measuring the economic impact of climate change on ... - UNDPCC.org

Climate Change Economics, Vol. 1, No. 1 (2010) 33–55 © World Scientific Publishing Company DOI: 10.1142/S2010007810000066

MEASURING THE ECONOMIC IMPACT OF CLIMATE CHANGE ON AFRICAN AGRICULTURAL PRODUCTION SYSTEMS

CHARLES NHEMACHENA* Council for Scientific and Industrial Research P. O. Box 395 Brummeria Pretoria 0001, South Africa [email protected] [email protected] RASHID HASSAN Centre for Environmental Economics and Policy in Africa (CEEPA) University of Pretoria, Department of Agricultural Economics and Rural Development Pretoria 0002, South Africa [email protected] PRADEEP KURUKULASURIYA Energy and Environment Group/Climate Change Adaptation United Nations Development Programme, New York [email protected] This study measured the economic impacts of climate change on crop and livestock farming in Africa based on a cross-sectional survey of over 8000 farming households from 11 countries in east, west, north and southern Africa. The response of net revenue from crop and livestock agriculture across various farm types and systems in Africa to changes in climate normals (i.e. mean rainfall and temperature) is analysed. The analyses controlled for effects of key socioeconomic, technology, soil and hydrological factors influencing agricultural production. Results show that net farm revenues are in general negatively affected by warmer and drier climates. The small-scale mixed crop and livestock system predominantly typical in Africa is the most tolerant whereas specialized crop production is the most vulnerable to warming and lower rainfall. These results have important policy implications, especially for the suitability of the increasing tendency toward large-scale mono-cropping strategies for agricultural development in Africa and other parts of the developing world in light of expected climate changes. Mixed crop and livestock farming and irrigation offered better adaptation options for farmers against further warming and drying predicted under various future climate scenarios. Keywords: Climate change; impacts; agriculture; Africa; Ricardian model.

*Corresponding

author.

33

34 C. Nhemachena, R. Hassan & P. Kurukulasuriya

1. Introduction Local ecosystems provide the main source of livelihoods for many of the world’s poor. Most of the rural poor in Sub-Saharan Africa rely, for their livelihoods and food security, on highly climate-sensitive, rain-fed, small-scale subsistence farming, pastoral herding and direct harvesting of natural services of ecosystems such as forests and wetlands (Mitchell and Tanner, 2006; IPCC, 2001). The productivity of this livelihood base is highly vulnerable to climate-related stresses such as changes in temperature, precipitation (both amount and variability) and increased frequency of droughts and floods (IPCC, 2001). Higher temperatures declining rainfall patterns, as well as an increasing frequency of extreme climate events (such as droughts and floods) are expected to be found in the future climate of the tropics (Mitchell and Tanner, 2006; IPCC, 2001). In southern Africa, for example, rainfall patterns show a declining trend of summer rainfall (about 20%) from 1950–1999 and a high frequency of droughts predicted to intensify in the 21st century. Predictions for 2050 by the US National Center for Atmospheric Research show that the declining trend in rainfall is set to continue and the region is expected to be 10–20% drier than in the previous 50 years (Mitchell and Tanner, 2006). These predicted changes in climate are expected to have differential impacts on agricultural productivity and food security and other sectors across spatial and temporal scales. In the tropics, and Africa in particular, changes in climate are expected to be detrimental to agricultural livelihoods (Dinar et al., 2008; Dixon et al., 2001). The vulnerability of the majority of the poor in Africa to climate-related stresses is worsened by widespread poverty, HIV/AIDS, lack of access to resources (e.g. land and water) and management capabilities, low levels of wealth, technology and education, ineffective institutional arrangements and lack of social safety nets (IPCC, 2001). Methods and results of recent continent-wide research efforts to measure the economic impacts of climate change on African agriculture are summarized in (Dinar et al., 2008). Recent studies suggest that agricultural crop productivity in a number of regions in Africa will be adversely affected by warming above current levels (Kurukulasuriya et al., 2006; Kurukulasuriya and Mendelsohn, 2008; Seo and Mendelsohn, 2008). These studies analysed impacts on dryland crops, irrigated crops and livestock separately. This represents an important limitation since the choice between crop and livestock production or their combination (mixed systems) is a non-separable choice endogenous to agricultural producers’ responses to varying climates and other circumstances. The decision of what to produce, and how, is accordingly an important adaptation mechanism in the face of a changing climate and other ecological and economic circumstances. This is of particular importance to Africa where the majority of poor, small-scale farmers practice mixed crop-livestock agriculture and few specialize in crop or livestock only. The primary objective of this study is therefore to measure the aggregate impact of climate change on income from all agricultural production systems (crop, livestock and mixed) in Africa and predict future impacts under various climate scenarios. In

Measuring the Economic Impact of Climate Change on African Agricultural Production Systems 35

addition, the study also measures and compares impacts on specialized crop and livestock farms. The results will be contrasted with findings of other regional studies using the same data but generating different climate response functions for crop and livestock farming separately (Kurukulasuriya et al., 2006; Kurukulasuriya and Mendelsohn, 2008) and (Seo and Mendelsohn, 2008). The study adopts the crosssectional (Ricardian) approach to measure the impact of changes in climate attributes (rainfall and temperature levels) on net revenue from crop and livestock farming, controlling for other production factors. Responses of different production systems are analysed under irrigation and dryland conditions. 2. Study Approach and Methods Impacts of climate change have been estimated using two main approaches: (a) structural modeling of crop and farmer response, which combines crop agronomic response with economic/farmer management decisions and practices; and (b) spatial analogue models that measure observed spatial differences in agricultural production (Adams et al., 1998; Schimmelpfennig et al., 1996). Other impact assessment methods that have been used are the integrated impact assessment method and the agroecological zone method (Mendelsohn, 2000; Kurukulasuriya and Mendelsohn, 2008). The main problem with using structural approaches (agronomic-economic models) is that in aggregate studies, inferences made to large areas and diverse agricultural production systems are based on results from very few laboratory and experimental sites (Adams et al., 1998; Schimmelpfennig et al., 1996). The spatial analogue approach, on the other hand, uses cross-section evidence to make statistical (econometric) estimations of how changes in climate would affect agricultural production across different climatic zones. The main advantage of this approach is that it gives evidence of changes in farmer management practices and decisions in response to changing climatic conditions (Mendelsohn and Dinar, 2003; Mendelsohn et al., 1994). An example of the spatial analogue approach is the Ricardian cross-sectional approach that measures the performance of farmers, households and firms across spatial scales with different climates. The technique draws heavily on the underlying observation by Ricardo that under competition, land values reflect the productivity of the land (Mendelsohn and Dinar, 1999, 2003; Mendelsohn, 2000; Mendelsohn et al., 1996, 1994). This study adopts the cross-sectional Ricardian approach to measure the economic impacts of climate on net farm revenue in Africa. It uses cross-section data and econometric analyses to estimate the impacts of climate variables (temperature and precipitation), soils, hydrological and socio-economic factors on net farm revenue. The study considered impacts on three production systems: (a) specialized crop, (b) specialized livestock and (c) mixed crop-livestock farming. Due to lack of African data on land rents, the study uses total net farm revenue, defined as the sum of net revenues from three main farming activities: (a) dryland crops, (b) irrigated crops and (c) livestock, as the measure


of farm performance. Farm net revenue (R) is assumed to reflect the present value of future net productivity and costs of individual crops and livestock: Z hX Z i X t Px X et dt ð1Þ R ¼ PLE e dt ¼ Pi Qi ðX, F, Z, H, GÞ where PLE is the net revenue per farm, Pi is the market price of crop i, Qi is output of crop i, F is a vector of climate variables, Z is a set of soil variables, H is a set of hydrological variables, G is a set of economic variables, Px is a vector of purchased input prices, t is time, and is the discount rate. The Ricardian method assumes that each farmer will seek to maximize net farm revenues by choosing inputs (X) subject to climate, soil and economic factors. The resulting net revenue function observes the loci of maximum profits subject to a set of climate, soil and economic factors and the Ricardian model is a reduced-form hedonic price model of the observed loci of profits. The standard Ricardian model relies on a quadratic formulation of climatic variables: R ¼ 0 þ 1 F þ 2 F 2 þ 3 Z þ 4 G þ 5 logðHÞ þ u

ð2Þ

where u is the error term. To capture the nonlinear relationship between net farm revenues and climate variables, the estimation includes both the linear and quadratic terms for climate variables, F (temperature and precipitation).1 3. The Data and Empirical Specifications of Model Variables This study is based on cross-section data obtained from the Global Environment Facility/World Bank (GEF/WB)–CEEPA-funded project: Climate, Water and Agriculture: Impacts on and Adaptations of Agro-Ecological Systems in Africa. The study involved 11 African countries: Burkina Faso, Cameroon, Egypt, Ethiopia, Ghana, Kenya, Niger, Senegal, South Africa, Zambia and Zimbabwe. For more information on the survey method and the data collected, see Dinar et al. (2008) and Kurukulasuriya et al. (2006). Over 9000 household surveys were conducted in the study, of which about 8000 surveys were found usable after data cleaning. It is important also to note that none of the interviewed farmers kept livestock only. However, we attempted to separate those specialized in livestock production from those practicing mixed croplivestock farming as discussed below (see categorization of farm types in the entire sample (Table 1) and the accompanying discussion). As mentioned earlier, this study used net farm revenue to measure farm performance due to lack of African data on land rents. Total net farm revenue is defined as the sum of net revenues from crop and livestock production activities.2 The Ricardian

1See

Dinar et al. (2008) for more details on the cross-section (Ricardian) method. considered impacts of climate change on two main datasets, one including negative net revenues up to US$200 and another one with only positive net revenues. The results of the two samples were not all that different and the analyses in this study are based on the sample with positive net revenues. 2We


Table 1. Characterization of farm types. Specialized crops

Specialized livestock

Mixed croplivestock

All farms

Average farm size (ha)

Net revenue ($)

Total sample (% of row total)

21%

1%

78%

100% (6966)

26.44

1894.25

Irrigated (% of column total)

20%

0

24%

23% (1612)

33.25

3175.97

Dryland (% of column total)

80%

100%

76%

77% (5354)

24.40

1507.39

Average farm size (ha)

28.55

384.28

21.51

—

26.44

—

Net revenue ($)

1832.83

7107.60

1839.20

1894.25

—

1894.25

Note: Results are based on the positive net revenue sample.

approach is traditionally based on analysing net revenue or land value per hectare. As most farmers in Africa graze livestock on open-access, communal land, it was very difficult to measure the amount of land farmers allocate to livestock production. Since this study combined net revenues from both crop and livestock production, we could not use net revenue per hectare and instead used net revenue per farm, making the unit of analysis in this study the farm. The explanatory variables consist of seasonal climate variables, soils, water flow and socio-economic factors (for details, see Nhemachena, 2009). The study relied on longterm average climate (normals) for districts in Africa (see Dinar et al., 2008; Kurukulasuriya et al., 2006 for details). Soil data came from the Food and Agriculture Organization (FAO, 2003). Data on hydrological variables (e.g. flow and runoff for each district) were obtained from Strzepek and McCluskey (2007). The explanatory variables included in this study have been shown to affect net farm revenue in many other African Ricardian models (Dinar et al., 2008; Kurukulasuriya and Mendelsohn, 2008; Mano and Nhemachena, 2007). Appendix 1 shows the distribution of usable surveys, net revenues and climate variables by country. Table 1 presents the categorization of farm types in the entire sample. We considered farms with only crops and livestock. None of the interviewed farmers kept livestock only. However, we attempted to separate those specializing in livestock production from those practicing mixed crop-livestock farming. For specialized livestock farms, we considered farms with a very small share of the total land area for crops and a relatively large number of cattle, goats or sheep. We also considered the share of income from livestock production — a very high share meant that the farm specialized in livestock production. By doing this, we found that only 1% of the farms could be classified as


specialized livestock farms. We also found that all of these specialized livestock farms were under dryland farming and none had irrigation. Specialized crop farms were defined as those with crops only and no livestock or those with small livestock numbers such as two sheep or a few chickens. Mixed croplivestock farms were defined as farms where neither of the two clearly dominate. The tables also present the distribution of dryland and irrigated farms in each country and farm type. The analyses in this study distinguish between the impacts of climate change on these two main farm types. This helps us to assess the importance of irrigation in responding to changes in climate. The economic impacts of climate change were estimated on each of the classified farming systems (mixed crop-livestock and specialized crop or livestock) as well as all the farms. The analyses presented in this study start with an analysis of impacts on all the farms (the entire sample) and then an examinination of each farming system. We estimated multiple regression models of net revenue across three samples for each farm type (dryland, irrigation and total sample for each farm type). 4. Results of the Economic Impact Analyses Table 2 presents results from the Ricardian regressions for the whole sample, mixed crop-livestock, specialized crop and specialized livestock samples. We also estimated impacts on dryland and irrigated farms for each farming system (results not shown here). The results show the effects of climate, soil, water flow and socio-economic variables on net revenue per farm for each farm type. The results indicate that the explanatory variables have differential impacts on dryland, irrigated farms and the total sample across farm types. The effects of a number of soil and household characteristic (e.g. age, gender and education of head of family) variables were tested and found to be insignificant and so were dropped from the analyses. The models account for about 45% to 73% of the variability in net revenues from farm to farm. We note that a relatively high proportion of the variation in net revenue is not accounted for by the explanatory variables in the models. The important sources of error accounting for this unmeasured variation include omitted variables and misreporting of net revenue. This same dataset was used to conduct parallel regional studies of climate change impacts on crops (Kurukulasuriya and Mendelsohn, 2008) and livestock (Seo and Mendelsohn, 2008) separately. As mentioned earlier, our study combined analyses of both crop and livestock systems. We therefore compare the results of our combined analyzes with results from these specialized studies. The results show that most of the explanatory variables are statistically significant at 10% or lower and the signs on most variables are as expected except for a few, which are discussed below. Larger farm sizes appear to have strong positive influence on net farm revenue across all farm types, suggesting that have more land allows farmers to diversify crop and livestock enterprises per farm, leading to more income although the


Table 2. Ricardian regression results. Variable name

Winter temperature Spring temperature Summer temperature Fall temperature Winter precipitation Spring precipitation Summer precipitation Fall precipitation Winter precipitation squared Spring precipitation squared Summer precipitation squared Fall precipitation squared Winter temperature squared Spring temperature squared Summer temperature squared Fall temperature squared Orthic Ferralsols (foFU) Fluvisols (jcMFU) Ferric Luvisols (lfU) Ferric Luvisols (lfCU) Cambic Arenosols (qc) Luvic Arenosols (qlCU) Chromic luvisols (lCU) Farmland (ha) Mean water flow Household has tractor (Yes/No) Household access to extension (Yes/No) Household access to electricity (Yes/No) Household size (No. of people) Using irrigation (Yes/No) Mixed crop-livestock (Yes/No) Specialized crop (Yes/No) North & East Africa (Yes/No) Southern Africa (Yes/No) Constant R-squared N

All farms

Mixed croplivestock farms

Specialized crop farms

Specialized livestock farms

1.641*** 1.255*** 0.824*** 1.794*** 0.036*** 0.011** 0.015*** 0.003 0.000*** 0.000* 0.000*** 0.000*** 0.019*** 0.015*** 0.005 0.018*** 0.278 0.443** 0.372** 0.488*** 0.111 0.730*** 0.469*** 0.643*** 0.010*** 0.331*** 0.169*** 0.333*** 0.183*** 0.053 0.447*** 0.455** 0.029 2.011*** 6.667

1.692*** 1.257*** 0.426 0.797 0.033*** 0.012* 0.024*** 0.012*** 0.000*** 0.000 0.000*** 0.000*** 0.019*** 0.014*** 0.002 0.002 0.378 0.446** 0.533*** 0.315** 0.053 0.647*** 0.495*** 0.642*** 0.009*** 0.271*** 0.168*** 0.378*** 0.154*** 0.091 — — 0.007 1.846*** 4.923

2.056*** 1.277* 1.937*** 4.143*** 0.036*** 0.005 0.003 0.017*** 0.000* 0.000 0.000 0.000 0.027*** 0.017* 0.027*** 0.057*** 0.030 0.582 0.076 1.096*** 0.617 1.352*** 2.033** 0.693*** 0.011*** 0.395* 0.177* 0.150 0.283*** 0.092 — — 0.180 2.025*** 23.161**

— — 7.116 — 5.787* 5.721* 1.804* 3.254* 0.054* 0.079* 0.014* 0.011* 0.838** 0.549** 0.614* 0.726** — — — 1.603 0.311 0.556 0.000 0.154 0.111* 1.089 0.158 0.267 0.626 3.280* — — 6.409

0.5102 5607

0.4537 4317

0.6490 1226

Note: ***; **; * significant at 1%, 5% and 10% levels respectively.

272.491 0.7343 64


value per hectare may be low. Similar studies found contrasting results of the impact of farm size on net revenue (per hectare). For example, Kurukulasuriya and Mendelsohn (2008) found that a greater farm area reduces the value per hectare of farms but at a decreasing rate. In contrast, Seo and Mendelsohn (2008) found that the dummy for large farms was insignificant, implying no difference in the net revenue per animal for small and big farms. Similarly, larger families seem to be associated with higher net farm revenue across all farm types. This suggests that agriculture in Africa is more labor-demanding. Better access to other farm assets, such as heavy machinery like tractors, appear to strongly and positively influence net farm revenues for all farms, mixed crop-livestock farms and specialized crop farms. These results suggest that capital, land and labor serve as important production factors in African agriculture. Attaining higher net farm revenues strongly depends on factor endowments (i.e. family size, land area and capital resources) at the disposal of farming households. Kurukulasuriya and Mendelsohn (2008); Seo and Mendelsohn (2008) and Kurukulasuriya et al. (2006) found similar positive effects from access to technology variables (electricity and heavy machinery) on net revenue. Kurukulasuriya and Mendelsohn (2008) and Kurukulasuriya et al. (2006) also found similar positive effects from household size. In contrast, Seo and Mendelsohn (2008) found that large households tend to have lower livestock net revenue per farm. Better access to extension services seems to have a strong positive influence on net farm revenue a across all farms, mixed crop-livestock farms and specialized crop farms. The effect on net revenue from specialized livestock farms, though positive, is insignificant. Both mixed crop-livestock and specialized crop variables positively affect net farm revenues. Among the regional dummies, southern Africa appears to have a strong negative influence on net farm revenue. Water flow had a significant positive effect on the total sample and mixed croplivestock system. Kurukulasuriya and Mendelsohn (2008) also found that water flow strongly influenced net farm revenue, especially for irrigated farms. We expect all farms and mixed farms to benefit from year-round water flow, for example, for watering livestock during dry seasons and also for irrigation. Using irrigation appears to positively influence net farm revenue for all farm types except specialized livestock farms. The possible explanation is that during the dry season, water flow helps to provide water for livestock watering and irrigation. The soil variables show that arenosols (qlCU), fluvisols (jcMFU) and ferric luvisols (lfCU), which are extensively developed and are usually highly productive soils, have a strong positive influence on net farm revenues across all farming systems. Seasonal climate variables show that climate effects vary across models and farm types. The coefficients of the linear and quadratic terms of climate variables are significant in some seasons, indicating a nonlinear relationship between these variables and net revenue. A positive (negative) sign of the quadratic term shows that the relationship between climate variables and net revenue is hill-shaped (U-shaped), but


the effect of quadratic seasonal climate variables on net revenue cannot be easily concluded as both linear and quadratic terms influence net revenue. To interpret the climate coefficients, we calculated marginal climate impacts at the mean of temperature and precipitation for all farm types and also results from dryland and irrigation farms (Table 3). In each case, the marginal effect of temperature and precipitation is evaluated at the mean for each sample. For example, the marginal effect of temperature on mixed crop-livestock farms is evaluated at the mean temperature of such farms and the marginal impact of precipitation on specialized crop farms is evaluated at the mean precipitation for such farms. The results suggest that better watered regions (i.e. with wetter seasons) are strongly related to higher net farm revenues for all farms, mixed crop-livestock and specialized crop. For example, a wetter summer increases net revenue per farm by $99 and $93 per mm of monthly precipitation for mixed crop-livestock and specialized crop farms respectively. The effects are strongest for mixed crop-livestock farms, suggesting that higher rainfalls allows farmers to diversify crop and livestock enterprises throughout the year. Kurukulasuriya and Mendelsohn (2008) also found similar results for the marginal impact of summer precipitation on crop revenue. Their study found that the marginal precipitation effects on dryland and irrigated farms are more similar ($3.8/ mm/mo for irrigated farms and $2.7/mm/mo for dryland) because irrigated farms are located in such dry locations. Warmer winters and springs appear to positively influence net farm revenues for all farms and mixed crop-livestock farms, especially irrigated ones. Warming in summer tends to be associated with a strong negative influence on net farm revenues across all farming systems. The magnitudes of the marginal effects show that the negative effects are strongest for specialized farm types, compared to mixed crop-livestock farms. This suggests that the mixed farming system offers important adaptation strategies for farmers. Dryland farms appear to be the most vulnerable to warming, compared to all farms and irrigated farms. Similar results were noted by Seo and Mendelsohn (2008), who found that the income of small farms is stable over a range of temperatures while that of large farms declines sharply as temperatures rise. Larger farms tend to be more specialized compared to small farms, which have diverse farm enterprises. In addition to marginal effects, we computed climate elasticities (the percentage change in net revenue as a result of a percentage change in climate variables). The elasticities are given in parentheses in Table 3. The temperature elasticity of dryland farms as well as specialized crop or livestock farms are relatively high compared to irrigated farms and mixed crop-livestock farms. Because irrigation and mixed croplivestock farms are buffered from temperature changes through irrigation and diversity of options respectively, we expect them to be less sensitive to warming. Kurukulasuriya and Mendelsohn (2008) and Seo and Mendelsohn (2008) also found that warmer temperatures increase the net revenues of irrigated farms because the mean temperature of irrigated farms is relatively cool and because irrigation buffers net revenues from temperature effects.

14.36 (0.10)

103.53*** (0.13)

15.49* (0.04)

122.31** (4.40)

61.85*** (3.34)

137.66*** (2.19)

Summer

Fall

17.65** (0.11)

10.49 (0.05)

Spring

Dryland farms 85.34*** Winter (2.04)

176.13*** (3.16)

29.53*** (0.07)

156.78*** (2.56)

Summer

Fall

113.23*** (0.73)

9.36** (0.04)

126.08*** (1.29)

Spring

124.32*** (1.57)

97.55*** (0.18)

125.43*** (1.66)

139.18*** (2.30)

121.55*** (1.14)

104.62*** (0.27)

132.11*** (1.52)

12.86*** (0.10)

154.20*** (2.24)

Winter

Temperature

9.88** (0.02)

98.81*** (0.14)

21.93** (0.09)

19.60*** (0.10)

70.18* (0.03)

99.25** (0.10)

84.12** (0.04)

54.73*** (0.09)

Precipitation

Mixed crop livestock farms

Precipitation

All farms

Temperature

Season

162.97* (3.30)

188.99** (4.37)

135.62 (2.08)

130.40* (4.70)

192.08 (3.61)

172.47** (3.08)

15.98* (0.13)

29.21*** (0.20)

17.79** (0.16)

36.85** (0.21)

58.93 (0.19)

92.94* (0.01)

39.79 (0.02)

32.67** (0.14)

155.99** (2.27) 128.87 (1.40)

Precipitation

Temperature

Specialized crop farms

262.44*** (1.42)

195.11** (0.36)

193.49** (3.01)

52.47* (0.07)

78.19* (0.73)

39.18*** (0.15)

56.08*** (0.03)

—

—

259.76** (4.38)

—

—

—

Precipitation

—

—

—

Temperature


Table 3. Marginal impacts and elasticities of climate variables on net revenue ($/farm).


102.60*** (0.20)

69.73** (0.03)

40.55*** (2.58)

347.28*** (1.52)

Summer

Fall

340.18*** (1.41)

226.37*** (1.05)

116.84** (0.89)

168.31*** (1.80)

Temperature

89.88** (0.01)

112.19*** (0.18)

69.08 (0.03)

93.70** (0.07)

Precipitation

Mixed crop livestock farms

210.37* (0.51)

55.20** (1.69)

233.16 (2.80)

41.12** (2.68)

Temperature

68.61 (0.22)

76.80* (0.08)

49.38 (0.02)

91.12** (0.09)

Precipitation

Specialized crop farms Temperature

Precipitation


Note: Values calculated at the mean of the sample using OLS coefficients from Table 3 and from dryland and irrigated farms regressions. Numbers in parenthesis are elasticities. ***; **; * significant at 1%, 5% and 10% levels respectively.

57.29** (0.09)

128.61*** (1.44)

74.03*** (0.07)

Irrigated farms 59.62*** Winter (1.77)

Spring

Precipitation

All farms

Temperature

Season

Table 3. (Continued )



A marginal increase in precipitation increases net revenue for all farm types. The precipitation elasticity is relatively high for dryland farms across farm types and for specialized crop and livestock farms. Because mixed crop-livestock farms are more diverse in their enterprises and options with the ability to shift easily between crops and livestock, we expect them to be less sensitive to warming and drying. From an adaptation perspective, mixed crop-livestock farming becomes a good alternative to specialized crop or livestock farms. An interesting observation from the results is that net revenue decreases with falling precipitation (in spring, summer and fall) for specialized livestock farms. This is in contrast to findings from the regional Ricardian livestock analysis (Seo and Mendelsohn, 2008) where net revenue increased with falling precipitation as farmers shifted from livestock to crops, forests to grasslands, and diseases became less prevalent. We, however, note that while wet conditions are expected to improve the quantity and quality of grazing pastures, they may also be associated with high levels of disease that may reduce the gains from improved pastures. The sensitivity of dryland farms and specialized crop or livestock farms is relatively high compared to irrigated farms. To provide a more complete analysis of the impacts of climate, we estimated climate response functions based on regression results in Table 2 and results from dryland and irrigated farms regressions. We plotted the net revenues of an average farm at different temperature and rainfall levels. The climate response functions for the entire sample (combining specialized crop, livestock and mixed farms) and each of the farming systems are presented separately in Figs. 1–8. The response functions show a hill-shaped response of net revenue to temperature and rainfall. The results show that net revenues for all farms rise with increasing temperature up to 24 C but further increases in temperature are associated with declines in net revenues (Fig. 1). The annual mean average temperature in Africa is currently about 24 C, indicating that further warming will be harmful to African agriculture. 8

Log net revenue US$/farm

6 4 2 0 0

3

6

9

12

15

18

21

24

27

30

33

-2 -4 -6 Te mpe rature (˚C e lcius)

Figure 1. Temperature response function — all farms

36

40

Measuring the Economic Impact of Climate Change on African Agricultural Production Systems 45 14

Ln net revenue US$/farm

12 10 8 6 4 2 0 0

50

100

150

200

250

300

350

400

450

500

550

650

700

750

800

Precipitation (mm)

Figure 2. Precipitation response function — all farms

The response to precipitation shows that net revenues rise with increasing rainfall up to 450 mm and then decline with higher levels of rainfall (Fig. 2). This implies that above a 450 mm seasonal average, wetter conditions become harmful to agricultural production. The response functions for temperature and rainfall show that reductions in net revenues with further warming are higher than with wetter conditions. These results confirm the findings from earlier Ricardian analysis on African cropland by Kurukulasuriya and Mendelsohn (2008). We also examined the separate response functions for mixed crop-livestock farms, specialized crop and livestock farms as well as dryland and irrigation farms in each system. Figure 3 shows that for mixed farms, net revenues rise with increasing temperature up to 25 C, after which they decline with further warming. For specialized


7 6 5 4 3 2 1 0 -1

0

3

6

9

12

15

18

21

24

27

30

33

36

40

-2 Temperature (˚Celcius)

Figure 3. Temperature response function — mixed crop-livestock farms

46 C. Nhemachena, R. Hassan & P. Kurukulasuriya 10 9

Ln net revenue US$/farm

8 7 6 5 4 3 2 1 0 0

50

100

150

200

250

300

350

400

450

500

550

650

700

750

800

Precipitation (mm)

Figure 4. Precipitation response function — mixed crop-livestock farms 5


0 0

3

6

9

12

15

18

21

24

27

27

33

36

40

-5

-10

-15

-20

-25

Temperature (˚Celcius)

Figure 5. Temperature response function — specialized crop farms

crop farms (Fig. 5) and specialized livestock farms (Fig. 7), net revenues also increase with rising temperature and decline with further warming above 23 C and 27 C. The results from the temperature response functions show that the net revenue curve for mixed farms covers a larger area than the specialized crop and livestock response curves. This implies that mixed farms are less affected by temperature changes than the specialized systems. In addition, results show that net revenues for mixed farms (Fig. 4) rise with increasing rainfall up to about 450 mm and decline with increasingly wet conditions. Precipitation response curves for specialized crop (Fig. 6) and livestock (Fig. 8) farms show that net revenues rise with increasing rainfall up to 350 mm and 300 mm, respectively. Precipitation above these levels has negative impacts on net farm revenues.

Measuring the Economic Impact of Climate Change on African Agricultural Production Systems 47 10


8 6 4 2 0 0

-2

50

100 150 200 250 300 350 400 450 500 550 650 700 750 800

-4 -6 -8 Precipitation (mm)

Figure 6. Precipitation response function — specialized crop farms

40


20 0 -20

0

3

6

9

12

15

18

21

24

27

30

33

36

40

-40 -60 -80 -100 -120 -140 Temperature (˚Celcius)

Figure 7. Temperature response function — specialized livestock farms

The shapes of the response functions are worth noting. Results show that specialized crop (Figs. 5 and 6) and livestock systems (Figs. 7 and 8) in Africa are highly sensitive to climate. The climate sensitivity, however, varies with dependence with rainfall and the use of irrigation. Irrigation acts as a buffer against adverse impacts from harsh climatic conditions and hence irrigated farms are less sensitive to climate. Mixed crop-livestock farms (Figs. 3 and 4) and irrigated farms appear to be more resilient to harsh climate conditions. The results suggest that specialized crop or livestock agriculture is more vulnerable to climate change than mixed systems. Generally, response curves for temperature and precipitation show that net revenues are more sensitive to temperature changes. This implies that temperature changes are more harmful to agricultural production in the region.

48 C. Nhemachena, R. Hassan & P. Kurukulasuriya 20


0 0

50

100

150

200

250

300

350

400

450

500

550

650

700

750

800

-20

-40

-60

-80

-100

Precipitation (mm)

Figure 8. Precipitation response function — specialized livestock farms

4.1. Forecasting impacts of climate change on net revenue This section predicts the impacts of future climate changes on net revenues from crop and livestock farming under various scenarios. We used estimated model parameters from the Ricardian analysis above to predict the potential impacts of future climate changes on net farm revenues across different farming systems. We examined the impacts of future changes in climate for both a set of climate sensitivity scenarios and a set of climate change scenarios predicted by the Atmospheric-Oceanic Global Circulation Models (AOGCMs). The first part of the forecasting involved using simple scenarios that helped us to see the behavior of the models, followed by forecasting based on AOGCM scenarios. 4.1.1. Climate sensitivity scenarios In the climate sensitivity scenarios, the study tested four climate change scenarios: þ2:5 C and þ5 C increases in temperature and 7% and 14% decreases in precipitation. The simulations allowed only one climate variable to change at a time. These assumptions are not realistic in the real world, but they provide important insights into likely responses to changes in climate variables. Table 4 presents the results of the four climate scenarios compared to the baseline income for each farm type and category. The results show that increases in warming of 2:5 C and 5 C seem to predict losses in net farm revenue per farm for all farms, mixed crop-livestock farms and specialized crop and livestock systems. The losses are strongest for specialized crop systems, for example, with a 5 C warming specialized crop farms lose 87% of net farm revenue per farm compared to a 57% loss for mixed crop-livestock systems and a 49% loss for specialized livestock farms.


Table 4. Predicted impacts of climate change on net revenue from simple scenarios. Climate scenario

All farms Net revenue (USD per farm)

% of net farm income

Baseline: 506.42

Mixed croplivestock farms Net revenue (USD per farm)


Baseline: 563.39

Specialized crop farms Net revenue (USD per farm)


Baseline: 333.18

Specialized livestock farms Net revenue (USD per farm)


Baseline: 569.95

2:5 C increae in warming

214.49

42.35

165.65

29.40

76.08

22.84

120.82

21.20

5 C increase in warming

51.36

10.14

318.31

56.50

291.30

87.43

276.46

48.51

7% decrease in precipitation

64.83

12.80

52.02

9.23

75.55

22.68

186.28

32.68


130.86

25.84

105.56

18.74

152.48

45.76

370.49

65.00

Baseline: 443.58 Dryland farms 2:5 C increase in warming

Baseline 502.21

Baseline: 283.86

87.34

19.69

337.22

67.15

238.71

84.09

—

—

110.01

24.80

54.86

10.92

263.28

92.75

—

—


60.77

13.70

175.00

34.85

226.76

79.88

—

—


78.48

17.69

221.92

44.19

234.36

82.56

—

—

5 C increase in warming

Baseline: 777.83 Irrigated farms 2:5 C increae in warming

Baseline: 790.36

Baseline: 669.42

201.58

25.92

154.89

19.60

192.30

28.73

—

—

232.59

29.90

172.93

21.88

228.69

34.16

—

—


158.85

20.42

120.66

15.27

131.56

19.65

—

—

14% decreased in precipitation

167.42

21.52

130.63

16.53

146.27

21.85

—

—

5 C increase precipitation

Note: Estimated using coefficients from regression results (Table 4-3 and the other models not presented here).


Reductions in precipitation (7% and 14%) predict higher losses in net farm revenue per farm for specialized crop and livestock systems compared to all farms and mixed crop-livestock farms. For example, a 14% reduction in precipitation predicts 65% and 46% losses in net revenue per farm for specialized crop and livestock systems, respectively, compared to 26% for all farms and 19% for mixed crop-livestock farms. These results suggest that specialized crop or livestock systems tend to suffer most from increases in warming and drying. Mixed crop-livestock farms that are less sensitive to climate changes suffer minimal damages compared to other farm types. Results also show that increases of 2:5 C and 5 C in temperature tend to predict losses for dryland systems and gains for irrigated systems. The magnitudes of the losses are highest for specialized crop systems compared to all farms and mixed croplivestock systems, suggesting that the risk of specialized systems is higher with warming in general. Reductions of 7% and 14% in precipitations appear to lead to losses both for dryland and irrigated farming systems. Similarly, the magnitudes of the predicted losses suggest that drying has strong negative effects on specialized crop systems compared to all farms and mixed crop-livestock farms. 4.1.2. AOGCM climate scenarios The study also examined a set of climate change scenarios from AOGCMs. We used two scenarios that predict a wide range of outcomes consistent with the most recent IPCC report (Houghton et al., 2001). The specific scenarios used in this study are A1 scenarios from the following models: Parallel Climate Model (PCM) (Washington et al., 2000) and Canadian Climate Centre (CCC) (Boer et al., 2000). We examined countrylevel impacts for each of these scenarios for the year 2100. We added the climate modelpredicted change in temperature to the baseline temperature in each district under each climate scenario. For changes in precipitation we multiplied the climate model-predicted change by the baseline precipitation in each district. When examining actual climate scenarios, there will also be a change in water supply; however, we did not take this into account in the study and this is a caveat when promoting irrigation. The models have a range of predictions: the PCM predicts a 3 C increase in temperature for 2100 and the CCC an increase of 6 C. Both models show a rising trend in temperature over time. The PCM predicts an increase in precipitation of 4% by 2100 and the CCC a reduction of 15% for the same year. However, rainfall distribution greatly varies across countries. An important point to note is the spatial and temporal variability in predictions of temperature and precipitation in Africa. To predict the impact of each climate scenario on net revenue, we calculated the change in net farm revenues from baseline values in Table 3 and under each new climate scenario. The difference between the two levels of net revenues gave us the change in net revenue per farm in each district. The predictions were based on Ricardian regression results from Table 2 and results from dryland and irrigated farms regressions.


Table 5. Predicted impacts from AOGCM climate scenarios (PCM and CCC for the year 2100). Climate scenario

All farms Net revenue (USD per farm)

PCM 2100 CCC 2100

Dryland farms PCM 2100 CCC 2100


Net revenue (USD per farm)


Specialized crop farms Net revenue (USD per farm)


Baseline: 506.42 14.92 2.95 298.17 58.88

Baseline: 563.39 15.90 2.82 107.55 19.09

Baseline: 333.18 120.08 36.04 189.61 56.91

Baseline: 443.58

Baseline: 502.21

Baseline: 283.86

62.81 76.14

14.16 17.17

Baseline: 777.83 Irrigated farms PCM 2100 CCC 2100

Mixed croplivestock farms

255.91 219.54

32.90 28.22

66.91 245.02

13.32 48.79

Baseline: 790.36 172.33 110.49

21.80 13.98

181.39 224.21

63.90 78.99

Specialized livestock farms Net revenue (USD per farm)


Baseline: 569.95 405.06 25.70 357.35 22.68

— —

— —

— —

— —

Baseline: 669.42 209.70 232.85

31.32 33.29

Note: Estimated using coefficients from regression results (Table 4–3 and Appendix 1) and AOGCM country specific climate scenarios.

Table 5 presents the predicted changes in net revenue per farm using the two climate scenarios for the year 2100. The PCM scenarios that forecast mild climate changes predict some increases in net revenue. The CCC scenarious that forecast substantial increases in warming and drying predict severe losses in net farm revenues across Africa. Dryland farms and specialized crop or livestock farms tend to suffer most from harsh climatic conditions. On the other hand, irrigated farms and mixed crop-livestock farms are less sensitive to changes in climate and experience less negative impacts from increases in warming and drying. These results support the observation that irrigation and mixed crop-livestock farms offer an important adaptation alternative for farmers. 5. Conclusions and Policy Implications This study analysed the impacts of climate changes on net farm revenues in Africa. The empirical analyses were based on a cross-sectional database of over 6900 surveys from 11 African countries. Additional climate, soil and water flow variables were obtained from other sources and combined with the cross-sectional survey data. We used a Ricardian approach to measure the impacts of climate change on combined crop and livestock net revenue. Net revenue per farm was regressed against climate, soils, hydrological and socio-economic variables to measure the effects of each variable on


net farm revenue. The impacts of climate change were examined for three main farming types: specialized crop; specialized livestock and mixed crop and livestock as well as the total sample and for dryland and irrigated farms within each farm type. The study also examined some climate sensitivity scenarios as well as two climate scenarios from Atmospheric-Oceanic General Circulation Models (AOGCMs). The results showed that larger farm sizes appear to have a strong positive influence on net farm revenues across all farm types, suggesting that having more land allows farmers to produce more crop and livestock enterprises per farm, thus leading to more income although the value per hectare may be low. Larger families seem to be associated with higher net farm revenues across all farm types. Better access to other farm assets, such as heavy machinery like tractors, appear to strongly and positively influence net farm revenues for all farms, mixed crop-livestock farms and specialized crop farms. These results suggest that capital, land and labor serve as important production factors in African agriculture. National policies need to invest more in improving factor endowments (i.e. land area and capital resources) at the disposal of farming households to enhance farm performances in the face of climate change. Better access to extension services seems to have a strong positive influence on net farm revenues across all farms, mixed crop-livestock farms and specialized crop farms. Improving access to extension services ensures that farmers have the information to make decisions that improve their production activities. Policies aimed at improving farm-level performance need to emphasize the critical role of providing information (through extensions services) to help decision making at the farm level. Improving access to technology (in this case, electricity) has significant potential in improving farm-level production activities and hence net revenues. For example, the use of irrigation and intensive livestock production systems (which are usually capitalintensive) increases when farmers have access to electricity and other machinery. Improving access to technology such as electricity and machines is therefore important to enhancing agricultural production in the face of climate change. The soil variables arenosols, fluvisols and ferric luvisols, which are extensively developed and are usually highly productive soils, appear to have a strong positive influence on net farm revenues across all farming systems. Marginal analyses of the impacts of seasonal climate variables show that African net farm revenues are highly sensitive to changes in climate. The sensitivity is relatively high for changes in temperature. Further warming and drying have severe adverse effects on net farm revenues. The results show variations in sensitivity to climate based on the type of farm and whether it is dryland or irrigated. Dryland and specialized crop or livestock farms suffer more from further increases in warming and drying compared to irrigated and mixed crop-livestock farms. Predictions of future climate impacts also indicate that mixed crop-livestock and irrigated farms are less sensitive to climate changes and will experience less damages compared to highly sensitive dryland and specialized crop or livestock farms. Results show that net farm revenues are in general negatively affected by warmer and dryer climates. The small-scale mixed crop and


livestock system typical in Africa is the most tolerant whereas specialized crop production is the most vulnerable to warming and decreased rainfall. Generally, farming systems located in dry, semi-arid and arid regions (for example, most of the southern parts of the continent) will suffer more from increases in warming and drying compared to those in more humid regions. It is therefore important for Africa to enhance adaptation efforts both at the micro (farm) and macro (national) levels. Governments need to integrate adaptation into national economic policies as well as strengthening community-based adaptations to help farmers minimize the potential damages from climate change. These results have important policy implications, especially for the suitability of the increasing tendency toward large-scale, mono-cropping strategies for agricultural development in Africa and other parts of the developing world, in light of the expected climate changes. Mixed crop and livestock farming and irrigation offer better adaptation options for farmers against further warming and drying as predicted under various future climate scenarios. Acknowledgments This paper was funded by GEF and the World Bank. It is part of a larger study on the effect of climate change on African agriculture coordinated by the Centre for Environmental Economics and Policy in Africa (CEEPA), University of Pretoria, South Africa. The authors are very grateful for the comments from anonymous reviewers. All remaining errors should be attributed solely to the authors. Appendix 1 Summary Statistics of the Survey Sample Country

Usable surveys

Temperature and precipitation normals (sample means)

Dryland Irrigated Total

Winter

Spring

Summer

Fall

Temp Precip Temp Precip Temp Precip Temp Precip Burkina Faso Cameroon Egypt Ethiopia Ghana Kenya Niger Senegal South Africa Zambia Zimbabwe

765 583 0 170 713 547 560 812 73 813 318

94 91 495 491 41 78 125 70 48 20 59

859 674 495 661 754 625 685 882 121 833 377

26.1 24.2 16.6 20.9 25.5 22.1 24.5 26.4 13.9 22.1 16.5

2.4 57.4 12.5 19.4 31.3 86.8 0.7 2.2 35.2 48.1 7.3

30 25.9 19.1 22.1 27.5 22.8 29 29.1 17.8 23.5 20.6

14.9 97.4 7.2 48.4 60.4 104.8 3.1 1.1 62.9 58 15.4

29.9 24.2 27.8 22.7 25.8 20 31.8 30.8 22.2 24.3 23.5

110.8 180.5 3.7 127.5 112.4 89.5 64.8 49.6 96.7 108.3 137.9

28.3 24.3 26.7 19.4 25.1 21 29.6 29.3 20.9 24.9 22

129.1 221.9 4.8 120.3 111.2 65.4 71.5 112.4 76.2 100.3 88.9

Total

5354

1612

6966

22.7

25.4

25.2

39.5

26.4

95.9

25.6

103.6


References Adams, RM, BH Hurd, S Lenhart and N Leary (1998). Effects of global change on agriculture: An interpretative review. Climate Research, 11, 19–30. Boer, G, G Flato and D Ramsden (2000). A transient climate change simulation with greenhouse gas and aerosol forcing: Projected climate for the 21st century. Climate Dynamics, 16, 427–50. Dixon, J, A Gulliver and D Gibbon (2001). Farming Systems and Poverty: Improving Farmers’ Livelihoods in a Changing World. Rome and Washington, D.C.: FAO and World Bank. Dinar, A, R Mendelsohn, R Hassan and J Benhin (2008). Climate Change and Agriculture in Africa: Impacts Assessment and Adaptation Strategies. Earthscan: London. FAO (Food and Agriculture Organization) (2003). The Digital Soil Map of the World: Version 3.6 (January), Rome, Italy. Houghton JT et al. (eds), (2001). Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the third assessment report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press. IPCC (2001). Climate Change 2001: Impacts, Adaptation, and Vulnerability. Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, UK. Kurukulasuriya, P et al. (2006). Will African agriculture survive climate change? The World Bank Economic Review, 20(3), 367–388. Kurukulasuriya, P and R Mendelsohn (2008). A Ricardian analysis of the impact of climate change on African cropland. African Journal of Agricultural and Resource Economics, 2(1), 1–23. Mano, R and C Nhemachena (2007). Assessment of the economic impacts of climate change on agriculture in Zimbabwe: A Ricardian approach. Policy Research Working Paper 4292, Development Research Group, Sustainable Rural and Urban Development Team. The World Bank, Washington, DC. Mendelsohn, R and A Dinar (2003). Climate, water, and agriculture. Land Economics, 79(3), 328–341. Mendelsohn, R (2000). Measuring the effect of climate change on developing country agriculture. Two essays on climate change and agriculture: A developing country perspective. FAO Economic and Social Development Paper 145. Mendelsohn, R and A Dinar, (1999). Climate change, agriculture, and developing countries: Does adaptation matter? The World Bank Research Observer, 14(2), 277–293. Mendelsohn, R, W Nordhaus and D Shaw (1996). Climate impacts on aggregate farm values: Accounting for adaptation. Agriculture and Forest Meteorology, 80, 55–67. Mendelsohn, R, W Nordhaus and D Shaw (1994). The impact of global warming on agriculture: A Ricardian analysis. American Economic Review, 84, 753–771. Mitchell, T and T Tanner (2006). Adapting to Climate Change-Challenges and Opportunities for the Development Community. Institute of Development Studies and Tearfund, Teddington, UK. Nhemachena C (2009). Agriculture and future climate dynamics in Africa: Impacts and adaptation options. PhD thesis, Department of Agricultural Economics, Extension and Rural Development, University of Pretoria. Schimmelpfennig, D, J Lewandrowski, J Reilly, M Tsigas and I Parry (1996). Agricultural adaptation to climate change: Issues of long-run sustainability. Agricultural Economic Report No. (AER740), United States Department of Agriculture, USA. Seo, N and R Mendelsohn (2008). Animal husbandry in Africa: Climate change impacts and adaptations. African Journal of Agricultural and Resource Economics, 2(1), 65–82.


Strzepek, K and A McCluskey (2007). District level hydroclimatic time series and scenario analysis to assess the impacts of climate change on regional water resources and agriculture in Africa. Policy Research Working Paper 4290, Development Research Group, Sustainable Rural and Urban Development Team. The World Bank, Washington, DC. Washington, WM, JW Weatherly, GA Meehl, AJ Semtner Jr, TW Bettge, AP Craig, WG Strand Jr, JM Arblaster, VB Wayland, R James and Y Zhang (2000). Parallel climate model (PCM) control and transient simulations. Climate Dynamics, 16, 755–74.