Agricultural markets in most countries have been the object of considerable government controls and other interventions. For instance, it is well known that in the OECD countries such policies result in annual transfers to farmers in the vicinity of $290 billion, with subsidies of various types making up in some cases 60-80 percent of farmers' revenues. This is why it took so long to bring agriculture under the rules of the General Agreement on Tariffs and Trade (GATT), the international rules framework for merchandise trade of the World Trade Organization (WTO). These interventions have resulted in excess production by many subsidising countries, depressed world prices, and frequent trade disputes. The Uruguay Round (UR) succeeded in finally defining some rules for agricultural trade, quantifying the various trade restrictions, and placing some limits in the use of subsidies, both domestic as well as export related. Despite its success in bringing agriculture within the WTO, the UR, nevertheless, legitimized a variety of remaining agricultural distortions. These include high tariffs, tariff escalation, large trade distorting domestic support, vague rules on what constitutes non-trade distorting support, and considerable export subsidies. The current "Doha Development Agenda" or DDA negotiations of the WTO has again highlighted the reluctance of many countries to place strong and binding limits to their agricultural protectionist and other support policies. Agriculture was one of the reasons that the WTO Ministerial Conference in Cancun in September 2003 failed to produce an agreement.
The "cotton issue" became one of the difficult negotiating issues at the WTO Ministerial Conference in Cancun. It was claimed that cotton subsidies, both domestic and export, granted by some countries, led to artificially depressed world market prices and thus negatively impacted negatively on both export earnings as well as production levels in non-subsidizing countries. While these types of subsidies are not unique to cotton, they became an issue following the submission of a joint proposal at a special session of the WTO negotiations on agriculture on May 16, 2003 (WTO 2003a). This proposal, by four West African Countries (WACs), Benin, Burkina Faso, Chad and Mali, claimed inter alia that the elimination of subsidies to cotton would raise world market prices and make cotton production in the WACs highly profitable. This submission cited recent model-based studies by the International Cotton Advisory Council (ICAC, 2002) and Goreux (2003) to claim that these subsidies led to significant amounts of export earning losses by the four WACs, e.g. $250 million in 2001/02 in the Goreux study. It was further claimed that the combined direct and indirect negative effects would be close to $1 billion per year.
On 4 August 2003, in the run up to the Cancun Conference, the four countries made another submission to the same WTO committee (WTO 2003b), essentially reiterating the same claims, and calling for "the establishment in Cancun of a mechanism to phase out support for cotton production with a view to its total elimination". Other aspects covered in these submissions were the effects of the cotton subsidies on poverty and food insecurity at the farm level and an international mechanism for compensation for the losses. Earlier in 2002, the "cotton issue" hit headlines around the world with the publication by OXFAM of a report called Cultivating Poverty: The Impact of US Cotton Subsidies on Africa (OXFAM 2002), drawing upon the results from the ICAC's model-based estimates (ICAC 2002) for establishing the link between subsidies and world market prices, supplemented with additional estimates on the impact. The results have been widely quoted, and included in statements by high-level political figures.
The case of the four WACs has also received a boost from the analysis of Minot and Daniels (2002). They used household survey data to estimate the direct and indirect effects on incomes and poverty in Benin, due to falls in received cotton prices. Under the assumption that cotton prices to producers are lower by 10 percent due to world price declines (the transmission assumed is perfect), they estimate that the incidence of poverty among Benin's cotton growers goes up by 5 percentage points (from 37 percent to 42 percent), and among all farmers by 2 percentage points (from 40 percent to 42 percent, or 170 000 people more fall below the poverty line). In the longer term, taking into account indirect effects, they estimate that the impact is even larger. Larger world and producer price declines have much larger impacts.
The purpose of this paper is to contribute some fresh analysis on cotton in the above context. As the cotton issue is not only a topic of academic interest to trade policy analysts, but has also become a politically sensitive matter, it is important that there are more studies on cotton so that the issue is debated with sound empirical basis. This study presents some fresh estimates of the likely impact of cotton subsidies on non-subsidizing countries, along with comparisons with all the previous similar studies. A particular contribution that we consider valuable is the detailed discussion of the "building blocks" of the partial-equilibrium modelling framework on which all previous analyses have been based.
There is a long and impressive tradition in the quantification of the impact of agricultural trade distortions on global markets, trade and economies of individual countries, a tradition many attribute to the work of Valdes and Zietz (1980) as the front runner. Many model-based studies were undertaken in the run up to the UR and following the conclusion of the Agreement. Similar works are appearing now in the context of the DDA negotiations, as there is a heightened demand for information on the likely impact of the reform process being negotiated. The literature provides a wide range of different assessments, in large part reflecting different approaches to modelling, assumptions made about parameters and the extent of the reform processes simulated. Quantification is indeed a difficult and imperfect task, irrespective of the analytical tools used. Notwithstanding the shortcomings and differences, various assessments have come to some fairly robust findings as well. Thus, world prices of agricultural commodities are found to increase with trade liberalizing reforms. While the absolute size of the price rise is debatable, there is a fair degree of agreement about price rises in a relative sense. In particular the impact is fairly strong on several temperate-zone food products while it is low or modest in case of tropical products and this is because of the different levels of initial distortions. Similarly, most liberalization analyses show some shift in the location of production, away from subsidizing areas to non-subsidizing countries.
There have been some recent attempts to analyse the impact on world markets and trade of cotton subsidies. ICAC (2002) estimated that 73 percent of the world production of cotton was under some kind of direct assistance. They also estimated that removal of US subsidies alone would have increased world prices in 2000/01 and 2001/02 by around 10 percent. Goreux (2003) in a report used as background for the African Countries' submission to the WTO, used a simple model to analyse the injury to African producers by the subsidies in developed countries, and concluded that world cotton prices would increase by 13-18 percent in the absence of these subsidies. Quirke (2002) estimated that removal of production and export subsidies by the United States and the EU would have increased world prices in 2001/02 by 10.7 percent. Tockarick (2003) found that multilateral trade liberalization in all agricultural products would induce a 2.8 percent increase in world cotton prices. FAPRI (2002) found that under global agricultural trade liberalization the world cotton price would increase over the baseline scenario by 12.7 percent over a ten year period and exports by Africa would increase by 12.6 percent. Finally Sumner (2003), in a report that was used by Brazil in its complaint to the WTO against the United States, used a modified version of the FAPRI model and found that the removal of domestic and export subsidies on cotton by the United States would increase world prices by 12.6 percent and reduce United States exports of cotton by 44 percent. It is clear that, while there is agreement on the overall direction of the price, production and trade changes, there is substantial divergence of empirical estimates with respect to the overall impact of the domestic and export subsidies on the world market as well as exports.
In this paper we discuss these empirical issues and present some fresh estimates of the impacts using the UNCTAD-FAO ATPSM model. Before the ATPSM results are discussed, it is important to understand why various models could generate different results, the source of the key controversy in this area. This is done next in Section 2 where the general framework used by all partial equilibrium models is presented in order to discuss the four key aspects that influence the results. In section 3 we discuss the four key aspects of the empirical analysis as they pertain to the cotton sector. In section 4 we briefly describe the ATPSM model. Section 5 indicates the data and parameters that are used in the simulations. Section 6 presents new estimates of the impacts of the removal of cotton subsidies based on the ATPSM model, including sensitivity tests. The final section summarizes the main conclusions of the analysis.