-
Patterns of Business Creation, Survival, and Growth:
Evidence from a Developing Country
Leora Klappery Christine Richmondz
November 2009
Abstract
We study rm dynamics using a novel database of all formally
registered rms in
Cote dIvoire from 1977-1997, which account for about 60% of GDP.
During this pe-
riod, Cote dIvoire experienced (separate) episodes of trade
reform, scal reform, and
monetary reform, and we examine the impact of these shocks on
private sector dy-
namics. First, we examine entry and exit patterns over this
period and the role of new
and exiting rms versus incumbents in job creation and
destruction. In comparison
with the U.S., new rms play a relatively important role in job
creation and growth,
particularly during periods of political uncertainty. Yet, in
absolute terms, incumbent
rms have a much larger impact on net employment. Next, we
examine survival rates
We thank Francois Bourguignon for providing us the data and
Daniel Dias, Sebastian Edwards, PhilEnglish, Vesela Grozeva, Ed
Leamer, Phil Kim, Javier Miranda, Nico Voigtlander, Mark Wright,
RomainWacziarg, and seminar participants at the World Bank and 2009
COST Workshop on Firm-level DataAnalysis in Transition and
Developing Economies, George Mason University-Kauman Foundation
BusinessCreation Symposium, and 2009 Northeast Universities
Development Consortium for helpful comments.Financial assistance
from UCLA CIBER is gratefully acknowledged by Christine Richmond.
The opinionsexpressed do not necessarily represent the views of the
World Bank, its Executive Directors, or the countriesthey
represent.
yWorld Bank, Development Research Group. The World Bank. 1818 H
St. N.W., Washington, DC20433. Tel: +1.202.473.8738; Email:
[email protected].
zAnderson Graduate School of Management, UCLA. Anderson Graduate
School of Management, UCLA.110 Westwood Plaza, Entrepreneurs Hall,
Suite C525, Los Angeles, CA 90095-1481. Tel.: +1
310.825.8207;Email: [email protected].
1
-
and nd that survival increases monotonically with rm size.
Furthermore, higher
GDP growth increases survival rates, and periods of economic
growth have the largest
impact on the reduction in the exit rates of small rms, relative
to large rms. Finally,
we nd that all reforms decreased rm survival, although the
impact varied by sector:
for instance, political reforms increased the exit rate of
foreign-owned rms (relative
to domestic rms), while both trade and monetary reforms
decreased survival rates of
rms in the trade sector.
JEL Classication: G18, G38, L51, M13
Keywords: Business Incorporation; Regulatory Barriers; Economic
Growth
2
-
1 Introduction
Firm entry and exit Schumpeterian creative destructionis
critical for the continued
dynamism of the modern economy. Though evidence has linked
entrepreneurship and eco-
nomic growth in developed countries, we have scarce evidence
that a relationship exists
in developing countries. For instance, to what extent does
entrepreneurship contribute to
job creation in countries with regulatory, legal, and tax
barriers to entry and exit? In this
paper, we examine job creation and destruction, at both new and
incumbent rms, in Cote
dIvoire, ranked 161 out of 181 countries in the Doing Business
"Ease of Doing Business"
ranking (where 1 indicates the most favorable business
environment) (World Bank, 2009).
We use a new and unique database that includes the complete
census of registered
rms in Cote dIvoire from 1976 through 1997. This includes data
collected at the time
of registration (ownership, sector, etc.) as well as annual
reporting of total employment.
Unlike comparable studies that use survey data or manufacturing
censuses, we can compute
over time exact formal sector entry and exit rates; growth and
contraction at incumbent
rms; and the impact of macroeconomic and policy reforms on
sectoral distribution and
growth.
First, we examine business dynamics in Cote dIvoire, using the
complete census of
registered rms over the period 1976 to 1997. We decompose trends
in job creation and
growth into entry of new rms and expansion of incumbent rms, and
job destruction
into exit and contraction. We also illustrate trends in Cote
dIvoire, relative to the U.S.
(Haltiwanger et al., 2008), and compare dierences and
similarities in formal business
creation and destruction across very dierent investment
climates.
Next, we examine the impact of three policy reforms on
entrepreneurship: The rst
reform removed import quotas and other trade restrictions. The
second reform reduced
public sector wages (in an attempt to reduce the budget decit).
The third reform devalued
the currency by half. We study the eect of these scal, monetary,
and trade reforms on
the creation of new rms, the average size of rms that
incorporate, and the growth of
3
-
incumbent rms, controlling for rm size, ownership, and age.
Finally, we use a survival
analysis to study the impact of the business cycle and
government reforms on rm exit.
New research on the survival of rms in low-income countries has
important implications
for development strategies. In order for the private sector to
act as an engine of growth
and advance the development process, it is necessary for rms to
survive and grow. It is also
important for policymakers to understand and consider the impact
of government policies
on private sector survival and growth.
Our paper proceeds as follows. Section 2 reviews previous
literature related to rm
dynamics. Section 3 reviews the relevant economic and political
history in Cote dIvoire.
Section 4 compares business dynamics in Cote dIvoire and the
U.S. Section 5 discusses our
survival test methodology and results. Section 6 concludes.
2 Related Literature
Empirical studies of rm dynamics overwhelming focus on rm
behavior in the US, Canada,
and other industrialized countries.1 In general, these studies
nd that gross entry is sub-
stantial in most industries and is signicantly higher than net
entry due to the high death
rate of infant rms.2 Successful entrants grow rapidly, so that
an entrant cohorts initial
market share falls slowly. For instance, an early study of
patterns of rm entry, growth, and
exit in US manufacturing industries over the period 1963-1982
shows that there is substan-
tial and persistent dierences in entry and exit rates across
industries and that industries
with higher than average entry rates tend to also have higher
than average exit rates (Dunne
et al., 1988). The authors also nd that most entrants and
exiters are substantially smaller
than existing or continuing producers.
Recent cross-country studies have examined the impact of country
characteristics, in
particular, the legal and regulatory environment, on rm entry
and growth. For instance, a
1See Caves(1998) for a thorough survey of the literature on job
turnover and rm mobility.2There is also a substantial literature on
entry into an industry (possibly by a rm from another industry)
as distinguished from rm creation or entrepreneurship; see
Gilbert (1989) for a comprehensive survey.
4
-
comprehensive picture of entry rates across Europe (using a
rm-level database of registered
rms) use a dierence-in-dierence approach to show that onerous
entry regulations hamper
the creation of new rms, especially in industries that naturally
should have high entry.
Furthermore, the consequences of regulatory barriers against
entrepreneurship are seen, not
in young rms, but in older rms, who grow more slowly and to a
smaller size (Klapper
et al, 2006). Another paper uses this database and a
cross-country approach and nds
complementary evidence that entry regulations have a negative
impact on rm entry (Desai
et al., 2003). A study using rm-level data from OECD countries
to analyze rm entry and
exit nds that higher product market and labor regulations are
negatively correlated with
the entry of SMEs in OECD countries (Scarpetta et al., 2002).
Finally, a study of entry
rates in over 100 countries nds signicant relationships between
entrepreneurial activity
and indicators of economic and nancial development and growth,
the quality of the legal
and regulatory environment, and governance (Klapper et al.,
2009).
Other studies have studied rm size distribution, i.e. the
importance of small and
medium sized rms (SMEs) in the economy. For instance, a study of
the creative destruction
process in 24 developed and developing countries over the 1990s
nds that countries are
dominated by micro and small rms (with less than 20 employees),
which accounts for more
than 80% of the rm population, yet only accounts for up to 30%
of employment (Bartelsman
et al., 2004). A study of 76 countries nds a strong association
between the importance of
SMEs and GDP per capita growth, although not necessarily a
causal relationship (Beck et
al., 2005).
The shortcoming of most studies in low- and middle-income
countries is their dependence
on survey data. For instance, surveys for ve Eastern and
Southern African countries
(Botswana, Kenya, Malawi, Swaziland, and Zimbabwe) as well as
the Dominican Republic
are used to examine the magnitude and determinants of rm births,
deaths, and expansions
of micro and small enterprises (MSEs) (Mead and Liedholm, 1998).
The authors nd that
the annual rate of new MSE start-ups averages over 20%, and that
the vast majority of
5
-
new rms were one-person establishments. For MSE closures,
location plays a signicant
role with urban MSEs having an almost 25% greater chance of
surviving the year, ceteris
peribus, compared to rural MSEs.
A number of studies have also used rm surveys to study the
determinants of rm
growth. Using a survey of manufacturing rms in Cote dIvoire,
Sleuwaegen and Goedhuys
(2002) nd that younger rms grow faster than older rms, yet
larger entrants experience
greater growth opportunities that improve over time. A broad
review of manufacturing
rm performance across developing countries nds that most
economies are small, have
limited domestically produced intermediate inputs and capital
equipment, face low rates
of secondary education and a lack of technicians, have limited
infrastructure, experience
greater macroeconomic and relative price volatility than
developed countries, have thin
credit markets with debt heavily skewed towards short-term
instruments, and face weak
legal systems and higher instances of corruption (Tybout, 2000).
Overall, these challenges
hamper the growth of small, entering rms.
Another important question in this literature is the
determinants of rm survival. For
example, evidence from Portuguese manufacturing rms nds that
drivers of a rms survival
include start-up size, with larger rms having a higher
probability of survival, and fast
growing industries having a positive eect on rm duration.
Furthermore, the probability
of rm failure decreases with the age of the rm, consistent with
the view that rm entry
is a process of selection and learning (Mata and Portugal,
1994).
Finally, the literature examines the impact of macroeconomic and
other shocks on private
sector behavior. For instance, large macroeconomic disturbances
are found to aect the
rm turnover process. A study of the cyclical behavior of small
versus large manufacturing
rms in the U.S. and the response to monetary policy, nds that
small rms account for
a signicantly disproportionate share of the decline in
manufacturing that follows a shift
to tight money (Gertler and Gilchrist, 1994). A related paper
focusing specically on new
rms in the U.S. nds that startup rates are substantially shaped
by both macroeconomic
6
-
uctuations as well as industry-specic characteristics;
macroeconomic expansion serves
as a catalyst for startup activity, but new-rm startups are
promoted by a low cost of
capital as well as high unemployment rates (Audretsch and Acs,
1994). A study of rms
in West Germany rms nds that the exit of new plants is not
responsive to aggregate
cyclical uctuations, rather, exit is large among new entrants at
any stage of the business
cycle; however, among incumbent plants, the survival of older
rms is more sensitive to the
aggregate cycle (Boeri and Bellmann, 1995).
Previous literature has also studied the impact of trade reform.
An investigation of
employment patterns in Chilean manufacturing rms following a
major trade liberalization
between 1975-1979 which made the economy more outward-oriented,
removed quantitative
trade restrictions and slashed tari rates, privatized rms,
relaxed government controlled
prices, and reformed nancial markets nds that in the years
following the trade liber-
alization, rm size matters for job creation and destruction; job
churning is very high in
both expanding and contracting industries; and job creation and
destruction varies with
whether the rm is operating in the tradable or non-tradable
sectors (Levinsohn, 1999).
The importance of rms size in determining survival is supported
by another study that
examines rms in Ghana, Kenya, and Tanzania during the 1990s,
when market reforms
were implemented to eliminate protection of the domestic rms
(Harding, et al., 2004). To
summarize, the literature in developing countries generally nds
that the main determinant
of rm exit is size, with small rms having much higher exit rates
than large rms, and nd
no evidence that productivity impacts rm survival.
7
-
3 Macroeconomic Background and Government Re-
forms
In 2006, Cote dIvoire was ranked 164 out of 177 in the World
Bank Human Development
Index (World Bank, 2009).3 This followed a period of political
instability and civil war
(2002 to 2003). Yet, Cote dIvoire remains one of the largest
economies in the region and
accounts for close to 40% of the economic activity in the West
Africa Economic Monetary
Union (WAEMU). Furthermore, the countrys relatively high GDP per
capita (US$ 1,027
in 2007) suggests opportunities for private sector development
and growth.
Cote dIvoire grew rapidly after achieving independence in 1960,
and by 1974, nominal
GNI per capita rose from 190 to 450 CFA (over 300% in real
terms). This expansion
was primarily due to growth in the agricultural sector as the
country benetted from high
international prices of coee and cocoa, for which the country is
a major exporter. In
addition, the countrys investment code and stable government and
policies encouraged
foreign capital investment in agribusiness activities. By 1978,
the World Bank declared,
The so-called Ivorian Miraclehas not been a matter of luck. . .
and the countrys political
stability and favorable economic environment led international
organizations to believe that
the country could be the rst African country to achieve
"developed" status (Tuinder,
1978).4 However, with the collapse of coee and cocoa prices and
the oil shock in 1979, the
Ivorian economy experienced a severe annualized contraction of
12% in 1980 and entered a
prolonged recessionary period that persisted through most of the
1980s. For most of this
period, GDP growth was less than 3% and private credit (as a
percentage of GDP) sharply
declined from about 30% to less than 20%.
After defaulting on commercial bank loans in 1983, the country
undertook a series of
foreign debt reschedulings with both o cial and private
creditors. For the next decade,
3This section draws on Trebesch (2008), Reuda-Sabater and Stone
(1992), and Library of CongressCountry Studies (1988).
4From 1960 to 1993, the country was led by a one-party rule:
Houphouet-Boigny ruled from 1960 untilhis death in 1993.
8
-
Cote dIvoire instituted a series of trade, scal, and monetary
reforms, which we will show
impacted private sector entry, exit, and growth.
First, from 1985 to 1987, the government removed a series of
import quotas, in order to
increase competition and reduce prices. The government removed
quantitative restrictions
and unied eective rates of protection across intermediate and
nal goods, which reduced
the competitive advantage of domestic manufacturers vis--vis
foreign exporters (Harrison,
1994).
Following the implementation of this trade reform, domestic
conditions deteriorated
drastically in 1987 with international coee prices dropping more
than 30% (YoY), which
led to a scal decit greater than 8% of GNP and external debt
ratios higher than 145% of
GNP, and caused the government to declare an o cial debt
moratorium in May 1987. In
December 1987, the government reached a multi-year rescheduling
of US$ 931 million with
Paris Club creditors and an agreement with London Club creditors
worth US$ 1,575 million
in March 1988. In support of the agreements with creditors, the
IMF extended US$ 240
million in loans in March 1988, and the government undertook a
series of revenue-raising
measures.
While the countrys default episode was not fully resolved until
1998 (additional reschedul-
ings occurred in 1989 and 1997), the workout arrangements with
creditors and IMF support
ensured access (albeit limited) to external capital markets. Yet
by the late 1980s the gov-
ernment faced signicant di culties implementing IMF and World
Bank mandated reforms,
particularly attempts to contain the public sector wage bill
which amounted to than 60% of
non-interest expenditures. These moves led to protests by trade
unions, strikes, demonstra-
tions, and led to a rise in student militancy. At the end of
1992, the government announced
its plans to abandon salary cuts (and introduce a multi-party
political system), which de-
fused public protests.
Due to further deteriorations in the terms of trade, the Central
Bank of West African
States (BCEAO) implemented an IMF-backed nominal devaluation of
the CFA by 50% in
9
-
mid-January 1994 in an eort to restore the regions international
competitiveness. After
being pegged to the French franc at the same rate for 46 years,
the currency value was
reduced from CFA/FRF 50 to CFA/FRF 100 (equivalent to CFA/EUR
656). Cote dIvoire
responded well to the devaluation as the government
simultaneously undertook a series of
reforms to improve competitiveness and promote investment,
including a revised investment
code, removal of price controls and export duties on key
exports, and privatization program,
in order to maximize the benets associated with the
devaluation.
To summarize, our empirical analysis will focus on the impact of
private sector activity
following three government interventions:
Trade reform (1985-1987): Removal of import quotas;
Fiscal reform (1989-1990): Reduction in public sector wages;
Monetary reform (1994): 50% currency devaluation.
4 Data and Summary Statistics
The dataset is comprised of 5,941 unique rms in Cote dIvoire
from 1976 through 1997,
which covers the complete census of formal, private sector
corporations in the country.5
The data was hand-collected from the Registrar of Companies for
the Modern Enterprise
Sector, and does not include a sample bias of exited rms. These
rms are expected to pay
the full range of taxes, and may benet from bank nance and state
technical assistance.
This dataset appears to be consistent with the overall private
sector in Cote dIvoire, which
includes about 800 large, formally registered enterprises (with
at least 50 employees), several
thousand formally and informally operating medium-sized
enterprises (10-50 employees),
and several hundred thousand informal microentrepreneurs, which
mostly generate self-
employment (Reuda-Sabater and Stone, 1992).
5The database excludes the primary sectors and state-owned
rms.
10
-
The structure of the data is an unbalanced panel of about 37,000
rm/year observations.
That is, there is detailed information tracking rms over time
and the dataset includes rms
that enter over the course of the sample period, as well as rms
that exit during the sample.
The variables of interest are information on sector classication
at the two-digit sector
level and rm ownership, collected at the time of registration,
and annual data on rm
employment, collected from required annual lings.6 Our dataset
also includes balance
sheet and income statement nancial information, which is
available for the sub-period
1976-1987.
We summarize this dataset in two dimensions: First, we provide
the rst illustration of
rm dynamics in a low-income country, across sectors and time.
This includes entry and
exit rates, as well as job creation and destruction rates and
new versus incumbent rms.
Second, we present a comparison of formal sector dynamics in
Cote dIvoire, relative to
private sector dynamics in the U.S. Following recent work by
Haltiwanger et al. (2008), we
use the U.S. Bureau of the Census, Business Demography
Statistics (BDS), which provides
detailed data on rm births and deaths and job creation and
destruction, by rm size,
age, and sector.7 An important caveat is that we are not able to
compare the two overall
economies: Data for the U.S. covers virtually the complete
private sector, while data for
Cote dIvoire includes a very small number of rms that chose to
formally register in a
challenging business environment. Yet, this comparison might
shed light on barriers to
private sector job creation and growth, which encumbers many
developing nations.
4.1 Business and employment formation
On average, there are about 2,000 registered rms per year, with
on average 29 employees
(and median employment of 7), which employ about 100,000
individuals. Although this is a
6All registered companies are required to submit annual lings to
the Registrar of Companies withnancial and employment information.
Our panel includes only two missing rm/year observations.
7An important caveat is that US data is on the establishment
(plant) level, while the Ivorian data is onthe rm (consolidated)
level. However, nearly all rms in Cote dIvoire operate in only one
location. Tomake the databases comparable, we exclude from the Cote
dIvoire sample rms in the nancial, insurance,and real estate (FIRE)
sectors.
11
-
small fraction of overall employment, including individuals
self-employed or employed in the
informal sector, the formal sector represents over 60% of GDP
(OECD African Economic
Outlook, 2004). Figure 1, Panel A, shows that while the total
number of registered rms in
a given year has remained at, on average, over the period, total
employment in the formal
private sector declines over time. These are important trends
for policymakers trying to
promote high-growth, employment generating, private sector rms.
While average rm size
in Cote dIvoire has decreased over time, Panel B shows that
average rm size in the U.S.
grew by over 7% over the same period.
Figure 2 presents summary statistics, by size buckets over the
entire sample period. For
the purpose of our analysis, we dene smallas less than 10
employees; middleas 10-49
employees; largeas 50-149 employees; and very largeas more than
150 employees.
Economy-wide, 40% of rms have less than 10 employees and an
additional 40% of rms
have 10-49 employees. Interestingly, we also nd in the U.S. that
80% of rms have less
than 50 employees, though a larger number, 62%, have less than
10 employees. The smaller
percentage of micro rms (with less than 10 employees) in Cote
dIvoire might reect
the fact that because of burdensome regulations, high marginal
tax rates, the absence of
monitoring and compliance (of both registration and tax
regulations), and other weaknesses
in the business environment, many rms might nd it optimal to
evade regulations and
operate in the informal sector. For instance, the cost of
starting a business (as a percentage
of income per capita) is 133% in Cote dIvoire versus 0.7% in the
U.S, and the minimum
capital required to start a business (as a percentage of income
per capita) is 205% in Cote
dIvoire, versus 0% in the U.S (Doing Business, 2009). Previous
literature has highlighted the
potential advantages to formal sector participation, including
police and judicial protection
(and less vulnerability to corruption and the demand for bribe),
access to formal credit
institutions, the ability to use formal labor contracts, and
greater access to foreign markets
12
-
(Schneider and Enste, 2000).8 The tendency of rms in Cote
dIvoire to choose to stay
small and informal might be unable to realize their full growth
potential.
Figure 3 shows the distribution of rms and employees, by sector.
Although almost 50%
of rms are identied as service providers, 46% of employment is
generated by the manufac-
turing sector (as compared to 37% of employment in the service
sector). Similarly, we nd
about 50% of rms in the U.S. are service rms and generate about
40% of employment.
Yet, we nd a notable dierence in the contribution of the
manufacturing sector in Cote
dIvoire, which comprises about 17% of rms and 46% of employment
(as compared to 10%
of rms and 31% of employment in the U.S). On the one hand, it is
surprising that manu-
facturing rms which require relatively higher levels of
investment, human resources, and
capital remain a large percentage of rms in Cote dIvoire; for
example, nancial devel-
opment (measured as private credit as a percentage of GDP) is
only 33% over our sample
period. However, on average, approximately 70% of Ivorian rms
are owned by foreigners
from industrialized countries with the majority of these owners
being French (not shown).9
The lack of access to domestic nance might also explain the
notable dierences in rm
size distribution, by sector. Though the manufacturing sector
has the fewest number of
rms, average rm size is signicantly larger than it is for rms
operating in the trade or
service sectors. However, the median rm size is small for all
sectors, with only a median
of 8 employees in trade sector rms. We nd notable dierences in
comparison to the
U.S. distribution. Most strikingly, across all sectors we nd
less than half the number of
rms in the category of 10-49 employees, which is novel evidence
that the missing middle
phenomena applies to sectors other than manufacturing.
For instance, within the manufacturing sector, we nd 44% of rms
in Cote dIvoire are
small (less than 10 employees) and only 27% are large (more than
50 employees), relative to
14% and 42% of rms in the U.S., respectively; possibly, the
growth of manufacturing rms
8The benets of formal sector registration might vary by
industry. This is discussed in Section 5.3.9We also nd that the
sectoral distribution of foreign- and domestically-owned rms is
nearly identical
(not shown).
13
-
in Cote dIvoire is limited by capital constraints. Service
sector rms display an even more
disparate distribution: 72% of rms operate with less than 10
employees in Cote dIvoire,
relative to 45% of rms in the U.S. In contrast, close to 50% of
rms operating in the
tradable sector in both Cote dIvoire and the U.S. have less than
10 employees, yet 28% of
rms in Cote dIvoire are large, while only 9% of tradable rms
have at least 150 employees.
It might be the case that in developing countries with costly
and timely barriers to starting
a business, only rms in wholesale and retail trade might have
the greatest incentive to
formally register, for instance, in order to receive a Value
Added Tax (VAT) number, which
might be required for domestic and international sales.
4.2 Firm and employment dynamics
In an eort to describe the entry and exit patterns of Ivorian
and U.S. rms, we begin by
constructing measures of entry and exit as well as their size
relative to other rms in the
industry. We count a rm as a new entrant if it appears for a
second year, that is, we
exclude rms that survive less than a year.
Similar to Dunne et al. (1988), we dene the following variables,
where bucket i is
equal to either 2-digit industry classications (manufacturing,
services, or trade) or size
(micro, small, medium or large):
NEit = number of rms that enter bucket i between years t 1 and
t
NTit = total number of rms in bucket i in year t, including rms
that
enter bucket i between years t 1 and t
NXi(t1) = number of rms that exit bucket i between years t 1 and
t
Next, we construct entry (ERit), exit (XRit), and turnover
(TRit) rates as:
ERit = NEit=NTi(t1)
XRi(t1) = NXi(t1)=NTi(t1)
TRit = (NEit +NXit)=NTi(t1)
14
-
The denominator for both entry and exit rates is the total
number of rms in bucket i in
year t-1. This is because we want to capture the pool of
potential exiting rms. These
measures are comparable to gross measures of entry and exit and
allow us to compare our
ndings to those of US rm studies.
We also construct survival rates as the probability of exiting
at t using the Kaplan-Meier
estimator such that
S(t) =tYj=1
[(n t)=(n j + 1)](t):
S(t) is the estimated survival function, n is the total number
of rm lifespans, ()t is a
dummy equal to one for completed spells and 0 for censored
spells. In our sample we nd
that the average lifespan of rms is 6.7 years, with a median of
5.0 years.
We follow Haltiwanger, et al.s (2008) analysis of business
dynamics in the U.S. Firm-
level employment growth is :10
git = (Eit Eit1)=Xit; where Xit = :5 (Eit + Eit1)
and aggregate employment growth is given by the weighted average
of rm-level growth:
gt = i(Xit=Xt)git; where Xt = iXit:
Employment growth can be decomposed into creation and
destruction:
JCit = i(Xit=Xt)maxfgit; 0g
JDit = i(Xit=Xt)maxfgit; 0g:
Next we decompose job creation into entry and expansion and job
destruction into exit
and contraction:
JC_Entryt = i(Xit=Xt)Ifgit = 2g;10This follows the methodology
of Davis et al. (1994).
15
-
where I is an indicator variable equal to one if the expression
in the brackets hold and zero
otherwise, and git = 2 denotes an entrant.
JC_Exitt = i(Xit=Xt)Ifgit = 2g;
where I is an indicator variable equal to one if the expression
in the brackets hold and zero
otherwise, and git = 2 denotes an exit.
We commence with a comparison of business dynamics across the
U.S. and Cote dIvoire
over the sample (Figure 5). The rst observation is the high
volatility of entry and turnover
rates in Cote dIvoire, which reect larger business cycle swings
and macroeconomic instabil-
ity (that is, a strong correlation between real GDP and entry
rates), political uncertainty,
and a more challenging business environment (Panels A and B).
This is consistent with
Klapper et al. (2006) and Bartelsman, et al. (2009), which nd a
(contemporaneous)
relationship between entry rates and the business
environment.
The second striking feature of these graphs is that on average,
the entry rates in Cote
dIvoire and the US are so similar (12.4% and 13.2%,
respectively), across such dierent
economic, legal, and institutional environments. This is
consistent with similar results
shown for a larger sample of over 100 countries in the World
Bank Entrepreneurship Survey
database, which nds very low variation in entry rates of
formally registered corporations
across a wide range of countries, e.g. the data nds, on average,
entry rates of 8% in
middle- and low-income countries, versus 11% in industrialized
countries (Klapper et al.,
2008). This is also consistent with Bartelsman et al., (2009),
which nds very similar entry
rates across OECD countries. It is important to bear in mind
that we are comparing only
formal sector dynamics; i.e. it is interesting to consider that
rms that overcome the hurdles
of registering in a challenging business environment look
similar to other registered rms
around the world.
We also compare one-year survival rates of rms in Cote dIvoire
and the U.S. over our
sample period (Panel C). Similar to our ndings on entry rates,
we nd high volatility of sur-
16
-
vival rates in Cote dIvoire, which reect both larger business
cycle swings (macroeconomic
instability), weaker bankruptcy laws, and greater political
instability. However, again, an
interesting observation is that on average, the 2-year survival
rates in Cote dIvoire and
the US are very similar, 83% and 84%, respectively. Taken
together, this might, intrigu-
ingly, suggest naturalentry and exit rates of registered rms
over time, regardless of the
investment climate.
Next, in Figure 6, we look at employment dynamics at new and
exited rms (Panel A)
and jobs created and lost at incumbent rms, normalized to 100%
(Panel B). Over time,
net employment is always positive, that is, more jobs are
created at new rms then lost
at exiting rms. However, the absolute number of new jobs added
to the private sector is
remarkably small on average, fewer than 3,000 formal, private
sector jobs are added on
an annual basis. In comparison, the net number of new jobs
generated at incumbent rms
is negative for all years from 1979 through 1995. The largest
net job losses occurred during
the 1980 macroeconomic downturn over 10,000 jobs were shed in
1979 although net jobs
at incumbent rms remained negative even during periods of GDP
growth.
In comparison to the U.S., an important implication is that new
rm creation appears
to be relatively more important for job creation in Cote
dIvoire, relative to incumbent
growth. This might suggest that expanding opportunities for
entrepreneurial potential is
important for macroeconomic growth. An important caveat,
however, is that jobs created
by new rms are highly volatile and not necessarily as good as
those provided by large rms
(Haltiwanger et al., 2009). This suggests that economic and
employment growth in Cote
dIvoire might be hampered by the inability of incumbent rms to
grow, relative to growth
and job creation rates in developed countries, such as the U.S.
This data can be interpreted
as quantitative evidence of the impact of barriers in the
business environment such as
costly and timely procedures to obtain licenses, register
property and movable collateral,
resolve disputes, and close a business (Doing Business, 2009) on
private sector growth.
17
-
Finally, we compare average job ows by sector (Figure 7). The
most prominent ob-
servation is that across all US all sectors, job growth is
driven from incumbent expansion.
While we continue to see similar patterns of job creation
arising from rm entry and job
destruction from rm exit, the dierence in dynamics is arising
from job ows in expanding
and contracting rms.
5 Survival Model and Results
The previous section highlights barriers in employment creation
at existing incumbent rms.
In order to better understand the barriers to growth, we begin
by addressing the primary
question of what determines rm survival. Understanding the
factors that appear to most
inuence rm survival is necessary to develop and implement
policies that will improve and
help rms better adapt and prepare for challenges and promote
opportunities for growth.
Our research aims to identify rm characteristics, such as sector
and size, that predict
greater vulnerability to business cycles and policy
interventions. First, we examine the
relationship between economic growth (measured by annual GDP
growth rates) and rm
survival. Next, we explore the impact of specic monetary, scal,
and trade reforms on rm
survival.
5.1 Firm dynamics and reforms
We begin by summarizing rm and job dynamics, relative to annual
GDP growth and
economic reforms. Figure 8, Panels A and B, show the absolute
number and one-year
growth rates of entering and exiting rms, by year, relative to
one-year GDP growth. For
most of the period, we nd that the formal private sector both
gains and looses about 200
rms per year, which is a growth and contraction rate equal to
about 10%. This trend is
surprisingly stable over time and macroeconomic conditions.
Panel C shows the impact of
policy reforms: Although GDP growth was relatively at during the
reform periods, the
18
-
trade and scal reforms led to a sharp decrease in the number of
rms. In particular, the
scal reform led to a relatively sharper decline in foreign-owned
rms as political instability
increased. In contrast, the monetary devaluation (and growth in
GDP) was associated with
an increase in new rm creation.
Next, in Figure 9, Panels A and B, we look at changes in new
employment at new and
exited rms (Panel A) and jobs created and lost at incumbent rms
(Panel B), relative
to annual GDP growth. Over time, net employment is always
positive, i.e. more jobs
are created at new rms then lost at exiting rms. However, the
absolute number of new
jobs added to the private sector is remarkably small on average,
fewer than 3,000 formal,
private sector jobs are added on an annual basis. In comparison,
the net number of new
jobs generated at incumbent rms is negative for all years from
1979 through 1995. The
largest net job losses occurred during the 1980 macroeconomic
downturn over 10,000 jobs
were shed in 1979 although net jobs at incumbent rms remained
negative even during
periods of GDP growth. Panel C shows that net employment at
new/exited rms has
little variation over our sample period, while employment at
incumbent rms appears more
sensitive to government reforms, in particular, the period
following the introduction of scal
reforms.
Finally, Figure 10 shows net employment creation, by sector,
relative to GDP and eco-
nomic reforms. Interestingly, we nd dierent patterns across
sectors, suggesting that eco-
nomic and political shocks aect job creation and destruction
dierently across sectors. For
instance, employment in trade is relatively at, perhaps because
these jobs are driven by
external factors, while manufacturing jobs are most sensitive to
macroeconomic uctuations
and government interventions, which might reect their greater
size and capital dependency.
5.2 Survival test methodology
Our comprehensive panel data allows us to study the rm-level
determinants of rm survival.
We estimate a discrete time duration model with time-varying
covariates. The parametric
19
-
specication we consider is a proportional hazard model with a
piecewise constant baseline
hazard. Specically, we use a baseline hazard with six
parameters, j(j = 1; 2; 3; 4; 5; 6):
one for each of the rst four three-year periods and one for the
remaining years. We also
accommodate unobserved individual heterogeneity with the
inclusion of a multiplicative
error term in the hazard function, for which a gamma
distribution with mean 1 and variance
2 is assumed.
The hazard function for rm i in period t is specied as being
proportional to:
exp(f(Zi;Wt; Xit));
where Wt is a vector of macroeconomic variables and reform
dummies, Zi is a vector that
includes dummies for rm sector, initial size, location, and
owner, and Xit is a vector of
individual time varying characteristics of the rm, including
current number of employees.
For rm i we dene it as the elapsed duration since the rms birth
in period t and let
Ti be the duration of the spell. Under the assumption of a
proportional hazard model with
a piecewise constant baseline hazard and unobserved
heterogeneity, the hazard function can
be expressed by:
(itjZi;Wt; Xit; vi) = re exp(f(Zi;Wt; Xit))vi;
where vi is the unobserved component for rm i and re = j for it
5 and it = 6 for
it 6: Standard results imply that the survival function for Ti
can be written as:
S(TijZi;Wt; Xit; vi) = exp(vi
T rXt=1
re exp(f(Zi;Wt; Xit))
):
Thus the contribution to the log-likelihood function from item i
can be written as:
ln(Li) = ln[S(Ti 1jZi;Wt; Xit) &iS(TijZi;Wt; Xit)];
20
-
where &i is a dummy variable which equals 1 for completed
spells and 0 for (right) censored
events.
5.3 Results
Table 1 shows summary statistics of our dependent and
explanatory variables. Our depen-
dent variable is a dummy (0/1), equal to one in the rst year
that the rm exits from the
sample, and equal to zero otherwise. On average, rms in our
sample survive 6.6 years.
We include a series of rm characteristics as our dependent
variables: First, we include a
dummy for rms located in the Abidjan (capital) region, which
includes 90% of the sample.
Second, a dummy for foreign ownership is included, with 72% of
rms being foreign owned.
Third, we include sector dummies to indicate rms in the
manufacturing (18%) or trade
sectors (34%); we omit the service sector (49%). Finally we
include size dummies to indicate
a rms entrysize at the time of registration: small (46%), medium
(35%), or large (9%);
we omit the very largecategory (10%). Next, we include dummies
indicating the three
reform periods: trade (1985-1987), scal (1989-1990), and
monetary (1994). Finally, we in-
clude real annual GDP growth. In all specications we use leading
GDP growth to address
the concern of endogeneity arising from the largest rms
contributing the most towards
aggregate GDP (since rm exits occurs in the rst year after we no
longer observe the rm).
To summarize our baseline hazard function in our regressions, we
consider domestically-
owned rms located outside of the Abidjan commercial region,
entering with at least 150
employees, and operating in the service sector. A correlation
matrix of the explanatory
variables is shown in Table 2.
Table 3 shows the determinants of rm exit, where a positive coe
cient indicates a
lower likelihood of survival. We do not nd any eect of operating
in the capital and do
not nd that survival rates vary by sector. However, we nd that
foreign rms exit more
frequently than domestically-owned rms, which might reect a
higher voluntary exit rate
among foreigner owners who are more sensitive to economic and
political volatility. We also
21
-
nd stark dierences in the exit rate of rms according to their
initial size: the larger a
rm is at entry, the less likely it is to exit. Next, we nd a
signicant eect of declining
GDP growth and lower survival rates. Finally, we nd that all
reform periods are associated
with higher exit rates; the scal reform had the largest impact
on survival, followed by the
monetary reform, and lastly the trade reform. In terms of
economic signicance, the scal
reform raised the probability of exit by 66%, the monetary
reform raised the probability of
exit by 40%, and the trade reform raised the probability of exit
by 19%, all relative to the
baseline hazard.
Table 4 presents the impact of GDP growth on rm exit. We use
interaction terms
to test if macroeconomic growth had a dierential impact on the
exit rate of rms across
dierent sectors, ownership, and size. We nd that the business
cycle has a marginal eect
on only the manufacturing sector, relative to service sector rms
(at the 10% level) and
nd no marginal eect by rm ownership. However, we nd a strong
eect of rm size:
smaller rms entering with 10-49 or 50-149 employees are most
exposed to economic
downturns, relative to large rms entering with 150+ employees.
This likely reects the
added vulnerability of small rms to downturns in the business
cycle.
Finally, Table 5 examines the impact of government reforms on
exit rates. From our
initial set of results presented in Table 3 we know that all
reforms increased the exit rate for
rms, with the scal reform having the greatest impact. First, we
nd that the manufactur-
ing sector experienced relatively higher exit rates during the
scal crisis; we hypothesize that
this reects the larger, more capital intensive nature of rms in
the industry. These results
are in line with previous literature showing that rms with
larger market capitalization are
more politically connected, and uncertainty surrounding these
political relationships can
negatively impact rmsperformances (Faccio 2006 and 2007 and
Fisman, 2001). Next, we
nd that trade reform signicantly increased the probability of
exit for rms operating in
the tradable sector, relative to the service sector, which was
expected given the removal of
22
-
import quotas and increased competition facing the domestic
economy. We nd no sector
or size specica impact of the monetary reform.
As a robustness check of our results, we estimate regression
models using contempora-
neous log employment levels instead of entry size dummies. While
not reported here, the
results are qualitatively similar.
Finally, an interesting nding is the shape of the baseline
hazard (Figure 11). Our esti-
mates reveal an increasing baseline hazard function, which can
be interpreted as suggesting
that that the probability of exit increases as the rm ages over
time. This result is somewhat
surprising and counterintuitive, given that rms observed in
developed countries exhibit an
increasing probability of survival as they age (see, for
example, Audretsch and Mahmood,
1995 and Holmes et al., 2008). Furthermore, we do not observe an
inection point where age
begins to increase the probability of survival, suggesting that
rm vulnerability to exiting
does not decrease over time. It might be the case that rms
become more visibleas they
age, which makes it more di cult to avoid costly operating and
labor regulations.
6 Conclusion
We study rm dynamics using a database of all formally registered
rms in Cote dIvoire
from 1977-1997, which account for about 60% of GDP. During this
period, Cote dIvoire
experienced (separate) episodes of trade reform, scal reform,
and monetary reform, and
we examine the impact of these shocks on private sector
dynamics. First, we examine entry
and exit patterns over this period and the role of new and
exiting rms versus incumbents
in job creation and destruction.
Next, we examine survival rates and nd that survival increases
monotonically with size.
Furthermore, higher GDP growth increases survival rates, and
periods of economic growth
have the largest impact on the reduction in the exit rates of
small rms, relative to large
rms. Finally, we nd that all reforms decreased rm survival.
23
-
A comparison of business patterns in the U.S. and Cote dIvoire
highlights the need to
encourage and inspire greater experimentation and innovation
among African entrepreneurs.
Start-up rms in the U.S. generally enter small and face low
survival rates, but those that do
survive grow and create new jobs, displacing those rms that do
not innovate. Improvements
in the institutional environment such as easing rm entry and
closure might promote
greater risk taking and high growth entrepreneurship.
7 References
1. Audretsch, D. B. and T. Mahmood. 1995. New rm survival: new
results using a
hazard function. Review of Economics and Statistics 77(1):
97-103.
2. Audretsch, D. B. and Z. J. Acs. 1994. New-rm startups,
technology, and macroeco-
nomic uctuations. Small Business Economics 6: 439-449.
3. Bartelsman, E. J., J. Haltiwanger, and S. Scarpetta, 2009.
Cross country dierences
in productivity: the role of allocative e ciency. University of
Maryland, Department
of Economics, mimeo.
4. Bartelsman, E. J., J. Haltiwanger, and S. Scarpetta. 2004.
Microeconomic evidence
of creative destruction in industrial and developing countries.
World Bank Policy
Research Working Paper No. 3464.
5. Beck, T., A. Demirguc-Kunt, and R. Levine. 2005. SMEs,
growth, and poverty:
cross-country evidence. Journal of Economic Growth 10(3):
199-229.
6. Boeri, T. and L. Bellmann. 1995. Post-entry behavior and the
cycle: evidence from
Germany. International Journal of Industrial Organization 13(4):
483-500.
7. Caves, R. E. 1998. Industrial organization and new ndings on
the turnover and
mobility of rms. Journal of Economic Literature 36(4):
1947-1982.
24
-
8. Davis, S., J. Haltiwanger, and S. Schuh. 1994. Small Business
and Job Creation:
Dissecting the Myth and Reassessing the Facts. Labor Markets,
Employment Policy,
and Job Creation, Westview Press.
9. Desai, M., Gompers, P., Lerner, J., 2003. Institutions,
capital constraints and entre-
preneurial rm dynamics: Evidence from Europe. NBER Working Paper
No. 10165.
10. Dunne, T., M. J. Roberts, and L. Samuelson. 1988. Patterns
of rm entry and exit
in U.S. manufacturing industries. RAND Journal of Economics
19(4): 495-515.
11. Faccio, M. 2007. The characteristics of politically
connected rms. Krannert School
of Management, Purdue University, Working Paper No.
2007-006.
12. Faccio, M. 2006. Politically connected rms. American
Economic Review 96(1): 369-
386.
13. Fisman, R. 2001. Estimating the value of political
connections. American Economic
Review 91(4): 1095-1102.
14. Gertler, M. and S. Gilchrist. 1994. Monetary policy,
business cycles, and the behavior
of small manufacturing rms. Quarterly Journal of Economics
109(2): 309-340.
15. Haltiwanger, J., R. Jarmin, and J. Miranda. 2009. Who
creates jobs? Small vs. large
vs. young. University of Maryland, Department of Economics,
mimeo.
16. Haltiwanger, J., R. Jarmin, and J. Miranda. 2008. Business
formation and dynamics
by business age: results from the Business Demography
Statistics. US Census Bureau,
mimeo.
17. Harding, A., M. Sderbom, and F. Teal. 2004. Survival and
success among African
manufacturing rms. Centre for Study of African Economies Working
Paper No.
2004-05.
25
-
18. Harrison, A. E. 1994. Productivity, imperfect competition,
and trade reform: theory
and evidence. Journal of International Economics 36: 53-73.
19. Holmes, P., I. Stone, and P. Braidford. 2008. An analysis of
new rm survival using
a hazard function. Applied Economics, forthcoming.
20. Klapper, L., R. Amit, and M. Guillen. 2008. Entrepreneurship
and rm formation
across countries. NBER Volume on International Dierences in
Entrepreneurship, J.
Lerner and A. Schoar, eds., University of Chicago Press.
21. Klapper, L., L. Laeven, and R. Rajan. 2006. Entry regulation
as a barrier to entre-
preneurship. Journal of Financial Economics 82(3): 591-629.
22. Levinsohn, J. 1999. Employment responses to international
liberalization in Chile.
Journal of International Economics 47(2): 321-444.
23. Library of Congress Country Studies. 1988. Cote dIvoire.
Washington, DC: Library
of Congress.
24. Mata, J. and P. Portugal. 1994. Life duration of new rms.
Journal of Industrial
Economics 42(3): 227-245.
25. Mead, D. C. and C. Liedholm. 1998. The dynamics of micro and
small enterprises in
developing countries. World Development 26(1): 61-74.
26. OECD. 2004. African Economic Outlook 2003/2004, Cote
dIvoire. Paris: OECD,
African Development Bank.
27. Reuda-Sabater, E. and A. Stone. 1992. Cte dIvoire: private
sector dynamics and
constraints. World Bank Policy Research Working Paper No.
1047.
28. Scarpetta, S., P. Hemmings, T. Tressel, and J. Woo. 2002.
The role of policy and
institutions for productivity and rm dynamics: evidence from
micro and industry
data. OECD Economics Department Working Paper No. 329.
26
-
29. Schneider, F. and D. Enste, 2000. Shadow economies: Size,
causes and consequences.
Journal of Economic Literature, 38(1): 77-114.
30. Sleuwaegen, L. and M. Goedhuys. 2002. Growth of rms in
developing countries,
evidence from Cte DIvoire. Journal of Development Economics 68:
117-135.
31. Trebesch, C. 2008. Delays in debt restructurings. Should we
really blame the credi-
tors? Free University of Berlin, mimeo.
32. Tuinder, B. A. den. 1978. Ivory Coast, the challenge of
success: report of a mission
sent to the Ivory Coast by the World Bank. Baltimore, MD: The
Johns Hopkins
University Press.
33. Tybout, J. 2000. Manufacturing rms in developing countries:
how well do they do,
and why? Journal of Economic Literature 38(1): 11-44.
34. World Bank. 2009. Doing Business, 2009. Available from
http://www.doingbusiness.org
27
-
Figure 1: Comparison of Total Employment and umber of Firms,
1977-1997
Panel A: Cote dIvoire
1,000
1,200
1,400
1,600
1,800
2,000
2,200
-
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
Number of Firms
Number of Employees
Total Employees Total Firms
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997.
Panel B: U.S.
4,000
4,500
5,000
5,500
6,000
6,500
-
20,000
40,000
60,000
80,000
100,000
120,000
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
97
Number of Firms
(thousands)
Number of Employees
(thousands)
Total Employees Total Firms
Source: US Census Bureau's Longitudinal Business Database,
1977-1997.
-
Figure 2: Comparison of Firm Size Buckets, Average 1977-1997
Panel A: Cote dIvoire
40%
40%
12%
8%
-
Figure 3: Comparison of umber of Firms and Employment, by
Sector,
Average 1977-1997
Panel A: Cote dIvoire
49%
34%
17%
Distribution of Firms
service
trade
manufacturing
37%
17%
46%
Distribution of Employment
service
trade
manufacturing
Panel B: U.S.
50%
40%
10%
Distribution of Firms
service
trade
manufacturing
40%
29%
31%
Distribution of Employment
service
trade
manufacturing
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997 and the U.S. Census Bureaus Longitudinal
Business Database.
-
Figure 4: Comparison of Firm Size Buckets, by Sector, Average
1977-1997
Panel A: Manufacturing Sector
44%
29%
6%
21%
-
Figure 5: Comparison of Firm Dynamics, 1978-1997
Panel A: Average Entry Rates
0%
5%
10%
15%
20%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
Cote d'Ivoire
United States
Panel B: Average Turnover Rates
0%
10%
20%
30%
40%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
Cote d'Ivoire
United States
Panel C: Average Survival Rates, 1978-1995
70%
75%
80%
85%
90%
95%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95
Cote d'Ivoire
United States
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997 and the U.S. Census Bureaus
Longitudinal Business Database.
-
Figure 6: Comparison of Employment Dynamics, Average 1978
1997
Panel A: Average Job Creation and Destruction Rates
Panel B: Average !umber of Employees, !ormalized to 100%
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997 and the U.S. Census Bureaus
Longitudinal Business Database.
0%
4%
8%
12%
16%
20%
Cote d'Ivoire US Cote d'Ivoire US
Job Destruction Job Creation
Exit Exit Entry Entry
Contraction Contraction Expansion Expansion
0%
20%
40%
60%
80%
100%
Cote d'Ivoire US Cote d'Ivoire US
Job Destruction Job Creation
Exit
Exit Entry Entry
Contraction Contraction Expansion Expansion
-
Figure 7: Comparison of Employment Dynamics, by Sector, Average
1978 1997
Panel A: Manufacturing
Panel B: Services
Panel C: Trade
0%
4%
8%
12%
16%
20%
Cote d'Ivoire US Cote d'Ivoire US
Job Destruction Job Creation
Exit Exit
EntryEntry
Contraction
Contraction
Expansion
Expansion
0%
4%
8%
12%
16%
20%
Cote d'Ivoire US Cote d'Ivoire US
Job Destruction Job Creation
Exit Exit Entry Entry
Contraction Contraction Expansion Expansion
0%
4%
8%
12%
16%
20%
Cote d'Ivoire US Cote d'Ivoire US
Job Destruction Job Creation
Exit ExitEntry
Entry
Contraction
Contraction
Expansion
Expansion
-
Figure 8: Cote DIvoire Firm Dynamics, GDP Growth, and Economic
Reforms
Panel A: !umber of Firms and GDP Growth
-15
-10
-5
0
5
10
15
0
50
100
150
200
250
300
350
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
GDP growth (annual %)
Number of Firms
Entry Exit GDP growth (annual %)
Panel B: Growth in !umber of Firms and GDP Growth
-15
-10
-5
0
5
10
15
0%
5%
10%
15%
20%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
GDP growth (annual %)
Annual Growth (%)
Entry Rates Exit Rates GDP growth (annual %)
Panel C: !et !ew Firms and Economic Reforms
-150
-100
-50
0
50
100
150
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
# of firm
s
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997.
Trade Fiscal Monetary
-
Figure 9:Cote DIvoire Job Dynamics, GDP Growth and Economic
Reforms
Panel A: !umber of Jobs Created at !ew Firms and Lost at Exiting
Firms
and GDP Growth
-15
-10
-5
0
5
10
15
-
5,000
10,000
15,000
20,000
25,000
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
GDP growth (annual %)
Job Creation/Destruction
Jobs created at new firms Jobs lost at exiting firms GDP growth
(annual %)
Panel B: !umber of Jobs Created and Lost at Incumbent Firms and
GDP Growth
-15
-10
-5
0
5
10
15
-
5,000
10,000
15,000
20,000
25,000
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
GDP growth (annual %)
Job Creation/Destruction
Jobs created at incumbent firms Jobs lost at incumbent firms GDP
growth (annual %)
Panel C: !et Employment at Incumbent and !ew/Exit Firms and
Economic Reforms
0
2000
4000
6000
8000
10000
-18000
-12000
-6000
0
6000
12000
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
Net Employment
Incumbants Entry/Exit
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997.
Trade Fiscal Monetary
-
Figure 10: et Employment Growth, GDP Growth, and Economic
Reforms,
by Sector, 1978-1997
Panel A: Manufacturing
-0.04
-0.035
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
-15%
-10%
-5%
0%
5%
10%
15%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
Manufacturing
GDP
Panel B: Services
-0.04
-0.035
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
-15%
-10%
-5%
0%
5%
10%
15%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
Services GDP
Panel C: Trade
-0.04
-0.035
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
-15%
-10%
-5%
0%
5%
10%
15%
78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
Trade GDP
Source: Cote dIvoire Centrale de Bilans des Entreprises,
1977-1997.
Trade Fiscal Monetary
-
Figure 11: Baseline Hazard Function
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
baseline hazard rate
# years since entry
-
Table 1: Variables and Summary Statistics
Variable Definition Mean
Survival Age of firm when it dies (in years)
6.66
Abidjan Dummy (0/1), =1 if the firm is located in the Abidjan
(capital) region, =0
otherwise
0.90
Foreign_owned Dummy (0/1), =1 if the firm is foreign owned, =0
if the firm is
domestically owned
0.72
Service Dummy (0/1), =1 if the firm operates in the service
sector, =0 otherwise
0.49
Trade Dummy (0/1), =1 if the firm operates in the trade sector,
=0 otherwise
0.34
Manufacturing Dummy (0/1), =1 if the firm operates in the trade
sector, =0 otherwise
0.18
Entry_small Dummy (0/1), =1 if the firm enters operation with
less than 10
employees, =0 otherwise
0.46
Entry_ medium Dummy (0/1), =1 if the firm enters operation with
10-49 employees, =0
otherwise
0.35
Entry_ large Dummy (0/1), =1 if the firm enters operation with
50-149 employees, =0
otherwise
0.09
Entry_very large Dummy (0/1), =1 if the firm enters operation
with 150+ employees, =0
otherwise
0.10
Reform_trade Dummy 0/1, =1 in 1985-1987, the years during which
trade reform was
undertaken
0.16
Reform_fiscal Dummy 0/1, =1 in 1989 and 1990, the years during
which fiscal/political
reform was undertaken
0.10
Reform_monetary Dummy 0/1, =1 in 1994, the year during which
monetary reform was
undertaken
0.05
GDP_growth Real one-year GDP growth
1.50
-
Table 2: Correlation Matrix
Note: Asterisk indicates significance at the 5% level.
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Abidjan (1) 1.0000
Foreign_owned (2) -0.0391* 1.0000
Service (3) -0.1083* 0.0490* 1.0000
Manufacturing (4) 0.0687* 0.0334* -0.4445* 1.0000
Trade (5) 0.0590* -0.0783* -0.6970* -0.3325* 1.0000
Entry_small (6) -0.0028 -0.0458* 0.0846* -0.2547* 0.1148*
1.0000
Entry_medium (7) -0.0224* 0.0273* -0.0554* 0.0703* 0.002
-0.6776* 1.0000
Entry_large (8) -0.0149* 0.0265* -0.0767* 0.1816* -0.0646*
-0.2848* -0.2300* 1.0000
Entry_very large (9) 0.0396* -0.0037 -0.0719* 0.2041* -0.0877*
-0.2379* -0.1921* -0.0807* 1.0000
Reform_trade (10) 0.0068 0.0179* 0.0095 -0.0064 -0.0049 0.0081
-0.0001 -0.0087 -0.0073 1.0000
Reform_monetary (11) -0.0050 -0.0414* -0.0151* 0.0062 0.0109*
0.0291* -0.0126* -0.0081 -0.0118* -0.0968* 1.0000
Reform_fiscal (12) -0.0082 -0.0092 -0.0014 -0.0035 0.0043
0.0135* -0.0010 -0.0068 -0.0076 -0.1429* -0.0742* 1.0000
GDP_growth (13) -0.0108* -0.0405* -0.0238* 0.0119* 0.0155*
0.0148* -0.0078 0.0022 -0.0035 -0.0170* 0.2846* -0.1530*
-
Table 3: Survival Model: Determinants of Firm Exit
This table reports estimates using a discrete time-duration
model with time-varying covariates The
dependent variables in all models is a dummy (0/1) equal to one
in the first year that the firm exits from the
sample, and equal to zero otherwise. Explanatory variables are
defined in Table 1. Standard errors are
shown below, ***, **, * indicate significance at the 1%, 5%, and
10% level, respectively.
(1) (2) (3)
Coeff S.E. Coeff S.E. Coeff S.E.
Abidjan -0.014 0.082 -0.020 0.080 -0.012 0.077
Foreign_owned 0.266 0.058 *** 0.255 0.057 *** 0.261 0.055
***
Manufacture 0.006 0.077 0.006 0.075 -0.008 0.073
Trade -0.039 0.057 -0.041 0.055 -0.049 0.053
Entry_small 1.683 0.132 *** 1.650 0.130 *** 1.541 0.127 ***
Entry_medium 0.956 0.124 *** 0.938 0.121 *** 0.872 0.118 ***
Entry_large 0.502 0.157 *** 0.483 0.153 *** 0.441 0.148 ***
GDP_growth -0.016 0.004 *** -0.015 0.004 ***
Reform_trade 0.176 0.051 ***
Reform_fiscal 0.509 0.054 ***
Reform_monetary 0.337 0.080 ***
Constant -2.160 0.212 *** -2.176 0.207 *** -2.359 0.202 ***
1 -1.053 0.207 *** -0.986 0.204 *** -0.826 0.200 ***
2 -0.876 0.171 *** -0.859 0.169 *** -0.728 0.165 ***
3 -0.628 0.144 *** -0.623 0.142 *** -0.548 0.140 ***
4 -0.434 0.127 *** -0.432 0.125 *** -0.476 0.126 ***
5 -0.138 0.112 -0.157 0.111 -0.277 0.113 **
Gamma variance 1.168 1.077 0.960
# Observations 36529 36529 36529
-
Table 4: Survival Model: The Impact of GDP Growth
This table reports estimates using a discrete time-duration
model with time-varying covariates The
dependent variables in all models is a dummy (0/1) equal to one
in the first year that the firm exits from the
sample, and equal to zero otherwise. Explanatory variables are
defined in Table 1. Standard errors are
shown below, ***, **, * indicate significance at the 1%, 5%, and
10% level, respectively.
(1) (2) (3)
Coeff S.E. Coeff S.E. Coeff S.E.
Abidjan -0.020 0.080 -0.019 0.080 -0.019 0.080
Foreign_owned 0.255 0.057 *** 0.256 0.058 *** 0.258 0.057
***
Manufacture 0.006 0.075 0.026 0.076 0.004 0.076
Trade -0.041 0.055 -0.032 0.056 -0.041 0.056
Entry_small 1.650 0.130 *** 1.650 0.130 *** 1.679 0.13 ***
Entry_medium 0.938 0.121 *** 0.937 0.121 *** 0.984 0.12 ***
Entry_large 0.483 0.153 *** 0.481 0.153 *** 0.539 0.16 ***
GDP_growth -0.016 0.004 *** -0.010 0.009 0.013 0.016
GDP_growth*foreign 0.000 0.009
GDP_growth*manufacture -0.019 0.011 *
GDP_growth*trade -0.009 0.008
GDP_growth*small -0.022 0.017
GDP_growth* medium
-0.042 0.017 **
GDP_growth*large -0.053 0.022 **
Constant -2.176 0.207 *** -2.177 0.207 *** -2.192 0.209 ***
1 -0.986 0.204 *** -0.991 0.204 *** -1.007 0.203 ***
2 -0.859 0.169 *** -0.865 0.169 *** -0.879 0.169 ***
3 -0.623 0.142 *** -0.628 0.142 *** -0.638 0.142 ***
4 -0.432 0.125 *** -0.436 0.125 *** -0.443 0.125 ***
5 -0.157 0.111 -0.162 0.111 -0.164 0.111
Gamma variance 1.077 1.077 1.088
# Observations 36529 36529 36529
-
Table 5: Survival Model: The Impact of Government Reforms
This table reports estimates using a discrete time-duration
model with time-varying covariates The
dependent variables in all models is a dummy (0/1) equal to one
in the first year that the firm exits from the
sample, and equal to zero otherwise. Explanatory variables are
defined in Table 1. Standard errors are
shown below, ***, **, * indicate significance at the 1%, 5%, and
10% level, respectively.
(1) (2) (3) (4) (5) (6) Coeff S.E. Coeff S.E. Coeff S.E. Coeff
S.E. Coeff S.E. Coeff S.E.
Abidjan -0.012 0.077 -0.012 0.077 -0.012 0.077 -0.013 0.077
-0.013 0.077 -0.012 0.077 Foreign_owned 0.270 0.062 *** 0.262 0.055
*** 0.260 0.055 *** 0.260 0.055 *** 0.261 0.055 *** 0.261 0.055 ***
Manufacture -0.007 0.073 -0.054 0.081 -0.057 0.059 -0.050 0.053
-0.050 0.053 -0.049 0.053 Trade -0.048 0.053 -0.049 0.054 -0.008
0.073 -0.009 0.072 -0.009 0.073 -0.008 0.073 Entry_small 1.541
0.127 *** 1.544 0.128 *** 1.539 0.127 *** 1.553 0.128 *** 1.537
0.127 *** 1.540 0.127 *** Entry_medium 0.872 0.118 *** 0.872 0.118
*** 0.870 0.118 *** 0.862 0.118 *** 0.844 0.121 *** 0.871 0.118 ***
Entry_large 0.441 0.148 *** 0.444 0.149 *** 0.440 0.148 *** 0.434
0.148 *** 0.439 0.148 *** 0.478 0.153 *** GDP_growth -0.015 0.004
*** -0.015 0.004 *** -0.015 0.004 *** -0.015 0.004 *** -0.015 0.004
*** -0.015 0.004 *** Reform_trade 0.249 0.097 *** 0.154 0.054 ***
0.122 0.062 ** 0.189 0.079 ** 0.167 0.059 *** 0.185 0.051 ***
Reform_fiscal 0.465 0.103 *** 0.470 0.058 *** 0.551 0.065 *** 0.611
0.082 *** 0.461 0.064 *** 0.512 0.055 *** Reform_monetary 0.359
0.130 *** 0.373 0.085 *** 0.345 0.098 *** 0.362 0.130 *** 0.331
0.092 *** 0.338 0.092 *** Reform_trade*foreign -0.095 0.110
Reform_fiscal*foreign 0.061 0.118 Reform_monetary*foreign -0.034
0.160 Reform_trade*manufacture 0.151 0.132
Reform_fiscal*manufacture 0.279 0.141 **
Reform_monetary*manufacture -0.247 0.226 Reform_trade*trade 0.156
0.098 * Reform_fiscal*trade -0.124 0.109 Reform_monetary*trade
-0.020 0.159 Reform_trade*small -0.023 0.097 Reform_fiscal*small
-0.171 0.105 * Reform_monetary*small -0.040 0.159
Reform_trade*medium 0.029 0.101 Reform_fiscal* medium 0.153 0.109
Reform_monetary* medium 0.023 0.172 Reform_trade*large -0.236 0.227
Reform_fiscal*large -0.071 0.223 Reform_monetary*large -0.001 0.170
Constant -2.367 0.202 *** -2.340 0.202 *** -2.356 0.202 *** -2.386
0.203 *** -2.356 0.201 *** -2.365 0.201 ***
1 -0.826 0.200 *** -0.841 0.200 *** -0.823 0.199 *** -0.801
0.201 *** -0.817 0.200 *** -0.821 0.199 *** 2 -0.727 0.166 ***
-0.741 0.166 *** -0.726 0.165 *** -0.709 0.167 *** -0.721 0.166 ***
-0.723 0.17 *** 3 -0.549 0.140 *** -0.558 0.140 *** -0.544 0.140
*** -0.536 0.141 *** -0.544 0.140 *** -0.544 0.140 *** 4 -0.476
0.126 *** -0.487 0.126 *** -0.478 0.126 *** -0.471 0.127 *** -0.475
0.126 *** -0.464 0.126 *** 5 -0.277 0.113 ** -0.291 0.113 ***
-0.278 0.113 ** -0.294 0.114 *** -0.282 0.113 ** -0.272 0.113 **
Gamma variance 0.959 0.967 0.958 0.940 0.953 0.959 # Observations
36529 36529 36529 36529 36529 36529