4. COMBINING DATA FROM DIFFERENT SOURCES

4.1. Food Balance Sheets and Trends in Nutrition

Food balance sheets are a tool for analyzing the sources and uses of all major food and feed products. For each product, all the sources of supply and all the destinations are tabulated in a numerically consistent way. On the international level, extensive work has been done on these accounts for many countries by the FAO, which calls them “supply and utilization accounts.”

The sources of supply are three: domestic production, imports, and reductions in domestic inventories. The uses, sometimes called “disappearances,” are more numerous: home retentions for seed, home retentions for human consumption, home retentions for livestock feed, wastage that occurs in the course of storage and transport, sales to processing industries, sales to livestock owners, sales directly to the domestic market, export sales, and accumulation of inventories.

Food balance sheets are supply-demand balances written out for each major product in numbers. The methods of estimating the entries differ, and therefore they usually are mutually inconsistent the first ti e around. The food balance sheet becomes a tool for rectifying the supply-demand accounts, for making initially inconsistent data consistent.

Each of the supply-demand balances is a commodity balance in the spirit of input-output economics; it is an accounting identity that must be observed strictly in the real world, even if measurement errors for the various items lead to its being numerically inconsistent. There are no fixed rules for reconciling the initial data, but usually reference is made to several sources and technical coefficients (such as feed coefficients and processing conversion factors), and judgments are made as to which of the initial data sources are likely to be the most reliable. A manual of techniques for constructing the balance sheets is found in Tuck, 1986. The techniques also are discussed in Timmer, Falcon and Pearson (1983, pp. 22–26), who present an example for Indonesia (Table 7) and suggest disaggregating the balance sheets by income class, making use of data from household budget surveys.

The process of reconciling the food balance sheets brings additional information, and logical discipline, to bear on the initial estimates of items like domestic production. For that reason, it usually results in improved estimates. It is this cross-reconciliation that causes input-output tables to be regarded as more reliable than say, national accounts series. When an input-output table becomes available, in many countries it is utilized to revise the national accounts series.

In the food balance sheets, the main item calculated as a residual from the accounting identity is likely to be human consumption. As noted, though, factors like the growth of personal consumption in the national accounts will be brought to bear on the estimates, and therefore the supply estimates may be altered as well. Since the consumption figures are not taken directly from surveys for most years, they are referred to as “apparent consumption.”

Table 7. SUMMARY FOOD BALANCE SHEET FOR INDONESIA, 1976 (in thousand tons)

Source: Timmer, Falcon and Pearson (1983), pp. 24–25.

Note; The original table contains more products than this extract does.

Time Series of Food Balance Sheets

Food balance sheets are most useful when a time series of them is constructed. Via these data, an appreciation can be gained of the relative importance of the growth of demands for agricultural products other than for direct human consumption. Obviously, for sane commodities the uses as livestock feed are dominant, but for others a surprisingly high proportion of output may go to industrial uses (for example, starch in the case of corn) and for seed retentions (cassava and yams). Understanding the role of each of these uses is basic to making demand forecasts, and therefore for projecting possible import requirements in the future. Also, in focussing on the wastage factor, the food balance sheets may provide the basic justification for investigating the possibilities for improved facilities for storage for the perishable products.

It is a common experience to find that the apparent growth rate of the demand for food grains, when calculated from these time series, is much higher than the growth rate which would be implied by GDP and population growth and cross-sectionally estimated income elasticities of demand. One of the reasons for this outcome is the set of difficulties associated with the application of parameters from a cross-section to economic behavior over time. Another reason is that the elasticities derived from household surveys refer only to direction consumption in households, and the aggregate trends include the typically rapid growth of industrial uses and livestock feed uses.

Likewise, the non-consumption uses need to be accounted for when estimates are made of nutrient availability for the human population, as discussed above and exemplified in Table 7. A time series on apparent consumption by crop, combined with a manual of nutrient coefficients by type of food, is the best source of information for computing the apparent time trends in nutrient availability, in the aggregate. (However, household surveys are needed in order to be able to develop estimates of the distributional aspect of nutrition.)

In sum, the food balance sheets are rather basic tools when the analysis deals with projections of food demand, food import trends, and/or estimates of the trends in the availability of nutrients for the population. The latter, of course, is an important indicator of possible problems in nutritional status.

4.2. Prices and Costs; Measures of Economic Protection

4.2.1. The Significance of Protection Measures

Economic protection—the degree to which domestic prices are sustained above world market prices—is a subject close to the heart of many discussions of agricultural policy. At least some agricultural prices deviate from international levels in all countries, developed and developing alike. In fact, the extremely high degree of protection given to rice in Japan often is cited as an outstanding example of a protectionist agricultural policy. In recent decades, the Japanese price of rice at the farmgate has been three to four times higher, or more, than the international price.

Protection can be negative, to the detriment of farmers. Scobie (1981) presents data showing that from 1950 to 1980, the domestic farm-gate price of cotton in Egypt averaged eighty percent below its international equivalent.

In policy debates, the general presumption is that high positive protection is prejudicial to economic welfare in the long run. It penalizes the country's consumers (in the short run as well) through higher commodity prices; it distorts the allocation of resources away from a more efficient pattern that could be achieved through greater reliance on the mechanisms of international trade; and it diminishes economic welfare in other countries through reducing their opportunities to participate in international trade.

There also is a presumption that uneven rates of protection across products are inimical to the attainment of maximal economic efficiency, for they also affect the pattern of resource allocation.

These concerns are equally relevant to agricultural analysis, but there are other concerns as well. First, as the example of Egyptian cotton shows, protection rates can be negative, with unfortunate consequences for farm incomes and for incentives to adopt improved techniques of production. Second, the presence of macroeconomic distortions in the economy (overvalued exchange rates, protection to the industrial sector, etc.) can impose an implicit tax on agriculture, and in the absence of economic reform policy makers may wish to maintain modest rates of protection in agriculture as a compensating measure. Third, there is growing recognition of the influence on international markets of agricultural subsidies in industrialized nations (Norton, 1987) and consequently arguments for compensatory protection on the part of developing countries are gaining acceptance.

For these reasons, the optimal or desirable rate of protection is not always immediately obvious, but it is clear that protection that is excessively high, strongly negative, or sharply uneven over products has negative net economic effects.

Protection coefficients are important analytic tools for monitoring the performance of the sector over time. Protection rates can vary substantially in short periods of time, mainly because they are measured against international prices, which vary considerably from year to year. Thus the same domestic price can be highly protectionist one year and negatively protectionist the next. (The Scobie study on Egypt shows a remarkable amount of year-to-year variation in the nominal protection rates for wheat, rice and corn (Scobie, 1981, p. 26). Strong annual variations in the domestic resource cost coefficient—defined below—are shown for wheat, corn and sorghum in the case of Argentina by Reca, 1980, p. 22.)

In addition to serving to monitor the degree of protection in the sector, some kinds of protection coefficients (DRCs) are useful in assessing the comparative advantage of products in the sector. This information is central to developing production and trade strategies for the sector.

It is hard to exaggerate the usefulness of measures of economic protection for obtaining an economic overview of agriculture. As noted in chapter 1, three of the most basic tools of analysis for starting a review of the supply side of the sector's performance are a production index, an index of real farmgate prices, and a set protection coefficients.

4.2.2. The Calculation of Protection Coefficients: the NPC

The simplest measure of protection, the nominal protection coefficient, involves a comparison of a product's domestic price with its international counterpart, after converting to common currency units via the exchange rate. In the early empirical economic studies of agriculture in developing countries, the comparison was made between the domestic producer (farmgate) price and the corresponding “border” price, which is the c.i.f. price in the case of import-competing products and the f.o.b. price in the case of exports. (See section 2.3 above for the sources of data on those prices.)

Defined in this way, the nominal protection coefficient. for product i is expressed as follows:

(15)

where pf_i is the farmgate price of the product, pw_i is the world price in dollars, and e is the exchange rate, in units of domestic currency per dollar. (A useful basic description of this coefficient and other measures of protection is found in Scandizzo and Bruce, 1980.)

In the absence of any protection, NPC_i = 1.0. With positive protection, it is greater than 1.0. The nominal rate of protection often is defined as 100(NPC_i. - 1.0), so it has expressions like 21 percent, or -43 percent. However, in usage the terms rate and coefficient tend to be used interchangeably, so the reader will have to decide from the context which formula is implied.

In empirical work it has been realized that the farmgate price and the border price are not comparable because they may not represent prices at the same stage of the marketing chain. Adjustments must be made to the numerator or denominator, or both, of expression (15) to make the prices equivalent. It may also be necessary to make adjustments for distortions in the exchange rate. Von Braun and de Braun (1983) adjust the export and import prices, allowing for processing and marketing margins and overvaluation of the exchange rate, to arrive at “border prices at the farmgate” in the case of Egypt.

When the exchange rate is not in equilibrium, it is useful to calculate (15) both with the official exchange rate (e) and with the * estimated equilibrium exchange rate (e ). A comparison of the two versions of the formula will tell immediately how much of the protection, or lack thereof, is attributable to exchange rate policy. But it should be borne in mind that the estimates of the equilibrium exchange rate may not be firm, so it is not wise to calculate the NPC only e , but rather it should be done both ways. (In another example, Tshibaka, 1986, also attempts to build a measure of the disequilibrium in the exchange rate, based on purchasing-power parity theory, into his calculation of coefficients of protection for the case of Zaire.)

Adjusting the marketing margins to make farmgate prices and border prices equivalent, with or without the exchange rate adjustment, is appropriate for exports, as the domestic product passes in well defined marketing stages to the border for shipment abroad. However, it is not so appropriate for imports, because the domestic product is not shipped to the border, nor the imported product to the farmgate.

For import-competing products, it is conceptually preferable to select a geographical location (such as the capital city) and calculate the costs of delivering the imported product to the wholesale markets there, and to do the same for the domestically-produced commodity. These two wholesale prices are then comparable in terms of the stage of marketing, and they may be inserted into (15), with the appropriate redefinition of the price symbols. This modification should be regarded as essential in calculating the NPC for import-competing goods.

The internal costs of the imported product have to include offloading from the ship, port charges, handling and storage at the port, tariffs (if any) and the costs of transportation to the selected wholesale market. As a rule of thumb., these costs will add 15–20 percent to the c.i.f. price of the good, but of course they vary considerably by country and over products.

In some instances it is worthwhile to select two points in space for purposes of making the price comparison, one relatively near a major port and one farther in the interior of the country. It can be the case that the imports have a competitive advantage in serving locations near the port, while the domestic product has the advantage elsewhere in the country. This turned out to be the case for Haiti in the case of corn (Norton, 1984).

It also is important to compute the NPC for more than one

agricultural product, to be able to assess the degree of uniformity of protection policy within the sector, and to compare the results with estimates of protection for the industrial sector when they are available.

Finally, it is best to calculate the NPCs for several years, or at least three years, before attempting to draw conclusions about protection policies. As discussed previously, the NPCs are surprisingly unstable over time. Three sources of instability are at work, and sometimes they reinforce each other with the net result of very sharp variations in the NPC: 1) variations in the real domestic price, 2) variations in the world market price, and 3) variations in the degree of disequilibrium in the exchange rate.

In summary, the following steps are required in computing the NPCs:

1. collection of the border prices and farmgate prices for a series of years;

2. selection of the geographical destinations (demand locations);

3. estimation of the processing and marketing margins, and consequent modification of the original farmgate and border prices;

4. estimation, or gathering of existing estimates, of the degree of disequilibrium in the exchange rate;

5. repetition of the above steps for other years; and

6. application of the formula.

Other variants of the NPC have been suggested. Scandizzo and Bruce (1980) suggest estimating all the domestic subsidies (and taxes) that directly or indirectly affect the domestic price, and then to calculate (15) after having added the net subsidy element to the numerator. They call the resulting coefficient the effective subsidy coefficient, or ESC. (Gotsch and Brown, 1980, provide estimates of ESCs for Pakistan.) However, as prices, and hence protection are in part the net result of the prevailing set of subsidies and taxes (as in the case of subsidies through parastatal processing industries—see section 3.3), there would appear to be a possibility of double counting in attempting to use existing domestic prices and then to add in the element of taxes and subsidies.

A more narrowly defined procedure might be more tenable: add in to the NPC's numerator only those subsidies and taxes that affect farm-level production costs. This version of the ESC would be helpful in revealing cases such as the following: nominal protection maintains domestic prices 30 percent above world prices, and a cost subsidy of 20 percent (of price) maintains in production farms whose costs are up to 50 percent above the corresponding world price. But this calculation is approaching the concept of effective protection, rather than nominal protection. It may be simpler and clearer just to deal with those two basic concepts.

Effective protection is discussed below, but the NPC, simple as it is, remains the most widely used coefficient of protection. It is the simplest protection measure to calculate, and it quickly gives an idea of the quantitative effects of policies that operate by influencing prices.

4.2.3. The Effective Protection Coefficient

The NPC measures the ratio of domestic to international prices as received by the producer. However, any distortions on the returns to his output may be nullified by taxes (subsidies) on his inputs. This phenomenon, of balancing implicit taxes on production with subsidies on inputs, or vice-versa, is quite pervasive. To account for it in the measures of economic protection, the effective protection coefficient (EPC) was devised. In concept, it is the ratio of value added (in producing the given output), computed at prevailing domestic prices, to that same value added computed at “international” prices.

By “international” prices is meant equilibrium prices, without any element of subsidy or tax, explicit or implicit. For example, irrigation water usually is not traded internationally, but in that case the “international” price means a water charge that is “high enough to allow for full recovery of the costs of the irrigation system or is equivalent to the marginal value product of the water.

Clearly, to implement this concept it is necessary to have a list of all the input coefficients—a farm budget—plus estimates of the rate of subsidy or tax on each input. This is a tall order, but it is not impossible, if farm budget studies are available. The algebraic expression for the EPC is as follows:

(16)

where

a_ij is the input-output coefficient, in physical units of input j per physical unit of the output i, from the farm budget, and

pd_j and p*_j are the prevailing domestic and equilibrium prices, respectively, of each input j.

Here the convention for the symbols is that the world price of the output is expressed in dollars (even with the aforementioned adjustments for internal marketing costs), and therefore has to be converted by the exchange rate e, while the input prices at international prices already are expressed in domestic currency units. The reason for this convention on the input side is that normally the “international” prices of inputs are derived by starting from their prevailing domestic prices and correcting for the presence of subsidies or taxes, so the question of the exchange rate does not arise in practice on that side of the calculation.

As regards the output price in the denominator of (16), it may be easier in practice to first convert the border price to domestic currency units and then to add the adjustments for domestic marketing margins.

In economy-wide studies and industrial-sector studies, the established procedures for estimating EPCs involve manipulating input-output tables, in order to be able to cumulate the protection effects through the many rounds of backward linkages represented by the input purchases (a car requires steel, which requires iron, which requires iron ore, which requires electricity, which requires diesel fuel, etc.). But in agriculture, the inputs are relatively few, and usually their protection rates can be measured directly by reference to border prices, so the input-output procedures are not necessary. This circumstance simplifies considerably the calculation of EPCs in agriculture.

For tracing the effects of policies, it is important to take into account protection on inputs as well as on outputs, so the EPC appears to be a valuable tool of analysis. In practice, since the ratio of purchased inputs to gross output is much lower in developing agriculture than in manufacturing (which can be confirmed by a glance at input-output tables), in many cases the EPCs do not differ markedly from the NPCs.

This is an empirical matter which depends in part on the ratio of purchased inputs to output for the products considered. One way to approach the matter is first, to calculate the NPCs for the relevant products, and then to calculate the EPCs for a subsample of the products. If these calculations indicate that the EPCs differ significantly from the NPCs in some cases, then the analyst should go ahead and calculate EPCs for all the products and all the years used in the study. If not, then the NPCs alone may be taken as approximate guides to policy. (Reca, 1980, pp. 48–49, finds comparable values of the NPCs and EPCs for many years in Argentina for wheat, corn, beef, rice, wool and cotton; the one exception is sorghum for some years.)

Another dimension of the calculation of rates of protection is the choice of technology. As the technology (indicated by the a_ijs) differs, the EPC may differ. Therefore some classes of producers may have different effective protection rates than others do. The same observation may apply to regions that have a distinct technology or have different levels of yields.

The NPC also may vary by region because of the variations in marketing costs by location, but regional variations are more likely to occur with effective protection rates, because prevailing technologies of production tend to vary by region. For this reason, it is useful to calculate the EPCs under alternative specifications of the technology of production. Inadvertently or otherwise, policy may be protecting some groups of producers more than others. An important contribution, of policy analysis is to point out how protection rates may vary over groups of producers.

What do NPCs and EPCs say about the comparative advantage of domestic producers in international markets? Not much, because a high rate of protection does not mean that producers would not still produce at a low or negative rate of protection. A high rate of protection is incompatible with competing against imports, or entering export markets. But protection in some cases simply conceals an ability to compete. To make a proper assessment of comparative advantage, it is necessary to distinguish between costs and economic rents, or excess profits. This can be done by estimating cost-based coefficients of competitiveness, such as the domestic resource cost coefficient, or DRC.

4.2.4 The Domestic Resource Cost Coefficient The DRC and Comparative Advantage

The DRC was brought into common use by Bruno (1972) specifically for the purpose of measuring comparative advantage. By definition, it is meant to be applied to goods that are tradeable or potentially tradeable. On the cost side, domestic cost items are distinguished from imported costs. On the output side, net foreign exchange earnings (through exports) or savings (through substitution for imports) are defined by subtracting the costs of imported inputs from the value of the output. The net foreign exchange cost, per unit of output, is then divided into the sum of the domestic costs of production, to give an expression for the cost in “domestic resources” of earning or saving a unit of foreign exchange: the DRC. (See Scandizzo and Bruce, 1980; Gotsch and Brown, 1980; and Reca, 1980, for useful discussions of the DRC.)

If a product's DRC is less than the exchange rate's valuation of a dollar, then the product has a comparative advantage (at the prevailing exchange rate). In other words, a dollar then could be earned (saved) via production of this product more cheaply than it could be purchased on the foreign exchange market.

If the DRCs for two products differ, then the product with the lower DRC has a greater comparative advantage than the other one. This does not necessarily mean the one with the lower DRC should be promoted for export or import substitution, for there may be other products with a still greater comparative advantage.

The terminology of comparative advantage is confusing. Strictly speaking, and in the simplest case, it means precisely what the example above says: that for a given country one product would be more competitive, or more nearly competitive, in international markets than another product because the first one requires fewer domestic resources per unit of foreign exchange earned or saved. But neither of the two may be really competitive. And the country may have other products that are more competitive than either of the first two.

In current usage, comparative advantage sometimes is used to mean absolute advantage, at an equilibrium exchange rate and at undistorted domestic prices. That is, it can be produced in the given country at a lower resource cost than in any other country (adjusting for international transport costs). In the remainder of this discussion the following definitions will be used: 1) a product has a comparative advantage if, at equilibrium domestic prices, including an equilibrium exchange rate, its DRC is equal to or lower than that exchange rate; 2) a product has a “competitive advantage” at the prevailing exchange rate and prevailing prices, whether or not they are in equilibrium, if its DRC is equal to or lower than the prevailing exchange rate.

Under these definitions, a country would gain net earnings of foreign exchange if the product were exported or substituted for imports, under the respective assumptions on prices.

It is important to make use of both concepts. While the prevailing-price DRC may be less attractive theoretically than the equilibrium-price DRC, it is well to remember that it may be a very long time before the equilibrium prices come into force, and in the meantime import and export policies have to be defined. Also, one person's equilibria may not be another person's equilibria: the assumptions used in computing equilibrium prices do not always command universal agreement.

Definitions of the DRC

With the foregoing considerations in mind, a family of DRCs may be defined algebraically. After the definitions and a discussion of pricing conventions, some numerical examples are given.

In the symbols, a. distinction now is made between the input-output (farm budget) coefficients for imported inputs, m_ij, and those that refer to domestic inputs, d_ij (both purchased inputs and inputs of primary resources). The corresponding input prices are pm_j. and pd_j. Those prices and the “border” output price pb_i are taken to mean prevailing prices. All prices are assumed to be expressed in domestic currency. If the output price is measured at the farmgate level, it is assumed to be adjusted to reflect the marketing costs required to place it at the border. Alternatively, it can be taken directly from trade data, as in section 2.3. Equilibrium prices for inputs and outputs are indicated by an asterisk.

As before, the exchange rate is denoted e, and the equilibrium exchange rate now is denoted e . In the price symbol, the farmgate designation is dropped, because the list of inputs now is assumed to include processing and marketing activities up to the appropriate stage in the marketing chain.

The DRC at prevailing prices, or DRC-P, is then

(17)

It can be seen that the numerator is the cost of the purely domestic inputs and resources, and the denominator is the net foreign exchange earnings, or net potential foreign exchange earnings. The numerator is expressed in domestic currency and the denominator in dollars.

Sometimes it is not clear whether an input should be classified as domestic or imported. It may be more clear if the criterion is “importable” rather than “imported.” Some chemical fertilizers may not be imported in some countries, but they always are importable. In agriculture, usually a reasonable procedure is to assume that fertilizers, agrochemicals, machinery, and, in some cases, certain hybrid seeds, are the importable inputs, and that the others are domestic.

The equilibrium DRC, adjusted to allow for an equilibrium exchange rate and at least some equilibrium domestic prices, is

(18)

According to the previous definitions, competitive advantage holds when

and comparative advantage holds when

If the term pb*_i in (18) is given a numerical value by using the border price (c.i.f. or f.o.b.) in domestic curency, then the DRC-E can be computed without any reference to data on domestic output prices. Likewise, if border prices from trade data are used for tradeable inputs (in this case, adjusted for commerce margins to get the input to the farm), then here also the data on domestic prices can be ignored.

In some circumstances, this flexibility in implementing the formula can be quite advantageous. There are cases in which sector-wide farmgate price data simply do not exist (Haiti, for example). There are other cases in which the domestic price conceals excess profits or has other distortions of an unknown extent, and therefore it is not possible even to say by how much those prices should be adjusted to reflect equilibrium values. Hence the cost-based and trade-based character of the DRC can be a distinct advantage in empirical work.

In the contrary case, when the input coefficients for domestic primary resources (land, labor, irrigation) are not very well known, the prevailing price version of the formula can be expressed in terms of only output prices and tradeable inputs:

(21)

where now p_i is the farmgate price, adjusted by marketing margins to move the product to the border, and pb_i is the c.i.f. or f.o.b. border price itself, expressed in domestic prices.

The numerator and denominator of this expression look somewhat similar, but the numerator is the value added earned by domestic factors and inputs (in domestic currency), while the denominator is the net foreign exchange earnings. The practical advantage of variant (21) is that questions of shadow pricing primary resources are avoided.

The similarity of the numerator and the denominator conveys an economic lesson: if there were no price distortions in the economy (P_i = pb*_i = pb _i, and pd_j = pm_j = pm*_j for all tradeable inputs j), then (21) would imply that for all traded outputs, DRC-P = DRC-E = e* = e.

The Prices Used in the DRC Calculations

If the exchange rate is not in equilibrium and an equilibrium-price DRC is being estimated, it must be remembered that a movement to an equilibrium exchange rate will affect the ratio of tradeable inputs' prices to nontradeable inputs' prices. A nominal devaluation usually implies a real devaluation, albeit to a lesser degree: the wage rate and the prices of other nontradeable factor and goods will drop relative to the prices of tradeable goods.

Contrary to some concerns expressed in the literature, this author's experience is that the prices of primary resources (land and labor) are reasonably well defined, given the region and the definition of resource quality. Clearly coffee land will differ in price from nonirrigated annual crop land, which in turn will differ from irrigated land. But within each category, there usually is a consensus on an appropriate annual land rental by district, and that rental more or less reflects opportunity costs.

In concept, in the DRC computation, the price of a primary resource should be its marginal value product, for that is the cost to the domestic economy of using another unit of the resources. If the analyst feels that prevailing resource prices approximately reflect marginal value products, then attempts at shadow pricing of resources may be unnecessary.

Often prevailing prices are used in the cases of land and hired labor, but two exceptions where shadow pricing is important are farm family labor, whose implicit wage rate usually is lower than the going market wage, because its opportunities are more limited (at least within the growing season); and irrigation supplies, which usually are highly subsidized. Therefore, to compute the DRC evaluated at equilibrium prices, it is important to adjust these two resource prices, to reflect better their true cost to the economy. Empirical studies tend to price farm family labor at 50–70 percent of the going daily market wage rate.

If the prices of the major products were to change significantly as, for example, the exchange rate is moved to an equilibrium value, then the annual rental rate for land would be likely to change also. Therefore it must be acknowledged that if only product prices are adjusted, and not resource prices, the equilibrium DRC is only approximated. However, the annual land rental normally forms a small part of production costs, so the implied error is small.

As in the case of the NPC and the EPC, it is necessary to take into account the costs involved in getting the exportable product from the farm to border, and the import-competing product to a stage in the marketing chain that approximates the stage of c.i.f. imports. (That stage may be roughly the provincial wholesaler, which marks the stage immediately preceding the metropolitan wholesale markets.)

A Numerical Example of DRCs

The following example illustrates the application of the DRC formulas and some of the foregoing observations. And it also sets the stage for the subsequent discussion of an alternative to the DRC method.

Suppose a policy option in a Central American country is whether to invest in expanding coffee production for export or poultry production, as a low-cost source of meat, for import substitution. At prevailing domestic prices, the input-output coefficients for the two products are as follows:

Although highly simplified, these examples may not be entirely unrealistic. The imported inputs for the local broiler industry would be yellow corn and various feed supplements and veterinary supplies. With the existing prices and input structures, coffee is a money-losing proposition (although it may not have been in the past), and poultry is viable (just). If the exchange rate is 2.5 pesos to the dollar, then the DRC-Ps are as follows:

By the rule in equation (19), poultry has a competitive advantage and coffee does not. Now suppose that the exchange rate were distorted, and that economists estimate that an equilibrium value would be 5 instead of 2.5. If a devaluation from 2.5 to 5 were to occur, then presumably these product's values (for the same level of output) would move in the same proprotion, as they are tradeable products: from 5 to 10, assuming the country is a price-taker in international markets for these products. And the unit cost of poultry inputs would rise from 4 to 8.

Under these new circumstances, the DRC-Ps (which now are DRC-Es) would remain the same: 3 and 2.5. This situation illustrates the statement of Krueger (1972) that the DRCs are invariant with respect to the exchange rate.

However, a more realistic set of assumptions on devaluation would include the fact that wages would increase with a devaluation. Suppose it is estimated that wages would rise by 33 percent with . the hypothesized change in the exchange rate. Then improved estimates of the DRC-Es would be

Now the DRCs have changed, and it turns out that under an equilibrium exchange rate both products have a comparative advantage (applying equation (20)). If exchange rate policy is expected to return eventually to an equilibrium, then these last results are the better guide to policy.

Krueger's rule is still applicable in the sense of the ratio of the DRCs between products; it remains constant at 1.20 under the devaluation and wage rate change. Comparative advantage inherently is a concept that compares products, so that is the appropriate interpretation of her rule.

However, notice that now coffee's profits per dollar of foreign exchange earned are higher than poultry's:

Coffee profit: 10 - 8 = 2

Poultry profit: 10 - 1.333 - 8 = 0.667

(under the equilibrium scenario). Is the DRC-E truly the preferred indicator of comparative advantage?

4.2.5. The Rule of Economic Profit Margins

The three foregoing scenarios may be defined as follows:

I. Original situation of overvalued exchange rate

II. Devaluation with no wage change

III. Devaluation with the wage change

Under the three scenarios, the profits of the two products (in pesos per dollar of foreign exchange earned or saved) are:

Clearly, in the existing, distorted economic environment poultry output would be expected to expand relative to coffee output. It is equally clear that the. reverse would be true under either scenario II or III.

The DRC has an unambiguous and useful interpretation: it tells how much of an economy's domestic resources are being spent to earn or save a dollar of foreign exchange. If this number is greater than the exchange rate (in domestic currency units per dollar) for a given product, then it is evident that product should not be promoted for export or import substitution. The policy usefulness of the DRC arises precisely because there are many instances when this rule is violated in practice, and pointing it out in objective, quantified terms helps highlight the need for a change in policy.

At the same time, when the DRC indicates the presence of comparative advantage in more than one product, there may be other factors that should enter the deliberations about which product to promote. The profit margin is not necessarily the decisive factor, but it does describe something the DRC doesn't: the magnitude of the economic margin which can be invested in future growth. Perhaps this consideration has led Pearson, Josling and Falcon (1986) to recommend ranking cropping activities by the ratio of their profits per hectare to the land rent per hectare—though that rule may be overly focused on one factor of production.

As a matter of practice, it appears useful to calculate both the DRC and the economic profit margin, both of them under prevailing prices as well as equilibrium prices, to help shed light on the extent to which economic distortions may be present. The policy decisions may take into account information from both calculations, as well as additional concerns such as distributional effects.

As a final note in this section on protection, it may be worthwhile to point out that in a large number of countries protection rates within agriculture show a distinct bias against export products and in favor of import-competing products (see, for example, Colclough, 1985). This tendency presumably leads to a less-than-efficient allocation of domestic resources over products within the sector. The implication for the analysis in this section is that when the measures of protection are calculated, they ought to be calculated not just for import-competing products but for exports as well. The investigator should not be surprised to find some exports with negative rates of protection—protection coefficients less than unity in value—at that same time that the DRC indicates a comparative advantage.