WPS4291 Policy ReseaRch WoRking PaPeR 4291 An Empirical Economic Assessment of Impacts of Climate Change on Agriculture in Zambia Suman Jain The World Bank Development Research Group Sustainable Rural and Urban Development Team July 2007 Policy ReseaRch WoRking PaPeR 4291 Abstract This report assesses the economic impacts of climate November to December, January to February, and March change on agriculture in Zambia, using the Ricardian to April. Assuming a nonlinear relationship of farm method. revenue with the climate variables, quadratic terms for A multiple linear regression model with net revenue climate variables were also included in the model. per hectare as response variable has been fitted with The results indicate that most socioeconomic variables climate, hydrological, soil, and socioeconomic variables are not significant, whereas some climate variables and as explanatory variables. There is one main cropping the corresponding quadratic variables are significant in season in Zambia, lasting from November to April. the model. Further findings are that an increase in the Crop production in this period depends solely on November-December mean temperature and a decrease rains. Considering crop progression in three stages-- in the January-February mean rainfall have negative germination, growing, and maturing, which require impacts on net farm revenue, whereas an increase in the different amounts of water and temperature--the climate January-February mean temperature and mean annual variables included in the model are long-term averages runoff has a positive impact. of the temperature and wetness index for the periods This paper--a product of the Sustainable Rural and Urban DevelopmentTeam, Development Research Group--is part of a larger effort in the group to mainstream climate change research. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Pauline Kokila, room MC3-446, telephone 202-473- 3716, fax 202-522-1151, email address pkokila@worldbank.org. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at sjain@natsci.unza.zm. July 2007. (31 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team AN EMPIRICAL ECONOMIC ASSESSMENT OF IMPACTS OF CLIMATE CHANGE ON AGRICULTURE IN ZAMBIA1 Suman Jain2 1An earlier version of this Working Paper was published as CEEPA Discussion Paper number 27. 2 Department of Mathematics and Statistics, University of Zambia. The author wishes to thank the GEF and the World Bank for funding which made this study possible; Dr Ariel Dinar for initiating the project; Prof. Robert Mendelsohn for his valuable guidance; Prof. Rashid Hassan and Dr James Benhin for coordinating the project and providing logistical support; Pradeep Kurukulasuriya for data management and help with the analysis; the Central Statistics Office of Zambia for helping with the nationwide household survey and providing relevant data from their archives; the Meteorological Department and the Ministry of Agriculture, Food and Fisheries of Zambia for providing data; graduate students Edson Nkonde and Fred Nambala for analyzing and preparing charts of weather and crop data; Themba Munalula of Zambia and other country teams for helpful discussions; and Dalène du Plessis and Lynette Burger from CEEPA for serving as hosts to the project. This paper was funded by the GEF and the World Bank. It is part of a larger study on the effect of climate change on agriculture in Africa, managed by the World Bank and coordinated by the Centre for Environmental Economics and Policy in Africa (CEEPA), University of Pretoria, South Africa. SUMMARY Agriculture is a major economic sector for many African nations, contributing about 30% to Africa's GDP. About 70% of the population of the continent depend on this sector for their livelihood. Most agricultural production on the continent is rain dependent. The increasing concentration of greenhouse gases in the earth's atmosphere is changing the global and regional climate. The global mean temperature has increased by over 0.5°C since the 19th century. It has been noted that the mean temperature is rising more rapidly at the regional scale for Africa. Rainfall in Africa is also varying from year to year, in volume and seasonal distribution, and showing erratic patterns. Droughts have become more frequent in the last 30 years. Such climatic changes have enormous consequences for the food security of the continent. Lack of capital and technology make it hard for poor countries in Africa to adapt. This report is part of a three year Global Environment Facility (GEF) and World Bank funded project Regional Climate, Water and Agriculture: Impacts on and Adaptation of Agro-ecological Systems in Africa undertaken in eleven African countries. It assesses the economic impacts of climate change on agriculture in Zambia, based on the Ricardian method which measures the effect of climate on the value of agricultural land. For a country like Zambia with abundant free farming land for subsistence farming, it is difficult to attach a value to land. The Ricardian method has therefore been modified here to replace land value with net farm revenue. A multiple linear regression model with net farm revenue as response variable has been fitted with climate, hydrological, soil and socio-economic variables as explanatory variables. There is one main cropping season in Zambia, lasting from November to April. Crop production in this period depends solely on rains. Considering crop progression in three stages ­ germination, growing and maturing, which require different amounts of water and temperature ­ the climate variables included in the model are long-term averages of the temperature and wetness index for the periods November to December, January to February, and March to April, representing crop progression in the three stages. Assuming a non-linear relationship of farm revenue with the climate variables, quadratic terms for climate variables were also included in the model. The results indicate that most socioeconomic variables are not significant, whereas some climate variables and the corresponding quadratic variables are significant in the model. Further findings are that an increase in the November­December mean temperature and a decrease in the January­February mean rainfall have negative impacts on net farm revenue, whereas an increase in the January­February mean temperature and mean annual runoff has a positive impact. 2 TABLE OF CONTENTS Section Page 1 Introduction 4 2 Overview of Zambia 4 3 Climate analysis 6 4 Climate and crop production in Zambia 6 5 Research methodology: The Ricardian approach 7 6 Data sources 8 7 Model formulation 11 8 Results and discussion 12 9 Conclusion and recommendations 13 References 16 3 1. Introduction Agriculture is becoming an increasingly important sector in the Zambian economy since the mineral sector, which was the backbone of the economy from post-independence times (1964) till the late 1980s, has declined. The agriculture sector generates about 18% to 20% of the country's GDP and provides a livelihood for more than 60% of the population. It employs about two-thirds of the labor force. In 1991 there was a major shift in the country's agricultural policy with the liberalization of the marketing of agricultural inputs and produce. This new policy has resulted in low agricultural productivity as subsistence farmers, who make up about 75% of the entire farming community in Zambia, find difficulty in procuring farm necessities such as seeds and fertilizer in terms of accessibility and funds. Their difficulties are further compounded by the erratic seasonal rainfall patterns which have been experienced by the country in the last 20 years. Almost all subsistence farming in Zambia relies on rainfall. Over the last two decades, the frequency of extreme climate events such as high surface temperatures, floods and droughts has increased over the entire globe. Although such extreme events are attributed to climate variability, they also signal that the earth is going through long- term climate changes in mean temperature and rainfall norms. Zambia has experienced an increase in drought frequency and intensity in the last 20 years. The droughts of 1991 92 , 1994 95 and 1997 98 worsened the quality of life for vulnerable groups such as subsistence farmers (Muchinda, 2001). This study has attempted to assess the economic impacts of long-term climate change such as the increase in mean surface temperature and the decrease in mean seasonal rainfall and mean annual runoff on farming activities in Zambia, in order to provide information for appropriate adaptation policies at national level so as to minimize the adverse impacts of climate change on agriculture. 2. Overview of Zambia 2.1 Physical attributes Zambia covers a land area of 752,615km2 and lies between 22° to 34° east of Greenwich and 8° to 18° south of the equator. The country consists mostly of plateau with an elevation between 950 meters to 1,500 meters above sea level. The country has a sub-tropical climate and vegetation. There are three distinct seasons: a warm wet season stretching from November through April during which 95% of the annual precipitation falls, a cool dry winter season from May to July with the mean temperature varying between 15°C and 27°C, and a hot dry season prevailing from August to October with an average maximum temperature of 27°C to 32°C. The annual rainfall varies from over 1200mm in the north to about 700mm in the central part of the country and less than 700mm in the south. 4 Zambia is divided into three agro-ecological zones with rainfall as the dominant distinguishing climatic factor (Figure 1). Zone I lies in the western and southern part of the country and accounts for about 15% of the land area. It receives less than 800mm of rain annually. It used to be considered the bread basket of the nation but for the last 20 years it has been experiencing low, unpredictable and poorly distributed rainfall. The observed meteorological data suggests that it is currently the driest zone, very prone to drought and has limited potential for crop production. Zone II covers the central part of the country, extending from the east through to the west. It is the most populous zone with over 4 million inhabitants and has the highest agricultural potential. The soils here are relatively fertile. It receives about 800­1000mm of rainfall annually, which is evenly distributed throughout the crop growing season. Zone III spans the northern part of the country and has a population of over 3.5 million. It receives over 1000mm of rainfall annually. The high rainfall in this region has resulted in the soils becoming leached. It is suitable for late maturing varieties of crop. About 65% of the region in this zone has yet to be exploited (Figure 2). 2.2 Administrative attributes Zambia is divided into nine provinces which are further divided into 72 districts. About 60% of the population is concentrated in four provinces: Southern, Central, Lusaka and the Copperbelt along the railway line. 2.3 Agricultural attributes Zambia has an estimated nine million hectares of land (12% of its total land area) suitable for cultivation and 16 million hectares suitable for rangeland grazing. There are approximately one million farmers, who can be grouped into three main categories: small-scale farmers (having five hectares or fewer of farming land), emerging farmers (having a farmland area of between five and 20 hectares) and large-scale commercial farmers (having more than 20 hectares of farmland). However, there are very few with farms comprising more than 60 hectares. The large-scale commercial farmers are located near major urban centers, whereas the majority of small-scale farmers are rural. About 95% of the rural households are occupied with crop production. Small- and medium-scale agriculture is dominated by the production of crops for two main reasons: for subsistence and as a source of income from marketed produce (CSO 1992, 2002a,b; MoA 2000). The major crops grown are maize, sorghum, millet, rice (paddy), wheat, cassava, ground nuts, sunflower, cotton, soya beans, mixed beans and tobacco. Most of these are summer crops which almost entirely depend on rains. Wheat is a winter crop and is grown by large-scale farmers using irrigation (MoA, 2000). Figure 3 indicates the area planted for the various crops in 2001/02 (CSO 2002a,b). 5 3. Climate analysis 3.1 Temperature It has been scientifically established that anthropogenic emissions of greenhouse gases and other atmospheric pollutants are changing global and regional climates. Global surface temperature has increased by over 0.5°C in the past 100 years. A similar rate of warming of about 0.05°C per decade in southern Africa has been observed during the present century (Hulme 1996). The six warmest years in this century in southern Africa have all occurred since 1980. The observed temperature from 32 meteorological stations in Zambia was analyzed to detect trends in temperature change over last 30 years. The mean temperatures computed for the agro- ecological zones for three time periods, November­December, January­February and March­ April, indicate that the summer temperature in Zambia is increasing at the rate of about 0.6°C per decade, which is ten times higher than the global or Southern African rate of increase of temperature (Figures 4 to12). The rate of increase is highest in November­December as compared to other periods across all agro-ecological zones (Hulme 1996). 3.2 Rainfall The rainfall in the southern African region has been decreasing in the last 25 years (Hulme 1996). The annual rainfall anomalies from the 1970­2000 annual averages were computed using observed data from all 32 meteorological stations in Zambia for the agro-ecological zones. For each agro-ecological zone, data was pooled from meteorological stations within the zone. These annual rainfall anomalies, plotted in Figures 13 to 15, indicate that of the 14 years from 1990/1991 to 2003/2004, at least ten years in each agro-ecological zone had below normal rainfall. We further note that the variability in annual totals across the three agro-ecological zones has not been uniform. The southern zone (Zone I) has experienced more severe dry seasons than the central zone (Zone II) in the last 20 years. 4. Climate and crop production in Zambia There is strong scientific evidence that human activities are causing an increase in the concentration of greenhouse gases (GHG) in the atmosphere and that the increasing concentration of GHG is causing long-term changes in mean earth temperature and rainfall normals. This is what we call climate change (Mendelsohn et al. 1994). Climate change is believed to have more severe effects on tropical than on temperate countries. It may lead to a rise in temperature, an increase in the frequency and intensity of droughts and floods, long-term water shortages, poor soils, desertification, disease outbreak, and so on. It is believed to have serious economic consequences for developing countries, many of which are 6 located in the tropics and rely heavily on rainfed agriculture (Mendelsohn et al. 1994; Kurukulasuriya & Rosenthal 2003). Agriculture in Zambia depends on rainfall to a very large extent. Since the 1990s, crop production in the country has faced negative impacts of extreme climate events which are believed to be manifestations of long-term climate change. Zambia has experienced some of its worst droughts and floods in the last two decades. Significant rainfall deficits at critical stages of crop growth have frequently led to a serious shortfall in crop production. Maize is a staple grain in Zambian meals. The yield during the severe drought of 1991/92 was less than half that of 1990/91. Notable shortfalls in maize yield were also recorded in the seasons 1972/73, 1979/80, 1981/82, 1983/84, 1986/87, 1993/94 and 1994/95, most of which were characterized as seasons with below normal rainfall by the Zambian meteorological department. Drought has been the biggest shock to food security in the country during the last two decades (MoA 2000; Muchinda 2001). The impact of extreme climate events has been felt in substantial loss of livestock and fertile soil. Low productivity in the agricultural sector has contributed to a low GDP. In short, changes in the supply of rainfall, whether in the total volume or in its distribution within a season, have enormous consequences for agriculture in Zambia. Maize production has been quite variable, as can be seen in Figure 16. In some years the yield has been only 40% of the long-term average. Major factors contributing to this low yield have been the long dry spells within a season and the shorter rainfall seasons which have been experienced by the country in the past 20 years. Crop production data obtained from the Central Statistical Office (CSO) indicates that the major crop of the Southern, Central and Eastern Provinces is maize which occupies more than 70% of the total area cultivated in these provinces. In the Northern Province, only about 18% of the total cultivated area is planted with maize. Table 1 shows the percentage of the area planted with various crops in the Southern, Central, Eastern and Northern Provinces. As may be noted from Figures 17 and 18, the maize yield in the Southern, Central and Eastern Provinces shows a high positive correlation with the total seasonal rainfall. Figure 19 shows that the maize yield in the Northern Province does not indicate a correlation with the total seasonal rainfall. The reason is evident from Table 1, which shows that maize is the main crop in the Southern, Central and Eastern Provinces, whereas little maize is grown in the Northern Province. Sorghum is a crop widely grown throughout the country, though the number of households engaged in its production is relatively small as compared to the number engaged in growing maize. It is promoted especially in drought prone areas as it is a drought tolerant crop. Sorghum production is being promoted so as to improve food security in areas where short rain periods are frequent. In conclusion, climate change is a phenomenon which represents a risk to sustainable agricultural practices and has significant consequences for food security and the country's GDP. Its probable effects can, however, be estimated and guarded against. 7 5. Research methodology: The Ricardian approach The Ricardian approach is a cross-sectional analysis of farm value across different climate zones. It measures the effect of climate on farm value. It assumes that farm value has a quadratic relationship with climate variables such as temperature and precipitation (Mendelsohn et al. 1994; Dinar et al. 1998). It aims to provide insights into sustainable agricultural practices that are compatible with the constraints of the present climate. It places emphasis on meeting the long- term needs of small-, medium- and large-scale farmers to minimize the negative impacts of climate change on agriculture (Munalula et al. 1999). One major drawback of the Ricardian approach is that it takes into account only current farming practices. Hence it does not capture future changes in agriculture as a result of changes in farm technology or carbon dioxide fertilization. Zambia has abundant land available free of cost for subsistence farming. Therefore land value does not appear an appropriate response variable in the Ricardian multiple linear regression econometric model. In this analysis net farm revenue per hectare (NRh) has been used as response variable to assess farm performance as a function of climate, soil, hydrological and socio-economic variables. The net farm revenue has been approximated from the following equation: n m NRh = i=1 P iCi - Yj Xi j j=1 Ah where NRh = Net farm revenue per hectare for household h Pi = Unit price of crop i Ci = Quantity produced of crop i Xij = Quantity of input j purchased for producing crop i Yj = Unit price of input j Ah = Total area planted by household h 8 The explanatory variables are classified into the following categories: climate, soils, hydrological indicators, socio-economic factors and planting and harvesting months. 6. Data sources 6.1 Household data A field survey by means of a questionnaire was carried out with the assistance of trained supervisors and enumerators from the CSO of Zambia on a total of 1015 households. The survey included small-, medium- and large-scale households from 30 districts out of the total of 72 districts in Zambia. The sample frame for survey was designed on stratified multistage cluster sampling technique which had four stages of selection. Stage 1: Selection of number of districts from each agro-ecological zone Zambia is classified into three agro-ecological zones on the basis of total annual rainfall and soil types. These three zones are widely diverse in area, soil type, population density, number of districts and agricultural practices, so to have a representative sample it was considered inappropriate to select the same number of districts from each zone. The probability proportions of districts (PPi) included in the sample from each of the three zones was therefore estimated as follows: PPi = pi p i Di = PPi * N Where pi = total number of districts in the ith zone N = total number of districts included in the sample. Di = number of districts included in the sample from the ith zone Stage 2: Selection of districts from each agro-ecological zone Systematic sampling was used for selecting districts from zones. The method employed was as follows: 9 Districts in Zone i Number of households Cumulative households X1 n1 n1 X2 n2 n1 + n2 . . . . . . . . . XN nN n1 + n2 + .... + nN Define sampling interval = (n1 + n2 +.....+ nN )and denote it by K. Di Choose a random number j between 1 and K, then select districts having cumulative households j, j + K, j + 2K, ...., j + (Di ­1) K. Stage 3: Selection of three SEAs from each sampled district Each district in Zambia has been divided into Standard Enumeration Areas (SEA) as an Agricultural Census exercise by the Central Statistics Office of Zambia who carry out crop forecasting and post-harvesting surveys every year (CSO 1992, 2002a,b). The number of SEAs in each district is different. The selection of three SEAs from each district included in the sample was also done by systematic sampling. The sampling interval was calculated as: Total households in Districts i Sampling interval = 3 Stage 4: Selection of ten households from each sampled SEA Finally, after selecting 30 districts, three SEAs from each district, to select ten households from each SEA, serial numbers were assigned to households in each SEA and the sampling interval was calculated by: Last serial number in SEA i 10 Systematic sampling was applied to select farm households from each sampled district. In total a sample of 900 small- and medium-scale farm households were selected. Of the approximately 3000 commercial farms in the country, 15 from each of the nine provinces were also included in the survey. 10 6.2 Climate data Satellite temperature and soil wetness data was obtained from the US Department of Defense. This data is derived from a set of polar orbiting satellites that are equipped with sensors to detect microwaves through clouds. The satellites make daily overpasses between 6 am and 6pm. The monthly averages for the period 1988­2004 for temperature and wetness indices were used in the analyses (Basist et al. 2001). 6.3 Soil and hydrological data Soil data was obtained from the Food and Agriculture Organization (FAO) database (FAO 2003). From various soil types classified by the FAO world map, the soil types which were found in Zambia in more than 30% of the sample units were included in the study. The runoff data was obtained from the University of Colorado (IWMI & University of Colorado 2003). 7. Model formulation 7.1 Response variable The response variable is the net farm revenue per hectare (nr1_3), which is computed as the total revenue from the farm produce minus the cost of seeds and fertilizer. The summary statistics of the net revenue in US$ computed from the household survey are shown in Table 2. 7.2 Explanatory variables Climate variables The major cropping season in Zambia runs from November through April. Agriculture in this season depends on the rains to a very large extent. There is a minor cropping season from June through September. The main crop of the minor season is wheat, which is grown only by commercial farmers using irrigation. This minor season is not included in the analysis. A plant has three growth stages: germination (initial period), growth (middle period) and maturing (ripening period). To correspond to these stages, the long-term (1988­2004) mean climate variables (temperature measured in degrees Celsius and wetness index measured on a scale of 0 to 10) were generated for three time composites: composite 1 (November + December), composite 2 (January + February) and composite 3 (March + April). Assuming a quadratic relationship of climate variables with the net farm revenue, the square terms of the temperature and wetness index for each of the composites were added to the model. The summary statistics of the climate variables are shown in Table 3. 11 Soil variables Soil types from the FAO classification which appear in at least 30% of the sampled units are included in the model. Tables 4 to 6 show the existence of the selected soil types in Zambia. The values of the soil variable are the proportions of plot area covered by a particular soil type. Hydrological variables Runoff is defined as excess precipitation which is not absorbed by soils. It runs on the soil surface and eventually joins a stream. Runoff takes away soil nutrients. Excessive runoff may have a negative impact on farm yield. Table 7 shows summary statistics of the sample mean runoff (in cm). Assuming a hill-shaped relationship of net farm revenue with mean annual runoff, a quadratic term of mean annual runoff has been added to the model. Socio-economic variables Table 8 summarizes the values in kilometers of variables distance to market from where inputs were purchased (distpmktkm) and output sold (distsmktkm). Table 9 shows a summary of two variables: access to public extension services (extc) and access to credit (inc1), each of which has two response values: 1 representing household responding positive to receiving public extension services or having access to different types of credit and 2 if the response is negative to either of these variables. Variables landhhcd1, landhhfd1 and landhhmd1 represent number of days spent by household children, females and males in land preparation for planting. The summary statistics from the sample are shown in Table 10. Other socio-economic explanatory variables included in the model are shown in Table 11. Variables representing planting and harvesting months Dummy variables representing the planting and harvesting months were included in the model. These variables are shown in Table 12. 8. Results and discussion The highlighted values in Table 13 show the significant regressors at the 5% level of significance. We note that the net farm revenue (nr1_3) has a quadratic relationship with the 12 climate variables tcom1, tcom2 and wcom2 and hydrological variable mean runoff (roff_mean). Net revenue (nr1_3) has a U-shaped relationship with November­December temperature (tcom1), whereas the relationship of net revenue (nr1_3) with the other two climate variables: January­February temperature (tcom2) and January­February wetness (wcom2) and the mean runoff variable (roff_mean) is hill-shaped. The other significant variables are soil types chromic luvisosls (perclcMFU) and lithosols (percilqHS), farm type (farmtype) and the June harvest period (harvjune). Analyzing the U-shaped relationship of net revenue (nr1_3) with the November­December temperature (tcom1), we note that net revenue decreases through values greater than the sample mean value 21.72°C for tcom1 and keeps decreasing up to 23.48°C (turning point) when net revenue is at its minimum. Marginal net revenue per hectare for an increase of 1°C in the mean temperature of November and December is US$322.628, indicating that if the temperature rises at the beginning of the cropping season, when plants are germinating, this may have a negative effect on the crop. Analyzing the hill-shaped relationship of net revenue with the average temperature of January and February (tcom2), we note that net revenue increases as tcom2 increases through values greater than the sample mean 19.7°C and attains the maximum value at a temperature of 20.7°C. The marginal net revenue per hectare for an increase of 1°C in the mean temperature of January and February is US$315.70, indicating that if the temperature rises during the growing stage of the plant, this may have a positive effect on the crop. Usually temperatures are lower in January­ February (tcom2) than in November­December (tcom1). It implies that in the mean range of 10­ 20°C a 1oC increase in the mean temperature in January­February (tcom2) may have a positive effect on crop growth. We note from Table 13 that the quadratic term of the mean wetness index for the period January and February (wcom2sq) is significant. Precipitation is represented by a wetness index which indicates the extent of moisture in the land surface area. The total range of the wetness index is from 0 to 10. The wetness index for sampled households ranges from 0.5 to 5.5. Therefore an index of one represents about 20% precipitation for Zambia. The marginal net revenue per hectare for a unit increase in the mean wetness index (20% precipitation) for January and February is US$334.67. Since the negative coefficient of the square term confirms the hill-shaped relationship of the net farm revenue with variable January­ February wetness (wcom2), we can deduce that a decrease of about 20% in the precipitation for this period can reduce the net revenue by about US$334.67. The annual mean runoff (roff_mean) and its squared term (roff_mean2) are both significant. The negative coefficient of roff_mean2 indicates a hill-shaped relationship of net revenue per hectare with the mean runoff variable (roff_mean). The sample mean of the mean runoff is 30.03cm, which lies on the left part of the hill-shaped curve of net revenue, indicating that the net revenue will increase per cm increase in mean runoff until the optimum level (the turning point) which happens at 32.5cm. The marginal revenue for a 1cm increase in runoff from the long-term annual average is US$3.39. 13 Table 14 shows the change in net revenue per hectare corresponding to a unit increase (1°C) in mean temperature in November­December and January­February. It also indicates the changes in net revenue per hectare corresponding to a 20% reduction in the mean precipitation in January­February (wcom2) and a 1 cm increase in annual mean runoff respectively. We further calculate the marginal net revenue per hectare for the above climate scenarios as a percentage of the observed mean net revenue per hectare. Table 15 summarizes the percentage values. Losses of more than 100% of the mean net revenue are attributed to the fact that in the event of crop failure or low harvest due to unfavorable climate conditions the farmer's total loss will be the sum of the costs of the farm inputs (seeds and fertilizer) and the proportion of the gross revenue that he has lost, which he would have gained in a normal year. 9. Conclusion and recommendations This study has attempted to estimate the economic impact of climate change on rain dependent agriculture in Zambia by regressing the net farm revenue on the climate, soil, hydrological and socio-economic variables. The results indicate that an increase in mean temperature in November and December and a reduction in mean precipitation in January and February have negative impacts on net farm revenue, whereas an increase in mean temperature in January and February and an increase in mean annual runoff have positive impacts on net farm revenue. Hence we find that agricultural production in Zambia is subject to the uncertainties of extreme climate events, which are indicative of an increasing mean temperature and reduction in total seasonal rain on a long-term time scale. Assuming that the climate induced impact on farm produce will be slow, the national government and all other stakeholders should respond to climate change by formulating and implementing adaptive measures to minimize the negative effects of climate on agriculture which may pose a serious threat to food security. Zambia, being a signatory to the United Nations Framework Convention on Climate Change (UNFCCC), has an obligation to reduce greenhouse gas emissions to mitigate climate change. However, in addition to implementing mitigation strategies there is an urgent need to make appropriate adjustments in farming systems to adapt to climate change which is occurring gradually but may have irreversible impacts. Some recommended adaptations are listed below. Farm level adaptations · Development of new varieties of crops which mature faster and are heat resistant. · Diversification from traditional crops to other types of crops which can withstand drought and higher temperatures, such as millet and sorghum. · Rotation of land use between crop and livestock to replenish soil nutrients. 14 · A shift to more productive new lands, for example moving farming from the Southern to the Central Province. National level adaptations · Provision of a rural credit facility to enable subsistence farmers to buy new varieties of seeds and fertilizer. · Dissemination of information to farmers on various adaptation options through extension services. · Removal of subsidies on crops which do not perform well in a changing climate. · Formulation of appropriate policies for marketing agricultural input and output products which are advantageous to subsistence farmers in particular, as this group makes up almost 80% of the entire farming community in the country and is most vulnerable to the negative impacts of climate. · Investment in research on agricultural issues such as climate resistant crop varieties, water harvesting, irrigation schemes, water rights, etc. · Investment in technological innovations, seed banks, etc. · A feasibility study to migrate farmers from agro-ecological Zone I to Zone II which has more potential for agriculture. · Provision of opportunities for alternative employment in non-farming activities to enable rural farmers make a livelihood. · Allocation of adequate funding to the National Meteorological Department to procure measuring equipment and build capacity in climate data collection, storage, analysis and forecasting. · Dissemination of climate forecasts in everyday language, not in scientific terms. · Formulation of policies to encourage the NGOs, private sector and civil societies to complement the government's efforts to implement adaptation policies. This study did not focus on the impacts of climate change on individual crops. Further research that investigates how the different crops respond to climate change is needed. 15 REFERENCES Basist A et al., 2001. Using the Special Sensor Microwave Imager to monitor surface wetness. Journal of Hydrometeorology 2: 297­308. CSO (Central Statistical Office), 1992. National Census of Agriculture, 1990/1992. Central Statistical Office, Lusaka, Zambia. CSO (Central Statistical Office), 2002a. Post-harvest report for large-scale agricultural holdings. Agricultural and Pastoral Production, 2001/2002, Central Statistical Office, Lusaka, Zambia. CSO (Central Statistical Office), 2002b. Structural type and post-harvest data for small- and medium-scale farmers. Agricultural and Pastoral Production, 2001/2002, Central Statistical Office, Lusaka, Zambia. Dinar A et al. (eds), 1998. Measuring the Impact of Climate Change on Indian Agriculture. World Bank Technical paper 402. Washington, DC. FAO (Food and Agriculture Organization), 2003. The digital soil map of the world: Version 3.6 (January), Rome, Italy. Hulme, M, 1996. Climate change and Southern Africa. Climatic Research Unit, University of East Anglia, UK. Kurukulasuriya P and Rosenthal S, 2003, Climate Change and Agriculture: A Review of Impacts and Adaptations, The World Bank Environment Department Mendelsohn R, Nordhaus WD & Shaw D, 1994. The impact of global warming on agriculture: A Ricardian analysis. American Economic Review 84: 753­771. MoA (Ministry of Agriculture), 2000. Agricultural Statistics Bulletin. Early Warning Unit, Ministry of Agriculture, Food and Fisheries, Zambia. Muchinda M, 2001. Drought incidence in Zambia over the thirty-year period 1979/71­ 1999/2000. Second International Conference on Tropical Climatology, Meteorology and Hydrology. Brussels, Belgium, December. Munalula T, Nkomoki J, Kamocha S & Nawiko M, 1999. Estimating the impact of climate change on rainfed agriculture: A cross-sectional analysis of impacts on medium and small scale agricultural revenue in Zambia. Ministry of Environment and Natural Resources, Lusaka. Strzepek K & McCluskey A, 2006. District level hydroclimatic time series and scenario analysis to assess the impacts of climate change on regional water resources and agriculture in Africa. CEEPA Discussion Paper No 13, Centre for Environmental Economics and Policy in Africa, University of Pretoria. 16 Table 1: Provincial crop area (%) planted in 2002/03 Crop % crop area in % crop area in % crop area in Northern Southern Province Central and Eastern Province Provinces Maize 83.9 72.4 18.3 Sorghum 3.6 1.7 1.1 Millet 1.8 1.3 12.6 Paddy rice 0.1 2.1 Groundnuts 5.7 16.0 3.5 Soya beans 0.3 0.01 Sunflower 0.4 0.1 0.006 Mixed beans 1.7 1.5 9.9 Cassava 0.9 3.9 40.8 Sweet potatoes 1.9 2.4 1.5 Table 2: Net farm revenue per hectare in US$ Variable Observations Mean Std dev Min Max nr1_3 955 132.77 212.2 -340.55 3572.26 Table 3: Summary statistics of climate variables Variable Obser- Mean Std dev Min Max vations Mean temperature 1008 21.71873 1.678403 19.09018 24.31125 (November-December) (tcom1) Mean temperature 1008 19.72414 1.251792 17.57925 22.13305 (January-February) (tcom2) Mean temperature (March- 1008 19.52797 1.193852 17.48295 21.81078 April) (tcom3) Mean wetness index 1008 1.384871 .9449548 .33275 5.381725 (November-December) (wcom1) Mean wetness index 1008 1.432622 .9649444 .384875 5.544375 (January-February) (wcom2) Mean wetness index 1008 1.74015 .9509898 .669375 5.3485 (March-April) (wcom3) 17 Table 4: Soil type ­ Chromic luvisols (perc1cMFU) Proportion of plot area Household Percent covered by soil Perc1cMFU plots 0 647 66.70 .1 62 6.39 .2 142 14.64 .3 66 6.80 .6 20 2.06 .7 33 3.40 Total 970 100 Table 5: Soil type ­ Orthic ferralsols (percfoFU) Proportion of plot area Household Percent covered by soil PercfoFU plots 0 342 35.26 .1 30 3.09 .2 65 6.70 .3 64 6.60 .4 34 3.51 .5 116 11.96 .6 30 3.09 .7 210 21.65 .8 27 2.78 .9 34 3.51 1 18 1.86 Total 970 100 18 Table 6: Soil type ­ Lithosols (percilqHS) Proportion of plot area Household Percent covered by soil percilqHS plots 0 621 64.02 .1 51 5.26 .2 173 17.84 .3 41 4.23 .4 32 3.30 .5 20 2.06 .6 32 3.30 Total 970 100.00 Table 7: Mean annual runoff (in cm) Variable Observations Mean Std dev Min Max roff_mean 1008 30.0301 13.32263 11.14925 50.56691 Table 8: Variables distpmktkm and distsmktkm Variable Observations Mean Std dev Min Max Distant to input market in 959 23.7122 31.49476 0 360 km (distpmktkm) Distant to output market 950 20.82 32.91 0 595 in km (distsmktkm) Table 9: Variables extc and inc1 Value of the variable Received public Borrowed extension for crop money (inc1) production (extc) Yes: 1 370 86 No: 2 631 911 Total households 1001 997 19 Table 10: Household labor Variable Obser- Mean Std dev Min Max vations Child household labor 764 9.626963 19.48931 0 288 (landhhcd1) Adult female household 942 28.92675 27.52775 0 240 labor (landhhfd1) Adult male household 951 29.9653 29.07663 0 240 labor (landhhmd1) Table 11: Socio-economic variables included in the regression Variable name Description Farm type Small, medium, large farm Hhhdocc1 Primary occupation of head of household: farming or non-farming Hhsize Size of the household Transport Means of transport used to carry farm produce for selling in the market. Possible values: walk, animal, cart, truck or other. Table 12: Variables representing planting and harvesting months Variable name Description plantoct,plantnov planting month (October, November) harvmar, harvapr, harvmay, harvjune, harvjuly harvesting month (March, April, May, June and July) 20 Table 13: Estimated regression coefficients of net farm revenue per hectare (nr1_3) Estimated Significance Net crop revenue per hectare (nr1_3) coefficients t-statistic level Mean temperature (November-December): (tcom1) -4282.98 -2.34 0.019 Mean temperature (January-February): (tcom2) 6232.428 3.09 0.002 Mean temperature (March-April): (tcom3) -1594.81 -0.96 0.338 Mean wetness index (November-December): (wcom1) -99.2666 -0.17 0.864 Mean wetness index (January-February): (wcom2) 242.6801 0.64 0.522 Mean wetness index (March-April): (wcom3) 12.12254 0.09 0.929 Mean temperature squared (November-December): (tcom1sq) 91.16825 2.26 0.024 Mean temperature squared (January-February): (tcom2sq) -150.018 -3.16 0.002 Mean temperature squared (March-April): (tcom3sq) 45.31013 1.08 0.281 Mean wetness index squared (November-December): (wcom1sq) 387.3936 1.25 0.211 Mean wetness index squared (January-February): (wcom2sq) -201.871 -2.35 0.019 Mean wetness index squared (March-April): (wcom3sq) -39.7648 -1.17 0.242 Mean annual runoff: (roff_mean) 44.23606 2.14 0.033 Mean annual runoff squared: (roff_mean2) -0.68425 -2.35 0.019 Soil type ­ Chromic Luvisols: (percfoFU ) -46.6943 -0.6 0.552 Soil type ­ Orthic Ferralsols: (perclcMFU) -503.654 -2.91 0.004 Soil type - Lithosols: (percilqHS) -242.336 -2.39 0.017 Distance to input market in km: (distpmktkm) 0.001513 0.01 0.995 Distance to output market in km (distsmktkm) -0.10097 -0.61 0.544 Received public extension services: (extc) -20.1031 -1.37 0.17 Farm type: (farmtype) 32.42927 1.93 0.054 Household size: (hhsize) 5.080366 1.58 0.114 Borrowed money: (inc1 ) 34.2957 1.76 0.079 Primary occupation of household head (farming/non-farming): (hhhdocc1) 0.89528 0.38 0.704 Child household labour: (landhhcd1) -0.88024 -1.45 0.148 Adult female household labour: (landhhfd1) -0.03874 -0.18 0.854 Adult male household labour: (landhhmd1) 0.037366 0.18 0.854 Transportation to output market: (transport ) -1.76893 -0.55 0.584 Planting month (October): (plantoct) -65.4114 -1.55 0.123 Harvesting month (March): (harvmar) 15.55523 0.13 0.893 Harvesting month (April): (harvapr) -89.4216 -1.63 0.103 Harvesting month (May): (harvmay) -8.38 -0.12 0.903 Harvesting monthe (June): (harvjune) -291.951 -2.04 0.042 Constant -1480.57 -0.15 0.883 R-Squared = 0.2663 21 Table 14: Change in marginal net revenue in US$ with respect to unit change in mean temperature, precipitation and runoff Climate scenario Marginal net revenue per hectare (US$) 1°C increase in mean temperature (November­December) -322.62 (tcom1) 1°C increase in mean temperature (January­February) 315.70 (tcom2) 20% reduction in mean precipitation (January­February) -334.67 (wcom2) 1cm increase in mean annual runoff (roff_mean) 3.39 Table 15: Change in marginal net revenue as % of mean net revenue Climate scenario Marginal net revenue per hectare expressed as % of the observed mean net revenue per hectare 1°C increase in mean temperature Loss of 243% (November­December) (tcom1) 1°C increase in mean temperature Gain of 237% (January­February) (tcom2) 20% reduction in mean precipitation Loss of 252 % (January­February) (wcom2) 1cm increase in mean annual runoff Gain of 2.5 % (roff_mean) 22 Figure 1: Agro-ecological zones of Zambia Note: See text for explanation on Zone characteristics Figure 2: Mean annual rainfall 23 Maize Seed Maize Wheat Groundnuts Soya Beans Sunflower Virginia Tobacco Rice Sorghum Millet Seed Cotton Cassava Sweet Potatoes Figure 3: Crop area planted in season 2001/02 28.50 28.00 27.50 27.00 26.50 TEMP 26.00 25.50 MEAN25.00 24.50 24.00 23.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 4: November­December mean temperature in Zone I 24 24.50 24.00 23.50 MP 23.00 TE 22.50 AN 22.00 ME 21.50 21.00 20.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 5: January­February mean temperature in Zone I 25.50 25.00 24.50 MP 24.00 TE 23.50 AN 23.00 ME 22.50 22.00 21.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 6: March­April mean temperature in Zone I 26.00 25.50 25.00 TEMP 24.50 24.00 MEAN 23.50 23.00 22.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 7: November­December mean temperature in Zone II 25 25.00 24.50 24.00 TEMP 23.50 23.00 MEAN 22.50 22.00 21.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 8: January­February mean temperature in Zone II 24.50 24.00 23.50 23.00 TEMP 22.50 22.00 MEAN21.50 21.00 20.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 9: March­April mean temperature in Zone II 26 25.00 24.50 24.00 23.50 TEMP 23.00 22.50 MEAN 22.00 21.50 21.00 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure10: November­December mean temperature in Zone III 23.50 23.00 22.50 TEMP 22.00 MEAN21.50 21.00 20.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 11: January­February mean temperature in Zone III 27 24.50 24.00 23.50 23.00 TEMP 22.50 22.00 MEAN 21.50 21.00 20.50 1970 1975 1980 1985 1990 1995 2000 2005 YEAR Figure 12: March­April mean temperature in Zone III 1.00 0.50 0.00 -0.50 ANOMALY-1.00 -1.50 -2.00 90/91 92/93 94/95 96/97 98/99 00/01 02/03 YEAR Figure 13: Annual rainfall anomalies in agro-ecological Zone I from 1970­2000 annual average rainfall 28 1.00 0.50 0.00 -0.50 -1.00 ANOMALY-1.50 -2.00 -2.50 90/91 92/93 94/95 96/97 98/99 00/01 02/03 YEAR Figure 14: Annual rainfall anomalies in agro-ecological Zone II from 1970-2000 annual average rainfall 1.00 0.50 0.00 -0.50 ALY -1.00 -1.50 -2.00 ANORM -2.50 -3.00 -3.50 -4.00 90/91 92/93 94/95 96/97 98/99 00/01 02/03 YEAR Figure 15: Annual rainfall anomalies in agro-ecological Zone III from 1970-2000 annual average rainfall 29 1400000 1200000 1000000 800000 Tonnes 600000 Metric 400000 200000 0 1990 1992 1994 1996 1998 2000 2002 2004 Years Figure 16: Maize production 1000 600000 900 500000 800 700 400000 600 (MT) Rainfall 500 300000 Production 400 Rainfall(mm) 300 200000 Production 200 100000 100 0 0 1987 1989 1991 1993 1995 1997 1999 2001 2003 Year Figure 17: Rainfall and maize production for Southern Province 30 1200 1200000 1000 1000000 800 800000 (MT) Rainfall 600 600000 Production Rainfall(mm) 400 400000 Production 200 200000 0 0 1987 1989 1991 1993 1995 1997 1999 2001 2003 Year Figure 18: Rainfall and maize production for Central and Eastern Provinces 1800 200000 1600 180000 1400 160000 1200 140000 120000 (MT) 1000 Rainfall 100000 800 Production 80000 Rainfall(mm) 600 60000 Production 400 40000 200 20000 0 0 1987 1989 1991 1993 1995 1997 1999 2001 2003 Year Figure 19: Rainfall and maize production for Northern Province 31 Policy Research Working Paper Series Title Author Date Contact for paper WPS4263HIV/AIDSandSocialCapitalina AntonioC.David June2007 A.David Cross-SectionofCountries 82842 WPS4264FinancingofthePrivateSectorin ConstantinosStephanou June2007 S.Coca Mexico,2000­05:Evolution, EmanuelSalinasMuñoz 37474 Composition,andDeterminants WPS4265TheStructureofImportTariffsinthe OleksandrShepotylo June2007 P.Flewitt RussianFederation:2001­05 32724 WPS4266TheEconomicCommunityofWest SimpliceG.Zouhon-Bi June2007 S.Zouhon-Bi AfricanStates:FiscalRevenue LyngeNielsen 82929 ImplicationsoftheProspective EconomicPartnershipAgreement withtheEuropeanUnion WPS4267FinancialIntermediationinthe HeikoHesse June2007 G.Johnson Pre-ConsolicatedBankingSectorin 34436 Nigeria WPS4268PowertothePeople:Evidencefrom MartinaBjörkman June2007 I.Hafiz aRandomizedFieldExperimentofa JakobSvensson 37851 Community-BasedMonitoringProject inUganda WPS4269ShadowSovereignRatingsfor DilipRatha June2007 N.Aliyeva UnratedDevelopingCountries PrabalDe 80524 SanketMohapatra WPS4270Jump-StartingSelf-Employment? RitaAlmeida June2007 A.Bonfield EvidenceamongWelfareParticipants EmanuelaGalasso 31248 inArgentina WPS4271Construction,Corruption,and CharlesKenny June2007 C.Kenny DevelopingCountries 33540 WPS4272Migration,Remittances,Poverty, DavidMcKenzie July2007 M.Sasin andHumanCapital:Conceptualand MarcinJ.Sasin 36877 EmpiricalChallenges WPS4273RulesofOriginandtheWebofEast MiriamManchin July2007 L.Yeargin AsianFreeTradeAgreements AnnetteO.Pelkmans-Balaoing 81553 WPS4274AreLaborRegulationsDriving MohammadAmin July2007 S.Narsiah ComputerUsageinIndia'sRetail 88768 Stores? WPS4275CanForeignLobbyingEnhance KishoreGawande July2007 V.Cornago Development?TheCaseofTourism WilliamMaloney 84039 intheCaribbean GabrielV.MontesRojas WPS4276HumanCapital,TradeLiberalization, TomKrebs July2007 V.Cornago andIncomeRisk PravinKrishna 84039 WilliamMaloney WPS4277ClimateChangeAdaptationinAfrica: SungnoNiggolSeo July2007 P.Kokila AMicroeconomicAnalysisof RobertMendelsohn 33716 LivestockChoice WPS4278EndogenousIrrigation:TheImpactof PradeepKurukulasuriya July2007 P.Kokila ClimateChangeonFarmersin RobertMendelsohn 33716 Africa WPS4279TheImpactofClimateChangeon SungnoNiggolSeo July2007 P.Kokila LivestockManagementinAfrica: RobertMendelsohn 33716 AStructuralRicardianAnalysis Policy Research Working Paper Series Title Author Date Contact for paper WPS4280 GovernanceMattersVI:Aggregate DanielKaufmann July2007 R.Bonfield andIndividualGovernance: AartKraay 31248 Indicators,1996-2006 MassimoMastruzzi WPS4281 CreditGrowthInEmergingEurope: SophieSirtaine July2007 S.Sirtaine ACauseForStabilityConcerns? IliasSkamnelos 87006 WPS4282 AreCashTransfersMadetoWomen NorbertSchady July2007 I.Hafiz SpentLikeOtherSourcesofIncome JoséRosero 37851 WPS4283 InnovationShortfalls WilliamMaloney July2007 V.Cornago AndrésRodríguez-Clare 84039 WPS4284 CustomerMarketPowerandthe NeeltjeVanHoren July2007 M.Gamboa ProvisionofTradeCredit: 34847 EvidencefromEasternEuropeand CentralAsia WPS4285 PovertyAnalysisUsingAn J.A.L.Cranfield July2007 P.Flewitt InternationalCross-CountryDemand PaulV.Preckel 32724 System ThomasW.Hertel WPS4286 InstitutionalEffectsasDeterminants JesúsÁlvarez July2007 S.Baksh ofLearningOutcomes:Exploring VicenteGarcíaMoreno 31085 StateVariationsinMexico HarryAnthonyPatrinos WPS4287 ACross-CountryAnalysisofPublic MartinMeleckyy July2007 M.Rosenquist DebtManagementStrategies 82602 WPS4288 ActualCropWaterUseinProject RobinaWahaj July2007 P.Kokila CountriesASynthesisatthe FlorentMaraux 33716 RegionalLevel GiovanniMunoz WPS4289 SensitivityofCroppingPatternsin AlexanderLotsch July2007 P.Kokila AfricatoTransientClimateChange 33716 WPS4290 TheImpactsofClimateChange KennethStrzepek July2007 P.Kokila onRegionalWaterResources AlyssaMcCluskey 33716 andAgricultureinAfrica