1 Employee Stock Options, Financing Constraints, and Real Investment Ilona Babenko Michael Lemmon Yuri Tserlukevich Hong Kong University of Science and Technology and University of Utah March 2009
2 Our goals I Focus attention on proceeds from the exercise of stock options as a source of nancing. I I Options as a cost to the rm and a source of cash. Provide signi cant cash in ows and tax bene ts. I Demonstrate that stock options help to allocate funds e ciently. I I Supply of cash is synchronized with investment needs. The optimal compensation structure can mitigate agency problem of overinvestment. I Test whether constrained rms use proceeds for investment. I I Measure the sensitivity of investment to the proceeds from stock options. Regression discontinuity design.
3 Importance of stock options I Aggregate value of stock options granted by U.S. companies grew from $11 billion in 1992 to over $119 billion in 2000 (Hall and Murphy (2003)). I In 2004, Amgen, Corp. received over $450 million from proceeds associated with stock option exercises, as well as $200 million in corresponding tax bene ts. I These cash in ows were approximately 50% of Amgen s capital expenditures for the year. I Outstanding options average 12.7% of outstanding shares. I Average proceeds from stock options exercise are $29M and the associated average tax bene ts are $16M.
4 Cash in ows from option exercises I Cash in ows associated with stock option exercises have a rst order e ect on rms cash ows. How does it happen that they are ignored in the literature? I Proceeds and tax bene ts appear in the Statement of Cash Flows and enter under Cash ows from nancing activities and Cash ows from operating activities, respectively. I Net increase in cash ow does not appear anywhere on the Income statement. I Incentive compensation growth is a relatively recent phenomenon.
5 Cash ow implications of stock options: at the time of the grant. I No cash out ow at the grant date. I Stock options replace wages and save cash for the company. I This e ect has been highlighted in Yermack (1995), Core and Guay (2001), and Hall and Murphy (2003).
6 Cash ow implications of stock options: at the time of the exercise. I Cash in ows at the time of the exercise consist of proceeds from exercise and the associated tax bene ts. I Proceeds: employees are required to pay the strike price at the time of the exercise, even for so-called cashless exercise. I Tax bene ts: the rm receives an associated tax deduction on the di erence between the stock price at exercise and the strike price. See Graham, et al (2004).
7 The use of stock options I Provision of incentives (Hall and Liebman (1998), Murphy (1999)). I Not likely to be a rst order consideration for lower level employees. - Oyer and Schaefer (2005) reject an incentives-based explanation for broad-based option plans. - Suggest sorting (optimistic employees) or retention (Oyer (2004)) are more consistent with the data. I When employees outside opportunities are correlated with rm performance, the use of stock options can minimize renegotiation costs in labor markets (Oyer (2004)). I Our work adds to this literature by relating the cash in ows from stock option exercises to investment.
8 Cash implications of stock options: the big picture t=0 t=1 initial investment options granted options are exercised further investment abandon the project options expire worthless
9 Related literature I Stein (1992) Issuing convertible bonds allows rms to signal quality to the market and issue equity through the backdoor. I Mayers (1998) Convertible bonds control overinvestment incentives and allow for staged nancing by injecting equity into the capital structure and increasing debt capacity at conversion. Convertible bonds do not directly provide cash ow at conversion.
10 Related literature I Stock options are essentially warrants issued to employees. I Sahlman (1990) and Schultz (1993) discuss how VC s and warrants mitigate overinvestment by providing for staged nancing.
11 : Overview I Financially constrained rm that invests at two dates and faces uncertainty. I The rm can borrow a limited amount using investment as collateral. I Investment opportunities are correlated with cash ows and stock prices. I We consider the possibility of overinvestment when investment opportunities are poor.
12 Stock options and nancing constraints I Granting of stock options can substitute for cash wages at the time of the grant. I Yermack (1995), Core and Guay (2001), and Kato, Lemmon, Luo, and Schallheim (2005). I Not clear, however, that employees are the most e cient providers of capital to the rm. - Employee stock options are non-transferable. - Human capital risk. I Will require a higher return to hold options compared to better diversi ed investors.
13 Stock options and nancing constraints I Employees may be e cient providers of capital when either: - Information asymmetries between employees and the rm are smaller than those between the rm and outside investors (Huddart and Lang (2003)). - When employees are more optimistic than alternative investors and strictly prefer the claim o ered by the rm to traded equity (Bergman and Jenter (2007)).
14 Stock options and nancing constraints I We do not attempt to provide a complete theory of why rms grant options or why options might dominate other securities. I We assume that the rm is constrained in its ability to access external nance, but that it can issue stock options at a cost. Employee optimism. Reduced informational frictions. Other unmodeled bene ts of stock options (e.g., Oyer (2004)).
15 I: Setup t=0 t=1 c 0 c 1 borrow qi 0 borrow qi 1 pay back qi 0 and qi 1 save C use C payo price S H H(I H 1 ) + F (I 0) prob p invest I 0 invest I1 H n options + w prob 1 p invest I1 L payo price S L L(I1 L) + F (I 0)
16 II: Additional assumptions I Functions F, H and L are increasing and concave. I Limited collateral B 0 qi 0. I H 0 (I ) > L 0 (I ) -investment opportunities are correlated with demand. I Participation constraint w + zn(s H K ) W. I Assume that salary is paid at the time of the grant. I Risk aversion means z < p.
17 III: Maximization problem Maximize the sum of dividends: max fd 0 + p(d1 H ns H + d2 H ) + (1 p) d1 L + d2 L g n,w,c,i,b s.t. d 0 = c 0 + B 0 C I 0 w(1 T ) 0 d H 1 = c H 1 + C + B H 1 I H 1 + nk + n(s H K )T 0 d L 1 = c L 1 I L 1 + C + B L 1 0 d H 2 = F (I 0 ) + H(I H 1 ) B 0 B H 1 d L 2 = F (I 0 ) + L(I L 1 ) B 0 B L 1.
18 IV: Unconstrained rm Investment of unconstrained rm is not sensitive to the compensation mix. Justi cation: Since constraints do not bind, the unconstrained rm makes rst best investment, i.e., H 0 (I H 1 ) = 1, F 0 (I 0 ) = 1, L 0 (I L 1 ) = 1. Investment is not sensitive to changes in compensation structure, i.e. I1 H I = 1 L n n = I 0 n = 0.
19 V: Investment by a constrained rm Maximize the sum of the dividends, using the participation constraint and budget constraints max ff (I 0) I 0 w(1 w,c,i +p(h(i H 1 ) I H 1 ) + (1 p) the optimal investment levels are determined by: I0 = c 0 C w 1 q I1 H = 1 1 q, L 0 (I L 1 ) = 1 p z )(1 T ) L(I1 L ) I1 L g. c H 1 + C + nk + n(s H K )T.
20 VI: Savings The rst order condition with respect to savings C (no tax, for simplicity) F 0 (I 0 ) = (1 p)1 + ph 0 (I H 1 ). Intuition: Optimal savings sets the marginal pro tability of current investment equal to the expected marginal pro tability of future investment at t = 0. Totally di erentiating this condition with respect to the number of options n yields: C n F 00 (I 0 ) w n = + ph00 I1 H K F 00 (I 0 ) + ph 00 (I1 H ). < 0 if K is su ciently high
21 VII: Investment sensitivity to compensation In contrast to investment of an unconstrained rm, investment of constrained rm increases with in ow of funds from stock options. I0 n I1 H n = = 1 w 1 C 1 q n 1 q n > 0 1 C 1 q n + K > 0
22 VIII: Optimal compensation structure Consider the rst order condition with respect to n: F 0 (I 0 ) 1 I 0 n + (H0 (I H 1 ) 1)p I H 1 n 1 p w z n = 0. I First two terms capture the e ect of stock options on budget constraints. I The last term is due to risk aversion. Risk-neutrality means z = p. I Corner and interior solutions are possible.
23 IX: Agency problem Holding compensation policy xed, with a more severe overinvestment problem (smaller L 0 ), the rm carries lower optimal savings. The rst order condition with respect to C is F 0 (I 0 ) = (1 p)l 0 (I L 1 ) + ph 0 (I H 1 ). I We show that (under mild conditions) the overinvestment problem generally increases the incentives to pay employees with stock options.
24 Sample Preliminary ndings Regressions The sample I Stock option exercises, grants, and outstanding options during I Source: IRRC (aproximately S&P 1500). Hand-collected sample for NASDAQ 100 rms. I Final sample contains 7,116 rm-year observations from 1,416 rms. I We calculate proceeds, tax bene ts, and the Black-Scholes value of the granted options from this data. I Computing tax bene ts is di cult (Hanlon and Shevlin (2001)).
25 Sample Preliminary ndings Regressions Investment measures I 1) Capital Expenditures 2) Research & Development 3) Total Investment I All variables are normalized by lagged book assets. I Why incorporate R&D? I I I Investment in R&D accounts for 1/3 of Total Investment, on average. Investment in R&D may be more sensitive to internally generated funds because of low collateral and high asymetric information. It is inappropriate to view the rm as having separate sources of funds for R&D and CAPEX. I Since R&D is expensed on income statement we add it back to the cash ow.
26 Sample Preliminary ndings Regressions Sample Variable Mean Std. Dev. 25% 50% 75% CAPEX/Asst R&D/Asst Total Investment/Asst Stock Rep./Asst Option Proceeds/Asst Fraction of Non-exec Tax Bene ts/asst Option Grants/Asst SEO Proceeds/Asst Option Proceeds ($M) Tax Bene ts ($M) Option Grants ($M) Option Use
27 Sample Preliminary ndings Regressions Use of options across rms High and Low Users of Stock Options High Users Low Users Variable Mean Std. Dev. Mean Std. Dev. CAPEX/Assets R&D/Assets Total Investment/Assets Stock Repurchases/Assets Option Proceeds/Assets Tax Bene ts/assets Option Grants/Assets
28 Sample Preliminary ndings Regressions Estimating investment sensitivity to option proceeds and tax bene ts. I it A i,t 1 = α i + α t + β 1 Non-Option CF it A i,t 1 + β 2 Q i,t 1 +β 3 Option Proceeds it A i,t 1 + β 4 Tax Bene ts it A i,t 1 + ε it I Regressions also include rm and year xed e ects. I Standard errors are clustered at the rm level.
29 Sample Preliminary ndings Regressions Estimating investment sensitivity to option proceeds and tax bene ts. I Tobin s Q is likely measured with substantial error (e.g., Erickson and Whited (2000)) I Rely on Regression Discontinuity Design (RDD) widely used in labor economics (e.g., Angrist and Lavy (1999), Van der Klaauw (1996)). See also Rauh (2006) paper on mandatory pension contributions. I The identifying assumption is that the relationship between the moneyness of stock options and the exercise proceeds is discontinuous at the point where options are out of the money.
30 Proceeds/assets versus moneyness Average Proceeds Moneyness Figure 1. Stock Option Exercise Proceeds and Moneyness.
31 Investment/assets versus moneyness Average Investment Moneyness Figure 2. Real Investment and Moneyness.
32 Results of the sensitivity regression I Panel A Cash Flow Q Option Proceeds Tax Bene ts Proceeds (Non-exec.) Proceeds (Executives) Expected Proceeds Total Invest (10.47) (8.48) (4.40) Dependent variables CAPEX R&D Stock Total Rep. Invest (8.58) (8.48) (0.85) (8.81) (7.63) (1.66) (4.21) (5.47) (3.08) (0.02) (7.20) (2.78) (2.71) Total Invest (10.32) (7.97) (3.47) (3.44) Total Invest (9.88) (3.48) (2.84) (4.98) Adj.-R Observations 7,106 7,106 7,106 6,586 5,517 6,855 6,724
33 Results of the sensitivity regression II: Total Investment Panel B Cash Flow Q Option Proceeds Moneyness Moneyness 2 Past Return 1 year Past Return 3 years Dependent variable is Total Investment Full Sample Full Sample 0.5<M< <M< (9.49) (9.44) (8.23) (6.69) (7.25) (3.27) (3.59) (-3.39) (7.25) (3.43) (1.51) (1.45) (7.06) (2.44) (5.13) (1.98) Adj.-R Observations 6,724 7,062 5,006 4,007
34 Degree of Financing Constraints Financing Constraints 1. Firm Size Small Firms Large Firms 2. Dividend Payout Zero Dividends Positive Dividends 3. Bond Ratings Below Inv. Grade Inv. Grade and Above 4. CP Ratings Unrated Firms Rated Firms Dependent variable is Total Investment CF Q Opt. Proceeds Obs. Adj.-R (6.84) (10.12) (7.90) (6.87) (8.26) (6.83) (9.17) (4.32) (5.99) (4.46) (6.43) (3.78) (6.97) (2.60) (7.33) (2.21) (3.65) (1.50) (3.87) (1.23) (4.02) (1.55) (4.11) (1.11) 3, , , , , , , ,
35 Alternative Hypothesis Financing Constraints Dependent Variable is Stock Repurchases CF Q Opt. Proceeds Obs. Adj.-R 2 1. Firm Size Small Firms , Large Firms (-0.42) (2.18) 2. Dividend Payout Zero Dividends (0.97) Positive Dividends (0.63) 3. Bond Ratings Below Inv. Grade (0.23) Inv. Grade and Above (0.54) 4. CP Ratings Unrated Firms (0.15) Rated Firms (1.73) (2.44) (2.12) (2.12) (2.91) (1.90) (2.58) (2.19) (2.20) (-1.06) (2.28) (-0.95) (2.50) (-0.45) (1.71) (-0.31) (0.95) 3, , , , , , ,
36 I We show that stock options relax the nancing constraints at the time of the grant and also provide signi cant in ows of cash at the time of exercise. I Stock options allow rms to increase investment precisely in those states where the demand for investment is high. I Test whether nancially constrained rms with larger stock option programs raise less external funds in nancing their new investment than similar rms without such programs. I Test whether rms in which investment demand is more correlated with stock price grant more stock options to employees. I Illustrates an additional bene t of broad-based option plans that has not previously been studied.