3.1 Forecasting Method by the Use of Quadratic Trend Analysis The growing competition, rapid change in circumstances and the trend towards automation demand that decisions are not based purely on guesses and hunches rather on careful analysis of data concerning the future course of events. When estimates of future conditions are made on a systematic basis, the process is referred to as “forecasting” and the figure or statement obtained is known as a “forecast”. Forecasts are statements of expected future conditions, definitive statements of what will actually happen are potentially impossible. Expectations depend upon the assumptions made. If the assumptions are plausible, the forecast has a better chance of being useful. Forecasting aims at reducing the areas of uncertainty that surround management decision-making with respect to demand, supply, production, pricing and so on. Forecasts are made in order to assist management to determine a strategy and alternative strategies. Forecasting will not only help in the short-term control of operations, its greatest contribution probably will come when it is able to improve short and long-term corporate strategies.
3.2 Quadratic Trend Analysis In the past 20 years, vagaries of the weather have caused quite a number of shortfalls in particular gram pulses production which is one of the key pulses crops in Bangladesh. These shortfalls have had a pronounced negative impact on the balance between food supply and demand. Data shows that advanced technology continues to make a positive contribution to yield growth in the country. However, in some areas, this growth has recently slowed, while other areas have maintained a high rate of yield growth. These different rates arise primarily from the impact of the interaction between climate and technology on yield growth.
Modelling Approach: A second degree polynomial regression model
LogeY = A0 +A1T + A2T 2 + e........... (1)
Where y is the yield, t is the time trend variable and A0 , A1, A2 are the regression coefficients and e is the residual term. The coefficients A0, A1 and A2 represent the initial level in time, the rate of increase and changes in this rate over time, respectively. The quadratic coefficient (A2) in particular shows the curvilinear in trend time required to estimate the climate limitations to increase. Thus, this coefficient is suggested as an index describing climate-technology interaction. The value and sign of this coefficient are a reflection of the degree and direction of the climate-technology interaction change. The larger this value, the greater is the input provided by this interaction. A positive coefficient is an indicator of compatibility between climate and technology or, in other words, shows that climate resources meet the requirements of the technology so as to maintain an increasing rate of growth. Conversely, a negative coefficient indicates that climate imposes certain constraints on the technology applied and contributes to reducing levels or at least limits the rate of increase.
When extreme weather conditions hit an area, causing a considerable shift considerably from the established long-range technologically induced trend observed several years in succession the standard regression technique cannot provide an adequate formalization of the trend, especially at the end of the time series. To overcome this problem a special statistical procedure has been developed. This procedure utilizes analysis of fluctuation of quadratic coefficients from the time series (Kogan, 1985). Gradual changes in dynamics of the quadratic coefficients over time indicate an absence of periods with abnormal weather. Conversely, a considerable shift from this dynamic condition, especially if it resulted in changes not only of the magnitude but also in the sign of the coefficient indicates a presence of abnormal weather impact. In the second case, the trend should be estimated without the period with abnormal weather; a special procedure was developed to handle this problem (Kogan, 1985).
3.4 Sources of Data Estimation of the model required the historical crop supply data. This study was an attempt to determine the nature and extent of the relationship of these variables. The nature of data obtained from the BBS and DAE constrained estimation of the model in both spatial and temporal dimensions. Thus for each district/region, the data of 21 years (1980/1981-1999/2000) had been considered in this study.