CLIMATE CHANGE AND ECONOMIC GROWTH: EVIDENCE FROM BANGLADESH

It is now unequivocally established that climate change is a reality, and the adversities of climate transformations pose as one of the greatest challenges to the contemporary world.1 The Intergovernmental Panel on Climate Change (IPCC) defines climate change as “a change in the state of the climate that can be identified (e.g., using statistical tests) by changes in the mean and/or the variability of its properties and that persists for an extended period, typically decades or longer.”2 Bangladesh experiences different climate-induced slow and rapid onset disasters, such as cyclones, recurring and extended floods, erratic rainfall, extended droughts, and salinity intrusion due to the sea-level rise (SLR). Thus, climate change is a significant threat to the lives and livelihoods of low-income people in the country. According to the Global Climate Risk Index 2016, annual average loss from disasters for 1995 to 2014 is 0.86 per cent of Gross Domestic Product (GDP) of the country.3 Bangladesh has just graduated from the status of low-income to lower middle-income country, as classified by the

1Dilruba Yasmin Chowdhury is Qualitative Expert and Study Manager, Shamunnay, Dhaka. Her e-mail
address is: dilruba@gmail.com
© Bangladesh Institute of International and Strategic Studies (BIISS), 2017.
1 IUCN Bangladesh, Protocol for Monitoring of Impacts of Climate Change and Climate Variability in Bangladesh,
Dhaka: International Union for Conservation of Nature, 2011, pp. xiv-182.
2
 IPCC, “Summary for Policymakers”, in S. Solomon, D. Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt,
M. Tignor and H. L. Miller (eds.), Climate Change 2007: The Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report the Intergovernmental Panel on Climate Change, Cambridge, United
Kingdom and New York, USA: Cambridge University Press, 2007.
3
 S. Kreft, D. Eckstein and I. Melchior, Global Climate Risk Index 2016, Bonn: Germanwatch, 2015.

 World Bank. There have been many other notable progresses in economic and social spheres during this short period including a trend of steady and impressive growth over the past ten years or so. However, climate change would come out with negative consequences for growth performances of the country.

A number of studies have been conducted so far on trends of change in climatic parameters in the context of Bangladesh. Warrick et al. 4 , Karmakar and Shrestha5 and Debsarma6 provided assessment of changes in temperature and precipitation over Bangladesh, while Chowdhury and Debsarma7 and Mia8 reported assessment of the changes in temperature based on analysis of historical data of some selected weather stations in Bangladesh. Karmakar and Nessa9 and Karmakar10 provided assessment of the effects of climate change on natural disasters.

Haque et al. 11 indicated that the average increase in temperature would be 1.3°C and 2.6°C for the projected years of 2030 and 2075, respectively. In line with the IPCC projections, the rise in winter temperature in Bangladesh was predicted to be higher probably due to significant increase in monsoon precipitation, which could also cause severe flooding in the future. Chowdhury and Debsarma12 studied that the projected changes would be 1.4°C in the winter and 0.7°C in the monsoon months in 2030. For 2075, the variation would be 2.1°C and 1.7°C for winter and monsoon, respectively. It has also observed the increasing tendency of lowest minimum temperature over Bangladesh. Warrick et al. 13 studied the variation of temperature and rainfall over Bangladesh. In their study, mean-annual temperatures have been expressed as departures from the reference period 1951-1980. It is evident that during

4
 R. A. Warrick, A. H. Bhuiya and M. Q. Mirza, The Greenhouse Effect and Climate Change: Briefing Document,
No. 1, Dhaka: Bangladesh Unnayan Parishad, 1994.
5
 S. Karmakar and M. L. Shrestha, “Recent Climate Change in Bangladesh”, SMRC Series, No. 4, Dhaka: SAARC
Meteorological Research Centre, 2000.
6
 S. K. Debsarma, “Intra-annual and inter-annual variation of rainfall over different regions of Bangladesh”,
Proceedings of SAARC Seminar on Climate Variability in the South Asian Region and its Impacts, Dhaka: SAARC
Meteorological Research Centre, 2003.
7
 M. H. K. Chowdhury and S. K. Debsarma, Climate Change in Bangladesh – A Statistical Review, Report of
IOC-UNEP Workshop on Impacts of Sea Level Rise due to Global Warming, Dhaka: Intergovernmental
Oceanographic Commission, 1992.
8
 N. M. Mia, “Variations of temperature in Bangladesh”, Proceedings of SAARC Seminar on Climate Variability in
the South Asian Region and its Impacts, Dhaka: SAARC Meteorological Research Centre, 2003. 9
 S. Karmakar, and J. Nessa, “Climate Change and its Impacts on Natural Disasters and South-west Monsoon
in Bangladesh and the Bay of Bengal”, Journal of Bangladesh Academy of Sciences, Vol. 21, No. 2, 1997, pp.
127-136.
10 S. Karmakar, “Trends in the annual frequency of cyclonic disturbances and storms in the Bay of Bengal”,
Proceedings of SAARC Seminar on Climate Variability in the South Asian Region and its Impacts, Dhaka: SAARC
Meteorological Research Centre, 2003.
11 M. Z. Haque, H. A. Quayyam, M. M. Hossian and M. S. Islam, “Occurrence of grain sterility in different rice
crops”, in A. K. M .N. Islam, Q. A. Fattah, I. A. Muttaqi and A. Aziz (eds.), Plant Science and Man: Problems and
Prospects, Proceedings of International Botanical Congress, Dhaka: Bangladesh Botanical Society, 10-12
January 1991, pp. 117-124.
12 M. H. K. Chowdhury and S. K. Debsarma, op. cit. 13 Warrick et al., op. cit.

this period, Bangladesh was getting warmer. Since the later part of the last century, there has been, on average, an overall increase in temperature by 0.50 C which was comparable to the observed global warming. Karmaker and Nessa14 studied on climate change and its impacts on natural disasters and southwest-monsoon in Bangladesh and the Bay of Bengal. They found that the decadal mean annual temperature over Bangladesh has shown an increasing trend especially after 1961-1970.

Currently, there is no empirical literature to examine the effect of climatic variables on economic growth of Bangladesh with long term time series data. If the negative linkage between climate change and economic growth is found, it would indicate that Bangladesh could have achieved higher growth than the observed ones. Therefore, it would help the government and non-government actors to decide over their programmes on climate change. Given the context, this present study aims to find out whether there is any linkage of climate change variables on economic growth. This paper would add significant value to this aspect. Thus, the contributions of this paper are as follows. First, it accumulates the literature on the specific macroeconomic variables and climate change. Second, it has used time series econometric techniques for a reasonable period, which provides useful insight for policymaking and undertaking future studies. The paper has been organised as follows. After the introduction, a comprehensive review of literature is presented in section 2. Section 3 describes the methodology of conducting the empirical study. Section 4 describes the results and analysis of the empirical findings. Finally, section 5 concludes the paper.
 

Empirical literature on the effect of climatic variables on economic growth of Bangladesh is sparse. However, the existing geophysical studies mostly concentrate on the change of temperature, rainfall and SLR to present evidences of climate change. Some studies also cover the geophysical variables of climate change and their impacts on GDP growth. Few of them also examine the effect of climate change on sectoral GDP.

Islam15 presented a regional climate model named Providing Regional Climates for Impacts Studies (PRECIS) in generating rainfall scenarios for South Asia. At first, PRECIS generated rainfall scenario is calibrated with ground-based observed rainfall during baseline period (1961-1990) in Bangladesh. The regression coefficients obtained through calibration are utilised for validation of PRECIS simulated rainfall during 2000-2006. PRECIS overestimated rainfall by 12.37, 1.58, 10.81, 4.79 and 13.18 per cent in 2000, 2002, 2003, 2005 and 2006, respectively. It underestimated by

14 S. Karmakar, and J. Nessa, op. cit. 15 M. N. Islam, “Rainfall and Temperature Scenario for Bangladesh”, Open Atmospheric Science Journal, Vol. 3,
No. 1, 2009, pp. 93-103.

 0.64 per cent and 10.84 per cent in 2001 and 2004, respectively. On average, PRECIS overestimated about 4.47 per cent of surface rainfall. Better performance of PRECIS through validation encouraged utilising it in rainfall forecasting for Bangladesh. In the second step, rainfall and temperature forecast for Bangladesh is experimentally obtained for 2010-2020. This work reveals that the PRECIS simulated rainfall and temperature are not directly useful in application purposes. However, after performing calibration, acceptable result obtained in estimating annual rainfall in Bangladesh with correlation coefficient is 0.9. Change of rainfall is forecasted from -0.99 per cent (in 2013) to 5.3 per cent (2018) for Bangladesh from 2010 to 2020.

Basak et al.16 assessed climatic variability based on analysis of historical data of temperature and rainfall recorded at 34 meteorological stations located at seven regions in Bangladesh for the period of 1976-2008. The trend of variation of yearly average maximum temperature has been found to be increasing at a rate of 0.0186°C per year, whereas the rate was 0.0152°C per year for yearly average minimum temperature. Analysis of monthly average maximum temperature also showed increasing trend for all months except January and April. The increasing trend was particularly significant for May to September and for February. Monthly average minimum temperature data also showed increasing trends for all months except January and November. Analysis of rainfall data showed that for a large majority of stations, the total rainfall showed increasing trend for monsoon and post-monsoon seasons, while decreasing trend was observed for the winter; pre-monsoon rainfall did not show any significant change. These observations are particularly significant in the context of Bangladesh where agriculture is heavily dependent on temperature and rainfall patterns.

Rajib et al.17 presented the development of multi-model combination of future surface temperature projections for Bangladesh on monthly basis, for each of the year from 2011 to 2100, using both global and regional climate models. The study demonstrated evidence of increasing temperature levels in Bangladesh from the climate model projections based on observed meteorological data. From the multi-model combination, Regional Climate Model (RCM) and Global Climate Model (GCM) projections of future average temperature change with respect to 1971-2000, it demonstrated that the winter months in Bangladesh might show more warming in future than the monsoon and pre-monsoon months. However, the trend of temperature increase might continue to increase invariably in every month.

Hasan and Rahman18 studied the trend of temperature in Bangladesh. Long term changes in surface air temperature over Bangladesh have been studied using

16 J. K. Basak, R. A. M. Titumir, and N. C. Dey, “Climate Change in Bangladesh: A Historical Analysis of
Temperature and Rainfall Data”, Journal of Environment, Vol. 2, No. 2, 2013, pp. 41-46. 17 M. A. Rajib, M. M. Rahman, and E. A. McBean, “Global Warming in Bangladesh Perspective: Temperature
Projections upto 2100”, Proceedings of the Global Conference on Global Warming, Lisbon, 11-14 July 2011. 18 A. B. M. S. U. Hasan and M. Z. Rahman, “Change in Temperature over Bangladesh Associated with Degrees
of Global Warming”, Asian Journal of Applied Science and Engineering, Vol. 2, No. 2, 2013, pp. 62-75.

 the available historical data collected by the Bangladesh Meteorological Department (BMD). The maximum, minimum and mean monthly temperature data of sixty-three years (1948-2010), collected from 35 stations of BMD located all over Bangladesh, have been used in this study. It has been found that monthly maximum temperature shows a positive trend of increase at a rate of 0.5°C per 100 years. The maximum increase occurred during November at a rate of 2.05°C per 100 years. However, monthly minimum temperature shows statistically significant trend of increase at a rate of 1.40°C per 100 years. Monthly mean temperature shows a positive trend of increase at a rate of 0.8°C per 100 years. It clearly indicates that monthly minimum temperature has increased significantly during the winter season (October to February) over the last sixty-three years. It also reveals that temperature was increasing predominantly over 21 years (1990-2010) than 63 years (1948-2010).

Being a low-lying deltaic country, Bangladesh is exposed to serious consequences of sea-level rise including permanent inundation of huge land masses along the coast line. There is a clear evidence of changing climate in Bangladesh which is causing changes in the precipitation, increasing annual mean temperature and sea-level rise. Shamsuddoha and Chowdhury19 found that during a period from 1961 to 1990, the annual mean temperature increased at the rate of 0.0037°C but the rate was 0.0072 during 1961 to 2000. It means that in the 1990s, annual mean temperature rise was almost double than the previous years. Over the last 100 years, Bangladesh has warmed up by about 0.5°C and 0.5 metre rise of sea-level in the Bay of Bengal. The study reveals that the loss of decreasing productivity, which is assuming 28 to 57 per cent reduction of crop production from the present level by 1 metre sealevel rise, cannot be restored.

Karim and Mimura20 described the impacts of Sea Surface Temperature (SST) rise and sea-level rise (SLR) on cyclonic storm surge flooding in western Bangladesh. A calibrated numerical hydrodynamic model was used to simulate surge wave propagation through the rivers and overland flooding. The model was calibrated with base condition (present climate), and then eight flooding scenarios of plausible future conditions were assessed by considering increased surge heights. Flooded area, flooding depth and surge intrusion length were computed by superimposing the predicted maximum water level information on a Digital Elevation Model (DEM). This analysis showed that for a storm surge under 2°C SST rise and 0.3 metre SLR, flood risk area would be 15.3 per cent greater than the present risk area and depth of flooding would increase by as much as 22.7 per cent within 20 km from the coastline. Within the risk area, the study identified 5,690 km2 land (22 per cent of exposed coast) as a high-risk zone (HRZ) where flooding of depth 1 metre or more might occur, and people should move to nearby cyclone shelters during extreme cyclonic events.

19 M. Shamsuddoha, and R. K. Chowdhury, Climate Change Impact and Disaster Vulnerabilities in the Coastal
Areas of Bangladesh, Dhaka: COAST Trust, 2007. 20 M. F. Karim and N. Mimura, “Impacts of climate change and sea-level rise on cyclonic storm surge floods
in Bangladesh”, Global Environmental Change, Vol.18, No.3, 2008, pp. 490-500.

 Predicted area of HRZ is 1.26 times compared to the currently demarcated HRZ. It was estimated that 320 additional shelters are required to accommodate people in the newly identified HRZ. This information would be of value to policy and decision makers for future shelter planning and designing shelter heights.

Islam et al. 21 examined the vulnerabilities of agriculture in coastal regions of Bangladesh to the different adverse effects of sea-level rise induced hazards, and also identified option for future agricultural adaptations. It reveals that due to sea-level rise, agriculture of the study area has already experienced noticeable adverse impacts especially in terms of area of inundation, salinity intrusion and reduction in crop production. A total of 303 out of 1200 respondents from three coastal villages were randomly interviewed. Descriptive and inferential statistics and logistic regression have been conducted to analyse the data. The study found that the agricultural land, production of crops, local crop varieties, income and employment facilities of the farmers are highly vulnerable to various SLR induced hazards. Selection of various adaptation options, such as control of saline water intrusion into agricultural land, coastal afforestation, cultivation of saline tolerant crops, homestead and floating gardening, embankment cropping and increase of income through alternative livelihoods are emerging need for sustainable coastal agricultural development. Therefore, this paper argued that further development and implementation of such adaptive measures could help minimise vulnerabilities of agriculture in the long run.

World Bank22 assessed the impacts of predicted climate changes on crop yields, agricultural production and GDP by factoring in all climate impacts, viz. CO2 emission, temperature and precipitation changes, flooding, and sea-level rise. It reveals that cumulative loss of rice production would be 80 million tons, which is nearly 3.9 per cent per year over 2005-50. Agricultural GDP is projected to decline annually by 3.1 per cent, i.e., US$ 36 billion value-added in agriculture is lost, while the loss of total GDP is estimated to be US$ 129 billion over this period.

Ahmed and Suphachalasai23used global PAGE model and Bangladeshspecific Computable General Equilibrium (CGE) model to assess the effects of climate change on GDP. It reveals that by 2050, annual GDP losses under the business as usual scenario is projected to be only 2 per cent while due to inaction (i.e., no action is taken for adaptation), the average total economic losses are estimated to be 9.4 per cent.

The above discussion reveals that the studies on the effects of climate change of Bangladesh are based on global geophysical and CGE models. None of them use

21 M. A. Islam, P. K. Shitangsu and M. Z. Hassan, “Agricultural vulnerability in Bangladesh to climate change
induced sea level rise and options for adaptation: a study of a coastal Upazila”, Journal of Agriculture and
Environment for International Development, Vol. 109, No. 1, 2015, pp. 19-39. 22 World Bank, The Economics of Adaptation to Climate Change, Washington, DC: World Bank, 2010. 23 M. Ahmed and S. Suphachalasai, Assessing the Costs of Climate Change and Adaptation in South Asia,
Manila: Asian Development Bank, 2014.

 
 long-term time series data on major climatic variables, viz., greenhouse gas emission, rainfall and temperature to examine their effects on long term GDP growth of the country. The CGE models are primarily based on Social Accounting Matrix (SAM) for a specific point of time, which is unable to capture the dynamics of co-movement of GDP and geophysical data. The present paper is an attempt to address these limitations. 

The objective of the present paper is to empirically examine the effects of climatic variables on economic growth in Bangladesh. Here, annual change of GDP is taken as economic growth variable, while greenhouse gas emission, rainfall and temperature are taken as climate change variables. Thus, the econometric model used for the paper is as follows:

Time-series econometrics requires an analysis of the time-series properties and paths of the economic variables in a regression equation before estimation of the model to consider if any relationship can be estimated for the model. To find out the long-term relationship among the variables, it is essential to have non-stationary at levels and stationary at first differences.

structural breaks in the time series of the variables. If any result is inferred from them, it will be inaccurate. For example, in the event of a standard ADF regression:

In the case of this regression model that involves time series data, there may be a structural change in the relationship between the dependent and independent variables. Structural change implies that the values of the parameters of the model do not remain the same through the entire period. Sometimes the structural change may be due to external forces or to unobservable factors, which exert individual or combined impact on the country.

24 P. Perron, “The great crash, the oil price shocks, and the unit root hypothesis”, Econometrica, Vol. 57, No.
6, 1989, pp. 1361-1401; and P. Perron, “Further evidence on breaking trend functions in macroeconomic
variables”, Journal of Econometrics, Vol. 80, No. 2, 1997, pp. 355-385. 25 E. Zivot and D. W. K. Andrews, “Further evidence on the great crash, the oil-price shock, and the unit-root
hypothesis”, Journal of Business and Economic Statistics, Vol. 10, No. 3, 1992, pp. 251-270. 26 P. Perron and T. J. Vogelsang, “Nonstationarity and level shifts with an application to purchasing power
parity”, Journal of Business and Economic Statistics, Vol. 10, No. 3, pp. 301-320. 27 P. C. B. Phillips and P. Perron, “Testing for a unit root in time series regression”, Boimetrika, Vol. 75, No. 2,
1988, pp. 335-346.
28 Ibid.

 A traditional approach of examining structural change is to adopt the Chow test.29 This test assesses whether the coefficients in two linear regressions on different data sets are equal. The fundamental problem associated with this test is the arbitrariness of a researcher in bifurcating the series to examine whether two sets have different intercepts, slopes or both. Recent developments in the literature of structural break suggest that it is important to determine (i) whether there is structural break in the entire series or not by taking into account the time series properties of the data, and (ii) single or multiple breakpoint exists in a series, and (iii) the year of the structural break.

Elliott and Müller30 devised a test to examine whether there is a structural break in a series. This test allows for many or relatively few breaks, clustered breaks, regularly occurring breaks, or smooth transitions to changes in the regression coefficients. The test distinguishes the null hypothesis of a stable regression model

29 G. C. Chow, “Tests of Equality between Sets of Coefficients in Two Linear Regressions”, Econometrica, Vol.
28, No. 3, 1960, pp. 591-605.
30 G. Elliott and U. K. Müller, “Efficient Tests for General Persistent Time Variation in Regression Coefficients”,
Review of Economic Studies, Vol. 73, No. 4, 2006, pp. 907-940.

Zivot and Andrews31 proceed with three models to test for a unit root: Model 4, which permits a one-time change in the level of the series; Model 5, which allows for a one-time change in the slope of the trend function, and Model 6, which combines one-time change in the level and the slope of the trend function of the series. To test for a unit root against the alternative of a one-time structural break, Zivot and Andrews use the following regression equations corresponding to the above three models:

The null hypothesis in all the three models is α=0, which implies that the series {yt } contains a unit root with a drift that excludes any structural break, while the alternative hypothesis α<0 implies that the series is a trend-stationary process with a one-time break occurring at an unknown point in time. The Zivot-Andrews test considers every point as a potential break-date (TB) and runs a regression for every possible break-date sequentially. From amongst all possible break-points (TB), the procedure selects as its choice of break-date (T B ) the date which minimises the onesided t-statistic for testing αˆ =(α-1) =1. According to Zivot and Andrews, the presence of the end points causes the asymptotic distribution of the statistics to diverge towards infinity. Therefore, some regions must be chosen such that the end points of the sample are not included. They suggest the ‘trimming region’ be specified as (0.15T, 0.85T).

Also, it is imperative to examine whether there is more than one breaks. Following Clemente et al.32, the null hypothesis can be tested as

31 Zivot and Andrews, op. cit. 32 J. Clemente, A. Montañés and M. Reyes, “Testing for a unit root in variables with a double change in the
mean”, Economics Letters, Vol. 59, No. 2, 1998, pp. 175-182.

The econometric equation (1) is based on the presumption that variability in climate change indicators can sufficiently explain the variability in GDP growth of Bangladesh without considering the other factors, such as capital and labour. The paper avoids the production function approach, which is complicated due to lack of reliable data on the stock of capital. On the other hand, continuous annual data on labour force is also unavailable in the secondary sources. Therefore, the present exercise depends on the explanatory power of the climate change indicators based on the scores of the goodness of fit. 

Before conducting the main multivariate analysis, stationarity tests have been performed to understand whether the variables included in the model are stationary (random) or not. If the variables are non-stationary at level, then the first difference needs to be taken and then again the tests have been conducted for stationarity. Afterwards, multivariate time series method should be applied to estimate the Equation (1). The results of stationarity test, viz. ADF and PP reported in Table 1, where both tests indicate that only total greenhouse gas emissions and temperature are found to be stationary.

33 Available at https://data.worldbank.org/indicator?tab=all, accessed on 02 September 2017.
34 Available at http://sdwebx.worldbank.org/climateportal/index.cfm?page=country_historical_
climate&ThisCCode=BGD, accessed on 02 September 2017.

37 B. Inder, “Estimating Long-Run Relationships in Economics: A Comparison of Different Approaches”,
Journal of Econometrics, Vol. 57, No. 1-3, 1993, pp. 53-68. 38 M. H. Pesaran and Y. Shin, “An Autoregressive Distributed Lag Modelling Approach to Cointegration
Analysis”, DAE Working Paper 9514, Cambridge: University of Cambridge, 1995. 39 M. H. Pesaran, and Y. Shin, “An Autoregressive Distributed Lag Modelling Approach to Cointegration
Analysis”, in S. Strom, (ed.), Econometrics and Economic Theory in the 20th Century: The Ragnar Frisch
Centennial Symposium, Cambridge, UK: Cambridge University Press, 1999. 40 Pesaran et al., op. cit. 41 For a succinct exposition of ARDL models in the context of cointegration, see U. Hassler, and J. Wolters,
“Autoregressive Distributed Lag Models and Cointegration”, Allgemeines Statistisches Archiv, Vol. 90, No. 1,
2006, pp. 59-74; and U. Hassler, and J. Wolters, “Autoregressive Distributed Lag Models and Cointegration”,
Working Paper No. 2005/22, Berlin: Freie Universitaet, 2005.
42 Pesaran et al., op. cit.