Profitability, Inventory Volatility, and Capital Structure

Transcription

1 Profitability, Inventory Volatility, and Capital Structure John R. Birge The University of Chicago Booth School of Business JRBirge UNC KFBS, 21 Sep

2 Themes The traditional tradeoff and pecking order theories differ in the assumed relationship between profitability and leverage Traditional analyses do not recognize the lag between purchasing commitment and product/service revenues Incorporating the lag into a tradeoff model can reverse profitability-leverage relationship An expanded tradeoff model with lag (as in news vendor) produces a relationship that is supported empirically in the cross-section of firms JRBirge UNC KFBS, 21 Sep

3 Outline Basic theories of capital structure Model with commitment and lagged revenue Predictions of the model Empirical results Conclusions JRBirge UNC KFBS, 21 Sep

4 Capital Structure Theories Tradeoff (Modigliani/Miller (1958/1963)) Without market imperfections, firms are indifferent among capital structures (debt vs. equity) Firms trade off the tax advantage of debt and the costs of financial distress to obtain an optimal capital structure Implications Firms with higher profitability can support more debt and, therefore, should have higher market leverage (ratio of debt to the value of the firm) JRBirge UNC KFBS, 21 Sep

5 Pecking Order Theory Myers/Majluf (1984) Firms have private information that is signaled to the market through capital structure Firms prefer first to use internal funds, followed by debt and then equity to finance investments Results As firm profitability increases, firms use less debt and should have lower market leverage JRBirge UNC KFBS, 21 Sep

6 Other Related Literature Capital structure: Jensen/Meckling (1976), Harris/Raviv (1988), Stulz (1988), Brander/Lewis (1986), Tittman (1984), Hart/Moore (1988), Diamond (1991), Hennesy/Whited (2005) Empirical results: Fama/French (2002), Rajan/Zingales (1995) Operations/Financing: Dotan/Ravid (1991), Mauer/Triantis (1994), Lederer/Singhal (1994), Xu/Birge (2004), Hu/Sobel (2007/8), Alan/Gaur (2011) Operating Leverage: Novy-Marx (2011) Inventory effects: Kashyap, et al. (1994), Gaur et al. (2005), Hendricks/Singhal (2005), Chen, et al. (2005), Lai (2006), Roumiantsev/Netessine (2007) JRBirge UNC KFBS, 21 Sep

7 Tradeoff Revisited Missing elements Debt capacity of the firm is determined by expectations of future repayment capability which depends on current decisions Time lag exists between purchasing commitments and realization of revenues Commit for purchase/scale Realize revenues JRBirge UNC KFBS, 21 Sep

8 Basic Model Choose production x t and debt D t (with net needs from equity investors) to maximize the current value of the firm (V t ) that depends on a current production capability state k t, costs for the state and production, future revenues that depend on demand s t and price p t, and with some transition k t+1 (k t +x t,s t ) JRBirge UNC KFBS, 21 Sep

9 Model with Time Lag State: inventory (scale) position at t: k t; Value: V t (k t ) Decisions: Production (x t ), Equity (E t ) and Debt (D t ) Costs: Fixed+holding, etc. (K t (k t; )), Variable Production (c t ) Selling price (fixed/competitive): p t Risk-neutral-equivalent demand dist: F t Dividends (positive or negative): d t Risk-free discount factor: t Rate on debt: r t Default (bankruptcy) point: s b t Breakeven (positive net income) point: s e t Recovery fraction: Tax rate: JRBirge UNC KFBS, 21 Sep

10 Period t Problem Note: Debt paid at end of period (if not in default) JRBirge UNC KFBS, 21 Sep

11 Optimality (FOC) Observations: (1) More impact on objective from incorrectly specifying production than debt; (2) Convex relationship between profitability (c t /p t ) and leverage (D t /V t ) JRBirge UNC KFBS, 21 Sep

12 Observations: Variable Costs Results in single period and multi-period with no fixed costs (Xu/JRB): As margin increases, market leverage (D t /V t ) is convex function of profit margin ((p t -c t )/p t ) Initially, leverage decreases with increasing profit margin, but may eventually increase with high margins Explanation: at low margins, order point is low quantile of distribution; small increases lead to large increases in risk; hence, more costly debt Increase in revenue risk Profit Margin JRBirge UNC KFBS, 21 Sep

13 Fixed Cost Effects As fixed cost increases, the order quantile only changes through changes in bankruptcy point Overall effect is that increasing fixed cost, should raise breakeven point and lower the advantage of debt Net effect is that debt is decreasing as fixed costs increase JRBirge UNC KFBS, 21 Sep

14 Net Income Predictions High Fixed Cost Firms: Possible net losses Higher net losses should have low leverage Low Fixed Cost Firms: Should have positive net income Higher net income should have convex relation to leverage as in variable-cost case JRBirge UNC KFBS, 21 Sep

15 High Fixed Predictions on Net Operating Leverage Margin Low Fixed Total Profit Margin JRBirge UNC KFBS, 21 Sep

16 Other Predictions Inventory variation should be increasing in net income for firms with low fixed costs Variable effects should be larger in industries with lower capital costs For firms with high fixed costs and significant losses, inventory variation may also be high due to high volatility in potential revenues JRBirge UNC KFBS, 21 Sep

17 Hypotheses H1: Firms with operating losses exhibit an increasing relationship between debt-to-market-value ratio and pre-tax operating margin. H2: Firms with low positive operating margins exhibit a decreasing relationship between debt-to-market-value ratio and pre-tax operating margin. H3: Firms with high positive operating margins may exhibit a increasing relationship between debt-to-market-value ratio and pre-tax operating margin, depending on the distribution of demand for the firm's products or services. H4: The volatility of inventories is initially decreasing in operating margin as firm losses decrease to zero and then increases as operating margin becomes significantly positive. JRBirge UNC KFBS, 21 Sep

18 Datasets: Empirical Results Value line (Damodoran) Compustat (CRSP) Approach Cross-section of firms in years Sort net operating margin by deciles to observe nonlinearity Compare test statistics on neighboring deciles for market leverage and inventory Group by industry JRBirge UNC KFBS, 21 Sep

19 Profitability Ratios: Sales v. Assets Here, profitability (EBIT) is normalized by sales as opposed to assets (according to the model) If normalized by assets, because both high positive EBIT firms also have high asset values, those values should be pulled toward the origin creating a single peak JRBirge UNC KFBS, 21 Sep

20 Sales/Assets Comparisons Averages over Strong centering for book values, less for market value JRBirge UNC KFBS, 21 Sep

21 Debt/Value by EBIT/Sales: JRBirge UNC KFBS, 21 Sep

22 Debt/Value by EBIT/Sales: JRBirge UNC KFBS, 21 Sep

23 Example: US 2006 US 2006 Data Max Pre-Tax Market debt to Standard N=5609 Decile Op Margin to capital Deviation Test Statistic % 16.72% 24.78% % 21.02% 29.10% % 26.24% 31.75% % 22.68% 25.72% % 20.39% 24.86% % 18.94% 22.45% % 16.88% 19.13% % 16.02% 19.49% % 17.19% 21.30% % 21.60% 23.45% 3.30 Green: Significant Positive; Orange: Significant Negative JRBirge UNC KFBS, 21 Sep

24 Summary of Annual Observations Consistent significant increases in leverage over lower deciles of EBIT/Sales Consistent significant decreases in leverage for first deciles with increasing positive EBIT/Sales Consistent significant increases in leverage for highest deciles of EBIT/Sales JRBirge UNC KFBS, 21 Sep

25 Industry Effects? Question: Is this due to leverage variation across industries or within industries? JRBirge UNC KFBS, 21 Sep

26 Industries SIC1-5 Note: Low negative EBIT for SIC 5 JRBirge UNC KFBS, 21 Sep

27 Industries SIC 6-9 Note: High variation within decile for SIC 8 JRBirge UNC KFBS, 21 Sep

28 Industry Observations SIC Industries 4, (5), 6, 7, 9 follow the entire Increase-Decrease-Increase pattern of leverage by EBIT/Sales SIC Industries 1, 2, 3 follow the Increase- Decrease pattern of leverage by EBIT/Sales SIC 8 s pattern is least consistent (but has highly skewed leverage values with withindecile means strongly different from medians) JRBirge UNC KFBS, 21 Sep

29 Consistency with Inventory Hypothesis Model description suggests that firms with low marginal costs are at both ends of the EBIT/Sales range Firms with low marginal costs should tend to invest at high fractiles of anticipated demand Their inventory levels should be more volatile than firms in the middle of the EBIT/Sales range JRBirge UNC KFBS, 21 Sep

30 Coeff. of Variation(Inventory/Sales) by EBIT/Sales JRBirge UNC KFBS, 21 Sep

31 CoV(Inventory/Sales) by EBIT/Sales JRBirge UNC KFBS, 21 Sep

32 Overall Results H1: Very strong support for increasing leverage with lower losses H2: Strong support for decreasing leverage as margins rise from zero H3: Support for inter- and intra-industry comparison of increasing leverage at high margins H4: Strong support for U-shaped inventory volatility relationship to margins JRBirge UNC KFBS, 21 Sep

33 Extensions Relationship to asset returns Operating leverage hypothesis predicts higher excess returns for higher operating leverage (lower margins) and less operational flexibility (Novy-Marx (2007)) Use inventory volatility to test role of operational flexibility Include competitive equilibrium JRBirge UNC KFBS, 21 Sep

34 Conclusions Including lags between commitment and realization of revenues can make tradeoff models consistent with empirical observations Models also appear to give consistent predictions for variation in inventory levels Pecking order may still be in effect but may also be consistent with tradeoff theory JRBirge UNC KFBS, 21 Sep

35 Thank you! JRBirge UNC KFBS, 21 Sep

36 US 2005 data US 2005 Data Max Pre-Tax Market debt to Standard N=5225 Decile Op Margin to capital Deviation Test Statistic % 16.67% 25.60% % 21.74% 29.61% % 27.71% 31.82% % 23.77% 25.80% % 23.19% 25.85% % 19.44% 22.97% % 17.46% 20.72% % 15.02% 17.96% % 18.48% 21.49% % 21.77% 24.06% JRBirge UNC KFBS, 21 Sep

37 Average Compustat Data Max Pre-Tax Market debt to Standard N=3318 Decile Op Margin to capital Deviation Test Statistic % 12.74% % 19.37% % 28.56% % 27.07% % 26.89% % 22.69% % 21.23% % 24.56% % 34.42% % 35.30% JRBirge UNC KFBS, 21 Sep

38 Industry First-Digit 1-3 Company Averages: Industry 1 N=177 De cile Market leverage ratio average Test Statistic Company Averages: Industry 2 N=489 De cile Market leverage ratio average Test Statistic Company Averages: Industry 3 N=916 De cile Market leverage ratio average Test Statistic % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % JRBirge UNC KFBS, 21 Sep

39 Industry First-Digit 4-6 Company Averages: Industry 4 N=308 Dec ile Market leverage ratio average Test Statistic Company Averages: Industry 5 N=301 Dec ile Market leverage ratio average Test Statistic Company Averages: Industry 6 N=662 Dec ile Market leverage ratio average Test Statistic % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % 0.18 JRBirge UNC KFBS, 21 Sep

40 Industry First-Digit 7-8 Company Averages: Industry 7 N=324 Company Averages: Industry 8 N=122 Decile Market leverage ratio average Test Statistic Decile Market leverage ratio average Test Statistic % % % % % % % % % % % % % % % % % % % % JRBirge UNC KFBS, 21 Sep

41 Compustat Inventory Coefficient of Variation Decile CoV of Inventory: Mean Test Statistic JRBirge UNC KFBS, 21 Sep