BORIS Theses

BORIS Theses
Bern Open Repository and Information System

Danger from Above - The Economic Impacts of Floods

Collalti, Dino (2023). Danger from Above - The Economic Impacts of Floods. (Thesis). Universität Bern, Bern

23collalti_d.pdf - Thesis
Available under License Creative Commons: Attribution-Noncommercial (CC-BY-NC 4.0).

Download (6MB) | Preview


The scientific evidence decisively attributes natural hazards’ increasing intensity and frequency to global warming. The Synthesis Report of the 6th IPCC Assessment Report, which has just been finalized in March 2023, states: In the near term, every region in the world is projected to face further increases in climate hazards (medium to high confidence, depending on region and hazard), increasing multiple risks to ecosystems and humans (very high confidence). These risks of climate hazards include heat-related human mortality, biodiversity loss, and the spread of diseases, to name a few. Concerning the hazards themselves, floods, landslides, and water availability have the potential to lead to severe consequences for people, infrastructure, and the economy. Their study is highly relevant because climate hazards constitute a causal channel through which anthropogenic climate change influences the economy. For instance, estimates of their social and economic costs provide evidence for setting the social cost of carbon emissions. Given the projections from the natural sciences, it is striking that the discipline of economics has made little effort to assess the economic impact of many hazards. This thesis comprises four papers investigating the economic impacts of floods from extreme rainfall in Central America and the Caribbean. Chapter One provides empirical evidence that a non-negligible part of hurricanes’ direct damages can be attributed to extreme rainfall. Chapter Two then shifts the focus from large-scale hurricanes to small-scale flash floods. It develops a statistical method to detect potential flash flood events from satellite rainfall data. In Chapters Three and Four, I then use this method to study the economic impacts of flash floods consistently across Central America and the Caribbean. Specifically, Chapter Three looks at the dynamics in night light activity following a flood, whereas Chapter Four analyzes the flood’s impact on establishments. In the first chapter of my thesis, co-authored with Eric Strobl and published in Natural Hazards, we revisit the common notion of modeling the impacts of hurricanes solely via their local wind speed. In recent years, some of the most destructive hurricanes, such as Morakot in Taiwan 2009, Harvey in Texas 2017, and Idai in Mozambique 2019, are characterized not by particularly strong wind but by a tremendous amount of rainfall. Relying on a model and a damage function that ignores rainfall and subsequent flooding is thus bound to yield biased results. Furthermore, there is a consensus that rainfall-heavy hurricanes will likely become more common with global warming (Grossmann and Morgan, 2011; Walsh et al., 2016; Knutson et al., 2019). A priori, it is not clear how to best assess the rainfall flood risk and relate it to damages. For instance, adequate rainfall measurements at a high spatial resolution during a hurricane are generally unavailable. To this end, we link remote sensing precipitation data to regional damage data for five hurricanes in Jamaica from 2001 to 2012. We find that the maximum rainfall intensity during a hurricane in a region is a significant determinant of economic damages, explaining much of the variation. Next, we use extreme value modeling of precipitation and combine the return periods with an estimated damage function and satellite-derived night light intensity to assess the local risk in monetary terms. This allows us to quantify the monetary risk for different horizons. For instance, the damage risk for a 20-year rainfall event in Jamaica is estimated to be at least 238 million USD, i.e., about 1.5% of Jamaica’s yearly GDP. In my second chapter, co-authored with Nekeisha Spencer and Eric Strobl, we set out to study the rainfall conditions that trigger floods, particularly flash floods. These are a type of highly localized flood that is directly caused by short but intense episodes of rainfall (Borga et al., 2007). Compared to river floods that require catchment-type hydrological modeling, they can occur almost anywhere given intense rainfall and are, as such, one of the most common natural hazards.¹ The Caribbean is especially at risk from flash floods since urbanization is often unregulated and soil degradation common such that excess rainfall can not run off quickly (Gencer, 2013; Pinos and Quesada-Rom´an, 2021). We set out to assess rainfall characteristics of previous flash flood events to create a classification method above what threshold a rainfall event likely causes a flash flood. For this, we gather information on all 93 confirmed flash floods in Jamaica from 2001 to 2018. We link these to remote sensing precipitation data, with which we further construct the location-specific yearly maximum rainfall events. By employing the copula method to create intensity-duration-frequency (IDF) curves, we model the intensity and duration of the annual maximum events separately and flexibly from their respective marginal distribution. The estimated Normal copula has Weibull and generalized extreme value (GEV) marginals for duration and intensity, respectively. The parametric IDF curve with an associated return period of 2¹⁄₆ years is then the optimal threshold for flash flood event classification. The simple nature of connecting the copula method for IDF curves with a classification for flash floods potentially opens up many applications in parametric insurance programs, regional risk mapping, and hazard warning systems. I then investigate the local dynamic economic response after a flash flood in the third chapter. The idea that a natural disaster influences economic performance and organization not only on impact but over time has entertained several empirical studies that try to estimate these. These include the study of tropical storms (Nordhaus, 2010; Strobl, 2012; Hsiang and Jina, 2014), earthquakes (Barone and Mocetti, 2014; Fabian, Lessmann, and Sofke, 2019), droughts (Barrios, Bertinelli, and Strobl, 2010; Hornbeck, 2012) and floods (Loayza et al., 2012; Kocornik-Mina et al., 2020). The studies on floods focus on large-scale disasters and reports in global databases that neglect localized flash flood events. Conceptually, a natural hazard causes destruction upon impact that might depress local economic activity, cause its re-location or spur innovation and growth in the future. Given the high frequency of extreme rainfall events in many developing countries, they could be a primary mechanism for how climate and environmental degradation impacts their economic development. I employ the method from Chapter Two to construct a high-resolution physically based indicator of flash flood occurrence for Central America and the Caribbean. With that, I estimate the local economic response to an event via changes in local night light emissions from satellite data. After accounting for tropical cyclone activity, flash floods have a delayed, short-term negative effect on economic activity. In countries with a low to medium human development index (HDI), the average effect can be up to −5.7%. Back-of-the-envelope calculations suggest that, due to their high frequency, flash floods in these countries cause GDP growth to fall by −0.84 percentage points. Countries with higher development appear to be only marginally affected. I also find evidence for negative spatial spillovers from floods in neighboring locations. In my fourth chapter, I shift my focus from the dynamic perspective to the economic agents. With a limited capacity to adapt to climate change, it is important to study the mechanisms through which climate change affects the economy to guide policymakers (Mendelsohn, 2012). In the case of flash floods, such mechanisms likely include local establishments as the main economic agents. I again use the method from Chapter Two and link the indicator of flash flood occurrence to the Worldbank Enterprise Surveys for Central America and the Caribbean. They uniquely provide a large number of consistent, geo-located surveys across the study region. After controlling for the location-specific extreme rainfall history, I find that a flash flood significantly decreases sales and the number of employees but increases capital productivity. The negative effects are driven by establishments for which financial market access is an obstacle, whereas the increased capital productivity occurs in establishments with sufficient financial market access. Flash floods similarly affect different industries, with the notable exception of the construction sector. The construction sector is not negatively affected in terms of output and employment. My results suggest that flash floods negatively impact firms and that their increase due to global warming will likely influence economic activity. Improving financial market access appears to be an effective adaptation strategy to increase establishments’ resilience. ¹For instance, according to the Centre for Research on the Epidemiology of Disasters (CRED) Emergency Events Database (EM-DAT), the number of affected people in 2022 by flash floods (0.9 M) is significantly more than river floods (0.1 M) or forest fires (0.03 M).

Item Type: Thesis
Dissertation Type: Cumulative
Date of Defense: 25 July 2023
Subjects: 300 Social sciences, sociology & anthropology > 330 Economics
600 Technology > 650 Management & public relations
Institute / Center: 03 Faculty of Business, Economics and Social Sciences > Department of Economics > Institute of Economics
Depositing User: Hammer Igor
Date Deposited: 12 Dec 2023 13:53
Last Modified: 30 Mar 2024 20:24

Actions (login required)

View Item View Item