Climate volatility deepens poverty vulnerability in developing countries

Syud A Ahmed¹, Noah S Diffenbaugh^2,3 and Thomas W Hertel^2,4,5

¹ Development Research Group, The World Bank, Washington, DC, USA
² Purdue Climate Change Research Center, Purdue University, IN, USA
³ Department of Earth and Atmospheric Sciences, Purdue University, IN, USA
⁴ Center for Global Trade Analysis, Purdue University, IN, USA
⁵ Department of Agricultural Economics, Purdue University, IN, USA

E-mail: sahmed20@worldbank.org

Received 20 March 2009
Accepted 20 July 2009
Published 20 August 2009

Contents

• 1. Introduction
• 2. Climate extremes in current and future A2 scenario
• 3. Economic data and methods
• 4. Poverty changes due to climate extremes
• 5. Conclusion
• Acknowledgments
• References

1. Introduction

Staple grains such as rice represent a significant source of expenditure, calories and earnings for many of the poor in developing countries (Cranfield et al 2003, Thurlow and Wobst 2003, Ulimengu et al 2009). This dependence has raised concern about how recent increases in food prices might affect poverty (IFPRI 2008). While the impacts of high food prices vary across different socio-economic groups in developing countries, analyses of survey data indicate that the short-run effect of the 2005–2007 staple food price increase was to worsen poverty for the majority of the poor (Ivanic and Martin 2008). Estimates from the Food and Agriculture Organization (FAO) indicate that rising prices were responsible for an increase in the proportion of hungry people in the developing world by about 75 million people between 2003–2005 and 2007 (FAO 2008).

The poverty vulnerability of developing countries to rising food prices motivates a need for better understanding of the drivers of commodity price volatility and their linkages to poverty. One important source of volatility is adverse climate events, including extreme heat, droughts and floods. Although it is known that these events can detrimentally affect agricultural systems and food security (White et al 2006, Funk et al 2008, Lobell et al 2008, Battisti and Naylor 2009), the impact of climate volatility on national-scale poverty has not yet been quantified, nor has the impact on the poor in different socio-economic strata. In addition, climate studies consistently show that further increases in GHG concentrations are likely to cause increases in hot, wet and dry extremes (e.g., Diffenbaugh et al 2005, 2007, IPCC 2007), and there is evidence that such changes to climate volatility are already occurring (Easterling et al 2000, IPCC 2007). As the frequency and intensity of climate extremes increase, crop production damages from such events will change (Schmidhuber and Tubiello 2007, Naylor et al 2007). Sharp reductions in crop supply put upward pressure on food prices, thereby having a significant poverty impact. Therefore, in order to create informed policy responses to the threat of increased poverty vulnerability as well as to better quantify potential damages associated with varying greenhouse targets, it is imperative to understand the linkages between developing country poverty and climate extremes.

However, analyses combining both rigorous climate science and economic theory have been notably lacking to date. This paper provides an integrated analytical framework with empirical and theoretical foundations that can be used to study the social dimensions of extreme events, future volatility, and other potential climate effects. It then examines which socio-economic groups and countries are most vulnerable to climate extremes, providing a basis for climate adaptation and poverty reduction strategies in the developing world.

2. Climate extremes in current and future A2 scenario

We seek to quantify the vulnerability of the poor to potential changes in climate volatility, in the context of the frequency and magnitude of annual-scale hot, dry, and wet extremes. We draw on the unprecedented global climate model database that has been assembled as part of the `CMIP3' project and heavily used in the IPCC Fourth Assessment Report (IPCC 2007), to consider three distinct agricultural productivity stressors. These annual-scale extreme climate indices from Frich et al (2002) are available in the CMIP3 multi-model archive: (1) the per cent of annual total precipitation due to events exceeding the 1961–1990 95th percentile; (2) the maximum number of consecutive dry days; and (3) the heat wave duration index (the maximum period greater than 5 days with the daily maximum temperature greater than the 1961–1990 normal). We thus quantify the response of the 30-year-maximum values of wet, dry, and hot extremes.

We analyze 30 year periods from 1971 to 2000 in the 20th century simulations and 2071 to 2100 in the simulations under the IPCC's A2 scenario, examining output from `run 1' of the seven models that have archived the extreme climate indices. Within each time period and scenario, and at the 108 Agro-Ecological Zone (AEZ) level of spatial disaggregation, we calculate the extreme climate indices: (1) the 30-year-maximum value in the 20th century period; (2) the 30-year-maximum value in the 21th century period; and (3) the number of years in the 21st century period that exceeded the 30-year-maximum value of the 20th century period. For each variable, we interpolate the data from each GCM to a common 1° geographical grid, and calculate the maximum annual value of the variable in the 1971–2000 and 2071–2100 time-series of each GCM at each grid point. We then determine the number of years whose value exceeded that maximum annual value in each of the time-series at each grid point: 1971–2000 and 2071–2100 in the A2 scenario. We compute the mean value at each grid point across the seven GCMs, as well as the inter-model standard deviation of the 21st century minus 20th century fractional difference.

We next interpolate these GCM occurrence data from the common 1° grid to the 0.5° geographical grid upon which the AEZ data is based, select the grid points corresponding to each of the focus countries, and compute the area-weighted mean occurrence in each time period/scenario for each AEZ in a given country and for the country as a whole. We also calculate the area-weighted mean of the values of the respective grid points falling within the borders of each country.

The CMIP3 ensemble exhibits substantial changes in extreme climate in the late 21st century of the A2 scenario (figure 1). The occurrence of what is now the 30-year-maximum extreme wet event increases (A2 minus present, relative to present) throughout the world, with maximum increases of greater than 900% occurring over Southeast Asia. The absolute magnitude of the 30-year-maximum event is also greater throughout the world in the A2 scenario than at present, with peak increases of greater than 40% also in Southeast Asia. All countries exhibit substantial increases in the occurrence and magnitude of extreme hot events, with the occurrence of the present 30-year-maximum event increasing more than 2700% in parts of the northern Mediterranean, and the magnitude of the 30-year-maximum event increasing 1000% to more than 2250% in much of central Africa. Most countries also display increases in the occurrence and magnitude of extreme dry events, with peak changes of greater than 800% and 60% (respectively) occurring over Mediterranean Europe. A handful of countries display decreases in occurrence, magnitude, or both of extreme dry events. Countries where the intensity and frequency of extreme consecutive dry days decline include Colombia and Uganda, while Tanzania, Indonesia, and the Philippines experience less intense but more frequent years of such extreme climate. The change in the magnitude of the extreme hot and wet events is reduced in the B1 scenario relative to the A2 scenario for all focus countries (table 1). This reduction in the severity of change in the B1 scenario is less universal for the magnitude of the extreme dry event.

3. Economic data and methods

The magnitude and spatial heterogeneity of changes in climate volatility suggest that the impacts on poverty could also be large and heterogeneous. To quantify the vulnerability of developing countries to being impoverished by current climate volatility we focus on the response to once-in-30-year-climate events. For the staple grains sector, climate outcomes for a given country in a given year can be interpreted as exogenous supply-side shocks to sector productivity. Controlling for trends, climate outcomes affecting grains production can thus be considered draws from a probability distribution of interannual grains productivity changes. We generate such productivity change distributions for 16 developing countries (sample selection discussed below) by implementing zero-mean normal distributions characterized by standard deviations of interannual staple grains productivity changes estimated from historical data. These estimates are taken from Valenzuela (2009) or are estimated from FAOSTAT data (FAO 2009). For each country in our sample of 16 `focus' countries, the productivity change that corresponds to a once-in-30-year-climate extreme is determined to be the interannual productivity change that occurs 3.33% of the time. However, since we are drawing directly from statistically estimated distributions of interannual productivity changes, the once-in-30-year agricultural productivity change for a given country is distinct from the once-in-30-year changes for other countries, and cannot be directly mapped to the historically observed maximum values of the annual-scale extreme climate indices (figure 1, table 1).

We examine the poverty impacts of the current climate's once-in-30-year productivity shocks using a modified version of the comparative static computable general equilibrium (CGE) simulation model GTAP (Hertel 1997). The model uses the empirically robust assumptions of constant returns to scale and perfect competition, and is modified to better delineate distributional impacts. We introduce factor market segmentation which is important in countries where the rural sector remains a dominant source of poverty (Keeney and Hertel 2005), and disaggregate land endowments by AEZ (Hertel et al 2009b). The model is supported by Version 6 of the GTAP Database (Dimaranan 2006) and the GTAP Land Use Database (Lee et al 2009, Monfreda et al 2009). The combined database represents the 2001 global economy, and the climate extreme simulations thus represent differences from this benchmark.

Poverty analysis is facilitated by a special poverty module which has been added to GTAP following the approach of Hertel et al (2009a). Implementation of this module requires comprehensive processing of a national household survey and this is the limiting factor constraining the size of our sample of focus countries which comprises 16 countries in Africa, Asia and Latin America (table 1)^Note6. While this is not a random sample of countries, it does encompass a wide range of developing countries with greatly differing patterns of poverty. Within each country, poverty is broken down into socio-economic strata based on primary source of income (95% or more of income from the following sources): agricultural self-employed (farm income), non-agricultural (non-agricultural self-employment earnings), urban labor (urban household, wage labor income), rural labor (rural household, wage labor income), transfer payment dependent, urban diverse, and rural diverse (Hertel et al 2004).

The poverty module utilizes a micro-simulation for representative households at the poverty line in each socio-economic stratum to determine changes in poverty headcount based on changes in real income. We use the World Bank's $1/day Purchasing Power Parity definition of poverty to ensure comparability across countries (Chen and Ravallion 2001)^Note7. Each household maximizes its utility obtained from an econometrically estimated AIDADS demand system (Rimmer and Powell 1996, Cranfield et al 2003). This yields a new preferred consumption bundle and a new level of utility in the wake of any change in goods prices and factor incomes. The AIDADS demand system is well suited for poverty analysis, due to ability to characterize consumer behavior at low income levels. In particular, for each commodity, AIDADS provides estimates of subsistence quantities of consumption, as well as the propensity of households at the subsistence level to spend additional income on various goods and services. These two parameters, operating in conjunction with changes in per capita income, determine household expenditure shares at the poverty level of utility, which is the focal point of our analysis. Across the 16 country sample, the share of total expenditure devoted to food ranges from 41% in Brazil to 67% in Uganda so that any rise in food prices will have a sharp impact on real income, and hence poverty.

The poverty consequences of an adverse climate shock are evidenced through two channels: changes in earnings and changes in the real cost of living at the poverty line. The impact of a food price rise on earnings depends on the income sources for a given household group; these earnings shares are estimated for households in the neighborhood of the poverty line from household survey data. If earnings rise faster than the cost of living for households at the poverty line in a given stratum, then the poverty headcount will fall and vice versa. The responsiveness of the stratum poverty headcount to a given real income shock is determined by the density of the stratum population in the neighborhood of the poverty line and is also estimated from the household survey data. When combined with information about the distribution of national poverty across socio-economic strata, we are able to estimate the change in national poverty headcount.

Figure 2 reports the proportions of focus countries' total 2001 populations that are impoverished, grouping them into the seven strata that reflect the pattern of household earnings specialization. The darker shaded areas in each bar correspond to the agriculture and rural households in our sample. It can be seen that these dominate the poverty picture in nearly every country. Bangladesh, Mozambique, Malawi, Tanzania, Uganda, and Zambia stand out as already experiencing particularly high incidences of poverty. However, given that Bangladesh's total population is more than that of the combined population of all five African countries, the absolute number of Bangladeshi poor is greater.

4. Poverty changes due to climate extremes

Using the framework just described, we simulate once-in-30-year-climate extremes one country at a time, and generate economic changes that are attributable only to the extreme climate impacts in that country. We thereby mitigate the effects of international price transmission that may have arisen due to agricultural productivity volatility in other countries (Valenzuela et al 2007). We can thus be confident that the resulting poverty changes are attributable to grains price and supply effects arising due to extreme climate outcomes in those countries alone. These poverty changes are indicated by the cross-hatched areas for each focus country in figure 2.

The countries with the highest shares of populations entering poverty in the wake of these extreme events include Bangladesh, Mexico, Mozambique, Malawi, Tanzania, and Zambia (figure 2). In Malawi and Zambia, simulated grains productivity declines of about 75% cause the poverty headcount in those countries to increase by about seven percentage points relative to their total populations. The large magnitudes of the declines in grains productivity are not unrealistic, and are reflected in the historically observed interannual changes in grains production. Grains productivity in Malawi and Zambia declined by between 50 and 59% in the period 1991–1992, respectively, when southern Africa experienced a severe ENSO-related drought (Glantz et al 1997). Cereal consumption—measured as daily calories consumed per person from cereal—declined by 8.8% in Malawi and by 1.3% in Zambia during this period (FAO 2009).

A closer look at the poverty distribution by stratum in each country from figure 2 can foreshadow where the impacts of an extreme climate year will be felt most sharply. Table 2 reports the percentage changes in poverty due to climate extremes by socio-economic stratum and country. These results indicate tremendous heterogeneity in the poverty vulnerability to climate extremes across different segments of the population—when differentiated by primary income source (stratum). The bottom row in table 2 reports the simple average of percentage changes across countries, for a given stratum. From this, we can see that the most vulnerable group is the urban, wage-labor-dependent stratum. While the urban labor group contributes modestly to total poverty in this sample of countries (figure 2), it appears to be highly vulnerable to extreme climate events. Indeed, the poverty rate for this group doubles in Malawi under the once-in-30-year-climate event (table 2). Zambia and Mexico also show high vulnerability among this group. The source of the vulnerability of the urban poor is their extreme exposure to food price increases. Since food is a major expenditure, this group's overall consumption falls with rising prices, pushing them below the poverty threshold of consumption.

Agricultural households, on the other hand, are much less exposed (an average 9.2% poverty change under extreme events—table 2). This group is generally hurt by the adverse productivity shock from a food consumption perspective. However, they are partially insulated from the effects of the higher prices since they also benefit (on the income side) as producers of food. The other groups tend to fall in between these two extremes. Rural (and, to a lesser extent, urban) diversified households tend to earn a significant portion of their income in agriculture, and therefore benefit from higher prices. Non-agricultural self-employed households are also somewhat insulated from the adverse productivity shocks in agriculture^Note8. Given that the shares of developing countries' populations living in rural areas are projected to decrease by more than one-third between 2010 and 2050 (United Nations 2007), climate extremes may have greater national-scale poverty impacts in the future because of higher population concentrations in the more sensitive urban strata.

Thus far we have assessed the relationship between current climate volatility and poverty via fluctuations in staple grains productivity. We now consider the impact of potential changes in climate volatility in the future. We can use the climate volatility metrics (table 1) as proxies for changes in future staple grain productivity under extreme events. To do so, we scale the current climate-related grains productivity decline under a 1-in-30 scenario by the simulated changes in intensity of a given type of climate extreme, thus obtaining an estimate of future adverse grains productivity shocks that can be used as input to the economic model.

Table 3 shows the response of national-level poverty when the current once-in-30-year productivity decline is adjusted based on changes to the intensity of the 30-year-maximum extreme dry events under the A2 scenario, determined as national averages weighted by the value of grains production by AEZ. Bangladesh, Mexico, and Zambia have the greatest additional vulnerability, with an additional 1.4, 1.8, and 4.6% of their populations being impoverished by future extreme climate, respectively. While the pattern of the vulnerabilities across strata and countries do not change and all strata become more vulnerable, self-employed agricultural households will tend to see the smallest increase in their vulnerability, while urban wage earners will see the greatest^Note9.

5. Conclusion

Our novel, inter-disciplinary analysis of climate–agriculture–poverty interactions takes advantage of available data from the FAO and the CMIP3 multi-GCM ensemble experiment. However, it has some clear limitations. Individual extreme events like floods and tropical storms are captured only implicitly by our annual data. In addition, we have used the simulated climate changes as proxies for changes in agricultural productivity. A complete treatment will include explicit modeling of agricultural productivity in response to projected changes in climate volatility. Finally, our framework does not consider the impact of alternative sequences of climate shocks. Thus we are unable to analyze how the susceptibility of the poor to detrimental climate may change after experiencing an extreme climate event, a topic of concern to many policy analysts (Adger 1999). In addition, it is important to note that there is uncertainty in the exact trajectory of climate volatility in the future, because of uncertainty in the physics of the climate system (as evidenced by the spread in the GCM projections and the difference between different emissions scenarios) (table 1).

Analysis of the countries and socio-economic strata that are most vulnerable to impoverishment from short-run food price increases associated with climate extremes will allow for better-informed strategic mobilization of international development resources and climate policy instruments. We find that climate extremes exert substantial stress on low income populations, and that there is considerable international heterogeneity in the response of poverty to climate volatility. Further, we find that the urban, wage labor households are likely to be the most vulnerable to climate extremes. Attempts to quantify the true value of different mitigation policies must integrate such heterogeneity of impacts, and adaptation policies must not neglect the urban poor.

The largest climate-induced poverty responses in our sample occur in Africa, with agriculture in many sub-Saharan countries being particularly sensitive to climate. Investment adaptation responses, such as irrigation investment in Tanzania, would thus reduce the magnitude of the agricultural productivity decline due to climate extremes (such as drought), thereby buffering the impact of the climate event on poverty. However, as experiences from South Africa and Ethiopia (Bryan et al 2009) point out, even when farmers are aware of the need for adaptation, through options like different farming practices and irrigation, they may be constrained by institutional factors, like barriers to access to credit and information. This suggests that, in addition to investments in infrastructure, an enabling institutional environment is required to significantly ameliorate the impacts of extreme climate events.

Acknowledgments

This research was funded by the World Bank's Trust Fund for Environmentally and Socially Sustainable Development. We also acknowledge the modeling groups, the Program for Climate Model Diagnosis and Intercomparison (PCMDI), and the WCRP's Working Group on Coupled Modeling (WGCM) for their roles in making available the WCRP CMIP3 multi-model dataset. Support of this dataset is provided by the Office of Science, US Department of Energy. The views and opinions reflected in this paper are solely those of the authors. This is PCCRC paper number 0921.

References

Adger W N 1999 Social vulnerability to climate change and extremes in coastal Vietnam World Dev. 27 249–69 
CrossRef

Battisti D S and Naylor R L 2009 Historical warnings of future food insecurity with unprecedented seasonal heat Science 323 240–4 
CrossRefPubMed

Bryan E, Deressa T T, Gbetibouo G A and Ringler C 2009 Adaptation to climate change in Ethiopia and South Africa Environ. Sci. Policy 12 413–26 
CrossRef

Chen S and Ravallion M 2001 How did the world's poor fare in the 1990s? Rev. Income Wealth 47 283 
CrossRef

Cranfield J A L, Preckel P V, Eales J S and Hertel T W 2003 Estimating consumer demand across the development spectrum: maximum likelihood estimates of an implicit direct additivity model J. Dev. Econ. 68 289–307 
CrossRef

Diffenbaugh N S, Pal J S, Giorgi F and Gao X 2007 Heat stress intensification in the Mediterranean climate change hotspot Geophys. Res. Lett. 34 L11706 
CrossRef

Diffenbaugh N S, Pal J S, Trapp R J and Giorgi F 2005 Fine-scale processes regulate the response of extreme events to global climate change Proc. Natl Acad. Sci. USA 102 15774–8 
CrossRefPubMed

2006 Global Trade, Assistance, and Production: The GTAP 6 Data Base ed  B V Dimaranan Center for Global Trade Analysis, Purdue University, IN 

Easterling D R, Meehl G A, Parmesan C, Changnon S A, Karl T R and Mearns L O 2000 Climate extremes: observations, modeling, and impacts Science 289 2068–74 
CrossRefPubMed

FAO 2008 Hunger on the rise Briefing Paper (17 Sept., 2008) (Rome: Food and Agriculture Organization) 

FAO 2009 FAOSTAT Database (Rome: Food and Agriculture Organization) Available at http://faostat.fao.org (4 February 2009) 

Frich P, Alexander L V, Della-Marta P, Gleason B, Haylock M, Klein Tank A M G and Peterson T 2002 Observed coherent changes in climate extremes during the second half of the twentieth century Clim. Res. 19 193–212 
CrossRef

Funk C, Dettinger M D, Michaelsen J C, Verdin J P, Brown M E, Barlow M and Hoell A 2008 Warming of the Indian Ocean threatens eastern and southern African food security but could be mitigated by agricultural development Proc. Natl Acad. Sci. USA 105 11081–6 
CrossRefPubMed

Glantz M H, Betsill M and Crandall K 1997 Food Security in Southern Africa: Assessing the Use and Value of ENSO Information (Boulder, CO: Environmental and Societal Impacts Group, National Center for Atmospheric Research) 

1997 Global Trade Analysis: Models and Applications ed  T W Hertel (Cambridge: Cambridge University Press) 

Hertel T W, Ivanic M, Preckel P V and Cranfield J A L 2004 The earnings effects of multilateral trade liberalization: implications for poverty World Bank Econ. Rev. 18 205–36 
CrossRef

Hertel T W, Keeney R, Ivanic M and Winters L A 2009a Why isn't the Doha development agenda more poverty friendly? Rev. Dev. Econ. doi:10.1111/j.1467-9361.2008.00483.x 

Hertel T W, Lee H L, Rose S and Sohngen B 2009b Modelling land use related greenhouse gas sources and sinks and their mitigation potential Economic Analysis of Land Use in Global Climate Change Policy ed T Hertel, S Rose and R Tol (Abingdon: Routledge) chapter 6 

IFPRI 2008 Responding to the Global Food Crisis: Three Perspectives (Washington DC: International Food Policy Research Institute) Essays 

IPCC W G I 2007 Climate Change 2007: The Physical Science Basis: Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change ed S Solomon, D Qin, M Manning, M Marquis, K Averyt, H L Miller and Z Chen 

Ivanic M and Martin W J 2008 Implications of higher global food prices for poverty in low-income countries Agric. Econ. 39 405–16 
CrossRef

Keeney R and Hertel T W GTAP-AGR: a framework for assessing the implications of multilateral changes in agricultural policies, 2005 GTAP Technical Paper No. 24 Center for Global Trade Analysis, Purdue University) 

Lee H L, Hertel T W, Rose S and Avetisyan M 2009 An integrated and use database for CGE analysis of climate policy options Economic Analysis of Land Use in Global Climate Change Policy ed T Hertel, S Rose and R Tol (Abingdon: Routledge) chapter 4 

Lobell D, Burke M B, Tebaldi C, Mastrandrea M D, Falcon W P and Naylor R L 2008 Prioritizing climate change adaptation needs for food security in 2030 Science 319 607–10 
CrossRefPubMed

Meehl G A et al 2007 Global Climate Projections, IPCC (2007) chapter 10 pp 747–845 

Monfreda C, Ramankutty N and Hertel T W 2009 Global agricultural land use data for climate change analysis Economic Analysis of Land Use in Global Climate Change Policy ed T Hertel, S Rose and R Tol (Abingdon: Routledge) chapter 2 

National Bureau of Statistics 2002 Household Budget Survey 2000/01 vol V (Dar es Salaam, Tanzania: National Bureau of Statistics) 

Naylor R L, Battisti D S, Vimont D J, Falcon W P and Burke M B 2007 Assessing risks of climate variability and climate change for Indonesian rice agriculture Proc. Natl Acad. Sci. USA 104 7752–57 
CrossRefPubMed

Ravallion M, Chen S and Sangraula P Dollar a day revisited, 2008 Policy Research Working Paper No. 4620 (Washington, DC: World Bank) 

Rimmer M T and Powell A A 1996 An implicitly directly additive demand system Appl. Econ. 28 1613–22 
CrossRef

Schmidhuber J and Tubiello F N 2007 Global food security under climate change Proc. Natl Acad. Sci. USA 104 19703–8 
CrossRefPubMed

Thurlow J and Wobst P Poverty focused social accounting matrices for Tanzania, 2003 IFPRI Discussion Paper No. 112 (Washington, DC: International Food Policy Research Institute) 

Ulimengu J M, Workneh S and Paulos Z Impact of soaring food price in Ethiopia? Does location matter?, 2009 IFPRI Discussion Paper No. 846 (Washington, DC: International Food Policy Research Institute) 

United Nations 2007 World Population Prospects: The 2006 Revision and World Urbanization Prospects: The 2007 Revision Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, Available at http://esa.un.org/unup 

Valenzuela E 2009 Poverty Vulnerability and Trade Policy: CGE Modelling Issues (Saarbrücken: VDM Verlag Dr. Müller) ISBN: 978-3639122114 

Valenzuela E, Hertel T W, Keeney R and Reimer J J 2007 Assessing global computable general equilibrium model validity using agricultural price volatility Am. J. Agric. Econ. 89 383–97 
CrossRef

White M A, Diffenbaugh N S, Jones G V, Pal J S and Giorgi F 2006 Extreme heat reduces and shifts United States premium wine production in the 21st century Proc. Natl Acad. Sci. USA 103 11217–22 
CrossRefPubMed

Notes

Note6  The data for Tanzania are from the National Bureau of Statistics (2002) and were processed by Ana Rios (Inter-American Development Bank). The data for the remaining 15 focus countries were processed by Ivanic (World Bank), and documented in Hertel et al (2004) and Hertel et al (2009a).

Note7  This is more stringent than the more recently estimated $1.25/day poverty line (Ravallion et al 2008).

Note8  The impact on transfer dependent households hinges on how we index such transfer payments. Here, we assume that they are adjusted according to a national price index, which serves to insulate them to some degree.

Note9  Unfortunately, this framework cannot be extended to simulate changes in intensity of extreme wet and hot climate events. If the productivity shocks were scaled to reflect the intensity changes of those climate indices, then the interannual change in grains productivity would exceed negative 100%, which is infeasible.