Contents - Previous - Next

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

Fighting malnutrition in Mexico: the fiscal cost of alternative policy options

Nora Lustig
El Colegio de Mexico, Mexico City

The incidence of malnutrition in Mexico has been estimated to be quite high. For example, the recent results obtained from the last rural nutrition survey by the Instituto Nacional de la Nutrición in 1979 indicate that, except for the north-western and part of the north central states, the per capita daily caloric intake is not adequate. For protein the situation is even worse because only a small fraction of the north-west coast population achieves an adequate intake (parts of Sinaloa, Nayarit, and Jalisco). On the other hand, the Institute has estimated that the national average daily per capita availability of calories (2,600) and protein (76 grams) is sufficient to meet recommended levels of dietary intake. This is an indicator that malnutrition in Mexico must be attributed to unequal distribution of purchasing power and resources.

The prevalence of widespread malnutrition together with the presence of what has been called the "agricultural crisis," manifested by a relative decline in agricultural production (especially of basic grains), accompanied by a substantial increase in imports beginning in the mid-1960s, led the government of López-Portillo to launch a programme called Sistema Alimentario Mexicano (Mexican Food System) whose two main objectives are to achieve self-sufficiency in staples (in particular, corn and beans) and to improve the income and nutritional conditions of the rural as well as the urban poor.

These two objectives are to be achieved through the implementation of a production strategy and a consumption strategy. The production strategy consists of a series of measures to encourage production of staples, especially in the non-irrigated areas where most of the poorer peasants live (i.e., credit, subsidized inputs of fertilizer and seeds, higher support prices, and sharing the risks of the use of new technology). In the consumption strategy the idea is to implement policies that will enable the currently malnourished population to have a food intake equivalent to the basic recommended diet, a nominative consumption basket estimated by the Mexican Food System to fulfil a series of nutrient requirements at the minimum cost. This could be achieved with different kinds of measures- for example, income transfer, food stamps, price subsidies, and direct distribution.

This paper provides a methodology to measure the food consumption gap in the absence of adequate, detailed food-intake data and to compare the fiscal cost of three target-group-oriented policy measures: income transfer, food stamps, and price subsidies.


One way to calculate the "nutritional gap" and provide an estimate of the number of malnourished in the population is by means of a normative food consumption basket whose cost can be compared with the amount spent on food by different income groups. The gap between the two values, assuming that both have been derived from the same set of prices, gives an idea of the gap in food intake, not measured, however, in terms of calories, proteins, or any other nutrient but rather in the monetary unit. In order to calculate the former deficit in nutrients, it is necessary to compute the gap item by item between actual consumption and that included in the recommended diet. In theory, one could carry out both calculations- i.e., the aggregate and the itemized one- but this was impossible in practice because no reliable information exists on what the actual consumption of each item is.

The construction of the recommended diet can be done in many different ways. The basic recommended diet (BRD) proposed by the Mexican Food System is presented in table 1. The BRD used by the Mexican Food System was computed using linear programming techniques; the objective was to minimize the cost of the diet subject to a series of constraints that include a wide list of nutrients, requirements, consumption habits, etc. (The details of the computation are given in the Documents of Project, no. 3, Mexican Food System, Office of Advisers to the President.) The BRD in grams was then converted into pesos, multiplying each item by its corresponding (regional) price in 1977, which was chosen because it was the year when the Household Expenditure Survey was carried out.

This survey covers nine regions, and the price of the BRD was calculated for each of them. Eight of the regions are groupings of the Mexican states (excluding the three largest metropolitan areas), and the ninth comprises the metropolitan areas of Guadalajara, Monterrey, and Mexico City.

TABLE 1. Basic Recommended Diet (BRD)

Food Items Net Daily
Per Capita
Gross Daily
Per Capita
Corn 497. 19 516.45 43.74
maize in grains 221.48 240.74 20.39
tortillas 268.25 268.25 22.72
dough (mesa) 7.46 7.46 0.63
Wheat 88.14 88.14 7.48
flour 16.6 16.6 1.41
sweet bread 16.78 16.78 1.42
white bread 30.0 30.0 2.54
crackers 8.76 8.76 0.75
noodles 16.0 16.0 1.36
Rice 18.0 18.0 1.52
Beans 52.0 52.0 4.40
Potatoes 23.04 28.1 2.38
Vegetables 35.59 40.9 3.46
tomatoes 30.18 34.3 2.91
chili 1.82 2.0 0.17
onions 2.66 3.1 0.26
lettuce 0.24 0.4 0.03
carrots 0.69 1.1 0.09
Fruits 41.53 63.9 5.41
bananas 17.88 26.3 2.23
apples 0.93 1.4 0.12
lemons 4.77 7.7 0.65
oranges 17.95 28.5 2.41
Dairy and Meat 278.58 306.0 25.94
beef 13.11 19.0 1.61
pork 7.1 10.0 0.85
poultry 6.72 12.0 1.02
other meat 5.46 7.0 0.60
eggs 25.52 29.0 2.46
milk (fresh) 210.0 210.0 17.79
lard 2.0 2.0 0.17
fish 8.67 17.0 1.44
Other 67.0 67.0 5.67
vegetable oil 17.0 17.0 1.44
sugar 50.00 50.0 4.23
Total 1,161.7 1,180.5 100.0

Source: Mexican Food System (1980).

The BRD in 1977 pesos calculated for each region with the corresponding prices was compared with the corresponding actual average per capita expenditure on food for 18 socioeconomic groups constructed in the following way: For each region, households were divided into those with an income less than or equal to the corresponding 1977 minimum wage and those with a higher income, and the former were selected for use in the study. These were then classified into two sectors, rural (sector 1) and urban (sector 2), according to whether the main activity of the head of the household was agricultural or not- the two sectors in each of the nine regions making the 18 groups. Information on the size of towns was not included in the 1977 survey.

The decision to use the minimum wage as a threshold was made in light of the amount of malnutrition in Mexico; but, because this threshold is based on weak grounds, an additional test was carried out: The number of households with an income above the minimum wage that spent less than the computed cost of the BRD on food and the number of households with an income below the minimum wage that spent more than that cost were determined. Only 10 cases were found in a sample of over 8,000 and so the minimum-wage threshold seems to be a safe enough measure for the purposes of the analysis.

For each socio-economic group the deficit between the BRD in pesos and the actual expenditure on food was calculated, and an estimate of the required income was provided. The required income measures the per capita income that would be sufficient to guarantee a level of expenditure on food equal to the BRD in pesos, calculated as follows: For each socioeconomic group, an estimate was made of the corresponding income elasticity for food as the coefficient of a log-linear regression of per capita expenditure on food, based on total per capita expenditure. (The income elasticities are actually expenditure elasticities, but the latter were considered to be the best estimate of income elasticity.) The information came from the 1977 income expenditure survey. The coefficients were estimated using generalized least squares to correct for heteroscedasticity, and the method of instrumental variables with per capita income was used for per capita total expenditure, to correct for errors of measurement. Using the estimated elasticity and the required increment in food expenditure, one can estimate the required increase in income and apply this increment to current per capita income to obtain the required income.

Table 2. Required Monthly Per Capita Income by Region and Sector

Group* Actual Expendi
-ture on Food
mended Expendi
-ture on Food
Deficit** (3) - (2) Deficit in % (4)/(2) Expendi
-ture Elasticity for Food
Required Increment in Income, % (5)/(6) Current Income*** Required Increment in Income*** Required Income *** No. of Yrs.
1-1 7.54 10.40 -2.86 37.93 1.1397 33.28 374.05 124.48, 498.53 6.6
1-2 12.46 10.40 0.7910 492.99
2-1 7.12 9.69 -2.57 36.09 1.4669 24.60 252.59 62.14 314.73 6.3
2-2 12.05 9.69 1.3990 490.83
3-1 6.76 10.18 -3.42 50.59 1.2681 39.80 241.44 96.31 337.75 8.4
3-2 9.74 10.18 -0.44 4.51 1.1753 5.30 368.55 19,53 388.08 1.0
4-1 5.41 10.09 -4.68 86.50 1.2424 69.62 213.80 148.85 362.65 12.8
4-2 6.84 10.09 -3.25 47.51 1.1397 41.69 321.08 133.86 454.94 8.0
5-1 3.59 9.84 -6.25 174.09 1.0244 169.94 190.53 222.91 413.44 20.7
5-2 7.22 9.84 -2.62 36.28 1.1786 30.78 284.38 87.53 371.91 6.34
6-1 6.44 10.20 -3.76 58.38 1.0334 56.49 225,70 127.50 353.20 9.42
6-2 8.72 10.20 -1.48 16.97 0.9364 18.12 358.05 64.88 422.93 3.21
7-1 6.35 10.36 -4.01 63.14 1.1999 52.62 221.06 116.32 337.38 10.03
7-2 8.83 10.36 -1.53 17.32 0.9631 17.98 386.79 69.54 456.33 3.27
8-1 6.10 9.86 -3.76 61.63 1.2265 50.25 202.14 101.58 303.72 9.84
8-2 9.27 9.86 -0.59 6.36 0.9371 6.79 326.35 22.16 348.51 1.26
9-1 8.61 10.37 -1.76 20.44 0.7472 27.36 390.82 106.93 497.75 3.8
9-2 10.31 10.37 -0.06 0.58 1.0858 3.53 512.80 2.72 515.52 0.12

* The first number indicates the geographical region and the second the sector: 1, rural; 2, urban.
** In daily per capita 1977 pesos; excludes alcoholic beverages and soft drinks; includes autoconsumption.
*** In monthly per capita 1977 pesos.

As shown in table 2, the required increment in income can be very high: e.g., for group 5-1 (region 5, rural sector) it is about 170 per cent. It reaches the highest value, in general, in regions 4, 5, 6, 7, and 8, the poorest in the country, especially region 5. These include the states around Mexico City and the southeastern part of the country.

In column 9 of table 2 we have estimated the number of years it would take for the per capita income in each region to reach the required level, assuming an optimistic growth in per capita income of 5 per cent per year and no changes in the distribution of income. In regions 4 and 5 the number of years necessary would be about 13 and 21, respectively.

These simple experiments, in spite of their shortcomings and weaknesses, again indicate the need for implementation of adequate policy measures if conditions in terms of food intake by the poor are to be improved in the short run. In the discussion that follows, three possible policy measures will be compared in terms of their fiscal cost. These policies are income transfer, food stamps, and food price subsidies,


There are a number of measures that a government may implement to improve the nutritional standard of the undernourished. These can range from redistribution of wealth and income to educational programmes geared to change "bad" food-consumption habits. Policies may be directed toward the entire population, such as overall food price subsidies, or may be oriented to specific target groups. The latter approach implies that the measure will, in principle, affect only a particular, pre-selected group in the population. However, in general, any policy will affect more people than those considered the target group because of indirect effects, e.g., creation of employment, inflationary pressures, etc. Among the target-group-oriented policies, the most common options are direct free distribution of food, income transfer, food stamps, and price subsidies. The choice between a general or a target-group. oriented policy must be based on a series of criteria, such as the time required to implement the policy, the associated fiscal cost, the administrative cost, nutrition effectiveness, political feasibility, etc.

The orders of magnitude in terms of fiscal cost are given for three target-group-oriented policy options: income transfer, food stamps, and price subsidies.

Income Transfer

This option entails a direct transfer of income to the population group currently undernourished. The amount of income to be transferred must be sufficient to cover the required expenditure on food, and thus it will depend, among other things, on the marginal propensity of the target group to spend money on food.

The fiscal cost per unit of increment in food consumption in a fraction equal to f can be expressed as:

Pf = (unitary} price of food,
mp = marginal propensity of the target group to consume food,
epd = demand price elasticity of the target group for food,
eps = excess-supply price elasticity of food for target group,
f = required relative increment in food consumption by the target group.

When the commodity is partly imported and the subsidy does not discrimintate between imports and domestic production, the FCu is evaluated for a es = ¥ which implies that eps = ¥ ; the FCu, then, becomes equal to

FCu + Pf/mp

In our calculations we have made this assumption because in Mexico part of the food, especially grain, is imported.

The total fiscal cost of an income transfer programme in Mexico was calculated in the following way: The required marginal propensities to consume were obtained by multiplying our estimated income elasticities by the corresponding average propensity to spend. The inverse of the mp's gives the fiscal cost per unit of increment as a ratio of the price. To obtain the fiscal cost associated with the required increment and evaluated at the relevant price, these magnitudes can be multiplied by the corresponding required increment in food expenditure. If this new magnitude is then multiplied by the size of the population and by 365 days and then added for the 18 socio-economic groups, an estimate of the total fiscal cost per year for this particular policy is obtained This calculation shows that the total fiscal cost of implementing the income transfer would have amounted to something like 2.56 per cent of the 1977 Mexican GDP.

This magnitude seems quite high, especially when compared to the proportion of government expenditure in toto: in 1977 this was equal to 11.7 per cent of the GDP. Moreover, the cost of this policy would be even higher when the administrative costs of implementing it are taken into account. However, one should bear in mind that we are calculating the fiscal cost of the income transfer policy with a series of parameters that may be overestimating the true fiscal cost, but this allowed us to be on the safe side in terms of combating undernutrition in Mexico.

Food Stamp Programme

Food stamp programmes have been widely used in the United States. Such a programme provides participants the opportunity to purchase a fixed amount of food at a cost below market prices. If the consumers' indifference maps were known, then an optimal food stamp programme could be implemented whose cost per unit of food reaching the target group would be less than the price of the food. However, this is impossible in practice, and thus one has to settle for a suboptimal programme where the cost per unit of increment is equal to the price of food- i.e., where the cost of the food stamps equals what the consumer's expenditure on food would be without the programme. Where programmes have no effect on the supply price of food (e.g., when food is partly imported), the fiscal cost per unit of increment as a ratio of the price of food of a suboptimal food stamp programme would then equal unity.

The total fiscal cost is calculated in a manner analogous to that of income transfer: FCu = 1 peso is multiplied by the food expenditure deficit, and this quantity is then multiplied by population and by 365 to obtain the total fiscal cost per year. The fiscal cost of the food stamp programme would be 1.60 per cent of the 1977 GDP (which may be compared with the cost of 2.56 per cent we calculated for income transfer).

Food Price Subsidy

A target-group-oriented food price subsidy provides beneficiaries with as much food as they wish to purchase at a reduced price In general, a price subsidy would not be applied to all food but to some pre-selected commodities. However, state-owned food stores may distribute all food items at a constant discount, and if only the target group is allowed to shop at such stores, this would be a real example of a target-group-oriented overall food price subsidy. The fiscal cost per unit of increment as a ratio of the price of food associated with this policy is

FCu =(1 + f) /epd

where f is the required increment in food consumption, measured as a fraction of initial consumption.

At the national level, per capita daily food expenditure by the target group was estimated to be 7.15 pesos in 1977 and the deficit to be -3.02 pesos. These two magnitudes were obtained by weighing the regional-sectoral ones by the respective populations.

TABLE 3. Fiscal Cost of Different Policy Options

  General Formula* Unitary Fiscal Cost in
1977 Pesos**
Total Fiscal Cost*** (millions of pesos per year) Total Fiscal Cost as % of 1977 GDP
Income transfer 1/mp 1.898 50 872.9 3.04
Food stamps 1 1 26,785.8 1.60
Target-group price subsidy (1+f)/epd 1.814 48,621.4 2.91

These calculations were made under the assumption that food is pertly imported (i.e., eS (r) ¥ )
*mp = 0.5270; f = 0.42; epd = 0.784
** These are the fiscal costs per unit of increment as a ratio of the price of food.
***The toted fiscal cost is obtained by multiplying unitary fiscal cost by the deficit (3.02), the target population size (24,315,9001, end 365 days.

Because it was not possible to calculate the price elasticities by region or sector, we computed the fiscal cost of a food price subsidy for the target group at the national level, i.e., those households that earned an income less than or equal to the corresponding minimum wage. In order to carry out the comparisons with the fiscal cost of the income transfer and the food stamp programme, we recalculated the first using the propensity to spend at the national level as well.

Table 3 summarizes the fiscal cost per unit of increment and the total fiscal cost associated with the different policy options. As can be seen, the target-group food price subsidy is less expensive in terms of its fiscal cost than income transfer; however, it is quite a bit more expensive than the food stamp programme. As a percentage of GDP, the fiscal cost of income transfer would equal 3.04 percent, while that of the food price subsidy and the food stamp programme would be 2.91 per cent and 1 60 per cent respectively.

An income transfer, however, has a major advantage over the other two policy options in that it allows for an increase in the consumption of products other than foodstuffs and thus raises the welfare level of the target population.

It should also be noted that the required decline in the price of food to encourage an increase of 42 per cent in the consumption of food by the target group would be of an order of magnitude of 48 per cent. Such a decline may not be possible to sustain from either a practical or a political point of view.


On the basis of fiscal cost, our results show that a food stamp programme would be by far the most cost-effective: it would cost almost 50 per cent less than an income transfer, for example. The second most cost-effective policy is a target-group-oriented price subsidy, followed by the third option, income transfer. These results are not surprising because the marginal propensity to spend on food as well as the price elasticity for food are bound to be less than unity, and, as we saw in the formulas, this is the main reason why income transfer and price subsidy will be higher in cost than the food stamp programme. In cases where the target group is extremely poor, however, the price elasticity and the marginal propensity to spend on food tend to approach one, and the three programmes are almost equally cost-effective. In general, a target-group-oriented price subsidy would be more cost-effective than an income transfer when the larger price elasticity is compared to the marginal propensity to spend. Nevertheless, income transfer can improve the general standard of living more effectively.

Contents - Previous - Next