In this section we identify several methodological, computational and logistical issues that in our opinion deserve special attention at the planning stage of the project. The choice of indicator depends in part on the availability of data, but the data should not drive the design of the study. More important is the intended use of the indicators and the predominance of particular policies in the Tier I countries. The methodological soundness of the framework for collecting and analyzing the policy data is a necessary if not sufficient condition for a convincing outcome. Computational considerations can sometimes influence the choice of indicators and logistical issues will also play a part in study design.
The key methodological issues are well illustrated by the range of studies discussed above. They include the range of commodity indicators to be included; the need to incorporate economy-wide impacts; the issue of inadequate marketing infrastructure and the impact of this on protection measures; the difficulties in adequately measuring government spending; and the problems of assigning public goods where pricing methods are not in place.
The selection of indicators of agricultural incentives
The methodological basis for the majority of the monitoring studies is that there are essentially two types of direct incentive policies widely employed by governments in the agricultural sector: agricultural price interventions and direct government subsidies. Market price support operates directly through price-related interventions of outputs and purchased inputs. This support derives from domestic price interventions (for example minimum-price policy) supported by foreign trade barriers such as tariffs and quantitative restrictions on both imports and exports.
Most of the studies listed above for developing countries have used NPRs and ERPs to capture the domestic price interventions, by direct price comparison between border and domestic farm prices adjusted for margins and quality differences. ERPs are relevant because most countries do have interventions affecting tradable inputs and the share of such inputs in the unit cost is significant for several activities. Analysis based on NPRs only would generally understate true protection, and considering government outlays on tradable inputs misses the effect of trade barriers on import-competing inputs. This analysis has been complemented in some studies by ERAs and PSEs so as to capture government expenditures by direct price comparison. The modelling studies generally used actual tariffs as reported in the WTO, or PSEs in the case of OECD members.
The debate on agricultural support in some regions traditionally focused on the evolution of domestic terms of trade (prices received relative to prices paid) relative to a base period. This is the case for example in transition economies. However, relative prices such as these fail to capture the misalignment of policies and incentives in the base period. Moreover, the cornerstone of agricultural policy today is that prices paid by farmers for inputs and the prices paid for their products should be similar to the real value of those goods to the economy as a whole. That is for products that can be traded internationally, they should pay and receive prices that are close to international prices. For this reason, we suggest the indicators should focus on the effect of prevailing policies in any given year relative to world prices of output and tradable inputs for that year, which is what was used in most of the studies listed above.
Accounting for economy-wide intervention
Previous analysis has shown that for some countries, the so-called Indirect Effects in the have been significant, in fact overwhelming the effect of sector-specific policies on agricultural incentives.17 We believe that the magnitude of these Indirect Effects has probably diminished somewhat during the late 1990s and early 2000s, partly due to the reduction in industrial protection in many countries, and due also to better macroeconomic policies. Thus, for example the magnitude of the exchange rate misalignment could be lower today than it was 10-20 years ago. But the issue remains for several countries. A major issue in the domestic debate is the influence of changes in the real exchange rate as it affects the competitiveness of the tradable agricultural sector. Even in some countries such as Chile, with a low current account deficit (relative to GDP), ample foreign exchange reserves, a floating nominal exchange rate regime and few exchange restrictions, the real rate has experienced considerable variations. The extent of misalignment is becoming a more complex analytical issue today vis-à-vis the 1980s, considering in particular the implications of a more open capital account.
The computation of Indirect Effects for every country might be too complex an undertaking for all countries covered in the FAO project. We suggest that this adjustment be limited to a subset of countries, those with better data and more experience in the analysis of economy-wide effects. However, at the very least, for all countries we would suggest documenting the evolution of the real exchange rate through time, so as to assess the possible impact on private returns in agriculture.18 This would not capture whether there is misalignment as such.
Choosing how to handle the monitoring of exchange rate disequilibria is a difficult decision. On the one hand, there is ample evidence of the strong influence that the exchange regime has had on the competitiveness of the tradable sector of agriculture for many developing countries in the past (Schiff and Valdés, 2002). But the study of the exchange rate phenomenon has become increasingly more complex in a world of more open capital accounts where one would expect changes in the “equilibrium” exchange rate unrelated to relative inflation rates or trade balances. Beyond the case of a clearly unsustainable nominal exchange rate, policy prescriptions to correct misalignment are neither obvious nor unambiguous. Thus, what would be practical guidelines for the whole set of countries? In our view, all countries should report the evolution of the real exchange rate (price of tradables to non-tradables), but the analysis in depth of the nature and magnitude of the exchange rate misalignment using the Krueger/Schiff/Valdés methodology should be restricted to those countries with the necessary data and expertise in this topic. What is practical and simple enough for other countries is open to doubt. The purchasing power parity (PPP) exchange rate is simple to compute, and useful to have, but does not capture exchange rate misalignment.
Structural impediments and marketing margins
The application of the direct price comparison approach raises some methodological issues when distinguishing between explicit trade and price policies and the presence of structural impediments.
One issue is whether to adjust for excessive margins in marketing. A high producer price/border price margin could be due to poor physical and institutional infrastructure, an uncompetitive processing industry, or the high intermediate transaction cost for traders due to erratic policy changes regarding QRs on imports and exports. Although they do not result from explicit government policies on agriculture, these high margins indirectly tax farmers by raising the cost of moving and processing domestic production. This indirect taxation could be interpreted as a policy failure that weakens competitiveness in upstream and downstream activities. Thus, it is important to distinguish between trade and price policies (which can be corrected quickly) and structural flaws in the market (which take longer to correct).19
Measuring government outlays
There are several issues to be considered in the measurement of government outlays on farm programmes. One is what items to include, and the selection depends on the indicators and objective of the FAO programmes. Josling and Tangermann (1989) present a discussion in the context of the PSE measures. If one wants to measure income transfers associated with government expenditures the list of budgetary items can become very long. For example in the study on transition economies (Valdés, 2000), the section on Turkey contains a table with a detailed description and corresponding values of more than 40 items through different government agencies to be included as government expenditures for agriculture during 1993 (Valdés, 2000, Table 5, p. 118).
Another issue is that the data on government expenditures is usually available only at the sector level, not at the commodity level. The allocation of government expenditures to commodity production systems is either arbitrary or requires extensive further analysis. The OECD solution is to have a category of support labelled “general support expenditures” that are not allocated by commodity. Similarly, the WTO monitoring of domestic support included a category for “non-product specific” support. Unless government payments are specifically tied to commodity production, this approach is probably the most satisfactory.
A further consideration arises from using the central government budget as a source for tracking down government expenditures on agriculture presents many challenges. For example in Russia and Brazil, reported federal government outlays do not include state and municipal outlays. This is also likely to be particularly important in other countries with a federal system. Hence it is best to collect information on state outlays in addition to federal outlays in the case of “federal” countries. Municipal outlays on agriculture, however, are probably less important in most countries.
In addition to these problems, actual expenditures (ex-post) could differ from budgeted outlays, though it is the former that is more relevant. But actual expenditures are often more difficult to collect, and available only with a lag. However, in principle one should ask consultants to try to get the actual outlays, instead of simply reporting budgeted outlays.
Finally, government figures on support to agriculture can vary according to which governmental source one used. For example in Russia budgetary support to the credit in kind programmes appeared as a reduction in revenue, rather than an item on expenditure. Furthermore, there were extra-budgetary funds for agriculture from a tax on gross revenues of enterprise in all sectors of the economy. In addition, a portion of government support also takes the form of mutual clearing of obligations. One implication of this is that the figures reported by the Ministry of Agriculture on government support to agriculture differ from figures from the Ministry of Finance and the National Statistical Committee (Valdés, 2000).
Public goods in the absence of pricing
One problem of allocating the benefits of subsidies over agricultural enterprises is that of assigning values for public goods. The case of water rights and the scale of charges for the use of water is an illustration of this problem. Subsidized water charges for irrigation are a common feature in many developing and some developed countries. For example this is the case in India and Pakistan, most of North Africa and the Near East, and many other countries. Adjusting for the implicit subsidy on water charges in ERP estimates has been attempted in a few studies, such as for Egypt (Word Bank Agro Export Strategy Report - 2000) and for India (Purcell and Gulati, World Bank), but it is difficult due to the absence of a reference price due to the lack of an open market for water rights in many countries. The situation is different for countries that moved to tradable water rights (independent of land transactions) such as Australia, Chile and Mexico.
The incidence of income transfers to agriculture
Even when we have data on government outlays, often we are unable to differentiate whether these expenditures really represent income transfer to farmers, or if they in fact capture transfers to input suppliers, to the agro-processing industry, or simply reflect the cost of an excessive bureaucracy. As an illustration, in the case of Turkey, according to official statistics on government outlays for agriculture (and in the OECD report for the country), fertilizer subsidies are listed as part of the transfer to farmers. However, imports of fertilizers (nitrogen) were taxed with a high tariff. Thus, doing the price comparison showed that the subsidy was only a partial compensation for the protection to the domestic fertilizer industry. The standard measure of government outlays overstated the true transfer to farmers.
Commodity coverage of the study
Choosing the appropriate commodity coverage is an important part of the design of the monitoring study. Should it include the major tradables only, or should it be extended to the whole sector? On the one hand we could get data on government outlays for agriculture as a whole, but it is difficult to get such data for specific activities. On the other hand, market related transfers usually cover only a sub-sector - albeit a representative and large component - of the tradable sector of agriculture, and including the non-tradable sector in the analysis of market related transfers would not be relevant.
Drawing the line between tradables and non-tradables can sometimes raise many questions. Is the share of trade in total consumption of the product the right criterion? Or, instead, should one examine the process of price formation in domestic markets? The situation is pretty clear for several products in most countries, but there are sub-sectors where this distinction is fuzzy.
Computational and interpretation issues related to Effective Protection (ERP)
In addition to the task of obtaining the data for the border/domestic price comparison for tradable inputs, the computation of the ERP forces the analyst to consider the relevance and impact on the following issues: the definition of non-tradables, choosing between the Corden and Balassa approaches; the validity of the usual assumption of fixed coefficient issue and the possible substitution between traded inputs; the degree of substitution between traded inputs and primary factors; and the interpretation of ERPs, in terms of either the ranking or relative scale of protection. These points will be discussed in turn.
The difference between the Corden and Balassa methods of defining tradables can make a significant difference in estimating value added. Corden treats non-traded inputs in the same way as primary factor inputs, in other words it includes them in value added. The Balassa method assumes that non-tradable inputs have a zero level of nominal tariff and places them with traded inputs. Most of the empirical work on ERP estimates in developing countries have used the Corden method, which is probably the best solution.
With respect to the degree of substitution between tradable inputs, the usual conclusion is that ERPs calculated from fixed (post-protection) input-output coefficients will bias the estimated ERPs (if σ = 0). The extent of the bias will depend on what can be considered an empirically reasonable range of values for the substitution elasticity (σ) associated with a particular industry. Unfortunately, there are few empirical estimates for agricultural activities in developing countries. The common practice of estimating ERPs has been to assume that the underlying production function is a “fixed proportions function in which the elasticities of substitution are zero. However, in a situation in which input prices are more distorted than product prices the coefficient measuring the proportion of the total cost spent on importable inputs will be understated and the implicit tariff will be underestimated.20
In many cases, traded inputs could substitute primary factors (non-traded inputs), such as for example cereals (import-competing) and forage (non-tradable) in beef and dairy production. A large increase in the price of cereals could induce farmers to expand the area devoted to forage. How this would show up in the calculation of an ERP would depend on whether forage was included as a primary input or not. This in turn depends on the answer to the question: Who are the producers: the owners of the firm, or the owners of the factors of production? The common assumption is that the majority of farmers own their land and the capital attached to the land, and that they provide the bulk of the labour input. In the example above, the forage would be a primary input, and hence be a part of the value added.
The interpretation of the ERP can either be as a comparison across time or countries or across sectors within a country. The second is the more useful. From the point of view of policy evaluation, one of the most useful results in the ERP computation is to obtain a profile of relative effective protection rates across and within sectors. This forces the policy maker to address the question about why some sectors benefit from an ERP that is substantially higher/lower than other activities, depending on how important are imported inputs into the production process and how they are taxed. The only sure way to guarantee against wide variations in rates of effective protection even when nominal rates are relatively low is to make the rate of nominal protection uniform across all products. When all nominal rates are equal, all effective rates are equal to this nominal rate.
We would highlight the following logistical and managerial issues as among those that need to be considered:
There are considerable advantages to following a common methodology across countries. This emphasizes the need to provide clear guidelines on the methodology before starting the country studies. On the other hand, some flexibility must be given to analysts where there are unique circumstances in a particular country that would make too rigid an application of the common method misleading. Such deviations should be documented and approved by the FAO Task Manager.
In our experience, most studies of agricultural policy and protection levels haven taken longer than initially planned. Perhaps the main reason is that the local consultants are over- committed and the project leaders have not been able to put together a team that can proceed within the timing agreed on. This suggests that a realistic timetable be established to prevent frustration and adequate intermediate deadlines and draft reporting schedules be inserted to avoid long periods of low-intensity work by the consultants.
The question as to whom to hire as a local consultant is another critical issue. From our experience, we would be inclined to do these studies outside government agencies and with individuals rather than through a contract with an institution. This is due to uncertainties as to the continued employment of the relevant staff, and to difficulties in identifying a person/team within the government agency that could implement the study in time and with rigor. Individuals in universities, or in think-tanks, or independent consultants are, in most developing countries, in a better position to deliver. However, it is also true that local consultants can be over-committed and their jobs can change.Moreover, how the question of country ownership of this work is dealt with is probably important to consider, and at least in some countries it could imply more direct involvement of a government agency in the data compilation and analysis. Some fall-back position needs to be devised so that data and expertise is not lost through consultant non-performance.
There is also a need to have a small core of researchers either employed at FAO, or reporting directly to FAO. This team would include the Task Manager of the study, a professional who would able to get involved in the substance of the work and in the preparation and revision of the publication of the results, and some research assistance to monitor periodically the work in progress and check all estimates. This would also facilitate the development of a database for the project at the level of the central team, to include the statistics used by all the countries involved and all the relevant information used to compute the Indicators.
Finally, we would like to emphasize the importance of institutional support from FAO, which would be essential for continued effectiveness of such a monitoring programme.
17 The issue is discussed in detail in Schiff and Valdés (2002).
18 For many developing countries the Central Bank computes the real exchange rate, in most cases defined as the ratio of the price of tradables to non-tradables relevant for the country in question. Furthermore, in their country specific analysis both the IMF and the World Bank usually report estimates of the real exchange rate.
19 A relevant example is the case of Mexico, where the observed negative NPRs were largely attributed to "excessively" high domestic marketing margins, the result of uncompetitive structure in transport and local markets (Ch. 15, World Bank, Mexico 2001).
20 A. Valdés (1973), "Trade Policy and its Effects on the External Agricultural Trade in Chile", Amer. J. Agri. Econ. 55(2), see pp. 159.