INTRODUCTION

It is of particular importance for both government decision-makers and industry leaders to have a good understanding about the future direction and development of the world sugar market. Sugar is one of the most important commodities produced and traded in world agricultural markets. The industry has been expanding in production and processing over the decades. Up to 25 percent of sugar produced is physically traded in the global area. The importance of sugar has prompted many attempts at modelling the world sugar economy. The models have varied widely in their specifications and goals. For instance, some models have aggregated the world into regions, while others have chosen specific countries as representative, and certain models have concentrated on trade flows, while others have examined supply and demand or looked at price behaviour. This study represents the first phase of an FAO project to build a new global sugar model.

MODEL SPECIFICATION

The FAO model builds upon the strengths of models developed to date, but places a heightened emphasis on demand factors through the use of a habit function, i.e. where current demand is related to previous levels of consumption. In addition, the analysis was strengthened by field visits to the major producing and consuming countries in the Asia and Pacific region. Hence, the models are estimated using a richer data set which gives more reliable and consistent estimates of the elasticities. The study covers Australia, China, Fiji, India, Indonesia, Japan, Pakistan, Philippines and Thailand. As a group these countries accounted for about 40 percent of world sugar production, 30 percent of consumption, 32 percent of imports and 18 percent of global sugar exports in 1996.

The importance of habit

To illustrate the importance of habit, we have recently examined a new extensive set of data on China where people living in the Southeast coastal region prefer sweeter foods while those in the Northern provinces like savoury foods. This is reflected in the different sugar consumption levels which have existed for a long time between northern and southern regions. This type of consistent divergence in sugar consumption levels, largely independent of income, can also be found in cross-country comparisons. Many Asian countries experienced rapid economic growth over the past decades, but the increase in per caput sugar consumption has been much less than the income growth in some of these countries. For example, in Indonesia per caput income in real terms nearly doubled during 1983-1995, but per caput sugar consumption increased only by 23 percent from 11 to 13.5 kg during the same period. In contrast, although the real per caput income in Pakistan increased by about 30 percent from 1983 to 1995, per caput sugar consumption nearly doubled from 11 to 20 kg in the same period. Japan has the highest per caput income in Asia but its per caput sugar consumption was significantly lower than many developing countries in this region.

The habit formation model reflects past consumption patterns when forecasting consumption levels. The model is estimated for each country separately. For comparison, another model, which excludes habit formation, is also estimated for each country using the same data set. Since the factors driving demand for industrial use may differ from those affecting direct consumption, the demand for direct consumption and that for industrial use were estimated separately, when data were available. For example, in the case of India, the consumption of gur and khandsari accounts for a large share of sweetener consumption and is closely related to both sugar consumption and production. Therefore, the demand for gur and khandsari were also estimated, separately.

In general, the data fit the models well. The hypothesis that consumption habits had no impact on the demand for sugar was tested using the likelihood ratio test, and was strongly rejected in all the country models. This supports the view that habits will also play a role in the future consumption of sugar too.

The price elasticities were in the range of -0.1 to -0.45 for all countries except for Japan and India. The price elasticity in Japan was high, about -0.8 which implies that the demand for sugar was more sensitive to a change in the sugar price. The competition from the highly developed alternative sweeteners industry (HFCS) and the dominant industrial use of sugar are probably responsible for the higher sugar price elasticity. The price elasticity in India is also relatively high, -0.6 which may be attributed to the existing alternative low quality sweeteners market which would induce consumers to shift from white sugar to gur and khandsari if the sugar price increases.

It has been a widely accepted view that income was the key factor in determining future consumption levels in developing countries. As per caput sugar consumption levels were low, future sugar demand in these countries would be expected to increase rapidly with economic growth. However, the empirical analysis provided a somewhat different insight. If the habit effect was taken into account, the impact of income on sugar consumption was very small suggesting that changes in income may not influence sugar consumption a great deal, particularly in the short-run. For instance, the income elasticities for China, Indonesia and Philippines were only 0.03, 0.03 and 0.06, respectively, which suggest that the impact of changes in income is not significant short-run in these countries. The short-run income elasticities for other countries based on the habit formation model were also relatively small, ranging from 0.1 to 0.45, excepting for Pakistan, which has an income elasticity of nearly 0.9. An explanation for the relatively high income elasticity in Pakistan would be that consumer preference for sugar was high, but severely constrained by income. Thus, a change in income would induce significant changes in sugar consumption.

Another interesting finding was that the estimated income elasticity for Australia was negative implying that sugar consumption would decline as income increased. Current high per caput consumption levels and increasing health concerns would appear to be the main factors behind this observation. The industry had recently initiated a series of nutritional campaigns in Queensland to try to reverse the trend.

The income elasticity of demand for industrial sugar use was found generally to be larger than that for direct household use. For example, in China the elasticity was about 0.6 or twenty times that for direct use. The income elasticity for gur and khandsari in India was found to be negative, which suggested that the demand for these sweeteners would decline as incomes rose. Gur and khandsari are considered to be inferior goods.

The overstated income impact on sugar consumption in the past may be largely be attributed to the misspecifications of the empirical model in many analyses of the sugar demand. Comparing the habit formation model and the demand model without habit formation, it was found that income elasticities were much higher in the latter than in the habit formation model. The habit effect on sugar consumption varied from country to country. For instance, it was very strong in China but weaker in Fiji, Japan and Australia, where high per caput consumption already existed.

Projections

Under the assumption that the current policy regimes includes all trade restrictions on sugar imports and exports, and various domestic controls over sugar production and marketing would remain unchanged; the estimated demand models were used to forecast sugar consumption for each country from 1997 to 2005. The population and GNP data used for forecasting is available in the FAO "Compendium of Demographic and Macro-Economic Assumptions". It was assumed that prices will change at the same annual growth rates as experienced during the last five years. Since the model specification may have significant implications for projecting, for comparison purposes, both a habit formation and a standard model were used. Moreover, given that recent developments in sugar use and government policies toward sugar consumption vary from country to country, particular scenarios were designed for the analysis of each country.

Australia. Per caput consumption has been declining steadily over the past two decades. Per caput consumption was about 57 kg in 1976, 50 kg in 1986 and 47 kg in 1991. The decline in per caput sugar consumption reflects changes in consumer preference, due to health concerns, and increasing competition from other sweeteners. Sugar consumption has recovered somewhat since 1992 because of the deregulation of the sugar refining industry, which increased price competition among domestic refiners, and the sugar nutrition campaign. The FAO model projects a decline in per caput consumption in the next few years, given a very high current consumption level and a negative income effect. It is projected that per caput consumption may be about 49 kg by 2005, about 4 percent lower than the current consumption level of 51 kg. However, total consumption may increase by about 5 percent by 2005, as compared with 1996, due to an increase in population.

China. Per caput direct sugar consumption is projected to grow in the future at a slower rate than in the past, reaching 2.9 kg by 2005, while the increase in industrial sugar use is projected to reach about 5 kg per person by 2005, 43 percent above the current level. Total sugar consumption is projected to increase significantly in the next decade to reach nearly 11 million tonnes by 2005, about 50 percent higher than the actual 1996 consumption level. As other sweeteners, especially saccharin, are important substitutes for sugar for industrial use and about 6 000 tonnes of saccharin which are equivalent to about 2.5 to 3.0 million tonnes of refined sugar have been used in recent years, an alternative scenario has been designed to investigate the impact of changes in saccharin prices on sugar consumption. It was found that if the price of saccharin increased by 15 percent annually (the annual increase in the saccharin price was only about 5 percent over the last five years) but the sugar price remained unchanged at the current level, the industrial demand for sweeteners would shift from saccharin to sugar. If this happened, sugar demand for industrial use would reach 10.7 million tonnes by 2005, about 160 percent higher than the current level and total sugar consumption would amount to 14.8 million tonnes by 2005.

For comparison, the general sugar demand model which results in a much higher income elasticity than the habit formation model was used to forecast consumption in scenario 3. It was found that total sugar consumption would reach 15.6 million tonnes by 2005, about 100 percent more than the actual national consumption in 1996 and about 43 percent higher than the projected consumption based on the habit formation model. The substantially higher consumption projection was mainly due to the dominant income effect estimated from the general model.

Fiji. It is expected that per caput sugar consumption would grow at a very slow pace. By 2005, per caput consumption would be about 43 kg, about 7 percent higher than the current level, which is consistent with the development of actual sugar consumption over the past few years. As per caput sugar consumption has already reached a very high level, amounting to about 40 kg in 1996, any further increase in consumption would be diminutive. Since habit formation had little effect on the demand for sugar, the projected sugar consumption from the tow models are nearly equivalent.

India. Per caput sugar consumption has been increasing at a relatively slow rate over the past decade. It was slightly more than 14 kg in 1996, about 18 percent higher than the 1987 level. The slow growth rate was largely attributed to the low industrial demand for sugar, a large quantity of low priced alternative sweeteners (gur and khandsari) consumed and tight control over sugar marketing , distribution and retailing by the government. It is projected that the per caput sugar consumption will increase at a slightly faster rate in the next decade as the industrial demand for sugar picks up and economic reforms give markets a greater role. Under the assumption that the annual income growth rate is 4.4 percent, it is projected that per caput sugar consumption in India would reach about 18 kg by 2005, about 25 percent higher than the current level. At the same time, per caput consumption of gur and khandsari will decline slightly during the same period. Total sugar consumption will be around 20 million tonnes by 2005, about 50 percent more than in 1995.

When habit formation was excluded, the projected per caput consumption was about 21 kg by 2005, which was 21 percent more than that projected quantity using the habit formation model. Correspondingly, the total consumption would be 23 million tonnes by 2005, about 15 percent higher than that based on the habit formation model.

Indonesia. During the past decade, per caput sugar consumption rose from 11 kg in 1985 to 14 kg in 1996, while per caput income more than doubled. Projections based on the habit formation model suggested that this relative slow growth trend would continue. It is projected that per caput sugar consumption would increase over the next decade to reach 16.5 kg by 2005, about 17 percent higher than in 1996. Total consumption would be around 3.5 million tonnes by 2005.

When habit formation is excluded, per caput sugar consumption is projected to increase substantially. It would reach more than 24 kg by 2005, about 70 percent higher than the actual consumption level in 1996 and nearly 50 percent higher than projected consumption based on the habit formation model.

Japan. The increasing use of alternative natural sweeteners, in particular low-calorie sweeteners which are preferred by consumers has resulted in the decline in sugar consumption over the past decade. The projections suggest that this trend would continue but at a much slower rate because of the higher marginal cost of increasing production of alternative sweeteners. It is projected that per caput sugar consumption would be 18.5 kg by 2000 and 17.8 kg by 2005 if the current price structure remained unchanged. Thus, total sugar consumption would be 2.34 and 2.26 million tonnes by 2000 and 2005, respectively.

Excluding habit from the model does not affect the result industrial use accounts for most of the sugar consumption in Japan.

Pakistan. Per caput sugar consumption has increased significantly in recent decades with economic development. If the trend continued, it is projected that per caput consumption would be above 28 kg, and total consumption would reach 5.2 million tonnes by 2005. Excluding habit formation it was projected that per caput and total consumption would reach 35 kg and 6.8 million tonnes respectively by 2005.

Philippines. It is projected that the increasing trend of sugar consumption would continue with its high economic growth rate. The projected per caput and total consumption would be about 35 kg and 3 million tonnes respectively by 2005. While per caput consumption would be about 30 percent more than that in 1996, total consumption would be 50 percent higher, due to the increase in population.

Comparing these projections using the model without habit formation gave a projected per caput consumption at 43 kg or about 50 percent higher than the current level by 2005.

Thailand. Economic growth and the expansion of pharmaceutical, soft drinks and bakery industries have significantly contributed to the increase in sugar consumption over the past decades. Under the habit formation model the growth of sugar consumption is projected to remain strong during the next decade. Per caput sugar consumption would rise by about 35 percent over the 1996 level to reach 35 kg by 2005. Correspondingly, total consumption would rise to 2.3 million tonnes from the 1996 level of about 1.6 million tonnes.

By contrast, excluding habit formation, per caput consumption would be 10 percent higher than the habit formation model, amounting to 39 kg by 2005.

The dynamic supply model for sugarcane and sugarbeet

Separate area yield equations derived from a dynamic sugar supply model have been estimated for each country (the details of model derivations and specifications are summarized in Appendix B). Since the factors affecting production decisions of producers vary from country to country, the general specifications for area and yield have been modified to reflect different policy regimes and production practices in these countries. For instance, to capture the impact of the liberalization of the entire sugar industry in China in 1992, a special variable was introduced in the sugarcane and sugar beet equations of that country. Similarly, since the sugar industry in the Philippines experienced significant changes during the early-eighties, a special variable was included in the area equation to take account of the consequence of these changes.

Other important policy instruments used by governments to manage sugar production were also taken into account in the model specification and estimation. For example, government purchase prices were used as the farmer-received prices for sugarcane and sugarbeet production in China before 1992 when all sugar processing factories were run by the central government (Ministry of Light Industry), but the free market prices of sugar (crops) were used after 1992. As growers in India would receive either the government minimum price or state purchasing price each year depending on the market situation, a selection procedure was used to construct the farmer received prices by comparing the government minimum price and the state purchasing-price. The switch regression method was used to estimate the model. The expansion of the sugar industry in Australia was regulated by government before 1992 but the assigned planting areas were never binding because the actual planting areas were far below the assigned areas. Thus, the model does not directly take account of the area assignment policy but only the impact of the 1992 deregulation by including a policy change dummy variable. Moreover, when a country produces both sugarcane and sugarbeet, the area and yield equations for these two crops were separately estimated because they were produced in different regions and in competition with different crops.

Data and estimation

A sugar production data set collected mainly from national statistics was used to estimate both area and yield equations. In addition to price data, including prices of sugarcane and competing crops and various input prices, a weather variable was constructed and used in estimating the yield equation for each country because a large proportion of production was in areas vulnerable to weather conditions. The weather variable was constructed based on FAO weather data sets which contain monthly average temperature, rainfall and other indicators from the records of the major national meteorological stations. Since precipitation had an important effect on sugar yield (both quantity and quality), the deviation of precipitation from the historic average in the major producing regions during the planting and growing periods was constructed as the proxy for weather conditions. When such detailed precipitation data was not available, a weather dummy was included based on country reports about the weather impact on sugar crops - good, bad or normal. As technology has played an important role in sugar production, a time trend was included in the yield equation as a proxy for technical change.

Empirical results

In general, the data fit the model well except for the yield equation for a few countries, for example, Fiji. In general, the production responses to prices in the short run were not very strong, ranging from 0.2 to 0.4 for most countries, which may reflect the difficulties facing sugar farmers in changing their production decisions, given the existing institutional and economic constraints. In particular, the supply elasticity in India was estimated to be only about 0.01 which suggested that the change in price had little impact on production. The substitute price elasticities of other crops for sugar crops ranged from -0.1 to -0.3. The extremely low substitute price elasticity in Fiji and Indonesia, only -0.01 may reflect fewer production alternatives and difficulties to shift from sugar to other crops facing sugarcane farmers in the short run.

The estimated results from yield equations suggested that technical progress had played an important role in the yield levels as the coefficient associated with the technical change variables in the yield equations for most countries were statistically significant. The changes in farm input prices were also found to affect yield levels. However, when weather had the dominant effect on the yield level, the effect of changes in input prices on yield became statistically insignificant.

Projections

The estimated equations were used to project sugar production for each country. The basic assumptions for conducting forecasts were unchanged - policy regimes, normal weather, and the same growth rates for all prices as the average annual rates over the last five years. It was also assumed that the conversion ratio from sugarcane or sugarbeet to raw sugar would be the same as the average rate for the last five years. Since a certain proportion of sugarcane was also consumed in other fora or used to produce other types of sweeteners, for example gur and khandsari in India, this assumption implied that the structure of sugarcane use would remain the same as that during the last five years. To analyze how the changes in government policy would affect sugar production, various different scenarios were also designed for the countries concerned for the model simulations.

Australia. High productivity and cost efficiency from large scale operation have contributed and would continue to contribute to the expansion of sugar production in the next decade. The deregulation of the sugar industry since 1991 had also been given the industry greater impetus to compete with exporting countries in the world market. It is projected that planting area and yields would increase by more than 20 percent, from 400 000 ha in 1996 to 490 000 by 2005. Yields are expected to increase by about 10 percent, from 100 tonnes in 1996 to 110 tonnes by 2005. Both the expansion in area and increase in yields would result in higher sugar production. It is projected that sugar production would reach 6.6 million tonnes by 2000 and 7.5 million tonnes by 2005. As sugar production is largely driven by world market price, it was found that the total production would reach 8.6 million tonnes by 2005 if the world sugar price increased by a 5 percent annual rate rather than 3.7 percent assumed in the base projections. This fact suggested an expansion potential and sensitivity to the world market price of the sugar industry in Australia.

China. Production has been volatile over the past few years. Although climate and other social-economic factors have an important impact on production, government policies toward sugar and other agricultural commodities also have a profound impact on annual production levels. Since the government still controls prices of grains, the major crops competing with sugar, changes in grains prices would induce a reallocation of resources, especially land, between sugar and grain crops. To analyze how the government policy will affect future sugar production, three different policy scenarios were designed: (a) no changes in the relative price relationship between sugar crops and other competing crops; (b) 5 percent higher prices per annum for grains; and (c) 5 percent lower prices for grains per annum.

Projections indicated that the production of sugar crops were very sensitive to changes in relative prices. Both sugarcane and sugarbeet planted areas would decline from current levels in scenarios (1) and (2). The sugarcane areas would be about 24 and 63 percent lower, respectively for scenarios (1) and (2) than the 1996 level by 2005, while the sugarbeet area would decline by about 30 and 50 percent by 2005 compared with the 1996 level. However, if the annual growth rate of grains prices was 5 percent lower than their average over the last five years, both sugarcane and sugarbeet area would increase to reach 1.4 million ha and 0.8 million ha by 2005, about 29 percent and 14 percent higher, respectively, than the current levels, because the decline in grains prices would make it relatively profitable to produce sugar crops.

Yields of both sugarcane and sugarbeet are expected to continue their increasing trends. It is projected that sugarcane and sugarbeet yield would increase by 25 percent and 33 percent, respectively by 2005, compared to their current levels. As the projected area differed greatly among the three different scenarios, the projected sugar production levels also varied with these scenarios. If the current policy regime remained unchanged (scenario 1), the total sugar output would amount to 6.6 million tonnes by 2005, about the same as the average level for 1993 to 1996 as the impact of the decline in area would largely be offset by increases in yield. If the government carried out a price policy which favoured the grain sector (scenario 2), sugar output would only be about 3.6 million tonnes by 2005, about half the present level. In contrast, sugar output could reach 8.4 million tonnes by 2000 and 11.3 million tonnes by 2005 if the government’s grains purchase prices (support prices) were 5 percent lower annually (scenario 3). These simulation results reveal the prominent role of prices in the future development of the sugar industry in China.

Fiji. The decline in yields of sugarcane over the past decade has been attributed to two major factors. As the continuing expansion of planted area pushed production into marginal land, low soil fertility and difficult undulating terrain resulted in lower yields. Moreover, a large proportion of sugarcane farmers operate on a very small scale leading to inefficient production and lower yields. It is expected that given current policies, the downward trend in yields would persist as the expansion in area continued. It is projected that the planted area would expand to about 98 000 ha by 2005, about 19 percent more than the current level while average yields would decline by about 8 percent to 48 tonnes per ha by 2005. Sugar production, however, is expected to increase by about 13 percent from the current level to 0.6 million tonnes by 2005. Given the great potential to increase yield through better farming practice, total sugar production would reach 0.6 million tonnes or higher by 2005 without any expansion of area.

India. Both the increase in yields and expansion of sugarcane areas have contributed to the significant increase in sugarcane production over the past decade. The sugarcane output in 1996 was about 50 percent more than in 1986. Based on the forecast made, this trend would continue. It is projected that sugarcane yields and planting areas would reach 79 tonnes per ha and 4.46 million ha respectively by 2005, a 40 percent increase in sugarcane production. Sugarcane has been used for producing both sugar and low quality sweeteners such as gur and khandsari. To project sugar production, it is assumed that production of gur and khandsari would always be equal to their projected consumption levels. Therefore, sugar production could be derived from the residual of the total output, deducting the amount of sugarcane used to produce gur and khandsari. It is projected that the total sugar production would be 19 million tonnes by 2005. An increase in the use of fertiliser would bring about higher yields and hence, the total production level. Given that the projected planting areas would remain unchanged, it was found that the total sugar production would reach 20 million tonnes if there would be a 5 percent decline in the price of fertilizer annually, and a 2 percent increase in sugarcane conversion rate due to quality improvement.

Indonesia. The expansion of planted area has contributed largely to the increase in sugarcane production over the past decades. The current analysis predicts a continuing area expansion in the next few years. It is projected that the total sugarcane areas would increase from 400 000 ha in 1996 to 490 000 ha by 2005 while sugarcane yields would reach 78 tonnes per ha. As both area and yields would increase, it is projected that the total sugarcane production would reach 3.1 million tonnes by 2005. As an alternative scenario, it is projected that yields would be 82 tonnes per ha by 2005 if the fertilizer price would be reduced by 5 percent annually stimulating greater use by farmers. As a result, the total sugar output will reach 3.9 million tonnes by 2005, about 62 percent higher than the 1996 level.

Japan. Both the decline in demand for sugar and the deterioration of the relative prices of sugar crops compared to alternative crops have contributed to the area reduction of sugar crops in the past decade. It is projected that the downward trend, particularly for sugarcane production, would continue if the current policy regime remained unchanged. The forecast sugarcane area would be 17 000 ha by 2005 which is about 29 percent less than 24 000 ha in 1996. The sugarbeet area would remain at the current level, 72 000 ha by 2005. Yields are projected to decline for sugarcane but increase for sugarbeet by 2005. Consequently, the total sugar production is projected to be about 0.9 million tonnes by 2005 which is roughly the same as in 1996. However, if the prices of substitute crops declined by 5 percent annually, it is projected that the sugarcane area would only be slightly lower than the current level by 2005, while the sugarbeet area would increase by about 7 percent to reach 76 000 ha. Consequently, total sugar production would reach about 1 million tonnes, about 2 percent higher than the current level.

Pakistan. It is projected that the steady expansion in sugarcane area and the increase in yields experienced over past decades would continue, but at a slower rate given agronomic constraints. It is projected that sugarcane areas would be slightly higher, about 4 percent, than the 1996 level by 2005, to reach 1 million ha, while yields would be 19 percent higher than current levels, at 56 tonnes per ha by 2005. The sugarbeet area and yields are projected to be 9 400 ha and 31 tonnes per ha by 2005, about 17 percent and 8 percent higher respectively than the 1996 level. Total sugar production is projected to reach 3.9 million tonnes by 2005, about 30 percent higher than the 1994 to 1996 average output level.

Philippines. Although both sugarcane yields and planted areas are still below their historical highs, the sugar industry has recovered somewhat in recent years. It is projected that this recovery would continue, although at a slower rate. The sugarcane area would reach 430 000 ha by 2005, about 17 percent more than in 1996. Meanwhile, compared to 1996, yields would increase by 21 percent to 86 tonnes per ha by 2005. Total sugar production, therefore, would amount to 2.4 million tonnes by 2005 which is 36 percent higher than in 1996 but almost the same as the level of the early-eighties.

Thailand. Production is expected to continue to increase, although less rapidly than in the past. It is projected that the total sugarcane areas would reach 1.25 million ha by 2005, about 24 percent more than in 1996. Compared with the increase in the last decade, which witnessed the doubling of sugarcane areas, this growth rate would be relatively low. The slower growth in the next decade would largely be attributed to the increasing competition from alternative crops, and lower productivity of marginal land used for sugarcane production. It is projected that yields would increase at a slower rate, from 59 tonnes in 1996 to 66 tonnes per ha by 2005 because of some of the reasons mentioned above. Although both planted areas and yields would increase at a slower pace, sugar production would reach some 9 million tonnes by 2005, more than 40 percent higher than in 1996.

The future growth prospects of sugar consumption

It is widely believed that the growth in income would be the major driving force to the increase in sugar consumption in developing countries. Since most developing countries in the Asia and Pacific region are experiencing dynamic and rapid economic growths, demand for sugar is expected to increase substantially in these countries. Moreover, given the huge population base (accounting for more than half of the world’s population), any significant increase in consumption in these countries would have important implications for the world sugar market. However, the analyses suggest that, for most of these countries, sugar consumption is largely driven by habit rather than by income alone. Therefore, even if income increases substantially in these countries, sugar consumption may not experience significant increases in the short run.

Trade

If no changes to stocks is assumed, the net trade in sugar for these countries can be estimated based on our production and consumption projections.

Australia. Given its competitive edge in the world market, Australia would continue to be probably the world’s leading exporter. Since no significant increase in domestic consumption is expected, the increase in production would result in the increase in export availabilities. It is projected that its total export availability could reach 6.5 million tonnes by 2005, about 60 percent more than in 1995.

China. The simulation suggests that China would continue to be a sugar importer in the next decade. However, the magnitude of its imports, ranging from 4 million tonnes to 9 million tonnes by 2005, would depend significantly on government policies to both sugar and other crops. For instance, if the current policy regime remains unchanged, sugar imports could increase from about 1.2 million tonnes in 1997 to 4.4 million tonnes by 2005 because supplies would not rise while the demand increases. If the government decides to reduce the use of artificial sweeteners, in particular saccharin, this would result in a substantial increase in the industrial use of sugar. As a result, total sugar imports could then reach about 9 million tonnes by 2005. The only way that China could achieve self-sufficiency would be if the government implements a favourable pricing policy for sugar crops vis-�-vis grain. It is projected that China could have a small surplus, about 0.3 million tonnes by 2005 if the government policy stimulated a significant expansion of both sugarcane and sugarbeet production at the expense of other crops.

Fiji. Given its stable consumption level, Fiji would continue to be a medium-sized exporter in the world sugar market in the next decade. Total exports could reach 580 000 tonnes by 2005, about 12 percent higher than the current level.

India. Although it is the world largest sugar producing country, India has only occasionally been a significant sugar exporter over the past decades as most its output has been consumed domestically. This situation would continue into the next decade. Domestic demand for sugar would increase substantially because of its population growth and the increase in per caput consumption. If additional efforts were made to increase production from now, India would meet its domestic sugar demand. However, if it did not, then it would need to import sugar by the year 2000, and total shipments would reach 1.7 million tonnes by 2005.

Indonesia. Although production is expected to increase, Indonesia would continue to be an importer in the next decade because the growth in population would continue to drive up national sugar consumption. It is estimated that Indonesia would import around 500 000 tonnes annually by 2005. If its per caput consumption increased at a slightly faster pace, imports could reach more than 1 million tonnes by 2005.

Japan. It is projected that as the decline in consumption would outweigh the slight decline in production, the import level would decline marginally over time to about 1.4 million tonnes by 2005.

Pakistan. It is expected that Pakistan would continue to be a sugar importing country. It is projected that imports would reach 1.4 million tonnes by 2005 if the growth in consumption continued at its current pace.

Philippines. The future position of the Philippines in the world sugar market depends largely on government polices toward the sugar industry. If the current policy regime remained unchanged, it is projected that the Philippines could become a net sugar importing country. The quantity imported could increase gradually to reach about 600 000 tonnes by 2005.

Thailand. With the continued expansion in area under sugarcane and improvement in yields, Thailand would continue to be one of the largest sugar exporting countries in the world. It is projected that Thailand would export about 6.6 million tonnes of sugar even though its own per caput consumption could reach 35 kg by 2005. Favourable agronomic conditions and lower production costs would enhance its position as a large exporter.

CONCLUSIONS

This study estimates the demand and supply functions using the theoretically consistent models for selected Asia and Pacific countries. The empirical evidence from this study suggest that in addition to the prices and various factors, the habit of consumers has significant impact on sugar consumption. Indeed, income was identified to have a rather weak effect on the change in sugar consumption in the short run if the habit formation impact were ignored.