If households are homogeneous in the sense that they engage in a similar income generation and face a similar economic problem, the estimated production function provides quantitative insights on the effects of the mortality on income generation and the decomposition of the effects into productivity and each productive asset. Since productivity includes all heterogeneities among households except those in the number of household members, land and livestock, if we included more heterogeneities in input, productivity would become less ambiguous in what it includes. However, we do not think it is our primal objective. The definition of land Kjt is the size of land owned and land rented-in. Land Kjt includes all of the four types of land: owned and used owned and fallowed, owned and rented-out, and not-owned and rented-in. 85% of land is type and the most of the remaining 15% is type . Agricultural income Yjt is net agricultural rent, that is, we subtract paying rent from and add receiving rent to agricultural output/sale. Thus, we control the heterogeneity in land in ownership and renting. Manure from livestock is important for agriculture in Kagera. Smith documents that farmers use manure sparingly and efficiently, they mix ash, mulch and composted manure into the holes in which coffee and banana trees are planted and farmers who optimize their use of manure can produce yield up to five times higher than their neighbors who cannot afford cows . In his data, three-fifths of male farmers use manure and all the farmers interviewed wanted to buy a cow to increase their herd in order to improve farm productivity .
The importance of manure is due to the fact that most of Kagera farmer do not use fertilizer. For example, in original KHDS data, only 5.3% and 3.2% households use fertilizer in wave 1 and wave 5 ,fodder system respectively. Complementarity between crop production and livestock is mainly due to manure since households do not use cattle for plowing. Complementarity between land and livestock is weak since a household uses communal land for grazing instead of its own household’s private land. Complementarity between land and the number of household members is also weak since households use a cattle owner association called omukondo which has twenty or so member households, pasture area, and a herd manager and each household does not have to use its own household member for herding. In our constructed data with total 401 households, there are 160, 119, and 138 households who have zero monetary value of livestock in 1991, 1992 and 2003, respectively. In order to accommodate these household into our analysis, we define livestock Sjt is the real monetary value of livestock plus one.In this subsection, we will show the descriptive statistics in each productive asset and agricultural income in 1991 for households with and without prime-age adult mortality in order to check how these two groups of households are different in 1991 and whether the data support us in taking prime-age adult mortality between 1990 and 2003 as an exogenous shock. Then, we will show the difference-in-difference estimates of the change in each variables from 1991 to 2003. Table 4 show the mean of each variable in 1991 for households with and without prime-age adult mortality between 1990 and 2003.
We divide households simply into households with mortality and those without mortality. The table shows that there is not clear difference in productive assets and agricultural income in 1991 between households with and without prime age adult mortality. We test the null hypothesis that the mean of each variable for households without death is the same as one for households with death and we cannot reject any of that hypotheses even with 10% significance level. These results support us in taking prime-age adult mortality as an exogenous shock.Apparently, households without mortality accumulate total assets and increase total agricultural income more than households with mortality. We test the null hypothesis that average change in each variable for household without death is the same as one for households with death against the alternative hypothesis that the former is larger than the latter. The test for each variable rejects the null hypothesis at 5% significance level . However, there are not clear differences in change of per capita land, livestock, and agricultural income between households with and without mortality. We cannot reject the null hypothesis that change in per capita land and agricultural income and rent for households without death is the same as one for households with death. In order to check whether our observation in Table 5 is robust, we make figures of distribution of change in each variable for households with and without mortality and the figures confirm our observation. These results shows the possibility that households hit by mortality endogenously respond to the negative shock and adjust productive asset level in order to improve efficiency. Thus, it is interesting to ask whether there is the difference in productivity growth between households with and without mortality.
Note that we can reject the null hypothesis that change in per capita livestock for households without death is the same as one for households with death in favor of the alternative hypothesis that the former is larger than the latter. These results imply that households hit by adult death kept per capita land but per capita livestock to improve or keep per capita agricultural income.As we have already seen in Table 5, the table also shows that average agricultural income growth from 1991 to 2003 is negative for both types of households and households hit by prime age households mortality experienced more severe decrease in agricultural income than house-holds without the mortality. The table shows how much productivity growth and accumulation of each productive asset contribute to this negative agricultural income growth. The decomposition of average agricultural income growth for the households without prime age adult mortality shows that the decrease is mostly due to the decrease of productivity rather than the decrease in productive assets. The percentage of contribution of the decrease in productivity is 93%. The decomposition of income growth for the households hit by prime-age adult mortality shows that the households with mortality increase less every component of the decomposition than the households without the mortality. The percentage of the contribution of the decrease in productivity for the households with mortality is 80%, which is smaller than one for the households without mortality . These results imply that on average, households without mortality could kept their productive assets but the households hit by mortality could not. The third row shows that a half of the difference in agricultural income growth between households with and without prime-age adult mortality is due to the difference in productivity growth. The percentage of how much the difference in productivity growth explains the difference in agricultural income growth is 60% . The third row also shows that how much the difference in the accumulation of each productive asset consists of the difference in agricultural income growth between households with and without prime-age adult mortality.
The difference in the accumulation of household members consists the most and those of land and livestock follows. The difference in the accumulation of household members explain more than a half of the difference in accumulation of all three productive assets; the number is 60% . This is reasonable since prime-age adult mortality decreases the accumulation of household members directly and may decrease accumulation of land and livestock indirectly. We could interpret that the difference in the accumulation of household members is direct negative effects of prime-age adult mortality on agricultural production. Note that although we call it as direct negative effects, we do not mean it excludes households’ endogenous response to prime-age adult mortality. “Direct” means just that adult death directly decreases the number of household members. On the other hand, differences in productivity growth and the accumulation of land and livestock are indirect negative effects. Surprisingly, the results show that direct effects do not count for the largest part in the difference in agricultural income growth between households with and without prime-age adult mortality. Instead, fodder system for sale the difference in productivity growth plays the largest role to explain the difference in agricultural income growth. The percentage for the difference in productivity growth is 60% as we mentioned above while the percentage for the difference in accumulation of household members is 24% . These results imply that households hit by prime-age adult mortality could not cope with it and not accumulate not only household members but also land and livestock as much as households without death could. Furthermore, households with the mortality could not increase productivity as much as the other households could. Surprisingly, the negative effects on productivity growth are larger than negative effects on productive asset accumulation. The fourth row shows whether each variable for households without death is statistically significantly larger than one for households with death. Productivity growth for households without death is statistically significantly larger than one for households with death. The increase in income generating power due to accumulation of all productive assets as a whole and household members only for households without death is statistically significantly larger than one for households with death.
We can say that households hit by prime-age adult mortality could not increase income generating power in every factor among productivity growth and the accumulation of each productive asset as much as households without mortality could. A surprising result is that the difference in the accumulation of household members between households with and without mortality is not the largest factor in explaining the difference in agricultural income growth. This result implies the following two things: First, households hit by mortality could not increase or keep productivity and productive assets, land and livestock as much as households without mortality could. Thus, mortality destroys not only household human capital but also land, livestock and productivity indirectly. Second, a household hit by mortality responds to and mitigates the decrease in household members due to mortality somehow. We may think that the household tries to increase its household members or at least try to keep them by accommodating a new member through marriage or keeping current members who would move out of the household if there was no mortality. A households hit by mortality adjusts its amount of each productive asset after mortality in order to improve productivity. However, the results show that the magnitude of negative effects of prime-age adult mortality is so large that we can observe the differences in productivity growth and accumulation of each productive asset between households with and without mortality even in a long term of 13 years.The Morrill Act of 1862 established land-grant universities across the country with the purpose of educating the citizens about agriculture, home economics, and other practical professions.1 According to the Act, each state had to set aside acreage of federal land, the income from which would have to support a college or university for teaching ‘mechanical arts’ . Twenty-five years later, in 1887, the Hatch Act was passed, which established the allocation of federal funds to state agricultural experiment stations. The Smith-Lever Act of 1914 formalized the cooperative extension through the creation of a partnership between the land-grant research universities and the U.S Department of Agriculture. The Congress clearly stated the purpose of Extension: ‘to aid in diffusing among the people of the U.S. useful and practical information on subjects related to agriculture and home economics, and to encourage the application of the same’ . Funding for the Cooperative Extension would come from the Congress to the United States Department of Agriculture, which would then distribute it among the land-grant universities, matching the amount to the state- and county-level expenditures.2 The formula designed for allocation of funding for Cooperative Extensions mandated that the federal and state contribution would each amount to 40 percent, with county contributions amounting to 20 percent of the total . In this paper, we do not distinguish between the 1914 Act and the Hatch Act, as both provide funding for research and dissemination activities within Cooperative Extension.