Fertilizer Subsidy Program a Strategy of Addressing Low Maize Productivity in Kakamega County, Kenya (Published)
Kakamega is one of the most populated Counties of Kenya. Due to continuous cultivation of the land and heavy rainfall received, it is prone to loss of soil nutrients through leaching. This has resulted to reduced crop productivity in the region. A number of strategies have been initiated by the government to promote soil fertility. Fertilizer subsidy program is one of the policy instruments so far implemented despite heavily criticized. The paper provides findings of a study carried out to analyze the effect of adoption of fertilizer subsidy program on maize productivity among the small scale farmers. Cross sectional data was collected from 300 respondents who were selected using multistage sampling procedure in Kakamega County. Questionnaires were used to collect primary data from randomly selected households. A probit and Tobit two stage models- controlling for programme selection bias were used to determine the probability of adoption and the effect of participation in the fertilizer subsidy program on maize productivity respectively. Maize productivity was significantly influenced by the subsidy program participation at 5% level. The average maize productivity was 2.21t/ha. Farmers who benefited from the program had an average productivity of 2.46t/ha significantly higher than non-participants who had an average of 1.97t/ha. Tobit model results showed that subsidy program participation had a likelihood of increasing productivity by 0.323t/ha. However, there is an existing yield gap between what is possible with the existing technologies and what farmers currently achieve. The cause of this is low average fertilizer usage to make an impact on productivity as compared to the national recommended application rates. Hence the study recommends expansion of the subsidy program by increasing quantities of fertilizer supplied to incorporate more farmers in the program.
Application of Least Square Dummy Variable (LSDV) in Estimation of Compensation of Employee (Published)
This research was conducted to estimate compensation of employee using least square dummy variable (LSDV) regression model. The data used in this work were secondary data sourced from National Bureau of Statistics (NBS) from 1981 to 2006. The variables considered were compensation of employee as the dependent variable, fixed capital, price of goods, tax and surplus as the independent variables. The data were analyzed using (STATA 13). The results obtained revealed that F-value of 3874.05 was statistically high suggesting the overall model was good fitted. The R2 -value 0.9989 was also high which indicated that 99.89% of the total variation was accounted for by the independent variables included in model while the remaining 0.11% unexplained was accounted for by the white noise. Again, all the differential intercept coefficients have negative signs. Also, several differential slope coefficients have negative signs which implied that they were negatively related to compensation. Again, the result revealed that compensation is not statistically significantly related to fixed capital, price, tax and surplus. However, none of the differential slope coefficients is statistically significant. Of all the three differential intercept coefficients only was statistically significant. Since none of the differential slope coefficients was statistically significant, it concluded that the differential slope coefficients are not different from the slope coefficient of the base/comparison group (power sector.
Presented and discussed in this paper is the use of dummy variable multiple regression techniques in the analysis of samples drawn from several related or dependent populations ordinarily appropriate for random effects and mixed effects two factor analysis of variances model with one observation per-cell or treatment combinations. Using the extra sum of squares principle the method develops necessary sums of squares, degrees of freedom and the F-ratios required test to the significance of factor level effects thereby helping resolve the problem of one observation per treatment combination, encountered in the usual two factor analysis of variance models with one observation per cell. The method provides estimates of the overall and factor mean effects comparable to those obtained with the two factor analysis of variance methods. In addition, the method also provides estimates of the total or absolute effects as well as the direct and indirect effects of the independent variables or factors on the dependent or criterion variable which are not ordinarily obtainable with the usual analysis of variance techniques. The proposed method is illustrated with some sample data and shown to compare favourably with the usual Friedman’s two-way analysis of variance test by ranks often used for the same purpose.