market – such as soybean – travel time is likely to have a negative effect on the
probability that they are produced and on the amount of land area devoted to them.
To the extent that upland rice and paddy rice are grown for subsistence, this effect is
expected to be insignificant. Moreover, if the only reason that the crop is grown is
that it is a substitute for a staple that can be bought in the market, then the travel
time coefficient is likely to be positive. The variable in the dataset measures the
‘average time taken one way, in minutes, to reach the market, using the most
popular mode of transport.’ It thus takes into consideration mode of transportation
and road quality.
21
Proportion of Adult Population The proportion of adults in the village is
expected to positively affect land brought under cultivation and the intensity with
which it is cultivated. Adult labor is required to grow crops on virgin land that
requires preparation.
22,23 The presence of more adults is likely to increase the
amount of land cultivated and ameliorate labor scarcity. In this study, proportion
of adult population is used as a proxy for available labor in the village and for the
opportunity cost of labor.
Productivity of Land There are two variables that are used as a proxy to measure
land productivity. These are water availability and a dummy for acidic soil.
24
(Please see below.) Additionally I also use a time invariant binary variable to
indicate whether the village grew high yielding varieties (HYV) of rice at any
time during the study period. (So HYV rice dummy ¼1 if the village ever grew
HYV rice during the study period, and ¼0 otherwise). I expect this variable to have
two impacts on productivity. The first is on paddy rice area: HYV rice is more
productive than non-HYV rice. I expect it to have a positive effect on area devoted
to paddy rice. The other effect this variable is likely to be a proxy for is the presence
(or absence) of ‘attention’ from local authorities. To the extent that growing HYV
rice requires additional knowledge and training provided by field officers and that
21 See for example Dawson and Barwell (1993).
22 It would be useful to gauge the different impacts of adult males and adult females.
23 See Godoy et al. (1997) for a similar argument.
24 Another possible variable is yield per acre but there are problems with measuring the variable
since it is measured only when crop data are available. It is also potentially endogenous. For
example for upland rice yield/hectare is available only for 541 observations, or 248 villages for at
least one point in time. For the subset of variables for which data are available: For soybean, there
is a positive time trend when the log of productivity is regressed on year, while controlling for
other variables (~3 %). when we regress this variable for soybean on time dummies, the time
dummies are insignificant (and indeed in the first two years, negative, compared to 1986. They are
positive in the next 2 years but insignificant. Only in 1996 is the time dummy significant and
positive – when an average increase of almost 30 % occurs). Similarly for upland rice, the time
trend is not significant or large (although it is positive). This indicates that there were not very
many productivity increases among farmers located in Forest Reserve villages of Chiang Mai,
during 1986–1996, although some may have taken place in the last year of the study period, for
soybean. Witnessing an increase in area despite there being an increase in productivity, further
strengthen my results.
142
J. Puri
probability that they are produced and on the amount of land area devoted to them.
To the extent that upland rice and paddy rice are grown for subsistence, this effect is
expected to be insignificant. Moreover, if the only reason that the crop is grown is
that it is a substitute for a staple that can be bought in the market, then the travel
time coefficient is likely to be positive. The variable in the dataset measures the
‘average time taken one way, in minutes, to reach the market, using the most
popular mode of transport.’ It thus takes into consideration mode of transportation
and road quality.
21
Proportion of Adult Population The proportion of adults in the village is
expected to positively affect land brought under cultivation and the intensity with
which it is cultivated. Adult labor is required to grow crops on virgin land that
requires preparation.
22,23 The presence of more adults is likely to increase the
amount of land cultivated and ameliorate labor scarcity. In this study, proportion
of adult population is used as a proxy for available labor in the village and for the
opportunity cost of labor.
Productivity of Land There are two variables that are used as a proxy to measure
land productivity. These are water availability and a dummy for acidic soil.
24
(Please see below.) Additionally I also use a time invariant binary variable to
indicate whether the village grew high yielding varieties (HYV) of rice at any
time during the study period. (So HYV rice dummy ¼1 if the village ever grew
HYV rice during the study period, and ¼0 otherwise). I expect this variable to have
two impacts on productivity. The first is on paddy rice area: HYV rice is more
productive than non-HYV rice. I expect it to have a positive effect on area devoted
to paddy rice. The other effect this variable is likely to be a proxy for is the presence
(or absence) of ‘attention’ from local authorities. To the extent that growing HYV
rice requires additional knowledge and training provided by field officers and that
21 See for example Dawson and Barwell (1993).
22 It would be useful to gauge the different impacts of adult males and adult females.
23 See Godoy et al. (1997) for a similar argument.
24 Another possible variable is yield per acre but there are problems with measuring the variable
since it is measured only when crop data are available. It is also potentially endogenous. For
example for upland rice yield/hectare is available only for 541 observations, or 248 villages for at
least one point in time. For the subset of variables for which data are available: For soybean, there
is a positive time trend when the log of productivity is regressed on year, while controlling for
other variables (~3 %). when we regress this variable for soybean on time dummies, the time
dummies are insignificant (and indeed in the first two years, negative, compared to 1986. They are
positive in the next 2 years but insignificant. Only in 1996 is the time dummy significant and
positive – when an average increase of almost 30 % occurs). Similarly for upland rice, the time
trend is not significant or large (although it is positive). This indicates that there were not very
many productivity increases among farmers located in Forest Reserve villages of Chiang Mai,
during 1986–1996, although some may have taken place in the last year of the study period, for
soybean. Witnessing an increase in area despite there being an increase in productivity, further
strengthen my results.
142
J. Puri