Copyright 2009 by R. S. Pradhan. All rights reserved.
Welcome to the session on Estimates of Inventory Demand by Nepalese Corporations
Research in Nepalese Finance 5 - 2 Copyright 2009 by R. S. Pradhan. All rights reserved. Estimates of Inventory Demand by Nepalese Corporations Regressions one year lagged sales on inventories produced the following results: Empirical results: ln Y t = 0.33 + 0.68 ln S t-1 (9) (12.62*) R-bar square = 0.64, F = 159.32
ln Y t = 1.85 + 0.65 ln S t-1 0.68 ln i t (10) (11.97*) (2.37*) R-bar square = 0.658, F = 86.67 5 - 3 Copyright 2009 by R. S. Pradhan. All rights reserved. Regressions of current year sales on inventories produced the following results: ln Y t = 0.33 + 0.69 ln S t (11) (12.57*) R-bar square = 0.638, F = 157.98 ln Y t = 1.99 + 0.65 ln S t 0.76 ln i t (12) (12.07*) (2.67*) R-bar square = 0.662, F = 88.07 The partial adjustment model results: ln Y t =0.78+0.08 ln S t-1 0.43 ln i t +0.85 ln Y t-1 (13) (2.62*) (2.27*) (15.23*) R-bar square = 0.889, F = 238.55 xxx 5 - 4 Copyright 2009 by R. S. Pradhan. All rights reserved. Estimates of Inventory Demand by Nepalese Corporations Statement of the problem There is no controversy as to the fact that target inventory level is a function of expected sales as indicated by almost all the earlier studies on the demand for inventories by firms. The controversy arises as to the presence of economies of scale in inventory holdings, the cost of capital and other effects on inventory demand, and the adjustment speed of actual inventory level with target inventory level. 5 - 5 Copyright 2009 by R. S. Pradhan. All rights reserved. Among others, Irvine (1981 May, 1981 September) and Akhtar (1983) clearly indicated economies of scale in inventory holding. However, some of the inventory demand functions of Lieberman (1980) showed diseconomies of scale. Much of the controversies, however, exist with the cost of capital effect on inventory demand. Theoretically, the level of inventories of a firm would depend on the costs associated with holding inventories. However, the earlier studies on the demand for inventories did not present unanimous findings. 5 - 6 Copyright 2009 by R. S. Pradhan. All rights reserved. Liu (1963, 1969), Kuznets (1964), Lieberman, Irvin and Akhtar reported statistically significant effect of capital costs on the demand for inventories. However, Robinson (1959), Lovell (1961, 1964), Burrows (1971), Joyce (1973), and Maccini and Rossana (1981) did not report the same. Controversy also exists with respect to coefficients of adjustment. Among others, Burrows, Maccini and Rossana and Irvine observed faster adjustment between actual inventories and target inventories. While Lovell and Grossman (1973) observed slow speed of adjustment. 5 - 7 Copyright 2009 by R. S. Pradhan. All rights reserved. Thus, there is no unanimous finding with respect to: - economies of scale in inventory holdings, - capital costs effect on inventory demand, & - adjustment coefficient of actual inventories to desired inventories. It has therefore become difficult to support one view or another as there exists no such studies in the context of Nepal. 5 - 8 Copyright 2009 by R. S. Pradhan. All rights reserved. I. The Model The decision about the aggregate level of inventories to be held may be regarded as subject to the constraint of wealth and the cost of holding inventories. As a first approximation to the theory, the function may be written as, Y* = f (S, i) ... (1) Where, Y* is real desired level of inventories, S is the real desired wealth defined in terms of sales, and i is the holding cost. In an empirical investigation, expression (1) takes the form. Y* = k S b 1 i b 2 e u (2) Where, the error terms e u is assumed to be independently and normally distributed. 5 - 9 Copyright 2009 by R. S. Pradhan. All rights reserved. Taking the natural logarithm of the expression (2) gives, ln Y* = ln k + b 1 ln S + b 2 ln i + U i (3) It is assumed that the desired level of inventories (Y*) is equal to its actual level (Y). Thus, the equation to be estimated is, ln Y = b 0 + b 1 ln S + b 2 ln i + U i (4) Where, b 0 is constant, b 1 and b 2 are elasticities of Y with respect to sales and holding costs respectively. The above model assumes the following reasonable a priori hypothesis: y/s > 0, & y/i < 0 (5) 5 - 10 Copyright 2009 by R. S. Pradhan. All rights reserved. While estimating the above equations, inventories and sales have been deflated by using a deflator. The empirical analysis also takes into account a partial adjustment or flexible accelerator model of inventory behaviour. This model hypothesizes that each corporation has a desired target level of inventories, and that each corporation, finding its actual level of inventories not equal to its desired level, attempts only a partial adjustment towards the desired level within any one period. The partial adjustment model is used to indicate the speed with which corporations adjust their actual level to desired level of inventories. 5 - 11 Copyright 2009 by R. S. Pradhan. All rights reserved. The partial adjustment model is stated as, ln Y t =c 0 +c 1 ln S t +c 2 ln i t +(1 ) ln Y t-1 +U t (7) Where, =rate of adjustment coefficient, c 1 & c 2
are the short run elasticities of inventories with respect to sales & interest costs respectively. Long run elasticities with respect to sales & interest costs are b 1 and b 2 . Since c 1 = b 1 and c 2 = b 2 , b 1 = c 1 / & b 2 = c 2 / (8) Short-term interest rate of commercial banks has been used as a proxy for holding cost of funds invested in inventories. 5 - 12 Copyright 2009 by R. S. Pradhan. All rights reserved. Empirical Results The regression of inventories on sales produced the following result: ln Y t = 0.33 + 0.68 ln S t-1 (9) (12.62*) R-bar square = 0.64, F = 159.32 The elasticity of sales with respect to inventories is less than unity showing evidence of economies of scale It thus supports the findings of Akhtar and Irvine and contradicts unitary or more than unitary sales elasticities noticed in some of the equations of Lieberman. Equation (9) ignores the opportunity cost of funds invested in inventories. 5 - 13 Copyright 2009 by R. S. Pradhan. All rights reserved. The regression of inventories on sales and capital cost produced the following: ln Y t = 1.85 + 0.65 ln S t-1 0.68 ln i t (10) (11.97*) (2.37*) R-bar square = 0.658, F = 86.67 Interest rate coefficient is statistically significant with a theoretically correct sign. Previously, this coefficient used to be either positive or not significant in many of the studies on the demand for inventories by firms. The highly significant coefficient of sales is consistently less than unity in equation (10). It shows evidence suggesting economies of scale in inventory holding. 5 - 14 Copyright 2009 by R. S. Pradhan. All rights reserved. This finding supports the conclusions of Liu, Kuznets, Lieberman, Irvine and Akhtar. And contradicts with the results of Robinson, Lovell, Jon Joyce, Burrows, and Maccini and Rossana. Holding the sales constant, it indicates that a one percentage point increase in capital cost leads on an average to about a 0.68 percent decline in inventory investment. Similarly, holding the capital cost constant, a one percentage point increase in sales leads on an average to about 0.65 percent increase in inventory investment. 5 - 15 Copyright 2009 by R. S. Pradhan. All rights reserved. So far the estimated results are based on previous year sales (S t-1 ). These results may not be directly comparable to results of some of the earlier studies which used current year sales (S t ) as an explanatory variable so long as the results are similar. Hence, it is felt necessary to use S t instead of S t-1 . The regressions of Y t on S t, and on S t and i t
produced the following results: ln Y t = 0.33 + 0.69 ln S t (11) (12.57*) R-bar square = 0.638, F = 157.98 ln Y t = 1.99 + 0.65 ln S t 0.76 ln i t (12) (12.07*) (2.67*) R-bar square = 0.662, F = 88.07 5 - 16 Copyright 2009 by R. S. Pradhan. All rights reserved. The use of S t-1 or S t produced similar results. After assessing the scale effect and cost of capital effect on inventory demand, we now estimate the partial adjustment model of inventory demand. The pooled estimate of the partial adjustment model is as follows: ln Y t = 0.78 + 0.08 ln S t-1 0.43 ln i t + 0.85 ln Y t-1 (13) (2.62*) (2.27*) (15.23*) R-bar square = 0.889, F = 238.55 In equation (13), the coefficient of the lagged dependent variable has been observed to be 0.85. Since the coefficient of lag ln Y t is equal to 1 minus the adjustment coefficient (1-), the adjustment coefficient is equal to 0.15. 5 - 17 Copyright 2009 by R. S. Pradhan. All rights reserved. The speed of adjustment between desired and actual inventories as implied by this value is much slower, about 15 percent. The result therefore contradicts the high speed of adjustment observed by Burrows, Maccini and Rossana, and Irvine. In the partial adjustment model, the estimated coefficients of the independent variables are equal to the elasticities of these variable times the adjustment coefficient, e.g., c 1 = b 1 & c 2 = b 2 , b 1 =c 1 / & b 2 =c 2 /. These coefficients are thus .08/.15=0.53 for sales & .43/.15=2.87 for interest rate which again support the conclusions drawn earlier. The capacity utilization as a significant variable affecting the demand for inventories is doubtful. THANKING YOU