May the pro-poor impacts of trade liberalisation vanish because of imperfect information?

Abstract

In this paper, we try to evaluate changes in welfare gains and their distribution due to trade liberalisation when imperfect information is considered. The results of two versions of a computable general equilibrium (CGE) model, using the GTAP database and representing goods as well as capital flows, are compared. In the first version, a standard world CGE approach is followed. In the second version, we included risk aversion, imperfect information and production lag in the agricultural sector. After a brief description of the two versions, changes in welfare, represented by the income of two types of household (middle-low and middle-high) in three regions (Europe, United States, Rest of the World) after agricultural trade liberalisation are presented. Theoretical and political consequences of the results are discussed.

Introduction

The global welfare impact of trade liberalisation, including the agricultural sector (Hertel et al., 1999, Hertel and Martin, 2000, Anderson, 2002) stands at the forefront of trade economist's pre-occupations. The development of a generic computable general equilibrium (CGE) model (Hertel, 1997) and large expected spill-overs between economic sectors have generated numbers of studies based on this methodology. These underline the positive effect of trade liberalisation due to efficiency gains. Yet, a growing concern about the impacts of trade liberalisation in developing countries, especially on poverty, has arisen. The consequences of various liberalisation scenarios are now under scrutiny from the viewpoint of equity within these countries (Hertel et al., 2002).

The new round of negotiations has been called the development round, in the hope, among other reasons, that trade liberalisation will help in fighting poverty. However, a particular bottleneck may arise from price instability. The negative impacts of price instability on the poorest are well documented. As consumers, they often spend more than half of their expenditures on food, which makes them sensitive to any price increase. With price instability and risk averse producers, supply is reduced at mean price equivalent. Thus, price instability increases mean prices, harming the poor.

Indeed, since the Roosevelt era, in the 1930's, following this simple reasoning, governments have tried to isolate their markets from world food price fluctuations through trade policies. Recently, stabilisation has been recommended to fight poverty, on the ground that risk limits producer's investments and prevents them—especially, the poorest—using more efficient technology (Timmer, 2000). In this context, attempts have been made to include price instability and its impact on the poorest in trade liberalisation analysis (Hertel et al., 2001). But how can such instability be accounted for in models, which by construction assume that prices are perfectly known? The key underlying issue, here, is where instability comes from.

Most of the time, price instability is only considered as a consequence of external shocks like climatic disturbances. In such a case, as demonstrated by Bale and Lutz (1979) and evaluated by Tyer and Anderson (1992), removing trade barriers stabilises world price, because of the ‘Law of large numbers’: many independent small shocks in various directions cancel out when pooled in one large market. Thus, including price instability in the model should improve the pro-poor impacts of trade liberalisation (Hertel et al., 2001).

However, random events may not possess the nice properties of ‘Gaussian’ perturbations (Mandelbrot, 1971). Moreover, weather and other ‘small’ random events are not the only sources of price instability (Roll, 1984). A part, at least, of price instability in commodity markets is due to market behaviour itself. Such a situation arises with imperfect information (Kindleberger, 1996, Chavas and Holt, 1991). Ezekiel (1938) stressed the importance of price expectations in the price formation process. He showed that markets may tend to fail if demand is rigid and supply elastic, with huge fluctuations, panics and crashes. Several later authors, in the tradition of business cycle analysis, have shown that endogenous price fluctuations may be generated by models including liquidity constraint, risk and relatively rigid demand curves (Boussard, 1996, Day, 1999, Rosser, 2000).

When the source of fluctuations is non-Gaussian random events, or deterministic chaotic disturbances, conclusions derived from standard insurance analysis are no longer valid. For that reason, Stiglitz (2000) reminds us that the market economy is subject to large fluctuations and that public regulations are required, as demonstrated by the recurrent currency crisis in the 1990s. In this paper, imperfect information and expectations are introduced into a standard CGE model that includes a rich and a poor household in each region. Results from two versions of the model (that is, the standard model versus the model with imperfect information assumptions) are analysed and compared.