) and income taxes payable (IP_t). Current financial liabilities do not decompose and only include short-term debt (STD_t). Noncurrent operating liabilities decompose into deferred taxes (DTX_t) and other noncurrent operating liabilities (ONCL_t). Noncurrent financial liabilities decompose into long-term debt (LTD_t) and capital leases (CapL_t).

To examine the third dimension, equity-like and liability-like claims, we add preferred stock (PS_t) to our regression analysis. We decompose preferred stock into two groups of features: redeemable preferred stock (PSR_t) and nonredeemable preferred stock (PSNR_t), and convertible preferred stock (PSC_t) and non-convertible preferred stock (PSNC_t). We scale all liability and preferred stock variables by the market value of common stock (CE_t) (Christie, 1982).

3.2 Sample

Our sample initially consists of 112,842 firm-year observations covering 32 fiscal years from 1985 to 2016 for all firms (except financial and utility firms) and all available data from Compustat and CRSP to compute the variables. Our research design includes first-difference variables, which are computed as the difference between the current period and the prior period amounts. After first-differencing, the sample size is reduced by one year to 102,928 firm-year observations covering 31 years (1986 to 2016). Table 1 reports descriptive statistics. All variables are windsorized at the 1^st and 99^th percentiles. Mean (median) IRISK_t is 0.384 (0.313). Kothari, Li, and Short (2009) report mean (median) return volatility (multiplied by 10) of 0.270 (0.230) for the period from 1996 to 2001, suggesting that our sample has riskier firm-year observations. Mean (median) IRISK_t3 is 0.388 (0.332) and IRISK_t5 is 0.390 (0.342), comparable to IRISK_t. We report first-difference variable amounts in Panel B.

Table 2, Panels A and B, report correlation coefficients (Pearson in the upper right of the diagonal and Spearman in the lower left) for first-difference variables, which we use in our multivariate analysis. Panel A reports correlations between the three risk measures and broader categories of liabilities (current, noncurrent, operating and financing) and preferred stock. Panels B reports correlations between the three risk measures and individual accounts. Panels A and B report that all three first-difference risk measures are positively correlated with virtually all first-difference liabilities and preferred stock variables.

4.1 Liabilities analysis

We begin by examining whether current liabilities and noncurrent liabilities are positively associated with risk, and whether the positive association of current liabilities is greater than noncurrent liabilities. We estimate the following first-difference model, where i and t are firm and year subscripts, respectively, RISK_it denotes IRISK_it, IRISK_it3 and IRISK_it5, and Δ is the difference between year t and year t−1:

Lagged level terms for leverage components are added as instruments because of potential feedback effects influencing risk levels. For instance, firms that are financially distressed may substitute current liabilities for noncurrent liabilities because creditors expect a shorter investment horizon.

We control for several risks that may not be adequately controlled by first-differencing. We control for operating risk (Lev, 1974) by including the standard deviation of ROA (net income before interest and taxes scaled by total assets) over years t, t−1 and t−2. We add industry fixed effects, using Fama and French (1997) 48 industries, to control for bankruptcy risk that may vary by industry life-cycle stage [17]. In later analysis, we partition the sample between financially distressed firms and nondistressed firms. Industry fixed effects also control for industry financing practices such as greater debt financing in capital intensive industries (Megginson, 1997). Agency risk is controlled by adding firm-size-decile dummies (Gayle & Miller, 2009). Information risk is controlled by adding book-to-market decile dummies (Rajgopal & Venkatachalam, 2011), firm-size-decile dummies (Lang & Lundholm, 1993) and bid-ask spread. Year dummies are added to control for temporal effects, such as changes in accounting standards, regulations and laws. Standard errors are clustered by firm to adjust for within-firm serial correlation (Petersen, 2009). Standardized coefficients are reported to afford comparability.

The results of estimating equation (1) are presented in Table 3. Operating risk (ΔSTD_ROA)_t and bid-ask spread (ΔBID_ASK)_t are positively associated with all three risk first-difference measures. The maximum variance inflation factor (VIF) for all independent variables (excluding dummy variables and lagged variables) is well below the multicollinearity threshold of 10 (Kennedy, 2008). Our variables of interest, Δ(CL/CE)_t and Δ(NCL/CE)_t, are positively associated with all three idiosyncratic risk measures (ΔIRISK_t, ΔIRISK_t3 and ΔIRISK_t5), although the effect is reduced for the latter two measures. The coefficient F-tests reveal that Δ(CL/CE)_t has a greater standardized coefficient than Δ(NCL/CE)_t, suggesting that current liabilities have a stronger association with idiosyncratic risk than noncurrent liabilities, supporting our prediction.

In equation (2a), we classify liabilities by operating liabilities Δ(TOL/CE)_t and financial liabilities Δ(TFL/CE)_t. Equation (2b) extends (2a) with current and noncurrent classification: current operating liabilities Δ(COL/CE)_t, noncurrent operating liabilities Δ(NCOL/CE)_t, current financial liabilities Δ(STD/CE)_t and noncurrent financial liabilities Δ(NCFL/CE)_t. Equation (2c) expands with individual accounts: current operating liabilities excluding income taxes Δ(COLx/CE)_t, income taxes payable Δ(IP/CE)_t, deferred taxes Δ(DTX/CE)_t, other noncurrent liabilities Δ(ONCL/CE)_t, short-term debt Δ(STD/CE)_t, long-term debt Δ(LTD/CE)_t and capital leases Δ(CapL/CE)_t.

Table 4 reports the results of estimating equations (2a)–(2c). Results from estimating equation (2a) show that Δ(TOL/CE)_t and Δ(TFL/CE)_t are positively associated with all three idiosyncratic risk measures, and the standardized coefficient on Δ(TOL/CE)_t is greater than Δ(TFL/CE)_t supporting our prediction that operating liabilities have a stronger association with idiosyncratic risk than financial liabilities. Results from estimating equation (2b) show that Δ(COL/CE)_t, Δ(NCOL/CE)_t, Δ(STD/CE)_t and Δ(NCFL/CE)_t are positively associated with all three idiosyncratic risk measures. Results from estimating equation (2c) show that Δ(COLx/CE)_t, Δ(ONCL/CE)_t, Δ(STD/CE)_t, Δ(LTD/CE)_t and Δ(CapL/CE)_t are positively associated with all three idiosyncratic risk measures. Notably, income taxes payable and deferred taxes are not associated with idiosyncratic risk.

4.2 Preferred stock analysis

To examine preferred stock (PS/CE)_it, we estimate the following models:

We decompose preferred stock into two features provided by Compustat: redeemable (Δ(PSR)_t) and nonredeemable (Δ(PSNR)_t) (3c), and convertible (Δ(PSC)_t) and nonconvertible (Δ(PSNC)_t) (3d), as follows:

Table 5, Panel A, reports the results of estimating equations (3a) and (3b) and Panel B reports the results of estimating equations (3c) and (3d). Results of estimating equation (3a) reveal that the standardized coefficient on Δ(PS/CE)_t is positively associated with all three idiosyncratic risk measures, supporting our prediction. Other coefficient values are consistent with those reported in Table 3. The coefficient F-tests reveal that Δ(PS/CE)_t has a smaller standardized coefficient than Δ(CL/CE)_t and Δ(NCL/CE)_t. These results suggest that preferred stock has a smaller positive association with idiosyncratic risk than current liabilities and noncurrent liabilities, consistent with our prediction.

Results of estimating equation (3b) show that Δ(PS/CE)_t is positively associated with all three idiosyncratic risk measures, supporting our prediction. Δ(TOL/CE)_t and Δ(TFL/CE)_t are positively associated with all three idiosyncratic risk measures, consistent with Table 4 equation (2a) results. The F-tests reveal that the standardized coefficients on Δ(PS/CE)_t are less than Δ(TOL/CE)_t and Δ(TFL/CE)_t. These results suggest that preferred stock has a smaller positive association with idiosyncratic risk than operating liabilities and financial liabilities, consistent with our prediction.

Results of estimating equation (3c) reveal that Δ(PSR/CE)_t and Δ(PSNR/CE)_t are positively associated with all three idiosyncratic risk measures, supporting our predictions. Standardized coefficient signs on other leverage components are consistent with results reported in Table 4 equation 2(c). Compared to other leverage components, redeemable and nonredeemable preferred stock have a weaker positive association with idiosyncratic risk than many liability categories, except Δ(IP/CE)_t and Δ(DTX/CE)_t.

Results of estimating equation (3d) reveal that Δ(PSC/CE)_t is positively associated with all three idiosyncratic risk measures, supporting our prediction. Δ(PSNC/CE)_t is only positively associated with ΔIRISK_t3 and is not statistically associated with ΔIRISK_t and ΔIRISK_t5, thus we make no inference of an association with idiosyncratic risk, contrary to our prediction. Standardized coefficient signs on other leverage components are similar to results reported in Table 4 equation 2(c). Compared to other leverage components, convertible preferred stock has a weaker positive association with idiosyncratic risk than many liability categories, except Δ(CapL/CE)_t, Δ(IP/CE)_t and Δ(DTX/CE)_t.

4.3 Robustness – Altman bankruptcy prediction z-score

For robustness purposes, we add ALTMAN as an additional variable to equations (1) through (3d) to control for bankruptcy risk that is not captured by industry dummy variables (untabulated and available upon request). We find our inferences are unchanged.

4.4 Financial distress analysis

Financial distress may change the association of idiosyncratic risk with creditor and preferred stockholder claims because the firm’s capital structure changes. When firms are financially distressed, creditors gain bargaining power over shareholders because they can threaten to exercise liquidation rights and halt operations if payments are in default. In turn, the firm may restructure their debt contacts to meet their obligations (Hotchkiss et al., 2008). They may renegotiate terms with creditors such as reducing or deferring obligations and replacing debt with common and preferred stock allowing even greater flexibility to defer payments without forcing bankruptcy while reducing the debt-to-equity ratio (Hotchkiss et al., 2008). They may even issue additional debt or equity to others (Hotchkiss et al., 2008).

Trade credit and preferred stock are attractive forms of financing for distressed firms. Trade credit provides liquidity to smaller, younger and higher growth firms and suppliers can better enforce payment (Cunat, 2007). Preferred stock allows dividend deferral without forcing bankruptcy, reduces leverage, and is used by companies with lower tax rates (Houston & Houston, 1990). Convertible preferred stock has a lower dividend rate, as well as the flexibility to defer dividend payments. In turn, cash constrained firms, such as start-ups, plough cash into investment opportunities with the upside potential for preferred stockholders to convert and share in future profits (Lee & Figlewicz, 1999; Jepsen & Wilks, 2018; Johnson, n.d.). Redeemable preferred stock also has the flexibility to defer dividends with a one-time payout.

In short, financially distressed firms likely bear greater bankruptcy, financial and liquidity risk, because they are more likely to fail. In turn, creditors raise interest rates and lend on a secured basis and a shorter term (Tirole, 2006). Preferred stock is likely issued at a higher dividend rate and with a redemption feature. Thus, we expect that liabilities and preferred stock are more positively associated with idiosyncratic risk for financially distressed firms than nondistressed firms.

To examine this prediction, we partition our sample between financially distressed and nondistressed firms. We identify financially distressed firms as those below the Altman (1968) z-score of 1.81. We use the Altman (1968) z-score because it is the dominant method for measuring financial distress and bankruptcy prediction, and thus should better reflect the capital structure of the firm and the underlying risk associated with it [18, 19]. Financially distressed firms comprise 27% of our sample, comparable to Altman (2013) that reports 20% for all U.S. industrial firms in Compustat in 1999.

We first examine total liabilities (TL) by estimating the following:

We then examine total liabilities and preferred stock. We estimate the following:

We then extend equation (4b) by separating liabilities into current and noncurrent categories, and operating and financial categories, as follows:

Table 6, Panels A, B, and C, respectively reports the results of estimating equations (4a), (4b) and (4c), by financially distressed and nondistressed firms and reports the difference in standardized coefficients between the two groups. Panel A reports that the standardized coefficient on Δ(TL/CE)_t is positive and statistically significant for all three risk measures. Standardized coefficients are greater for distressed firms than nondistressed firms, as predicted. The implication is that the positive association of total liabilities and idiosyncratic risk is greater for distressed firms than nondistressed firms, consistent with the notion that creditors gain bargaining power when firms are financially distressed.

Panel B reports that the standardized coefficient on Δ(PS/CE)_t is positive and statistically significant for financially distressed firms only and standardized coefficients are greater for distressed firms than nondistressed firms when the dependent variable is ΔIRISK_t and ΔIRISK_t5, and marginally significant for ΔIRISK_t3 (p < 0.10). The implication is that the positive association of preferred stock and idiosyncratic risk is greater for distressed firms than nondistressed firms, and that the positive association for the entire sample (Table 5, Panel A) appears to be driven by distressed firms.

Panel C reports that standardized coefficients are greater for distressed firms than nondistressed firms for Δ(COL/CE)_t, suggesting that the positive association between current operating liabilities and idiosyncratic risk is greater for distressed firms than nondistressed firms. The standardized coefficients on Δ(STD/CE)_t are greater for distressed firms than nondistressed firms, suggesting that the positive association between short-term debt and idiosyncratic risk is greater for distressed firms than non-distressed firms. Differences in standardized coefficients between distressed and nondistressed firm are not statistically significant across most of the three idiosyncratic risk measures for Δ(NCOL/CE)_t and Δ(NCFL/CE)_t. Panel C also reports that the liability component coefficient signs are consistent with Table 4 equation (2b). Results for Δ(PS/CE)_t are similar to Panel B.

We extend equation (4c) by separating preferred stock into redeemable and nonredeemable categories, and convertible and nonconvertible categories, as follows:

Panel D reports the results of estimating equation (4d). The positive association between redeemable preferred stock and the three idiosyncratic risk measures is only evident for distressed firms. The standardized coefficient is greater for distressed firms than nondistressed firms when the dependent variable is ΔIRISK_t and ΔIRISK_t5, and marginally significant for ΔIRISK_t3 (p < 0.10). The implication is that the positive association observed for the entire sample (Table 5, Panel B) is primarily driven by financially distressed firms, which is consistent with the notion that redeemable preferred stock is used by distressed firms, including cash constrained start-up companies. There is no difference in the association of nonredeemable preferred stock and idiosyncratic risk between distressed and nondistressed firms.

Panel E reports the results of estimating equation (4e). The positive association between convertible preferred stock and the three idiosyncratic risk measures is only evident for distressed firms. The standardized coefficient is greater for distressed firms than nondistressed firms when the dependent variable is ΔIRISK_t and ΔIRISK_t5 and is not statistically different for ΔIRISK_t3. The implication is that the positive association observed for the entire sample (Table 5, Panel B) is driven by financially distressed firms. There is no difference in the association of nonconvertible preferred stock and idiosyncratic risk between distressed and nondistressed firms.