An applied general equilibrium model for Dutch agribusiness policy analysis

The purpose of this thesis was to develop a basic static applied general equilibrium (AGE) model to analyse the effects of agricultural policy changes on Dutch agribusiness. In particular the effects on inter-industry transactions, factor demand, income, and trade are of interest.The model is fairly general and could be used to analyse a great variety of agricultural policy changes. However, generality requires that the model should be adapted and extended for special research questions. This is shown in chapters 13 to 15 where the AGE model is used to examine the impact of the milk quota system and a reduction in livestock production for environmental purposes.The policy simulations performed serve in the first place as an illustration of the AGE model, in the second place they function as a source of information for policy makers and interest groups.The thesis consists of five parts; each part contains one or more chapters.Part I describes the purpose and method of research, and presents the structure of the thesis. The purpose and method have already been described above.Part 11 presents the concept of agribusiness (chapter 2) and gives a concise quantitative description of Dutch agribusiness (chapter 3). Industries or parts of industries with strong economic links, either on the input or output side, with agriculture belong to Dutch agribusiness, together with agriculture itself. In chapter 3 the size of Dutch agribusiness is measured invalue added and employment. Agriculture had a share in total value added of 4 per cent in 1981 and 3.8 per cent in 1988, for agribusiness these shares were respectively 7.5 and 6.5 per cent. The quantitative description is madeusing input-output analysis. Chapter 3 also presents the data for the basic model. These data are given in expenditure and tax tables which show the flows of the value of transactions and income in the Dutch economy in 1981.Part III presents the basic AGE model. In the model there are 19 industries that produce 20 goods (dairy farming has two outputs). Four of these industries are agricultural industries (dairy farming, arable farming, livestock production and horticulture) and there are 6 other agribusiness industries (dairy manufacturing, meat manufacturing, grain mills, sugar factories, cocoa product manufacturing and food not elsewhere classified). The industries use intermediate inputs, imports and the services of primary inputs to produce outputs. The outputs of the industries are used domestically or exported. Industries operate in competitive input and output markets with free entry. Therefore they take prices as given. Industries maximize profitsgiven their technology, which is described by multi-level production structures that consist of CES (Constant Elasticity of Substitution) production and CET (Constant Elasticity of Transformation) product transformation functions that include exogenous factor-augmenting technological change.Because of the competitive markets profit maximization is replaced by cost minimization and revenue maximization. The production structure is presented in chapter 5. In chapter 6 technological change is introduced into the production structure and the functional forms are chosen in chapter 7.Consumer demand (see chapter 8) partially determines the level and relative distribution of the output of industries. In the model there is one representative consumer or private household that owns the labour, land and part of the capital stocks in the economy. A fixed share of the total income of the private household is saved, the other share is used for expenditure on consumer goods. The total income of the private household is determined by the income earned from the supply of the services of the primary inputs corrected for taxes and subsidies and net income transfers.The private household maximizes a utility function given an income constraint, or because of the duality between the utility and expenditure function, minimizes expenditure given a fixed level of utility. Utility maximization results in a demand system related to the Linear Expenditure System (LES) that allows for substitution between consumer goods as a consequence of price and income changes.The supply of the services of primary inputs (see chapter 9) is not part of the utility function of the private household but is exogenous. This implies, for example, that there is no choice between labour and leisure in the model.The services of the primary inputs are used by industries according to the marginal returns of these services. It is assumed that the services are imperfectly mobile between industries. This is a distinct feature of the model. The factor rewards go to the owners of the primary inputs, who are: the private household (all the wages and land rents, and part of the capital rents), the government (part of the capital rents) and the rest of the world (part of the capital rents). The total availability of the primary inputs is exogenous to the model. This implies that, for example, immigration or other changes in the labour force are not modelled. Gross investment is modelled but it only affects spending and not the production capacity in the economy.Chapter 10 deals with trade in the basic AGE model. Imports are divided into competitive and complementary imports. The former are imported only by industries. The latter are also imported by the private household. The competitive imports are imperfect substitutes for domestic goods. The complementary imports have no domestic equivalent. For products from agribusiness a division has been made between the EC and rest of the world as regards competitive imports and exports.In the model the small country assumption is used in the sense that world market prices for imports and exports are exogenous variables in the model. This implies that the supply of imports and the demand for exports are perfectly elastic.To model import demand and export supply the Armington procedure is used. The Armington procedure assumes that imported, exported and domestically produced and used goods are imperfect substitutes.The government or public sector has three functions in the model (see chapter 11). Each function is performed by a special hypothetical institution.First, the Treasury can impose direct taxes, indirect taxes and tariffs. But it can also give export subsidies and income transfers. All taxes and subsidies in the basic model are represented by ad valorem taxes or tariffs. The Treasury can use policy instruments to influence prices, and hence interindustry transactions, income, etc.Second, the public services industry purchases commodities and the services of primary inputs to produce its output. It is treated as an ordinary industry in the model.Third, the public household performs the consumption and saving tasks of the public sector. It demands only one good: the output of the public services industry. The income of the public household is determined by the Treasury (by means of an income transfer) and the income from the supply of capital services. A fixed proportion of the income of the public household is borrowed (negative saving).In a general equilibrium model all input and output markets are in equilibrium. The other equilibrium conditions are a fixed public budget (the revenue and expenditure of the Treasury are equal), zero profits, the budget constraints of the private and public households, a fixed surplus on the trade balance, and equality of saving and investment. The model has a neoclassical closure in the sense that saving determines investment in the model.The model is calibrated using 1981 data. Calibration implies that the coefficients in the model are determined such that the model represents the base year situation. For the calibration procedure expenditure and tax tables were used (discussed in chapter 3) which show the value of transactions and income flows in the economy. In addition, values of the substitution and transformation elasticities are required. Both the expenditure and tax tables and the value of the elasticities are taken from Zeelenberg et al. (1991).A solution strategy and a solution algorithm are required to solve the model for alternative equilibria. The model is a collection of non-linear algebraic equations and is solved directly with a numerical solution technique included in GAMS, a computer package developed to solve for non-linear systems.The equilibrium conditions, calibration procedure and the model solution method are discussed in chapter 12.The basic model is a general framework. To analyse specific policy questions for agribusiness the basic model has to be modified. Part IV of the thesis examines the consequences of two specific policies; a supply quota for milk and a supply quota for livestock to protect the environment. First, detailed policy instruments were incorporated in the model (see chapter 13). These are variable import levies and export subsidies for dairy and meat imports and exports from outside the EC, and supply quotas for milk and livestock production. Some background information on the policy issues and the outcome of the policy simulations are presented in chapter 14 (a supply quota for milk) and chapter 15 (a supply quota for livestock).In chapter 14 a ceteris paribus analysis of different aspects of the introduction of milk quotas is given. The results provide some important insights.First, they reveal the dependence of the outcomes on EC and world market prices. This stresses the importance of trade for Dutch dairy production and agribusiness. The competitiveness of the Netherlands on the world market for dairy products is affected by the introduction of milk quotas because domestic prices change relatively to rest of the world market prices. For example, there is a growth of imports from the rest of the world and a relative increase in importance of the EC as the export market for Dutch dairy products.Second, the dependence of the outcomes on the degree of factor mobility is highlighted. The assumption in most AGE models of complete or totally incomplete factor mobility needs therefore to be reconsidered.Third, with low factor mobility, factor demand (e.g. employment) in dairy farming does not fall much but factor prices do, whereas with high factor mobility the primary inputs move out of dairy farming and the factor prices drop less because they more or less have to equalize between industries. The value of factor demand is larger with high factor mobility implying smaller quota rents and quota prices. It has to be remembered that factor mobility is low in dairy farming.Fourth, the effects on other industries of the introduction of milk quotas are substantial, especially for dairy manufacturing and grain milling, which includes the feed industry. Neglecting the effects in those industries would be misleading. However, the effects are less than predicted by input-output models because of the greater possibilities for input substitution in the AGE model.Fifth, the introduction of milk quotas increases welfare, measured in terms of GDP and the equivalent variation, although the rise is small. This is a second best result; quotas reduced the distortions caused by government intervention in dairy markets.Sixth, from a welfare-economic point of view price decreases, compared to supply quotas, are preferable when the production of milk, and therefore the support given to milk production, has to be reduced (GDP and the equivalent variation are higher). The welfare improving effects on industries are larger with price decreases than with supply quotas: output, factor demand, value added and trade change more. Especially in dairy farming, value added decreases much more in case of the price decreases. These large changes make price decreases in dairy farming politically difficult to achieve.Finally, the degree of output substitution in dairy farming between milk and cattle production is relatively unimportant for the effects of milk quotas.Of course the model results do not provide a definite answer to all issues regarding milk quotas. The strength of the model approach chosen is that it gives a consistent static analysis of the effects of the introduction of milk quotas on inter- industry transactions, price and income formation, factor demand, and exports and imports in Dutch agribusiness and economy as a whole.Many of the general issues discussed above are also relevant for the simulation of the introduction of supply quotas in the livestock industry (see chapter 15). In this simulation the difference between the effects of using supply quotas and a levy on the demand for compound feed were analysed. The low price elasticity of demand for compound feed made the differences in the effects on output and inputs between both policies small. However, there is a large difference in value added generated because with supply quotas quota rents are created that support income earned in livestock production. However, in the long run quota rights are just an ordinary production factor. This implies that in the long run livestock farmers are better off with levies on compound feed (because value added excluding rents is reduced less with the levy on compound feed).The meat manufacturing industry and the grain mills are affected substantially by both ways of reducing livestock. However, the effects for other industries are relatively small which indicates that the livestock industry is relatively small and isolated in the Dutch economy. Welfare losses measured by changes in the equivalent variation are greater in the case of supply quotas than in the case of the levy. This indicates that supply quotas for livestock cause more distortions than the levy for compound feed.Both supply quotas and levies on the demand for compound feed increase domestic prices relative to EC and world market prices. This implies a strong decrease of exports and increase of imports of livestock products.The last part of the thesis assesses the methodology and results, and draws conclusions. It is concluded that the model developed is a flexible and powerful too] for analysing the effects of agricultural policy changes on agribusiness and the economy as a whole. This is especially true in comparison with the input-output models that are traditionally used to analyse the economy-wide effects of agricultural policy changes, and with partial equilibrium models. The AGE model developed here provides valuable information for policy makers and interest groups on employment, income and trade in the economy as a whole.The AGE model has additional advantages. First, the model incorporates accounting consistency (for example budget constraints are taken into account as is market balance, in addition to basic macroeconomic identities such as the equality of saving and investment). Second, the model is theoretically consistent which makes the interpretation of the results relatively easy in spite of the fact that the model is rather large. Third, all interindustry effects are explicitly modelled, there is no need to make a choice regarding which linkages are important enough to model as in partial equilibrium models. Fourth, in a partial equilibrium model some results are quite obvious, for example introducing trade distortions reduces welfare. In a general equilibrium model this is no longer always the case because the effects on the rest of the economy are also taken into account. Finally, welfare analysis should be applied to the owners of the factor inputs, that is, the households. This is easy in the AGE model but not so easy in a partial equilibrium model. Changes in consumer and producer surplus and tax receipts are often a poor proxy for the real welfare changes in an economy.The broadness of our AGE model, however has its price. The model ignores much detail which is often present in partial equilibrium models. Moreover, the fact that the model is rather large and few data are available means that econometric estimates, and therefore econometric tests, of the behaviourial equations are impossible. However, use has been made of econometric estimates of substitution (AESs) and transformation elasticities (AETs) as presented by Zeelenberg et al. (1991). In the future, to make the results of the simulations more useful for policy analysis more technical information could be used (e.g. crop and livestock yields), more data about the development of the exogenous variables could be obtained (e.g. on world market prices) and more external information on AESs and AETs could be used. Although the model is unsuitable for predictions the clear structure and flexibility make it a useful tool for analysing the effects of agricultural policies in relation to different assumptions about key linkages between agricultural industries and the rest of agribusiness and the economy as a whole.