7 SUMMARY AND CONCLUSIONS

One of the issues that continue to attract considerable attention in terms of analysis and debate is the impact of agricultural trade distortions on global markets, trade, and on the economies of individual countries. As was the case during 1994-96 in the context of the Uruguay Round, several analyses are appearing now in the context of the Doha Development Agenda as there is a heightened demand for information on the likely impact of the reform process being negotiated.

The purpose of this paper was to contribute some fresh analysis on cotton within the above context, namely likely effects of distortions (protection and support) in the world cotton market on non-subsidizing countries. The analytical and modelling framework used here can be applied to any commodity whose market is distorted. In the process of contributing fresh analysis, the paper also reviewed extensively the assumptions behind and results from some recent model-based studies on cotton. This comparative assessment is considered to be particularly useful since there is a fair amount of confusion about the nature and size of the impact of the distortions as models used tend to differ in terms of approach, specification, parameters and assumptions on distortions.

The two main policy parameters reflecting distortions in world cotton markets are tariffs and domestic subsidies. An analysis of recent import policies of all major cotton importing countries showed that import tariffs are negligible. As regards domestic subsidies, the results are based on the official subsidy levels (the trade-distorting or AMS component) as notified to the WTO. The paper also shows in an Annex the corresponding simulation results based on subsidies as compiled by the ICAC (the same tariff protection applied to both scenarios). While there are some other small differences, the ICAC estimates show very high levels of cotton subsidies in China. We reproduce in the Annex the ATPSM results based on the ICAC subsidy data being fully aware of the official Chinese position concerning cotton subsidies, namely that there are no cotton subsidies in China since 2000, because one major objective of this study is to compare and review various recent studies that have been the source of the debate and controversy. Most of these studies assumed ICAC subsidies in their analyses.

The two key impact indicators that the ATPSM (and other models) generate endogenously are new world market prices and quantities produced, consumed and traded - all other indicators like export earnings and welfare measures follow from these. Concerning the impact on world market price, the ATPSM results showed that the complete elimination of distortions (mostly subsidies) led world cotton market price to rise by 3.1 percent under the base scenario and as much as 4.8 percent under different assumptions about supply and demand elasticities, compared with a range of 3-15 percent impact in various other studies. Although it is difficult to reconcile all the differences, as published studies often lack full details about assumptions etc, the ATPSM results are not out of line, as we identified differences in the results between some other studies that we could review in detail and this study, as largely due to differences in parameter assumptions. Our sensitivity tests verified that many of the results of other studies fall within the range of values estimated in these sensitivity runs, when differences in initial subsidy levels and supply and demand elasticities are taken into account. The vastly different assumption about cotton subsidies in China was a major factor.

In the ATPSM, new equilibrium production and trade outcomes are determined simultaneously based on changes in world market prices and long term price elasticities of supply and demand. Overall, the simulated effects on production and trade are consistent with expectations, and supply-demand response parameters. Thus, production and export shrink in countries that eliminate subsidies (e.g. Brazil, EU and US). The impact is mainly felt by the subsidizing countries and largely in proportion to the initial level of subsidy. At the global level, cotton production falls by only 2 percent, as outputs in non-subsidizing countries do not rise enough to offset the large declines in subsidizing countries.

Other results discussed in the paper follow from these key indicators. Thus, for example, all non-subsidizing countries experience gains in export earnings and in producer surpluses (PSs). All subsidizing countries lose in PS terms but some of them gain overall in total welfare term as government expenditures incurred on subsidies initially are saved. These are fairly standards results.

One important outcome from models of this type that is often overlooked is that even where the impact on the world market price is fairly small, e.g. in this study, there could be substantial shifts in production and trade. Thus, despite only a 3.1 percent increase in world price, following complete elimination of all subsidies and other trade controls, the outputs in the United States and EU fall by 15 and 32 percent respectively. Similarly, the impact on the exports and export earnings of the four West Africa Countries can be considerable under all liberalization scenarios, especially under assumptions of higher supply elasticities.

Similarly, one of the other largely overlooked facts is that, following liberalization, it is not the case that only 1-2 countries make gains of impressive size. What happens, and this is realistic, is that many cotton producers gain, and this happens in proportion to their ability to expand production at given world prices. In other words the benefits from liberalization depend a lot on supply responses. While these effects were explored in our sensitivity tests, they mask the true importance of supply response effects, because in the sensitivity tests these parameters were changed for all major producing and consuming countries by the same proportion. If the supply elasticities are raised selectively, for example only for some countries in West Africa, then the results would show that these countries would expand cotton production much more than shown in the sensitivity results.^[30]

The results presented above refer only to long run impacts. The world cotton market, much like the market of other agricultural products, exhibits considerable short term (namely year to year, as well as within year) instability, in both prices, as well as quantities produced, and traded. These are usually much larger than the long run effects outlined above. One should not confuse, however, the short term effects, with longer term and more permanent impacts. As noted earlier in the paper, there maybe permanent shifts in supply and demand following short term instability, for instance through induced technical innovation, or induced changes in demand for cotton in the production of fibres. There is very little known about these potentially important effects, but given that the degree of short term world cotton price instability is much larger than the secular trend, these effects could be important.

Another implication that is not treated here is the potential economy wide impacts of lower prices. Given that cotton is an important commodity for some poor countries, notably the four West Africa countries that submitted the WTO complaint, the policy induced depression of world cotton prices has longer term implications for poverty and growth in such countries. As noted by Minot and Daniels (2002), these impacts can be large, even with small long term world price changes.

Before concluding, it is worthwhile to highlight some issues and caveats. As discussed in detail in Section 3, in the context of the "building blocks" of a model of this type, the initial level of subsidies is crucial in ascertaining the impact of any changes. Hence it is essential that the analysts first, and then the WTO negotiators, agree on the total level of current cotton subsidies, including the amounts that are coupled or decoupled. One of the sources of the current differences in assessed impacts has been the overall level of subsidies, with the official WTO notifications differing substantially for some countries from those compiled by the ICAC and used extensively in many empirical analyses.

Secondly, there is the issue of handling coupled versus decoupled subsidies. Resolving this aspect particularly for the EU and US subsidies is important for assessing impacts as well as for the WTO negotiations. On this issue, the modelling approach used by Sumner and being experimented on by FAPRI is encouraging, but more empirical research on both sides of the Atlantic is needed.

A third issue concerns the values of price elasticities of supply and demand. It became clear in the course of our sensitivity tests that a very large part of the impact analysis depends on these values. Unfortunately there has been not enough work to estimate these values empirically, and this will likely remain a source of differences, unless resolved through further research.

In our analysis we assumed that all international price signals are fully transmitted domestically. While this appears reasonable in the case of cotton, and in a longer-run context, it is by no means certain that the assumption is a correct one. There does not appear to exist any empirical analysis of this issue.

While it is unlikely that models used by different agencies and individuals, and the parameters and assumptions made, will ever be identical, many of the above areas of disagreement and confusion can be resolved to some extent. One possible way to achieve this would be for the authors of the various studies to meet and discuss the models, assumptions, results etc.

Finally, although cotton is not an exception as a commodity that receives high levels of subsidies in some countries at the cost of non-subsidizing countries, the "cotton issue" became a controversial subject in the run up to the WTO Cancun Conference and continues to be so. With the WTO negotiations expected to resume in the middle of 2004, there is a need for more analysis of the issue in order for these negotiations to proceed on the basis of facts and figures rather than conjectures. The fact that this is a sensitive issue makes sound empirical analysis all the more important. That was the main purpose of this study and hopefully it is a contribution to the small amount of literature that exists concerning model-based assessments of the various distortions in the world cotton market.