tLUNOlS UBRARY STACKS Digitized by the Internet Archive in 2011 with funding from University of Illinois Urbana-Champaign http://www.archive.org/details/industryclassifi613zumw n, -X . -; Faculty Working Papers College of Commerce and Business Administration Univ*rsity of Illinois at U rba n a - Cho m pa i g n Faculty Working Papers College of Commerce and Business Administration Univarsity of Illinois at U rba n a - Cha m pa ig n FACULTY WORKING PAPERS College of Conmerce and Business Administration University of Illinois at Urbana-Champaign October 8, 1979 INDUSTRY CLASSIFICATION, BUSINESS RISK AND OPTIMAL FINANCIAL STRUCTURE J. Kenton Zumwalt, Assistant Professor, Depart- ment of Finance Tai S. Shin, Virginia Commonwealth University #613 Summary; This study examines the relationships among financial leverage, industry classification and business risk. The results show that some industries exhibit different financial structures but that many indus- tries do not exhibit dissimilar financial structures. The results also indicate there is little relationship between industry classification and business risk measures. Finally, little associations were found between financial structure and business risk. Industry Classification, Business Risk. and Optimal Financial Structure Finance theory indicates that an optimal financial structure should exist in a world where interest payments are a tax deductible expense and market imperfections limit the amount of fixed-income obligations a firm can issue [4] . The optimal level of debt occurs at the point where the marginal benefit of the tax shield due to additional borrowing is exactly offset by the expected marginal cost of financial distress [5]. The probability of financial distress is a function of both the distri- bution of operating income and the level of contractual debt obliga- tions. Firms with high and relatively certain levels of operating income are able to support larger debt obligations than those firms with lower and/or less certain levels of operating income. The optimal financial structure should be related to the business risk faced by the firm. That is, those firms facing high business risk should opt for low financial risk, and vice versa. In recent years several studies have examined the relationship between industry clas- sification and the financial structure of the firm [1,2,3,9,10,11,12,14]. These studies, which assumed industry classification as an adequate proxy for business risk, attempted to determine if firms in different industry classifications exhibit similar financial structures. Of these studies several suggest industry classification is a determinant of financial structure [10,11,12] while other studies suggest industry classification is not related to financial structure [1,9,14]. This paper has two objectives; (1) to shed light on the industry classification-financial structure disagreement and (2) to suggest a more appropriate method than industry classification for determining -2- "equivalent risk class" firms. The first objective will be achieved by performing a more in-depth analysis of the financial structure of the industries under consideration. In addition to considering debt and equity, this study breaks the total debt into current liabilities and long-term debt. Also, preferred stock is separately analyzed. The second objective of developing equivalent risk classes is achieved by utilizing a cluster analysis procedure to group firms according to a measure of business risk. Business Risk and Industry Classification Modigliani and Miller [6] introduced the concept of "equivalent return classes" and suggested these "homogeneous" groups were analo- gous to industry classifications. The groups were assumed to be homo- geneous with respect to the uncertainty attaching to expected operating income, or business risk of the firm. Subsequent to the M & M studies many, if not most, of the studies concerning financial leverage and/or dividend policy assumed industry classification to be an adequate proxy for equivalent business risk. Business risk, defined as the uncertainty inherent future operating income, is a function of the uncertainty of future sales and the operating leverage utilized by the firm. The products and/or services produced by the firm determine not only the firm's industry classifica- tion, but also, to a great extent, the sales potential and the operating leverage of the firm. A firm's operating leverage is deter- mined by the extent of fixed costs associated with its total cost structure. Firms with a high proportion of fixed costs exhibit high operating leverage. -3- Firms will exhibit varying degrees of business risk, due to dif- fering uncertainties regarding sales and/or differing use of operating leverage. The demand for the firm's products influences the uncer- tainty of sales while the asset structure and efficiency of asset uti- lization influences the firm's operating leverage. For example, two firms with similar operating leverage can exhibit different degrees of business risk if one firm is much less certain of its expected sales than the other firm. Alternatively, two firms facing similar sales potential will exhibit different degrees of business risk if one firm is more highly levered operationally than the other. Firms producing different products may have different asset struc- tures resulting in different levels of operating leverage. Also, the type of product or service may influence the minimum size of the firm and the sales potential for the firm. For example, firms in the steel industry are more capital intensive, have a larger minimum size, and exhibit greater sales variability than firms in the food industry. This line of reasoning supports the view that industry classifications may be an adequate proxy for business risk. However, in order for industry classification to be an adequate proxy, it must also be assumed that (1) all firms in the same industry face reasonably similar demand for their products, (2) the technology of the industry requires all firms to have similar type assets, and (3) all firms are reasonably similar in the efficiency of asset utilization. Presumably, if these assumptions hold all firms in the industry will exhibit similar busi- ness risk. -4- Now, assuming optimal financial structures are related to business risk, and assuming firms in a particular industry have similar busi- ness risk, then it can be assumed that industries facing different business risk will exhibit different financial structures. In the real world, however, it is unlikely that these assumptions are met. In any industry many factors exist which may cause differences among the business risk characteristics of the firms. For example, due to product differentiation the demand for two similar products may be different (i.e., name brand vs. generic drugs). Also, different size firms may utilize different operating leverage in that the larger firms may use more capital intensive processes. The cost structure for firms may vary due to different ages of the equipment used. Even if ail firms in the same industry face similar business risk, differ- ences in managements risk-taking preferences can cause the financial structure to be nonhomogeneous within the industry. Business Risk and Optimal Financial Structure In a world with taxes and no bankruptcy costs, M & M [7] have shown that the value of the firm is where X is the expected (perpetual) before-tax operating cash flow, T is the firm's tax rate, D is the market value of the firm's debt, and k and k^ are the capitalization rates for the cash flows of an unlevered firm and for the interest payments on debt, respectively. -5- Ignoring the problem of possible insolvency with associated bankruptcy costs allows the firm to continue to increase its value by issuing more and more debt. In the "real world," however, lenders will not allow a firm to borrow without limit. As financial leverage is increased, the probability of insolvency increases as do any costs associated with possible bankruptcy. Hong and Rappaport [5] defined insolvency as a state in which a firm's operating cash flows are inadequate to meet contractual debt obligations and noted that bankruptcy will follow if the firm is unable to meet its obligations. Denoting the cost of this "financial distress" as insolvency cost, H & R modify valuation equation (1) by adding a term for insolvency cost. V = ^^}~'^^ + TD - k^D (2) k I e where k is the cost of insolvency per unit of debt. Now, as debt is increased the value of the firm is increased by the value of the tax benefits of the interest payments and decreased by the cost of insol- vency, k^D. The optimal capital structure occurs at the point where the marginal tax benefit of additional debt is equal to the marginal insolvency cost of additional debt. Hong and Rappaport assume the insolvency cost, k.^ , is an in- creasing function of financial leverage given the annual operating cash flow distribution. That is kj = f(D X,a), (3) where X and a are mean and standard deviation of the cash flows. Insolvency costs defined in this manner must consider both the cost •^v ■>.^ :5-'0 -6- of bankruptcy and the probability of bankruptcy actually occurring. However, if the business risk is allowed to vary across firms, bank- ruptcy costs are a function of the operating cash flows as well as the level of debt utilized in the firm's financial structure. Now, bankruptcy costs, b, may be defined b = f(D,X,o). (4) This formulation indicates the probability of bankruptcy costs actually being incurred are a function of the level of debt obliga- tions and the level and certainty of the operating cash flows. Since greater operating cash flows allow the firm to take on a greater debt burden, b is a decreasing function of X. However, as in the H & R formulation, b is an increasing function of debt. Also the proba- bility of bankruptcy costs occurring increase as the uncertainty of the operating cash flows, a, increases. In addition to changing the probability of bankruptcy costs, allowing the cash flow distribution to vary across firms will result in different capitalization rates across firms. Review of Previous Studies Several studies have investigated the assumption that financial structure is related to industry classification. Wippern [14] regressed the logarithms of operating earnings against time for a ten-year period. He used the standard error of the regression as a proxy for business risk. He examined 61 firms in 8 industries, and concluded that industry classifications do not discriminate among groups of firms with equiva- lent business risk. However, Schwartz and Aronson [10] used a common equity-to-total asset ratio to describe the financial structure of 32 -7- firms in 4 industries. They concluded industries do exhibit signifi- cantly different financial structures. Gonedes' study [3] had two conclusions; (1) firms in a particular industry do not exhibit similar degrees of business risk and (2) significant differences in business risk between industries do exist. Gonedes used a relative growth rate measure as the business risk proxy. Four studies [1,9,11,12], two of which indicate industry classifi- cation i£ a determinant of financial structure and two of which indi- cates it is not, utilized similar methods of analysis. In these studies a variable designed to reflect the financial structure of the firm was calculated and an analysis of variance procedure was utilized. (Scott and Martin also used the Kruskal-Wallis one-way analysis of variance by rank.) The Scott, and the Scott and Martin studies used common equity- to-total assets as the financial structure variable while Belkaoui used total debt-to-total equity and Remmers, Stonehill, Wright and Beekhuisen (RSWB) used total debt-tototal assets. The number of firms, industries and years examined vary among the studies. Most recently Ferri and Jones [2] employed a cluster analysis pro- cedure to partition firms into 6 distinct classes based on a total debt-to-total asset ratio. A cross-tabulation of financial leverage classes with industry classification showed only a "slight statistical relationship between relative debt structure class and generic industry class." [2, p. 638] Ferri and Jones also reported that a firm's leverage and income variation could not be shown to be associated. All these studies are summarized in Table 1. 00 a> s: -—N M H- I-" n n ^ 3- "O o^ s (D -~J 03 n x_/ ^^ 3 rr N x-s M (?i VO o\ > ON 1-1 >^ o 3 cn o 3 o o 3 m a. n CO ON NO CO ^ s: ?o n M i-( fB o NO p. 3 rr •^oo 3 rr *- =r m ^^ •— ' rr CO H* Q^ V NO •^ W CO NJ (D rr >w/ (D O ?r 3 o* m c 3" > H- H- 0) h-" m K-" 3 "• 00 00 m Si 03 O c NO U1 2 w fli n i-( o rr rr NO -J M (D NO .-( ^ <1 NO h^ '^ CO (f rr C <-, a o ^ 3 (D CO tNj 3 CL c =s= Oi rr O ri Ml H- (B 03 ON M N3 00 o N3 Ul 00 00 VO to N3 ►t] ^ NO NO NO NO 0^ Ln ON N5 U) *• h-" 00 1 w NO NO NO NO NO NO NO NO NO NO NO 0^ Ui ON Ul ^ ^ ON ^ On -vj ON «vl 00 1 CO NO 1 H- O ON U> 00 1 1 M ^~' *< NO NO m ^4 ON 03 ON NO 1 CO > n o O H H 3 o m o O O H-" > 01 m rr 3 TD M O < 2 rt rr NO 3 p^ i-t p. 3 O 3 '^ H" 3 03 03 U1 3 H- 1-1 M O c n 03 o M M -> C rr o O 3 3 O Q rr 9 1 flj =r i-l OO D. 3 l-t H- O a M r-" 3 W (BOO w (B (B NO 1-^ 03 O j2 ?0 3 3 .o o* cr ON M n 1-1 i-h C 03 i-h rr W c rt rr I J Q) o LJ- ^> « ^T* Uk. *^- ^^ H O o* < 03 c 3 3 03 1-1 -9- Methods of Analysis Seven variables describing financial structure and interest obliga- tions and eight measures of business risk were utilized. A parametric analysis of variance and a Kruskal-Wallis one-way analysis of variance procedures were utilized to determine if there were industry differences with respect to financial structure and business risk. The Kruskal- Wallis procedure utilizes rank-ordered data. Next, product-moment and rank-order correlation procedures were utilized to examine the asso- ciation between the financial structure measure and the business risk measures. Also, the relationship among the business risk measures were examined. Finally, a cluster analysis procedure was used to par- tition the firms into groups based on business risk measures. The ANOVA procedure was employed to determine if firms in different busi- ness risk classes exhibited different financial structures. The Sample The firms used in this study were selected to duplicate as nearly as possible those of the Scott and Martin study with two additional constraints; data for each firm had to be available for the entire period and only firms with December fiscal years were considered. These requirements resulted in 198 firms representing 11 industries being examined each year. The common fiscal year requirement should make the firms more homogeneous, as the levels of debt may vary over the year for those firms with highly seasonal sales. Data from the 1966-1976 period were utilized. The data were examined for two five- year periods and for the entire ten-year period. The industry classi- fication numbers and the number of firms in each industry are pre- sented in Table A-1 in the appendix. -10- The Variables The seven financial structure and eight business risk, measures utilized in this study are listed in Table 2. Five of the financial structure variables indicate the proportion of total assets provided by each liability and equity account. The other two financial struc- ture measures are concerned with the interest payments associated with the financial structure. For example, variable seven is the inverse of the familiar interest coverage ratio and indicates the extent to which a firm is able to cover its interest obligations. Variable eight is an average rate of interest over all debt, both short and long term. Depending on the leverage measure used, it is possible that a firm could be highly levered but actually have a relatively low interest obligation. For example, a firm could have a high TD/TA ratio but because much of the total debt is non-interest current liabilities, the firm could have a low interest pajmient obligation. Business risk measures are designed to indicate the uncertainty of future income. Four variables, the standard deviations of the standardized sales growth and standardized growth in earnings before interest and taxes and the coefficients of variation of sales and KBIT, are similar to those used by Ferri and Jones. The other four variables are similar to the variable used by Wippem; that is, the anti-log of the standard error of the estimate around the logarithmic regression of sales and KBIT over the 1966-1976 period (ASEE Sales and ASEE EBIT) and these same variables relative to the mean of sales and of EBIT. The standardized growth variables were calculated as o^q = a[(St--St--i)/S] where t = 1,2,..., 10. This procedure adjusts the measure for differences in size among the firms. See [2, p. 633]. jH.v.i ■: _ •>,L -11- Table 2 Financial Structure and Business Risk Measures Financial Structure Variables 1. Current Liabilities Total Assets 2. Long-Term Debt Total Assets 3. Total Debt Total Assets 4. Preferred Stock Total Assets 5. 6. 7. Common Equity Total Assets Interest Earnings Before Interest and Taxes Interest Total Debt Business Risk Variables 1. Standard Deviation of Sales Growth: a„„ 2. Coefficient of Variation of Sales: y^ , Sales 3. Anti-Log of the Standard Error of the Estimate of Sales (ASEE Sales) 4. ASEE Sales Mean Sales 5. Standard Deviation of EBIT Growth: a„_ 6. Coefficient of Variation of EBIT: Yj.gj^ 7. Anti-Log of the Standard Error of the Estimate of EBIT (ASEE EBIT) 8. ASEE EBIT Mean EBIT -12- The variables listed as financial structure measures were calcu- lated as five-year averages for the 1966-1971 and 1971-1976 periods and as a ten-year average for the entire 1966-1976 period. Because the business risk measures involve standard deviations and standard errors of estimates of regressions, the entire ten-year period was used in their calculation. EMPIRICAL RESULTS Financial Structure Measures Table 3 presents the ANOVA results for both the financial struc- ture measure and the business risk measure. The financial structure variables which are a proportion of total assets are very consistent for both subperiods and the entire 10-year period. All of these variables except preferred stock/total assets showed significant dif- ferences existed among the industry means. Similar results occurred when the variables were rank ordered. Results of the Kruskal-Wallis one-way analysis of variance by rank are presented for the 10-year period only and are denoted by "RO". These results support those reported by Scott and Scott and Martin. Of the interest obligation related variables interest/total debt exhibited a significant F-ratio for the first subperiod while interest/ KBIT was significant only when the data were rank ordered. This sug- gests that while differences exist from a financial structure stand- point, interest obligations relative to KBIT and total debt are not significantly different among industries. While the variables utilized by Scott, and Scott and Martin (CE/TA) and by Belkaoui (TD/TA) both indicate significant differences exist among industries, the ANOVA does not indicate which industries or how 03 < c 03 0) •f-t U J-) CO 3 T3 v£) C r~^ en M OV .H lU c 1—1 0^ 1 J3 > to a; r^ H ^ ^ w o> 1— 1 ui o 3 en > o < >> l-l to E £ 3 O OS va ON rH I ON r^ Ov .H I tT\ 0^ <-> I r^ ON ■K •K * ■K •X ■X ■K ■X •X ■K * * ■K •K •X ■X •<1- lO ro m CNl O CM . ■X •X ■X * •X 1— 1 •H > M JJ -c O 03 W iH > 3 w )— 1 01 l-l J-l U-I 0) w to . iH CO CM U-I oa- CO (U 00 (U •H CO (U lU CO 00 •^ CO u c cd •H )^ to > o CO >N r-l (0 >N CO 3 I lU c o •X ■X •X •X ■X •X * NO rH in nO 00 CN CO CO o o OO iH O iH i>. • I * ■X ■X * •X •X * ■X in CO in ON o p~ rH i-^ CO >— 1 fH CO a- 00 r^ ON CO o CM o o 00 CM o «M 00 •K * CO CO CO CO ^ 1 rH CO M CO 3 U ^ • • 01 r-l i-H J3 > V OJ s tH .H o m rH U-I o o • • CD U 4-1 c • M NO (U (U 03 o CJ c C s CO to CJ CJ QJ •rl •^ u I4H «-i CO •H •H c C= 03 00 00 QJ •H •^ 3 4-1 03 CO rH 4-1 XI CO 03 (U 03 0) > 0) Q OJ U ON v£> o CO C/0 H 1 CN 1 iH CN 1 iH 1 iH 1 1 CN CN 1 O >£) O CO 1-1 o in o O cn o CN 00 ON 00 CO CJN O ON o O 00 o 1 1 1 ^-^ 1 1 CN 1 1 1 >-• iH 1 iH 1 ^ * O 00 o CN CN in in -a- o O ON 00 00 CN o iH CO O O 1 iH 1 1 1 1 1 iH CN 1 1 1 iH 1 1 ^^ m CN CN r^ O ON O CJN LTl 0\ 00 o ON ON ON 00 00 CN o 3- 1-1 ON ON O CN M N l-l »^ u 01 CO XI 3 B ■a 3 C 2 en ^o LTl CTN CO CN I SO rH C CO -H iJ c 0) -H S 2 CO 0) CO • (U iH x: (U w > c (U (U fH V4 CO m a o • dj j: 0) j-i J3 •H (U 3 a c OJ (U l4 u CO LM CO y-i c to o CO ji; o CO c o a 3 vO H ^ CO ON c CO iH CO ca 4J 4-1 CO c c > (U 0) cu 1-H <-l ^4 CO m <-i o o CO • • G o *-l '4H OJ •H ■H "O c c I-l 00 00 o ■H •H CO CO ,M d to CO CO o OJ s J J-l rH o CO D •H ■a o O a u CO Oi c •H b C 0) 0) S 0) CO CU )^ 3 09 CO 0) S 0] * * •K * -K * •K w ^"^ z*^. ^ ,^-N, /'-N y^^ w c r^ cN r^ m C30 CN C» a^ ON r-- O O 00 CU CO r^ M •H O iH I I I I I I ^o o o en O rH I I u u-i rH (n CN 00 v£> CN rH CN 1^ ■vT vO O rH O O o rH CN -vT O ■-i I I I I « ■K •K -K ■K * ^\ ■K * ^^ z*^ * '^ * -^ '^~\ vO vO m O ON rH vO CN a\ u-1 r-~ r-t a^ O CN 00 -d- rH ^ CN 00 o r* r-t CN rH O ^ • • CN • CN O • • o ^ • • rH O • • CJ^ CO • • o o • • * * -K •K * * * * •K /— N ■K -^ * -^ /-^ * /-N * -^ /— N CN O rs. vo vO CN CN vO rn r^ CN vO CO CN o o ^ O -vT m -* vO vD vO CN m O O ON ON rH O ON O ■* ^ o o o Li-i en I I * * I I CN \0 en > OJ OJ (U rH rH ^^ CO m <-i o o CO • • c o (U (U •H JZ ^ u 4-1 4-1 CO rH 4-1 4-1 Ln} . 9V >nj ■* ;-!>u,'^v :■'.-...'.. : ■ -23- Table 7 Financial Leverage Variables 1. Current Liabilities ANOVA Results of the Seven Clustered Groups F-RATIOS Business Risk Measures Total Assets 2. Long-Term Debt Total Assets 3. Total Debt Total Assets 4. Preferred Stock Total Assets 5. Common Equity Total Assets Interest Related Variables 6. Interest EBIT 7. Interest EG 0.614 0.905 0.609 0.358 1.627 2.020 "''EBIT 1.894 2.174* 1.756 0.154 1.352 0.285 1.696 1.215 A-SEE SEE /MEAN e e 2.872* 1.237 2.142 0.899 2.124 2.634* 1.658 3.049** 1.942 0.847 2.587* 1.289 1.999 0.565 Total Debt *Denotes significance at the .05 level. **Denotes significance at the .01 level. -24- The results of this study and the other studies of financial structure and business risk show that much more research needs to be done in the area of business risk definition and measurement. -25- REFERENCES 1. Ahmed Belkaoui, "A Canadian Survey of Financial Structure," Finan- cial Management (Spring 1975), p. 74. 2. Michael G. Ferri and Wesley H. Jones, "Determinants of Financial Structure: A Methodological Approach," Journal of Finance (June 1979), pp. 631-644. 3. Nicholas J. Gonedes, "A Test of the Equivalent Risk Class Hypothe- sis," Journal of Financial and Quantitative Analysis (June 1969), pp. 159-177. 4. Charles W. Haley and Lawrence D. Schall, The Theory of Financial Decisions , 2nd Edition, McGraw-Hill, New York, 1979. 5. Hai Hong and Alfred Rappaport, "Debt Capacity, Optimal Capital Structure, and Capital Budgeting Analysis," Financial Management (Autumn 1978), pp. 7-11. 6. John D. Martin and David F. Scott, Jr., "Debt Capacity and the Capital Budgeting Decision," Financial Management (Summer 1976), pp. 7-14. 7. Franco Modigliani and Merton H. Miller, "The Cost of Capital, Corporation Finance, and the Theory of Investment," The American Economic Review (June 1958), pp. 261-97. 8. Franco Modigliani and Merton H. Miller, "Corporate Income Taxes and the Cost of Capital: A Correction," The American Economic Review , (June 1963), pp. 433-43. 9. Lee Remmers, Arthur Stonehill, Richard Wright, and Theo Beekhuisen, "Industry and Size as Debt Ratio Determinants in Manufacturing Internationally," Finance Management (Summer 1974), pp. 24-32. 10. Eli Schwartz and J. Richard Aronson, "Some Surrogate Evidence in Support of the Concept of Optimal Financial Structure," Journal of Finance (March 1967), pp. 10-18. 11. David F. Scott, Jr., "Evidence on the Importance of Financial Structure," Financial Management (Summer 1972), pp. 45-50. 12. David F. Scott, Jr. and John D. Martin, "Industry Influence on Financial Structure," Financial Management (Spring 1975), pp. 67-73. 13. Sidney Siegel, Non-parametric Statistics , McGraw-Hill, New York, 1956. 14. Ronald F. Wippern, "A Note on the Equivalent Risk Class Assump- tion," The Engineering Economist (Spring 1966), pp. 13-22. D/37 -26- Appendix A-1 Industries in Sample Industry Class 1. Metal Mining 2. Oil-Crude Producers 3. Forest and Paper Products 4. Chemicals 5. Drugs 6. Glass Products and Containers 7. Blast Furnaces and Steel Works 8. Smelting and Metal- working 9. Machine Tools 10. Auto Parts 11. Retail Stores Compustat Codes 1000, 1021, 1031 1311 2400, 2600 2800, 2810, 2820 2835, 2836, 2837 3210, 3221 3310 3330, 3341, 3350 3340, 3550 3714 5311, 5312, 5331 Number of Firms 21 15 20 24 21 11 20 17 20 20 9 198 Faculty Working Papers '.1. 'I- ji.Tiw:>f . i-.y".^ revL.!:/ v>y-.-iyxF^\i-v(ei. ^BMifiBU£iS^^ College of Commerce and Business Administration Univsrtity of Illinois at U r ba n a - C ha m p a i g n '^ou^o|^ 3-9i