Climate change and agriculture in Sri Lanka: a Ricardian valuation

C 2005 Cambridge University Press Environment and Development Economics 10: 581–596  doi:10.1017/S1355770X05002044 Printed in the United Kingdom Cli...
Author: Angelica Wood
6 downloads 2 Views 134KB Size
C 2005 Cambridge University Press Environment and Development Economics 10: 581–596  doi:10.1017/S1355770X05002044 Printed in the United Kingdom

Climate change and agriculture in Sri Lanka: a Ricardian valuation SUNG-NO NIGGOL SEO Yale University, USA. ROBERT MENDELSOHN Yale University, USA. MOHAN MUNASINGHE Munasinghe Institute For Development (MIND), Sri Lanka, and Yale University, USA. ABSTRACT. This paper measures climate change impacts on Sri Lankan agriculture using the Ricardian method. The model examines the net revenue per hectare of the four most important crops in the country. The limited range of temperature variation allows only a simple test of temperature impacts, but the greater range of precipitation across the country distinguishes more complex precipitation effects. We then examine the impacts of the climate predictions of five AOGCM models and two simple uniform change scenarios for Sri Lanka. The impacts of rainfall increases are predicted to be beneficial to the country as a whole in all five AOGCM scenarios, but temperature increases are predicted to be harmful. Nationally, the impacts vary from –11 billion rupees (− 20 per cent) to + 39 billion rupees (+72 per cent) depending on the climate scenarios. With warming, the already dry regions (the Northern and Eastern provinces), are expected to lose large portions of their current agriculture, but the cooler regions (the central highlands), are predicted to remain the same or increase their output. The paper reconfirms that climate change damages could be large in tropical developing countries, but highly dependent on the actual climate scenario.

1. Motivation and background Climate change has become a major concern to human society because of its potentially deleterious impact worldwide. It poses especially significant Sung-No Niggol Seo is a Ph.D. student at Yale School of Forestry and Environmental Studies, 205 Prospect St., New Haven, CT 06511, USA. Robert O. Mendelsohn is Edwin. W. Davis Professor at Yale University. Mohan Munasinghe is Chairman of MIND (Munasinghe Institute for Development), 10/1 De fonseka Place, Colombo 5, Sri Lanka and Vice-chairman of the United Nations Intergovernmental Panel on Climate Change (IPCC), Geneva. This research was made possible due to the assistance of MIND (Munasinghe Institute of Development), Colombo, Sri Lanka. Help provided by Mr U. Sapukotana, Ms Y. Deraniyagala, and staff of MIND in gathering data, is gratefully acknowledged. The authors wish to thank Peter Otis for funding this research and Mr Keisei Ohta, President of Inchemy Chiba in Japan, for his encouragement. We also appreciate the help from Pradeep Kurukulasuriya for drawing the maps in this paper. Three anonymous referees provided very constructive comments on an earlier draft.

582

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe

threats to sustainable development in developing countries, which have fewer resources and are more vulnerable (Munasinghe, 2001). Impacts on developing countries remain poorly understood because few studies have successfully measured the effects of climate on developing country economies. Nonetheless, it is likely that a developing country will be more vulnerable because a greater fraction of its economy is in climate sensitive sectors (for example, agriculture), it is already in a hot climatic zone, and the economy relies on labor-intensive technologies with fewer adaptation opportunities (Mendelsohn et al., 2001). Recently, an innovative methodology called the Action Impact Matrix (AIM) approach was used to rank climate change impacts and vulnerability of various sectors in Sri Lanka.1 This country-specific AIM was first developed in a workshop that harmonized views among diverse participants (over 70 leading economists, ecologists, sociologists, climate specialists, and other experts). The workshop results placed both traditional agriculture (for example, rice farming) and tree crops (that is, plantations) among the most vulnerable areas (Munasinghe et al., 2002). The urgent need for more detailed studies on climate change impacts on agriculture was also highlighted. This paper contributes to our knowledge concerning developing countries by examining the effect of climate on agriculture in Sri Lanka. The paper uses the Ricardian method (Mendelsohn et al., 1994) to look at how net revenue varies across climatic zones in Sri Lanka. Assuming that farmers adapt to where they live, the Ricardian method captures adaptation implicitly by comparing net outcomes for farmers facing different climates. Instead of studying the yields of specific crops, the model examines how climate in different places affects the net revenue or value of farmland. By directly measuring net revenues, the Ricardian method accounts for the direct impacts of climate on yields of different crops as well as the indirect substitution of different inputs, introduction of different activities, and other potential adaptations by farmers to different climates. The Ricardian approach has been used primarily in the US to predict the damages from climate change (Mendelsohn et al., 1994, 2001; Mendelsohn and Nordhaus, 1996; Mendelsohn and Neumann, 1999). There have also been a number of studies in Brazil and India (Dinar et al., 1998; Kumar and Parikh, 2001; Mendelsohn et al., 2001), as well as one recent study of Canada (Reinsborough, 2003). These studies suggest that climate change would be slightly beneficial to US agriculture, while it is likely to be harmful to tropical and semi-tropical countries. In addition to the Ricardian method, the Agro-Economic model is also capable of measuring climate effects. This model uses a combination of controlled experiments on specific crops, agronomic modeling, and economic modeling to predict climate impacts (Adams and McCarl, 2001). The presumed changes in yields from the agronomic model are fed into an economic model, which determines crop choice, production, and market prices. Though the ‘Agro-Economic Model’ contributed much 1

The Action Impact Matrix method was developed at the World Bank to identify and prioritize interactions between major economywide policies and key sustainability issues – see Munasinghe and Cruz (1994) and Munasinghe (1997).

Environment and Development Economics

583

to the scientific understanding of impacts, it is difficult to apply to developing countries. First, it has been criticized for underestimating adaptive responses to changing climate (Mendelsohn and Neumann, 1999). Second, there have not been sufficient experiments to determine agronomic responses in most developing countries. Finally, economic models of developing country agriculture are poorly calibrated. 2. Theory: Ricardian analysis We assume that farmers maximize net revenues per hectare, NR Max NR = Pi ∗ Qi (R, E) − Ci (Qi , R, E) R

(1)

Here Pi and Qi are respectively the price and quantity of good i; Ci (.) is the relevant cost function; R is a vector of inputs, and E reflects a vector of environmental characteristics of the farmer’s land including climate. Given that the farmer chooses inputs, R, to maximize NR, one can express the resulting outcome of NR in terms of E alone NR = f (E) The welfare value of a change in the environment from state A to B is   W= f (E iB ) ∗ L i − f (E iA ) ∗ L i

(2)

(3)

where L i is the amount of land of type i. Cross-sectional observations across different climates can reveal the climate sensitivity of farms. The advantage of this empirical approach is that the method not only includes the direct effect of climate on productivity, but also the adaptation response by farmers to local climate. Agronomic research and casual observation reveals that many crops have preferred temperature and precipitation zones. Temperatures and precipitation levels either below or above such optimal ranges reduce productivity. The evidence suggests that the relationship between net revenue and these climate variables should be hill-shaped. We attempt to capture this hill shape using a quadratic functional form    NRi = a 0 + a s Ts + b s Ts2 + c s Ps + ds Ps2 + (4) f c Zc + e where Ts and Ps represent normal temperature and precipitation variables in each season; and Zc represents relevant socio-economic variables. The original Ricardian studies used land value for the dependent variable. In many developing countries, however, land value is not available. Annual net revenue per hectare can be used instead, since land value is the present value of a future stream of net revenue (Dinar et al., 1998). The use of annual net revenues, however, introduces a potential problem since the net revenue in any one-year is influenced by the weather in that year. The Ricardian method has been criticized on several counts. The original estimates did not include surface water or irrigation (Cline, 1996 and

584

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe

Schlenker et al., 2003). The method cannot measure the effect of variables that do not vary across space such as CO2 . The method measures long-run adaptation, but not the speed of adaptation. The model assumes current technology, so that it does not take into account technology that may be available in the future. The model assumes no price effects (Darwin, 1999). If climate change alters supplies of individual crops, prices are likely to change. The measurements reflect current agricultural policies. These problems are significant but not fatal (Mendelsohn, 2001). CO2 effects can be included exogenously as can new technology. Global prices are not expected to change dramatically as a result of climate change (Reilly et al., 1996). Irrigation and surface water have been taken into account in more recent estimates and found not to influence climate sensitivity (Mendelsohn and Nordhaus, 1999b; Mendelsohn and Dinar, 2003). It is not clear what role current agricultural policies play in Ricardian measurements. 3. Sri-Lanka Sri Lanka is an island located under the Indian peninsula. It has an area of 66,000 sq. km, stretching over 433 km from North to South (latitude 5◦ 55 North to 9◦ 51 North), and 244 km East to West at its widest point (longitude 79◦ 41 East to 81◦ 53 East). The country consists of nine provinces and 25 districts. Colombo, the capital, has the highest monthly income, whereas the Eastern province (Baticaloa, Trincomalee, Ampara) and Northern province (Jaffna, Kilinochchi, Mannar, Vavuniya, Mullaitivu) have the lowest income levels. The Western province (Gampaha, Colombo, Kalutara) and Southern province (Galle, Matara) show high population density and low altitude, whereas the Northern and Eastern provinces have low population density. The Central province has the highest elevation in the nation. Sri Lanka is a tropical country with distinct dry and wet seasons. The climate is characterized by two monsoons. The ‘Yala’ season, during the South-West Monsoon, lasts from May to August, bringing rain to the Southand West-coast regions as well as central highlands. The dry season in these regions is from December to March. The second wet season arrives with the North-East Monsoon, which blows from October to January (called the Maha season), bringing rain to the North and East of the island. The dry season in the North-East is from May to September. There is also an intermonsoonal period in October and November when rain and thunderstorms can occur in many parts of the island. The nation is commonly categorized into dry zones and wet zones depending upon the annual rainfall. Wet zones are divided into a maritime wet zone and hill country wet zone, depending upon the altitude. Table 1 shows climatic zones and districts. The South, South-West, and Central highlands are much wetter than the North and North-Central regions. Colombo and the low-lying coastal regions have an average temperature of 27 ◦ C. The temperature falls rapidly with increasing altitude. At Kandy (altitude of 500 m) the average temperature is 20 ◦ C and at Nuwara Eliya (1889 m) it drops down to 16 ◦ C. In this paper, we use ‘normal temperature’ and ‘normal rainfall’ that is defined to be the 30-year average temperature and rainfall. The normal

Environment and Development Economics

585

Table 1. Climatic zones and districts Climatic zone Wet zone Dry zone

Districts Maritime Hill country

Colombo, Gampaha, Kalutara, Galle, Matara Kandy, Nuwara Eliya, Ratnapura, Kegalle Matale, Hambantota, Jaffna, Mannar, Vavuniya, Mullaitivu, Batticaloa, Ampara, Trincomalee, Kurunegala, Puttalam, Anuradhapura, Polonnaruwa, Badulla, Moneragala.

Note: Ministry of Finance, 1997.

temperature and rainfall are presented in the appendix Tables A1 and A2 respectively. Sri-Lanka’s main agricultural products comprise paddy rice, commercial tree crops, and highland crops. Paddy rice is grown all around the country and contributed about 3.6 per cent to GDP in 1995 (Statistical Abstract, 1997). The cost of production differs depending on the methods of paddy rice production (that is, irrigation-rainfed or irrigated). Three commercial tree crops: coconut, rubber, and tea are grown on the West coast and hill area in Sri-Lanka. Tree crops comprise a significant portion of the exports of the country and are taken care of by the government Department of Plantation Industries. They contributed 2.1 per cent (tea), 2.1 per cent (coconut), and 0.8 per cent (rubber) to GDP in 1995 (Department of Census and Statistics, 1997). In 1995, agricultural exports totalled 42.5 billion Sri Lankan rupees, of which tea exports were 24.6 billion, rubber exports were 5.7 billion, and coconut exports were 5.3 billion. These three commercial crops represented 84 per cent of total agricultural exports. The 1995GDP in market price was 667.8 billion Sri Lankan rupees (13 billion US dollars) and agriculture accounted for 20 per cent (or 133 billion rupees) of this total (Central Bank, 1998). 4. Ricardian model The Ricardian model regresses net revenue on climate and other explanatory variables (see equation (4)). Net revenue for each district was constructed from the data recorded by the government.2 Net revenue per unit area equals total net revenue of the district/area of cropland in hectares. For rice production, net revenues were constructed from the national statistics data book (Department of Census and Statistics, 1997). We use the average of irrigated and dryland farming for the cost of production of paddy crops. To construct net revenues of commercial crops, the Plantation Sector Statistical Pocket Book, 2001 was used as the main source. Extent data of 2

“Statistical Abstracts of Sri-Lanka, 1997” released by the Department of Census and Statistics, ‘Agrarian Data Bank’ of Agrarian Research and Training Center, ‘Economic and Social Statistics of Sri-Lanka, 1998’ by the Central Bank of SriLanka and ‘Cost of cultivation of agricultural crops, 2000/2001 Maha’ from the Socioeconomics and Planning Center of the University of Peradeniya.

586

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe Table 2. Total net revenues of districts in 1995 (million rupees)

Province

District

Paddy

Coconut

Rubber

Tea

Total

Western

Gampaha Colombo Kalutara Matale Kandy Nuwara Kegalle Ratnapura Galle Matara Hambantota Badula Monaragala Baticaloa Trincomale Ampara Anuradhapu Polonnaruw Puttalam Kurunegala Mannar Vavuniya Kilinochchi Jaffna Mullaittivu

194 67 375 530 339 103 485 489 82 363 1,076 623 357 436 464 2,497 1,638 3,116 333 2,721 68 214 336 95 182 17,192

2,415 388 541 506 356 44 914 714 548 689 987 51 272 174 76 165 244 127 2,205 6,808 50 18 – 427 93 18,823

164 339 1,386 84 68 2 1,538 1,007 375 178 0.7 11 51 – – – – – – 106 – – – – – 5,316



2,775 800 2,546 1,417 2,415 3,712 3,323 4,055 2,368 2,541 2,083 2,786 689 611 541 2,662 1,882 3,244 2,539 9,639 118 232 336 522 276 54,124

Central

Sabaragamuw Southern

Uva Eastern

Northcentral Northwestern North

Total

4 242 296 1,651 3,562 385 1,844 1,361 1,310 19 2,100 8 – – – – – – 3 – – – – – 12,792

Notes: Total = total net revenue of the four crops included for the whole production of the districts.

districts were multiplied by average productivity of the districts to get total production of each crop. Table 2 shows the total net revenues of the districts by the production of four crops stated above. The five districts in the Northern province, and the two districts in the Eastern province have the lowest revenues, whereas the Western and Southern districts along with the Hill region show higher revenues. The main area for paddy production lies in the North-East of the country. Tea and rubber are cultivated mostly in the West and Central area of the country, while coconut is harvested largely on the Western coast. Our dependent variable is defined as Net revenue per unit area = total net revenue of the district/area of cropland in hectares of the district. Four months are selected as explanatory variables to represent Sri-Lanka’s climate. November and January represent the Maha monsoon period. May

Environment and Development Economics

587

Table 3. Climate regressions Full model

Simple model

Variables

Estimates

Standard error

T-value

Jan. temp (Jan. temp)2 May temp (May temp)2 Nov. temp (Nov. temp)2 May rain Sept. rain Jan. rain Nov. rain Irrigated land Pop density Altitude

347,785 −5,999 443,965 −7,760 −745,121 12,901 −190 227

153,092 2,794 185,709 3,202 330,648 5,907 81 93

2.27∗ −2.15∗ 2.39∗ −2.42∗ −2.25∗ 2.18∗ −2.33∗ 2.43∗

−0.08 0.06 −1.42 −3.31 2.62 −1.27 −13.16 11.13 −1.18 N = 25 R-sq: 0.91, Adjusted R-sq: 0.84

Estimates

Standard error

T-value

−4,105

1,173

−3.5∗

93 103 −141 −0.15494

28 49 68 0.09968

3.27∗ 2.11∗ −2.05∗ −1.55

−27.3012 8.98063 −3.04∗ N = 25 R-sq: 0.69, Adjusted R-sq: 0.59

Note: Superscript on the T-value implies the estimate is significant at 5 per cent.

represents the Yala monsoon period. September is included to see the effects of the period between the two monsoons. We present two regressions in table 3. Our first regression, the full model, includes climate variables of the four months and is specified as a quadratic equation. We include three control variables, that is irrigated land, population density, and altitude. All three have been shown to be important in earlier studies. The regression shows a high degree of fit, with an adjusted r-square of 0.84. The climate variables are all significant at the 5 per cent level. The control variables are not significant. We also present a more parsimonious model, the simple model. In the previous model, the temperature coefficients were highly correlated and the quadratic specification seemed to make the situation worse. Consequently, the confidence intervals resulting from the full model were quite large. By simplifying the model, we estimate fewer coefficients, but the coefficients are more independent of each other. The resulting national impact estimates are more precise. The simplified model includes just three months of rainfall (in January, September, and November), and one month of temperature (in May). They are all specified in a linear form (i.e., without the temperature squared terms in equation (4)). The model includes two important control variables, irrigated land and altitude. The simple model shows an adjusted r-square of 0.59 that is significantly lower than that of the full model, but better t-statistics for coefficient estimates. The temperature coefficient implies that net revenue will decrease by −4,105 per degree of warming. Summing the coefficients on the rain variables implies that rain is in general beneficial, although increased rain in November is harmful.

588

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe Table 4. AOGCM climate change predictions Jan. rainfall

Current value 8.71 Value in 2100 CCSR 5.14 CGCM 7.32 CSIRO 10.85 HAD3 2.36 PCM 31.27 Percentage change of rainfall CCSR −40% CGCM −15% CSIRO 24% HAD3 −72% PCM 259%

Sept. rainfall

Nov. rainfall

May temperature

17.91

29.74

27.18

37.36 38.32 23.58 40.88 24.92

17.01 14.05 31.75 22.32 44.21

29.83 31.2 30.14 30.59 28.76 Absolute change 2.65 4.02 2.96 3.41 1.58

108% 113% 31% 128% 39%

−42% −52% 6% −24% 48%

Table 5. National-level impacts Scenarios

Temp. effect

Rain effect

Temp + rain effect

CSIRO PCM CCSR CGCM HAD3

−19.9 B (−35%) −10.6 B (−18%) −17.8 B (−31%) −27.0 B (−50%) −22.9 B (−40%)

8.9 B (+14%) 6.2 B (+11%) 53.1 B (+98%) 66.4 B (+122%) 39.9 B (+72%)

−11.0 B (−20%) (−23 B, 1 B) −4.3 B (−7%) (−34 B, 25 B) 35.2 B (+64%) (−7 B, 77 B) 39.3 B (+72%) (−13 B, 91 B) 16.9 B (+29%) (−19 B, 53 B)

Note: The absolute numbers are in billion Sri Lankan rupees. The percentage numbers show the fractions of aggregate net revenue of the crops included in this analysis.

5. Future climate impacts In this paper, we use AOGCM climate predictions for Sri Lanka, by 2100. The five AOGCM’s are CGCM (Boer et al., 2000), CSIRO (Gordon and O’Farrell, 1997), CCSR (Emori et al., 1999), HAD3 (Gordon et al., 2000), and PCM (Washington et al., 2000). The five climate models predict a wide range of outcomes. Table 4 presents the rather wide range of predictions of future climate from the five global models. Table 5 describes the national-level impact estimates for all the climate models. In all five models, the changes in rainfall change are beneficial to the nation. The benefit ranges from 11 per cent to 122 per cent of the current net revenue of the crops included in this model. In all five models, the impacts of temperature change are predicted to be harmful to the nation. The loss ranges from −18 per cent to −50 per cent of the current agricultural productivity. The combined effect of temperature change and rainfall change varies. The CSIRO and PCM models predict a loss from climate change. Both models predict an increase in November rains that offset precipitation increases in the other months. The remaining three models predict welfare

Environment and Development Economics

589

Table 6. District level impacts (1,000 rupees) HAD3

CSIRO

District

Impact

% Change

Impact

% Change

Gampaha Colombo Kalutara Matale Kandy Nuwara Kegalle Ratnapura Galle Matara Hambantota Badulla Moneragala Baticaloa Trincomalee Ampara Anuradhapura Polonnaruwa Puttalam Kurunegala Mannar Vavuniya Kilinochchi Jaffna Mullaitt

19 35 38 0 14 23 43 33 30 17 −3 −4 −1 −13 −3 −14 −3 −5 2 6 −6 1 −1 −1 1

61 98 92 −1 44 59 121 85 73 52 −11 −13 −2 −54 −20 −74 −15 −28 7 27 −35 4 −17 −12 9

−7 −3 −1 −7 −5 −3 −2 −2 −3 −5 −10 −6 −11 −8 −9 −7 −10 −8 −10 −10 −12 −10 −11 −11 −10

−22 −8 −2 −20 −17 −9 −5 −6 −6 −15 −44 −18 −25 −35 −54 −39 −46 −48 −46 −45 −72 −51 −202 −121 −62

gains as precipitation benefits outweigh temperature losses. These three models predict that precipitation will increase in the beneficial months of January and September. However, these models also predict decreased rainfall in November – the one month when rainfall is harmful. Table 5 also shows the uncertainty associated with the national-level impact predictions. In the last column, the numbers in parentheses below the impact estimates are 95 per cent confidence intervals of the net benefits. The net climate change impacts predicted are uncertain in that all the intervals include zero. The impact estimates are not statistically different from zero. These results reveal a limitation of the Ricardian method when applied to small countries. There may not be enough variation in the climate over a small region, to estimate reliable impacts for non-marginal climate changes. The analysis also describes how the impacts vary across districts. Table 6 summarizes the impact estimates for two of the AOGCM models, CSIRO and HAD3. Table 6 shows how net revenue in each district will be affected by the change in climate. As before, the simple model is used for this calculation. The impact column shows the change in net revenue caused by the climate change.

590

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe Table 7. Change in net income: Ricardian results

Country

Temperature change Change in with 7% precipitation net income increase (celsius) (%)

Sri Lanka Sri Lanka United States India

2.0 3.5 2.0 2.0

India

3.5

India India Brazil Brazil

2.0 3.5 2.0 3.5

Source

−27% −46% From −3 to +3

Simple model (this paper) Simple model (this paper) Mendelsohn, Nordhaus, and Shaw (1994) From −3 to −6 Sanghi, Mendelsohn, and Dinar (1998) From −3 to −8 Sanghi, Mendelsohn, and Dinar (1998) From −7 to −9 Kumar and Parikh (1998) From −20 to −26 Kumar and Parikh (1998) From −5 to −11 Sanghi (1998) From −7 to −14 Sanghi (1998)

Note: The table is modified from Mendelsohn and Dinar (1999).

In the HAD3 scenario, the central highland and west coast of the nation are expected to gain from climate change, while the northern and eastern dry lands are predicted to lose. In the CSIRO scenario, all districts are expected to lose from the future climate. However, the impacts in the central highlands are small compared with the northern and eastern provinces. The damages in these regions amount to more than 50 per cent of the current net revenue. Impact maps of the two models are presented in figure 1. Different colors denote percentage changes of net revenue in the districts. The maps depict district-level results, that is, climate change will be harmful to the northern and eastern provinces, while it will be less harmful or beneficial to the central uplands and west coast of the country. The maps predict a likely future shift of Sri Lankan agriculture from the north and east to the central highlands. 6. Conclusions This paper describes a Ricardian analysis of agriculture in Sri Lanka. The study was able to measure both temperature and precipitation effects. In general, warming is expected to be harmful to Sri Lanka but increases in rainfall will be beneficial. Applying the estimated regression results to five climate scenarios, we estimate a range of effects from a loss of 20 per cent to a gain of 72 per cent. The scenarios with losses had overall harmful temperature impacts, with offsetting precipitation benefits. The scenarios with gains had harmful temperature effects, which were dominated by beneficial changes in rainfall. In the beneficial scenarios, there were large increases of precipitation in beneficial months and decreases of precipitation in harmful months (November). Table 7 compares our results with other Ricardian analyses, using a uniform scenario with a 2 ◦ C warming and a 7 per cent increase in precipitation. We find that the Sri Lankan simple model predicts national impacts of −27 per cent (loss) of agricultural output. A 3.5 ◦ C uniform increase predicts impacts of −46 per cent. The Sri Lankan results are

Figure 1. Impact Maps, HAD3 and CSIRO models

Environment and Development Economics 591

592

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe

predicted to be more severe than the results found in other countries. This is surprising, especially given the close proximity of India. Clearly more careful analysis of both Sri Lanka and other countries is warranted.3 Note that the AOGCM-based Sri Lankan results in this paper show a much wider range of variability because they test a much larger range of climate scenarios. It is also helpful to compare these results with agronomic studies done in South and South-East Asia. An examination of the literature shows that there is a wide range of predicted effects across crops and countries (Sanderson, 2002; Watson et al., 1998). In general, agronomic studies predict that warming will be harmful to production, but these studies do not include farmer adaptation or carbon dioxide fertilization effects. Counting these factors in the results, warming is likely to reduce production about 4 per cent (Matthews et al., 1997) in the region. However, agronomic studies definitely find effects in specific crops as large as those listed in this study. The study also demonstrates that the impacts will vary by district. Districts in the Northern and Eastern regions that are currently marginal from a climate perspective are particularly vulnerable to future warming. This paper reconfirms that tropical developing countries are sensitive to the predicted climate changes over the next century. Depending on the climate scenario, the damages could be substantial. However, not every scenario is harmful to the country. Some of the climate scenarios that predict an improvement in rainfall patterns for Sri Lanka may be beneficial. Furthermore, the estimates in this study do not include the beneficial effects of carbon fertilization, which could offset many of the damages expected from the climate alone. There is still a great deal to be learned about climate sensitivity, especially in the tropical regions of the world. References Adams, R. and B. McCarl (2001), ‘Agriculture: agronomic-economic analysis’, in R. Mendelsohn (ed.), Global Warming and the American Economy: A Regional Assessment of Climate Change, Cheltenham: Edward Elgar Publishing, pp 18–31. Amien, I., P. Rejekiningrum, A. Pramudia, and E. Susanti (1996), ‘Effects of interannual climate variability and climate change on rice yield in Java, Indonesia’, Water, Air, and Soil Pollution 92: 29–39. Bazzaz, F. and W. Sombroek (eds) (1996), Global Climate Change and Agricultural Production: Direct and Indirect Effects of Changing Hydrological Pedological and Plant Physiological Processes, Chichester: John Wiley & Sons. 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: 427–450. Central Bank of Sri-Lanka (1998), Economic and Social Statistics of Sri-Lanka. Cline, W. (1992), The Economics of Global Warming, Washington, DC: Institute of International Economics. Cline, W. (1996), ‘The impact of global warming on agriculture: comment’, The American Economic Review 86: 1309–1311. 3

There has been on-going research on Sri Lankan agriculture at the household level by Kurukulasuriya and Ajwad (unpublished). It shows a comparable decrease in aggregate net revenue, with the range of values (13–47 per cent) depending on the extent of climate change.

Environment and Development Economics

593

Darwin, R. (1999), ‘The impact of global warming on agriculture, a Ricardian analysis: Comment’, The American Economic Review 89: 1049–1052. Department of Agriculture (2002), Agrarian Data Bank, Sri Lanka. Department of Census and Statistics (1997), Statistical Abstract of the Democratic Socialist Republic of Sri-Lanka. Dinar, A., R. Mendelsohn, R. Evenson, J. Parikh, A. Sanghi, K. Kumar, J. McKinsey, and S. Lonergan (eds) (1998), ‘Measuring the impact of climate change on Indian agriculture’, World Bank Technical Paper No. 402, Washington, DC. Emori, S., T. Nozawa, A. Abe-Ouchi, A. Namaguti, and M. Kimoto (1999), ‘Coupled ocean-atmospheric model experiments of future climate change with an explicit representation of sulfate aerosol scattering’, Journal of the Meteorological Society Japan 77: 1299–1307. Escano, C.R. and L.V. Buendia (1994), ‘Climate impact assessment for agriculture in the Philippines: simulation of rice yield under climate change scenarios’, in C. Rosenzeig and A. Iglesias (eds), Implications of climate change for international agriculture: crop modeling study’, United States Environmental Protection Agency, Philippines Chapter, Washington DC, pp. 1–13. Fischer, G., K. Frohbergy, M.L. Parry, and C. Rozenzweig (1996), ‘The potential effects of climate change on world food production and security’, in F. Bazzaz and W. Sombroek (eds), Global Climate Change and Agricultural Production: Direct and Indirect Effects of Changing Hydrological, Pedological and Plant Physiological Processes, FAO and John Wiley & Sons. Frisvold, G. and B. Kuhn (1999), Global Environmental Change and Agriculture: Assessing Impacts, Cheltenham: Edward Elgar. Gordon, H. and S. O’Farrell (1997), ‘Transient climate change in the CSIRO coupled model with dynamic sea ice’, Mon. Weather Research 125: 875–907. Gordon, C., C. Senior, H. Banks, I. Gregory, T. Johns, J. Mitchell, and R. Wood (2000), ‘The simulation of SST, sea ice extents, and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments’, Climate Dynamics 16: 147–168. Intergovernmental Panel on Climate Change, Climate Change (2001), The Scientific Basis, Cambridge University Press. Kulukulasuriya, P. and I. Ajwad (2003), ‘Estimating the impact of climate change on small holders: a case study on the agriculture sector in Sri Lanka’, School of Environment and Forestry Studies, Yale University, New Haven, USA. Kumar, K.S.K. and J. Parikh (1998), ‘Climate change impacts on Indian agriculture: the Ricardian approach’, in A. Dinar, R. Mendelsohn, R. Evenson, J. Parikh, A. Sanghi, K. Kumar, J. McKinsey, and S. Lonergon (eds), ‘Measuring the impact of climatic change on Indian agriculture’, World Bank Technical Report No. 409, World Bank, Washington, DC. Kumar, K. and J. Parikh (2001), ‘Indian agriculture and climate sensitivity’, Global Environmental Change 11: 147–154. Luo, Q. and E. Lin (1999), ‘Agricultural vulnerability and adaptation in developing countries: the Asia-Pacific region’, Climate Change 43: 729–743. Matthews, R.N., M.J. Kropff, T. Horie, and D. Bachelet (1997), ‘Simulating the impact of climate change on rice production in Asia and evaluating options for adaptation’, Agricultural Systems 54: 399–425. Mendelsohn, R. (2003), ‘The impact of global warming on Pacific Rim countries’, in Ching Cheng Chang, Robert Mendelsohn, and Daigee Shaw (eds), Global Warming and the Asian Pacific, Cheltenham: Edward Elgar Publishing. Mendelsohn, R. (ed.) (2001), Global Warming and the American Economy: A Regional Assessment of Climate Change, Cheltenham: Edward Elgar Publishing. Mendelsohn, R. and A. Dinar (1999), ‘Climate change, agriculture, and developing countries: does adaptation matter?’, The World Bank Research Observer 14: 277–293.

594

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe

Mendelsohn, R. and A. Dinar (2003), ‘Climate, water, and agriculture’, Land Economics 79: 328–341. Mendelsohn, R., A. Dinar, and A. Sanghi (2001), ‘The effect of development on the climate sensitivity of agriculture’, Environment and Development Economics 6: 85–101. Mendelsohn, R., W. Morrison, M. Schlesinger, and N. Adronova (2000), ‘Countryspecific market impacts from climate change’, Climatic Change 45: 553–569. Mendelsohn, R. and J. Neumann (eds) (1999), The Impact of Climate Change on the United States Economy, Cambridge University Press. Mendelsohn, R. and W. Nordhaus (1996), ‘The impact of global warming on agriculture: reply to Cline’, The American Economic Review 86: 1312–1315. Mendelsohn, R. and W. Nordhaus (1999a), ‘The impact of global warming on agriculture: reply to Quiggin and Horowitz’, The American Economic Review 89: 1046–1048. Mendelsohn, R. and W. Nordhaus (1999b), ‘The impact of global warming on agriculture, a Ricardian analysis: reply to Darwin’, The American Economic Review 89: 1053–1055. Mendelsohn, R., W. Nordhaus, and D. Shaw (1994), ‘The impact of global warming on agriculture: a Ricardian analysis’, The American Economic Review 84: 753–771. Ministry of Finance and Planning (1997), Statistical Abstract of the Democratic Socialist Republic of Sri Lanka. Ministry of Plantation Industries (2001), Plantation Sector Statistical Pocket Book. Munasinghe, M. (2001), ‘Sustainable development and climate change: applying the sustainomics transdisciplinary meta-framework’, International Journal of Global Environmental Issues 1: 13–55. Munasinghe, M. (ed.) (1997), ‘Environmental impacts of macroeconomic and sectoral policies’, The World Bank and the International Society of Ecological Economics (ISEE), Washington, DC. Munasinghe, M. and W. Cruz (1994), Economywide Policies and the Environment, Washington, DC: World Bank. Munasinghe, M., U. Sapukotana, and Y. Deraniyagala (2002), ‘Interactions between climate change and sustainable development in Sri Lanka: analysis using the action impact matrix (AIM)’, MIND Research Discussion Paper, Munasinghe Institute for Development, Colombo. Parry, M.L., M. Blantran, A.L. de Rozari, S. Chong, and S. Panich (1992), ‘The potential socio-economic effects of climate change in South East Asia’, United Nations Environment Programme, Nairobi. Qureshi, A. and D. Hobbie (eds) (1994), ‘Climate change in Asia: executive summary’, Asian Development Bank, Manila. Reilly, J., W. Baethgen, F.E. Chege, S.C. van de Greijn, L. Ferda, A. Iglesia, C. Kenny, D. Patterson, J. Rogasik, R. Rotter, C. Rosenzweig, W. Sombroek, and J. Westbrook (1996), ‘Agriculture in a changing climate: impacts and adaptations’, in IPCC (Intergovernmental Panel on Climate Change), R. Watson, M. Zinyowera, R. Moss, and D. Dokken (eds), Climate Change 1995: Impacts, Adaptations, and Mitigation of Climate Change: Scientific-Technical Analyses, Cambridge: Cambridge University Press. Reinsborough, M.J. (2003), ‘A Ricardian model of climate change in Canada’, Canadian Journal of Economics 36: 21–40. Sanderson, J. (2002), ‘An analysis of climate change impact and adaptation for South East Asia’, Ph.D. thesis, Centre for Strategic Economic Studies, Victoria University of Technology. Sanghi, A. (1998), ‘Global warming and climate sensitivity: Brazilian and Indian Agriculture’, Ph.D. Thesis, University of Chicago, USA. Sanghi, A., R. Mendelsohn, and A. Dinar (1998), ‘The climate sensitivity of Indian agriculture’, in A. Dinar, R. Mendelsohn, R. Evenson, J. Parikh, A. Sanghi,

Environment and Development Economics

595

K. Kumar, J. McKinsey, and S. Lonergon (eds), ‘Measuring the impact of climatic change on Indian agriculture’, World Bank Technical Report No. 409, World Bank, Washington, DC. Schlenker, W., M. Hanemann, and A. Fisher (2003), ‘Will US agriculture really benefit from global warming? Accounting for irrigation in the hedonic approach‘, Presented at the NBER summer Research Institute, Boston, MA. Tongyai, C. (1994), ‘Impact of climate change on simulated rice production in Thailand’, in C. Rosenzeig and A. Iglesias (eds), Implications of Climate Change for International Agriculture: Crop Modeling Study, Washington DC: United States Environment Protection Authority. Washington, W. et al. (2000), ‘Parallel climate model (PCM): control and transient scenarios’, Climate Dynamics 16: 755–774. Watson, R.T., M.C. Zinyowera, and R.H. Moss (eds) (1998), The Regional Impacts of Climate Change: An Assessment of Vulnerability, Cambridge University Press.

Appendix Table A1. Thirty-year average temperature of the districts Province

District

Jan.

March

May

July

Sept.

Nov.

Western

Gampaha Colombo Kalutara Matale Kandy Nuwara Eliya Kegalle Ratnapura Galle Matara Hambantota Badula Monaragala Bandarawela Baticaloa Trincomalee Ampara Anuradhapur Polonnaruwa Puttalam Kurunegala Mannar Vavuniya Kilinochchi Jaffna Mullaitivu

26.9 26.6 26.4 23.3 23.3 14.7 27.2 27.2 25.9 25.9 26.3 21.3 18.2 18.2 25.5 26 25.5 25.3 24.9 25.7 25.8 26.3 24.6 24.6 25.8 24.6

27.9 27.7 27.5 25.6 25.6 16.3 28.3 28.3 27.3 27.3 27.4 23.4 20 20 27.2 27.9 27.2 28.3 27.9 28.1 28.4 28.1 27.8 27.8 27.6 27.8

28.2 28.3 28.2 25.6 25.6 17.1 27.9 27.9 27.6 27.6 28.1 25 21.5 21.5 29.4 30.5 29.4 28.8 28.7 29 28.3 29.6 29 29 30.1 29

27.6 27.6 27.5 24.5 24.5 15.7 27.1 27.1 26.7 26.7 27.7 24.6 21.2 21.2 29.4 30.1 29.4 28.8 28.5 28.5 27.4 28.6 29.1 29.1 29.4 29.1

27.5 27.5 27.3 24.3 24.3 15.7 27 27 26.6 26.6 27.2 24.2 20.7 20.7 28.5 29.6 28.5 28.6 28.5 28.5 27.4 28.5 28.7 28.7 28.9 28.7

26.9 26.7 26.6 24.2 24.2 15.6 27.1 27.1 26.3 26.3 26.7 22.8 19.6 19.6 26.5 26.7 26.5 26.4 26.2 26.6 26.5 27.1 26.1 26.1 26.6 26.1

Central

Sabaragamuwa Southern

Uva

Eastern

Northcentral Northwestern North

Notes: Center for Climate Change Studies, Department of Meterorology, Colombo. In case the data is not available in the district, the data in an adjacent district is used.

596

Sung-No Niggol Seo, Robert Mendelsohn, and Mohan Munasinghe Table A2. Thirty-year average rainfall of districts

Province

District

Jan.

March

May

July

Sept.

Nov.

Western

Gampaha Colombo Kalutara

51.62 95.08 130.362

148.81 223.08 234.323

329.73 470.78 537.215

129.74 237.54 269.885

221.45 351.88 396.892

307.71 396.9 414.823

Central

Matale Kandy Nuwara Eliya

154.589 126.522 123.783

87.9444 125.267 131.952

126.122 225.644 295.709

81.8111 215.111 312.709

124.856 221.356 292.183

304.767 333.7 317.717

Sabarabgamu

Kegale Ratnapura

78.94 123.139

207.34 233.8

463.553 430.667

333.613 247.228

407.02 351.517

418.167 399.433

Southern

Galle Matara Hambantota

119.15 117.18 68.975

203.483 145.57 80.1375

425.867 280.26 91

238.75 158.8 51.9875

334.067 239.43 78.25

367.917 333.88 211.638

Uva

Badulla Monaragala

216.638 71.35

135.681 134.7

117.831 93.45

72.0375 40.35

129.306 73.4

310.725 286

Eastern

Baticaloa Trincomalee Ampara

197 130.575 218.314

87.7 53.75 111.886

43.6 43.35 48.2

37.575 55.675 37.6

50.325 92.15 63.1857

298.075 282.175 268.629

Northcentral

Anuradhapur Polonnaruwa

77.05 155.31

68.7 83.57

75.475 66.23

31.4 47.09

74.675 93.94

220.275 285.75

Northwestern

Puttalam Kurunegala

46.8 53.55

82.1091 96.025

146.736 150.208

39.7909 76.025

89.7 121.842

247.355 268.95

North

Mannar Vavuniya Kilinochchi Jaffna Mullaitivu

52.45 67.55 89.4 65.85 77.7333

45.4 53.85 37.8 22.65 48.2333

53.475 71.3 49.125 50.95 67.8

9.675 35.45 24.375 21.1 42.5333

35.75 91.5 69.325 59.65 88.5667

231.3 259 338.25 312.65 314.567

Notes: Center for Climate Change Studies, Department of Meteorology, Colombo. District rainfall is an averaged value of several stations inside the district.