An integrated macroeconomic, demographic and health modelling framework for palm oil policies in Thailand

Agricultural and food system changes have historically generated complex and dramatic "trade-offs" for health, the economy and the environment [1,2]. Palm oil is arguably the commodity whose position in the modern food system has had the most profound implications across these three policy areas and is thus the focus of the analysis in this paper. The production and trade of palm oil has brought great economic benefits to Asia given its relatively high yields, low production costs, and advantageous qualities as an ingredient in processed foods [3]. However, palm oil is a leading dietary source of saturated fat and has been strongly linked with negative health outcomes [4,5] and negative environmental effects, including massive deforestation, ecosystem service loss and greenhouse gas emissions [6]. This pioneering research paper brings together these three key dimensions of the Asian palm oil sector – health, economic and environmental in a single investigative framework. Few methodologies are amenable to accommodate analyses of the multi-facetted consequences of palm oil since the interactions between facets prevents the reliable a priori specification of economic, health or environmental outcomes through exogenous satellite models. However, the macroeconomic Computable General Equilibrium (CGE) methodology stands out as a tool which has been used effectively for both inter and intra disciplinary analysis in each of the mentioned areas and, by fully integrating these approaches in a single model, we show that CGE is the ideal methodology from which to explore and develop a new and interdisciplinary research approach to studying and managing trade-offs across interacting policy perspectives. This methodological extension to the CGE approach develops a dynamic and integrated Macroeconomic-Environmental-Demographic-health (MED-health) model framework with the ability to capture health, macroeconomic, population and environmental aspects of policies relating to palm oil production and consumption in Thailand. Our fully integrated macroeconomic-demographic-health model brings together a Computable General Equilibrium (CGE) model of the Thailand economy, a microsimulation-based epidemiological model [7] and a demographic model of the population in a single model framework. Our policy scenarios are conducted over a 20 year (2015-2035) timescale and employ economic incentives (taxes) from the macroeconomic sub-model as policy instruments to target reductions in saturated fatty acid consumption from palm oil. These taxes affect household consumption in the CGE model by means of an Almost Ideal Demand System which, in turn, are translated into nutritional changes by means of nutritional coefficients (particularly relating to fatty acids) which are assigned to regional household food consumption sectors in the CGE model. By interpolating across endogenous lookup tables of clinical health outcomes, derived from a published health microsimulation model [7], illness-specific incidence and mortality impacts corresponding to the changes in household consumption of food items are estimated and imposed on the demographic model. Demographic changes relating to both working-age patient and caregiver time losses are used to calculate regional labour market mortality impacts in the demographic model enabling the endogenous calculation of health, demographic and macroeconomic model outcomes. The endogenous feedback effects are specified, separately, for each of our 9 regional Thai household types which include Bangkok together with urban/rural stratifications of the 8 other standard administrative regions: Central (excluding Bangkok), North, Northeast, and South region. Our intervention strategies therefore have region-specific impacts and our economy-wide economic, disease burden and nutritional outcomes can also be decomposed into regional effects. Finally, an environmental module has been constructed to calculate CO2-equivalent emissions from land use change (LUC), based on Thai-specific LUC-coefficients [8]. The database on LUC-related CO2-equivalent emissions covers 20 primary agricultural crops, and a mapping between these crops and the 7 primary agricultural crops in our CGE model allowed us to derive a 7x7 matrix of LUC-related CO2-equivalent emission coefficients which was used to calibrate the environmental module of our integrated model framework. Having developed the integrated modelling framework, we demonstrate its usefulness by simulating five scenarios for reductions in household energy intake (Kcal) from palm oil consumption by 10% intervals in the range 10%-50%. The results of our simulated reductions in palm oil consumption include economic and welfare indicators, nutrition health and demographic indicators, and environmental indicators. A summary of these results follows and more details will be presented at the conference. The overall cumulative economic impacts from the 10%-50% palm oil shocks are positive and raise GDP by 2.3-14.3bn USD (2015 prices) over the 20 year timescale. Health contributes modestly to this total, contributing just 30mn-225mn while the remainder of the gains are generated via the sales tax pathway. Since our model employs a revenue-neutral government budget closure, increased sales tax revenues are returned to households through reduced direct taxes. This allows an increase in household savings, capital formation and economic growth and this mechanism contributes to the progressive nature of the tax. Household welfare, measured by cumulative real consumption, similarly shows modest health-related economic gains of 17mn-127mn USD (2015 prices) and a much larger contribution from the sales tax of 1.9-11.5bn USD over the coming 20 years. At the regional level, health-related economic impacts are proportionally distributed across Thailand, including 0.0020-0.0028% increases in real consumption for all regions, whereas sales tax-related economic impacts will only benefit non-South regions. South region households, where the majority of palm oil is produced, will experience average 0.022% welfare losses over the coming 20 years. Nutritional impacts, particularly changes in the ratio of saturated fatty acids (SFAs) to polyunsaturated fatty acids (PUFAs) drive health outcomes. Our analysis shows that the sales tax generates changes in household consumption of food items: reducing consumption of energy from palm oil, which is high in SFAs, and primary agricultural products and increasing energy intake from other edible oils which are high in PUFAs and also increasing energy-intake from other processed foods in the long run. In addition, household budget expansion from economic growth fuels a net increase in overall dietary energy intake. The change in food demand, particularly the substitution towards other edible oils and other processed foods has important implications for the composition of micro-nutrient intakes. The largest (50%) palm oil energy shock leads to a long run 17.7 kcal/day reduction in energy intake from saturated fatty acids (SFA), while substitution in other food sectors increases SFA consumption from those sectors slightly by 5.5 kcal/day. However, the substitution towards other edible oils significantly increases energy intakes from polyunsaturated fatty acids (PUFA) generating an overall 33.3kcal/day net increase in the long run. Health impacts resulting from the substitution away from SFA towards PUFA energy are positive, but small. Nutritional changes alter the total-HDL cholesterol ratio which, in turn yields improved clinical health outcomes for both myocardial infarction and stroke. For the 10%-50% shocks, our model predicts that, over the next 20 years cumulative incidence of myocardial infarction (MI) will decline by 884-6,580 while stroke cases will decline by 285-2,126. Combined reductions in MI and stroke-related morbidity from the 10%-50% shock, measured by Years Lived with Disability (YLD) are estimated to be between 261-1,921 years and cumulative premature deaths are predicted to decline by 613-4552. These MI and stoke impacts constitute approximately 0.4% and 0.10% of the Thai MI and stroke disease burdens respectively. Beyond the direct patient effects, our results suggest that the 10%-50% shocks will reduce total caregiver time use for MI and stroke patients by 524-3,900 person-years, 217-1,593 person-years of which is work time which would be lost from the labour force. Demographic impacts are also positive. Cumulative population impacts from reduced palm oil consumption will, on average, increase the Thai population by 0.0004%-0.0027% (or a cumulative 4,813-36,315 person years) for the 10%-50% dietary energy shocks. This translates to an average saving of up to 2.7 persons per 100,000 Thai inhabitants over the 20 year simulation. However, from the regional perspective, cumulative impacts show wide discrepancies across the country: Northern and Central (excluding Bangkok) regions expand, cumulatively, by 0.0013%-0.0037% (a saving of up to 3.7 persons per 100,000 inhabitants), while Bangkok and South region cumulative population impacts are more modest at 0.0001%-0.0017% (a saving of up to 1.7 persons per 100,000 inhabitants). Comparing rural and urban population impacts also suggests large discrepancies. Although the model captures migration from rural to urban areas in response to economic growth, the cumulative rural population still expands by 0.0006%-0.0037% saving up to 3.7 persons per 100,000 inhabitants. The cumulative urban population, in contrast, only expands by 0.0002%-0.0019% or up to 1.9 persons per 100,000 inhabitants. Finally, we find negative environmental impacts from our policy shocks. Palm oil sales taxes will reduce demand for processed palm oil and land use in oil palm production, significantly. This provokes a reallocation of land away from oil palm and towards other agricultural crops, which have less beneficial carbon sequestration characteristics. Our results suggest that CO2-equivalent emissions will therefore increase by a cumulative 1.8-10.2 Mega-tonnes over our 20 year time horizon. This impact is primarily due to reduced carbon sequestration among smallholder producers in the South region of Thailand, where oil palm production is mainly conducted on abandoned paddy fields, waste lands, and other land with minimal alternative carbon sequestration potential. Conclusion Our results suggest that reductions in consumption of dietary energy from palm oil, via a sales tax, is likely to produce aggregate benefits for Thailand from the economic perspective together with small health and demographic gains. At the regional level, the tax is shown to be progressive, distributing benefits to rural more than urban households, but South region households, where palm oil is produced, are likely to suffer economic losses in contrast to gains in all other areas and so will be noticeably worse off as a result of reduced demand for palm oil. Environmental consequences are also negative. The ability to distinguish both distributional impacts and contrasting positive and negative trade-offs from economic, health and environmental perspectives highlights the strength of our fully integrated approach. References 1) Dangour, A. D., Green, R., Häsler, B., Rushton, J., Shankar, B., & Waage, J. (2012). Linking agriculture and health in low-and middle-income countries: an interdisciplinary research agenda. Proceedings of the Nutrition Society, 71(02), 222-228. 2) Lee, D. R., & Barrett, C. B. (Eds.). (2001). Tradeoffs Or Synergies? Agricultural Intensification, Economic Development and the Environment. CABI Publishing. 3) Shankar, B., & Hawkes, C. (2013). India has a problem with palm oil. BMJ, 347 4) Basu, S., Babiarz, K. S., Ebrahim, S., Vellakkal, S., Stuckler, D., & Goldhaber-Fiebert, J. D. (2013). Palm oil taxes and cardiovascular disease mortality in India: economic-epidemiologic model. BMJ, 347. 5) Chen, B. K., Seligman, B., Farquhar, J. W., & Goldhaber-Fiebert, J. D. (2011). Multi-country analysis of palm oil consumption and cardiovascular disease mortality for countries at different stages of economic development: 1980-1997. Global Health, 7(1), 45. 6) Fitzherbert E, Struebig M, Morel A, Danielsen F, Bruhl C, Donald P, et al. How will oil palm expansion affect biodiversity? Trends in ecology & evolution. 2008; 23 (10): 538—545 7) Basu S, Babiarz KS, Ebrahim S, Vellakkal S, Stuckler D, Goldhaber-Fiebert JD. 2013. Palm oil taxes and cardiovascular disease mortality in India: economic-epidemiologic model. BMJ 347:f6048 8) Gheewala SH. 2015. CO2-equivalent Emissions From Land Use Change in Thailand. Electronic data. Private communication (12. November 2015). 9) Corley, R. H. V., & Tinker, P. B. H. (2008). The oil palm. Wiley Publishers. 10) Kosulwat, V. (2002). The nutrition and health transition in Thailand. Public Health Nutrition, 5(1a), 183-189. 11) Wilkinson, J. (2008). The food processing industry, globalization and developing countries. The transformation of agri-food systems: globalization, supply chains and smallholder farmers Food & Agriculture Organization of the UN (FAO), 87.