Uncertainty on Agricultural Public Good and Externality Production. Implications for Domestic and International Agricultural Policy Design

Table of Contents

Summary

Zusammenfassung

1 Introduction

2 Methodological approach

3 The base version of MULTSIM – a reference for analysing uncertainty
3.1 Valuation of externalities
3.2 Policy scenarios
3.3 Policy goal trade-offs

4 Implications of uncertain knowledge on the societal valuation of public goods and externalities

5 Conclusions for policy design

Sources

Annex

Summary

The results of empirical studies on the valuation of public goods and externalities produced by agriculture should be integrated into quantitative modelling approaches for policy-decision support. Although valuation research progresses, uncertainties about the societal valuation of non-marketable outputs and negative externalities will remain part of the constraint set that determines policy decisions on economic incentives for a multifunctional agriculture. The role of valuation uncertainty must therefore be explicitly addressed in policy advice. This article deals with the consequences of uncertainties about the societal valuation of landscape benefits and environmental costs for multiple policy goal trade-offs that may result when internalising external benefits and costs by abolishing commodity subsidies, introducing land subsidies and intermediate input taxes. Numerical results of an illustrative model of the German agricultural sector are presented and trade-off curves for farm income, environmental cost reduction and landscape benefits are discussed. If the relevant criterion for policy decisions was a simple aggregation of the single policy goal indicators with constant weights, the optimality of policy options would be rather robust to valuation errors. Expected exchange ratios between the different policy goals, however, can be severely distorted if policy-makers have wrong estimations of the true societal valuation of externalities. This may make targeted incentives that aim at a better orientation of agriculture towards societal demands appear too risky for policy-makers and finally leads to the blockage of policy changes. An appropriate policy portfolio should therefore comprise non-distorting compensations for negatively affected policy goals. The EU’s commitment in the WTO to multifunctional agriculture has to be underlined by effective domestic policy reforms, if this concept is not to be discredited as an attempt to justify disguised protectionism.

Key words: environmental costs, landscape benefits, multifunctional agriculture, valuation of non-marketable outputs, uncertainty, agricultural sector modelling, policy advice

Zusammenfassung

Die Ergebnisse empirischer Untersuchungen zur Bewertung von der Landwirtschaft produzierter öffentlicher Güter und Externalitäten sollten in quantitative Modellierungsansätze zur Unterstützung von Politikentscheidungen einbezogen werden. Wenngleich Forschungsarbeiten zur Bewertung voranschreiten, werden Unsicherheiten über die gesellschaftliche Bewertung von nicht marktfähigen Outputs und negativen Externalitäten Bestandteil der Restriktionsmenge bleiben, die die Politikentscheidungen über ökonomische Anreize für eine multifunktionale Landwirtschaft bestimmt. Die Bedeutung von Bewertungsunsicherheiten ist deshalb in der Politikberatung explizit anzusprechen. Dieser Beitrag beschäftigt sich mit den Folgen von Unsicherheiten über die gesellschaftliche Bewertung von Landschaftsnutzen und Umweltkosten der landwirtschaftlichen Produktion für Mehrfachzielkonflikte, die bei der Internalisierung von externen Nutzen und Kosten durch Abbau von produktgebundenen Subventionen, Einführung von Flächenbewirtschaftungssubventionen und Vorleistungssteuern entstehen können. Es werden numerische Ergebnisse eines illustrativen Modells des deutschen Agrarsektors vorgestellt und Trade-off-Kurven für Agrareinkommen, Umweltkostenreduzierung und Landschaftsnutzen diskutiert. Bei einer Aggregation der einzelnen Politikzielindikatoren durch konstante Gewichte zu einem allgemeinen Wohlfahrtsindikator als relevantes Entscheidungskriterium für Politikentscheidungen wäre, so die Ergebnisse des Modells, die Optimalität von Politikoptionen relativ robust gegenüber Bewertungsfehlern. Erwartete Austauschrelationen zwischen den verschiedenen Politikzielen können jedoch erheblich verzerrt sein, wenn Politiker die wahre gesellschaftliche Bewertung von Externalitäten falsch einschätzen. Dies kann dazu führen, dass Politikern gezielte Anreizmechanismen, die auf eine an den gesellschaftlichen Bedürfnissen besser ausgerichtete Landwirtschaft zielen, als zu riskant erscheinen und Politikänderungen schließlich verhindert werden. Ein geeigneter Bestand von Politikmaßnahmen sollte deshalb auch nicht-verzerrende Kompensationen für durch Politikänderungen negativ betroffene Politikziele umfassen. Das Bekenntnis der EU in der WTO zu einer multifunktionalen Landwirtschaft muss durch effektive eigene Politikreformen unterstrichen werden, wenn dieses Konzept nicht als ein Versuch zur Rechtfertigung von verstecktem Protektionismus diskreditiert werden soll.

Schlüsselwörter: Umweltkosten, Landschaftsnutzen, multifunktionale Landwirtschaft, Bewertung nicht marktfähiger Güter, Unsicherheit, Agrarsektormodellierung, Politikberatung

1 Introduction

The liberalisation of agricultural trade and the reduction of domestic support for farming are major objectives of the current WTO round. The European Union and many other developed countries like Norway and Japan fear, however, that important domestic policy objectives may be missed if their agricultural sectors are exposed to stronger international competition. They demand that so-called non-trade concerns (NTCs) are taken into account in the future trade agreement. The European Union favours the concept of a multifunctional agriculture, which emphasises that agriculture produces so-called non-commodity outputs as for example landscape amenities, maintenance of recreational areas, biodiversity and flood control. Common to most of these non-commodity outputs is that they possess characteristics of externalities and public goods, which are jointly produced with commodity output and not traded on markets. Since they are perceived as benefits for the society, it is often argued that the farm sector should not base factor allocation and production decisions on the basis of world commodity prices only and that, in addition, policy should create economic incentives that support farmers in producing such societal benefits. However, agriculture also produces societal costs as for example water pollution and soil erosion that are not completely internalised by farmers. The co-existence of positive and negative externalities and, in particular, their blurred valuation makes national and international policy design for multifunctional agriculture a complex task.

Adequate policy design for multifunctional agriculture requires knowledge from different scientific disciplines like natural sciences, engineering, economics and sociology that has to be integrated through a multidisciplinary approach. But this knowledge often proofs to be rather uncertain when it has to be transferred to the relevant policy site. Knowledge - at the broader sectoral scale – of how supply of non-commodity outputs reacts to changes in factor allocation and production structure and of how society values public goods and externalities is rather uncertain. Being an unavoidable part of the constraint set faced by policy-makers, uncertainties of these kinds should not be masked but explicitly addressed by policy advisors.

The aim of the paper is to numerically illustrate the implications of uncertain knowledge of public good production and societal valuation of externalities for national and international policy design. The remainder of the paper is organised as follows: Section 2 describes in short the methodological approach of the German agricultural supply model MULTSIM with multifunctional outputs, which is used for the numerical calculations. In section 3 a set of reference simulations with MULTSIM comprising policy scenarios that abolish commodity subsidies and internalise environmental costs and landscape benefits is presented. For section 4 various sensitivity analyses with respect to the societal valuation of externalities and public good production have been conducted. The results are compared with the reference simulations in order to study the implications of uncertain knowledge for policy goal trade-offs between farm income, environmental cost reduction and landscape amenity value production. Section 5 presents conclusions for policy design.

2 Methodological approach

Quantitative modelling tools should support policy makers in understanding goal conflicts and help to give scientific advice even if our knowledge of the real world is limited and uncertain. Multi-market and trade models are widespread used to analyse the effects of market and trade policy instruments on supply and demand for commodities, prices and welfare indicators (e.g. Kirschke and Jechlitschka, 2002; Weber, 2001; Dixit and Roningen, 1986; von Lampe, 2001). With respect to multifunctional agriculture and public good production these models are less useful since they do not take into account the external benefits and costs of agricultural production. Sector models have been extended to depict the relationships between production processes and environmental indicators (e.g. Flur, Gotsch and Rieder, 2001; Heckelei and Britz, 2001; Julius et al. 2003; Gömann et al., 2003) – often at detailed regional level.

The choice of an appropriate modelling approach is important for policy advice. Here it is argued that the approach must not be too ambitious as regards the degree of commodity and input differentiation in order to gain illustrative insights into the implications of uncertain knowledge for expected policy goal trade-offs. Therefore the relatively simple and easy to handle functional relationships of the supply model MULTSIM for German agriculture are used. The approach has been described in greater detail in Weber (2003a,b). It can be summarised as follows: The agricultural sector is modelled as producing a commodity output y represented by the production value, a non-commodity output a with external benefits from landscape maintenance (e.g. flood control, aesthetic value, variety of landscape, amenities), and an environmental output e representing external costs (e.g. water, soil and air contamination). Thus it only reflects basic relationships between factor input and output. However, variable proportions between these joint products are allowed. Production takes places in five farm types. These are specialised field crops, specialised grazing, specialised granivore, permanent cultures, and mixed agricultural production [F=(FIELD,GRAZ,GRAN,PERM,MIX)]. As production factors farms use intermediate input, land and other primary factors including labour and capital [L=(INT,LAND,OTHP)] with the latter assumed to be fixed.

The production functions are of the iso-elastic Cobb-Douglas type. The set of production elasticities consists of elasticities of commodity output by single farm types with respect to input use by farm type αfl, elasticities of landscape benefits with respect to land input by farm type βf, elasticities of landscape benefits with respect to commodity output per hectare by farm type gf and elasticities of environmental costs with respect to intermediate input use per hectare by farm type kf (with fÎF and lÎL). Table 1 presents the set of production elasticities used in MULTSIM for illustrative scenario building. The elasticities imply that agricultural land input is the main determining factor for landscape benefit production (Σfβf=0.9), whereas commodity output quantity per hectare is less relevant (Σfγf=0.1). The landscape benefit elasticities of the single farm types reflect their shares in total agricultural land use and in total agricultural commodity output, respectively. With respect to environmental costs the elasticity set implies that intermediate input quantities have a disproportionate impact, which is the same for all farm types (κf=1.5).

The economic rationale behind the model’s factor allocation mechanism is that farm types maximise their profits subject to the production functions and that land rent is maximised at the total sector level given prices for land use. A system of Kuhn-Tucker conditions for non-linear optimisation with inequality constraints is set up and the two optimisation problems are solved simultaneously. The policy interface of MULTSIM comprises a spreadsheet to set subsidies and taxes linked to commodity output and production factors. Overall social welfare is defined as the sum of producer rents and external benefits minus external costs minus the taxpayers’ net position vis-à-vis agriculture. The latter is calculated as expenditures on subsidies minus revenues from taxes. Government intervention in the form of subsidies and taxes can be played through with the model by scenario analyses. But the model can also be used to conduct normative analysis and to solve for the overall social welfare maximising subsidies and taxes. The modelling-technique uses the Kuhn-Tucker conditions for profit maximising as constraints to the social welfare optimisation problem.

The model is set up with the software GAMS (Brooke et al., 1998) and is solved using the solver CONOPT2 which is based on a reduced gradient algorithm (Drud, 1992).

3 The base version of MULTSIM – a reference for analysing uncertainty

3.1 Valuation of externalities

Empirical studies that value positive externalities from the societal demand perspective are often based on contingent valuation methods (CVM) and measure the willingness-to-pay (WTP) (for an overview see Randall, 2002). So, for example, Cicia und Scarpa (2002) measure an annual WTP for the traditional agricultural landscape view in the national park of Cilento/Italy of 60-130 Euro/ha and Drake (1991) estimates that the annual WTP for agricultural landscape in Sweden is 140 Euro/ha. However, the usefulness of CVM studies for policy support at the national and supranational level is still limited. The transfer of estimated WTP-values and welfare functions from single study sites to the national and supranational policy site is – as Navrud (2002) reports - subject to major errors. Meta analyses of CVM studies use the methodological characteristics of the different studies as additional independent variables but pay little attention to the characteristics of the analysed public goods and externalities or to the socio-economic determinants of societal valuation (Navrud, 2002). As a consequence, using the results of WTP measurement for policy advice is a shaky undertaking. Rough assumptions have to be made. For example, for a model of US agriculture with positive and negative externalities, Peterson, Boisvert and de Gorter (2002) use amenity values for agricultural land of 0-10 US-$/acre (0-25 US-$/ha)¹ and assume external environmental costs of 2-5 billion US-$².

In the base version of the model MULTSIM, which is the starting point for the various sensitivity analyses in section 4, the assumed landscape amenity value produced by agriculture for the base year 1999/2000 is 5% of the commodity output value (corresponding to 103 Euro/ha). The external environmental costs are assumed to amount to 10% of the commodity output value. It is to be emphasised that the valuation of the externalities is set for illustrative scenario building and sensitivity analyses, but does not mean to reflect true external benefits and costs.

3.2 Policy scenarios

In Weber (2003b) the model MULTSIM has been used to analyse the effects of abolishing commodity subsidies and of internalising external costs and benefits by intermediate input taxes and land subsidies. The same set of scenarios is now used to illustrate the implications of uncertain knowledge about the valuation of public goods and externalities for policy goal trade-offs and policy design. In the base scenario (BA) all commodity subsidies are abolished. Scenario 4 (SC4) represented a policy with overall social welfare maximising levels of intermediate input taxes and land subsidies. In SC1 to SC3 and SC5 to SC7 intermediate input taxes are reduced and increased, respectively, by identical steps of 25% of their levels in SC4. As in BA commodity subsidies are abolished in SC1 to SC7. Table 2 shows the resulting subsidy and tax levels in these scenarios.

3.3 Policy goal trade-offs

The results of the base version of MULTSIM on the impacts of the policy scenarios for factor allocation and production quantities are presented in Table A.1 of the annex. They are very close to the results that have already been reported on in Weber (2003b)³. Following we focus on the various welfare indicators and on policy goal trade-offs.

Table 3 presents the changes in the welfare indicators of the scenarios compared to the base year 1999/2000. Abolishing commodity subsidies (BA) reduces external environmental costs by €1.1 billion (30%), additional full internalisation of externalities by land subsidies and intermediate input taxes (SC4) diminishes these costs by €2.1 billion (61%). Producer surplus for the total agricultural sector falls in BA by €3.6 billion (23%). Internalisation leads to additional income losses. In SC4 producer surplus is €3.6 billion (27%) lower than in BA. Abolishing commodity subsidies (BA) reduces the taxpayers’ net expenditures on the agricultural sector by €3.3 billion, and with full internalisation (SC4) the budgetary savings even slightly increase since the revenues from intermediate input taxes are higher than the expenditures for land subsidies. If the budgetary savings are redistributed to farmers by payment schemes that are fully decoupled from production, the environmental situation could be improved without substantial losses in farm income.

[...]

¹ 0-6% of the agricultural output value

² 1-3% of the agricultural output value

³ The main difference of the MULTSIM version used in Weber (2003b) from the base version of MULTSIM presented in this article is that the γ-elasticities had been set to zero and that the sum of the β-elasticities was unity.