Duration analysis to determine factors
that lead wine and processed meat agroindustries to abandon their activities:
Policy implications
Zein KALLAS and Fatima LAMBARRAA
zein.kallas@upc.edu and fatima.lambarraa@upc.edu
The XXVII conference of the International Association of Agricultural Economists, 16-22 August 2009, Beijing, China.
Center for Agro-food Economy and Development (CREDA)- Polytechnic University of Catalonia (UPC)- Food and Agriculture Investigation and Technology Institute (IRTA).
Department of Agricultural Engineering and Biotechnology. Edifici ESAB- Avinguda del Canal Olímpic s/n, 08860-Castelldefels (Barcelona)- Spain.
1. INTRODUCTION
3. METHODOLOGY
 This Work focuses on assessing the
determinants factors that drive
agro-industries to abandon their
activities as well the timing of their
decision. Moreover, we analyze
agro-industry exit and Technical
Efficiency.
 First step we used the stochastic
frontier methodology to measure
firm’s technical efficiency. (TE)
 We analyze two different agroindustry sector in Catalonia (Spain);
the wine and processed meat
sectors.
 The selection of both sectors has
been motivated by the strategic
position they occupy within the
Catalonian agro-industry sector:
 Both have a relevant economic
function representing 43% of the
total sales.
 The input-output tables show a
strong interrelationship with
other economic sectors. such as
agricultural input suppliers (50% of
total input), logistic, banking,
technological, etc.
 They play an important social
and territorial role employing the
equivalent of 31,534 full-time
workers, representing 42% of the
labor force in the Catalonian agroindustry sector.
 Results allow policy makers to
obtain information about agroindustries with high hazard to
abandon market. Thus, decision
could be taken to maintain the
social fabric associated to these
agro-industries.
 Second step we
apply The
Duration Analysis in order to
determine not only why firms exit but
also the timing of dissolution and the
factors that influence the observed
time patterns.
3.1. STOCHASTIC FRONTIER
METHODOLOGY
 The stochastic frontier production
function can be expressed as follows:
yit  f ( xit , t ;  )e
vit -uit
where yit is the output of the i-th firm
(i=1...N) in period (t=1...T), where
f(xit,t,) represents the production
technology, xit is a (1K) vector of
inputs and other factors influencing
production associated with the i-th
firm in period t, and  is a (K1)
vector of unknown parameters to be
estimated.
 Disturbance term is composed of
two parts: vit and uit . The former
captures the effects of statistical
noise outside the firm's control. The
latter is associated with output
oriented technical inefficiencies.
 The distributions of the two error
terms are independent:


uit  exp    t  T   ui
 Maximum likelihood techniques are
used for estimation.
2. DETERMINANTS OF EXIT
 Several factors can influence the
decision to market exit and the
likelihood of firm dissolution:
 Manager characteristics and
risk behavior; Age, gender,
education, experience, attitudes,
risk aversion, etc.
 Firm characteristics; firm size
(Capital, employee number), firm
age, legal form, location, etc.
 Firm management; number of
brand, Total Assets Turnover,
diversification of activities and
products, economies of scale,
Technical Efficiency, etc.
 Economic and financial results;
profitability, leverage, solvency,
current ratio,, etc.
 Exogenous factors: market size
and
concentration,
industry
growth, input prices, subsidies,
information access, reforms, etc.
3.2 . DURATION ANALYSIS
 Duration Analysis (DA) models the
time length of a spell or “event”. The
spell starts at the time of entry into a
specific state (entry decision) and
ends at a point when a new state is
entered (exit decision).
 The conceptual foundations of DA
rely on probability theory. We
define the hazard function that
specifies the rate at which a spell is
completed at time t=T , given it
survives until time t . in our analysis,
the hazard function represents the
probability that a farmer exit at time t
, given he has not adopted before t :
h(t )  lim
Pr (t  T  t  t T  t )
t
F (t  t )  F (t )
 lim
t  0
t S (t )
f (t )

S (t )
t  0
 A set of explanatory variables of
economic and non-economic nature
influence and alter the distribution
of the duration:
h(t , x,θ,β)  lim
0
5 RESULTS AND DISCUSSION
Table 1: Results from partial likelihood
estimation for COX proportional Hazard
model (processed meat agro-industry).
Pr (t  T  t   T  t )
Parameter
Std.
Error
P-value
Hazard
Ratio
Date of firms’ setting -up
0.0012***
0.0003
0.0001
1.0010
Leverage
2.4526**
0.9936
0.0136
11.6180
-1.5904***
0.5173
0.0021
0.2040
Return on capital employed (%)
-2.1821*
-0.0076
0.4579**
-0.0011
0.3269
1.1754
0.0062
0.2325
0.0009
0.2613
0.0634
0.2176
0.0488
0.2319
0.2109
0.1130
0.9920
1.5810
0.9990
1.3870
Dummy year 2000
2.0176*
1.2008
0.0929
7.5200

Variable
where  is a vector of unknown
parameters of x , the vector of
explanatory variables and  is a
vector
of
parameters
that
characterize the distribution function
of the hazard rate.
 To estimate the duration model we
use the semiparametric Cox
proportional hazards model. Under
this model the duration of each
Current Ratio
Technical Efficiency
Employee number
Business extraordinary results
Profit growth
Likelihood Ratio: 96.20 (0.000)
Wald test: 45.11 (0.000)
Lagrange Multiplier Test: 75.76 (0.000)
Table 2: Results from partial likelihood
estimation for COX proportional Hazard
model (wine agro-industry)
member of a population is
assumed to follow its own hazard
function:
hi (t )  h (t; xi )  h0 (t ) exp(xi' β)  h0 (t ) exp( 1xi1 
 k xik )
 The estimation procedure is based on
the
partial
likelihood
function
introduced by Cox (1972, 1975)

βx i
n 
e
PL    n
βx j
i 1 
Y
e

ij

 j 1






i
4 . EMPIRICAL APPLICATION
 Data used in this analysis were
obtained from the ‘Sistema Anual de
Balances Ibéricos’ (SABI) database.
The
SABI
collects
extensive
economic
and
non
economic
information on the firms.
 The sample size for the wine and
processed meat sectors is formed
by 231 and 288 active firms. While
the inactive industries are 20 and 47
respectively.
 For
the
stochastic
frontier
production yit is defined as the
deflated total sales in meat and wine
product. xit is a (13) vector that
contains three inputs: a) wage
expenses, b) intermediate inputs
and c) capital employed in the
production process.
 For the DA, the dependent variable
is the time firms waited before exit
market. from the set of explanatory
variables, we use: a) age of firm
represented by the Date of firms’
setting–up, c) leverage (ratio of debt
to total assets), d) current ratio (is
an indication of a company's ability to
meet short-term debt), e) Technical
efficiency, f) firm size (employee)
g) extraordinary results, h) Profit
growth and i) Return on capital
employed .
Parameter
Std.
Error
P-value
Hazard
Ratio
Date of firms’ setting -up
0.0009***
0.0003
0.0010
1.0010
Leverage
3.4227***
1.1542
0.0030
30.6530
-0.4604
0.3476
0.1853
0.6310
Return on capital employed (%)
-2.8863*
-0.1406**
0.8460***
-0.0005
1.6610
1.7368
0.0630
0.3087
0.0011
1.4263
0.0965
0.0257
0.0061
0.6596
0.2442
0.0560
0.8690
2.3300
1.0000
5.2640
Dummy year 1990
-2.1924*
1.2155
0.0713
0.1120
Variable
Current Ratio
Technical Efficiency
Employee number
Business extraordinary results
Profit growth
Likelihood Ratio: 44.91 (0.000)
Wald test: 22.29 (0.003)
Lagrange Multiplier Test: 24.70 (0.008)
In both sector, results demonstrates that
factors which increase the likelihood of
exit
are:
leverage
and
positive
extraordinary results as a results of
disinvestment decisions.
Factors that decrease the hazard of
abandoning market are: age (Date of
setting-up), Current Ratio, Technical
Efficiency, firm size and Technical
efficiency. However there are small
difference between sectors.
Dummy variables (2000 for meat and
1990 for wine are relevant in explaining the
hazard of market exit as a result of the
PAC reforms and regulations.
Results allow policy makers to focus
on agro-industries with higher hazard.
6. REFERENCES
1.
Agarwal R.; Gort M. 1996. The evolution of markets and entry, exit and
survival of firms. The review of economics and statistics, 78: 489-498.
2.
Aigner, D.J.; Lovell, C.A.; Schmidt, P.J. 1977. Formulation and
Estimation of Stochastic Frontier Production Function Models. Journal
of Econometrics, 6: 21-37.
3.
Allison, P.D. 2001. Survival Analysis Using the SAS System: A Practical
Guide. SAS publishing.
4.
Audretsch, D.B., Houweling, P.; Thurik, A.R. 2000. Firm survival in
the Netherlands. Review of Industrial Organisation, 16: 1-11.
5.
Audretsch, D.B. 1994. Business survival and the decision to exit.
Journal of Economics and Business, 1: 125-137.
6.
Battese, G.E.; Coelli, T.J. 1992. Frontier production functions, technical
efficiency and panel data: With application to paddy farmers in India.
Journal of Productivity Analysis, 3: 153-69.
7.
Dimara, E.; Skuras, D.; Tsekouras, K.; Tzelepis, D. 2003. Firm
efficiency and survival. Paper presented at the 2nd Hellenic Workshop
on Productivity and Efficiency Measurement. University of Patras,
Greece.
8.
Doi N (1999) The determinants of firm exit in Japanese manufacturing
industries. Small Business Economics,13:331–337.
9.
Efron, B. 1977. The efficiency of Cox's likelihood function for censored
data. Journal of the American Statistical Association, 76: 312-319.
10. Kumbhakar, S.C.; Lovell, C.A. 2000. Stochastic Frontier Analysis, New
York, Cambridge University Press.
11. Lambarraa, F.; Serra, T.; Gil, J.M. 2007. Technical efficiency and
decomposition of productivity growth of Spanish olive farms. Spanish
Journal of Agricultural Research, 5(3). 259-270.
12. Tsionas, E.G.; Papadpgpnas, A.T. 2006. Firm exit and technical
efficiency. Empirical economics, 31:535-548.