Credit as a determinant of tobacco output: a case study of the former Tabora tobacco growers' cooperative society

In all spheres of life, economic, social and political, there often arises need of knowledge for expected changes in agricultural production, one or two years ahead. Usually, the inquiries will focus on what levels of output will most likely be produced; and their interest would be a real, regional or national in scope, rather than limited to farm levels. There is need thus for predictive and aggregate output response models suitable for answering such questions about short-term changes in agricultural production. The aim of the study was to evaluate the usefulness of agricultural credit as an important determinant of tobacco output using the former Tabora Tobacco Growers Co-operative Society as a case study. Several determinants were identified but only four were quantifiable. The number of farmers (representing labour inputs), number of hectares under the crop, credit inputs and rainfall were regressed against out put using both the Cobb-Douglas and linear models. A step-wise regression was used for the Cobb -Douglas model and the findings are summarized below. (Charter 3 was devoted to detailed discussion on this). In the simple regressions, most of the coefficients were positive and significant both at the 1 per cent and 0.5 per cent levels. The effect of rainfall on output was found to be negative, though the coefficients were also significant. In all equations, the adjusted R2 was highly significance and no serial correlation was detected. In the multiple regression, the constant and B1 were both positive and significant at 0.5 percent level of significance. B2 was only significant at a confidence level of between 60 per cent to 65 percent; The B3 coefficient came up with a negative sign and was not significant even at 60 per cent confidence interval. multi collinearity was probably responsible for the low significance of some of the coefficients and the negative sign. When the fourth variable was introduced into the Model, none of the coefficients was significant at 1 per cent level. The adjusted R2 was still significant at 0.5 percent level and no auto correlation was detected.The linear model was tried for comparison purposes. Two of the constants in the simple regressions were of negative signs and were only significant at 60 per cent and 90 per cent confidence intervals. Apart from the B3 coefficient which was only significant at levels lower than 1 percent, all the remaining coefficients assumed negative signs and the significance of some of the coefficients was reduced by the presence of multi collinearity On the whole, the Cobb-Douglas model appeared to be superior than the linear model. It was concluded in Chapter Four that if the short-comings of the estimated relationship are recognized, then it is reasonably adequate for analysis or prediction of crop yields. A brief review of the society and its production records were surveyed in Chapter two. For analysis of the trend in output and its postulated determinants, simple indices and ratios were computed. These are also discussed in this chapter.