Global trade agreements are complex and multi-faceted. To capture their effects requires a model that includes all production, consumption and trade in the world; distinguishes different regions; and captures the effects of different forms of protection-both tariff and nontariff measures-many of which are designed specifically to obfuscate their effects.

Despite these challenges, global models can be enormously useful. When used with detailed data on trade and protection, models such as the World Bank's LINKAGE model can provide a link between the negotiations on very specific products at the WTO, and the real incomes of households in rich and poor countries. Even quite different models can provide valuable insights into the relative importance of different measures, such as tariffs versus domestic support to agriculture.

Unfortunately, there is much uncertainty about key parameters of these models and, as a consequence, the measured gains from any reform often vary considerably across different models. When considering changes in the results over time, it is usually useful to examine results from similar models, running similar experiments, and making similar assumptions about the sources of gains from trade reform.

When we do this, we find that the measured potential gains from trade reform have declined in recent years. If this were not the case, then the results from these models would be very suspect. There has been an enormous amount of liberalization in recent years, particularly in developing countries that have come to realize the benefits to their economies of lower trade barriers. In addition, very substantial trade reforms have been agreed by countries acceding to the WTO, and especially by China. As well, our ability to measure protection has improved dramatically with the GTAP-6 database, which takes into account the fact that some countries already face low barriers because of tariff preferences.

When we take these factors into account, it is not surprising that the more recent estimates of the welfare gains from total trade reform using the World Bank LINKAGE model are smaller than earlier ones: they declined from $355 billion in our 2002 analysis to the current estimate of $287 billion. This is generated by a model that takes into account only the savings from replacing higher-cost domestically produced goods with lower-cost imported goods - that is to say, a so-called static model.

The gains from the sort of partial reforms being considered under the Doha Agenda negotiations are also much smaller, at $119 billion a year, because this liberalization will be partial, and is based on the GATT bound tariff rates that are often considerably above applied tariff rates. Exceptions for sensitive and special products are likely to further reduce these gains because they permit continuation of the most costly forms of protection.

Any analysis using standard modeling frameworks such as LINKAGE almost always finds larger gains from agricultural trade reform than from reform of trade in manufactured goods, because the much smaller agricultural sector features generally higher and more variable rates of protection, and tariff peaks - which are commonplace for agricultural commodities - are by far the most costly form of tariff.

Other studies that include additional sources of gain frequently obtain larger estimates of welfare gains. For example, a World Bank study from 2002 that reported $832 billion in potential gains from trade reform included potential productivity gains from increased participation in global trade. The recent Carnegie Endowment study generates increases in employment and welfare following trade reform by making urban wages fall relative to producer prices-a linkage which also increases the perceived importance of non-agricultural trade reform relative to agricultural trade reform.

It is clear that our current models of the impacts of global trade reform miss many of the potential benefits from trade reform. They are much too aggregated in their treatment of trade policies. They generally ignore the gains from reforming services trade, where the potential is probably greater than with merchandise trade because of imperfect competition in markets for services, and the gains from new and improved varieties of service inputs. They ignore the efficiency gains that can come from exports, including the gains from reallocating resources to the most efficient firms, which are the ones that typically choose to export. Finally, they ignore the fact that increases in exports typically come from new, rather than traditional, products. Much new theoretical and empirical work is underway to capture these effects, and there are strong grounds for optimism given the widespread acceptance in the analytical community of Professor Tom Hertel's challenge: "If you think it's broken, ... then please help fix it".

Will Martin is the lead economist in the World Bank's Trade and Development Research Group.