Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Part 1: Introduction

Price policy and the food that people consume

The stylized facts and a model of them
How the model works
Some numerical examples

Lance Taylor

The market for staple foods in a poor country accounts for a substantial fraction of economic activity, is politically important, and, if left to itself, is highly unstable. For these reasons, governments all around the world intervene in food pricing, for better or for worse (1). In recent years, food subsidies in particular have been used as redistributive tools, and to protect consumers from the recent exceptional rise in import prices. In other times and circumstances, farmers have been shielded from the consequences of excess supply by price supports. As a share of total fiscal activity, the costs or receipts to the treasury of these price interventions can be quite substantial.

Most analysis of food tax and subsidy interventions has focused on their micros effects. One example is asking how food supply from farmers might respond to higher prices, without tracing feedbacks to the rest of the economy from the larger farm income the increased prices would bring. On the demand side, food consumption increases in response to a subsidy might be projected without considering the added sales of non-food products that would follow from the extra spending power that lowered food prices would create. Precisely because food production, processing, and distribution sectors are such large fractions of GNP in poor countries, these economy-wide impacts of food price policies should be explored. The purpose of this paper is to point out some of the macromechanisms through which food price policy acts, and their implications for food consumption and ultimately nutritional status.

The discussion will centre upon a formal model of a hypothetical economy- an algebraic presentation appears in sections 1 and 2, and in section 3 numerical examples illustrating the impact of different policies are worked out. Similar models are used for policy analysis in developing countries (2) but for present purposes it is simplest to omit the local detail that inevitably complicates a country application and concentrate on the essentials. We begin with a recitation of the "stylized facts" about food price policy in our model economy.

The stylized facts and a model of them

Like many underdeveloped countries, the economy under consideration has the following characteristics.

A. Staple food consumption is a large proportion of the total economy-as much as a third or more. Food production and processing generate a similar share of total income. The market for food clears rapidly via a changing price, and neither demand nor supply responds much to price movements in the short to medium-term. Food demand is also income inelastic, which means that changes in real income induced by changes in food prices spill over into demand shifts for other sectors.

B. The non-food sectors have fairly rigid prices, determined by an oligopolistic mark up over variable costs of labour and other inputs. There is excess capacity in non food products, so supply and employment fluctuate up and down in response to aggregate demand. Part of the demand comes from sales of fertilizer and other manufactured products to farmers. C. The government intervenes extensively in food-related markets. It subsidizes the use of fertilizer by farmers, and the price that consumers pay for food after it has been processed by workers off the farm. Food imports also enter the system in quantities regulated by quota or a government monopoly trading company.

D. There are class differences among agriculturalists, workers, and recipients of mark-up income in the proportions of income they save (3). Investment, exports, and other government expenditures are, in standard Keynesian fashion, assumed to be exogenous and fixed in real terms in the short run.

The formal model equations appear in Table 1. The variables XF and XNF, respectively, represent output of agricultural products for use as food, and inputs such as fertilizer from non-food (subscript N) sectors into food production (subscript F). Equation 1 shows that food output responds to non-food inputs with an elasticity b (which is given the empirically plausible value of 0.16 in the numerical examples below). Let PF and PN be the prices that producers of food and non food products receive, and s the government subsidy rate on non-food inputs into food production. If, for the sake of simplicity, it is assumed that farmers maximize their profits PFXF-PN (1-s) XNF, their demand function for XNF is given in equation 2-the demand elasticity is 1/1 (1-b), which will be a bit greater than one. Finally, food imports MF add to domestic supply XF in meeting consumption demands CF in equation 3.

In the non-food sector, variable cost-per-unit output is just waLN, where w is the wage rate and aLN the labour/output ratio in the sector. The output price PN is given in equation 4, where for simplicity, z-the rate of producers' mark-tips over costs-is assumed constant. Nonfood output XN goes to satisfy agriculture's input demand XNF, and consumer and autonomous demands CN and AN in equation 5. The autonomous demand comprises investment, exports, and government purchases, and is assumed fixed in the short-run

Equations 6 - 8 define income flows. Farmers' income YF is the value of their sales less fertilizer purchases at the subsidized price PN (1-s). Workers receive income Yw from non-food production waLNXN and also from processing of agricultural output and imports into consumable form. If total food consumption is CF, and labour input coefficient into processing is aLF, then the income generated is waLFCF. Equation 8 gives the income flow Yz from mark-ups.

In equation 9 total consumer spending D is determined by consumption coefficients g F, g w, and g z from the three income flows. Farmers and workers have fairly high consumption shares (assumed in the numerical examples to be 0.85 and 0.95, respectively), while the share g z of capitalists will be lower (it is given a value of 0.301). Equation 10 shows that the consumer price of food is the agricultural price plus processing costs (PF + waLF), multiplied by a subsidy term 1-t. When the subsidy rate t is greater than zero, consumers will be acquiring processed food at less than its cost.

Equations 11 - 13 apply an empirically plausible set of consumer demand equations known as the "linear expenditure system," or LES, to the two-sector model at hand. The LES is built around the assumption that for each product there is some base level of consumption (q F and q N, for food and non-food, respectively), which is insensitive to price and income changes. The cost E of acquiring these base consumption levels is defined in equation 11. The increment in expenditure above the base level D-E is split between the two commodities according to "marginal budget shares" mF and mN. These sum to one, because expenditure above the floor level E must be exhausted by purchases of one good or the other. The relevant equations are 12 and 13. The fact that food demand is relatively insensitive to both income and price means that its marginal budget share mF will be small, while the fixed consumption level q F will be large in comparison to current purchases. The opposite observations apply to non-food (4).

How the model works

Policy changes can be traced through this equation system in at least two ways. The first is to reduce the discussion to only the market for food. On the supply side, if the money wage w and mark up rate z are fixed in the short run, then the non-food producers' price PN is determined in equation 3. Given the fertilizer subsidy rates, fertilizer demand XNF comes from equation 2 and food output XF in turn from equation 1. Note that the scale of production is determined by the constant K, which stands for land, capital, and other fixed assets used in agriculture. For a given value of K, a supply schedule for XF can be traced out as the price PF varies. Supply rises as the price goes up, but for an empirically plausible value of the output elasticity b, the response will not be strong. The rising curve SS in figure 1 illustrates supply response (with a constant level of food imports MF added to domestic production).

To trace the determination of food demand, the whole circular flow of incomes in the economy must be followed. Begin by noting that agricultural income YF is essentially determined by the prices PF and PN through the supply equations 1 and 2 and the definitional equation 6. Both wage income YW and mark-up Yz depend on non-food output XN, which, in turn, depends on non-food consumption CN from equation 5. This consumption itself is affected by the income flows YF, Yw, and Yz, in standard Keynesian fashion. If we let B denote the total consumer spending generated by a unit of output XN,

B = (g W +g ZZ)waLN

then total spending D can be solved in a multiplier equation:

D=[(1 - B(mN / PN)] -1 {B [ q N + AN - (mN / PN)F]

+ (q FPF + q WWaLF) XF + q WWALFMF (14)

+ [B-q FPN (1-s)]XNF}

The term outside the curly brackets in equation 14 is the multiplier-the inverse of the fraction of income generated by XN not spent on itself. The terms inside the brackets are, respectively, the spending on XN generated by autonomous demand AN and the constants in the consumption equation 13, by food production and imports XF and MF, and by fertilizer sales XNF. In effect, equation 14 shows how, for instance, food production generates income, with the spending by farmers on non-food consumer goods and the additional income that creates taken into account.

With expenditure D determined from equation 14 plus the prices, it is easy to calculate total food consumption CF from equation 12. By varying PF the pattern of food demand can be traced, as shown by the falling curve DD in figure 1. Note that this curve is very steep, so that food price changes do not strongly affect consumption demand. Any shifts in the supply curve with demand unchanged would therefore mostly affect prices and leave the consumption level nearly the same, as can be seen by imagining various intersection points of SS with DD as the former curve moves. This acute responsiveness of food prices to supply shifts is, of course, well known, and is a key reason why governments intervene in markets to try to hold prices stable.

Another response pattern is shown by the shifted demand curve EE, which illustrates an increase in the food subsidy rate t in equation 10. Increasing the subsidy leads to greater real purchasing power for consumers, most of which is not directed to food. However, the resultant increase in demand for non-food products raises worker and mark-up incomes in multiplier fashion from equation 14, and part of the increase does spill over into an upward shift of the food demand curve to EE. But here, the inelasticity of supply leads to a sharp price increase. We have another reason for government wariness in intervening in the market for food (5).

A second way of analyzing equilibrium in this economy is via the responses of saving and investment to changing prices for food. At an equilibrium point (such as the intersection of the SS and DD curves in figure 1), the following accounting equation will hold:

(1 - q F)YF + (1 - q z)Yz + PFMF


On the left of the equal sign is "saving," i.e., the parts of farmers', workers', and capitalists' incomes that are not consumed, plus the resources provided the economy by food imports. On the right are the uses of resources that this saving must finance- autonomous expenditure and the subsidies on fertilizer and food sales.

To see how this equation describes the determination of economic equilibrium, assume that its right side exceeds the left. A sudden increase in autonomous expenditure AN from an established equilibrium would have this effect. The added demand would make markets tighten up and create more employment in industry. The resulting increases in labour and mark-up incomes YW and Yz would add to savings on the left of equation 15, but by less than the increase in AN, as Part of the income increment would be consumed. Some of the consumption demand would be directed to food, and its price would rise. On the left of equation 15, the higher prices would stimulate savings from agricultural incomes (1 - g F)YF. On the right, the total expenditure on the food subsidy tPFCF would rise. In a stable economy, the farmer savings response to higher food prices would ultimately outweigh the increased cost of subsidies, and equilibrium would be reestablished with a new, higher value of PF. Because of the steepness of the supply curve SS illustrated in figure 1, this food price increase could be substantial. And of course, the price rise would redistribute income toward the farm sector and away from the cities. This could cause political problems for the government as well.

Some numerical examples

The foregoing discussion suggests that macro-responses of both prices and income distribution to dislocations of the market for staple foods are likely to be large. Presumably, the same statement would apply to food (and calorie) consumption levels by different segments of the population as well. To emphasize these points, discussion in this section will consider some numerical illustrations of how the model of table 1 might respond to policy and other changes.

The numerical specification selected for the model appears in table 2. Note that food consumption makes up three eighths of the total, while agriculture generates about 20 per cent of total incomes. Another 7 per cent of income (counted in Yw) is generated by food processing. The income elasticity from the linear expenditure system, equation 12, turns out to be -0.32. All these numbers are quite plausible in the aggregate for a low- to middle-income developing economy.

Results from a number of policy experiments with the model appear in table 3 l6). The rows give information about various economic aggregates: GNP in real (base year prices) and current price terms, and changes in real output and price indexes. Next con shares of the three income recipient classes in total income, and their respective levels food consumption. Per cent changes in food consumption are also presented to provide a convenient summary indicator of the impact on both purchasing power and nutritional status by income class of the policies considered. Finally, the shares in GNP of total food and fertilizer subsidy payments appear to illustrate the relative importance of these fiscal interventions.

3.1 Increasing food subsidies. The initial food subsidy is 20 per cent, so that while the cost of consumed food is 1.25 (equals PF + WaLF), its consumer price is only 0.8 x 1.25 = 1.0. Suppose that the subsidy rate is doubled from 20 to 40 per cent to reduce the consumer price initially to 0.75. What will be the effects on the macroeconomy g

The second column shows that the impact would be profound. As already discussed in connection with figure 1, the subsidy would increase spending power directed to both food and non-food products. Because of their asymmetric supply responses, the two sectors would react in completely opposite form. Because it is assumed that there is excess industrial capacity, non-food output would rise in response to the increased demand, generating higher incomes for workers and capitalists in the process. Most of the food demand increase, however, would drive up prices and farmer incomes without calling forth much additional supply. The table shows a solid 25 per cent increase in real GNP, while the farm share of total income goes up 37 per cent, from 0.203 to 0.279 (7). Food consumption for all classes increases, but mostly for the farmers. Instead of merely doubling, the share of the food subsidy in GNP almost triples because of the spiraling prices.

The solution illustrates the points already made in connection with figure 1-that food subsidies can have strong impacts on income distribution, prices, and aggregate demand. Instead of the increase considered here, a subsidy decrease (or a staple food tax) would substantially reduce real spending and food intake levels. For the poorer segment of the population in both urban and rural areas, the consequent reduction in food intake could lead to a visible increase in clinically detectable malnutrition and morbidity, as the nutritional status of the poor is likely to be precarious in any case.

3.2 Increasing fertilizer subsidies. As discussed in connection with equations 1 and 2, raising the fertilizer subsidy rate should lead to a modest increase in overall food supply. Because it reduces agricultural costs, the increased subsidy would also lead to a drop in farmgate food prices, unless the government intervened to store or export some of the additional crop. The third column of table 3 illustrates a case in which such intervention did not occur, and the producer price of food falls from 1.0 to 0.95 (in response to a doubling of the subsidy rate from 0.2 to 0.4). Food production rises from 5.0 to 5.24, or about 5 per cent.

Observe that both workers and mark-up recipients benefit from the supply increase, expanding their food consumption by about 6 per cent each. However, the farmers' share of income drops, and their food intake falls. Whether or not they would accept such an income reduction in political terms is unclear; if they did, it would surely be the rural poor who would bear the brunt of the real food consumption decrease. The solution illustrates quite clearly that policies aimed to increase food production alone are not sufficient to direct more calories to some of the people who need them most. Something has to be done to increase their real purchasing power as well.

3.3 A reduction in agricultural yields. In both rich and poor countries, of course, agricultural productivity is anything but stable. We can capture the effects of bad weather and similar mishaps in the model by adjusting the constant K appearing in equations 1 and 2. The fourth column of table 3 shows what happens when K is reduced from 5.0 to 4.75, leading to a 5 per cent shortfall in output before any price and subsequent production responses are taken into account.

As is often the case when farm production falls, the effect is to drive up food prices (by 15 per cent in this case) and shift the income distribution toward agriculture. Demand for non-food products falls because the food price increase reduces real spending power, leading to a drop in output from 15.0 to 14.8. The decrease in food supply is largely borne by urban workers and capitalists; because of their higher incomes farmers actually eat more. The government could, of course, offset the price inflation and income redistribution by more food imports, but the foreign exchange cost could be high. The options a government faces when crop yields decline are all very unpleasant.

3.4 Increased food imports. If a crop shortfall increases agricultural incomes, then more imports are bound to hurt them. The fifth column of table 3 shows the consequences of an increase in imports MF from 1.0 to 1.25. The farmgate food price falls from 1.0 to 0.85, and the agricultural income share drops by over 12 per cent. Real GNP also falls, because increased imports add to potential savings capacity on the left of equation 15, and with fixed exogenous expenditure the economy contracts in response. Farmers suffer substantial food consumption losses, while the other groups gain. Overall food consumption, in fact, increases slightly (from 6.0 to 6.09), so that imports add to total supply even after their negative impact on production is taken into account. However, the rural poor could be hard hit in the short run by a freer food import regime (8).

3.5 Increased labour productivity in food processing. As is the case with more imports, an increase in labour productivity in the "middleman" sector of food processing should be beneficial, as it potentially adds to aggregate supply. However, employment would drop from the productivity increases, unless the government took steps to create jobs elsewhere. The next to last column in table 3 indicates that the drop in purchasing power resulting from a (unlikely) doubling of productivity would be enough to reduce real GNP by 2.2 per cent, with some increase in the price farmers receive for food. If successful, attempts to reduce costs in food processing could adversely affect worker's income, unless the government intervened. Absorbing technical advances or increased efficiency in the macro-economy is not always an easy task.

3.6 Increased autonomous demand. The last column of the table shows how an increase in autonomous spending from 4.0 to 5.0 would drive up aggregate demand. Real GNP goes up 18 per cent in response to the 25 per cent spending increase (implying an output multiplier of 3.6), and the food price goes up 20 per cent. All classes gain in terms of real income and food intake, but the price increase is substantial. Together with productivity increases or increased import supply, increased expenditure can stimulate real output and help maintain the system in balance. However, by itself increased spending benefits farmers and capitalists more than it does wage recipients, and the food price increases it induces could be especially harmful to the poor.


Like many macro-interventions, changes in food price policy cut with more than one edge. The numerical examples just presented show that plausible changes in tax and subsidy policies, import controls, and aspects of the overall economic environment will almost certainly have noticeable repercussions on income distribution, the level of economic activity, and food consumption patterns. In general, some income classes are helped and others harmed by the policies considered, and no clear guideline can be laid down. A fortiori, the same is true of the effects of the policies on nutritional status of different groups in the population, insofar as it is affected by changing levels of food intake.

This conclusion is somewhat negative, but it does point to an important corollary:

Attempts to "get the prices right" may indeed have positive effects on output or economic efficiency, but their negative distributional consequences for some groups in the population may be large. Contrarily, "wrong" or taxed and subsidized prices may help the welfare of some segments of the poor. In any case, the inelasticities of food supply and demand mean that governments will always be present in the market to try to offset the impacts of potentially large price fluctuations on some economically disadvantaged groups. A plausible case for nonintervention cannot be made, but the problem of choosing price manipulations with a minimum of unfavourable side effects and a maximum of benefits is serious and has yet to be solved.


1. A review of country policies is provided by Davis (Ref. 1), who concludes that most of their impacts are for the worse.

2. For example, see the model by McCarthy and Taylor (Ref. 3) for Pakistan.

3. Consumption demand patterns for all three income classes are assumed to be the same. A more complete model would allow for class differentiation, but the detail is omitted here to keep the algebra as simple as it can be.

4. For more detail on the LES, see, for example, Taylor (Ref. 5). For any commodity i, its marginal budget share m' can be shown to be equal to the product of its income elasticity of demand and its share in consumers' total expenditure. In our numerical example, the overall food budget share is 0.375, and an empirically plausible value of 0.4 for the income elasticity of food demand makes the marginal budget mF equal to 0.15. That is, consumers are assumed to allocate 15 cents out of each additional dollar of expenditure to food consumption, and the rest to nonfood.

5. Of course, the supply bottleneck could be broken by allowing more imports, but that could create serious balance of payments problems that the government might choose to eschew.

6. The model was solved along the lines of figure 1, by varying the food price PF until food supply was equal to demand The actual computations were done on a computer, though with a bit of ingenuity a hand calculator might have been used.

7. The results on inflation and real output growth are over-optimistic because with such a large increase in aggregate demand, one would expect non-food sector mark-up rates to increase as capacity limits are approached.

8. There is a large literature about the deleterious impact on producer incentives of increased food imports in poor countries-see Refs. 2 and 4 as examples. As long as food subsidies and taxes are not modified to protect the rural sector against price decreases, the results here show that the previous literature may be too optimistic, as it does not taken into account macro-responses to increased trade. The authorities could modify the fertilizer subsidy or farm price supports to compensate the downward import pressure on PF, but then there would be further repercussions on the budget and the macro-situation to consider. Managing the economy when the inelastic food sector is large is not an easy task.


1, J.M. Davis, "The Fiscal Role of Food Subsidy Programs." International Monetary Fund Staff Papers, 24; 100 127,1977.

2. F.M. Fisher, "A Theoretical Analysis of the Impact of Food Surplus Disposal on Agricultural Production in Recipient Countries." J. Farm Econ., 45: 864875, 1963.

3. F. D. McCarthy and L. Taylor, "Macro Food Policy Planning: A General Equilibrium Model for Pakistan." Rev Econ Statist., in press, 1979.

4. T.W. Schultz, "Value of U.S. Farm Surpluses to Under-developed Countries." J. Farm Econ. 42: 1019-1030, 1960.

5, L. Taylor, Macro Mode/s for Developing Countries. New York: McGraw-Hill, 1979.

TABLE 1. Model Equations

XF= K (XNF)b (1)
XNF = [KbPF/PN (1 -s)] 1/(1-b) (2)
MF + XF = CF (3)
PN = ( 1 + z) wa LN (4)
XN = XNF + CN + AN (5)
Y F = PFX F-PN ( 1 -s) XN F (6)
Yw = W(aLNXN + aLFCF) (7)
yz = zwa LNXN (8)
D=g FYF+g W Yw+,g ZYZ (9)
QF = (PF + WaLF) (1-t) (10)
E = q FQF + q NPN (11 )
CN = q F + (mF/QF) (D-E) (12)
CN = q N + (mN/PN) (D-D) (13)

Fiq.1. Quantities of Food Demanded and Supplied

TABLE 2. Model Initial Values and Parameters

XF (Food production) 5.0
MF (Food imports) 1.0
CF (Food consumption) 6.0
XNF (Non-food inputs into food production) 1.0
XN (Non-food production) 15.0
CN (Non-food consumption) 10.0
AN (Non-food autonomous demand) 4.0
PF, PN (Prices received by producers} 1.0
QF (Consumer food price) 1.0
w (Wage) 1.0
z (Non-food producers' mark-up rate) 0.5
YF (Income of food producers) 4.2
Yw (Income of wage recipients) 11.5
Yz (Income of mark-up recipients) 5.0
s (Subsidy rate on XNF) 0.2
t (Subsidy rate on CF) 0.2
b (Food output elasticity of XNF) 0.16
K (Constant in food production function) 5.0
aLF (Labour input coefficient into food processing) 0.25
aLN (Labour input coefficient into non food products) 0.66667
g F' g W g z (Consumption shares of income) 0.85, 0.95, 0.301
q F,q N (Base consumption levels) 4.8, 3.2
mF, mN (Marginal consumption shares) 0.15, 0.85

TABLE 3. Economic Effects of Policy Changes


Base Solution

Raise food subsidy rate from 0.2 to 0.4

Raise fert. subsidy rate from 0.2 to 0.4

Farm yield decreases by 5%

Food imports increase 1.0 to 1.25

Labour productivity in food processing doubles

Autonomous expenditure increases by 25%

Real GNP 19.0 25.4 20.3 18.5 18.5 18.6 22.6
per cent change   33.9 6.8 -2.6 -2.7 -2.2 18.8
Nominal GNP 19.0 34.7 20.1 19.1 17.9 18.2 23.4
Food producers'
price 1.0 2.35 0.95 1.15 0.85 1.02 1.20
Implicit GNP
deflator 1.0 1.37 0.99 1.01 0.97 0.98 1.04
Income shares
term 0.203 0.279 0.189 0.223 0.178 0.220 0.206
worker 0.556 0.495 0.564 0.541 0.574 0.533 0.550
mark up 0.241 0.226 0.247 0.236 0.248 0.247 0.244
Food consumption levels
farmer 1.34 2.10 1.30 1.43 1.20 1.46 1.41
worker 4.10 4.17 4.34 3.86 4.31 3.97 4.20
mark-up 0.56 0.61 0.60 0.53 0.59 0.58 0.59
Per cent change in food consumption
farmer   57.1 -2.9 6.4 -10.7 9.3 4.9
worker   1.8 6.0 -5.7 5.2 -3.0 2.5
mark-up   7.3 6.7 -5.5 4.4 3.5 5.0
Food sub./GNP 0.079 0.206 0.075 0.085 0.075 0.076 0.077
Fert sub/GNP 0.011 0.016 0.026 0.012 0.009 0.011 0.011

Contents - Previous - Next