off-farm labour participation of farmers and spouses alessandro corsi university of turin
TRANSCRIPT
Off-farm labour participation of Off-farm labour participation of farmers and spousesfarmers and spouses
Alessandro Corsi
University of Turin
The problemThe problem
Off-farm work is widespreadIt helps the adjustment process of
farmers to new market conditionsIt is important to analyse the
variables that influence the choice of working off the farm
Theoretical modelTheoretical model
Off-farm work participation is a dichotomous variable (may be yes or no)
The farmer chooses to work off the farm if the market wage is larger than the reservation wage (= the minimum wage for which he is willing to work off the farm)
Theoretical modelTheoretical model
w*
O family labour
inco
me
Theoretical modelTheoretical model
O
inco
me
family labour
Theoretical modelTheoretical model
w
O
inco
me
family labour
Theoretical modelTheoretical model
The reservation wage therefore depends on:
personal characteristics (age, sex, education, etc.)
household characteristics (e.g., number of children)
farm characteristics (size, farming system, etc.)
Theoretical modelTheoretical model
The market wage depends on: personal characteristics (age, sex, education, etc) characteristics of the labour market
Theoretical modelTheoretical model
The farmer will have an off-farm job if:
market wage > reservation wage
W > W*
Theoretical modelTheoretical model
w
w*
O family labour
inco
me
Theoretical modelTheoretical model
The market wage can be written:
iLtPiwi '2
'10
The reservation wage can be written:
iH tFiPiwi *'3
*'2
*'10
*
Theoretical modelTheoretical model
The difference between the market and the reservation wage, w - w* is :
ii
H iFiPio
LiPioyi
'3
'2
'1
'2
'1
Theoretical modelTheoretical model
For brevity, y*can be written as:
iX iyi
'
(X are all the explanatory variables, and
is the random term)
Theoretical modelTheoretical model y* cannot be observed; it can only be observed if the farmer works off the
farm or not. Then:
Pr[off-farm work] = Pr[y* > 0] =
= Pr[’X < ] = [’X]
( is the cumulative probability of the random variable , assumed to be normal)
Theoretical modelTheoretical model
X
Prob[ < X]=Prob[off-farm job]
Pro
b[
]
Theoretical modelTheoretical model
The parameters of the equation can be estimated through a probit model It yields the probability of the outcome (off-farm yes or no) as a function
of the explanatory variables It is also possible to estimate the change in probability resulting from a
change in the explanatory variable (marginal effect)
DataData
351 farms in Pennsylvania surveyed in 1985 and again in 1991
351 farm operators 344 spouses
DataData
data on personal characteristics: age, sex, education
data on household characteristics: # children of different age
DataData
data on farm characteristics: farm size principal farm enterprise (dairy, other labour
intensive, all-year-round or seasonal - dummy variables)
DataData
characteristics of the labour market employment share by sector ratio of average nonfarm to farm incomes unemployment rate
ResultsResults
Models estimated for operators and spouses:
fitting results comment
ResultsResults
Observations: 351Log-Likelihood -163.002Log-Likelihood (slopes=0) -229.142LR test of the model: 2
(d.f.)132.279
(15)Correct predictions (%):Total 78.9Not working off the farm 87.6Working off the farm 63.5
Operators
ResultsResultsOperators
Variables Coeff. t-values Partialderivatives
Constant -10.462 -3.57AGE 0.294 3.73 0.103AGE2 -0.003 -4.12 -0.001EDUC 0.099 2.68 0.035CHILD<5 0.020 0.07 0.007CHILD<17 0.066 0.75 0.023CHILD<30 -0.007 -0.06 -0.002ACRES -0.001 -3.78 -0.00048DAIRY -1.426 -7.03 -0.500LABINT -0.639 -1.89 -0.224SEASINT -0.449 -1.52 -0.158TRADE -0.183 -3.00 -0.643HIGHSERV 0.447 2.59 0.157LOWSERV -0.002 -0.05 -0.0007MANUFACT 0.036 1.70 0.013INCOMGAP 3.001 2.91 1.054
ResultsResults
FOR OPERATORS: Personal characteristics have a significant impact
on off-farm labour participation The same is true for farm characteristics and
labour market characteristics Household characteristics do not significantly
affect operators’ choices
ResultsResultsSpouses
Observations: 334Log-Likelihood -186.797Log-Likelihood (slopes=0) -220.834LR test of the model: 2
(d.f.)68.074
(15)Correct predictions (%):Total 71.0Not working off the farm 85.2Working off the farm 47.2
ResultsResultsSpouses
Variables Coeff. t-values Partialderivatives
Constant -6.895 -2.60AGE 0.085 1.17 0.315AGE2 -0.001 -1.80 -0.05EDUC 0.193 5.13 -0.071CHILD<5 -0.549 -1.98 -0.204CHILD<17 -0.144 -1.67 -0.053CHILD<30 -0.102 -0.88 -0.038ACRES -0.000 -1.38 -0.0016DAIRY -0.199 -1.13 -0.074LABINT -0.011 -0.03 -0.004SEASINT -0.473 -1.60 -0.175TRADE 0.049 0.87 0.181HIGHSERV 0.211 1.35 0.078LOWSERV 0.009 0.26 0.003MANUFACT 0.043 2.20 0.016INCOMGAP 0.117 0.12 0.043
ResultsResults
FOR SPOUSES: Among personal characteristics, only education has a
significant impact on off-farm labour participation Household characteristics, particularly small children,
significantly affect spouses’ choices Farm characteristics have no influence Among labour market characteristics, only low-wage
manufacturing employment increases the probability of off-farm work
ResultsResults Further results can be drawn from more sophisticated
econometric methods by using data from both surveys Farmers and spouses who choose an off-farm work in the
past are more likely to make the same choice in the following For farmers, this is most likely because when they started an
off-farm work they modified the farm, so that it is not easy to come back
For spouses, this is most likely because they accumulated work experience, and hence, have higher market wages.
ConclusionsConclusions This is an example of how econometric methods can
be used to assess empirical questions The results are consistent with the theory, but more
detail has been gained It is possible to make predictions of what will happen if
some explanatory variable will change It is possible to detail these effects for farmers and
spouses (who exhibit different behaviour), for small and large farms, etc.
THANK YOU FOR YOUR ATTENTION!