Current enterprises

Business Training and Female Enterprise
Start-up and Growth in Sri Lanka
Suresh de Mel, University of Peradeniya
David McKenzie, World Bank and
Chris Woodruff, Univ of Warwick
World Bank Conference on “New Ideas in Business Growth:
Financial Literacy, Firm Dynamics and Entrepreneurial
Environment”
March 2011
Motivation: Previous work in SL
•
•
•
•
•
•
–
–
Randomized experiment where we provide SLR 10K or
20K (US$100 or 200) in equipment or cash grants to
micro-enterprises to create exogenous variation in
capital stock (QJE, Nov 2008)
Selected 618 firms in three districts in southern Sri
Lanka (Kalutara, Galle, Matara) with less than SLR 100K
(US$1000) in capital (excluding land and buildings).
Surveyed first in March 2005, then quarterly for two
years, semi-annually for a third year (11 waves)
Profits increased on avg by 5.9% per month.
The surprising result: males generated 7.8% increase in
profits but females generated -0.8% return.
We explore several possible explanations (AEJ Applied,
July 2009)
Intra-household bargaining / capture by spouse
Sector of activity
Motivation: Other recent work
•
•
•
•
•
We look at the impact of business training on a general
population of female business owners – not just on MF
clients.
The current study focuses on 2 groups – current
enterprises and potential enterprises
Content of the business training is a standardized
training package (ILO’s SIYB training program). Useful to
know its impact. Difficult to compare content across
customized training programs offered by MFIs.
We measure outcomes at 3 different points in time posttraining. Able to achieve more power than is possible
with a single follow-up survey. And able to examine the
growth trajectory over time.
We use business training + capital grants as
interventions. Can examine impact of training only vs
training + grants
Follow-on project in Sri Lanka
• Identify two groups of women in 7 districts in and
around Colombo and Kandy. Listing in 142 GNs in 10 DS
divisions.
– Age 25-45 yrs
– Current enterprises: > 20 hrs per wk in self employment,
sector other than seasonal agri/fisheries, monthly profits
=< SLR 5000 ($43).
– Potential enterprises: planned to enter SE in next yr, able
to identify the nature of the proposed business,
unmarried/married with no kids/married with kids > 5 yrs
of age/if < 5 yrs of age had someone to look after the kids.
• Selected sample of 628 current enterprises and 628
potential enterprises equally distributed across 10 DS
divisions.
Interventions
• Provide business training – ILO’s Start and Improve Your Business
(SIYB) program. Implemented in 95 countries. Estimated global
outreach of 1.5 million trainees. Teaching materials customized to
local language and context.
• Potential Ents: 3 day Generate Your Business Idea (GYB) + 5 day Start
Your Business (SYB). Current Ents: 1 day Refresher GYB (RGYB) + 5
day Improve Your Business (IYB)
• Both groups got 1 day technical training – exposure to, and training
in, some relatively high rtn sectors which are socially acceptable for
women. 2-3 options available at each training location.
• Cash grants of SLR 15,000 (~$125) for half, conditional on
completing training
• Attendance payment of Rs 400 per day – transport, lunch, opp cost.
• At each DS location, training offer to 40 current and 40 potentials.
Half of those who completed qualified for the 15K grants.
Sample: Summary Statistics
Current Enterprises
Training
Training +
Control
only
Cash
Variables stratified on
Total Monthly Profits (Rs.)
Have no children or have someone to look after them
Colombo district
Kandy district
Has taken concrete steps to opening business
Has never worked before
Variables not stratified on
Age
Married
Number of children under 18
Years of Education
Risk-seeking score
Digitspan Recall
Raven test score
Total household income from all sources
Household has a fridge
Household has a sewing machine
Household has an oven
Household has a gas cooker
Age of Firm (years)
Ever had a loan from financial institution
Total Monthly Sales (Rs.)
Capital Stock excluding land and buildings (Rs.)
Truncated Capital Stock (Rs.)
Business Practices Score
Number of Firms
3987
0.55
0.20
0.21
3981
0.54
0.20
0.20
4001
0.55
0.20
0.20
35.94
0.89
1.55
10.16
6.81
6.00
2.58
17192
0.45
0.56
0.08
0.25
6.47
0.23
12523
28649
28649
4.59
37.71
0.86
1.47
10.34
6.87
6.04
2.75
18245
0.53
0.60
0.08
0.23
6.88
0.18
12485
27418
27418
4.99
224
200
Potential Enterprises
Training Training +
Control
only
Cash
0.19
0.20
0.51
0.18
0.20
0.20
0.50
0.17
0.21
0.20
0.51
0.19
36.58
0.80
1.40
10.51
6.53
6.01
2.68
17595
0.51
0.60
0.12
0.30
6.35
0.20
12640
35187
34997
4.98
34.38
0.84
1.40
10.51
6.73
6.03
2.76
16422
0.39
0.51
0.09
0.28
34.05
0.91
1.47
10.56
6.82
5.93
2.59
16690
0.41
0.54
0.05
0.24
33.72
0.89
1.59
10.53
6.75
6.06
2.81
16393
0.43
0.55
0.08
0.24
200
228
200
200
Sample
• Typical current enterprise:
– 36 years old, married, with 10 yrs of education,
running the business for 6.5 yrs.
– Mean monthly business income SLR 4000 (US$34).
– This is about 1/4th of HH income
– Low business practices score at baseline (mean is
4.6 out of 29).
– Only 18% have done any business related training
– and of this mainly technical training
Sample
• Typical potential enterprise:
– Only 18% have never worked before, but only 8%
have previously been in SE
– 50% have taken some concrete steps towards
opening a business in the past year.
– 2 yrs younger in age than current grp, but otherwise
similar in terms of education, digitspan recall, raven
tests, attitudes towards risk, and no of children.
– Monthly HH income about Rs 1100 less than
current.
– Less likely to own fridge or sewing machine (assets
that have business potential)
Timeline
April/May
2009
June
2009
Sept
2009
Jan
2009
Screening and
Baseline survey
Grants
delivered
Notification
/ Training
Jan
2010
Sept
2010
Second
follow-up
survey
First follow-up
survey
Third follow-up
survey
Treatment Takeup
• Current: 279 (69.8%) of the 400 offered
treatment attended training and 268 (67%)
completed training.
• Potentials: 282 (70.5%) of the 400 offered
treatment attended training and 261 (65.3%)
completed.
• Common reasons for not taking up training:
– Family member was sick
– No one to look after the business in their absence
– No one to look after their children
Who is likely to take-up training?
• CURRENT
– Married, more educated women, running younger
firms, more likely to attend training.
– Having no children or having someone to look after
children not significantly associated with takeup
– Manufacturing firms more likely to attend training
– Opp cost of time seems to matter: women running
higher profit earning enterprises are less likely to
attend, women working more than 40 hrs per wk are
less likely to attend.
– Firms in Colombo are less likely to attend
Who is likely to take-up training?
• POTENTIALS
– Take-up increases with age of woman and raven
score.
– Colombo potentials are less likely to attend.
– Having no children or having someone to look after
children, yrs of education, previous work experience,
having taken steps to open a business, spouse’s
income not significantly associated with takeup
Takeup: Current Ents
Table 2A: Determinants of Training Take-up Among Current Enterprises
Marginal effects from Probit estimation of Attending Training among those offered
(1)
Has no children or has someone to look after children
0.0164
(0.0469)
Log of monthly profits
-0.0700*
(0.0405)
Age
0.00621
(0.00403)
Married
0.121*
(0.0654)
Years of Education
0.0197**
(0.00985)
Firm is younger than 5 years old
0.0838*
(0.0495)
Baseline Business Practices Score
0.00152
(0.00643)
Risk-seeking Attitude
-0.0215
(0.0135)
Digit-span Recall
-0.00911
(0.0201)
Firm is in Manufacturing
0.158**
(0.0646)
Firm is in Retail Trade
0.0826
(0.0652)
Works more than 40 hours a week at baseline
-0.0878*
(0.0485)
Says would pay 500 Rs or more for a training course
-0.0149
(0.0494)
Colombo District
D.S. (locality) fixed effects
No
(2)
-0.0452
(0.0472)
0.0259
(0.0391)
0.00627
(0.00408)
0.164**
(0.0676)
0.0101
(0.0102)
0.131**
(0.0521)
0.00380
(0.00690)
-0.0111
(0.0134)
0.00918
(0.0199)
0.146**
(0.0681)
0.0488
(0.0718)
-0.0879*
(0.0497)
-0.0265
(0.0528)
-0.454***
(0.0672)
0.0730
(0.0628)
No
Number of firms
Robust standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.1
400
400
Kandy District
(3)
-0.0117
(0.0494)
0.0112
(0.0428)
0.00452
(0.00419)
0.144**
(0.0687)
0.00742
(0.0107)
0.128**
(0.0525)
0.00761
(0.00759)
-0.0104
(0.0140)
0.00915
(0.0214)
0.149**
(0.0691)
0.0502
(0.0732)
-0.0705
(0.0495)
-0.0508
(0.0537)
400
Yes
Takeup: Potential Ents
Current Ent: Impact on Business Practices
• Measured at baseline (R1) and short-term (R2: 4
months after training) and medium term (R4: 16
months after training)
• Business Practices usage has increased in both the
short term and medium term for both training only grp
and training + cash grp
• Magnitude of increase large relative to baseline.
• Training also significantly improved the components –
marketing, stock control, financial planning, and record
keeping.
Current Ent: Impact on Business Practices
Table 3: Impact on Business Practices of Current Enterprises
Total Practices Score
(1)
(2)
(3)
Round 2 Round 4 All rounds
Intent-to-Treat Effects
Assigned to Cash if finish Training 2.530*** 1.936*** 2.071***
(0.555) (0.567) (0.373)
Assigned to Training only
1.719*** 1.708*** 1.656***
(0.555) (0.560) (0.382)
Treatment on the Treated
Received Training & Cash
3.588*** 2.790*** 2.986***
(0.591) (0.607) (0.462)
Received Training Only
2.192*** 2.261*** 2.169***
(0.540) (0.546) (0.431)
Marketing Stock Control Record keeping Financial Planning
(4)
(5)
(6)
(7)
All rounds All rounds All rounds
All rounds
0.395***
(0.130)
0.495***
(0.134)
0.236***
(0.0690)
0.167**
(0.0704)
1.000***
(0.174)
0.633***
(0.161)
0.566***
(0.150)
0.537***
(0.161)
0.564***
(0.165)
0.647***
(0.153)
0.340***
(0.0886)
0.219***
(0.0816)
1.485***
(0.228)
0.833***
(0.186)
0.836***
(0.194)
0.705***
(0.187)
Observations
544
513
1,057
1,057
1,057
1,164
1,164
Firms
544
513
573
573
573
598
598
Baseline Mean:
4.96
5.02
4.96
1.66
0.53
2.10
0.64
Robust standard errors in parentheses clustered at the firm level when all rounds used, *** p<0.01, ** p<0.05, * p<0.1
All specifications also include survey round dummies, baseline outcome value, and controls for randomization strata.
Current Ent: Impact on Firm Performance
• Impact of the treatments on monthly profits (also on
sales and capital stock). Examined in levels, truncated at
99th percentile and in logs.
• Training alone does not increase profits.
• But significant impact of training + cash on profits
• Magnitudes of impact on profits is also high. Eg. TOT
shows that truncated profits increase by 2236 relative
to baseline mean of 4014.
• Increase in profits occur in R2 and R3 (4 months + 8
months post training) but falls off by round 4 (16
months post training).
Current Ent: Impact on Firm Performance
Table 4: Impact on Firm Performance for Current Enterprises
All rounds pooled
(1)
(2)
(3)
Levels
Panel A: Monthly Profits
ITT Effects
Assigned to Cash if finish Training
Assigned to Training only
TOT Effects
Received Training & Cash
Received Training Only
Baseline Mean:
Truncated
Levels
Logs
Round 2
(4)
Round 3
(5)
Round 4
(6)
Truncated Truncated Truncated
Levels
Levels
Levels
1,195
(884.1)
-574.7
(908.7)
1,520** 0.213***
(645.9) (0.0755)
-118.0
0.0450
(661.1) (0.0797)
1,955**
(929.7)
-79.99
(952.1)
441.4
(1,191)
-531.2
(1,223)
1,765
(1,203)
-746.1
(1,105)
2,236** 0.312*** 2,587** 2,885***
(882.3)
(0.103)
(1,046)
(1,058)
-146.4
0.0602
22.99
-90.80
(804.9) (0.0978) (897.6)
(964.3)
656.7
(1,307)
-699.0
(1,215)
4014
4014
8.14
1,801*
(945.6)
24.92
(904.7)
4004
4023
4016
Observations
1,592
1,592
1,527
538
542
512
Firms
577
577
571
538
542
512
Notes:
Robust standard errors in parentheses clustered at the firm level when all rounds used, *** p<0.01, **
p<0.05, * p<0.1
All specifications also include survey round dummies, baseline outcome value, and controls for
randomization strata.
Truncated levels truncate at the 99th percentile.
Potential Ent: Entry into SE
• Training + grant leads to a 13 percentage point increase
in prob that a woman enters into SE.
• Training alone has a smaller effect which is not
statistically significant.
• However in R2 (4 months post training), training only
leads to a 12 percentage point increase and training +
grant leads to a 22 percentage point increase.
• By R4 (16 months post training), the difference
between treatment and control groups have
disappeared.
• On avg, training + grant leads to a 9.5 percentage point
increase in likelihood of being ever SE
Potential Ent: Entry into SE
Table 6: Impact on Entry into Self Employment
Probability of being Self Employed
VARIABLES
Assigned to Training only
Assigned to Training and Cash
Grant
Observations
Ever SE
(1)
(2)
(3)
(4)
All waves
R2
R3
R4
0.0677
0.1187**
0.0553
0.0311
0.0290
(0.043)
(0.053)
(0.050)
(0.051)
(0.047)
0.0248
0.0952**
0.1306*** 0.2211*** 0.1521***
(5)
(0.042)
(0.050)
(0.049)
(0.051)
(0.046)
1,732
588
587
557
588
Notes:
R2 through R4 denote survey
rounds
Robust standard errors in parentheses clustered at the firm level, *** p<0.01, ** p<0.05, *
p<0.1
All specifications also include survey round dummies and controls for randomization strata.
Potential Ent: Impact on Business Practices
• Impact on business practices is much less compared to
current ent.
• Trtmnt raised overall score by only just over 1 point –
but this is statistically significant only for the training +
cash grp.
• Impact of the trtmnt is positive in each of the subcomponents but significance indicated only for
marketing among the training + cash grp and record
keeping for the training grp
Potential Ent: Impact on Business Practices
Table 7: Impact on Business Practices of Current
Enterprises
Overall
Score
Marketing
(3)
(4)
Ruond 4
Ruond 4
Intent-to-Treat Effects
Assigned to Training Plus Cash
Grant
1.334** 0.648***
(0.663)
(0.214)
Assigned to Training only
1.106
0.253
(0.734)
(0.237)
Stock
Control
(5)
Ruond 4
Record
keeping
(6)
Ruond 4
Financial
Planning
(7)
Ruond 4
0.117
(0.129)
0.132
(0.131)
0.430
(0.324)
0.563*
(0.322)
0.139
(0.296)
0.158
(0.297)
Observations
335
335
335
335
335
R-squared
0.225
0.214
0.103
0.078
0.285
Robust standard errors in parentheses clustered at the firm level when all rounds used, ***
p<0.01, ** p<0.05, * p<0.1
All specifications also include survey round dummies, baseline outcome value, and controls
for randomization strata.
Potential Ent: Impact on Firm Performance
• We find positive but insignificant effects of both
treatments
• In R2 and R3 we find negative effects on profits for
those who rcvd training + grants (but not statistically
significant)
• By R4, large positive effect on profits for those who rcvd
training only. Positive but not significant effect for those
who rcvd training+grant.
• Recall that in R2 and R3 we had significantly higher
rates of SE in training + grants grp. Could it be that with
the training + grants there has been more entry by
women with lower potential profits and less entry by
women with higher potential profits?
Potential Ent: Impact on Firm Performance
Table 8: Impact on Firm Performance for Potential
Enterprises
All waves
(1)
Panel A: Monthly Profits
ITT Effects
Assigned to Training Plus Cash Grant
328
(844)
Assigned to Training only
1,210
(900)
Observations
Notes:
950
R2
(2)
R3
(3)
R4
(4)
-186
(866)
486
(899)
-560
(1,248)
239
(1,379)
1,532
(1,135)
2,632**
(1,230)
287
329
334
Robust standard errors in parentheses clustered at the firm level when all rounds
used, *** p<0.01, ** p<0.05, * p<0.1
All specifications also include survey round dummies, baseline outcome value, and
controls for randomization strata.
Truncated levels truncate at the 99th percentile (columns 5-8)
Conclusions
• Examined impact of training and training + cash grant
among current and potential female enterprises.
• CURRENTS:
– Significant improvements in business practices. Effects are
only slightly smaller even 16 months after training.
– Training only does not affect profits. But training +grants has
significant positive impact on profits.
• POTENTIALS:
– Both training only and training + grant has speeded up entry
into SE, but not the ultimate rate of entry.
– Some evidence that profits are higher among the treatment
grps by R4.