Bio-economic household modelling for agricultural intensification

This study contributes to the quest for sustainable agricultural intensification through the development of a quantitative bio-economic modelling framework that allows assessment of new technology and policy measures in terms of household welfare and sustainability indicators. The main aim of the study is the development of a farm household model to aid policy dialogues. The study consists of three parts. The first part is a general introduction into the context of the research, a justification of the approach and a general description of issues underlying the modelling framework. The second part explains the methodological details of the modelling framework. The third part contains some applications of the approach to specific questions related to agricultural intensification in the Cercle de Koutiala in southern Mali.

The bio-economic model developed in this study combines elements from different existing methodologies into a flexible framework that is able to capture the peculiarities of household agriculture in West Africa. The methodology is sufficiently general to be applied in other settings as well, and contains a number of innovations, viz. a direct utility function, a robust goal weighting procedure and the use of metamodelling to analyse mathematical programming outcomes.

Part 1

Soil degradation is regarded as a serious problem threatening the livelihoods of present and future generations in West Africa. To bring soil degradation to a halt, a combination of appropriate technology and an enabling policy environment is needed (Chapter 1). To assess new technology and policy measures, information from biophysical and socio-economic disciplines are combined into a quantitative framework.

Over the past decade a number of quantitative studies have been conducted that aim at combining biophysical and socio-economic information in such a way that the results are relevant for both social and biophysical sciences. These approaches are termed bio-economic modelling. A review of the methodologies (Chapter 2) reveals that none of them are able to tackle simultaneously the analysis of the causes and effects of soil degradation in combination with household decision making to assess the effects of policy change. The studies do however provide valuable building blocks for the present methodology.

A framework to characterise the bio-economic models according to the spatial and temporal scales assists in finding appropriate methods for different research questions. Two critical issues emerge from the review. The first concerns the choice of objective function in economic models. The second refers to the interface between economic behaviour and biophysical processes.

This critical interface between biophysical processes and economic behaviour is wrought with difficulties due to differences in scientific paradigms (Chapter 3). Biophysical sciences use the concept of efficiency in the analysis of technology options. The concept differs from the way economists use it. As a result there is a disparity between the way biophysical scientists and economists view production and damage functions. Whereas economists tend to use well-behaved continuous Cobb-Douglas production functions, biophysical scientists describe production activities in terms of the outcomes of biophysical processes, which more often than not yield nasty functions. This is due to the synergistic effects of inputs and the interrelations between causes and effects of soil degradation.

The implications for bio-economic modelling are that Leontief production functions best describe the biophysical processes. Biophysical modelling frameworks exist that generate point data for this type of production function. One such framework, the technical coefficient generator (TCG) is used for generating the biophysical information needed in the household model.

Part 2

The household model is based on the standard theoretical model of a farm household (Chapter 4). The theoretical model although developed for econometric estimation is difficult to implement in such a way, due to the existence of failures and imperfections in commodity and input markets, the occurrence of risk, data limitations and the complexities in the production functions. As a result a complex non-separable household model is needed, which in turn cannot be estimated econometrically.

Instead of estimating a full econometric model the present methodology proposes an alternative through the use of mathematical programming models that have been parameterised with partial econometric studies and expert knowledge. The basic structure of this bio-economic model consists of six separate modules. The production activities module describes the biophysical processes and their interrelationships using information generated by the TCG. Different technological options are defined in terms of input-output combinations for both current agricultural practices and alternative technologies. The price module includes information on factor and commodity markets. Price bands are used to describe market imperfections and results from the household models in terms of aggregate supply are used to calculate new market-clearing prices.

A separate module describes different household types in terms of their resource endowments, real time preference and savings capacity. The savings capacity is linked to a savings and investment module that describes consumption smoothing and investment behaviour. Investment in soil conservation measures is one way of halting ongoing soil degradation.

The expenditure module warrants separate mention (Chapter 6). The use of non-separable farm household models implies that consumption and production decisions are considered simultaneously. As a result the commonly used profit maximisation objective function cannot be used. Instead a utility function is used that describes household preferences for consumables. The direct utility function is estimated econometrically from a cross-sectional budget survey that is considered the revealed preference generated by an underlying utility function.

The study develops a procedure to derive such a utility function. Because direct measurement of utility is impossible careful procedures are needed to test if the derived function is statistically robust. Next to consumption households also consider soil degradation in their decision making. The consequence is that multiple objectives have to be considered and a procedure is needed to combine those objectives (Chapter 5). The study presents a methodology for estimating the weights of different household objectives by comparing simulation model results with empirical evidence. To obtain statistically robust results maximum entropy econometrics is used.

Part 3

Application of the modelling framework to the case study area of Cercle de Koutiala in southern Mali is done for specific research questions. The first question concerns the validity of the model itself (Chapter 7). The model generates a base run that is consistent with empirical evidence. Applying sensitivity analysis to key parameters, analysing the near-optimal solution space and by applying the model to a separate data-set tests the robustness of those results. The model turns out to be robust for the most important variables while insight is gained into those areas for which the model does not give adequate answers.

The model is also used to analyse new technology (Chapter 8). New technologies were chosen on biophysical grounds. Using partial budget analysis a first indication of the possibilities of the new technology is obtained. The approach is too partial to capture farm household goals and aspirations nor the resource constraints they face. Bio-economic model results indicate that most of the alternative technologies that seemed promising from a biophysical point of view do not fit well into the production systems of farm households in Cercle de Koutiala. A metamodelling approach is used to analyse the outcomes of the farm household model for a large number of variations in key exogenous parameters, thus obtaining fluid response surfaces.

The model is also used to assess the possibilities of using policy instruments to create an enabling environment to induce farm households to adopt more sustainable technologies (Chapter 9). Two key instruments that figure in the forefront of policy debates in West Africa are analysed, viz. fertiliser price subsidies and infra-structural development resulting in lower transaction costs. Model results analysed in a metamodelling framework indicate that although the direction of the change in both income and soil organic matter balance is as would be expected, viz. simultaneous improvement of household-welfare and agro-ecological sustainability indicators, the magnitude of the improvements is limited. The policy measures in combination with the available new technologies are effective but not efficient.