Int. J. Production Economics 71 (2001) 21}29
Food inventory policies under liberalized trade Shikha Jha*, P.V. Srinivasan Indira Gandhi Institute of Development Research (IGIDR), General A.K. Vaidya Marg, Goregaon (East), Bombay 400 065, India
Abstract This paper evaluates the role of food inventories or bu!er stocks in stabilizing prices of foodgrains in India when private external trade in grain is liberalized. The model adopts a multi-market equilibrium framework with the direction of trade determined endogenously. World prices are assumed to be sensitive to the amount traded by India (the &large country' e!ect). The simulation results demonstrate that holding foodgrain inventories is an ine$cient mechanism for stabilizing domestic prices under liberalized trade. Variable levies/ subsidies on trade appear to be more cost e!ective. Freeing of trade lowers domestic price variability even though world prices are more volatile. 2001 Elsevier Science B.V. All rights reserved. Keywords: Foodgrain inventories; Price stabilization; Liberalized trade; Rational expectations; Variable levies
1. Introduction Since World War II, the Indian government has been depending on foodgrain inventories or bu!er stocks to simultaneously protect consumers from high food prices and provide price support to farmers during good crop years. Physical storage of grain, however, has resulted in a "scal strain due to wastage and ine$ciencies in handling and distribution by public agencies. The signing of the GATT agreement and the consequent liberalization of trade has provided an opportunity to use alternative mechanisms of price stabilization such as variable levies on trade. However, fears have been
* Corresponding author. Tel.: 91-22-8400919; fax: 91-228402752. E-mail address:
[email protected] (S. Jha).
expressed that freeing of foodgrain trade could have an adverse impact on domestic price stability, which in turn could increase public costs of price stabilization. A major objective of the present study is therefore to examine the e!ectiveness of foodgrain inventories in stabilizing prices in India, a large country, under a scenario of liberalized external trade. For this, a dynamic stochastic simulation model in a multi-market equilibrium framework is built to focus on the major foodgrains, namely, rice and wheat. First, the implications of liberalizing trade are obtained in the absence of government intervention by comparing with the outcomes under autarky. Next, the cost e!ectiveness of achieving greater price stability through food inventories is compared with the alternative option of imposing variable levies on private foodgrain trade under a liberalized trade scenario. The issue of the impact of international price volatility on domestic price stability is di$cult to resolve
0925-5273/01/$ - see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 5 2 7 3 ( 0 0 ) 0 0 1 0 4 - 3
22
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
analytically. Such issues can be resolved at best empirically through stochastic simulation exercises. Studies such as Bigman and Reutlinger [3], for example, suggest based on simulation results that international trade can provide domestic price stability at a cost lower than that incurred in maintaining foodgrain inventories. India is neither a regular importer nor exporter of rice and wheat since its production levels are usually su$cient to meet domestic demand. Given the domestic demand, #uctuations in domestic production, however, imply that there may be a need for imports during a supply shortfall and exports when output is in surplus. Thus, the direction of trade is determined endogenously in the model. Given the small size of the world rice market, the amount traded by India can form a considerable proportion of total world trade so as to in#uence world prices and the model captures this &large country' e!ect. It also incorporates private storage behavior with rational price expectations. The rest of the paper is organized as follows. The next section describes the model and the computational steps. Section 3 discusses how domestic price stability would be a!ected due to liberalized trade. The cost e!ectiveness of bu!er stocks in stabilizing prices under liberalized trade is examined in Section 4. Finally, Section 5 concludes.
2. The model A multi-commodity dynamic stochastic simulation model is used for the purpose of this paper where prices, consumption, production, trade and inventories of both rice and wheat are determined simultaneously. The important features of the model are provided below. A complete description of the speci"cation of various funct-
Greenaway et al. [1] and Williams and Wright [2, Chapter 9] demonstrate that even stylized single-commodity models yield ambiguous results. Outcomes depend on, among other things, the slopes of the excess demand and supply schedules, the relative size of production in the countries concerned and the correlation between production shocks.
ions and the parameter values used is presented in Table 1. Consumption: Aggregate demand equations are calibrated using the elasticities estimated by Radhakrishna and Ravi [4]. Demand for each grain depends on its own price, price of the other grain and income, all relating to the current period. Production: Aggregate supply response for each cereal is assumed to depend on the cereals' own expected future market price with producers having rational price expectations. For estimation purposes these expectations are approximated by 1-year ahead forecasts obtained from an ARIMA model "tted to the price series. Private storage: Private storage agents are assumed to be risk neutral and have rational price expectations. Private storage satis"es the arbitrage equations obtained from expected pro"t maximization. The optimal solution for storage is obtained using a numerical technique based on the recursive logic of stochastic dynamic programming. The expected future price appearing in the arbitrage equations for private storage is derived using the polynomial approximation procedure described in Williams and Wright [2, Chapter 3]. This technique ensures that price expectations of storage agents are internally consistent. Government behavior: The government operates a price band stabilization policy by maintaining prices within a speci"ed price band through either bu!er stocks or variable levies/subsidies on trade. Limited capacity for physical storage of grain puts an upper bound on government stocks. Also, the closing stocks cannot be negative. Thus, it may not be possible to keep prices within the price band whenever these storage constraints become binding. In the case of variable levies/subsidies, however, there are no constraints on the tax/subsidy levels and hence prices are always maintained within the speci"ed price band. International trade: Since India's share in world rice exports is becoming signi"cant, we make the large open economy assumption to model private external trade in foodgrains. That is, excess demand/supply functions for the rest of the world are assumed to be price elastic. Export price falls as exports increase and similarly
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
23
Table 1 Description of the model: speci"cation and parameters Domestic demand d"a#bp#cp (income is assumed to be "xed) Domestic output Planned output: yL "a#b[E (p )] R R> Realized output: y"yL #
The own price elasticities of demand for rice and wheat are, respectively, !0.51 and !0.58 and the cross price elasticities are, respectively, 0.072 and 0.19. Price elasticities of supply (planned output) are 0.16 for rice and 0.09 for wheat. is the deviation from planned output generated randomly.
Government storage rule Deplete stocks (sg\) if p*'p Add to stocks (sg>) if p*'p
The magnitudes are chosen so that p* is within the price band. p* is the market equilibrium price, p and p are ceiling and #oor prices of the price band. Capacity constraint on total storage is 30 Million tonnes.
Private storage Satis"es the arbitrage equations obtained from expected pro"t maximization. (1) p #k((1#r)\ E (p ), sp "0 R R R> R (2) p #k"(1#r)\ E (p ), sp '0 R R R> R Import trigger price p( K"p@(1#import margins) G G Export trigger price p( V"p@(1!export margins) G G Exports (x) If p**p( V then x"0, otherwise x"(p*! )/ , where the inverse export demand function for the rest of the world is taken as pV" # x Imports (m) If p*)p( then m"0, otherwise m"(p*! )/ , where the inverse import demand function for India is taken as pK" # m Variable levies Import subsidy is positive when m'0 and p* tends to go above p . Import tax is positive when m'0 and p* tends to go below p . Export subsidy is positive when x'0 and p* tends to go below p . Export tax is positive when x'0 and p* tends to go above p .
p is the current price, k the marginal cost of storage services (Rs. 52 per quintal), E (p ) the expected future price, r the interest R R> rate (0.1) and &sp' the amount of private storage.
Commodity balance d #sp #(sg>!sg\)#(x !m )!(y #sp )"0 R R R R R R R R\
p* is computed such that commodity balance condition holds.
p@ is the border price generated randomly. Import margins are 15.0 and 36.7% of border price, respectively, for rice and wheat. Export margins are 5.0 and 9.0% of border price, respectively, for rice and wheat. The elasticity of price with respect to exports is !0.14 for rice and !0.001 for wheat.
The elasticity of price with respect to imports is !0.14 for rice and !0.001 for wheat.
The magnitudes are chosen so that p* is within the price band.
See Jha and Srinivasan [6] for further details on the model.
import price rises as imports increase. Given the trade elasticities, commodities are shifted o! or on the world market whenever the marginal revenue adjusted for transportation and transaction costs, falls short of or exceeds the current value of grain.
When private external trade is permitted, equilibrium price will depend on world market prices in addition to domestic market conditions. The equilibrium price is determined with the help of an equilibrium price function, which depends on domestic output as well as the export and import
24
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
trigger prices. Exports would take place whenever the domestic market price falls below the &export trigger price'. Similarly, imports take place if the market price rises beyond the &import trigger price'. If equilibrium price lies between the export and import trigger prices, it implies that there is no trade. Computation of equilibrium outcomes: Equilibrium outcomes are computed for a sequence of 1000 periods with di!erent possible random realizations of domestic yields and world market prices. Equilibrium prices and quantities for each period are obtained by matching current availability with domestic consumption and storage demand (public and private) for each grain. Availability of grain in any period is de"ned as the realized production plus net imports in that period. The realized production is obtained by adding a random deviation to the expected output, generated from the estimated probability distribution. Given the initial values of availability of grains and the realizations of world prices, the equilibrium prices, additions to stocks by private and government agencies, levels of imports/exports etc. are computed simultaneously. This exercise is repeated sequentially for 1000 periods. The means and variances of di!erent variables including consumption, production, price, producer and consumer surplus etc are then derived to obtain the welfare implications of di!erent alternatives. There are two main steps involved in solving the model. First, a reduced form relationship between expected future price and current storage carryout is obtained using a polynomial approximation procedure. The iterative procedure used for this purpose involves solving, for the entire model, for all possible states of nature for each round of iterations. Second, the model is simulated for the desired number of times. The polynomial obtained in the "rst step is used in solving for private storage. Equilibrium values for each period (simulation) are Export trigger price is de"ned as that domestic price at which the trader is indi!erent between selling a marginal unit in the domestic or world market. If the domestic price is lower than this level the trader will sell it in the world market and viceversa. Similarly, import trigger price is de"ned as that price at which the trader is indi!erent between buying from domestic and international markets. If the domestic price is higher the trader prefers to import.
obtained sequentially. Starting with some initial values for expected price and carry-in of public and private stocks and the realizations of shocks to domestic output and world prices from a random number generator, all the endogenous variables of the model are solved for simultaneously. The expected future price and private and public stock carryout obtained from this solution replace the initial starting values for expected price and carryin of public and private stocks, respectively, for the subsequent period. This process is repeated for the desired number of periods. 3. E4ect of trade liberalization on domestic price stability Before comparing the e!ectiveness of levies on trade as an alternative to food inventories, it is important to analyze how liberalization of trade in itself will a!ect domestic price stability. Since world prices are much more volatile than domestic prices, there is a potential risk of exposing domestic producers and consumers to greater price instability through free trade. Under stable world prices, if the distance between the export and import trigger prices is large enough such that equilibrium price is always between these two prices, then there can be no e!ect on domestic price variability due to free trade. But as the two trigger prices are brought closer, say through a reduction in export/import margins (port charges, etc.) or the domestic trade and transport margins, there would be a monotonic decrease in price variability. Maximum reduction in price variability will be achieved when all the relevant margins which provide a wedge between domestic and world prices are set to zero, that is, when both pe and pm are equal to the border price. In the case of unstable world prices, however, price variability can rise or fall. When world prices and domestic output #uctuate independently, the results from simulations show that price variability decreases with freeing of trade (Table 2). With Historical data show a positive but low and statistically insigni"cant correlation between domestic output and world price #uctuations. Since the correlation is close to zero, in our simulations the random outcomes of domestic output and world prices are generated independently.
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
25
Table 2 Impact of free trade (at historically observed correlation between domestic output and world price #uctuations) Autarky
Free trade Existing trade margins
Market price (Rs/quintal)
50% lower margins
Zero margins
rice
897.89 (0.1670)
931.20 (0.1517)
926.23 (0.1631)
915.15 (0.1762)
wheat
510.72 (0.1393)
505.46 (0.1207)
485.56 (0.1897)
453.96 (0.2625)
0.1532
0.1362
0.1764
0.2194
Average CV of rice and wheat prices Import price (Rs/quintal)
rice wheat
* *
1248.83 620.74
1171.51 537.85
1100.04 455.26
Export price (Rs/quintal)
rice wheat
* *
865.65 413.00
881.55 433.42
894.70 452.68
Dom cons (million tonnes)
rice
67.87 (0.0771)
66.55 (0.0748)
66.52 (0.0773)
66.62 (0.0814)
wheat
51.97 (0.0720)
52.68 (0.0589)
54.03 (0.0961)
56.10 (0.1265)
rice
67.87 (0.0896)
69.72 (0.0879)
69.34 (0.0882)
68.92 (0.0886)
wheat
51.97 (0.0783)
51.83 (0.0785)
51.50 (0.0790)
50.99 (0.0797)
Prodn (million tonnes)
Pvt stks (million tonnes)
rice wheat
0.73 0.23
1.51 0.11
1.75 0.21
2.21 12.82
Imports (million tonnes)
rice wheat
0.00 0.00
0.56 1.65
1.11 4.04
2.07 13.08
Exports (million tonnes)
rice wheat
0.00 0.00
3.71 1.26
3.82 1.75
4.13 7.96
Producer surplus (Rs. crores)
217758.20
220998.20
219457.36
216842.22
Consumer surplus (Rs. crores)
79401.46
77569.98
78706.54
80720.95
297159.66
298568.18
298163.90
297563.17
Total surplus (Rs. crores)
All the numbers in the table are average values derived from model runs for 1000 periods. The "gures in parentheses give the CV (coe$cient of variation) of the variables. Since there is no price stabilization policy by the government, there are no government stocks listed in this table.
a perfect positive correlation between world prices and domestic production (which is equivalent to a perfect negative correlation between domestic and world prices), there is a big reduction in price variability (Table 3). Conversely, with a perfectly negative correlation, there is an increase in price variability although the magnitude of increase is small (Table 4).
However, the results are sensitive to the assumed level of import/export margins and domestic trade and transport margins. Under free trade, as these margins are reduced from the existing levels there is a monotonic increase in price variability (Table 2). Under autarky (which can be considered to be a case of very large margins between the border price and the export/import trigger price) price
26
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
Table 3 Impact of free trade (perfect positive correlation between domestic output and world price #uctuations) Autarky
Free trade Existing trade margins
Market price (Rs/quintal)
50% lower margins
Zero margins
rice
897.89 (0.1670)
946.30 (0.0135)
944.14 (0.0226)
939.47 (0.0370)
wheat
510.72 (0.1393)
507.70 (0.0479)
499.21 (0.1078)
452.64 (0.1701)
0.1532
0.0307
0.0652
0.1036
Average CV of rice and wheat prices Import price (Rs/quintal)
rice wheat
* *
1332.69 618.59
1252.98 535.96
1173.46 453.42
Export price (Rs/quintal)
rice wheat
* *
891.96 411.50
901.98 431.82
909.30 451.58
Dom cons (million tonnes)
rice
67.87 (0.0771)
65.97 (0.0081)
65.95 (0.0100)
65.68 (0.0166)
wheat
51.97 (0.0720)
52.68 (0.0321)
53.28 (0.0677)
56.48 (0.0912)
rice
67.87 (0.0896)
70.27 (0.0858)
70.27 (0.0858)
70.07 (0.0860)
wheat
51.97 (0.0783)
51.84 (0.0784)
51.68 (0.0786)
50.82 (0.0799)
Prodn (million tonnes)
Pvt stks (million tonnes)
rice wheat
0.73 0.23
0.00 0.00
0.00 0.32
0.00 5.32
Imports (million tonnes)
rice wheat
0.00 0.00
1.16 2.22
1.35 3.37
1.57 11.08
Exports (million tonnes)
rice wheat
0.00 0.00
5.42 1.39
5.66 1.75
5.95 5.40
Producer surplus (Rs. crores)
217758.20
223064.14
222618.64
219692.03
Consumer surplus (Rs. crores)
79401.46
76261.95
76758.38
78855.39
297159.66
299326.09
299377.02
298547.42
Total surplus (Rs. crores)
All the numbers in the table are average values derived from model runs for 1000 periods. The "gures in parentheses give the CV (coe$cient of variation) of the variables. Since there is no price stabilization policy by the government, there are no government stocks listed in this table and government costs are zero.
variability is higher than that under free trade at existing margins. Thus, over a certain range of margins beyond the existing levels, price variability rises with margins reaching the autarky level in the limit. That is, there is a U-shaped relationship between price variability on the one hand and the margins between domestic and border prices on the other.
The import/export margins and domestic trade and transport margins form a wedge between domestic and world prices under a free trade regime. The broader this wedge the lesser is the frequency of trade for given distributions of domestic output and world prices. If these margins can be reduced through better port and infrastructure fa-
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
27
Table 4 Impact of free trade (perfect negative correlation between domestic output and world price #uctuations) Autarky
Free trade Existing trade margins
Market price (Rs/quintal)
50% lower margins
Zero margins
rice
897.89 (0.1670)
945.17 (0.1732)
947.92 (0.1721)
940.83 (0.1772)
wheat
510.72 (0.1393)
514.13 (0.1365)
507.54 (0.1551)
457.13 (0.1718)
0.1532
0.1549
0.1636
0.1745
Average CV of rice and wheat prices Import price (Rs/quintal)
rice wheat
* *
Export price (Rs/quintal)
rice wheat rice
* * 67.87 (0.0771)
927.74 415.69 66.08 (0.0872)
935.47 436.30 65.91 (0.0874)
938.75 456.38 65.66 (0.0920)
wheat
51.97 (0.0720)
52.24 (0.0693)
52.72 (0.0835)
56.17 (0.0838)
rice
67.87 (0.0896)
70.00 (0.0890)
70.13 (0.0891)
69.73 (0.0894)
wheat
51.97 (0.0783)
52.24 (0.0780)
51.92 (0.0785)
50.82 (0.0800)
Dom cons (million tonnes)
Prodn (million tonnes)
1250.36 624.45
1168.82 540.89
1089.37 457.57
Pvt stks (million tonnes)
rice wheat
0.73 0.23
2.30 0.38
2.48 0.61
3.04 4.88
Imports (million tonnes)
rice wheat
0.00 0.00
0.00 0.00
0.00 0.86
0.72 8.57
Exports (million tonnes)
rice wheat
0.00 0.00
3.94 0.00
4.21 0.00
4.86 3.20
Producer surplus (Rs. crores)
217758.20
222077.58
221838.97
218169.22
Consumer surplus (Rs. crores)
79401.46
76603.92
76718.04
79031.49
297159.66
298681.50
298557.01
297200.71
Total surplus (Rs. crores)
All the numbers in the table are average values derived from model runs for 1000 periods. The "gures in parentheses give the CV (coe$cient of variation) of the variables. Since there is no price stabilization policy by the government, there are no government stocks listed in this table and government costs are zero.
cilities it will imply a smaller gap between the export and import trigger prices, which in turn will increase the possibility for greater volumes of trade. Therefore there is a greater exposure of the domestic economy to world prices and their instability as observed from the simulation results. This, however, does not mean that India should not invest in ports or other infrastructure.
Table 2 shows that as margins are reduced consumer surplus is increased. Better trade and transport infrastructure increases the price received by exporters net of margins and reduces the domestic price of imported grains. The net e!ect of this is a fall in the equilibrium market price. Thus, although producers lose due to reduction in margins, the consumers bene"t.
28
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
Fig. 1. Trade-o! between stabilization cost and price variability.
4. Cost e4ectiveness of food inventories under liberalized trade The government is assumed to stabilize interyear price #uctuations through a price band policy. The price band consists of a #oor price and a ceiling price. If the price goes beyond the ceiling price, then the government sells grain in the market by depleting its stocks until the price is driven down to the ceiling level. Similarly, it prevents the price from falling below a #oor level by buying grain from the market and adding to its stocks. Variable levies/subsidies is an alternative mechanism for price stabilization, under liberalized trade. Di!erent levels of price stability can be achieved by varying the price bands. Although the distribution of prices in the rest of the world is taken as exogenous the price realizations in the world are endogenous to the model.
The realized prices are in#uenced by the magnitude of domestic trade since excess demand/supply functions of the rest of the world are elastic with respect to price. The policy response from the rest of the world to domestic country's price stabilization policies is not taken into account. Bigman [5], for example, notes that if each country insulates its own economy through price stabilization policies then it can lead to lower world trade and less price stability for all the countries.
The desirability of either option depends on how well the prices are stabilized, what are the implications for social welfare and how high are the costs to the government. In a price band model it is not feasible to "x either the price variability or government costs across scenarios. We therefore obtain curves depicting trade-o! between price variability and public price stabilization costs (Fig. 1). As compared to the case of autarky, we "nd that under free trade bu!er stocks are not very e!ective in price stabilization. For similar amounts of government expenditure the reduction in price variability is far less in the latter case. The trade-o! curve is #at implying that even by incurring high costs, price variability cannot be reduced much. Variable levies/subsidies on private external trade turn out to be a better option. The relative ine!ectiveness of food inventories under free trade compared to the autarky scenario is mainly due to the fact that storage constraints are binding more often in the former case. Since the average price of rice is at a higher level under free trade, public rice stocks are exhausted more often in this case. Similarly, the lower wheat price in the free trade scenario contributes to greater frequency of hitting the aggregate storage capacity constraint. Variable levies/subsidies perform better because there are no constraints speci"ed for the values
S. Jha, P.V. Srinivasan / Int. J. Production Economics 71 (2001) 21}29
29
these levies can take. The mean values for export subsidies and import taxes are well within limits and consistent with the GATT commitments.
These results may, however, be sensitive to the values assumed for trade and transport margins between domestic and border prices.
5. Concluding remarks
Acknowledgements
Traditionally, food inventories have been used in India for stabilizing domestic grain prices. However, the cost e!ectiveness of this method in achieving its objective has come into question in recent times. Liberalization of external trade in foodgrains has given rise to alternative mechanisms for achieving food price stability. In this paper, we examine the comparative cost e!ectiveness of food inventories vis-a`-vis variable levies/subsidies on trade using a dynamic stochastic simulation model. The main "ndings can be summarized as follows.
This paper draws on a report submitted to the World Bank. We thank Deepak Ahluwalia, David Bigman, Benom( t Blarel, Bruce Gardner and Dina Umali-Deininger for many useful comments and discussions. The paper has also bene"ted from the comments and suggestions of two anonymous referees of this Journal and the participants, in particular M.A. Ghali, at the 10th International Symposium on Inventories, held at Budapest, August 23}28, 1998. The responsibility for any errors, however, lies fully with us. The views expressed in this paper are ours and should not be attributed to the World Bank or IGIDR.
1. Higher instability of international prices does not necessarily increase domestic price variability or the government's costs of price stabilization. 2. At current levels of margins between domestic and border prices, freeing of trade leads to a reduction in domestic price variability. 3. The above result is more likely to hold when these two variables are positively correlated. Historical evidence indeed suggests a positive, although weak, correlation between these two variables. 4. Government's food inventories are less e!ective in stabilizing prices under liberalized trade as compared to the case of autarky. 5. Variable levies/subsidies on trade appear to be more cost e!ective compared to food inventories. One reason for this is that while the latter alternative works under constraints on physical storage of grains, the former is free from such constraints.
References [1] D. Greenaway, C.W. Morgan, A.J. Rayner, G.V. Reed, Trade liberalization and domestic price instability in an agricultural commodity market, Applied Economics 25 (1993) 199}205. [2] J.C. Williams, B.D. Wright, Storage and Commodity Markets, Cambridge University Press, Cambridge, 1991. [3] D. Bigman, S. Reutlinger, Food price and supply stabilization: National bu!er stocks and trade policies, American Journal of Agricultural Economics 61 (1979) 657}667. [4] R. Radhakrishna, C. Ravi, Food demand in India: Emerging trends and perspectives, Mimeo., 1994. [5] D. Bigman, Stabilization and welfare with trade, variable levies and internal price policies, European Review of Agricultural Economics 7 (1980) 185}202. [6] S. Jha, P.V. Srinivasan, Grain price stabilization in India: Evaluation of policy alternatives, Agricultural Economics 21 (1999) 93}108.
