Margaret McMillan, Dani Rodrik, and Karen Horn Welch take a look at trade cause celebre of the 1990s: Mozambique's cashew exports. This is important because they are trustworthy--and virtually no one else with a loud opinion on this is.

Unfortunately, I haven't been able to do anything other than skim it yet, but here's the abstract and the introduction:

When Economic Reform Goes Wrong: Cashews in Mozambique

by Margaret McMillan, Dani Rodrik, Karen Horn Welch. NBER Working Paper No. w9117. Issued in August 2002

Abstract:

Mozambique liberalized its cashew sector in the early 1990s in response to pressure from the World Bank. Opponents of the reform have argued that the policy did little to benefit poor cashew farmers while bankrupting factories in urban areas. Using a welfare-theoretic framework, we analyze the available evidence and provide an accounting of the distributional and efficiency consequences of the reform. We estimate that the direct benefits from reducing restrictions on raw cashew exports were of the order $6.6 million annually, or about 0.14% of Mozambique GDP. However, these benefits were largely offset by the costs of unemployment in the urban areas. The net gain to farmers was probably no greater than $5.3 million, or $5.30 per year for the average cashew-growing household. Inadequate attention to economic structure and to political economy seems to account for these disappointing outcomes...

INTRODUCTION

In a case that has become a cause celebre for the anti-globalization movement, the World Bank prevailed on Mozambique’s government in the early 1990s to liberalize the cashew sector and to remove restrictions on exports of raw cashews. The Bank hoped that resources would be allocated more efficiently and the incomes of cashew farmers would be boosted. The policy was met with fierce opposition from the domestic cashew-processing industry, which ironically had just been privatized. After a decade of political strife, international controversy, and ongoing if hesitant reform, the consequences remain hotly contested. Each side in the debate has its favorite statistics: the World Bank points to the rise in farmgate prices, while its opponents point to the processing plants in urban areas which have been shut down and the thousands of workers that remain unemployed.

Historically, the cashew sector has constituted a significant part of Mozambique’s economy, providing income to several million individuals across the country. In the 1960s, Mozambique produced as much as half of the world’s total. The sector went into a long decline thereafter, as a combination of adverse policies and civil war (1982-1992) brought new tree plantings to a halt. Following independence in 1975, the government had banned the export of raw cashew nuts to stimulate domestic processing. Mozambique became the first African country to process cashews on a large scale. By 1980, the country had 14 processing factories. Following World Bank advice, the government began to loosen restrictions on raw cashew production in the late 1980s. The ban on exporting raw cashews was lifted in 1991/92 and replaced with an export quota and export tax. The quota was subsequently removed, and the export tax on raw nuts came down from 60% in 1991/92 to 14% in 1998/99.

From the vantage point of textbook economics, the analysis of the export restriction and its removal is a straightforward exercise. A ban (or tax) on exports depresses the domestic price of raw cashews, effectively subsidizing the domestic processors for whom raw cashews is the chief input. The policy results in an inefficient allocation of resources: raw cashew production is discouraged, and labor and capital are pulled into cashew processing where, absent externalities, their social value marginal product is lower than in other activities. The relaxation of the restrictions is therefore expected to create a double benefit. First, an efficiency gain, arising from the reversal of the adverse resource pulls mentioned above. And second, a distributional gain, resulting from the rise in farmgate prices for the poorest households in Mozambique. This is the sort of analysis that underlies, for example, Paul Krugman’s (2000) New York Times column on the subject, which took the anti-World Bank crowd to task for overlooking the propoor bias of the export liberalization.

As we shall show in this paper, many of the textbook implications of export liberalization were indeed realized. Farmgate prices rose, raw cashew exports increased, and resources were pulled out of cashew processing. However, even under the most favorable assumptions, the magnitude of the benefits generated by these effects were quite small—both in economic terms and in relation to the amount of time and energy that Mozambique’s government spent on this question over the years. We estimate that the efficiency gains generated by the removal of the export restrictions could not have amounted to more than $6.6 million annually, or about 0.14% of Mozambique GDP. The additional income accruing to the farmers was probably no greater than $5.3 million, or $5.30 per year for the average cashew-growing household. These are puny amounts for a policy that was a key plank in the World Bank’s reform agenda, and that became a serious bone of contention between the Bank and Mozambique, requiring the personal attention of both of their presidents.

Moreover, small as they are, these numbers overstate the benefits involved. The standard gains from the liberalization have to be set against the efficiency losses that have resulted from the idling of processing plants. In theory, the workers employed in these plants should have found alternative sources of employment after a reasonable time, perhaps suffering some wage losses in the process. In reality, a large number seems to have remained unemployed, perhaps because of the expectation that the liberalization would be eventually reversed. One account claims that 90% of the sector’s 11,000 workers were unemployed in 2001. Even if we take a fraction of this number, the loss in real output (equivalently, loss of real income of workers) that is involved is of the order of $6.1 million, or 0.12% of GDP. Note that this amount is roughly equivalent to the direct efficiency gain generated by the liberalization (as noted above). In all likelihood, therefore, the aggregate static gains produced by the liberalization were a wash. These disappointing outcomes are due in part to wrinkles that the textbook analysis sets aside. First, there are complications that arise from imperfect market structures. We highlight two of those in the analysis below. Domestically, there are several layers of intermediaries that separate cashew farmers from the export trade, creating a situation analogous to doublemarginalization in the analysis of vertical relationships in industrial organization. The chief implication of this is that we cannot expect increases in export prices to be passed one-for-one on to the farmers. The pass-through coefficient is much smaller than unity—more of the order of 40-50%--reducing the gains that accrue to the poorest households. In other words, traders capture much of the benefits from the liberalization. Externally, we have the complication that the world market for raw cashew is significantly less competitive than that for processed cashew. In effect, India is a monopsony buyer of raw cashew from Mozambique. Mozambique’s transformation from an exporter of processed cashews to an exporter of raw cashews can be expected therefore to produce a terms-of-trade loss for the country, which diminishes both the efficiency and distributional gains from liberalization.

The real hope for the liberalization strategy might, and should, have been placed on the dynamic effects. We emphasize two dynamic consequences in particular. First, the liberalization could have reinvigorated the rural sector over the medium- and long-run by reversing the dramatic collapse in cashew tree planting. Second, in the urban sector it could have heralded a restructuring of production by promoting a more rational investment pattern. The key in both instances was a credible commitment to a new pricing regime—possibly complemented with compensatory programs—that would have made it worthwhile for farmers, entrepreneurs, and workers to undertake investments that would be at least in part irreversible. The main failing of the cashew liberalization policy, in our view, was that it did not send sufficiently credible signals about the pricing regime. The result was that farmers refused to plant trees, cashew processors refused to take their resources elsewhere, and urban workers refused to look for other jobs. Had these adjustments taken place instead, the static losses would have been minimized, while the efficiency gains would have grown over time.

The Mozambique cashew story illustrates several themes that have become increasingly central in the analysis of reform. One theme has to do with the importance of credibility and the need for expectations management. The supply responses that will make reform successful are likely to be forthcoming only when there is sufficient credibility attached to the change in the policy regime. That in turn requires creative thinking on credibility enhancing mechanisms as an integral part of reform. The second theme is that reform is a “political” problem as well as a “technical” one. Had the political opposition of the urban groups been anticipated, or factored in, compensatory mechanisms and side bargains could have been worked out beforehand. Third, policy reform via conditionality is rarely conducive to desirable outcomes. The credibility problems noted above were created in part because the liberalization of the cashew sector was viewed as a “World Bank policy”—something that the government was doing not because it was a priority but because it was required to qualify for World Bank (and IMF) lending. Not having full ownership of the reform, the government was poor at selling it.

There are also implications for the acrimonious turn that the debate on globalization has taken of late. We have little doubt that most of the activists that have attacked the World Bank over its handling of the Mozambique cashew issue have their hearts in the right place. But few have carefully scrutinized the question that one would think they would have been most interested in: are the poorest farmers getting a better price for their product, and are they better off as a result? But the economists are not without blame either. They have relied on a priori generalizations and textbook expositions instead of figuring out what has really transpired on the ground.

Beyond Mozambique are there also implications for the rest of Sub-Saharan Africa (SSA), where 75% of the labor force still depends entirely on agriculture for its livelihood? Recent evidence suggests that—as in Mozambique—the supply response to price liberalization throughout most of SSA has been disappointing (UNCTAD, 1998). Over-reliance on price reforms is likely to be one reason for this. Most policymakers would agree that price and nonprice incentives are both important determinants of supply. In practice however, it is price reforms that have been carried out most often. It is far easier to stop regulating producer prices than it is to remove structural constraints like poor roads, lack of access to credit, or monopsony power on the part of domestic traders. The problem with price reforms is that they can be also reversed with the stroke of a pen. As our analysis of Mozambique suggests, a significant supply response is unlikely unless there is a sharp break in farmers' expectations about the future. Since non-price reforms are harder to reverse, they may be more effective in increasing the expected profitability of investment from the farmers' point of view, thus eliciting the elusive supply response.

The plan of the paper is as follows. Section 2 provides a capsule history of the cashew industry and of recent developments. Section 3 presents an analytical framework and a decomposition of the welfare effects of cashew liberalization into various channels. In section 4, we provide quantitative estimates of the efficiency and distributional implications of the liberalization. Sections 5 and 6 deal with domestic and international market structure complications, respectively. Section 7 focuses on the domestic processing industry and presents estimates of the unemployment loss. Section 8 speculates on the reasons behind the disappointing supply response. Section 9 concludes. A synopsis of the debate surrounding the case is presented in Appendix A.