New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter...

22
New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering Entrepreneurship Research Georgia Institute of Technology November 10, 2012

Transcript of New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter...

Page 1: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

New Firm Performance and the Replacement of

Founders

Jing Chen Copenhagen Business School

Peter Thompson Emory University

12th Roundtable for Engineering Entrepreneurship Research

Georgia Institute of TechnologyNovember 10, 2012

Page 2: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

2

Motivation Thoery on founder turnover (Holmes and Schmitz 1990)

Distinct comparative advantage embodied by entrepreneurs and CEOs Entrepreneurs specialize in new business creation Professional managers specialize in managing existing ventures

Research gap in the empirical literature Literature on leadership (CEO)change in established firms (Carroll 1984, Haveman

and Khaire 2004) Not exactly an entrepreneurial (startup) context

Literature on founder turnover in startups with transformational growth VC-financed (Hellmann and Puri 2002) Small samples of high-tech startups (Wasserman 2003, Boeker and Wilbank 2005)

No framework to understand founder turnover in a broader population of firms

Page 3: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

3

This Paper Purpose/contribution

a large representative sample of startups little involvement of VC

Research questions Initial / prior startup performance Subsequent performance after founder turnover Labor market outcome of departing founders

The matching model Individual level: U-shaped relationship between founder ability and turnover Firm level: mismatching between founder ability and quality of business idea

Can the U-shaped relationship be carried over for startup preformance and founder turnover What is the correlation between the first and second-period performance with/without

turnover

Page 4: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

4

A Two-Period Model (I) Business output is given by

, : the quality of business idea at time t (technology, fit to the market, public knowledge, etc) : founder’s general innate ability and are complements in both periods is known upon entry is a random draw from , is a random draw from

Profit in period 1

: the opportunity cost of the founder’s time

Page 5: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

5

A Two-Period Model (II) In Period 2,

is observed An option to choose between the founder and an outside CEO to maximize An outside CEO is hired if

: ability of the outside CEO c : is the transition cost of replacing the founder

Maximized profit in period 2

where

Page 6: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

6

Eq

q q2

q0

2 2( )p q%

2 2( , )qp q

c-

wq-

wq¢-

2 2( , )qp q ¢

q q¢>

Payoff with an outside CEO

Payoff with the founder

Figure 1. Second period payoffs; c > 0.

A startup with a very low-quality idea exits the market in the 2nd period

The founder stays if his ability is matched with the quality of business idea.

• Mismatch arises because the founder’s abiltiy is relatively lower than the quality of business idea.

• The founde is replaced with a CEO with higher ability.

• Mismatch arises because the founder’s abiltiy is relatively higher than the quality of business idea.

• The founde is replaced with a CEO with lower ability.

Page 7: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

7

0.0001

0.001

0.01

0.1

1

10

100

0.01 0.1 1 10 100

Idea

Quality

in P

erio

d 2

, q 2

Founder ability, q

Replaced with less able CEO Founder remained

Replaced with more able CEO Firm exited

Figure 2.

Founder Ability vs. Second-Period Idea Quality, by Outcome

Baseline Simulations, 5000 observations

0.01

0.1

1

10

100

0.01 0.1 1 10 100

Seco

nd p

erio

d per

form

ance

First period performance

Replaced with less able CEO Founder remained

Replaced with more able CEO Series4

Page 8: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

8

0.01

0.1

1

10

100

0.01 0.1 1 10 100

Sec

ond p

erio

d p

erfo

rman

ce

First period performance

Replaced with less able CEO Founder remained

Replaced with more able CEO Series4

Baseline Simulations, 5000 observations

Figure 3.

First-Period vs. Second-Period Performance among Surviving Firms, by Outcome

VC story

Page 9: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

9

Data Entrepreneurship in Denmark

Business start-up rate in 2000: 2.99% (13/21 countries in GEM) Venture capital investment in 1999: 0.04% of GDP (Denmark) vs. 0.52% of GDP (US)

Three databases maintained by Statistics Denmark The Entrepreneur Database (1996-2006)

New businesses created in Denmark Unique identifiers (firm, plant, and one individual) the focus on 1999 and 2000 cohorts due to availability of accounting data

The Firm Database (1995-2008) Employment information for all workers at all firms in Denmark Identify possible co-founders

The Integrated Database for Labor Market Research (1980-2008) Employer-employee matched panel Labor market information at firm, plant, and individual level Demographic information of employees

Page 10: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

10

Data Original founder

Formed a startup in 1999 or 2000 as a sole proprietor Currently working at the new business At least one employee working at the business as his/her primary occupation

(minimum economic activities)

Firm outcomes Survival : same firm ID from previous year appears in the current year Ownership change : Different firm ID, same establisment (plant) ID Exit : All establishments of the firm are no longer found in the data

Founder status at the original startup Stay : as employee or employer Exit : due to turnover, business exit or ownership change

Page 11: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

11

2000 2001 2002 2003 2004 20050

200

400

600

800

1000

1200

1400

Surviving: founder stays

Surviving: founder turnover

Firm exit

Ownership change

2001 2002 2003 2004 20050

200

400

600

800

1000

1200

1400

1600

1800

2000

Surviving: founder stays

Surviving: founder turnover

Firm exit

Ownership change

1999 cohort (1,588 startups)

2000 cohort (2,584 startups)

Figure 5. Status Change of Startups by Year

Page 12: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

12

The Samples

Subsample 1: Startup performance and founder turnover 1,784 startups survived during the observation window between founding year

and 2005

Subsample 2: Founder turnover and startup future performance

2000

2001

2002

2003

2004

1999

Turnover by 2002

Turnover after 2002 No Turnover

Exited or acquired by 2002 × ×

Exited or acquired after 2002 ×

Survived to 2004 ×

2,349 startups remain in subsample 2

Page 13: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

13

Dept Var: = 1 if founder replaced in year t

Using sales in year t-1 Using sales in founding year(1) (2) (3) (4)

Log of sales -0.940*** -0.966*** -1.159*** -1.270***

(-5.08) (-5.10) (-3.30) (-3.43) (Log of sales)2 0.072*** 0.076*** 0.084*** 0.095***

(5.71) (5.87) (3.36) (3.61) Founder characteristics Founder age __ -0.017*** __ -0.019***

(-3.05) (-3.40) College educated = 1 __ -0.235* __ -0.090

(-1.38) (-0.54) Male = 1 __ -0.154 __ -0.104

(-1.28) (-0.86) Married = 1 __ -0.331*** __ -0.303***

(-3.25) (-2.98) Firm characteristics Firm age -0.366*** -0.340*** -0.324*** -0.288***

(-10.39) (-9.47) (-9.68) (-8.46) Cohort dummy (= 1 if founding year = 2000)

0.130 0.10 3 0.095 0.069(1.34) (1.04) (0.96) (0.69)

Controls for industry No Yes No Yes

Ave Log Likelihood -0.239 -0.235 -0.241 -0.237No. of observations 7,357 7,348 7,305 7,305

Table 1. The Effect of Startup Performance on the Probability of Founder Turnover

Sales in thousand DKK. z-scores are in parentheses. Significance levels: *** 0.01, ** 0.05, * 0.1.

Page 14: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

14

Figure 3.

First-Period vs. Second-Period Performance among Surviving Firms, by Outcome

Page 15: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

15

Dept Var: log of sales in 2004Surviving Firms

OLSAll Firms

Tobit(1) (2) (3) (4) (5) (6)

Log.sale0 0.437*** 0.453*** __ 0.724*** 0.817*** __(18.8) (18.6) (5.92) (6.12)

Turnover 0.336*** 1.524*** __ -5.00*** -1.032 __(4.15) (2.80) (-14.4) (-0.45)

Log.sale0*turnover __ -0.165** __ __ -0.552* __(-2.21) (-1.73)

log.sale0 percentile: <25th (D1)

__ __ -1.108*** __ __ -1.817***

(-16.6) (-5.04)log.sale0 percentile: 25th -75th (D2)

__ __ -0.700*** __ __ -1.666***

(-12.5) (-5.48)D1 × turnover __ __ 0.441*** __ __ -5.879***

(2.60) (-8.83)D2 × turnover __ __ 0.483*** __ __ -3.605***

(4.19) (-7.12)D3 × turnover __ __ -0.046 __ __ -6.427**

(-0.29) (-9.67)2000 cohort 0.072 0.068 -0.081* -0.441* -0.452* -0.666***

(1.56) (1.49) (-1.76) (-1.80) (-1.85) (-2.76)Adj R-sq 0.214 0.216 0.188 -- -- --Obs. 1,440 1,440 1,440 2,142 2,142 2,142

Table 2. Founder Turnover and Future Performance of Startups

Columns (1) - (3): t statistics in parentheses. Columns (4) - (6): z scores in parentheses. The omitted category in columns (3) and (6) is the top quartile of the log of initial sales. All regressions include controls for industry. Significance levels: *** 0.01, ** 0.05, * 0.1.

Page 16: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

16

Logit ModelDept Var: =1 if firm survived to 2004

(1) (2) (3) (4)Log sale0

0.198*** 0.190***

(3.48) (3.35)log.sale0 percentile: <25th (D1)

-0.458*** -0.450***

(-2.80) (-2.75)log.sale0 percentile: 25th -75th (D2)

-0.540*** -0.537***

(-3.83) (-3.81)D1 × turnover -2.002*** -1.446***

(-8.01) (-5.36)D2 × turnover -1.276*** -0.368

(-6.90) (-1.62)D3 × turnover -2.178*** -1.337***

(-8.61) (-4.59)Turnover -0.538 -1.128

(-0.63) (-1.18)Log.sale0*turnover -0.163 0.028

(-1.39) (0.21)2000 cohort -0.224** -0.270*** -0.262** -0.319***

(-2.14) (-2.61) (-2.41) (-2.97)Ave. Loglikelihood -0.582 -0.580 -0.582 -0.579Obs. 2153 2153 2035 2035

Z-scores in parentheses. Industry controls included. Significance levels: *** 0.01, ** 0.05, * 0.1.

Table 3. Survival and Turnover

Page 17: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

17

Logit RegressionsDept Var: =1 if being an employer at t+1

(1) (2) (3) (4)Log sale0 -0.047 -1.759* __ __

(-0.38) (-1.92)Log sale0^2 __ 0.124* __ __

(1.88)Log salet-1 __ __ -0.014 0.294

(-0.12) (0.35)Log salet-1^2 __ __ __ -0.021

(-0.38)Aget-1 -0.007 -0.009 -0.004 -0.004

(-0.52) (-0.64) (-0.32) (-0.32)Marriedt-1 0.634** 0.607** 0.575* 0.580*

(2.08) (1.96) (1.92) (1.94)Male -0.093 -0.092 -0.064 -0.053

(-0.30) (-0.29) (-0.21) (-0.17)Colleget-1 -1.147*** -1.230*** -1.030*** -0.996***

(-2.91) (-3.08) (-2.73) (-2.59)Ave. Loglikelihood -0.376 -0.373 -0.384 -0.384Obs. 498 498 502 502z scores are in parentheses. Significance levels: *** 0.01, ** 0.05, * 0.1.

Table 4. Founder’s Occupational Choice after Turnover

Page 18: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

18

Conclusions Summary of findings

Founder turnover is more likely in the tails of the performance distribution Founder turnover has less impact on improvement of performance in starups at

the upper end of the initial performance distribution (support the matching story) Future earnings of departing founders are positively correlated with the

performance of their previous startups There is no evidece that they are more likely to choose entrepreneurship over

other occupations after departure. (not consistent with Holmes and Schmitz’ theory)

Future research How do owners decide between selling a business and hiring a CEO (Founder

turnover vs. Business turnover) Assess whether founder-CEO paris differ across startups that is consistent with the

mismatching model

Page 19: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

19

Descriptive Summaries

Subsample 1 1,784 startups survived during the observation window between founding year

and 2005

AllNo

Turnover TurnoverAge (mean years) 40.07 40.23 37.91Male (%) 0.76 0.76 0.77Married (%) 0.64 0.64 0.54Education (%) less than high school 16.84 16.55 21.01 exact high school 6.66 6.34 11.07 vocational training 50.91 51.75 39.21 some college 14.72 14.66 15.57 bachelor 1.13 1.02 2.63 Master 7.66 7.66 7.69 Phd 0.69 0.69 0.75 others 1.38 1.33 2.06Obs. 7,961 7,428 533

Page 20: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

20

Descriptive Summaries

Subsample 1 1,784 startups survived during the observation window between founding year

and 2005

AllNo

Turnover TurnoverFounded in 1999 (%) 42.28 42.69 36.59Initial sales (1,000 DKK) 2,149 2,135 2,347Initial no. of full-time employees 1.87 1.85 2.09Industry (%) Agriculture 0.29 0.28 0.38 Manufacturing 7.25 7.34 6.00 Construction 17.57 17.86 13.51 Wholesale and retail 37.57 37.55 37.9 Transport, post, telecom 4.89 4.90 4.69 Low-tech businesses 6.93 6.93 6.94 Public Services 0.43 0.42 0.56 Personal Services 6.05 6.26 3.19 Kibs 0.44 0.34 1.88 High-tech 18.24 17.77 24.77 Others 0.34 0.35 0.19Obs. 7,961 7,428 533

Page 21: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

21

Descriptive Summaries

1999 Cohort 2000 Cohort Turnover No Turnover Turnover No TurnoverTotal 142 701 232 1,274 Survived 50 545 82 912 Exited before 2002 56 0 70 0 Exited after 2002 36 156 80 362

All No Turnover TurnoverObs. 2,349 1,975 374

Survival in 2004 (%) 67.65 73.77 35.29Founded in 1999 (%) 35.89 35.49 37.97Initial sales (1,000 DKK) 2,008 1,962 2,253 Sales in 2004 (1,000 DKK) 5,168 4,929 7,639 Initial no. of full-time employees 1.68 1.62 2.00No. of full-time employees in 2004 4.08 3.86 6.39

Subsample 2

Page 22: New Firm Performance and the Replacement of Founders Jing Chen Copenhagen Business School Peter Thompson Emory University 12th Roundtable for Engineering.

22

Descriptive Summaries Subsample 3

Turnover Year (t)2000 2001 2002 2003 2004 2005

Obs. 65 191 118 67 52 40Aget-1 37.54 35.49 38.08 40.13 39.23 44.05Male 0.69 0.76 0.75 0.87 0.79 0.80College 0.18 0.26 0.25 0.30 0.35 0.30Married 0.57 0.48 0.59 0.48 0.54 0.70Sale0 (1,000 DKK) 2,554 2,125 2,379 2,193 2,574 3,243 Gross incomet+1 (1,000 DKK) 301.8 388.5 333.6 349.7 356.8 422.4

Ave. gross income (1,000 DKK) 301.0 350.9 323.3 349.8 392.7 384.6

Self-employedt+1 0.18 0.12 0.22 0.13 0.10 0.08No. Of years in SE 0.40 0.35 0.54 0.40 0.29 0.18Sales and income are all converted to 2000 prices.

Descriptive Summary of Subsample 3 (Mean Values Reported)