Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

OWN-PRICE ELASTICITIES

The term "own-price elasticity" refers to the elasticity of demand for a commodity with respect to its own price, i.e. the responsiveness to its own price of the quantity of a commodity demanded.

The own-price elasticities studied are shown in table 7. They were all negative and statistically significant at a 5 percent level except for fish, sea-foods, and milk in quartile IV. The results supported a priori expectations that the quantity demanded of a commodity would vary inversely with its price. Across commodities, rice, corn, sugars, fats and oils, and fish had smaller elasticities in absolute value than other cereal products, which were mostly wheat based, and than fruits and vegetables, meat, eggs, and milk. Staple foods, namely rice and corn, and fish, the second most important protein source, were not as price elastic as non-staples and luxury items.

Across quartiles, the absolute values of the price elasticities declined as income increased for sugar, the fruit and vegetables group, fish and sea-foods, and meat. A U-shaped pattern was detectable for corn and corn products and other cereal products, with troughs in the third quartile, and for eggs, with a trough in the second. An inverted U shaped pattern was shown by rice and starchy roots, with a peak in quartile 11, as well as by eggs, with a maximum also in quartile 11.

The decline in the own-price elasticities reflected a fall in food-budget shares and food-budget elasticities for necessity or staple foods as income rose. This result was similar to the decline in elasticities described by Timmer [43]. However, the non-linearities seemed to indicate that the relationship between the price elasticities and income was not monotonic. Moreover, the behaviour was more noticeable in the case of energy foods such as rice, corn, other cereal products, and roots. The non-monotonicity of the price elasticities conformed with Bouls' results but was contrary to the linear decline hypothesized by Timmer [6]. The peak in the price elasticity in quartile 11 for rice and roots reflected the consumer's increased ability to purchase and substitute preferred energy foods for less preferred ones, for example rice for corn; having satisfied his hunger or "bulk" constraint to some degree, he could consider diversifying his diet [6]. The higher values of the elasticities might also have been due to the existence of a wider range of affordable substitutes in the energy foods group once income reached the second quartile level.

TABLE 7. Own-price elasticities, 1978 FNRI Survey: seemingly unrelated regressions (SUR) results

a. Significant at the 0.05 level.
Source: Food and Nutrition Research Institute [14].

This result was also analogous to Rao's observation regarding changes in the proportion of income spent on food [35]. Rao postulated that until food needs are satisfied, people spend relatively more of their incremental income on food. This behaviour is revealed in an increasing or invariant proportion spent on food as income increases up to a certain point. Rao suggested that this critical level, which corresponds graphically to the level at which the decline in the proportion spent on food is marked and smooth, could be used to develop a poverty line. Perhaps patterns in cereal consumption could also be used to determine the monetary equivalent of a food energy threshold.

The absolute values of the price elasticities were quite large compared to most of the previous estimates for the Philippines (table 8) but were similar to results for Brazil [17], Indonesia [45], and Thailand [47] (table 9). While differences in the choice of functional form could be a source of variation, most of the variation could be traced to the nature of the data. Most food-demand studies in the Philippines have been based either on time-series data or on the cross-sectional surveys conducted by the Ministry of Agriculture Special Studies Division (MA-SSD) [1]. Covering a longer time period, these sources exhibit greater price variation relative to quantity variation in comparison to a single-period, cross-sectional data set. It is commonly accepted in empirical work that time-series estimates yield smaller coefficients than corresponding cross-sectional estimates, since the former represent short-run adjustments that do not capture the full long-run response. Thus, estimates based on time-series data and a series of pooled cross-section surveys would yield smaller elasticity estimates than those from a single cross-sectional survey. The Brazilian, Indonesian, and Thai studies, on the other hand, were based on cross-sectional data from surveys conducted within a one-year period: the National House hold Expenditure Survey (ENDEF) was conducted by the Brazilian Geographical and Statistical Institute from 1974 to 1975; the National Socio-economic Survey (SUSENAS V) was carried out by the Central Bureau of Statistics, Indonesia, in 1976; and Thailand's National Socio-economic Survey was done from 1975 to 1976. These results should, therefore, be interpreted as long-run elasticities, reflecting long-run adjustments to relative prices, short-run elasticities being substantially smaller. The results of the simulation should also be interpreted with caution, for these may overstate the actual values of the short-run response. If the true value of the long-run price elasticities is relatively small, then the prospects for increasing consumption using consumer price subsidies are not too bright.

The complete price and cross-price elasticity matrices and a discussion of the cross-price interactions can be found in Quisumbing [33].

THE NUTRITIONAL IMPACT OF MARKET INTERVENTION POLICIES: A SIMULATION APPROACH

This section examines the effects of food-market intervention policies on the nutrient consumption of several income groups using the estimated income-stratum-specific demand elasticity matrices. The discussion is divided into two parts. The first examines the nutritional effects of income transfers and price subsidies, both targeted and non-targeted The second part evaluates the cost-effectiveness of food-market intervention policies as a function of the desired percentage calorie gain.

Nutritional Impact of Food-budget Transfers and Price Subsidies

Priority should be given to income-transfer and price subsidy policies that increase calorie consumption by calorie-deficient households, since calorie inadequacy is a more basic nutritional problem than is protein deficiency. The SUR estimates discussed in the previous section were used to simulate the potential impact of these policies.

Since the elasticities estimated were long-run elasticities, the simulation results should be interpreted as the potential effects of the implementation of a package of policies for a period of approximately five years or longer. The results, therefore, do not represent one-time cash or income transfers or temporary price subsidies or increases.

The long-term effects of agricultural market intervention policies are valid concerns, especially for the Philippines, where a complex array of institutions and policies have created taxes or subsidies on agricultural output, often to the disadvantage of the agricultural sector. For example, wedges between producer and consumer prices or between domestic prices and border prices may have been maintained over long periods by the structure of agricultural incentives and domestic protection. Cases in point are the Philippine government's intervention in rice marketing, the coconut levy, and many others [11 ] . Previous studies have shown that the rice-market intervention has been consumer oriented, even biased, towards the urban consumer in recent years [24], while the burden of the coconut levy has been borne largely by coconut farmers [9, 10].

TABLE 8. Summary of previous elasticity estimates for selected food items in the Philippines

a. Numbers in parentheses following authors' names refer to references at the end of this article. Acronyms not spelt out in full in this table are spelt out in the reference list.
b. R. Frish, ''A Complete Scheme for Computing All Direct and Cross Demand Elasticities in a Model with Many Sectors,'' Econometrica, 27: 177-176 (1959).
c. Carabao meat.
d. A model incorporating slope shifters for each income group, with prices and income deflated by the consumer price index.
e. R.R. Betancourt, "The Estimation of Price Elasticities from Cross-sectional Data under Additive Preferences,'' International Economic Review, 12 (2): 283-292 (1971).

TABLE 9. Summary of price elasticity estimates for selected food items, single cross-section survey, Brazil, Indonesia Thailand, and Philippines

Long-term income transfers can result from institutional changes that alter relative incomes of various groups, but not from short-term wage or tax policies. An example of a change that would effect a long-term income transfer would be successful land reform. In addition, more specific, or food-linked, income transfers have actually been operative in economies such as those in Egypt and Sri Lanka, which have sizeable food subsidy and distribution programmes. Also included in the category of food-budget transfers are food stamp programmes.

General price policies, on the other hand, have often been used to achieve conflicting objectives: high food prices to maintain agricultural producer incentives and low prices to protect poor consumers. Unless the two groups are effectively insulated by some tax-cum-subsidy policy, prices will no longer be able to perform their function of maintaining allocative efficiency. Economy-wide price intervention policies for the sake of increasing nutrient intake would then be very expensive to implement. In addition, the actual effect of such policies may be biased toward achieving one set of objectives rather than the other.

For example, Regalado mentioned that the government has been more effective in defending the price ceilings than the price floors for rice and corn [1, 36]. However, in the case of other food commodities such as milk, sugar, cooking oil, and meat products, the government seemed to be ineffective in defending ceiling prices. Thus, retail prices paid by farmers for these commodities were generally higher than the price ceilings. This set of policies appears to be biased against rice- and corn-growing farmers.

Because of the cost of maintaining such policies and the possibility of conflicting producer and consumer objectives, it is perhaps desirable to grant selective subsidies, concentrated either on vulnerable groups or targeted at basic food staples consumed by the poor [41]. Another alternative would be selective food-linked income transfers to the poor.

Effects of Food-linked Income Transfers

Food-linked income transfers serve to increase the demand for a commodity at the prevailing price. The effect of such transfers upon nutrition depends upon relative preferences for food compared to non-food items and the ability of supply to meet the increased demand. Food-linked income transfers (or food-budget transfers) have become an increasingly important way of transferring incomes to the poor. A food stamp programme is one example of a food-linked income transfer.

Unless the marginal propensity to use transfer income for food consumptions is equal to one, however, or unless the marginal budget share of food is equal to unity, it cannot be guaranteed that an income transfer will be spent exclusively on food. So, using food-budget elasticities to estimate the nutritional effects of an income transfer may be over-optimistic and will not consider the "leakage" in the form of an increase in purchasing power for non-food products.

We therefore transformed the food-budget elasticities into income elasticities by multiplying the food-budget elasticities with the stratum-specific marginal budget shares and the reciprocal of the food-budget share, which is equal to the elasticity of food expenditure with respect to income. These estimates were obtained using marginal budget shares for low and high income groups for Thailand and actual expenditure shares for the Philippines [27, 46, 47]. Stratum-specific estimates of the income elasticity of food expenditure were not available for the Philippines. Estimating these elasticities from the 1978 data set was not advisable because it contained under estimations for income data [34]. The marginal budget shares, the average budget shares, and the computed income elasticities of total nutrient consumption are shown in table 10.

Elasticities of total nutrient consumption. Income and food-budget elasticities of total calorie and protein consumption are presented in table 10. As expected, after adjusting the food-budget elasticities by the income elasticity of food expenditure, the income elasticities of total nutrient consumption were lower than their corresponding food-budget elasticities. In general, the income elasticity of calorie consumption was less than one and decreased with increasing income except for fourth quartile households, A pattern of decreasing total calorie income elasticities was expected since households at the higher income levels would be nutritionally better off and more likely to spend additional income on purchases other than food. The slight increase in the fourth quartile of protein and calorie elasticities could be due, however, to increased consumption of expensive calorie and protein sources.

Assuming perfectly elastic supplies, the income and food budget elasticities for total calorie consumption can provide us with the minimum required percentage change and the minimum size of the transfer needed to close the gap between the estimated average energy requirement and the amount of calories actually consumed. Of course, closing the gap is conditional upon targeting the transfers perfectly.

TABLE 10. Marginal and average budget shares, food budget and income elasticities of total calorie and protein consumption

a. Computed by adjusting Trairatvorakul's estimates for high and low income consumers in Thailand [471.
b. National Census Statistics Office, 1975 Family Income and Expenditure Survey [27].

TABLE 11. Minimum change in the food budget required to eliminate calorie deficits

a. Estimated average energy requirement (AER): 2,036 kcal per capita per day.

A few calculations will give us an idea of the percentage increase in the food budget needed to bring calorie consumption by deficit groups up to 2,036 keel per day, or the estimated average energy requirement (AER).

For income quartile I, expanding consumption from 1,589 to 2,036 kcal per capita per day would amount to an increase of 28 percent for quartile II; expanding consumption from 1,769 keel to the AER would amount to an increase of 15 percent, while only an 8 percent increase would be needed in quartile III to meet the AER. Using the total calorie consumption elasticities these calorie increases would require a 19 percent increase in the food budget in quartile I households (from 1.88 to 2.24 pesos per capita per day), an 11 percent increase in quartile II households (from 2.40 to 2.66 pesos per capita per day), and a 6 percent increase in quartile lit households (from 3.02 to 3.21 pesos per capita per day) (table 11).9 These estimates rely on the assumption that food prices have not changed.

The corresponding income increases required to close the calorie gap are computed analogously. In both quartiles I and III, a 17 percent increase in income was required. While the results for the third quartile may appear paradoxical, they can be explained by the fact that quartile III households consumed more expensive calorie sources than those in quartile II and, more importantly, had a higher income elasticity of non-food expenditure. Thus, we expect that there would be substantial income leak ages to non-food commodities if quartile lIl households were to receive an income transfer.

TABLE 12. Estimated change in nutrient consumption, 10 percent income transfer

Regalado also estimated the minimum income change needed to eliminate calorie deficiency [36]. She found that a 37 percent increase in annual per capita income was needed for the first stratum as defined in her study to close the calorie gap, while a 46 percent increase was required by the second stratum. The upper two income groups, strata III and IV, could undergo decreases of 11 percent and 47 percent respectively, and still have sufficient calories. The differences in Regalado's results and those of this study can be attributed to the different sample stratifications used and, thus, the correspondingly different calorie gaps.¹⁰

Targeted versus non-targeted income transfers. Table 12 presents estimates of the potential change in nutrient consumption arising from a 10 percent income transfer, using alternative assumptions of unitary and zero supply elasticities, denoted as S = 1.0 and S = 0.0 respectively

For households deficient in nutrients, i.e. those in income quartiles I to III, the potential increase in nutrient com gumption, taken as a whole, was greater assuming unitary supply elasticities than assuming zero supply elasticities. Although the percentage increase in calorie consumption for households in quartile I seemed to be greater assuming S = 0, the percentage decreases in quartile II calorie consumption were likewise large; increased calorie consumption by the lowest income group seemed to have been obtained at the expense of another calorie-deficient group. Assuming unitary supply elasticity, the percentage gain in quartile I consumption might be smaller, but so was the percentage decrease in the consumption of households in quartiles I and II.

Quartile IV households clearly experienced higher percentage increases in calorie consumption if supplies were assumed to be inelastic. This observation was attributed to the fact that, assuming inelastic supplies, consumers are competing for a fixed supply of goods, and the resultant increase in price due to an upward shift in demand will dampen the initial increase in demand by the lower income groups. The higher income group will experience increased nutrient intakes because they can afford to purchase goods even at higher prices. These results suggested that if supplies are relatively inelastic, higher income groups should not be beneficiaries of a transfer programme. Such a transfer would create nutritional waste at the expense of the nutrient-deficit groups. This case stands in sharp contrast to the perfectly elastic case discussed earlier.

Table 12 also shows that the more precise the degree of targeting to the group most nutritionally at risk, the greater the gain by that group. For example, quartile I households stood to gain in calories by 30.38 percent (S = 1.0) if the transfer were targeted only to that quartile. The potential increase if the transfer were directed to both quartiles I and II was only 28.66 percent (S = 1.0). The results were similar for calorie consumption for the S = 0 case as well as for estimated changes in protein consumption.

Estimated calorie gains per peso. Since each quartile has different absolute per capita incomes, an equal percentage income transfer would, of course, amount to different absolute budgetary outlays for the agency implementing an income-transfer policy. A 10 percent income transfer to quartile I would amount to 0.33 pesos per capita per day (in 1978 pesos) and a 10 percent transfer to quartile II to 0.46 pesos, while 10 percent food-budget transfers to quartiles III and IV would equal 0.67 and 1.48 pesos per capita per day respectively. These costs were obtained by adjusting the food budget upwards using the average food expenditure shares from the 1975 Food Income and Expenditure Survey (FIES) and then computing the cost of a 10 percent expenditure transfer.

TABLE 13. Absolute calorie gain and calorie gain per peso, 10 percent income transfer

The estimated calorie gains per peso can be obtained by dividing the absolute calorie gains by the total cost of the transfer policy, i.e. the sum of the transfers to all groups covered by the policy. The total cost per person of a 10 percent income transfer to all quartiles was 2.94 pesos per day; a 10 percent transfer to quartiles I and II costs 0.79 pesos per day; and a transfer to quartile I alone amounts to 0.33 pesos per day. The absolute calorie gains for each of the income groups from the above mentioned policies are presented in table 13, together with the computed calorie gain per peso.

We were primarily concerned with the calorie gains of the lowest income group, that is, households in the first quartile. As expected, the calorie gains per peso are greater the more precise the degree of targeting. Looking at the S = 1.0 case, an income transfer to quartiles I and II gave about 3.9 times as much gain per peso as a transfer to all quartiles, while a transfer to the first quartile alone was 10 times as cost-effective as a general transfer. For the S = 0 case, a transfer to quartiles I and II was also 3.9 times more cost-effective than a general transfer, while a transfer targeted to quartile I was 9.8 times as cost effective as a general transfer.

We must note, however, that we have not discussed the mechanisms under which such transfers can be realized. It is likely that, in practice, food-linked income transfers may be feasible and economically sustainable as deliberate policy interventions only if they are targeted; otherwise, the alternative is for distribution-oriented economic growth to increase the incomes of the poor. in the long run, such growth may well be the only way to achieve a sustainable, permanent improvement in nutritional status. We turn now to a discussion of price subsidy schemes.

Effects of Price Subsidies

This section compares the effects of targeted and non-targeted price subsidies for three commodities — rice, corn, and oil — on the nutrient consumption of various income groups. Additional commodities and combinations have been simulated in the case of a targeted subsidy to the first quartile. Starchy roots have been added as a possible subsidy policy candidate, while combinations of rice and corn and rice and oil have also been studied.

The choice of commodities for the simulation was guided by several considerations. First, more expensive commodities consumed mostly by higher income groups are not desirable to subsidize. Since these groups already have adequate nutrition, subsidies would involve nutritional waste since the impact on the nutrient-deficient households is not likely to be great, especially if the size of the subsidies is infra marginal. Moreover, such subsidies would involve sizeable costs. Therefore, commodities like meat, poultry, eggs, milk, sugar, and other cereal products were not considered. Second, since general subsidies are quite expensive, it was necessary to be selective. Subsidies are better directed towards foods that are inexpensive and consumed by the poor and that have desirable nutritional qualities. The commodities that are cheapest in terms of pesos per nutrient unit are corn (0.66 pesos per 1,000 kcal); rice (0.80 pesos per 1,000 kcal); and oil (0.91 pesos per 1,000 kcal) (table 14). Third, subsidized foods must be acceptable in general, that is, reasonably consistent with existing dietary patterns.

continue

Contents - Previous - Next