Abstract
This paper investigates the vulnerability of households to climatic disasters in the low-lying atoll nation of Tuvalu. Small Island Developing States, particularly the atoll nations, are the most vulnerable to climatic change, and in particular to sea-level rise and its associated risks. Using the most recent household surveys available, we construct poverty and hardship profiles for households on the different islands of Tuvalu, and combine these with geographic and topographic information to assess the exposure differentials among different groups using spatial econometric models. Besides the observation that poor households are more vulnerable to negative shocks because they lack the resources to respond, we also find that they are also more likely to reside in areas highly exposed to disasters (closer to the coasts and at lower elevation) and have less ability to migrate (between and within the islands).
Similar content being viewed by others
Notes
See the map of Tuvalu in Appendix 1.
According to the 2012 Census, only 17.7% of the people living on Funafuti are local Funafuti people, while the rest are without land ownership, and are renting houses from the locals. Based on the 2012 Census, 84.5% of rental houses in Tuvalu are on Funafuti.
Abject poverty was first coined by the United Nations (UN) in 1995 referring to the form of severe deprivation of basic needs and services that is faced by many individuals or households particularly in some least developed countries (LDC). Specifically, this form of poverty is characterized by a severe lack of food access, safe drinking water, basic sanitation, healthcare access, shelter, educational resources and access to information. Although Tuvalu is classified as an LDC by UN standards, the presence of ‘abject poverty’ there is rare.
Abbott and Pollard (2004) emphasize that ‘hardship’ is a more acceptable terminology. The World Bank (2014) also argues that “the label of poverty is considered culturally inappropriate because it is viewed as implying a failure of traditional, community-based safety nets”. Below we used hardship and poverty interchangeably, to mean “living with less than expected to meet both required food consumption and non-food essentials”.
In practice, this includes all those below a threshold that is 10% higher than the Basic Needs Poverty Line (BNPL). The BNPL was calculated based on Ravallion (1998), and in line with Tuvalu Statistics Office’s policy. The hardship threshold is the sum of the food threshold and the non-food threshold. The food component was calculated from a basket of essential basic food items that is estimated to be equivalent to the widely used nutritional requirement for good health of 2100 calories per person per day suggested by the Food and Agricultural Organisation (FAO) of the United Nations. The non-food threshold is the average non-food expenditure by households in the lowest 3 deciles. The non-food threshold is calculated differently for rural and urban (Funafuti) areas as the non-food expenditure, especially housing, is quite different between the regions. Expenditure is derived as the sum of food expenditure and non-food expenditure.
USD refers to United States Dollars while AUD refers to Australian Dollars. The AUD is the legal tender in Tuvalu.
Other possibilities can be traced to the availability of more employment in the rural sector from the island council, clinics, island development projects, and small-scale businesses (after the collapse of the Tuvalu Cooperative Society). Informal work allocation in the outer islands is also more equally distributed amongst families and may not rely on educational qualification as much as in the urban area of Funafuti.
Authors’ calculations from digitized maps.
Niulakita, the smallest island, was excluded from the household survey.
The assumed 16,000 l of water capacity storage threshold used is the median of household water storage available for all households surveyed in the 2010 HIES. This is assumed to be sufficient if water is used efficiently.
Borrow pits (we will refer to it as “pits” onward) were created by digging/borrowing of soil from parts of the island of Fongafale (Funafuti), by the American military during World War II, to construct the airplane runway. We used 20 m to the pits as an indication of those living beside pits, based on the assumption that during King tides, a house within that range will most likely be flooded. This problem has been mostly solved in 2015 by the Tuvalu Borrow Pits Remediation (BPR) project funded under the New Zealand Aid Programme, where ten borrow pits on Fongafale island were filled with sand except for Tafua pond to the northeastern side of the airstrip, which is a natural pond. It is yet to be seen if we will observe any future flooding in these filled up pits.
This was obvious since Nui and Nukulaelae were flooded during the 2015 Cyclone Pam. However, the three islands Nanumaga, Niutao and Vaitupu which have higher elevation did not experience flooding during Cyclone Pam, but only storm surge and coastal intrusion of sea waves from the western side. All islands build their harbour and houses on the western side of the island away from easterly winds, but a cyclone that strikes from the west side will badly hit most islands without lagoons and islets on the west as shields.
Nanumea has a few wells that enable access to brackish freshwater lens (Johnston et al. 2012).
ArcGIS was used for geo-coding of locations (households, schools, hospitals/clinics, etc.) creating a digitized map for all islands and islets in Tuvalu. These were then used in STATA for the empirical analysis.
LeSage and Pace (2009) state that “omitted variables may easily arise in spatial modelling because unobservable factors such as location amenities, highway accessibility, or neighbourhood prestige may exert an influence on the dependent variable. It is unlikely that explanatory variables are readily available to capture these types of latent influences”.
Elhorst (2014) shows the relationship between the different spatial dependence models for cross-section data.
For our case, every household has at least one neighbour. Therefore, each row sums up to 1.
The smallest island in Tuvalu with only four households (based on the 2012 Census).
The Moran’s I test statistic is used to test if the data have spatial dependence. According to Olivia et al. (2009), the Moran’s I for a row-standardized spatial matrix where e is a vector of OLS residuals and W is the spatial weight matrix, asymptotically normally distributed with an expected value of − 1/(N − 1) and its statistical significance can be evaluated from a standardized normal table. It is expressed as I = elWe/ele. The Lagrange multiplier (LM) tests for SEM and SAR whether (λ = 0) and (ρ = 0). The robust LM tests were also developed by Anselin et al. (1996) to cater for the presence of both SEM and SAR (which is a weakness for the LM test as LMλ and LMρ have power against the other alternative). Olivia et al. (2009) provides more detailed discussion of the tests.
We classified Funafuti as the urban and outer islands as rural since Funafuti is the capital and where the central government, commerce, main hospital, seaport and airport are located.
The elevation projected from the Digital Elevation Model (DEM) may differ marginally with land elevation. Variations between elevation and mean sea level (MSL) are explained in http://www.esri.com/news/arcuser/0703/geoid1of3.html. Note that the houseowner coefficient is only significant for the rural sample.
Moran’s I test is highly significant at 1% level, indicating spatial autocorrelation.
Note that the asymptotic theory of spatial models for limited dependent variables has only been developing recently, so we are uncertain about the robustness of these results (e.g., Qu and Lee 2012).
The Hausman panel test indicated a strong preference for the fixed-effects (FE) model over random-effects (RE); while the Breusch–Pagan indicated the panel models are preferable to the OLS estimation. The FE model for N observations (i = 1,…,N) and T time periods (t = 1,…,T) is yit = αi + Xit+uit where yit is the dependent variable observed for individual i at time t, αi is the unobserved time-invariant individual effect, Xit is the time-invariant 1 × k regressor matrix, and uit is the error term. However, we present all three specifications in the appendix for comparison.
Note that there are no time-invariant variables on the right-hand side, e.g., variables such as “sex” and “ethnic” of the household head are not time-invariant variables as they change overtime in our data depending on who is the household head at a specified time. The identity of the household head present varies over time.
Authors’ calculations from data. These reflect the overall number of households that migrated between the islands and those who are non-poor.
All outer islands have primary schools, clinics and police stations. The main boarding secondary school is located in the outer island on Vaitupu.
References
Abbott D, Pollard S (2004) Hardship and poverty in the Pacific. Asian Development Bank, Philippines
ADB (2007) 2006 Tuvalu Economic Report: from plan to action. Asian Development Bank, Philippines
Anselin L (1988) Spatial econometrics: methods and models. Kluwer, Dordrecht
Anselin L, Bera AK, Florax R, Yoon MJ (1996) Simple diagnostic tests for spatial dependence. Reg Sci Urban Econ 26:77–104
Barnett J (2005) Titanic states? Impacts and responses to climate change in the Pacific Islands. J Int Aff 59:203–219
Becker M, Meyssignac B, Letetrel C, Llovel W, Cazenave A, Delcroix T (2012) Sea level variations at tropical Pacific islands since 1950. Global Planet Change 80–81:85–98. https://doi.org/10.1016/j.gloplacha.2011.09.004
Black R, Collyer M (2014) Populations’ trapped’ at times of crisis. Forced Migr Rev 45:52
Cabezon E, Hunter L, Tumbarello P, Washimi K, Wu Y (2015) Enhancing macroeconomic resilience to natural disasters and climate change in the small states of the Pacific. IMF Working Paper, WP/15/125
Connell J (2003) Losing ground? Tuvalu, the greenhouse effect and the garbage can. Asia Pac Viewpoint 44(2):89–107
Dasgupta A, Baschieri A (2010) Vulnerability to climate change in rural Ghana: mainstreaming climate change in poverty-reduction strategies. J Int Dev 22(6):803–820. https://doi.org/10.1002/jid.1666
Dercon S (2005) Risk, poverty and vulnerability in Africa. J Afr Econ 14(4):483–488. https://doi.org/10.1093/jae/eji023
Dutta I, Foster J, Mishra A (2011) On measuring vulnerability to poverty. Soc Choice Welfare 37(4):743–761. https://doi.org/10.1007/s00355-011-0570-1
Échevin D (2014) Characterising vulnerability to poverty in rural Haiti: a multilevel decomposition approach. J Agric Econ 65(1):131–150. https://doi.org/10.1111/1477-9552.12017
Elhorst JP (2014) Spatial econometrics. Springer, Berlin. http://link.springer.com/10.1007/978-3-642-40340-8
Gibson J, McKenzie D (2007) Using the global positioning system in household surveys for better economics and better policy. World Bank Policy Research Working Paper (4195). http://papers.ssrn.com/sol3/papers.cfm?abstract_id=979667
Gibson J, Rozelle S (2002) Poverty and access to infrastructure in Papua New Guinea. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=334140
Hallegatte S, Bangalore M, Bonzanigo L, Fay M, Kane T, Narloch U, Vogt-Schilb A (2015) Shock waves: managing the impacts of climate change on poverty. The World Bank. http://elibrary.worldbank.org/doi/book/10.1596/978-1-4648-0673-5
Haughton J, Khandker S (2009) Handbook on poverty and inequality. The World Bank. http://elibrary.worldbank.org/doi/book/10.1596/978-0-8213-7613-3
Jalan J, Ravallion M (1998) Are there dynamic gains from a poor-area development program? J Public Econ 67(1):65–85
Jalan J, Ravallion M (2002) Geographic poverty traps? A micro model of consumption growth in rural China. J Appl Econ 17(4):329–346. https://doi.org/10.1002/jae.645
Jha R, Dang T (2010) Vulnerability to poverty in Papua New Guinea in 1996. Asian Econ J 24(3):235–251. https://doi.org/10.1111/j.1467-8381.2010.02038.x
Jha R, Dang T, Sharma KL (2009) Vulnerability to poverty in Fiji. Int J Appl Econ Quant Stud 6(1):43–60. http://www.usc.es/~economet/ijaeqs.htm
Johnston BR, Hiwasaki L, Klaver IJ, Ramos Castillo A, Strang V (eds). (2012) Water, cultural diversity, and global environmental change. Springer, Dordrecht. https://doi.org/10.1007/978-94-007-1774-9
LeSage J, Pace RK (2009) Introduction to spatial econometrics. Taylor and Francis Group, Boca Raton
Mechler R, Schinko T (2016) Identifying the policy space for climate loss and damage. Science 354(6310):290–292
Mortreux C, Barnett J (2009) Climate change, migration, and adaptation in Funafuti, Tuvalu. Global Environ Change 19:105–112
Noy I (2015) Disasters in the Pacific: an overview of economic and fiscal issues (Pacific economic monitor). Asian Development Bank
Noy I (2016) Natural disasters in the Pacific Island Countries: new measurements of impacts. Nat Hazards 84(S1):7–18. https://doi.org/10.1007/s11069-015-1957-6
Noy I (2017) To leave or not to leave? Climate change, exit, and voice on a Pacific Island. CESifo Econ Stud. https://doi.org/10.1093/cesifo/ifx004
Olivia S, Gibson J, Smith A, Rozelle S, Deng X (2009) An empirical evaluation of poverty mapping methodology: explicitly spatial versus implicitly spatial approach. In: A contributed paper to the Australian Agricultural and Resource Economics Society’s Annual Conference, Cairns. http://ageconsearch.umn.edu/bitstream/47651/2/47651.pdf
Olivia S, Gibson J, Rozelle S, Huang J, Deng X (2011) Mapping poverty in rural China: how much does the environment matter? Environ Dev Econ 16(02):129–153. https://doi.org/10.1017/S1355770X10000513
Pradhan M, Ravallion M (2000) Measuring poverty using qualitative perceptions of consumption adequacy. Rev Econ Stat 82(3):462–471
Qu X, Lee L (2012) LM tests for spatial correlation in spatial models with limited dependent variables. Reg Sci Urban Econ 42(3):430–445. https://doi.org/10.1016/j.regsciurbeco.2011.11.001
Ravallion M (1996a) How well can method substitute for data? Five experiments in poverty analysis. World Bank Res Obs 11(2):199–221
Ravallion M (1996b) Issues in measuring and modelling poverty. Econ J 106(438):1328. https://doi.org/10.2307/2235525
Ravallion M (1998) Poverty lines in theory and practice. World Bank, Washington, DC
Smith R, McNamara KE (2015) Future migrations from Tuvalu and Kiribati: exploring government, civil society and donor perceptions. Clim Dev 7(1):47–59. https://doi.org/10.1080/17565529.2014.900603
World Bank (2007) Poverty and the environment: understanding linkages at the household level. The World Bank. http://elibrary.worldbank.org/doi/book/10.1596/978-0-8213-7223-4
World Bank (2013) World Development Report 2014: Risk and Opportunity - Managing Risk for Development. The World Bank, Washington DC
World Bank (2014) Hardship and Vulnerability in the Pacific Island Countries. The World Bank, Washington DC
World Bank (2016) Climate and disaster resilience. Pacific Possible, Washington DC, p 70
Yamano H, Kayanne H, Yamaguchi T, Kuwahara Y, Yokoki H, Shimazaki H, Chikamori M (2007) Atoll island vulnerability to flooding and inundation revealed by historical reconstruction: Fongafale Islet, Funafuti Atoll, Tuvalu. Global Planet Change 57(3–4):407–416. https://doi.org/10.1016/j.gloplacha.2007.02.007
Author information
Authors and Affiliations
Corresponding author
About this article
Cite this article
Taupo, T., Cuffe, H. & Noy, I. Household vulnerability on the frontline of climate change: the Pacific atoll nation of Tuvalu. Environ Econ Policy Stud 20, 705–739 (2018). https://doi.org/10.1007/s10018-018-0212-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10018-018-0212-2