The option to quit: The effect of employee stock options on turnover

ARTICLE IN PRESS

JID: FINEC

[m3Gdc;November 9, 2017;13:6]

Journal of Financial Economics 0 0 0 (2017) 1–16

Contents lists available at ScienceDirect

Journal of Financial Economics journal homepage: www.elsevier.com/locate/jfec

The option to quit: The effect of employee stock options on turnoverR Serdar Aldatmaz a, Paige Ouimet b, Edward D Van Wesep c,∗ a

George Mason University, School of Business, 4400 University Dr., Fairfax, VA 22030, United States University of North Carolina at Chapel Hill, Kenan-Flagler Business School, Campus Box 3490, McColl Building, Chapel Hill, NC 27599, United States c University of Colorado at Boulder, Leeds School of Business, 995 Regent Dr., Boulder, CO 80302, United States b

a r t i c l e

i n f o

Article history: Received 23 December 2015 Revised 12 October 2016 Available online xxx JEL: J33 J63 G30

a b s t r a c t We show that in the years following a large broad-based employee stock option (BBSO) grant, employee turnover falls at the granting firm. We find evidence consistent with a causal relation by exploiting unexpected changes in the value of unvested options. A large fraction of the reduction in turnover appears to be temporary with turnover increasing in the third year following the year of the adoption of the BBSO plan. The increase three years post-grant is equal in magnitude to the cumulative decrease in turnover over the three prior years, suggesting that long-vesting BBSO plans delay, instead of prevent, turnover. © 2017 Published by Elsevier B.V.

Keywords: Employee stock options Turnover

1. Introduction

R

We thank Jed DeVaro, Radhakrishnan Gopalan, Shimon Kogan, an anonymous referee, conference participants at the 2012 Western Finance Association Meetings in Las Vegas, Nevada, 2012 Financial Intermediation Research Society meetings in Minneapolis, Minnesota, 2011 Beyster Conference on Employee Ownership, and 2011 Census Research Data Center Conference, and seminar participants at the University of North Carolina and Vanderbilt University. We also thank Bert Grider and Clint Carter for their diligent assistance with the data and clearance requests. Monica Wood, Mark Curtis, and Gang Zhang provided superb research assistance. The research in this paper was conducted while we were Special Sworn researchers of the US Census Bureau at the Triangle Census Research Data Center. Any opinions and conclusions expressed herein are ours and do not necessarily represent the views of the US Census Bureau. All results have been reviewed to ensure that no confidential information is disclosed. This research uses data from the Census Bureau’s Longitudinal Employer Household Dynamics Program, which was partially supported by National Science Foundation Grants SES-9978093, SES-0339191, and TR0427889, National Institute on Aging Grant AG018854, and grants from the Alfred P. Sloan Foundation. ∗ Corresponding author. E-mail address: [email protected] (E.D. Van Wesep).

When firms grant broad-based employee stock options (BBSOs), they provide an incentive for employees to remain with the firm until those options vest.1 BBSO grants thus should be associated with turnover reductions at granting firms. Whether an effect should be seen in practice, however, is unclear. Competitors often poach employees with options by granting a signing bonus to make the employees whole, in which case BBSOs impose costs on competitors, but do not affect turnover. A reduction in turnover

1 A BBSO plan is a grant of stock options to a broad set of a firm’s employees, not solely to executives. As with executive options, strike prices are typically at par when the options are granted, and options vest over time. The average vesting period is three to four years (Crimmel and Schildkraut, 2001; Oyer and Schaefer, 2005), but options may not expire for several years after vesting. The 20 0 0 National Compensation Survey found that options expire 8.9 years after they are granted, on average. A more detailed review of the institutional details and usage of BBSO plans can be found in Aldatmaz and Ouimet (2011).

https://doi.org/10.1016/j.jfineco.2017.10.007 0304-405X/© 2017 Published by Elsevier B.V.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

JID: FINEC 2

ARTICLE IN PRESS

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

may also be temporary, as employees merely delay joining competitors until after their options vest. We exploit a large panel of establishment-level data to show that turnover is reduced following the granting of BBSOs and that the relation appears causal. We show that this reduction in turnover is largely temporary. After the options vest, turnover increases by an amount comparable in magnitude to the prior reduction. As such, BBSO grants appear to be delaying, as opposed to preventing, turnover. The increase in turnover once the options vest suggests that the mechanism through which BBSO grants reduce turnover is associated with anticipation of a deferred compensation award as opposed to a permanent increase in loyalty toward the firm following the granting of stock-based compensation. We also find that BBSO plans are associated with a stronger reduction in turnover in up markets, when growth options are high and retaining key employees is likely to be especially valuable. This provides some support for the Oyer (2004) theory of why granting employees options can be valuable, even if they are insufficient to incent greater effort. Turnover is important. The direct costs of turnover are high. Hansen (1997) estimates that the cost of hiring and training a new worker equals 150–175% of her annual pay. The indirect costs are high as well. Brachet and David (2011) find that turnover accounts for twice as much organizational forgetting as the more traditionally understood skill depreciation. Theory suggests that BBSOs are a natural solution. Typically, these options vest in three years (Crimmel and Schildkraut, 2001; Oyer and Schaefer, 2005), and an employee who departs before her options have vested forfeits their value. Furthermore, as Oyer (2004) shows, insofar as her outside employment opportunities are likely to increase when stock prices are higher, the incentives provided by BBSOs against quitting are likely to be stronger when quitting is more appealing. We connect quarterly, establishment-level US Bureau of the Census data on turnover with ExecuComp data on stock option grants to estimate the effect of implementing a broad-based stock option plan on employee turnover. In the years following a large BBSO grant, turnover falls at the granting firm. We find evidence consistent with causality by exploiting exogenous variation in the value of unvested BBSO grants. If the existence of unvested options reduces turnover, then the effect should be larger when the firm’s stock price rises following the option grant, thereby increasing the value of the unvested options. While we observe this effect in the data, it is not sufficient evidence that the BBSO grant is causing reduced turnover. If a firm’s stock price is predictable for the manager establishing the BBSO plan, then a variety of omitted variables could explain such results. Thus, we instead focus on market returns that occur after the option grant. To the extent that own-firm returns are correlated with market returns, a firm’s option values increase when market returns are high. Moreover, market-wide returns are presumably unpredictable for the manager. This effect should be stronger at firms whose stock price is highly correlated with their peers, and indeed we show that this is the case.

We also provide results consistent with the theory of Oyer (2004) that BBSOs are especially valuable in reducing turnover because the carrot they provide adjusts to the temptation to quit that employees face. We show that BBSOs, on average, lead to relatively greater turnover reductions following positive market returns. This is important because the loss of valuable employees is presumably more costly following positive market returns. Hiring is more difficult when labor markets are tight, suggesting that direct turnover costs are likely higher in good times. Prior work has suggested that BBSOs can improve firm performance by better aligning employee incentives, conserving cash, screening employees, and reducing turnover. Core and Guay (2001) find evidence that firms with higher growth opportunities, high capital needs, and lower access to capital markets use BBSO plans more. Based on this evidence, they conclude that aligning incentives, conserving cash and retaining employees drive the implementation of BBSO plans. Ittner et al. (2003) study new economy firms and argue that BBSO plans are used as a substitute for more expensive means of monitoring employees. Carter and Lynch (2004) show that firms that reprice options subsequently experience lower turnover, as compared with a matched set of firms, with forfeited in-themoney options used as a proxy for turnover. Oyer and Schaefer (2005) calibrate an agency model and conclude that screening employees and reducing turnover can justify the implementation of BBSO plans. Bergman and Jenter (2007) find evidence that BBSO plans are most common among firms with superior prior stock performance, supporting their theory that BBSO plans could be justified as a means to screen employees based on their sentiment about the company’s prospects. Kedia and Rajgopal (2009) show that the implementation of BBSO plans could be motivated by the granting behavior of peer firms. Blasi et al. (2010) find that workers owning stock options report that they are less likely to look for a new job. Hochberg and Lindsey (2010) explore the relation between BBSO plans and subsequent firm operating performance. Instrumenting for the decision to adopt a BBSO plan with peer firms’ BBSO implementation, they find that BBSO plans lead to superior operating performance. Babenko and Tserlukevich (2009) determine that options tend to be exercised when corporate taxable income is high, reducing issuers’ long-run tax bills. Finally, Babenko, Lemmon, and Tserlukevich (2011) find that BBSOs are a good way to raise cash when external financing is costly, and that approximately one third of cash raised by option exercise goes to investment. While reducing turnover has been suggested as one motivation, among many others, in some of the aforementioned studies, none of them has tested the claim directly. Balsam, Gifford, and Kim (2007) perform a case study of one large firm (unnamed) that implemented a BBSO and find that it experienced lower turnover during option vesting periods. Vance (2013), studying the relation of deferred profit sharing and turnover at a large retail firm, finds that more deferred compensation is associated with lower turnover. O’Halloran (2012) shows the same relation across a substantially larger set of firms.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

JID: FINEC

ARTICLE IN PRESS

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

3

Table 1 Variable definitions. Variable

Definition

Turnover

New hires minus change in employment (set to zero if change in employment is negative) divided by the beginning-of-period employment at the establishment, measured quarterly for all full-quarter employees. Indicator that equals one if the firm’s new broad-based employee stock option (BBSO) grants per employee (in Black-Scholes dollar value) exceeds the 95th percentile of the sample distribution. In regressions, this variable takes a value of one if such a large grant occurred in the current year or in the preceding two years. Book value of total assets, normalized to year 2005 dollars. This variable is log transformed in regressions. Log of total assets, normalized to year 2005 dollars. Share price at the end of the fiscal year multiplied by the number of shares outstanding, normalized to year 2005 dollars. Long-term debt plus debt in current liabilities, divided by assets. Ratio of operating income to total assets. Geometric mean of monthly returns for the firm over a set window. (Year 0 indicates the variable was measured as of the year of the BBSO issuance for treated firms or the year of the match for control firms.) Geometric mean of monthly returns for the firm over a set window minus value-weighted market returns over the same window. BNH stands for buy-and-hold. Total option grants at a firm minus grants to top-five executives minus 10% of the average grant to second through fifth executives, divided by the number of employees. Black-Scholes dollar value of BBSOs divided by the number of employees. Annualized 24-month stock return volatility. National annual unemployment rate. Firm beta, estimated with weekly returns relative to the Standard and Poor’s (S&P) 500 over the trailing 52 weeks. (Year 0 indicates the variable was measured as of the year of the BBSO issuance for treated firms or the year of the match for control firms.) Market or industry stock return over trailing 52 weeks. The definition of market or industry is specified by column headers in each table. Returns are measured over a nine-month window, which ends the day before the three month period over which turnover is measured. Industry returns are measured as the median return using three-digit standard industrial classification codes. Dummy variable set to one when the firm’s stock price has an above-median correlation with industry peer stock prices and zero otherwise, estimated with weekly returns relative to industry peers over the trailing 52 weeks. Dummy variable set to one when the firm has an above-median beta and zero otherwise, estimated with weekly returns relative to the S&P 500 over the trailing 52 weeks. Dummy variable set to one when the firm’s stock price has an above-median correlation with the stock prices of other firms in the same state and zero otherwise, estimated with weekly returns relative to peers located in the same state, over the trailing 52 weeks. Current year minus the first year the establishment appears in the standard statistical establishment list.

BBSO

Total assets Size Market capitalization Leverage Profitability Firm BNH returns

Industry adjusted firm BNH returns BBSO shares/employee BBSO value/employee Sigma Unemployment Beta

Return

High industry correlation

High beta High state correlation

Establishment age

The primary contributions of our paper are to establish not only a correlation between deferred compensation and turnover, but also a causal link, and to measure whether turnover is reduced or merely deferred. We provide evidence that BBSO plans delay employee turnover. We find evidence suggesting this relation is causal, thereby supporting the claim that retaining employees is one of the reasons that companies implement BBSO plans. The paper proceeds as follows. Section 2 describes the data. Section 3 presents our evidence on BBSO issuance and subsequent turnover. In Section 4, we show that more successful firms tend to issue BBSOs, but their subsequent success, as measured by stock returns, is not associated with success. Section 5 concludes. 2. Data We use four primary data sources in the analysis. Two databases are from the US Census Bureau: the Longitudinal Employer-Household Dynamics (LEHD) database, containing information on hiring and employment at US estab-

lishments, and the Longitudinal Business Database (LBD), containing information about establishment characteristics, including ultimate owner. We identify BBSO plans through Standard and Poor’s (S&P) ExecuComp database. Compustat provides information on firm characteristics. Definitions of all variables, if not in the text, appear in Table 1. 2.1. Longitudinal employer-household dynamics data The LEHD program has created a unique panel database matching employees to employers and tracking the relation over time. The data are primarily derived from state unemployment insurance records, with information on employee age, gender, wages, hiring, and departure dates.2 The data also include the employer’s geographic location and industry. The data cover all establishments within a state, once the state is covered by the LEHD pro-

2 See Abowd, Stephens, Vilhuber, Andersson, McKinney, Roemer, and Woodcock (2009) for more detail.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC 4

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

gram. An establishment is identified as any separate geographic address at which at least one employee works. For example, a law firm with an office in Boston, Massachusetts, and an office in New York, New York, would have two business establishments. The panel begins in 1992 for several states and the coverage of states increases over time. We use the LEHD to estimate employee turnover. Employee turnover includes all voluntary and involuntary departures from a firm and is estimated as





 

Ti, j,t = Hi, j,t − max Ei, j,t − Ei, j,t−1 , 0 /Ei, j,t−1 ,

(1)

where T is turnover, H is the number of new hires, and E is the level of employment, at firm i, establishment j, and at the beginning of quarter t. Turnover covers both full-time and part-time workers. We limit our sample to full quarter employees, defined as workers employed over the previous quarter, as these employees are most likely to be granted stock options. Turnover is estimated at the establishment level and is measured quarterly. The LEHD reports the hiring, separations, and earnings of workers in each quarter by age and gender. Employee earnings include base salary as well as any other payment taxed as ordinary income, such as bonuses. Two simple cases illustrate how turnover is defined. In the first case, the firm does not change the number of employees at an establishment in quarter_t, but one worker quits and is replaced. Then, Ei, j,t − Ei, j,t−1 = 0 and Hi, j,t = 1, so Ti, j,t = 1/Ei, j,t−1 . As should be intuitive, a fraction 1/E of the workforce turns over. In the second case, the firm hires one more person at the establishment without anybody departing. Then, Ei, j,t − Ei, j,t−1 = 1 and Hi, j,t = 1, so Ti, j,t = E 1−1 = 0. This firm has expanded, not experienced i, j,t−1

turnover.

2.2. Longitudinal business database The LBD is a panel data set that tracks all US business establishments with at least one employee or positive payroll. The database is formed by linking years of the standard statistical establishment list (SSEL) maintained by the Internal Revenue Service of the US Treasury Department. The SSEL contains the federal employer identification number (EIN), physical address, and name of each business establishment that files tax returns in the US. The LBD contains information on the number of employees working for an establishment and total annual establishment payroll (See Jarmin and Miranda, 2002, for more information). The LBD allows us to identify establishment age, based on when the establishment enters the database (establishment age is truncated at 30 years in the data). The LBD also identifies the owner of all establishments. Finally, we use the unique establishment-level identifier assigned by the LBD, the lbdnum, to track firms over our sample period 1994–2005. Business establishments in the LBD are linked to ExecuComp and Compustat through the Compustat-SSEL bridge. 2.3. ExecuComp data To estimate the size of broad-based stock option plans, we follow a methodology similar to Oyer and Schae-

fer (2005). First, we obtain data on the top-five executives’ option holdings and option grants from the S&P ExecuComp database, which provides both the number and the Black-Scholes value of option grants to top five executives. ExecuComp also has the percentage of total employee option grants assigned to the top five executives, a field called PCTTOTOPT. Using this percentage and the BlackScholes option values reported for the executive grants (ExecuComp field OPTION_AWARDS_BLK_VALUE), we estimate the value of stock options granted to all employees at the firm who were not among the top five receiving grants. If an executive is given multiple grants within a fiscal year, the value of options to all employees is estimated separately based on each grant and the average value is taken. This method assumes that the average maturity, timing, strike price, etc., of employee grants are similar to those of executive grants. Because most firms have executives outside the top five, and we wish to subtract their option grants as well, we assume that non–top five executives receive total grants equal to 10% of the average grant awarded to the second through fifth executives. Our estimate of options granted to nonexecutive employees is, therefore, equal to total grants minus the sum of grants awarded to top five executives and ten percent of the average grant made to the top five group after subtracting the grant made to the Chief Executive Officer. This variable is restricted to being positive. We have data only for stock options, not other forms of equity grants. We use an indicator variable to identify those firms with the largest BBSO plans. We define a large BBSO plan as occurring at firm i at time t if the inflation-adjusted Black-Scholes dollar value per employee of new BBSO option grants at firm i in year t exceeds the 95th percentile of the sample distribution. Given the vesting requirements typically associated with option grants, we assume that a large BBSO grant impacts turnover for three years. As such, the BBSO variable is equal to one if the firm had a large BBSO grant this year, last year, or the year before, and zero otherwise. Defining the BBSO variable raises three important considerations. First, we need to capture truly broad-based option grants as turnover is being broadly estimated. Therefore, we restrict attention to the largest grants when it becomes increasingly unlikely that such grants could be allocated to just a small group of top executives. Second, we want to capture the total effect of a large grant on turnover. Because options commonly vest over multiple years, we assign a value of one to our baseline variable so long as options were granted in the year of the observation or in the prior two years. As vesting schedules can allow partial vesting prior to three years, we could imagine that the effect of an option grant is decreasing over time. To allow for this, we also sometimes perform regressions with year-specific indicator variables. Third, we must be sure that our variable is capturing large spikes in option issuance. That is, we must not capture regular issuances from the same firms over time. Fig. 1 shows that roughly two fifths of firms that have at least one BBSO plan in our data have only one. About one quarter have two. Fig. 2 shows that, of the firms with two

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

5

45 40

35 30 Percent

25 20 15 10 5 0

1

2

3

4

5 6 7 Number of jumps

8

9

10

11

Fig. 1. Number of broad-based employee stock option issuances per firm. This figure presents the distribution of the number of BBSO issuances per firm, among firms with at least one issuance in which new BBSO grants exceed the 95th percentile of the sample distribution. Most firms have one or two major broad-based grants in our data.

70 60 50 Percent

40 30 20 10 0

1

2

3 4 Number of years

5

6

Fig. 2. Number of years between broad-based stock option issuances. This figure presents the distribution for the number of years between BBSO issuances, among firms with exactly two BBSO issuances. In two thirds of cases, the issuances are back-to-back.

BBSO grants in ExecuComp, about two thirds issue the BBSOs in back-to-back years, which essentially amount to one large grant over two years. We are therefore confident that our BBSO indicator variable is capturing grants that are unusual for the firm in question and broad-based. An important caveat is that large BBSO grants were especially popular at the peak of the technology boom of the late 1990s. In our sample, the fraction of firms that issue large BBSO grants in a given year rises from approximately 4% in 1997–1998 to 8% in 1999–2001 and falls back to approximately 4% from 2002 to 2005. In our tests, we compare firms issuing BBSOs in a given year with those that do not so this spike in issuance should not drive our results,

but it is important to note that issuance is not constant over time.

2.4. CRSP and Compustat We use Compustat to identify firm characteristics and the Center for Research in Security Prices (CRSP) to measure firm stock performance. We use several control variables that have been suggested in the literature to affect turnover: book leverage, profitability, and size (log of the book value of total assets) from Compustat and firm buyand-hold stock returns and sigma from CRSP.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC 6

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16 Table 2 Summary statistics. The table reports separate summary statistics for firm-years in which firms are defined to be large broad-based employee stock option (BBSO) issuers and this subsample’s complement. The full sample contains all firms in ExecuComp for the years 1995–2005. The subsample of large BBSO issuers has all firm-year observations for firms issuing new BBSO option grants per employee (in Black-Scholes dollar value) that exceed the 95th percentile of the sample distribution in that year or in the preceding two years. The remaining firm-years are both firms that are never large BBSO issuers and firms that simply have not issued a large BBSO this year or in the preceding two. We report means and standard deviations. Variable definitions are in Table 1. Firm-level data Recent issuers of large BBSOs Variable

Mean

Standard deviation

Total assets Market capitalization Leverage Profitability Total employees BBSO shares/employee BBSO value/employee Sigma

4510 1570 12.80% 11.44% 4930 2480 41,910 67.83%

20,720 8290 18.10% 13.46% 12,410 1740 27,0 0 0 22.31%

2.5. Linking data Matching observations across the different data sources is complicated. First, no single identifier presents in the LBD and LEHD that links the two databases. However, the business establishments in the LEHD can be linked to the business establishments in the LBD using crosswalk files provided by the US Census. These files identify business establishments in the LEHD that can be matched to business establishments in the LBD based on EIN, industry, state, and county variables. Unfortunately, two issues arise from matching. First, due to inconsistency in reporting, not all business establishments in the LEHD can be matched to the LBD using EIN, industry, state, and county. Second, if a firm has multiple establishments located in the same county and operating in the same line of business, we cannot do a one-to-one match. Instead, we aggregate the LEHD and LBD data to a meta-establishment, which includes all of the firm’s establishments in that state, county, and industry. We use an iterative process to complete our matching. In each iteration, we identify a match using EIN, state, county, and some definition of industry. In the first iteration, we use the most stringent definition of industry, four-digit standard industrial classification (SIC) (for 1992– 2001) or six-digit North American Industry Classification System (NAICS) (for 20 02–20 05). For those establishments not matched in the first iteration, we attempt a match again using EIN, state, and county but now using threedigit SIC or four-digit NAICS. In the third iteration, we use two-digit SIC and NAICS codes. In the fourth iteration, we use one-digit SIC or NAICS codes. In the final iteration, we match only on EIN, state, and county. The second matching hurdle follows from the differences in the timing and frequency of the observations between data sources. The LBD provides annual data measured as of March 12 each year. To ensure that Compustat data always strictly predates the LBD, we match by picking the most recent fiscal year that ends before March 12 of a

All others Mean

Standard deviation

8980 5890 24.37% 13.98% 18,960 180 2270 39.28%

27,140 14,760 17.09% 8.70% 34,700 390 4240 18.92%

given year to that year of data in the LBD. The Quarterly Workforce Indicators (QWI) is available quarterly based on calendar years. We map all four quarters of a given calendar year in QWI to the same year in the LBD. Table 2 presents summary statistics for our dataset and for the subsample of firms that have recently implemented a large BBSO plan. Issuing firms are smaller, on average, as measured by assets and employees, but they are of more similar market capitalization to firms in the broader sample. They are less levered and feature greater stock price volatility. These properties of large BBSO issuers are consistent with those found in the literature. Our approach combines quarterly data and annual data. Many of our variables and, most important, our variable on BBSO grants, are available at only the annual level. Turnover is measured at the quarterly level. One option would be to estimate turnover at the annual level as well, but this can lead to substantial inaccuracies if an establishment is growing or shrinking over the course of a year. We observe the number of new hires and the total beginning-of-period and end-of-period employment each quarter. This allows us to directly estimate turnover per quarter. Issues arise with annualizing this number, the biggest of which is how to handle establishments that are changing size dramatically within a given year. For example, if an establishment grows from ten to 20 to 30 to 40 to 50 people over four quarters, and in each quarter there are 15 new hires, then the turnover rate in each of the four quarters is 5/10 = 50%, 5/20 = 25%, 5/30 = 16.7%, and 5/40 = 12.5%. Alternately, treating the year as a whole yields 60 hires, a growth of 40 employees, and an initial employment of ten, which gives a turnover rate of (60–40)/10 = 200%. Alternative methods produce alternative turnover rates. When establishments are a relatively constant size, the two methods give similar turnover estimates, but they predictably differ for growing or shrinking establishments. We believe that a more continuous turnover measure would be most appropriate and, because the finest fre-

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

7

Table 3 Average turnover among large broad-based employee stock option (BBSO) issuers and the full sample. This table reports employee-weighted average quarterly turnover among large BBSO issuers and the full sample. The sample contains all establishments owned by firms in ExecuComp for the years 1995–2005 that we were able to match to US Bureau of the Census data and had no missing data on the regression variables. We define a large BBSO issuer as a firm that issued a BBSO in the prior year (as opposed to the prior three years). We aggregate quarterly establishment turnover at the firm-quarter level by weighting each observation by beginning-ofperiod employment. Then, we report the average for four quarters of the calendar year to which this observation matched. If a firm has a fiscal year that ends in January or February, then this would be the calendar year of the fiscal year-end. If the firm has a fiscal year-end in March or later, this would be the subsequent calendar year. Establishment-level data

Turnover mean Number of observations

Sample

Large BBSO issuers

p-value for difference

0.057 2637,053

0.029 25,117

p < .0 0 01

quency in the data is quarterly, we take a simple mean of the quarterly percent turnover to estimate an annual rate of turnover (in the example above, this yields a turnover rate of 26%). The approach of combining establishmentlevel (or even worker-level) data with annual financial data is common among papers using Census data. For example, see Kim and Ouimet (2014), Babina (2015), and Tate and Yang (2015). Still, in unreported results (available upon request), we replicate our baseline results (presented in Table 4) using annualized turnover and find coefficients on relevant variables almost identical to those we present below. 3. The effect of BBSO plans on turnover Firms experience a decrease in turnover after issuing stock options to employees. As shown in Table 3, turnover is significantly lower at firms that have recently issued a large BBSO grant to employees. Quarterly turnover in the full sample is 5.7%, and quarterly turnover in the subsample of large BBSO issuing firms is 2.9%, approximately half of the level in the full sample.3 These data are suggestive that a BBSO can reduce turnover, but they are not conclusive. Firms implementing a BBSO plan are, both observably and un-observably, different from firms that do not implement such plans. In Table 4, we attempt to control for cross-sectional and timeseries variation in the type of firm implementing a BBSO plan by estimating variants of

Ti, j,t = α BBSOi,t−1 + Xi, j,t−1 β + γ ei, j + δ Ti, j,t−1 + εi, j,t .

(2)

BBSOi,t−1 is our indicator for whether a firm has recently implemented a large BBSO plan, Xi, j,t−1 are macro-, firm-, or establishment-level controls, lagged one period, and ei, j are establishment fixed effects. We use quarterly establishment-level observations and cluster our standard errors at the firm level. 3 Average turnover in our sample is lower than Bureau of Labor Statistics estimates of 9%–10% in the US over the same period, calculated as the ratio of monthly turnover (approximately 4.6 million) to average employment (approximately 140 million). This discrepancy is likely due to the fact that our data include only workers employed at the establishment for at least three months (full quarter employees) and workers at large public firms. We report an employee-weighted average as this facilitated Census disclosure.

In all regressions, we use a one-year lag of all of our Compustat-based variables. As described in more detail in Section 2, we match Compustat and the LBD so that the fiscal year-end always predates the LBD observation data. However, due to differences in the timing of the LBD and QWI, this same matching process creates the possibility that QWI data are estimated contemporaneously to Compustat data, for firms with fiscal years ending in January or February. To ensure that Compustat-based variables always strictly predate the observation window in the QWI data, we use one-year lags in all regressions. In Column 1 of Panel A, we include the BBSO indicator variable with additional controls and observe a negative and significant coefficient on the BBSO indicator. Thus, the observed reduction in turnover following the adoption of a BBSO plan reported in Table 3 appears to be due to neither a correlation of BBSO plan implementation with timevarying firm characteristics nor changes in unemployment. In Column 2, we include establishment fixed effects, which control for time-invariant differences at establishments with BBSO plans and those without. We continue to find that turnover at an establishment is reduced following the adoption of a BBSO plan. This result also holds with year fixed effects (Column 3).4 On average, firms experience a 0.48–0.52% drop in quarterly turnover in the three years following the implementation of a BBSO plan. This is an absolute change. If a typical firm in this group experiences 5% turnover, then the effect is to reduce the turnover rate by approximately 10% (9.6% = 0.0048/0.05 and 10.4% = 0.0052/0.05). These results could underestimate the effect of such plans for two reasons. First, our measure of turnover includes all employees, while it is unlikely that even 50% of employees receive BBSO shares. As such, the effect on employees who receive the treatment (get shares) should be at least twice as large as the overall effect estimated. To the extent that the employees receiving shares are those most likely to respond, our estimate for an average effect across all employees at a firm is more accurate. Second, selection issues can bias our results. For example, perhaps in an attempt to stem a tide of departures, managers adopt BBSO plans when they expect turnover to increase. In this

4 In Column 3, we drop the control variable, unemployment, because it is an annual variable and is unidentified once we add year fixed effects.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC 8

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

Table 4 Change in turnover around the adoption of a large broad-based employee stock option (BBSO) plan. This table presents our baseline ordinary least squares estimation results. Panel A covers all establishments owned by firms in ExecuComp for the years 1995–2005 that we were able to match to US Bureau of the Census data and had no missing data on the regression variables. Panel B, Column 1, contains all observations; Columns 2–3 a larger set of control firms (we did not require that lag turnover be non-missing) and a smaller set of treatment firms. Firms with a BBSO at any time are included in Columns 2–3 only if there is non-missing data for BBSO Years 0–3. The dependent variable is turnover for establishment i in quarter t. BBSO is an indicator that takes the value of one if that firm’s new BBSO option grants per employee (in Black-Scholes dollar value) exceeds the 95th percentile of the sample distribution in that year or in the preceding two years. Variables are defined in Table 1. Standard errors are corrected for firm-level clustering. ∗ , ∗ ∗ , and ∗ ∗ ∗ denote statistical significance at 10%, 5%, and 1%, respectively. Panel A: Change in turnover in three years following BBSO adoption Variable BBSO Leverage Profitability Sigma Total assets Log age Unemployment

(1)

(2)

(3)

(4)

(5)

(6)

(7)

−.0269∗ ∗ ∗ (0.0100) −0.0032 (0.0097) .0447∗ ∗ (0.0209) .0011∗ ∗ ∗ (0.0 0 03) −.0058∗ ∗ ∗ (0.0 0 09) −.0032∗ ∗ ∗ (0.0 0 08) −.0032∗ ∗ ∗ (0.0011)

−.0052∗ ∗ (0.0026) .0061∗ (0.0036) −0.0 0 06 (0.0065) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.007) −.0045∗ ∗ ∗ (0.0 0 08) −.0048∗ ∗ ∗ (0.0 0 06)

−.0048∗ ∗ (0.0 0 09) 0.0055 (0.0036) 0.0 0 04 (0.0067) .0 0 03∗ ∗ ∗ (0.0 0 01) 0.0 0 0 0 (0.0 0 06) −.0035∗ ∗ ∗ (0.0 0 09)

−.0136∗ ∗ ∗ (0.0047) −0.0014 (0.0046) .0218∗ ∗ (0.0104) .0 0 05∗ ∗ ∗ (0.0 0 01) −.0028∗ ∗ ∗ (0.0 0 04) −.0018∗ ∗ ∗ (0.0 0 04) −.0016∗ ∗ ∗ (0.0 0 05) .5238∗ ∗ ∗ (0.0166)

−.0051∗ (0.0026) .0060∗ (0.0035) −0.0 0 07 (0.0064) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 07) −.0045∗ ∗ ∗ (0.0 0 08) −.0047∗ ∗ ∗ (0.0 0 06)

−.0053∗ ∗ (0.0026) 0.0056 (0.0036) 0.0 0 02 (0.0067) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 03 (0.0 0 07) −.0046∗ ∗ ∗ (0.0 0 08) −.0048∗ ∗ ∗ (0.0 0 06)

−.0053∗ ∗ (0.0026) 0.0056 (0.0036) −0.0 0 01 (0.0064) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 07) −.0045∗ ∗ ∗ (0.0 0 08) −.0048∗ ∗ ∗ (0.0 0 06)

Lagged turnover

−0.0 0 03 (0.0 0 08)

Beta One-year firm BNH returns One-year industry adjusted firm BNH returns Establishment fixed effects Year fixed effects Number of clusters Number of observations R-squared

−0.0054 (0.0081)

No No 1787 2637,053 0.0972

Yes No 1787 2637,053 0.5611

Yes Yes 1787 2637,053 0.5614

No No 1787 2637,053 0.3431

Yes No 1787 2637,053 0.5610

Yes No 1787 2637,053 0.5610

−0.0056 (0.0063) Yes No 1787 2637,053 0.5611

Panel B: Change in turnover in one, two, and three years following BBSO adoption Variable BBSO Year 0 BBSO Year 1 BBSO Year 2 BBSO Year 3 Leverage Profitability Sigma Total assets Log age Unemployment Establishment fixed effects Year fixed effects Number of clusters Number of observations R-squared

(1)

(2)

(3)

−0.0 0 09 (0.0 0 09) −.0044∗ ∗ ∗ (0.0010) −.0033∗ ∗ (0.0014) .0075∗ ∗ (0.0034) 0.0061 (0.0036) −0.0 0 04 (0.0065) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 07) −.0045∗ ∗ (0.0 0 08) −.0048∗ ∗ ∗ (0.0 0 06) Yes No 1787 2637,053 0.5611

−0.0011 (0.0015) −0.0041 (0.0026) −.0040∗ ∗ ∗ (0.0014) .0077∗ ∗ (0.0038) .0088 ∗ ∗ (0.0039) 0.0 0 08 (0.0062) .0 0 04∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 07) −.0021∗ ∗ (0.0 0 09) −.0045∗ ∗ ∗ (0.0 0 05) Yes No 1824 3087,308 0.5384

−0.0011 (0.0013) −.0046∗ (0.0027) −.0039∗ ∗ ∗ (0.0013) .0073∗ ∗ (0.0036) .0084∗ ∗ (0.0040) 0.0022 (0.0064) .0 0 05∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 07) −0.0010 (0.0010)

Yes Yes 1824 3087,308 0.5386

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

JID: FINEC

ARTICLE IN PRESS S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

case, the coefficient on our BBSO dummy underestimates the magnitude of the effect. In Column 4, we investigate whether the tendency for BBSO issuers to have higher betas could account for the lower turnover. We calculate each firm’s 52-week trailing beta with respect to the S&P 500 and add this value to our baseline regression with establishment fixed effects from Column 3. The coefficient on beta is a precisely estimated zero, and the coefficient on our BBSO dummy variable does not materially change. One could expect that if a firm is doing particularly well, turnover would be lower, as implicit retention incentives (such as the higher likelihood of promotion and the lower likelihood of layoffs) should be stronger. In Column 5, we add the firm’s one-year lagged buy-and-hold returns to our regression from Column 3, to see whether the effect on retention is driven by a firm’s past success. In Column 6, we use instead the firm’s one-year lagged industryadjusted buy-and-hold returns. In each case, while the negative coefficient on lagged returns suggests that this intuition is correct, it is not statistically significant. The coefficient on our BBSO dummy variable is not materially changed.5 Following the work of Babenko, Lemmon, and Tserlukevich (2011) and Babenko and Tserlukevich (2009), we investigate to what degree our results stem from firms’ incentives to issue BBSO grants as a source of financing or tax planning. It would not be surprising if, for example, the ability to tap external capital markets would correlate with a firm’s turnover rate. We therefore include as regression controls a variety of variables from Babenko, Lemmon, and Tserlukevich (2011) and Babenko and Tserlukevich (2009), in an attempt to see how much turnover they explain, independent of the issuance of BBSOs. We include dummy variables for whether a firm is small, has low cash flow, pays a dividend, has zero share repurchases in the past year, or has negative earnings. We also include earnings before interest and taxes (EBIT) volatility and EBIT autocorrelation as continuous controls. All are included separately or in a single regression. The coefficient on our BBSO dummy variable ranges, across ten regressions, from −0.0 0504 to −0.0 0527, all statistically significant at the 5% or 10% level. It does not appear that the correlation of these variables with our BBSO dummy variable is generating the patterns in turnover that we see (results available upon request). In Panel B, we investigate the effect of BBSO implementation at an establishment in each year after issuance. BBSOs should reduce turnover while they remain unvested and, potentially, increase turnover in the year following vesting, as employees who postponed a departure to capture value from their options could now be more likely to

5 In unreported results, we repeat the analyses of Columns 5 and 6 with two- and three-year raw and industry-adjusted buy-and-hold returns. In all cases, the coefficients on lagged returns are negative and insignificant at traditional levels. The coefficients on turnover range from −0.0 0514 to −0.0 0543 and are always statistically significant at the 5% or 10% level. We also include beta with both raw and industry-adjusted lagged buy-and-hold returns, and in neither case does our coefficient on any variable materially change.

[m3Gdc;November 9, 2017;13:6] 9

quit. While we do not observe the vesting period(s) for a given firm’s BBSO issue, this period is typically less than or equal to three years. Year 0 is the year of the large BBSO grant. Year 1 is the year after a large BBSO grant. Year 2 (3) captures two (three) years after the large BBSO grant. If all options vest in the first three years (Years 0–2 in our definition) then Year 3 reflects the first year in which all options are assumed to have vested. In Column 1, the coefficient in year zero is negative, but not statistically significant. The more modest effect could be driven by the fact that the options do not vest for several years, making them less important for decisions today. It can also be due to the fact that the stock price is unlikely to have moved substantially since the grant was issued. If employees underestimate the value of options that are not in-the-money, then this would predict a modest or nil effect on turnover immediately following a grant. In Years 1 and 2, the effect is clearly negative. Turnover is reduced in the two years following the year of the BBSO grant. As one would expect if options partially vest in each year, the effect in Year 2 is less than in Year 1, though the difference is small and not statistically significant. In the third year after the year of the grant, when most or all of the options are likely to be vested, a significant increase in turnover appears, consistent with the view that employees time their departures to take place after full vesting. Therefore, managers should expect an increase in turnover after vesting and plan accordingly. In Columns 2 and 3, we repeat the regression in Column 1 but use only a subset of the firms with BBSO plans. For firms that have a BBSO at any time, we require that the firm have non-missing information for Years 0, 1, 2, and 3. We retain all control firm-year observations used in Column 1 and include additional control firm-year observations that were previously dropped from the sample due to missing data for lagged turnover (We added these control observations to facilitate disclosure of this sample). Column 2 includes controls for national unemployment rates. Column 3 includes year fixed effects. We find similar results, indicating that differences in the sample of BBSO firms reflected in each BBSO year dummy was not driving the result in Column 1. The reversal three years after a grant indicates that the majority of the effect of BBSO plans on turnover is simply to delay turnover. We find, on average, that 87% of the reduction in turnover observed in the first three years (Years 0, 1, and 2) is reversed in the subsequent year. Furthermore, a number of firms issue multiple large BBSO grants. To the extent that a second large BBSO grant timed years after the first grant reduces future turnover, we are underestimating the reversal. In unreported results, we repeat the regression in column 2 but exclude firms with multiple BBSO grants and find an even more dramatic reversal. While this result is consistent with an underestimation of the magnitude of the reversal, it also introduces a lookahead bias. Firms that do not issue subsequent large BBSO grants could be unique and have higher future turnover. In unreported results, we investigate whether threeyear vesting appears to be the norm. We cannot observe any particular plan’s vesting schedule, but we look for an effect on turnover after Year 3. We include a dummy

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

JID: FINEC 10

ARTICLE IN PRESS

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

variable for Year 4 after a BBSO grant and find economically and statistically insignificant coefficients in all three regressions. If we add a dummy variable for BBSO Year 4 to the specification of Column 3, we find a coefficient of 0.0 0 02 and a p-value of .93. Similar coefficients and pvalues result with specifications consistent with Columns 1 and 2. The positive effects on turnover are limited to Year 1 and Year 2, and the negative effect is limited to Year 3, after a grant is issued. We also consider just a subset of firms in the higher human-capital industries of technology and biotechnology, under the assumption that their vesting periods could be substantially different from those of other firms. A firm is defined as Biotech if its primary SIC code is 2830–2839, 3826, 3841–3851, 5047, 5048, 5122, 6324, 7352, 80 0 0– 8099, or 8730–8739 excluding 8732. A firm is in the tech industry if its primary SIC code is 3570–3579, 3600–3629, 3643, 3644, 3670–3699, 3825, 5044, 5045, 5063, 5065, 5734, or 7370–737. If we estimate the level effect on the coefficients for BBSO Year 0, BBSO Year 1, BBSO Year 2, and BBSO Year 3, coefficients remain negative and significant in Year 1 and Year 2, and positive and significant in Year 3. If added, the coefficient on BBSO Year 4 remains very insignificant. This suggests that three years appears to be typical across industries, at least for the firms and years in our sample. Across all regressions, three further results are apparent. First, as previously shown by Lane, Isaac, and Stevens (1996) and Porter and Steers (1973), larger firms and older establishments experience lower turnover. Second, as a firm’s stock market volatility (sigma) increases, turnover at its establishments increases as well. This fact is robust across specifications and could be because these firms are experiencing significant changes. Third, higher unemployment is associated with lower turnover, as would be expected. These results are largely consistent with prior work. Together, we find results consistent with a reduction in turnover stemming from the implementation of a broadbased employee stock option plan, but we have not shown causality. A variety of stories regarding the choice of the manager to implement a BBSO plan are also consistent with these results. For example, perhaps managers who anticipate that the firm will soon experience a period of success are more likely to share this with employees by adopting large BBSO plans. Because turnover is likely lower during such periods, turnover would be lower following the adoption of a BBSO plan. 3.1. The effect of BBSO plans on turnover in up versus down markets The negative association between the implementation of a BBSO plan and turnover is larger when industry or market returns are high during the post-issuance, prevesting period. As argued in Oyer (2004), industry (market) returns reflect labor market conditions. Higher industry (market) returns imply better outside employment opportunities for employees with industry-specific (general) human capital. Assuming that a firm’s stock price is correlated with those of its peers, the value of previously

granted options is higher when the returns of its peers are high. Thus, options provide a cost to quitting that increases during periods when outside employment options are greatest. This argument suggests an approach to identify evidence consistent with a causal effect of BBSO plans on employee turnover. Assuming managers cannot predict industry or market returns, realized returns that occur after the options have been issued, but before they have vested, reflect exogenous shocks to the value of unvested options and, therefore, the expected impact of BBSO plans on turnover. Firms with BBSO plans should see greater relative declines in turnover, as compared with firms without BBSO plans, when industry or market returns are higher. The relation is weaker if firms reprice their granted options following weak performance, biasing our results toward zero. However, this is unlikely given that Carter and Lynch (2001) find that firms are unlikely to reprice options in response to poor industry performance. Three assumptions are critical to this identification. First, industry and market returns must correlate positively with firm returns. In Section 2.2, we use this assumption to further refine the test. Second, industry and market returns must be unpredictable for the manager. This cannot be shown conclusively, but it would be surprising if managers were able to forecast industry/market returns with such a high degree of confidence that they use these expectations to determine the timing of BBSO plans. Third, the correlation of other (dis)incentives for turnover with industry and market returns must be similar for BBSO and non-BBSO issuing firms. That is, one can imagine that firms issuing BBSOs are high beta. In this case, while a strong market would raise outside opportunities for all firms’ employees, the effect would be muted for BBSO issuers because implicit retention incentives would be unusually high. We discuss in Section 2.3 that this does not appear to be a driver of our results. To test whether decreases in turnover following the implementation of a BBSO plan are greater in magnitude when the market performs well, we use a difference-indifferences type methodology. We interact measures of industry and market performance with our indicator variable, BBSO. If the effect of a BBSO grant is larger in up versus down markets, then the coefficient on this interaction term should be negative. In the absence of one obvious measure of market performance for our tests, we use three different variants: equal- and value-weighted market returns and industry median returns. We define industry at the three-digit SIC code level. To ensure that the BBSO plan was implemented before the returns were realized, we continue to lag the BBSO indicator variable by one year, as in Table 4. We estimate returns over a nine-month window, beginning at the end of the period over which BBSO is measured and ending at the beginning of the period over which turnover is measured. We require the returns to be estimated over a period of time that strictly follows the window for estimating allocated BBSO shares to estimate the effect of unanticipated returns at the time of the BBSO issuance. We require the returns to be estimated over a period of time that strictly precedes the window for estimating turnover

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

11

Table 5 The effect of broad-based employee stock option (BBSO) grants on turnover in up versus down markets. The dependent variable is turnover for establishment i in quarter t. The sample contains all establishments owned by firms in ExecuComp for the years 1995 to 2005 that we were able to match to US Bureau of the Census data and had no missing data on the regression variables. BBSO is an indicator that takes the value of one if that firm’s new BBSO option grants per employee (in Black-Scholes dollar value) exceeds the 95th percentile of the sample distribution in that year or in the preceding two years. Variables are defined in Table 1. Standard errors are corrected for firm-level clustering. ∗ , ∗ ∗ , and ∗∗∗ denote statistical significance at 10%, 5%, and 1%, respectively. Return definition

Variable

−.0049∗ ∗ (0.0022) .0049∗ ∗ ∗ (0.0014)

BBSO Return BBSO

∗

Value weighted (1)

Profitability Sigma Total assets Log age

Equal weighted (3) −.0050∗ ∗ (0.0022) .0034∗ ∗ ∗ (0.0 0 09)

Equal weighted (4)

.0096∗ ∗ ∗ (0.0035) 0.0095 (0.0072) .0 0 06∗ ∗ ∗ (0.0 0 01) −0.0 0 0 0 (0.0 0 06) −.0067∗ ∗ ∗ (0.0 0 09)

−.0050∗ ∗ (0.0022) .0049∗ ∗ ∗ (0.0014) −0.0081 (0.0053) .0096∗ ∗ ∗ (0.0035) 0.0096 (0.0072) .0 0 06∗ ∗ ∗ (0.0 0 01) −0.0 0 0 0 (0.0 0 06) −.0067∗ ∗ ∗ (0.0 0 09)

.0090∗ ∗ ∗ (0.0035) 0.0091 (0.0073) .0 0 06∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 06) −.0074∗ ∗ ∗ (0.0 0 09)

−.0047∗ ∗ (0.0022) .0034∗ ∗ ∗ (0.0 0 09) −.0057∗ (0.0033) .0090∗ ∗ ∗ (0.0035) 0.0091 (0.0073) .0 0 06∗ ∗ ∗ (0.0 0 01) −0.0 0 02 (0.0 0 06) −.0074∗ ∗ ∗ (0.0 0 09)

Yes

Yes

Yes

1713 2252,857

1713 2252,857

0.5829

0.5829

Return

Leverage

Value weighted (2)

Yes

.0062∗ (0.0035) −0.0011 (0.0067) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 01 (0.0 0 07) −.0043∗ ∗ ∗ (0.0 0 08) −.0049∗ ∗ ∗ (0.0 0 06) Yes

1713 2252,857

1713 2252,857

1713 2252,857

1713 2252,857

0.5829

0.5829

0.5831

0.5831

to avoid reverse causality interpretations if market returns reflect contemporaneous turnover. Table 5 presents the results of this analysis. In all six columns, when the market or industry has performed well, turnover is, on average, higher, suggesting greater outside employment options for employees. Columns 2, 4, and 6 include interaction terms. In each case, with statistical significance in Columns 4 and 6, the existence of a BBSO plan has a larger effect on subsequent turnover when industry and market returns are high after the fact. This is consistent with a causal connection between BBSO plans and subsequent turnover and inconsistent with many alternative selection stories.

3.2. The effect of BBSOs at firms whose performance is more or less correlated with peers Because we are using industry or market returns as a proxy for the relative value of unvested options, our argument relies on a positive correlation between the performance of the issuing firm and its competitors. BBSOs should affect turnover more when a firm’s share price has appreciated but, to avoid noncausal interpretations of our results, we use industry or market returns as a proxy for firm returns. If BBSOs are driving lower turnover, then this proxy should work better if a firm’s returns are more correlated with those of peers.

−.0047∗ (0.0028) .0034∗ ∗ (0.0017)

Same industry only (6) −.0061∗ ∗ (0.0026) .0036∗ ∗ (0.0018) −.0075∗ (0.0042) .0062∗ (0.0036) −0.0011 (0.0067) .0 0 03∗ ∗ ∗ (0.0 0 01) −0.0 0 01 (0.0 0 07) −.0043∗ ∗ ∗ (0.0 0 08) −.0049∗ ∗ ∗ (0.0 0 06) Yes

Unemployment Establishment fixed effects Number of clusters Number of observations R-squared

Same industry only (5)

In Table 6, we introduce dummy variables for whether a firm’s share price is highly associated with the prices of peers. We calculate a firm’s correlation with peers in two ways, both defined in Table 1. First, we use weekly stock price data for each firm and an index of its industry peers in the 52 weeks prior to the BBSO date, and we calculate the correlation between the two. If the correlation is above the sample median, we define the high correlation dummy variable to equal one and zero otherwise. Using the same stock price data, we estimate each firm’s beta relative to the S&P 500 in the 52 weeks prior to BBSO issuance and create a high beta dummy variable that equals one if the firm’s beta is above the sample median and zero otherwise. In Column 1, we regress establishment-level quarterly turnover on the same right-hand-side variables (suppressed for clarity) as in Table 5, as well as the valueweighted market return, the high correlation dummy, and the high beta dummy. We then create a full set of interactions of the BBSO grant dummy variable, the market return, and the high industry correlation dummy variable, as well as a full set of interactions for the BBSO grant dummy variable, the market return, and the high beta dummy variable. If a BBSO is reducing turnover, and the reduction is higher when the firm’s shares are worth more, then each of these triple interactions should have negative signs. If a firm’s returns do not correlate with market returns, then the fact that the market return has been high would not

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC 12

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

Table 6 The effect of broad-based employee stock option (BBSO) grants on turnover in up versus down markets, depending upon the association of the firm’s stock price with the prices of peers. This table presents results of the association of turnover and BBSO implementation, depending upon the correlation of a firm’s stock price with industry peers, with its geographic peers, and its beta. The dependent variable is turnover for establishment i in quarter t. BBSO is an indicator that takes the value of one if that firm’s new BBSO option grants per employee (in Black-Scholes dollar value) exceeds the 95th percentile of the sample distribution in that year or in the preceding two years. Variables are defined in Table 1. Leverage, Profitability, Sigma, Total Assets, and Log Age are included as controls in all regressions, but coefficients are unreported for clarity. High Beta, Returns, High State Correlation, and all two-way interactions of these three variables are also included and unreported. Standard errors are corrected for firm-level clustering. ∗ , ∗ ∗ , and ∗ ∗ ∗ denote statistical significance at 10%, 5% and 1%, respectively. Return definition

Variable BBSO ∗ Return ∗ High Industry Correlation BBSO ∗ Return ∗ High Beta BBSO ∗ Return ∗ High State Correlation Establishment fixed effects Firm controls Number of clusters Number of observations R-Squared

Value weighted (1)

Value weighted (2)

Equal weighted (3)

Equal weighted (4)

Same industry only (5)

Same industry only (6)

0.0084 (0.0113) −.0196∗ (0.0117)

0.0080 (0.0107) −.0185∗ (0.0108)

−.0138∗ ∗ ∗ (0.0050) −.0104∗ (0.0056)

−.0135∗ ∗ ∗ (0.0049) −.0098∗ (0.0059)

−.0167∗ ∗ (0.0067) −0.0063 (0.0067)

−.0164∗ ∗ ∗ (0.0060) −0.0045 (0.0066)

Yes Yes 1787 2637,053 0.559

−0.0059 (0.0106) Yes Yes 1787 2637,053 0.559

Yes Yes 1787 2637,053 0.559

0.0 0 02 (0.0077) Yes Yes 1787 2637,053 0.559

affect the value of an employee’s options and, therefore, would not be expected to reduce turnover. The more the firm’s returns correlate with the market’s returns, the more turnover should be reduced when the market is doing well. Column 3 replicates the analysis of Column 1, but it uses an equal-weighted market return instead of a valueweighted return. Column 5 replicates the analysis but uses industry returns instead of market-wide returns. We observe negative coefficients on both triple interactions in all regressions, though statistical significance is stronger in some than others. In Columns 2, 4, and 6, we replicate the analyses of Columns 1, 3, and 5 but add a set of interaction terms built with a dummy variable for whether a firm’s returns have an above-median correlation with the returns of other firms with headquarters located in the same state. This variable is motivated by Kedia and Rajgopal (2009), who find that firms grant more BBSOs when geographically proximate firms do as well. The inclusion of this variable does not substantively affect the coefficient estimates from Columns 1, 3, and 5, and the coefficient on the triple interaction is of varying sign and never statistically significant. This suggests that correlation with the industry or overall market, not correlation with proximate peers, is associated with BBSOs having a greater effect on turnover. 3.3. Alternative interpretations The results in Table 6 are consistent with BBSO plans having a causal impact on turnover. However, these same results can also be explained by an omitted variable if firms that adopt BBSO plans have different sensitivities to market returns. For example, if BBSO firms have, on average, stock returns that are more correlated with the market

Yes Yes 1787 2637,053 0.559

0.0023 (0.0080) Yes Yes 1787 2637,053 0.559

than non-BBSO firms, then they perform better than their peers in up markets. This better performance can provide greater career opportunities for employees and result in lower turnover. To better account for potential differences between BBSO and non-BBSO firms, we use a matched sample analysis, in which up to three control firms are matched to each BBSO grant firm in the year of the grant (i.e., Year 0). This step reduces our sample size to 842,915 observations. We draw control firms from the sample of all firms in ExecuComp that can be matched to the Census data, but that do not issue a large BBSO grant during our sample period. We require matched firms to operate in the same threedigit SIC code and in the same year as the BBSO granting firm to which they are being matched. We then pick the three firms with values for total assets closest to the total assets of the BBSO issuer. We allow control firms to match to more than one BBSO issuer. In Table 7, Column 1, we repeat the regression of Table 5, Column 2, but use our matched sample. As before, firms with recent BBSO grants experience lower turnover than their matched peers, and the association is even more pronounced when the market has done well since the grant. One could be concerned that firms differ from their matched peers in cyclicality. If they are more cyclical, then they have larger drops in good times for a reason unrelated to a BBSO grant. To investigate this possibility, we include in Column 2 the firm’s beta, calculated in the 52 weeks prior to the BBSO grant, and an interaction of its beta with the market return. Both coefficients are positive, though not statistically significant. If anything, this suggests that higher beta firms experience higher turnover, especially in good times, which runs contrary to the concern. It seems unlikely that unobserved variation in betas is causing our results.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

13

Table 7 The effect of broad-based employee stock option (BBSO) grants on turnover depending on beta and pre-grant performance. This table presents results of the association of turnover and BBSO implementation, depending upon the firm’s stock price performance and its beta in the year before the grant. The dependent variable is turnover for establishment i in quarter t. BBSO is an indicator that takes the value of one if that firm’s new BBSO option grants per employee (in Black-Scholes dollar value) exceeds the 95th percentile of the sample distribution in that year or in the preceding two years. Variables are defined in Table 1. Leverage, Profitability, Sigma, Total Assets, Unemployment, and Log Age are included as controls in all regressions, but coefficients are unreported for clarity. Standard errors are corrected for firm-level clustering. ∗ , ∗ ∗ , and ∗ ∗ ∗ denote statistical significance at 10%, 5%, and 1%, respectively. Return definition

Variable BBSO Return BBSO

∗

Return

Value weighted (1)

Value weighted (2)

Value weighted (3)

Equal weighted (4)

Equal weighted (5)

Equal weighted (6)

−.00579∗ (0.0031) 0.0030 (0.0020) −.0111∗ ∗ (0.0047)

−0.0054 (0.0033) −0.0 0 09 (0.0030) −.0125∗ ∗ (0.0052) 0.0061 (0.0041) 0.0038 (0.0034)

−.0058∗ (0.0031) 0.0029 (0.0019) −.0114∗ ∗ (0.0049)

−.0056∗ (0.0031) −0.0011 (0.0016) −0.0047 (0.0029)

−0.0051 (0.0033) −0.0025 (0.0020) −.0064∗ (0.0033) 0.0066 (0.0046) .0035∗ (0.0021)

−.0056∗ (0.0031) 0.0010 (0.0014) −.0051∗ (0.0030)

Beta (Year 0) Return

∗

Beta (Year 0)

One-year Firm BNH Returns (Year 0) Return ∗ one-year Firm BNH Returns (Year 0) Establishment fixed effects Firm controls Number of clusters Number of observations R-Squared

Yes Yes 522 842,915 0.636

Yes Yes 522 842,915 0.636

An alternative concern is that firms that perform well are more likely to grant BBSOs and are also likely to experience lower turnover than their matched peers. This could explain the negative coefficient on the BBSO × Return interaction in a variety of ways. In Column 3, therefore, we include the firm’s past 12-month return at the time of the grant or time of the match and an interaction of this return with market returns. Again, positive coefficients result, though the t-statistics are far from statistically significant. If anything, this runs counter to the concern. We repeat these analyses in Columns 4–6, using equalweighted return measures, and find similar results. We conclude that these facts are difficult to reconcile with noncausal stories associating the implementation of BBSO plans with declines in turnover and, therefore, lend credence to the causal argument. They also support the Oyer (2004) argument in favor of deferred equity-based compensation over deferred cash pay. In unreported results, we include market-to-book ratio as a control variable in all regressions. Its inclusion has little effect on the coefficient on our BBSO dummy variable. If we separate our data into firms with high and low market-to-book ratios, we find that the effect of BBSOs on turnover is somewhat stronger among firms with higher market-to-book ratios. 3.4. Alternative specifications for large BBSO issuance To this point, we have relied upon specifications in which firms are defined as having issued a broad-based stock option grant if the Black-Scholes values of options

0.0283 (0.0897)

0.0241 (0.0909)

0.0366 (0.0681) Yes Yes 522 842,915 0.636

0.0410 (0.0472) Yes Yes 522 842,915 0.636

Yes Yes 522 842,915 0.636

Yes Yes 522 842,915 0.636

that it issues in a given year exceed 95% of the sample distribution. Nothing is particularly special about 95%. We simply need the grant to be very large so that it is probably broad-based. Alternative thresholds are reasonable and could be industry- or even firm-specific. We perform regressions identical in specification to the regression in Column 2, Table 3, but with different thresholds defining how big a BBSO must be to count as broadbased. If we choose a threshold of 60%, 80%, 120%, or 140% of the size of our baseline threshold, then the resulting coefficients range from −0.0024 to −0.0060, with tstatistics ranging from −2.19 to −3.14. For brevity, each full regression is untabulated, but coefficients and t-statistics are available upon request. As the threshold for what qualifies as a large BBSO increases, the impact on turnover increases as well. Giving employees more unvested shares reduces turnover more. In Table 8, we repeat the analysis of Table 4, Panel A, but replace our BBSO dummy variable with a continuous variable, equal to the log of the dollar value of options per employee granted in the last three years if that value was positive and zero otherwise. Results are consistently negative and statistically significant at the 95% level. Conditional on being identified as having a BBSO plan, bigger plans have a bigger impact. If we choose an industry-specific threshold, in which a firm must have a grant of at least 95% of the sample distribution within its three-digit SIC industry, then the coefficient is −0.0043, with a t-statistic of −1.68. Using a four-digit SIC industry classification, the coefficient becomes −0.0052 with a t-statistic of −2.18.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC 14

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

Table 8 Change in turnover around the adoption of a large broad-based employee stock option (BBSO) plan with a continuous measure of BBSO size. This table presents our baseline ordinary least squares estimation results from Table 4, but with the log of BBSO dollar value per employee instead of a dummy variable. If the BBSO grant value per employee is not positive, the variable Log BBSO Value takes a value of zero. The dependent variable is turnover for establishment i in quarter t. Variables are defined in Table 1. Leverage, Profitability, Sigma, Total Assets, and Log Age are included as controls in all regressions, but coefficients are unreported for clarity. Unemployment is included in all regressions except Column 3. Standard errors are corrected for firm-level clustering. ∗ , ∗ ∗ , and ∗ ∗ ∗ denote statistical significance at 10%, 5%, and 1%, respectively. Variable Log BBSO value

(1)

(2)

(3)

(4)

(5)

(6)

−0.0022∗ ∗ (0.0 0 09)

−.0 0 05∗ ∗ (0.0 0 02)

−.0 0 05∗ ∗ (0.0 0 02)

−.0 0 05∗ ∗ (0.0 0 02) −0.0 0 03 (0.0 0 08)

−.0 0 05∗ ∗ (0.0 0 02)

−.0 0 05∗ ∗ (0.0 0 02)

Firm beta One-year firm BNH returns One-year industry adjusted firm BNH returns Establishment fixed effects Year fixed effects Firm controls Number of clusters Number of observations R-squared

−0.0055 (0.0081)

No No Yes 1787 2637,053 0.097

Yes No Yes 1787 2637,053 0.561

These coefficients are all in the ballpark of our baseline specification estimate of the effect of a BBSO grant on turnover, suggesting that our choice of a 95% acrossindustry threshold was not a driver of results. 4. Do more successful firms issue BBSOs? In exploring the performance leading up to BBSO issuance and following issuance, we begin by confirming results in Bergman and Jenter (2007), who find that firms tend to grant BBSOs following superior stock price performance and that they do not outperform following the grant. Table 9, Panel A, presents the evidence in our sample. Mean and median raw buy-and-hold returns measured three, 12, or 24 months prior to the BBSO jump are significantly higher than returns measured three, 12, or 24 months post-jump. To make sure that this pattern is not driven by industry or market returns, we compare industry- and market-adjusted buy–and–hold returns and find evidence for the same pattern. In Panel B, we extend this analysis to success, more broadly defined. With varying levels of statistical significance (99% when comparing medians), we find that sales growth, EBIT growth, capital expenditure growth, and employment growth are all higher pre-BBSO. Panel A and B tell a consistent story; that is, firms issue BBSOs when they have been doing well but do not keep that performance going after issuance. In addition to confirming our results, these findings suggest that BBSOs are not granted by managers who are privately informed that the firm is likely to do poorly or well. Firms granting BBSOs perform about as well as the market and their industry peers in the years following grants. 5. Conclusion Given that the cost of hiring and training a new worker has been estimated to equal between 150% and 175% of

Yes Yes Yes 1787 2637,053 0.561

Yes No Yes 1787 2637,053 0.561

Yes No Yes 1787 2637,053 0.561

−0.0056 (0.0063) Yes No Yes 1787 2637,053 0.561

her annual pay (Hansen, 1997), reducing turnover is a firstorder concern for a firm. One potential approach to reduce turnover is to provide the employee with unvested compensation so that she forfeits some wealth if she leaves the firm prior to the vesting date. Unvested stock options can be particularly useful, as they are likely to be worth more when the employee’s outside opportunities are most attractive (Oyer, 2004). We use detailed data to estimate the effect of implementing a broad-based stock option plan on employee turnover. Combining quarterly, establishment-level Census data on turnover with ExecuComp data on stock option grants, we show that, in the years following a large BBSO grant, turnover falls for the granting firm. The reduction in turnover is approximately 0.005% of employment per quarter, which aggregates to 2% per year. If a typical firm experiences 20% turnover annually, and if approximately half of workers receive BBSOs, then the implementation of a BBSO plan reduces turnover by 20%, a substantial reduction. This estimate is robust across a wide variety of specifications, suggesting that it is not spurious. That said, we find more direct evidence consistent with causality by exploring exogenous variation in the value of unvested options. If the existence of unvested options is a deterrent to quitting, the deterrent should be larger after a firm’s stock price has increased, and the options are more valuable. We cannot look at turnover as a function of firm stock returns directly because they could be predictable for the manager establishing the BBSO plan. Instead, we explore industry and market returns, which occur after the BBSO plan was granted but before the options have vested, and find that pre-existing BBSO plans are more effective at reducing turnover when returns are high. We show that while BBSO plans are associated with lower turnover during years when the BBSO shares are unvested, this effect is mostly temporary. In the third full year after the year of the BBSO grant, we observe a large increase in turnover, suggesting BBSO plans often succeed

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

ARTICLE IN PRESS

JID: FINEC

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

15

Table 9 Performance and stock returns pre- and post-broad-based employee stock option (BBSO) grant. The table reports summary statistics for sample of large BBSO issuers around their BBSO issuance years. The sample contains all firms in ExecuComp issuing new BBSO option grants per employee (in Black-Scholes dollar value) that exceed the 95th percentile of the sample distribution in that year. Panel A compares raw, industry-adjusted, and market-adjusted means and medians of stock returns, measured at three, 12-, and 24-month windows, prior to the BBSO issuance versus post-BBSO issuance. Panel B compares means and medians for the number of acquisitions (from Securities Data Company), sales growth, earnings before interest and taxes (EBIT) growth, capital expenditure (CAPX) growth, and employment growth for one (three) year(s) before versus after the BBSO issuance year. Growth variables are winsorized at the 5th and 95th percentiles. Tests for difference in means and medians are shown in the right-most columns. ∗ , ∗ ∗ , and ∗ ∗ ∗ denote statistical significance at 10%, 5%, and 1%, respectively. BNH indicates buy-and-hold return over the specified period. Pre-BBSO jump

Post-BBSO jump

Mean

Median

Mean

Median

Mean difference

Median difference

0.0288 0.0104 0.0183 0.0431 0.0273 0.0325 0.0118 0.0052 0.0114

0.0275 0.0084 0.0151 0.0348 0.0229 0.0270 0.0125 0.0066 0.0094

−0.0021 −0.0062 −0.0104 0.0119 0.0104 0.0069 −0.0074 −0.0095 −0.0125

0.0023 −0.0 0 01 −0.0028 0.0072 0.0136 0.0125 −0.0044 −0.0038 −0.0056

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

1.67 29.93 100.51 20.09 13.91 29.59 94.54 18.02 50.49

2.54 28.22 56.54 −13.95 −149.84 37.08 45.20 20.13 31.19

1.50 17.78 17.03 6.42 −80.03 11.25 −40.58 11.69 8.17

–

–

∗

∗∗∗

∗

∗∗∗

– –

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

∗∗∗

Panel A: stock returns around BBSO jumps Return measure Three-month BNH 12-month BNH 24-month BNH Three-month industry adjusted BNH 12-month industry adjusted BNH 24-month industry adjusted BNH Three–month market adjusted BNH 12-month market adjusted BNH 24-month market adjusted BNH Panel B: Performance around BBSO jumps Performance measure Number of acquisitions Sales growth, one-year Sales growth, three-years EBIT growth, one-year EBIT growth, three-years CAPX growth, one-year CAPX growth, three-years Employment growth, one-year Employment growth, three-years

2.68 49.01 260.83 15.93 22.98 71.11 343.05 28.99 100.58

only in delaying turnover. We find, on average, that 87% of the reduction in turnover observed in the first three years is reversed in the subsequent year. Furthermore, a number of firms issue multiple large BBSO grants. To the extent that a second large BBSO grant timed years after the first grant reduces future turnover, we are underestimating the reversal. We cannot say that BBSO plans are worth the cost, and we do not suggest that retention is the only justification for employee stock options. They are very expensive and clearly can serve an incentive role, screen for optimistic employees, etc. Nonetheless, the effect on turnover is also important and should be considered a first-order benefit of employee stock options. Also, even though the effect on turnover is temporary, BBSOs could still be beneficial if implemented during a period when retaining existing human capital is particularly valuable, for example, a period of high demand and high anticipated turnover. References Abowd, J., Stephens, B., Vilhuber, L., Andersson, F., McKinney, K., Roemer, M., Woodcock, S., 2009. The LEHD infrastructure files and the creation of the quarterly workforce indicators. In: Dunne, T., Jensen, J., Roberts, M. (Eds.), Producer Dynamics: New Evidence from Micro Data. National Bureau of Economic Research, Conference on Research into Income and Wealth. Chicago, IL. University of Chicago Press, pp. 149–230. Aldatmaz, S., Ouimet, P., 2011. The determinants and consequences of broad-based stock option plans: the view from economics and fi-

∗∗∗

nance. In: Carberry, E. (Ed.), Employee Ownership and Shared Capitalism: New Directions and Debates for the 21st Century. Cornell University Press, Ithaca, NY, pp. 211–240. Babenko, I., Lemmon, M., Tserlukevich, Y., 2011. Employee stock options and investment. Journal of Finance 66, 981–1010. Babenko, I., Tserlukevich, Y., 2009. Analyzing the tax benefits from employee stock options. Journal of Finance 64, 1797–1825. Babina, T., 2015. Destructive creation at work: how financial distress spurs entrepreneurship. University of North Carolina, Chapel Hill, NC Unpublished working paper. Balsam, S., Gifford, R., Kim, S., 2007. The effect of stock option grants on voluntary employee turnover. Review of Accounting and Finance 6, 5–14. Bergman, N., Jenter, D., 2007. Employee sentiment and stock option compensation. Journal of Financial Economics 84, 667–712. Blasi, J., Freeman, R., Mackin, C., Kruse, D., 2010. Creating a bigger pie? The effects of employee ownership, profit sharing, and stock options on workplace performance. In: Kruse, D., Freeman, R., Blasi, J. (Eds.), Shared Capitalism at Work: Employee Ownership, Profit and Gain Sharing, and Broad-Based Stock Options. University of Chicago Press, Chicago, IL, pp. 139–165. Brachet, T., David, G., 2011. On the determinants of organizational forgetting. American Economic Journal: Microeconomics 3, 100–123. Carter, M., Lynch, L., 2001. An examination of executive stock option repricing. Journal of Financial Economics 61, 207–225. Carter, M., Lynch, L., 2004. The effect of stock option repricing on employee turnover. Journal of Accounting and Economics 37, 91–112. Core, J., Guay, W., 2001. Stock option plans for nonexecutive employees. Journal of Financial Economics 61, 253–287. Crimmel, B., Schildkraut, J., 2001. Stock option plans surveyed by NCS. Compensation and Working Conditions (spring), 3–21. Hansen, F., 1997. Currents in compensation and benefits. Compensation and Benefits Review 29, 6–18.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007

JID: FINEC 16

ARTICLE IN PRESS

[m3Gdc;November 9, 2017;13:6]

S. Aldatmaz et al. / Journal of Financial Economics 000 (2017) 1–16

Hochberg, Y., Lindsey, L., 2010. Incentives, targeting, and firm performance: an analysis of nonexecutive stock options. Review of Financial Studies 23, 4148–4186. Ittner, C., Lambert, R., Larcker, D., 2003. The structure and performance consequences of equity grants to employees of new economy firms. Journal of Accounting and Economics 34, 89–127. Jarmin, R., Miranda, J., 2002. LBD documentation: broken longitudinal links from missed reorganizations and inactive spells. Technical note, CES-TN-2002. University of Munich, Center for Economic Studies, Munich, Germany. Kedia, S., Rajgopal, S., 2009. Neighborhood matters: the impact of location on broad-based stock option plans. Journal of Financial Economics 92, 109–127. Kim, H., Ouimet, P., 2014. Broad-based employee stock ownership: motives and outcomes. Journal of Finance 69 (3), 1273–1319. Lane, J., Isaac, A., Stevens, D., 1996. Firm heterogeneity and worker turnover. Review of Industrial Organization 11, 275–291.

O’Halloran, P., 2012. Performance pay and employee turnover. Journal of Economic Studies 39, 653–674. Oyer, P., 2004. Why do firms use incentives that have no incentive effects? Journal of Finance 59, 1619–1649. Oyer, P., Schaefer, S., 2005. Why do some firms give stock options to all employees? An empirical examination of alternative theories. Journal of Financial Economics 76, 99–133. Porter, L., Steers, R., 1973. Organizational, work, and personal factors in employee turnover and absenteeism. Psychological Bulletin 80, 151–176. Tate, G., Yang, L., 2015. The bright side of corporate diversification: evidence from internal labor markets. Review of Financial Studies 28 (8), 2203–2249. Vance, M., 2013. Broad-based equity compensation, employee turnover, and unit performance. University of Pennsylvania, Philadelphia, PA Unpublished working paper.

Please cite this article as: S. Aldatmaz et al., The option to quit: The effect of employee stock options on turnover, Journal of Financial Economics (2017), https://doi.org/10.1016/j.jfineco.2017.10.007