Abstracts Working Papers

Abstracts of the professional publications by SOW-VU's staff since 1999.

Hard copies of SOW-VU's working papers are available from the Centre for World Food Studies.


2008      2007      2006      2005      2004      2003      2002      2001      2000




Wesenbeeck, C.F.A. van, M.A. Keyzer, M. Nubé, ‘Estimation of Undernutrition and Mean Calorie Intake in Africa for 2005: Methodology, Findings and Implications‘, 23 pp.

Poverty and hunger are basic yardsticks of underdevelopment and destitution. While the measurement of poverty through surveys is relatively well documented in the literature and a reasonable degree of methodological consensus has been emerging, for undernutrition, information is much scarcer, particularly regarding adults, and very different methodologies are applied for children and adults. Undernutrition in children is generally estimated on the basis of direct anthropometric measurement (underweight, stunting, and wasting) but for adults a headcount is commonly used for instance by FAO that relies on estimates of the “food gap” – the difference between actual and normative calorie intake – in which anthropometry hardly plays any role. To obtain this food gap, food availability, measured as production plus net import at national level, is distributed among household groups on the basis of information from household expenditure surveys. This yields food availability by household group, and comparison with a given minimum calorie intake leads to the gap, from which the headcount is derived. Our paper seeks to improve on this practice in two ways. One is that we estimate the prevalence of undernutrition in Sub-Saharan Africa (SSA) based on anthropometric data for women and children in the Demographic and Health Surveys (DHS), nationally representative surveys commissioned by USAID that have been held in many African countries. As second contribution, we estimate mean calorie intake and implied calorie gap for SSA, also using anthropometric data in the DHS as point of departure, after conducting a test on their quality.  Our main results are, first, that we find a much lower prevalence of hunger, and, second, that there is much less spread in mean calorie intake across the continent than reported by FAO (2008), the only estimate that covers the whole of Africa. On the basis of our results, we argue that these spatially explicit estimates of total calorie intake form a solid basis for an assessment of economic performance of Africa. Especially in East and Central Africa, economic growth must have been higher than is usually assumed. In addition, we indicate how our anthropometry-based estimates for the year 2005 can be used in a fast update procedure to assess the impact of the food crisis (2007/2008) on Ethiopia.




Keyzer, M.A., M.D. Merbis, R.L. Voortman, ‘The biofuel controversy’, 23 pp.

About a decade ago, the main OECD countries decided to promote the use of biofuels so as to reduce greenhouse gases, to contribute to energy self-sufficiency and to create additional demand for agricultural commodities. The introduction of mandatory blending requirements and lavish subsidies spurred fast adoption of this technology. In the course of 2008, the already existing controversy about the effectiveness of this strategy culminated as the resulting upward shift in demand contributed to staggering rises in food prices on world markets. It is uncertain as yet whether this will tone done current ambitions among policy makers to expand biofuel production. The paper shows that high ratios of energy prices to food prices are needed to make biofuel production profitable without the mandatory blending and subsidies. Yet, even if food-based biofuels disappeared, the issue remains that rising high energy prices will promote intensified use worldwide of land for energy crops, requiring huge amounts of mineral fertilizers and putting nature under additional pressure. In policy terms, this defines three major tasks. The first is replacing the current excise taxes on energy carriers by a uniform carbon tax, so as to mitigate greenhouse gas emissions in an efficient manner, the second to prevent price fluctuations on the oil markets from destabilizing food markets, as happened in recent years. Introduction of upper limits on the use of food for biofuel could prove effective here. The third, much wider, task is to make the transition to a partly biomass based energy production possible and sustainable, that is establishing fair distribution of property and user rights over the lands, while safeguarding biodiversity and soil fertility and maintaining adequate labor standards and living conditions, also during periods that these become non-profitable following a drop in energy prices.




Nubé, M., ‘The Asian Enigma: Predisposition for low adult body mass index among people from South Asian descent’, 24 pp.

Prevalence rates of undernutrition among children and adult women are considerably higher in South Asia in comparison with Sub Sahara Africa. At the same time, South Asia performs better than Sub Sahara Africa with respect to a number of factors generally strongly associated with the occurrence of undernutrition, such as infant mortality, maternal mortality,  women’s education, per capita food availability, and overall poverty. These seemingly contradictory results are often described as the “Asian Enigma”. It has frequently been argued that one of the explanations of the poor nutritional status of children and adult women in South Asia is the disadvantageous position of women and the occurrence of female discrimination in countries in this region.

In the present report an analysis is made of undernutrition prevalence rates in males and females (children, adolescents, adults), as these occur in South Asia and Sub Sahara Africa,  and it is investigated whether this reveals a distinct nutritionally disadvantageous position of women relative to men in South Asia. It is followed by an analysis of the prevalence and characteristics of undernutrition in populations from Asian descent in other parts in the world, in particular South Africa, Fiji, the United States, and England.

Results indicate that in children and adults differences in nutritional status between males and females are generally small, both in South Asia and in Sub Sahara Africa. In adolescents, if anything, undernutrition prevalence rates are higher among males than females, both in South Asia and in Sub Sahara Africa. With respect to anaemia, which is generally for a large part attributed to iron deficiency, prevalence rates are almost consistently higher among women than among men, with in South Asia a somewhat higher ratio between the anaemia prevalence rate in males and females than in Sub Sahara Africa. The analysis of undernutrition prevalence rates among population groups of different ethnic descent living in the same country reveals generally higher levels of low Body Mass Index among adults form South Asian background. These higher rates of adult undernutrition cannot be explained on the basis of less favourable socio-economic characteristics, such as lower income or less access to food.

It is hypothesized that, apart from the possible negative effects of a low status of women,  there exists among people from Asian descent an ethnic predisposition for a low adult body mass index, which comes to expression under a relatively low or moderate level of standard of living. Through its effect on birth weight of babies, the low body mass index of women contributes to the high levels of child malnutrition in South Asia. One other identified factor which may also contribute to the high levels of undernutrition in South Asia is the occurrence of  deficiencies in micronutrients, such as iron, vitamin A, and zinc, caused by an overall low quality of the habitual diet, and in particular a very low consumption of meat. It is stressed that the current findings do not dispute at any degree the occurrence of female discrimination in South Asia, but its nutritional aspects are currently not fully understood and need further investigation.


Keyzer, M.A., V. Molini, G.J.M. van den Boom, ‘Risk minimizing index functions for price-weather insurance, with application to rural Ghana’, 31 pp.

Poor farmers find it difficult to cope with price-weather shocks through self-insurance, because they cannot afford to keep large stocks and to protect their crops through irrigation and other measures. Mutual insurance is not an option either, because all participants would be faced with the same price-weather conditions at the same time. The next option of market insurance is plagued by excessive monitoring cost in avoidance of moral hazard and adverse selection. Consequently, new types of insurance are needed.

Among the arrangements suggested, index-based insurance is currently receiving much attention. Index-based insurance offers an indemnification according to an index function that depends on agreed upon price and weather conditions rather than on an assessment of damage at individual farm level. Existing proposals and experiments present a synthetic index function whose effectiveness is established by assessing its capacity to stabilize revenues on the historical record.

The present paper proposes an approach that is different in that it enables the insurer to offer an indemnification that is optimal from the perspective of the farmer in preventing a fall below a specified poverty line and is self-financing up to a given subsidy. To this effect, we develop and apply a model that minimizes farmers’ risk of receiving an inadequate indemnification. The approach builds on methods from catastrophic risk management in insurance and support vector regression in statistics. It is applied to Ghana, where according to our database, 47 percent of the farm population fell below the poverty line. Simulations of index-based insurance show that, while a parametric form fits reasonably the ideal indemnification it reduces poverty by only 4 per cent. By contrast, the proposed semi-parametric forms perform much better in terms of fit and could reduce poverty by another 5-10 per cent points, depending on the regularization.between traditional fertilizers and supplementary micronutrients.

It is speculated that, in particular in Sub Sahara Africa, controlled and site-specific application of micronutrients to crops, in combination with well-balanced usage of traditional fertilizer (N, P and K) might help break the vicious cycles of low yields, poverty, and poor human nutrition.


Boom, G.J.M. van den, S. Pande, ‘User manual for the SAS-facility to plot maps’, 38 pp.

In recent years spatial analysis has gained considerable attention at SOW-VU and the plotting of maps is increasingly used to present research outcomes. This note discusses the facility that has been developed for this plotting using the SAS software (SAS Institute Inc., 2003). It is linked to the gridding, regression, classification and polling software (Keyzer, 2005) but can also be used as a standalone package. It is an extension and revision of an earlier version of the facility (Overbosch, 2006). The main revision concerns the construction of (administrative) boundaries to overlay the basic raster map of grid cells. Whereas these boundaries had to be provided as a separate data set, they are now identified directly from the basic raster map itself following the edges of grid cells at the boundary of a specified (arbitrary) grouping. The extension of the facility consists of additional color schemes and added options to position the legend, to scale the legend and the plot, and to plot at higher or lower resolutions. Also, the parts of the facility that used to be specific to a certain application have been eliminated whereby the facility is now generic and can be initialized and tested through a simple batch-file.


Keyzer, M.A., S. Pande, ‘Classification by crossing and polling for integrated processing of maps and surveys. An addendum to GRCP-software’, Working Paper 07-04. Amsterdam, SOW-VU, pp.

The paper presents numerical procedures to compute conditional frequencies on large scale maps and surveys comprising mostly qualitative data, much in the way commonly done for ballots but with sufficient generality and computational power to support simultaneous processing of  a large number of questionnaire entries, for categorical as well as for real-valued data. Specifically, we address the curse of dimensionality inherent in the crossing of a large number of qualitative answers by focusing on highest frequency outcomes and by applying sorting routines from database management in computation. In case real-valued explanatory variables appear jointly with categorical variables, we make use of kernel smoothing, which allows among others for the representation of spatial correlation, under a window size that maximizes the goodness of fit. The appendix describes the fully GAMS-controlled operation of the software tool, a new component of SOW-VU’s GRCP-package for grid level calculations, regression and classification. The tool also has options for spatial interpolation, for projection of survey data on maps, and vice versa, as well as for calculations on recursive sequences of conditioning variables (Markov chains).


Molini, V., M.A. Keyzer, G.J.M. van den Boom, W. Zant and N. Nsowah-Nuamah, ‘Social safety nets and index-based crop Insurance: historical assessment and semi-parametric estimation for Northern Ghana’, 33 pp.

Our paper considers past and present social safety net arrangements in Northern Ghana, where village communities are poor and tend to face risks that affect virtually all members and, consequently, call for safety net arrangements beyond individual and mutual insurance. After a brief review of history, we assess the possible contribution of index-based crop insurance in this respect that bases its indemnification on objectively and easily measurable variables, such as rainfall data and prices at major markets, unlike standard insurance contracts which are individualized and have much higher transaction costs. After noting that safety net arrangements should be effective, timely and well-coordinated in securing (i) entitlements (in kind, cash or as indemnification payments from insurance) for the poor, (ii) funding (through taxes or private contributions, possibly insurance premiums), and (iii) delivery of necessities such as staples to all households, we observe that index-based insurance could play a useful role in entitlement and to a lesser extent in funding but that it does not in itself provide for adequate delivery, meaning that under supply shocks such as droughts the indemnity payments could cause prices to rise and channel away scarce food from the uninsured to the insured. This is particularly relevant in Northern Ghana, where rainfall varies strongly, subsistence farming is dominant and few remittances flow in. Turning to the modalities of index-based insurance, we seek to improve on existing indemnification schedules that are commonly specified synthetically or estimated in a simple parametric form. Via an adaptation of available kernel learning techniques, we can estimate a schedule that minimizes farmers’ risk of seeing their income drop below the poverty line. This schedule depends on selected index variables through a perfectly flexible functional form that maintains self-financing up to a prespecified subsidy. We test the scheme’s performance as a safety net for Northern Ghana on the basis of the size of its basis risk and its capacity to reduce poverty through full sample estimation as well as bagging. Although our schedule reduces by 20 percent points the poverty incidence from an initial level of 63 percent, and proves to be quite robust under bagging, basis risk and associated poverty remain considerable, reflecting the limited capacity of the variables selected to eliminate it.


G.B. Overbosch, Inside the Map Factory: Note on making maps with SAS, 30 p.

In recent years spatial features have gained considerable attention in economic research at SOW-VU, and maps have increasingly been used to present spatial data or outcomes of scenario simulations. Consequently, a growing need was felt for software facilities that would simplify the making of maps, as well as would allow for making these maps automatically as final step of scenario simulations. This note discusses such facilities for making maps in SAS, a statistical software package extensively used at SOW-VU and elsewhere in economic research

M. Nubé and R.L. Voortman, Simultaneously addressing micronutrient deficiencies in soils, crops, animal and human nutrition: opportunities for higher yields and better health, 48 p.

A study was done on the relationships between micronutrient deficiencies in soils, in food and fodder crops, in animal nutrition, and in human nutrition. While positive relationships between low soil contents and the occurrence of human deficiencies are since long known to exist for micronutrients such as iodine and selenium, the situation is less clear for other micronutrients, including iron and zinc. 

Results indicate that not only for iodine and selenium, but also for zinc, there are strong indications of direct linkages between soil zinc contents and human zinc status. For iron, uptake by plants and also by humans is highly complex, yet a positive relationship between poor iron availability from soils and the occurrence of human iron deficiency cannot be excluded. For other micronutrients, such as copper, magnesium and manganese, linkages between soils, food crops, and animal and human nutrition may well exist but currently available information is insufficient to analyze such relationship in full detail.

It is further explored to what extent there are possibilities for simultaneously increasing crop yield and crop micronutrient contents through application of micronutrients to crops. Where this is possible, there would be a simultaneous benefit both for farmers in the form of higher incomes and for consumers in the form of better health.

Results show that for zinc, there are strong opportunities for both increasing crop yields and increasing crop micronutrient contents for major crops such as wheat, maize and rice. For iodine, application to fodder crops might result in better growth of animals, higher iodine contents of animal products such as meat and milk, and on its turn contribute to improved human iodine status. For iron, positive effects on both yields and crop iron contents are, in principle, possible, but chemistry and physiology of iron is highly complex, and there is a strong need for further research into agriculturally based approaches aimed at increasing crop iron contents. For selenium, experiences in Finland and China have shown that crop selenium contents, and as a result human selenium intake, can be increased by adding selenium to common fertilizer.

When selecting appropriate technologies for applying micronutrients to crops, best results are most likely to be obtained by applying micronutrients separately on a one by one basis to crops, or in combinations of a few carefully selected micronutrients, in order to prevent antagonistic interactions between traditional fertilizers and supplementary micronutrients. It is speculated that, in particular in Sub Sahara Africa, controlled and site-specific application of micronutrients to crops, in combination with well-balanced usage of traditional fertilizer (N, P and K) might help break the vicious cycles of low yields, poverty, and poor human nutrition.


Keyzer, M.A., M. Nubé, C.F.A. van Wesenbeeck, ‘Estimation of calorie intake in Africa: methodology, findings and implications for Africa’s record’, 29 p.

Whereas the international community has come in great need of obtaining reliable estimates on food and nutrition, after adopting the Millennium Development Goals while committing to monitor their realization, it appears that essential data are lacking and that current estimates depend on highly tentative assumptions. The present paper reports on estimation of undernutrition and calorie intake in Sub-Saharan Africa (SSA). It also needs tentative assumptions but unlike the official estimates published by FAO, invokes as an additional data source the anthropometric measurements available in the Demographic and Health Surveys (DHS) commissioned by USAID. Besides comparing the estimates on undernutrition, we also assess the calorie intake and the implication for agricultural production, and purchasing power in urban areas, all in the year 2000. We find substantially lower undernutrition rates than reported by FAO, while per capita consumption is only little higher on average but significantly higher in Central and Eastern Africa. This suggests that the agricultural production and the purchasing power of farmers should by and large have managed to keep up with both the rise in farm population and the growth of cities, which in turn would indicate that non-agricultural income has grown over the last decades as well. Some preliminary calculations on this basis show that common GDP estimates of past and projected growth in SSA may be too conservative as well. However, these positive findings need qualification. First, the very high child mortality in SSA as compared to other developing regions may in part explain the more favorable nutritional status of the surviving population. Second, large numbers of refugees are sustained from foreign food aid. Finally, the AIDS/HIV epidemic creates uncertainties in many respects, including the work capacity of the population and the demography itself.


W.C.M. van Veen, P.J. Albersen, G. Fischer (IIASA) and Laixiang Sun (IIASA), Data set for the Chinagro welfare model: structure and composition, 41 p.

The paper describes the structure and contents of the data set that has been compiled for the Chinagro general equilibrium model. The model is a 17-commodity, 8-region welfare model with 6 income groups per region and agricultural supply represented separately for as much as 2433 counties (virtually all). In every county several land use types in cropping and livestock production are distinguished, with in total 28 aggregate outputs. Apart from the 17 tradable commodities, local commodities such as manure, household waste and crop residuals are accounted for. Data are collected from various basic sources, reclassified into Chinagro taxonomy and made consistent for baseyear 1997. Consistency requirements for commodity balances and price margins follow from the general equilibrium structure of the model. The same data set is also used to provide benchmark information at county level to the spatially explicit partial models that have been developed as a parallel activity in the Chinagaro project. The construction of the data set is programmed in GAMS, with a modular set-up that shows the steps from source data to final data and facilitates revisions of specific components. The paper has a statistical appendix with a summary of baseyear tabulations.

M.A. Keyzer and W.C.M. van Veen, Towards a spatially and socially explicit agricultural policy analysis for China: specification of the Chinagro models, 61 p.

The Chinagro-project has developed a series of single-commodity, spatially explicit partial equilibrium models covering China with around 94000 grid cells of 10-by-10 kilometer surface, as well as one 17-commodity, 8-region general equilibrium welfare model with 6 income groups per region and agricultural supply represented separately for as much as 2433 counties (virtually all), and describing, for every county, 14 land use types in cropping and livestock production, with 28 aggregate outputs. Both models are run in parallel. The present paper describes the structure of the general equilibrium model and its relation to the partial versions. It also shows how to compute the solution of this very large model, and how to calibrate it. The model is formulated as a welfare program of an open economy with transportation costs between regions and with tax distortions. It operates on an annual basis, evaluating a solution under given scenario-trends with respect to land availability, demography, economic growth, technological progress, international prices and government policies. Regarding validation, the Chinagro-model fully replicates for every county and region of China at 1997 base-year conditions, adequately mimics changes over the period 1997-2003 and provides interpretable results until 2030. It has fully integrated software that efficiently runs from basic data, via solution algorithms and simulation, to automatic production of detailed county-level maps and tabulation of results. The Chinagro-model is programmed in GAMS. Maps and other tabulations are controlled in GAMS or in user-friendly menus, even though they actually run in DOS, Fortran and SAS.

M.A. Keyzer, Rule-based and support vector (SV-)regression/classification algorithms for joint processing of census, map, survey and district data', 88 p.

The paper considers the updating of datasets with census data and grid maps, as well as the parameter estimation and micro-simulation of models based on these datasets, while accounting jointly for (i) survey data such as expenditures from households, rainfall data from weather stations and data on land classification; and (ii) aggregate data at district or sector level as published by statistical agencies. Various rule-based gridding, regression and classification algorithms to conduct these tasks are presented that rely on kernel smoothing, nearest neighbor and support vector techniques. Beyond this incorporation of census and grid map data within the survey based regression/classification itself, as opposed to their use in prediction only, the intended contribution is to allow for formulations with discrete data other than single, binary and multiple class choice, such as ranking with known and unknown class bounds and a consumer model with discrete choice and to present GRCP, a software package designed for this purpose, through which the user can readily navigate between gridding, regression and classification, controlling by simple  GAMS commands all operations from the basic data down to the automatic preparation of tables and geographic maps with appropriate legends and 2D and 3D plotting of functions.


M.A. Keyzer,
'Welfare program of the Jordan basin: formulation, decomposition, and solution through a homotopy algorithm', 45 p.

The paper specifies a spatially explicit model with a very large number of, say, about 50000 cells, to represent hydrological flows in the Jordan basin within a welfare context. Its main emphasis is on specifying a new homotopy algorithm to solve this model in the presence of market imperfections (missing markets) through which no payment is received for outflows from designated sites of arbitrary spatial configuration (lines, dots, closed objects). Hence, it can represent any coalition of territories within the basin, who refuse to pay to downstream districts as well as imperfections in the monitoring system. It can be run for several years in sequence, as the future can be considered to lie downstream of the present, and in the absence of imperfections will yield an intertemporally efficient path. The equilibrium is unique. The homotopy algorithm is based on price iteration over a sequence of dual convex programs, which are solvable in closed form, on a site-by-site basis, in decreasing order of elevation. We also discuss extensions, including market imperfections, dynamics and the treatment of uninhabited sites.


M.A. Keyzer, '
Support Vector Regression: a bridge between micro and macro?', 27 p.

Support Vector (SV)-regression, a common tool in statistical learning, estimates functions as linear combinations of given (nonlinear) kernel functions. To this representation corresponds through duality a weighted, possibly infinite sum of generally unknown eigenfunctions. The paper proposes to start from given eigenfunctions, as these can be thought of as representing a known micro based model, say, the sum of known Marshallian demand functions of all individual consumers and applies SV-regression to estimate a more compact and macro representation to replace this micro-model. The method applies to any single valued and bounded, possibly discontinuous eigenfunction and we show that it can accommodate multiple equations as well as constraints on parameters and function derivatives. The existing SV-algorithm can solve this micro-based problem at moderate computational cost, even when the number of individuals is very large, say, running in the millions because it can store all the necessary information in a Gram-matrix whose dimensions do not depend on this number. Among various problems in the range between micro and macro, optimal aggregation is the most micro. It amounts to finding an optimal, as sparse as possible vector of weights, so that the individuals with positive weights fit the true aggregate model with sufficient accuracy. Next, comes the competition between the micro-model and a given macro-model, and a we gradually attribute a more modest to the micro a priori, we eventually reach pure kernel approaches, the Gram-matrix is postulated as a covariance-like measure of the distance between observations and is applied to the macro-model directly. Finally, beyond the macro models lay the essentially descriptive non-parametric data enveloping techniques that only postulate a Gram-matrix. While the same SV-algorithm of dual quadratic programming applies throughout, it appears that the consistency properties of the estimators become weaker along this path, and whereas stochastic quasigradient methods achieve convergence almost surely, SV-regression only reaches convergence in probability.


M. Nubé, 'Relationships between undernutrition prevalence among children and adults at national and subnational level', 19 p.

Anthropometric information on the prevalence of undernutrition in children, such as the prevalence of a low weight-for-age, is increasingly used as an indicator of poverty or food insecurity at the level of countries or regions within countries. However, little is known how, at this level,  the nutritional status of children is related to the nutritional status of other age segments such as adolescents or adults. Without such information, the undernutrition prevalence among children cannot be interpreted as an overall indicator of poverty or food insecurity.

In the present study, utilizing in particular results from the Demographic and Health Studies, an analysis is made of the relationships between undernutrition prevalence rates among children and adults, both at the level of countries and at the level of smaller geographical subunits within countries (districts, provinces). At the level of countries, results reveal a strong positive relationship between undernutrition prevalence rates among children and adults. These results are in support of the concept that national undernutrition prevalence rates among children can be considered a proximate of overall nutritional conditions in a country.

At the level of smaller geographical units relationships are different. High levels of undernutrition among children may or may not be associated with high levels of undernutrition in adults. It is hypothesized that a combined high prevalence of undernutrition both among children and adults is in particular associated with insufficient household level access to food, while a combination of a high level of child undernutrition with an adequate or  reasonable nutritional condition of adults points to non food factors, such as poor water and sanitation conditions and poor education, as major causes of undernutrition.



Albersen, P.J., H.E.D. Houba  & M. A. Keyzer: Pricing a raindrop in a process-based model: Theory and application for the Upper-Zambezi. August 2003, 63 p.

A general approach is presented to value the stocks and flows of water as well as the physical structure of the basin on the basis of an arbitrary process-based hydrological model. This approach adapts concepts from the economic theory of capital accumulation, which are based on Lagrange multipliers or shadow prices that reflect market prices in the absence of markets. This permits us to derive a financial account complementing the water balance in which the value of deliveries by the hydrological system fully balances with the value of resources, including physical characteristics reflected in the shape of the functions in the model. The approach naturally suggests the use of numerical optimization software to compute the multipliers, without the need to impose an immensely large number of small perturbations on the simulation model, or to calculate all derivatives analytically. A novel procedure is proposed to circumvent numerical problems in computation and it is implemented in a numerical application using AQUA, an existing model of the Upper-Zambezi River. It appears, not unexpectedly, that most end value accrues to agriculture. Irrigated agriculture receives a remarkably large share, and is by far the most rewarding activity. Furthermore, according to the model, the economic value would be higher if temperature was lower, pointing to the detrimental effect of climate change. We also find that a significant economic value is stored in the groundwater stock because of its critical role in the dry season. As groundwater comes out as the main capital of the basin, its mining could be harmful.


R.L. Voortman & J. Brouwer, 'Coversand land ecology and site-specific millet yield functions in SW Niger',  29 p.

In this study, we apply novel tools for data exploration to a detailed environmental and crop yield database from a spatially very variable farmer's field in SW Niger. Rather than the entire field, as we did in earlier studies, we now consider two individual field parts that derive from different parent materials (coversands). The objective is to verify if site-specific yield function analysis leads to a better understanding of soil chemistry-yield relationships and if these then allow fine-tuning of possible external input technologies. Our findings show that the type of variables that explain millet yield well across soil types are also important at the level of individual field parts. However, the functional forms for the two parts are quite different: they conform better to theoretical knowledge than the overall equation; they are easier to interpret; they indeed identify site-specific operating mechanisms; and, consequently, they lead to different promising input technologies. These findings confirm earlier observations that farmers in this environment need to apply low-tech precision farming in order to achieve greater efficiency of external input use. Regarding analytical methods, this study highlights the importance of the selection of variables and the functional form, which should not be imposed from the outset, but derived from the data themselves. We further conclude that process-based crop growth models, as yet, cannot accommodate the full complexity of soil chemistry and its effect on crop yield. Empirical on-farm research, as presented in this paper, could be the answer and identify the most significant soil fertility complexities that need to be addressed in real-world situations aiming at the design of external input technologies that combine low cost with substantial yield improvement.


M. Nubé, 'Food security in Sub Sahara Africa', 31 p.

A comparison has been made of developments with respect to food security over the period 1980-2000 in four regions of Sub Sahara Africa: coastal West Africa, the Sahel, Central and Eastern Africa, and Southern Africa. On the basis of trends in food production and consumption and trends in human outcomes (child malnutrition, life expectancy), it appears that developments have been relatively positive in coastal West Africa (with the exception of Liberia and Sierra Leone), and also in the Sahel region, and relatively unfavorable in Central and Eastern Africa and Southern Africa. Regional differences have been identified with respect to natural resources conditions, with respect to the occurrence of emergencies and disasters, and with respect to international assistance in the form of food aid. It is concluded that awareness and understanding of  regional differences between various regions of Sub Sahara Africa is important in further analysis of food security developments and prospects of Sub Sahara Africa.


CPB Report 2003/2
Keyzer, M.A., M.D. Merbis and M. van ‘t Riet ‘Will the proposed CAP reform reward green services?' pp. 34-40.

In 2003, the EU member states are expected to agree on a reform of its Common Agricultural Policy that could have far-reaching consequences for the ongoing Doha round of WTO-negotiations as well as the rural countryside all over Europe. There is growing appreciation of the role of agriculture in maintaining rural landscapes and keeping the rural country-side inhabited and maintaining cultural heritage: agriculture is seen to deliver "green services". The mainstay of current EU proposals for CAP reform is to decouple the present support by transforming the area and animal premiums into a single farm-specific payment. This offers a unique opportunity for a transition to a system that makes these payments contingent on the provision of green services, and can liberalize trade while keeping the countryside populated. However, the current proposals are insufficiently articulate for a credible transition to such a system, and even contain elements that may hamper this transition.




G.J.M. van den Boom, N.N.N. Nsowah-Nuamah & G.B. Overbosch "Curative health care utilization in Ghana. A multinomial analysis of equitable access opportunities", 29 p.

In this paper we provide an overview of the curative health care system in Ghana, discuss the determinants of its utilization, and simulate utilization patterns under alternative access policies. The figures point to two major challenges to improve access, viz. the geographical coverage of health facilities and health workers, and the financing of necessary health care of the poor. Ghana has a fair number of facilities (about 1 per 10,000 inhabitants and per 140 km2) but doctors are scarce (about 1 per 11,000 inhabitants) and both are highly unevenly spread in favor of urbanized regions. Almost one third of the population lives outside a 5-km radius of medical assistants and nurses, while one quarter is more than 15 km away from a doctor, who is the preferred health care provider when available. Moreover, the private cost of consultation and treatment is significant and shows large variation, restricting access to necessary health care for the poor. We use multinominal logit and probit techniques to explore opportunities to improve access. Simulations show positive, but modest, effects of reducing the distance to orthodox facilities in rural area and of replacing part of the private cost by national health insurance with income dependent premium. Thus, at the current stage of development, the provision of adequate health care at affordable prices to all remains a formidable challenge.


R. L. Voortman, J.  Brouwer, and P. J. Albersen “Characterization of Spatial Soil Variability and its Effect on Millet Yield on Sudano-Sahelian coversands in SW Niger”

Very local spatial soil variability on Sudano-Sahelian coversands hampers the interpretation of agro-pastoral research and is an obstacle for the dissemination of research findings. Site characterization and the establishment of site-specific plant performance and yield responses to external inputs are therefore of fundamental importance to achieve agricultural development. In an earlier paper we applied novel tools for data exploration such as non-parametric kernel density regression and spatial econometrics to spatially explicit data on topsoil N, P and K and could explain 81 percent of the millet yield variation. However, the macronutrients explained a modest portion of millet yield only, while the good explanatory power largely derived from spatial autocorrelation. In the present paper we use the same tools to identify and characterize the sources of soil variability/spatial autocorrelation within a single farmers’ field. Three soil types are identified, the essential differences of which refer to the cation exchange complex: different proportions of the cations (Ca, Mg, K and Na), in combination with a different Al saturation profile as well as absolute levels of K. Millet yields are high when low levels of topsoil Mg and Na occur in combination with high levels of Al saturation in the deeper subsoil. Conversely, yields are low when Al saturation in the subsoil is lower and when the proportion of Mg and Na is high. Since Al saturation is caused by water percolation, the empirical evidence thus suggest that high levels of Mg and Na in the topsoil reduce water infiltration. The possible operating mechanism is the destabilizing effect these elements have on the clay fraction, which in turn is the cause of surface sealing. Moisture availability and the effect of surface sealing on seedling establishment may thus be a key source of local millet growth variation on Sudano-Sahelian coversands. The above variables explain millet yield better than N, P and K, without spatial autocorrelation being present in the residuals, and thus constitute the true source of millet growth variability. The variables involved and stratigraphical evidence suggest a parent material connection to the local soil variability: coversands of different source materials and age occurring as shallow layers. We propose further experimental research to investigate if Ca and K applications, which reduce the proportion of Mg and Na in the topsoil, ameliorate moisture availability, improve seedling establishment and raise crop yields.


M.A. Keyzer & C.F.A. van Wesenbeeck "Equilibrium selection in games: the mollifier method". 28 p.

This paper proposes to use a mollifier mapping, a general and flexible tool for smoothing the response by the players of a game for the selection of a robust equilibrium (refinements) as well as for procedures converging to a unique equilibrium: tracing, smoothing of best response, testing of evolutionary stability and superimposing of aggregate shocks on replicator dynamics.
With respect to selection, first, we note that the best-response mapping is almost everywhere single-valued. Secondly, we show that, after integration over perturbations with given density the mollified mapping becomes single-valued everywhere. Thirdly, taking the density of perturbations to be continuously differentiable, we prove that this implies continuous differentiability of the mollified best-response mapping, as well as generic local uniqueness of the Nash equilibria of the mollified game. Fourthly, we prove that the Nash equilibria of the mollified game converge to that of the unperturbed game. Finally, we show that the mollifier can also be applied directly to the unperturbed game, in which case the procedure amounts to determining the averaged subgradient of the value function describing the optimization problem of the players in the game.
Next, turning to the other applications, we consider the use of the mollifier in the tracing approach, where it ensures that the equilibrium mapping is almost everywhere single valued, avoiding instability of one kind. Second, we find that application of the mollifier to best-response dynamics smoothens the mapping without requiring additional assumptions that keep reactions in the interior of the strategy space. Furthermore, the mollified mapping can be estimated consistently on the basis of repeated observations, and this estimation procedure has a transparent interpretation in terms of fictitious play. Thirdly, within evolutionary game theory applications, the mollifier extends recent contributions that allow heterogeneity within populations, for example local interaction models. In particular, not all strategies have to be present everywhere in the population, and more general differential equations can be included, such as forms that inhibit discontinuities when a zero value is attained.


Keyzer, M.A. 'Food safety, labeling and market efficiency',
47 p.

The safety of a given food item may be expressed as a product characteristic that is uncertain at the time of transaction. It can be represented through a probability distribution. Improving food this safety amounts to changing the shape of the distribution, through prevention measures. Extreme risks will lead to prohibition, but if financial compensation of losses is conceivable, the situation becomes more complex, as economic agents have to decide simultaneously on quantities and probabilities, for example in the calculation of expected utility and expected profits. This creates a non-convexity that may cause market failure, even in situations where all safety characteristics are priced competitively. The paper follows two lines of investigation to avoid this problem. The first considers restrictions on the functional forms of utility and prevention functions, and the second introduces the institution of standard-based labeling as a filter on possible distributions, essentially to discretize them. Starting from a partial equilibrium model with one consumer, one producer and one unsafe commodity, in which various options prove effective, we gradually extend the analysis to cover several  agents and commodities, eventually arriving at a general equilibrium formulation. At every step, we eliminate unsuitable options, eventually to find labeling as sole viable option.


B.G.J.S. Sonneveld and M.A. Nearing, 14p. A non-parametric/parametric analysis of the universal soil loss equation.

Due to its modest data demands and transparent model structure the Universal Soil Loss Equation (USLE) remains the most popular tool for water erosion hazard assessment. However, the model has several shortcomings, two of which are likely to have prominent implications for the model results. First, the mathematical form of the USLE, the multiplication of six factors, easily leads to large errors whenever one of the input data is misspecified. Second, the USLE has a modest correlation between observed soil losses and model calculations, even with the same data that were used for its calibration. This raises questions about its mathematical model structure and the robustness of parameters that are implicitly assigned to the model. This paper, therefore, analyzes if the USLE could benefit from mathematical model transformations that, on one hand, mitigate the impact of incorrect input factors and, on the other hand, result in a better fit between model results and observed soil losses. For the analysis we revisit the original data set and consider the USLE factors as variables rather than their common interpretation as parameters. We first use both non-parametric and parametric techniques to test the robustness of the implicit parameter assignments in the USLE equation. Next, we postulate alternative mathematical forms and use parametric test statistics to evaluate parameter significance and model fit. A tenfold cross validation of the model with the best fit tests the sensitivity of the parameters for inclusion or exclusion of the data. The results show that the USLE model is not very robust, however, only slight model improvements are obtained by drastic modifications of its functional form, thereby sacrificing the simple model structure that was intended by its designers.


M.A. Keyzer, "The emergence of crime under decollectivization: a general equilibrium argument", 30 p.

The paper formulates a static general equilibrium model in which crime is represented as a unilaterally enforced income transfer. Individuals choose between an honest and a criminal career while investing in prevention to avoid robbery or arrest. We explore the conditions under which the criminal career becomes unattractive. We show, without recurring to higher penalties, better competition and improvement of employment opportunities, and under a broad class of market distortions, that crime is eliminated if individual robbers are obliged to share the proceeds with their peers, while the robbed also share their losses, essentially because this taxes away their gains. Conversely, the individualization of incentives under decollectivization is seen to disrupt this redistribution mechanism and hence to foster crime.


Raymond J.G.M. Florax, Roelf L. Voortman & Joost Brouwer, "Spatial Dimensions of Precision Agriculture: A Spatial Econometric Analysis of Millet Yield on Sahelian Coversands", July 2002, 31 p.

The identification of local soil variability caused by within-field differences of macronutrients and ecological features is of paramount importance for the effectiveness of precision agriculture. We present several spatial statistical and econometric techniques to capture local differences in soil variation, ecological characteristics, and yield more effectively than the analytical techniques traditionally used in agronomy. The application of these techniques is illustrated in a case study dealing with precision agriculture in the West African Sahel. The production of millet on acid sandy soils constitutes a typical example of low soil fertility areas exhibiting small absolute but large relative differences in crop production conditions over short distances.


M.A. Keyzer, Kernel learning for poverty mapping: an introduction. July 2002, 37 p

Kernel learning offers tools for semiparametric regression and classification that have made their proofs in the field of pattern recognition, especially in optical reading, voice recognition, and genomics (Schölkopf and Smola, 2002). In this paper, we describe its possible contribution to the construction of poverty maps, especially its potential to improve the flexibility of the functional forms in regression and shows how to use census information for estimating these. We have implemented the kernel learning algorithms in GAMS and found to be numerically effective, as they essentially rely on convex quadratic programming.


M.A. Keyzer, Shaping the odds: insecurity management in general equilibrium. July 2002, 31 p.

Insecurity management differs from risk management in that it takes the distribution of uncertain events to be under the control of economic agents, rather than given. This effectively turns these distributions into public goods for all who are affected by them, and therefore calls for dedicated institutional arrangements, to avoid market failure and inefficiency. Once distributions become variable, market failure may also result via the product of the density and the utility function in expected utility maximization, that loses concavity. The paper discusses two ways to maintain concavity, one in which the actions other than those that shape the distributions can be postponed until after uncertainty has been revealed, another in which new institutions restrict the possible probability profiles to a finite number of alternatives. Under central planning, this can be effectuated via standardization. Standard-based labeling offers a decentralized solution that enables individuals to compose a mix from given profiles, either collectively, or individually.


M.A. Keyzer: Consistent calculation, valuation and calibration of surface flows in spatially explicit and dynamic models. August, 41 p.

The paper presents an algorithm to solve a spatially explicit dynamic model in which substances flow downward over a relief of arbitrary shape and material balances are maintained in every cell of the spatial grid. On the basis of the downstream extraction prices and amenity values, the algorithm, implemented in Fortran, also calculates in a numerically effective way the upstream prices of the flows, for problems of very large dimensions, say, with a few million cells. We distinguish between a linear version, in which the directional outflow fractions of every cell are kept independent of the flows, and a nonlinear one, which may also exhibit discontinuities, and can accommodate an arbitrary valuation criterion that is not necessarily separable over sites. Next, we indicate how, at given prices of the flows, we can in a fully decentralized calculation on a site-by-site basis, impute prices of stocks and generally of model parameters, in the static version of the model as well as in the steady state of a dynamic version, with carryover stocks. Moreover, we present conditions under which the algorithm can determine the optimal extraction at every site as well optimal routing between sites, and locate sites where uphill pumping would be profitable. We also indicate that the valuation criterion that is being maximized might be a likelihood function, in which case the model parameters become the key variables for valuation, with their prices pointing to directions of change that would improve the fit to observed data.


B.G.J.S. Sonneveld. Formalizing the use of expert judgments for land degradation assessment: a case study for Ethiopia
November, 25 p.

Expert judgments are potentially a valuable source of information in land degradation assessment, especially in those areas where data paucity impedes the utilization and validation of quantitative models. However, these expert opinions are also much disputed because they are not tested for consistency, abstain from a formal documentation, while its quantitative interpretation is inherently unidentifiable. In this paper we aim to evaluate and formalize the use of expert judgments in order to conduct a nationwide water erosion hazard assessment in Ethiopia. We therefore test the experts' judgment for its consistency, the correlation with quantitative observations on soil loss and its reproducibility. The study uses an Ethiopian and an international data set for which groups of experts gave qualitative judgments on water erosion hazard, for well-described sites under different types of land uses. The experts have a relatively high consistency in their judgments on land degradation for similar sites. Comparing the ranked qualitative expert opinions to quantitative soil losses reveals that particularly the boundaries of the middle classes vary widely between experts and comprises a wide range of soil loss values. Reproducing expert opinions with an ordered logit model shows a reasonable accuracy in predicting the presence or absence of erosion, but the model is less precise in distinguishing between the higher erosion classes. In 58 per cent of the cases, the model gives a similar classification as the experts, in 19 per cent the model gives a higher and, more seriously, in 23 per cent a lower erosion class. Mapping the model results for Ethiopia demonstrates a high erosion hazard for land under annual crop cultivation, while erosion under perennial crops, rangeland and forest is absent or moderate. The likelihood of selecting the correct hazard class for rangeland is relatively high but low probabilities prevail for erosion classes of other land uses.


Nubé, M. & G.J.M. van den Boom: Gender and adult undernutrition in developing countries. November,  21 p.

Information on the prevalence of undernutrition in adults in developing countries is mainly restricted to data on women. Literature reporting on the occurrence of female deprivation in developing countries, in particular in South Asia, suggests that differences between undernutrition prevalence in adult men and adult women might occur, but systematic information on the subject is lacking.
The study compares undernutrition prevalence rates, based on prevalence of low body mass index (BMI<18.5), in adult men and adult women in developing countries. Regional comparison is made between the main developing regions Sub Sahara Africa, South/Southeast Asia and Latin America.
The study uses data as reported in sixty, in most cases, small-scale studies (divided over the three developing regions), in which anthropometric information has been collected in adult men and women within one and the same community.
Results indicate that, in general, prevalence rates of undernutrition are rather similar in adult men and women. However, there are regional differences. In communities in Sub Sahara Africa prevalence of low BMI is, on average, a few percent higher in men than in women, in South/Southeast Asia the reverse is the case. In some communities differences in undernutrition prevalence between men and women are exceptionally large.
It can be concluded that, in general, information on undernutrition prevalence in women can be considered a proxy for undernutrition prevalence in all adults, men and women together. Yet, the finding that in South/Southeast Asia undernutrition prevalence is, on average, higher in women than in men, supports the concept of female deprivation in the region. Where large differences between prevalence of low BMI between men and women occur, gender specific policies aimed at reducing undernutrition should be considered.


G.B. Overbosch, N.N.N. Nsowah-Nuamah, G.J.M. van den Boom & L. Damnyag, ‘Determinants of antenatal care use in Ghana. 19 p.

We investigate the determinants of antenatal care use in Ghana using a large-scale living standard survey. Most previous studies on the subject have used surveys that focus on demography and fertility, and have used approximate indicators of economic variables such as income and cost of consultation. This leads to an overestimation of effects when explanatory factors pick up the effect of underlying economic conditions. We describe antenatal care demand as a three level nested multinomial logit model that includes more appropriate economic explanatory variables. The estimation results show that indeed income, cost of consultation and in particular travel distance to the health care facility are significantly associated with the demand for antenatal care. Use of sufficient antenatal care can thus be promoted effectively by extending the supply of antenatal care services in the rural area. In addition, education of the mother is positively associated to choice for sufficient antenatal care, while women having more pregnancy experience tend to underutilize antenatal care. This suggests that campaigns to promote sufficient antenatal care should pay special attention to education and to women who already gave birth. The results further indicate that, in contrast to findings elsewhere, a special targeting of antenatal care according to religion seems unwarranted.


M. Nubé & G.B. Overbosch,  Health in Central West African Countries: Commonalities and Disparities.  19 p.

Five Central West African countries, Burkina Faso, Côte d’Ivoire, Ghana, Mali and Togo, cooperate in the Réseau-SADAOC (Sécurité Alimentaire Durable en Afrique de l’Ouest Central), which has as its major objective to increase food security in the region. The present report aims to review the health and nutrition situation in the five countries, to identify major bottlenecks in the functioning of the health care system, and to suggest strategies for addressing identified problems.  
All five countries face serious health problems, but on the basis of most health indicators the coastal countries Côte d’Ivoire, Ghana and Togo perform somewhat better than the landlocked countries Burkina Faso and Mali. Health improvements, which took place over past decades, have stagnated in the 1990s.
Health care provision has a traditional and modern component. With respect to the traditional sector, a discrepancy exists between its perceived role and its rather small share in survey data. With respect to modern health care provision, the public sector dominates, with the private sector playing a complementary role, increasing in importance. Quantitative information on total availability of health care facilities (buildings, personnel) is scarce if not completely lacking, which is an impediment for policy development.
With respect to health expenditures, private expenditures are at least equal to, and in most countries higher than public expenditures. Private expenditures are mainly on drugs. In case of illness people often buy medicines without seeing a health care provider. Polices that affect availability, quality, prices, and intelligent usage of drugs, can play an important role in health developments in the SADAOC-countries.
In all countries public health sector reform continues. Important elements are the decentralization of health care provision, with less government spending on high-cost facilities in capitals and more funds being made available for primary health care, the transfer of responsibilities to the local level, and the raising of user fees. These developments are expected to make health care services more efficient and more responsive to local needs. Of specific interest are experiments that have the objective to separate  government tasks with respect to health care policy from its tasks on health care provision.
Access to health care services (both physically as well as financially)  is probably the most important condition for proper functioning of a health care system. With respect to physical access, for a large segment of the population there is still a need to reduce distance to a health care facility. With respect to financial access, public health care expenditures can, in principle be viewed as a form of health insurance. Both cost-ineffective treatments and inexpensive or largely predictable health treatments should be excluded from such an insurance scheme. For equity purposes, also inexpensive health treatments could be provided for free to the lowest income groups and to children. Finally, free or subsidized health services for government personnel should be financed separately from the basis public health insurance system.


W.C.M. van Veen ‘Solving an intertemporal Arrow-Debreu model under aggregate risk: implementation of an SQG-based algorithm’.  52 p.

The paper describes the implementation of a solution algorithm for a two-period, multi-commodity, multi-actor general equilibrium model with aggregate risk. Uncertainty is reflected by commodity-specific shocks, affecting the actors differently and leading to stochastic period-2 prices. The model has contingent markets of the Arrow-Debreu type, with constraints on borrowing and insurance. The algorithm is based on a proposal by Ermoliev and Keyzer to rewrite the model in Negishi-format and apply techniques from stochastic optimization. It has a deterministic inner loop for period 2, a stochastic-optimization middle loop for period 1 and a Negishi outer loop for intertemporal budgets. The current implementation, using standard functional forms for utility and investment, functions well. In the end, it appears to be mainly the speed of the inner loop that determines the success of the algorithm. Extension to more periods seems possible, via recursive methods, if the multiple inner loops can be solved rapidly.


IIASA Research Report RR-02-03
Albersen, P.J., G. Fischer, M.A. Keyzer and Laixiang Sun '
Estimation of agricultural production relations in the LUC model for China'. 51 p.

China's demand for grains has been growing rapidly during the past two decades, largely as a result of the increasing demand for meat. This raises the important question of whether in the coming years China will be able to satisfy these in-creasing needs. The answer to this question will have implications that reach far beyond China's borders, especially in light of China's accession to the World Trade Organization (WTO). The answer depends on many factors, including the policy orientation of the Chinese government, the loss of cropland caused by the ongoing industrialization and urbanization processes, and the effect of climate change on the country's agricultural potential.
To analyze these issues, the Land-Use Change (LUC) Project at the Interna­tional Institute for Applied Systems Analysis (IIASA) has been developing an in­tertemporal welfare-maximizing policy analysis model. This report presents the input–output relationships for agricultural crops in the model. The specified rela­tionships are geographically explicit and determine the crop output combinations that can be achieved under the prevailing biophysical conditions across China from given input combinations in each of some 2,040 counties based on data for 1990. The non-land inputs are chemical and organic fertilizer, labor, and machinery. Irri­gated land and rain-fed land are distinguished as separate land-use types. Distinct relationships are estimated by cross-section for the eight economic regions distin­guished in the LUC model. The biophysical potential enters as an asymptote in a generalized Mitscherlich–Baule yield function and is computed on the basis of an agro-ecological assessment of climatic and land resources, including irrigation. The chosen form globally satisfies the required slope and curvature conditions.
Estimation results show that all key parameters are significant and are of the expected sign. The calculated elasticities of aggregate output with respect to inputs quite closely reflect the relative scarcity of irrigated land, labor, and other inputs across the different regions. It also appears that if both the local population density and the distance to main urban centers are taken into account, the observed cropping patterns are generally consistent with profit maximization. The often-noted labor surplus is confirmed in all regions, particularly in the southern and southeastern regions.


IIASA Interim Report IR-02-021
Ermoliev, Yu., M.A. Keyzer and V. Norkin 'Estimation of econometric models by risk minimization: a stochastic quasigradient approach'. IIASA Interim Report 02-021, 27 p.

The paper presents a risk minimization approach to estimate a flexible form that meets a priori restrictions on slope and curvature by means of constraints on both the estimated parameters and the function values. The resulting constrained risk minimization combines parametric and nonparametric estimation and contains integrals and implicit constraints. Within econometrics, simulation has become a common tool to solve problems of this kind. However, it appears that in our case, the simulation approach only applies when the model is linear in parameters, has simple constraints on parameters and a quadratic risk function. To deal with other cases, we use a stochastic optimization technique known as the stochastic quasi-gradient method for stationary and nonstationary problems with Cesaro averaging. This method is also applicable to an expanding series of random observations, and produces asymptotically (weakly) convergent estimates.



K. Appiah-Kubi, N.N.N. Nsowah-Nuamah, and G.J.M. van den Boom 'Returns to education and experience in Ghana, 1987-1999: evidence from four rounds of the Ghana Living Standards Survey'

The study compares schooling, experience and labor market outcomes in a sample of over 20,000 Ghanaians aged 15 to 65, who have been interviewed during four rounds of the GLSS and who represent the labor force as a whole over the period 1987-1999. Estimates of the coefficient of the years of schooling in a standard human capital model are combined with data on public and private schooling expenses. We find an average social rate of return (ROR) to basic education as low as 3 per cent, while the social ROR to secondary and tertiary schooling is 16 and 3 per cent, respectively. With estimates of 6, 25 and 15, the private ROR are higher, especially for tertiary education. As regards experience, estimates suggest positive and gradually decreasing returns. The first year of experience is estimated to increase earnings by around 5 per cent, while additional experience gradually yields less until earnings reach a peak around 35 years of experience. The results suggest that educational sector reform in Ghana be highly concerned with improvements of basic education, with the upholding of the quality of secondary schooling, and with opportunities to increase the share of private expenses in secondary and tertiary education.


M.A. Keyzer 'Regulation and efficiency in the health sector: a general equilibrium analysis with endogenous risk'

Public authorities tend to be actively involved in the health sector through prevention programs, regulation of health insurance and control of access to treatment. Common explanations include asymmetries in information, the danger of epidemics and imperfect competition. This paper argues that these interventions may also be required to deal with the typical externality that probabilities depend on individual and collective decisions. We formulate a general equilibrium model, in which individuals face endogenous probabilities of incurring specified diseases and obtaining timely treatment. Health insurance organizations collect premiums and invest in treatment capacity, in accordance with the preferences of their customers, but because of limited treatment capacity and insurability, they may have to conduct lotteries to assign patients. The externality causes a non-convexity in the patient’s expected utility function. We show that the associated consumer demand functions are well behaved nonetheless, and that an equilibrium exists. Moreover, if bounds on insurability are not effective and the collective decisions on prevention follow Lindahl pricing, this equilibrium will be Pareto-efficient in terms of expected utility, despite the capacity constraints and the endogenous risk. We also discuss how subsidies, prohibition and equal access restrictions can be included, at the expense of efficiency.


R.L. Voortman and  J. Brouwer 'An empirical analysis of the simultaneous effects of nitrogen, phosphorus and potassium in millet production on spatially variable fields in SW Niger'

Low soil fertility is a major constraint for increasing millet production on the acid sandy soils of the West African Sahel. On these soils, all three macronutrients: Nitrogen (N), Phosphate (P) and Potassium (K), may be expected to limit crop yields. The important question is therefore: which of them is the most critical and would, if applied in small amounts, increase yields significantly? This paper addresses this question with an empirical approach, thus avoiding the commonly observed difficulty in the interpretation of agronomic research, caused by the extreme local soil variability which characterizes Sahelian coversands. We actually exploit soil variability by using novel non-parametric techniques for data exploration in combination with spatial methods of parametric model estimation. Apart from N, P and K, the effects of surface crusting, local topography, manure levels, farmer behavior and spatial dependence are taken into account, since these may confound the true effects of N, P and K. A quadratic formulation conforms best to the data and explains 81 percent of the yield variation. The equation highlights the importance of interactions among variables and thus confirms the possible impact of native soil conditions on the outcome of fertilizer treatments in experimental research. The results of much earlier, multi-year, research are confirmed remarkably well by this single year study. In addition, a spatially explicit assessment on the crop response to increasing nutrient levels highlights that blanket fertilizer applications are inefficient, because yield increases in some places will be accompanied by yield decreases at other sites. Cash-constrained farmers therefore have to resort to precision farming techniques to maximize returns from minimal external input packages. However, a large part of the good explanation of millet yield variability over space derives from spatial autocorrelation, and not directly from topsoil N, P and K. This calls for further research on the factors that affect millet yield and on the characterization and classification of sites, followed by experimental work to design site-specific fertilizer technologies.


M.A. Keyzer, M.D. Merbis, I.F.P.W. Pavel and C.F.A. van Wesenbeeck 'Can We Feed the Animals? The Impact on Cereal Markets of Rising World Meat Demand'.

The paper argues that current international projections of meat and feed demand may underestimate future consumption patterns for mainly two reasons. First, their demand projections are based on income extrapolation with an assumed demand elasticity, and on expert judgment. Second, feed requirements per unit of meat are taken to be fixed. Instead, we propose a structural specification of meat demand, that accounts for the differences in income between households within countries as well as for the nonlinear shape of the meat demand schedule since the poor segments of the population tend to abstain from meat consumption until their income reaches some lower threshold, while rich consumers become satiated beyond an upper threshold. Regarding feed requirements, we distinguish between traditional feeding technologies based on grazing, household residuals and harvest by-products, and more intensive livestock technologies. We formulate optimistic and pessimistic projections on technological advances in feeding efficiency, carcass weights, and offtake rates. Our finding is that under the growth rates of per-capita income assumed in the commonly accepted projections, world meat demand will be significantly underestimated in the coming thirty years, especially during the first half of this period, even under optimistic assumptions on technological advances. The fast increase in the demand for meat will together with the tendency towards urbanization make it more difficult to expand the branch of animal husbandry that feeds on residuals and grass. This creates a strong pressure on cereal markets, especially to satisfy demand in Asia.


G.B. Overbosch 'Health effects of health policies in Burkina Faso'.

In this paper I explore the possible effects of health policies in Burkina Faso on the health status of its population. Relations between health indicators for mothers and their young children and their determinants are estimated using data from a recent Demographic and Health Survey. Of the determinants, general education, the supply of safe water and sanitation, and the provision of health care can be influenced by health policies. In absence of price data the estimated equations include proximity indicators for health care facilities, for which three specifications are tried, but the results show only modest to low influence of these indicators. According to the estimations, better living standards have a positive impact on the nutritional status of mother and child, and reduce the prevalence of illness among children. Education of the mother positively influences her nutritional status and that of her young children, but does not seem to affect their illness frequency. Safe drinking water is positively related to the nutritional status of mothers and children, but its effect on illness of children is unclear. Toilet facilities do not seem to have any impact at all. Also a positive effect of health care on health is hard to establish, and health status is positively correlated to a few types of facilities only. Effects of private health facilities seem more pronounced than those of public ones, but may be biased due to selectivity. Of all vaccinations, only those against measles reduce the illness frequency of young children.


M.A. Keyzer, Yu. Ermoliev, and V. Norkin 'General equilibrium and welfare modeling in spatial continuum: a practical framework for land use planning'.

The application of continuous distributions from statistics in spatial modeling makes it possible to represent discrete choices in a spatial continuum, and to obtain efficiency results and competitive equilibrium prices where aggregate or discretized models fail. Along these lines, and combining principles established by Aumann and Hildenbrand in the sixties with recent results from stochastic optimization, the paper develops a practical modeling framework for land use planning and presents the associated stochastic algorithms for numerical implementation. We consider groups of consumers and producers whose activities are distributed over space, and who have to make decisions, say, about where to live, which marketplace to visit, and which infrastructure facilities to invest in. After presenting a general equilibrium model in which all consumers meet their own budget with given transfers, we focus on the case in which transfers among consumer groups adjust to support the maximization of a given social welfare criterion. It appears that this optimization problem becomes more tractable if it is treated as the minimization of a dual welfare function, that solely depends on prices but is evaluated after integration over space. Next, we apply the dual welfare function to represent (non-rival) demand that simultaneously benefits several agents, reflecting a general informational infrastructure as well as investments with uncertain outcomes. This leads to a minimax problem, in which the dual welfare function is to be minimized with respect to prices and maximized with respect to non-rival demand. Finally, we endogenize welfare weights jointly with prices to model, for example, a land consolidation process whereby none of the participants should lose relative to the initial situation, and the gains could be shared according to agreed principles. This gives rise to a bargaining problem whose solution can be found by jointly minimizing the dual welfare function over prices and welfare weights, subject to constraints.


B.G.J.S. Sonneveld and M.A, Keyzer 'Land under pressure: soil conservation concerns and opportunities for Ethiopia'.

This paper evaluates the future impact of soil degradation on national food security and land occupation in Ethiopia. It applies a spatial optimization model that maximizes national agricultural revenues under alternative scenarios of soil conservation, land accessibility and technology. The constraints in the model determine whether people remain on their original site, migrate within their ethnically defined areas or are allowed a trans-regional migration. Key to this model is the combination of a water erosion model with a spatial yield function that gives an estimate of the agricultural yield in its geographical dependence of natural resources and population distribution. A comparison of simulated land productivity values with historical patterns shows that results are interpretable and yield more accurate outcomes than postulating straightforward reductions in yield or land area for each geographic entity. The results of the optimization model show that in absence of soil erosion control, the future agricultural production stagnates and results in distressing food shortages, while rural incomes drop dramatically below the poverty line. Soil conservation and migration support a slow growth, but yet do not suffice to meet the expected food demand. In a trans-regional migration scenario, the highly degraded areas are exchanged for less affected sites, whereas cultivation on already substantially degraded soils largely continues when resettlement is confined to the original ethnic-administrative entity. A shift to modern technology offers better prospects and moderates the migration, but soil conservation remains indispensable, especially in the long term. Finally, an accelerated growth of non-agricultural sectors further alleviates poverty in the countryside, contributing to higher income levels of the total population and, simultaneously, relieving the pressure on the land through rural-urban migration.


W.C.M. van Veen 'Introducing aggregate risk in an orthodox intertemporal Arrow-Debreu model: what happens to saving and investment?'

We study the effect of aggregate risk on saving and investment in an illustrative two-period general equilibrium model with contingent contracts (possibly constrained) in the Arrow-Debreu tradition and rather standard functional forms. Risk is modeled in the form of independent sector-specific endowment shocks. Household classes maximize expected utility. In a partial setting this specification of preferences and behavior would imply that risk leads to precautionary saving, with the largest tendency for the poorest classes. Simulations with the two-period model show that precautionary saving persists also in the general equilibrium context, in spite of the equilibrium price variability and the smoothing provided by contingent contracts. However, the precautionary saving effect is drastically reduced if investment itself is stochastic, more precisely if its impact is positively correlated to the aggregate risk in the economy. Surprisingly, savings hardly react to constraints on contingent transactions. Only when the assumption of rational expectations is dropped and household classes are assumed to be optimistic or pessimistic in the perception of their expected utility, bounds on contingent transactions have a significant impact on saving rates. Furthermore, the simulations show that in an orthodox Arrow-Debreu context with aggregate risk, liquidity constraints are not necessarily detrimental to the poor.



Temel, T. 'Notes on untested assumptions of a likelihood-ratio test'

Although it seems natural to assess quality of statistical inferences, researchers have not been loud enough about the implications for decisions based on the underlying untested assumptions. In this study, we examine and evaluate the untested assumptions of the likelihood-ratio test. This can be regarded as a standard exercise for the evaluation of any hypothesis testing procedure, as such exercise should offer a framework as to how a test procedure is to be carried out.

Keywords: Hypothesis testing, the Likelihood-ratio test.


Temel, T. and P.J. Albersen 'New facts for an old debate: farm size, productivity, and geography'

This study examines productivity, farm size, and the relationship between size and productivity in U.S. agriculture over the period 1982-1992. A nonparametric regression method was applied to detect ex post geographical patterns in changes in size and productivity. Estimations show that (i) in 1982 productivity per acre was high in the East, West, and South, modest in the middle part of the U.S., and low in the North, and this pattern remained the same in the period 1987-1992, while the level of productivity continuously increased over time; (ii) during 1982-1992 farm size remained unchanged, large farms being in the middle belt stretching from North to down South and small ones in the East, West and South; and finally (iii) over the period 1982-1992 an inverse relationship grew stronger between size and productivity per acre. Furthermore, Markov chains method was applied to project these ex post patterns into the future. Predictions suggest that at the national and regional levels (i) farms are likely to experience lower productivity per acre; (ii) small and large farms are likely to coexist as medium-sized farms to vanish; and (iii) the inverse relationship is likely to show a strong geographical pattern.

Key words: Farm size, productivity, geography, inverse relationship, U.S. agriculture, nonparametric regression, Markov chains.


Temel, T. 'A Non-parametric Hypothesis Test via the Bootstrap Resampling'

This paper adapts an already existing non-parametric hypothesis test to the bootstrap framework. The test utilizes the non-parametric kernel regression method to estimate a measure of distance between the models stated under the null hypothesis. The bootstrapped version of the test allows to approximate errors involved in the asymptotic hypothesis test.

Keywords: Hypothesis test, the bootstrap, non-parametric regression, omitted variables.


Keyzer, M.A. 'Reweighting survey observations by Monte Carlo integration on a census'

The possible non-representativity of household surveys can be addressed by reweighting the survey observations through a correction factor. This factor is usually computed on the basis of the frequency of a combination of household characteristics, say, the age and sex of the respondent, in the survey relative to its value in a population census. In this paper, a generalization of this technique is proposed that makes it possible to account for several household characteristics including real-valued ones. It applies kernel density regression to estimate the joint density over the survey of selected characteristics that are also recorded in the census. The weights are estimated by Monte Carlo integration of the estimated density over the census distribution. The efficacy of the procedure is tested in a simulation experiment that creates a large number of survey data sets as biased samples from the Ghana Living Standards Survey, applies the reweighting procedure to every sample, and confronts the resulting estimated mean with the known true mean of the census.

Keywords: Monte Carlo integration, kernel density regression, weighting scheme


Keyzer, M.A. 'Pricing a raindrop: the value of stocks and flows in process-based models with renewable resources'

Assigning economic value to renewable resources has become a major concern in environmental management. A common difficulty in this context is that a process-based representation of the underlying bio-physical processes necessarily causes the model specification to depart from the basic postulates of micro-economic theory, as it corresponds to a technology characterized by non-convexity, lack of free disposal, and no possibility of inaction. Consequently, regular valuation procedures do not apply, and current practice has become either to discard the irregularities in technology or to restrict the valuation procedure to impact assessments for a limited number of scenario simulations. In this paper, relying on capital theory, we describe a technique that enables us to effectuate a comprehensive valuation despite the non-standard features. It calculates marginal returns over an infinite time horizon of variations in inflows (raindrops) or adjustments in structural characteristics of the bio-physical process, in the steady state with all stocks. It can be implemented without analytical calculation of derivatives, because these marginal returns can be computed as the Lagrange multipliers of a mathematical program. We also discuss various extensions to account for random variability and for non-stationarity arising in as a consequence of, say, technical progress, population growth, and climate change.


Gerlagh, R. and M.A. Keyzer 'Dynamic efficiency of conservationist measures: an optimist viewpoint'

Advocates of strict resource conservation are often accused of embracing an overly pessimistic worldview that underrates the scope for technological innovation. This paper argues that conservationism also fits within an optimistic perspective. We consider an overlapping generations model with several consumer goods and exhaustible resources that provide amenity value, in which technical progress allows for sustained growth of manmade goods that remain, however, imperfect substitutes for these natural amenities. We compare a grandfathering policy that ensures efficiency through privatization with a policy of enforced resource conservation. It is shown that strictly conservationist measures do not cause any Pareto inefficiency in this model, irrespective of whether they pass a cost-benefit test. 


Nubé, M. and G.B. Overbosch 'High levels of Burden of Disease for all age groups of the population in Sub Sahara Africa'

The data as presented in the Global Burden of Disease study, a collaborative undertaking between the Harvard School of Public Health and the World Health Organization, provide the opportunity to assess how the total burden of disease (mortality and morbidity) is distributed over age groups. The data indicate that in Sub Sahara Africa all age groups, including the economically most active (15-45 years), are subject to high levels of burden of disease, much higher in comparison with high income countries, and also much higher in comparison with other developing regions. In view of the impact of disability on household functioning in all its aspects, including income generating capacity, results call for detailed studies on burden of disease at household level, for which household survey data might be an appropriate source of information. With respect to specific causes, mortality and disability resulting from violence, accidents, and mental illness, are estimated to account together for more than 40% of the disease burden in the 15-45 years age group in Sub Sahara Africa. Other diseases or disorders that significantly contribute to the total disease burden for adults are AIDS, other sexually transmittable diseases, tuberculosis, maternal conditions, malaria, and respiratory diseases. On the other hand, the contribution of tropical diseases such as schistosomiasis, river-blindness, filariasis, leishmaniasis and trypanosomiasis (but excluding malaria) is relatively limited.