«A Dissertation Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy Approved November 2014 by the Graduate ...»
It should be noted that environmental perception of water quality is an approximate, not a direct measurement of water quality for agricultural activities.
However, besides data limitation, environmental perception is appropriate variable to be treated as natural capital since perception is, in many cases, based on an individual’s everyday observations and the information shared among people with similar interest.
Therefore, environmental perceptions of water quality, which is used as natural capital in
surrounding a given household.
Physical capital Physical capital is measured with agricultural equipment, consumer items, and housing quality, and these variables are from Household Agriculture and Consumption Survey. Agricultural equipment measures how many pieces of agricultural equipment, such as tractor, cart, irrigation pump set, gobar gas plant, or others (thresher, chaff cutter, sprayer, or any), a household owns. The range of the value is, therefore, between 0 and 5.
Consumer items measures how many consumer items, such as radio, television, bicycle, motorbike, a household owns. The range of the value is between 0 and 5.
An index for housing quality is also created. This index consists of the number of stories and the material of wall, roof, and floor. One story of a house adds one point to the index, so the range is from 1 to 5. For materials used to build a wall, concrete adds 6 points, brick 5, stone 4, wood 3, mud 2 and cane with mud 1. For materials used to build roof and floor, concrete adds 4 point, brick 3, wood 2, and mud 1. The range of the index is, therefore, from 4 to 19.
Financial capital Financial capital is measured with livestock and poultry, and these variables are from the Household Agriculture and Consumption Survey. Livestock is the total number of livestock, such as cows, bullocks, buffaloes, sheep, goats, and pigs, weighted by the value of each livestock based on the study by Regmi (1999). For example, cows are more
multiplied by.69 while the number of pigs is multiplied by.30, and the sum of those two numbers are used for livestock. In detail, the number of cows is multiplied by.69, bullocks by.96, male buffalos by.95, female buffalos by.71, sheep or goats by.25, and pigs by.30. Poultry as an independent variable, just like the number of poultry as a dependent variable, includes chickens, ducks, and pigeons, and the total number of all poultry is used.
Social capital Social capital is measured with the proportion of each dependent variable in the same neighborhood as the focal household. For example, when the size of farming land and the transition out of farming are analyzed, the proportion of farming households in the same neighborhood is used as the social capital variable. The farming status of a given household is excluded from the computation of the variable. When the size of land for rent is analyzed, the proportion of households renting their land out in the same neighborhood is used as a social capital variable. For the number of poultry, the proportion of households raising any poultry is used, and for the amount of chemical fertilizer, the proportion of households using chemical fertilizer in the same neighborhood is used. When it comes to the transition to the first salary employment and the transition to the first business outside the home, the proportion of individuals who had a salary job and the proportion of individuals who had out of home business in each year are used as a social capital variable, respectively. Any neighborhoods that have only one household in the sample are excluded from the sample. The major drawback of these
used. Thus, the number of households used for the computation for each neighborhood might be as small as three households for some neighborhoods.
Measurements of individual characteristics Education: this variable measures an individual’s level of education in years. The question asks the highest grade in school or year of college he or she has completed. It is a continuous variable.
Previous education experience of mother and father: this variable measures whether or not an individual’s mother and father went to school, separately. The question asks if a mother or a father went to school by 1996, the start year of the Chitwan Valley Family Study project, so these variables are dichotomous.
Previous work experience of mother and father: this variable measures whether or not an individual’s mother and father ever worked out of home, separately. The question asks if a mother or a father ever worked for pay out of home before an individual was younger than twelve years old, so these variables are dichotomous.
Descriptive Statistics In the first step of the analysis, the characteristics of the individuals, households, and neighborhoods are summarized in Table 2, Table 3, Table 4, and Table 5.
Due to the complexity of the datasets used in this dissertation, I chose two datasets to present descriptive statistics: one at the household level and the other at the individual level. Both datasets properly represent all the datasets used for the following analyses which are the subsample of these two datasets. Since migration is hypothesized to have a positive or negative association with agricultural and energy transitions, descriptive statistics are summarized by migration status except for the case of the variables measured at the neighborhood level. The reason is that those neighborhood-level variables, all social capital variables, have the same values for the households or individuals in the same neighborhood, and neighborhoods cannot be divided by migration experience at the household level. Migration status is dichotomous, and if a household or an individual has any migration experience in each time period, it is considered to be a household with at least one migrant or a migrant.
The descriptive statistics show a few notable differences between the households with migrants and the households without migrants. First, Table 2 and Table 3 summarize the characteristics of the households in Chitwan, Nepal, by migration status.
The results are based on the primary dataset mainly using the Household Registry. Those households that had at least one migrant in the first period between 1996 and 2001 tend to have less household members of young age, but more household members of working
education, better quality land (khet), and raise more poultry, compared to non-migration households. These patterns are the same in the second period, between 2001 and 2006. In addition, the average duration of migration among the households with migrants is about three years (35-39 months) within both time periods, and it is about three months higher in the second time period.
The characteristics of the individuals in the Chitwan Valley by migration status is summarized in Table 4. The results are based on the second dataset mainly using the Life History Calendar. This is mainly for the analysis of the changes in the modes of production, especially for the analyses of the transitions to the first salary and outside home business at the individual level. Those individuals that had any migration experience tend to be male, older, and more educated than those individuals without any migration experience. In addition, they are likely to have fathers that are more educated and had working experience in the non-farm sector before their child was twelve years old.
Last, the characteristics of the neighborhoods in Chitwan, Nepal is summarized in Table 5 by mainly using the Neighborhood History Calendar merged with the Household Registry. Within both time periods, the neighborhoods tend to consist of the households whose main modes of production are farming and that are not likely to rent their land out. On average, more than half of the households in the same neighborhood raise poultry, and about seventy percent of the households use chemical fertilizer. And about one third uses modern energy sources, such as gas, kerosene, or electricity. If we compare two time periods, the numbers show a hint of agricultural and energy transitions
more households renting their land out and using modern energy sources, and less households raising poultry. One exception is the case of a proportion of households using chemical fertilizer: lower proportion of households uses chemical fertilizer over time.
Size of Farming Land. The first analysis is to test the effects of migration and household capitals on the changes in the size of farming land. This is a household-level analysis. As discussed, the most appropriate model for this analysis is the first difference model. In this model, the changes in the size of farming land between 2001 and 2006 are the results of the changes in migration and the changes in household capitals between 1996 and 2001. In case a household does not do farming, the size of farming land is considered to be zero square meters. As a result, all the households regardless of farming status are included in the sample. The first difference model examines the changes in a dependent variable, so this is the most appropriate way to capture those households that transitioned out of farming between 2001 and 2006 or vice versa.
The descriptive statistics of the size of farming land by migration status is presented in Table 6. In both years, 2001 and 2006, the size of farming land is smaller for the households without migrants compared to the households with at least one migrant. The difference is about 2,400 square meters in 2001, and it is about 1,500 square meters in 2006.
The results of the first different model are summarized in Table 7A. Two models test the main effects of the changes in migration and household capitals on the changes in the size of farming land. Model 1 includes only migration, and household capitals are added in Model 2. The result of Model 1 shows that duration of migration does have positive impact on the size of farming land. One additional month in the duration of previous migration is associated with 20.51 square meters increase in the size of farming land. When household capitals are controlled in Model 2, migration is still
physical capitals play significant roles in the changes in the size of farming land. In terms of human capital, additional labor in a house increases the size of farming land. One additional household member in young and working age increases the size of farming land by 284.5 and 534.0 square meters, respectively. Considering that productivity of a person for farming would be peaked of working age, the results meet the expectations.
On the other hand, one additional year in education for the youngest household member decreases the size of farming land by 59.3 square meters. This result also meets the expectation since it is likely that highly educated young generations tend to find their future opportunities in the non-farm sector as a society urbanizes. Natural capital has significant impacts as well. Possessing bari (upland) or khet land (low irrigated land) decreases the size of farming land by 593.4 and 822.3 square meters, respectively. These results imply that land possession is more likely to sustain and stabilize their livelihood and lead them to seek non-farming opportunities rather than to expand their agricultural activities. Additional consumer item, which is a physical capital, also decrease the size of farming land. Having more consumer items might be one of the potential indicators that a household is going through ideational changes and moving out of farming. Thus, the negative impact of consumer items meets the expectation. Last, housing quality affects the size of farming land in the same direction: one unit increase in household quality is associated with 151.7 square meters decrease in the size of farming land. This result could mean that those households having a good quality house spent their accumulated resources in housing already, so they do not need to expand farming anymore but to sustain and stabilize their current livelihood.
interactions between migration and human capital, Model 4 tests the interactions between migration and natural capital, Model 5 tests the interactions between migration and physical capital, Model 6 tests the interactions between migration and financial capital, and Model 7 tests the interactions between migration and social capital. The interaction results show significant interaction effects between migration and household capitals except social capital.
The interaction between migration and human capital, especially the number of household members in young and working ages, and the education level of the youngest household member, has significant impacts on the size of farming land. The interactions between duration of migration and number of household members in young and working ages are visualized in Figure 3 and Figure 4. The results show that available extra labor plays a significant role in repressing the negative effect of migration on the changes in the size of farming land. For example, when there was one additional household member of young age between 1996 and 2001, a household reduces the size of farming land less compared to a household who lost one young household member in the same period. In other words, as the labor in a household increases, the negative effect of migration reduces. The interaction between migration and number of household member of working age shows essentially the same pattern. In sum, these results indicate that a household rich in human capital would have more options to keep farming even at a lesser degree despite the loss of labor due to migration, but a household poor in human capital would not have such an option and be more likely to stay away from farming as the duration of migration increases.
It shows that there is a gap in the size of farming land between a household with a young member who received more education and a household with a young member who did not receive any more education in the same period in the case of no migration experience.