Aspects of Modeling Sugar Markets

drey a d Vink (2007), however, state that such TRQs exis

orters. Also an official source from Fiji states that the country has a long-term deliv-ery agreement with Japan (Tadulala, 1998).

Surprisingly, the wholesale price for white sugar is considerably above that level. It is cited with ¥ 122,000 (€ 940) in 2003, ¥ 142,000 (€ 1010) in 2005 (USDA, 2006, GAIN Re-port JA6008), ¥ 130,600 (€ 970) in 2004 and ¥ 154,500 (€ 990) in 2007 (ALIC, 2007). The protection which is provided for refined sugar is

0 (€ 140 + € 340, Elobeid and Beghin, 2005) per ton is not high enough to sustain that price level. Market Experts explain the difference with profit and wholesale margins between the raw sugar price and the final wholesale price for white sugar (Dyck and Fukuda, 2007).

2.2 Aspects of Modeling Sugar Markets

discussed. In doing so, it is inevitable to talk about some common aspects of model-ling in general and modelmodel-ling of agricultural markets in particular. Thus, some issues which would rather fit in section 3.1 will have to be anticipated here. Throughout this section and the whole study, the term modelling refers to economic equilibrium modelling unless explic-itly stated otherwise.

The internal and external policies with which a number of countries, above all the EU, govern their sugar markets lead to a situation where they are at the same time importers and exporters of sizeable quantities o

ct bilateral trade flows explicitly. Net trade models which depict countries as either (net-) importers or exporters of a certain commodity are unable to do so.²⁹

White sugar from beets and from cane is chemically identical indistinguishable. Raw sugar, unless consumed in the raw stage³⁰, is an input in the production of white sugar (Mitchell, 2004). These facts justify the assumption, that sugar is a homogeneous good or a so-called fungible commodity. It is, therefore, preferable that a model to be applied for the analysis of changes in the sugar market be able to depict trade in homogeneous goods.

How-29 Banse et al. (2005) developed a possibility to depict bilateral trade flows in a net-trade model. This approach is, however, inflexible as either imports or exports are fixed exogenously.

ever, many models of international trade employ the Armington (1969) assumption of prod-uct heterogeneity with regard to origin. This assumption leads to severe consequences when applied in the modelling of sugar markets as is discussed in Nolte (2006) and shall be de-scribed in section 3.1.

However, two issues must be discussed in the context of product homogeneity. The first is the substitutability of raw and white sugar. As mentioned above, raw sugar is for its major part an (homogeneous) input in the production of white sugar, a process which is re-ferred to as refining. For a sugar importing country there are two possibilities to supply itself with white sugar. The first is to import white sugar directly from a beet sugar producing country or a country with a refining industry, the second is to import raw sugar and refine it in the country. Thus, for a country which in the past followed the first path, it will require the establishment of a refining industry if it were to switch to importing raw sugar in the future.

On the other hand, for a country following the second path, its domestic refining industry would become idle if the country were to switch to importing white sugar. In that case it can be expe

The second issue necessary to be discussed in the context of the assumption of homo-geneity

cted that the refining industry would consider its facilities as sunk cost and operate at variable cost until they are outdated. To substitute between raw and white sugar may hence require temporal adjustment. Furthermore, differences in efficiencies of refining industries have a significant influence on trade patterns, which is proved by the situation of countries like South Korea which import raw sugar and export refined sugar without producing sugar crops themselves. It is possible and likely that more countries specialize in refining imported raw sugar in the future, when the EU will reduce its exports of white sugar considerably. To capture the friction in the substitutability between raw and white sugar in an equilibrium model, the most exact way would be to model them as two products and to include a refining activity in the model³¹. If one wanted to account for the temporal adjustment it would also be necessary to model the refining industry as well as the sugar production dynamically rather than statically.

is the substitutability between cane and beet sugar. As mentioned in section 2.1.1, beet sugar is not consumed in the raw stage. For the consumption of raw sugar it is therefore necessary to obtain it from sugar cane. One the other hand, Mitchell (2004) states that raw

30 According to Mitchell (2004), only about 10% of global sugar is consumed in the form of raw or partially refined sugar.

sugar which is consumed in the US and other industrialized countries is, for hygienic rea-sons, usually white sugar to which cane molasses is added. The substitutability would be perfect in these cases, however, require the availability of cane molasses. The issue of substi-tutability of beet and cane sugar is of minor relevance, though. The reason for that is that in the current situation of the world sugar market beet sugar producers are in most cases less efficient than competing cane producers and will therefore most likely not increase their pro-duction

.

rium Models (GE), the better option was certainly to model the transportation sector endoge-nously.

and market share. A situation in which a raw sugar consumer would consider to switch from cane sugar to beet sugar is thus highly improbable to happen. The substitution will in future rather take place in the opposite direction, where substitutability is not an issue.

As pointed out in section 2.1, sugar is a close substitute in consumption for some other sweeteners, the most important of which in terms of market share being HFCS. Includ-ing that as a product in an equilibrium model would account for this relationship. The cross relationships of HFCS are, however, wide ranging. To model the supply of HFCS it would be necessary to model the supply and demand of maize which is the input for its production, and the demand for protein feeds, which Corn Gluten Feed (CGF), a by-product of HFCS production competes with. The technical as well as the empirical requirements are hence very high

To trade sugar as well as any other commodity is costly and those costs can make up for a vital share of final costs. It is, therefore, necessary to deal with the question of how to account for those costs in a model. The ability to do so in a net-trade modelling framework is limited, since the transportation costs had to be equal on all routes a country exports to or imports from. In a spatial modelling framework where different transport costs for different routes can be specified one has to decide between modelling the transportation sector endogenously and modelling transportation costs as parameters. Since sugar is a price taker in the market of bulk transportation due to its small share, the latter option would be the path to follow in a model of the sugar market solely. If one were to model sugar in a model which contains more or possibly all sectors of the economy, as it is the case with General

31 To establish such a model would besides the modeling effort require also a large amount of empirical re-search.

As has been pointed out in section 2.1 it is necessary for an analysis of the sugar mar-ket to account for interrelationships with the energy marmar-ket. There are several possibilities available to do so. The best option would be to model processing demand for sugar crops explici

sources of agricultural enterprises, above all for land. For that reason many models which deal with the sugar m

odel the markets which are not of

tly. This would, however, require the model to depict the production of sugar crops and the processing of cane or beet to sugar as two distinct activities rather than modelling the production of sugar directly as it is the case in most modelling approaches. Furthermore the modeller would have to make assumptions about the development of energy prices, espe-cially those for crude oil, and about the substitutability or the cross-price effects of the oil price on processing demand for sugar crops.³² An easier way is to make estimates about the future level of sugar production costs and hence supply curves, which account implicitly for the fact that the demand for the raw product (beet or cane) faces demand competition by the ethanol sector.

The Cultivation of sugar crops competes with the production of other crops for re-arket include the production of other crops and animal products. There are many well established ways to include substitutes on the production side in an agricul-tural equilibrium model. The limiting factors in doing so are thus usually not of technical nature. The work load that is coming along with the modelling of additional products and even more the empirical requirements are in most cases the reasons for which modellers de-cide not to include all competing markets in their model or to m

central interest at a rougher scale. Possibilities for the latter are regional or sectoral aggregation, ignoring applied policy measures, or simply choosing the supply functions and their shifters such that they implicitly account for price expectations of competing products.

As has been mentioned earlier in this section, dynamic models are better suited to capture adjustment periods than static models. The latter have merely the possibility of shap-ing their (supply) functions such that they account for a degree of adjustment that is regarded reasonable in the envisaged time horizon of the model. In the case of the world sugar market it is mainly the temporal adjustment of the processing capacity, sugar factories and refiner-ies, that makes the application of a dynamic model interesting.

32 It is important to point out that making assumptions about the latter is indeed the only possibility. This prob-lem cannot be solved by empirical research at this time, since there exist virtually no observations of the substi-tution of gasoline and ethanol purely for reasons of the price.

In a broader context, the modeler would also have to take into account the prices and cross price elasticities of other substitutes for gasoline including other bio-fuels.

A final aspect of modelling the sugar market is the selection of an appropriate func-tional form. The most common funcfunc-tional form of supply curves in agricultural equilibrium models are isoelastic ones in partial modelling frameworks or functions derived from a Con-stant Elasticity of Transformation (CET) production function in GE models. Both do not allow for production to end while facing a positive producer price. This is, of course, unreal-istic for all products, but usually not a severe problem as long as the run of the supply curve shows the desired behaviour in the price range the modeller expects to prevail. The potential price changes on the sugar market are, however, too large for such a proceeding. If those are to be modelled it is necessary to employ supply functions which allow for a positive inter-cept on the price axis, i.e. simulate the production to end below a certain price. The simplest case of a function showing that property would be a linear function, which is very restrictive, though, and whose curvature contradicts observed behaviour of producers. Another, more elegant way to model sugar supply, which under certain conditions would also allow for a positive intercept on the price axis, are so-called second order flexible functional forms. For their complexity, however, they are rarely applied in agricultural equilibrium models (Grethe and Weber, 2005). A more simple way, which is chosen by Nolte and Grethe (2007) and which will also be applied in the model used for this study, is the generalization of an isoe-lastic function by the introduction of an additive parameter which corresponds to the price at which the production of sugar will be abandoned.

3 Modeling Approaches and Results of Former Studies