Cost of Capital, Firm Size and Financial Distress

Financial distress (FD) is a global muddle that adversely affects firms and economies. Kenya documents substantive evidence of FD across economic sectors. This indicates a missing link between financial surveillance and business management. Past research concentrates on FD modeling and capital structure effects on performance. This study explored the influence of cost of capital on FD and the moderation effect of firm size. The study was anchored on Modigliani and Miller’s second proposition and Trade-off theory. Retrospective longitudinal research design was adopted targeting all non-financial firms listed in Nairobi Securities Exchange (NSE). Hierarchical panel regression analysis explored the multi-dimensional financial data collected from audited financial statements, daily stock prices and market indices from year 2006 to 2015. Findings show cost of capital to relate significantly and negatively with FD. Cost of equity (Ke) rises with cost of debt (Kd). Interaction term cost of capital*firm size has no effect on FD. Kd and Ke significantly influence FD positively and negatively respectively. Interaction Kd*firm size has a positive insignificant influence on FD while interaction Ke*firm size has a negative significant effect on FD. The study recommends diligent capital budgeting to ensure firms only invest in feasible ventures surpassing the cost of capital.


Literature Review
This study is anchored on the second proposition of Modigliani & Miller (MM) and trade-off theory. MM capital structure theorems provide the foundation of modern corporate finance theory. MM's second proposition holds that a firm's cost of equity increases with an increment in debt-equity ratio (Modigliani & Miller, 1958, 1963. Trade-off theory on the other hand postulates that the optimal debt to value ratio in a firm is purely determined by striking a balance in between the costs and benefits of debt financing while assets and investment decisions are held constant (Myers, 1984).
Empirical literature relating to the study was reviewed in establishment of research gaps. Ikpesu and Eboiyehi (2018) studied a sample of 58 manufacturing firms listed in Nigeria Stock Exchange from year 2010 to 2016 and established that capital structure prompts financial distress owing to escalated financial leverage despite borrowing being an opportune norm in the manufacturing sector. On the contrary, Bagga and Kaur (2016) determined that cost of capital negatively relates to financial leverage being a proxy for financial distress in the Indian manufacturing sector. The latter study further proves the existence of a positive relationship between firm value and leverage hence concurring with MM's first proposition with taxes. Comparably, Dhankar (1996) figured out a negative relationship between cost of capital and capital structure using a sample of 26 firms listed in Bombay Stock Exchange from 1981 to 1991. Despite the effect of capital structure decision on firm value varying across firms, it is significant at a multifaceted macro-level that accommodates market factors but insignificant at microlevel where most influential predictors such as firm reputation fails to be measurable (Dhankar, 1996). Elsewhere, Ozkan (1996) analyzed a panel dataset of 195 U.K. firms trading from year 1981 to 1991 and found out that firm characteristics including the asset base, influences FD and capital structure decisions. Pindado, Rodrigues & De La Torre (2006) uncovered that FD relates negatively with borrowing based on unbalanced panel data from 402 small Portuguese business ventures from year 1990 -1997. In addition, small firms in distress zones tend to make irrational financial decisions (Pindado et al., 2006). However, the study fails to account for the financial choices accorded to distressed firms. In a study on publicly trading firms in Kenya, Muigai (2016) upholds that internal equity financing has a negative and significant effect on financial distress concluding that a combination of internal equity and non-current debt financing significantly reduces the probability of distress. Conversely, Kirui and Gor (2018) disputed the pecking order hypothesis on the ground that financially troubled manufacturing firms listed in Kenya from year 1999 to 2016 do not prioritize using internal funds to cater for capital expenditure. The latter study recognizes the divergent effects from cost of debt and cost of equity on the financial status but fails to attach the requisite significance and direction of association they exhibit. Alifani and Nugroho (2013) proved an inconsistent existence of the second proposition of MM's capital structure theory using a sample of Indonesian firms from year 2003 to 2012. The study revealed that cost of equity does not necessarily rise with increase in leverage. This holds when cost of debt is greater than cost of equity owing to gradual skewness towards reliance on debt over the years while exhibiting an insufficient EBIT to pay off the finance cost of borrowed funds (Alifani and Nugroho, 2013). On the contrary, Naidu (2013) supports MM's second proposition arguing that return on equity increases when debt levels are high because the leverage induces market reactions from equity shareholders who demand for higher returns as a cover for exposure to more risk. However, Alifani and Nugroho (2013) agrees to the latter supposition only if associated with neutral or risk taking investors that seek for more risk with an expectation of higher returns unlike risk averse investors who shy away from capital structures that accommodate much debt. Damodaran (2016) upholds that the overall cost of capital responds to cost of debt which lenders adequately determine after consideration of default risk while on the other hand investors demand for a higher return on the perception that it should cater for the possible risk of losing their investment due to financial risk from increased leverage. Empirically, it has been proved that debt leverage beyond optimality, induces FD hence denoting a positive significant association between FD and leverage (Carmassi & Patti, 2015;Kazemian et al., 2017;Khaliq, Altarturi, Thaker, Harun, & Nurun, 2014;Muigai, 2016;Sporta, 2018). Findings in the latter group of studies concur with the trade-off theory.
Chancharat (2008) determined firm size to have a positive and significant influence on FD using survival analysis techniques on a sample of 1,117 companies trading from year 1989 to 2005. The study also concluded that large firms exhibit a high probability of being distressed more so if levered. Similarly, Carmassi and Patti (2015) observed large firms to commonly have more debt in their capital structures in comparison to smaller firms hence standing a higher chance of suffering FD. On the contrary, Edesiri (2014) established that firm size represented by total assets, negatively affects FD. The latter study was based on a sample of 120 listed companies in Nigeria from year 1990 -2013. Ikpesu and Eboiyehi (2018) realized that firm size and FD associates negatively based on a dataset from Nigerian manufacturing firms in between year 2010 -2016. Likewise, Makeeva and Khugaeva (2018) through a panel logistic regression analysis of 389 innovative firms sampled across the world on the basis of incurring an annual expenditure > $ 200 million, determined that firm size negatively and significantly influences distress. On another note, Rianti and Yadiat (2018) determined that firm size expressed in assets, insignificantly predicts FD among the agricultural firms listed in Indonesia Stock Exchange from year 2012 to 2014. Comparably, Agarwal and Taffler (2008) in a study on FD and momentum anomalies in firms listed in London Stock Exchange from year 1979 to 2002, established that firm size in terms of market capitalization does not affect FD.
In reference to moderation from firm size, Ali, Mukulu and Kihoro (2016) explored the interaction effect caused by firm size in the association between functional integration and firm performance proxied by both financial and non-financial facets using a sample of 176 manufacturing firms in Kenya. The study's findings uncovered that the interaction term (firm size*functional integration) had no influence on firm performance thus concluding that firm size as a function of total assets and total sales to be a weak moderator. In contrast, Mutunga and Owino (2017) analyzed the interaction role of firm size in terms of assets and market share in the association between micro-factors and financial performance using a sample of 180 firms trading in Kenya. Micro-factors were operationalized by; production scope, management decisions and business practices. The findings revealed that the interaction term (firm size*micro-factors) has a significant positive effect on firm performance thus concluding that larger firms can embrace leverage to improve their financial performance in contrast to smaller firms that are better off when inclined towards equity financing. Comparatively, Abbasi and Malik (2015) studied firm size in terms of market capitalization as a moderator on the relationship between firm performance and growth using a sample of 50 firms trading in Pakistan. The study results established that the interaction term (firm size*growth) significantly influences firm performance hence concluding that firm size enhances the financial performance of a firm.
The existing literature points out much focus on capital structure optimality that has prompted the development of a number of capital structure theories that still have not resolved the global financial distress chronic muddle. Minimal attention has been accorded to the influence resulting from cost of finance or capital. The existing theories in which this study is premised on also provides a theoretical gap on the basis of contradicting postulations. The tradeoff theory insinuates a positive relationship between financial leverage and financial distress after attaining capital structure optimality. In contrast, MM theorem generally suggests that firm value increases with financial leverage. Empirical literature on firm size provides conflicting significance of its main effect on financial performance and its moderation effect on the relationship between firm characteristics and performance. This study envisages to fill the scholarly gap by evaluating the influence of the overall cost of capital (WACC) as well as the individual effects from costs associated with debt and equity on FD while moderating for firm size in a Kenyan perspective.

Research Hypothesis
H01: Cost of capital has no significant influence on financial distress in firms quoted at the Nairobi Securities Exchange. H02: The interaction of cost of capital and firm size does not significantly influence financial distress in firms quoted at the Nairobi Securities Exchange.

Research Methodology 3.1 Research Design and Data
A longitudinal descriptive research design with a correlational approach formed this study's roadmap in testing the developed hypotheses. The study targeted all non-financial firms listed at NSE from January 2006 to December 2015 as shown in Table 1. Financial firms were excluded because they are strictly regulated in terms of capital reservation and liquidity margins. Secondary longitudinal data was extracted from; published audited financial statements, Central Bank of Kenya published financial market rates, NSE handbooks, NSE stock market indices daily stock prices. Automobiles and Accessories 3 3.
Commercial and Services 12 4.

Measurement of Study Variables 3.3.1 Financial Distress (FD)
This study used Altman's Z-score model to compute indices for FD. Z-score appropriateness and prediction accuracy stands the test of time (Carmassi & Patti, 2015;Kazemian et al., 2017;Khaliq, et al., 2014). The selected Altman Z-score model builds on ratios that are representative of the inward and outward business environment thus enhancing distress prediction and classification capability.

Cost of Capital (CC)
This is the overall required rate of return for funds provided by shareholders and creditors in financing business operations (Fabozzi & Drake, 2009). CC at time period 't' was computed as the weighted average cost of capital formulated as: = Weight of debt in capital structure Ke = Cost of equity We = Weight of equity in capital structure Additionally, cost of debt (Kd) was derived from the finance cost of long-term borrowed funds after adjusting for; retirement benefits obligations, any deferred compensation and accrued obligations such as deferred tax and deferred revenue. This was determined from:

Where Kd
= Cost of debt after tax Fc = Finance cost Ld = Long-term debt T = Corporate tax rate Cost of equity (Ke) on the other hand was determined using capital asset pricing model (CAPM). Rossi (2016) upholds that CAPM's objectivity provides an estimate of cost of equity and it is the basis of supplementing other pricing models. Ke was expressed as the expected return on a stock [E (Ri)] derived from:

Results and Discussions 4.1 Descriptive Statistics
The mean Z-score in Table 4 shows that the listed firms are generally safe financially. However, this does not reflect the real picture because on the contrary Table 3 discloses that the percentage of distressed firms rose from 28.9% in year 2012 to 46.2% in year 2015. This therefore signifies that mean as a measure of central tendency is affected by the presence of extreme values in a dataset. Heiman (2011) describes mean as an inaccurate measure when interval or ratio data scales presents a skewed distribution. FD has a standard deviation of 8.638 which also confirms inadequacy of mean due to a high degree of variability. In consensus with Appendix I, Table 3 demonstrates that financial distress has a rising trend across the years the worst being year 2015. Table 4 shows minimal deviance that denotes a modest variation in cost of capital. However, the mean for cost of debt (0.017) is lower than that of cost of equity (0.024) thus implying that debt usage among the listed non-financial firms is a cheaper finance. This can be explained by the fact that debt is commonly pegged on collateral such as fixed assets thus controlling for credit risk unlike equity that is not secured thus shareholders demand for more returns. Furthermore, when a firm is financially constrained, it can opt to cut on dividend payout but still service long term liabilities. Firm size values are descriptively consistent.   Table 5 presents the results from 4 test statistics used to determine the appropriate panel model for the study on the basis of accepting or rejecting the test associated null hypothesis (at 5% level of significance). Conclusively, random effects model was deemed most appropriate.  Figure 1 visually shows normality of the data distribution using a regression standardized residual histogram and normal quartile-quartile (Q-Q) plot. Normality was further checked against Shapiro-Wilk normality test (W-value = 0.91453; P-value = 3.693e-13). When the data volume is large, Shapiro-Wilk test is objectively interpreted based on how close to 1 the W-value is other than P-value that subjectively leads to type I error owing to an inherent bias that increases chances of rejecting the null hypothesis that data is normally distributed (Das & Imon,2016;Field, 2009). W-value is closer to 1 hence closer to unity and therefore concluding that the dataset is normally distributed. . Table 6 shows absence of collinearity between variables because the coefficients of r are far away from being close to perfect correlation (-1 or +1) as well as not within the zones for strong positive or strong negative correlation (0 ≥ r > -0.5; 0 ≤ r < +0.5). Field (2009) confirms multicollinearity when r almost nears perfect correlation index to an extent of being > 0.9. In addition, Variance Inflation Factor (VIF) and tolerance statistic results in Table 7 confirms absence of collinearity (1 ≥ VIF ≤ 5; Tolerance > 0.1). VIF ranging between 1 -5 represents insignificant collinearity (Sporta, 2018). Tolerance value is a reciprocal of VIF thus if < 0.1 collinearity is present (Field, 2009).   Figure 2 shows a scatter plot that confirms linearity is present. The regression standardized residual points do not fan-out in a curvy pattern and instead they lie along the abline. Schreiber-Gregory (2018) emphasizes that as long data is free from outliers, a scatter plot is ideal to test for linearity.  Figure 2 shows absence of heteroscedasticity in that the data is randomly scattered around the zero abline indicating constant variance along the predicted values. Plotting standardized residuals against regression standardized predicted values reliably examines homoscedasticity (Osborne & Water, 2002).

Panel Regression Hierarchical Analysis
The analysis was premised on random effects model selected in section 4.2 and after establishing that the dataset conforms to a linear pattern (section 4.3). The analysis was hierarchical as specified in section 3.4. Notably, the higher the financial distress Z-score, the safer a firm is financially as elaborated in section 3.3. The results from hierarchy 1 are shown in Table 8.

a. Between columns b. Within columns (errors) Dependent variable: Financial distress '***', '**', '*', '.' represents 0.1%, 1%, 5% & 10% significance levels respectively
Model H1M1 shows cost of capital (CC) as the sole predictor of FD to a significant positive relationship on firm safety (β = 16.291; t-value = 4.495; P-value = 0.000 < 0.05 α). This implies that the higher the overall cost of finance, the safer a firm is. Similarly, the cost of capital relates negatively and significantly with FD indicating that a capital structure with both debt and equity portions, results to increased WACC when sourcing for more finance and more so if through borrowing. Reliance on debt exposes equity shareholders to financial risk thus inducing them to demand for more returns. Naidu (2013) establishes that shareholders require a higher return to cater for the risk associated with borrowed funds. Alifani and Nugroho (2013) associates the applicability of this to neutral or risk taking investors only. This is in agreement with Modigliani and Miller's second proposition in that cost of equity increases with rise in debt-equity ratio (Modigliani & Miller, 1963). This study also concurs with the trade-off theory as well in explaining why cost of capital relates negatively with FD on the premise that before optimality of costs and benefits from borrowing, levered firms enjoy tax shield effects that lowers the financial distress costs irrespective of the increased cost of capital owing to a rise in cost of debt that also prompts an increase in cost of equity. The results are in consensus with a study by Ikpesu and Eboiyehi (2018) that found capital structure decisions to relate negatively with FD. Bagga and Kaur (2016) also figured out that cost of capital is inversely related to leverage that corresponds to FD.
Model H1M2 introduces firm size as a second predictor of FD. Hierarchically, a moderator is first introduced as a predictor so as to test for its the main effect prior to establishing the interaction effect it causes (Warner, 2013). In comparison to the first model, cost of capital retains a negative significant association with FD (P-value < 0.05 alpha level) while controlling for firm size. Firm size relates negatively with firm safety although not statistically significant (β = -0.132; t-value = -0.528; P-value = 0.598 > 0.05 α). The 0.0001 ∆R 2 in model H1M2 at least confirms a main effect from addition of firm size into the model. This further means that firm size on its own has a positive influence on FD. In agreement with the findings, Rianti and Yadiat (2018) determined firm size to insignificantly influence FD. However, the positive effect of firm size on FD could mean that firms strive to expand by financing their asset level through debt beyond optimality between tax shield benefits and bankruptcy related costs hence becoming prone to distress. On the same note, Carmassi and Patti (2015) observed that large firms are commonly financed through debt to a great extent. Chancharat (2008) established that large firms that are financially leveraged, stand a higher chance of becoming financially distressed. Notably, firm expansion comes with increased complexity which if not well managed, exposes a business to probability of financial loss and failure. Contrary to this study's findings, Agarwal and Taffler (2008) ascertained that firm size has no effect on financial distress risk. Elsewhere, some studies support firm size to negatively affect FD (Edesiri, 2014;Ikpesu & Eboiyehi, 2018;Makeeva & Khugaeva, 2018).
Model H1M3 shows results of cost of capital interacting with firm size (CC*FS) as a third predictor. The model accounts for zero change in R 2 (∆R 2 = 0.0000) which is also supported by an insignificant interaction term. This implies that there was hardly any interaction effect from firm size and cost of capital on financial distress. Ali et al. (2016) concurs with the interaction results after ascertaining that firm size fails to account for an interaction effect on firm performance with regards to both financial and non-financial aspects. Mutunga and Owino (2017) uncovered inconsistent results that firm size is a positive significant moderator in reference to financial performance. The first and each of the other two models in hierarchy 1 respectively accounts for 86.61% and 86.62% variations in financial distress as shown by R 2 in Table 8. The models significantly fit data well compared to an intercept-only model as evidenced by the F-value of 36.44, 35.67 and 34.89 respectively each with a P-value of 2.2e-16 that is < 0.05 alpha level. Table 9 shows results from hierarchy 2. Operational variables of CC (cost of debt [Kd] & cost of equity [Ke]) were regressed against FD indices while testing for interaction effect of firm size (FS). Table 9: Hierarchy 2 Panel Regression Results

a. Between columns b. Within columns (errors) Dependent variable: Financial distress '***', '**', '*', '.' represents 0.1%, 1%, 5% & 10% significance levels respectively
Model H2M1 concurs with model H1M1 in that Kd (β = -15.905; P-value < 0.005) and Ke (β = 21.386; Pvalue < 0.005) both significantly influence FD. However, Kd has a significant negative association with firm safety this being the equivalent of a significant and positive association with FD while Ke relates positively and significantly with firm safety meaning that the sub-variable has a significant negative relationship with FD. These findings practically explain the role of each cost of capital component with regards to influence on FD. When cost of debt increases, financial leverage is on the rise and therefore a firm becomes more susceptible to financial distress. The results are consistent with past studies showing financial leverage to have a significant and direct association with financial distress (Carmassi & Patti, 2015;Kazemian et al., 2017;Khaliq et al., 2014;Muigai, 2016;Sporta, 2018). Conversely, the study results show that rise in cost of equity reduces the probability of a firm being financially distressed. This implies that a firm that relies more on equity, is safer financially in contrast to one which accommodates more debt in its capital structure. This is in agreement with Muigai (2016) who found out that equity negatively influences financial distress to a significant extent. Kirui and Gor (2018) ascertained the existence of diverse effects of Kd and Ke on financial constraints but failed to account for their significance and directional connotations. The findings further enhance the inference made from model H1M1 on the argument that a capital structure comprising of debt and equity leads to increase in WACC when extra finance is sourced through borrowing. This is so because despite the cost of debt increasing, cost of equity follows suit as a result of equity shareholders demanding for more returns to cater for financial risk associated with more debt financing. This corresponds to Modigliani and Miller's second postulation that cost of equity increases with financial leverage (Modigliani & Miller, 1963).
In reference to model H1M1, model H2M2 shows that Kd and Ke retains their significant positive and significant negative influence on FD respectively while controlling for firm size. On the other hand, firm size has no effect on firm safety (∆R 2 = 0.0000). Agarwal and Taffler (2008) concurs with this finding. Model H2M3 tested on interaction effects from Kd*FS and Ke*FS. ∆R 2 stands at 0.0081 indicating that interaction effect is present. However, interaction term Kd*FS has a negative insignificant influence on firm safety (β = -5.650; P > 0.05) implying that although insignificant, it positively affects FD. This further indicates that high cost of debt increases chances of an entity being financially distressed more so for larger firms. On the other hand, interaction term Ke*FS has a positive significant influence on firm safety (β = 6.989; P < 0.05) implying that it significantly and Notes: -Leverage and total assets related positively in most firms implying that firm size proxied by asset base was financed by debt. However, in the case of A. Baumann & Co. Ltd., debt leverage rose as firm safety and size shrank. -Firm safety had a diminishing trend over time. This implies that financial distress rose across the years towards the approach of 2015. -Some firms such as Marshalls E.A. and Express Kenya opted to dispose assets to repay debt but that never mitigated financial distress as shown in graphs. -The financials of some firms did not feature in every year of the study period to either being listed at NSE after year 2006 (Home Afrika Ltd.) or being suspended (A. Baumann & Co. Ltd. and CMC Holdings).