Methods for determining how much to spend on flood prevention: an empirical case study in the Philippines

Fernandez, Cheryl Joy (2016) Methods for determining how much to spend on flood prevention: an empirical case study in the Philippines. PhD thesis, James Cook University.

PDF (Thesis)
Download (4MB) | Preview
View at Publisher Website:


Flood prevention/mitigation is an important component in disaster management and plays a key role in the well-being of communities all over the world. Typhoons and floods are predicted to become more severe and to occur more frequently in many areas (IPCC, 2014; UNU-EHE, 2014). Moreover, these events are likely to impact more people, particularly in less affluent countries, where populations are increasing in urban, often low-lying areas. Non-excludability and non-rivalry of (most) flood prevention strategies provide a disincentive for the private sector to provide them. Governments have planned and executed many strategies to prevent flood impacts – although budgets are limited. The central question addressed in this dissertation is thus 'how much should be spent on flood prevention/mitigation programs.' I use three economic methods to address that question, each with its own aim, namely to:

1) Determine how much damage (to households) could be avoided if one were able to prevent flood (flood damage assessment);

2) Determine how much households are willing to pay to avoid future flood damages (contingent valuation or CV method); and

3) Determine the impact of flood damages on life satisfaction (LS), and the amount of income that would need to be paid to flood victims to 'compensate' them (i.e. to hold LS constant) for the flood damage (life satisfaction or LS method).

The Metropolitan Iloilo (MI) of the Philippines is the case study region. This region is an ideal one in which to undertake the research because the country is one of the most vulnerable countries in the world in terms of disasters (Yusuf & Francisco, 2009; UNU-EHS, 2014), yet expenditure on flood prevention programs in regional areas is inadequate (Benson, 2009; Lasco & Delfino, 2010; Iloilo City Government, 2010). The area also has a varied cross-section of barangays (communities/villages) and households, located near and far from rivers/creeks and with different sociodemographic characteristics. It is thus a good one in which to consider the differential impact of floods on different demographic groups.

First, I studied related literature in order to understand various ways of thinking about the benefits of flood control – formally, this lead me to the literature on methodologies for assessing 'value' for flood prevention, including that associated with contingent valuation and 'willingness to pay' (WTP); and that associated with 'life satisfaction' (LS). Following this, I ran eight focus group discussions in the case study area, to determine factors that were regionally relevant to local individuals/households. Inputs from the discussion were incorporated in the pre-test survey, in which I designed my hypothetical scenario for assessment in my CV study (e.g. payment vehicle, frequency of payments, amounts in the payment cards), determined how best to phrase particular questions, and identified an appropriate data collection method. During July and August 2013, I conducted face-to-face interviews with the help of eight hired and trained enumerators in six towns and one city of the MI region, capturing the views of 600 respondents.

Prior studies that have estimated flood damages rarely include indirect and/or intangible flood damages and their focus is usually on a single flood event and/or a single year of damages. I sought to better understand flood impacts over a longer period of time, collecting data about flood damages over a five-year period. I found that the 'average' households incurred around ₱1,800 to ₱3,700 (US$39 to US$82) worth of flood damages during that period – most damage (>60%) being associated with damage to property, the rest associated with loss of employment opportunities and other damages (e.g. medical expenses).

This gives an indication of how much damage could be avoided should floods cease to impact people in this region. It does not, however, provide an estimate of the welfare costs (and thus, economic benefit) of flood prevention because the estimate does not include a measure of the cost of 'intangibles' (such as trauma) and does not allow for the fact that individuals are (finances permitting) able to undertake at least some private mitigation activities (such as elevating houses, or moving to less flood prone areas).

In theory, the CV method is able to get around those problems, generating more accurate estimates of the welfare costs of flooding – although there are numerous interrelated variables that affect WTP. I thus designed a flood valuation model that examines these relationships using a two-stage interval regression. I found that the 'average' household was only willing to pay around ₱108 (US$2.4) per year to prevent any future flood impacts. Although the design of the survey and methods used to analyse data sought to minimise some of the problems commonly associated with CV (including, but not limited to survey and hypothetical bias), the WTP estimates, at less than 2% of reported damage, were much lower than expected. The big difference between WTP and reported damages could indicate that: (a) damages were grossly over reported; (b) intangible costs are negligible; (c) respondents felt that they had many opportunities to mitigate flood impacts privately; and/or; (d) that respondents were constrained by ability to pay. But the differences might also be attributable to the fact that the CV method can only generate accurate estimates of welfare costs if respondents are able to accurately predict their utility in the future – with and without flood prevention. This requires respondents to have perfect information and respond to questions about WTP truthfully and rationally.

The last method I used (the LS approach), does not require this to be so, and – like the CV method, is able to capture intangibles and also people's private ability to undertake flood mitigation activities. I used the method to estimate the 'value' of flood prevention by using coefficients from a regression model to calculate the amount of income that would need to be paid to respondents to 'compensate' them for losses in their level of life satisfaction associated with (self-reported) flood damages. In previous applications of the LS model, most researchers have used secondary data to capture differences in the environment (e.g. national pollution levels, flood depth within a region), which is then compared to individual-level life satisfaction scores. I addressed this potential problem of heterogeneity (whereby individuals within the region might experience different flood impacts), using self-reported damage assessments instead. I found that, on average, households would need to be compensated by around ₱1,515 (US$34) per year for reported flood damages – an amount that is approximately equal to the average annual flood damage reported.

While the LS method does not require rationality and perfect information, the method (like all valuation methods) is not problem-free: social desirability bias, context effects and endogeneity are all real and present issues.

From my results, I implied that the LS approach seems to be a better option for estimating the 'value' for flood prevention, if and only if, endogeneity of income has minimal and/or insignificant effect on the 'income compensation' estimate. In terms of applying the CV method to flood valuation, I find room of improvement. If my findings are applicative to other flood-vulnerable and/or urbanised areas, then the fact that individuals are unlikely to be able to predict their utilities and their level of flood risks, limits the applicability of the CV method. Improving our understanding of these important issues will present an important step forward in assessing the 'value' of, and thus determining the 'optimal' level of, government funded flood mitigation services – eventually enhancing the well-being of communities.

Item ID: 44640
Item Type: Thesis (PhD)
Keywords: disaster management; disaster preparedness; disaster prevention; flood control; flood damage; flood mitigation; flood prevention; flood proofing; flood protection; flooding; Philippines
Related URLs:
Additional Information:

Appendix E (administrative documentation) and appendix G (permissions) are not available through this repository.

Publications arising from this thesis are available from the Related URLs field. The publications are:

Fernandez, Cheryl Joy, Welters, Riccardo, and Stoeckl, Natalie (2014) How does flooding influence life satisfaction of residents in the Metro-Iloilo region, Philippines. In: Proceedings of International Symposium on Business and Management, pp. 586-603. From: ISBM 2014: International Symposium on Business and Management, 2-4 April 2014, Nagoya, Japan.

Date Deposited: 11 Aug 2016 03:56
FoR Codes: 14 ECONOMICS > 1402 Applied Economics > 140205 Environment and Resource Economics @ 33%
14 ECONOMICS > 1402 Applied Economics > 140214 Public Economics- Publically Provided Goods @ 33%
14 ECONOMICS > 1402 Applied Economics > 140219 Welfare Economics @ 34%
SEO Codes: 91 ECONOMIC FRAMEWORK > 9102 Microeconomics > 910209 Preference, Behaviour and Welfare @ 60%
96 ENVIRONMENT > 9606 Environmental and Natural Resource Evaluation > 960601 Economic Incentives for Environmental Protection @ 40%
Downloads: Total: 1567
Last 12 Months: 11
More Statistics

Actions (Repository Staff Only)

Item Control Page Item Control Page