Essays on Beliefs about Catastrophic Risks

 Chapter 1 How do homebuyers adapt after experiencing a natural disaster? Evidence from the Florida real estate market 1.1 Introduction One of the biggest questions surrounding the issue of climate change is adaptation: will people adapt to rising temperatures and extreme weather events, how costly will these adaptations be, and will they be made before it is too late? There is an enormous literature studying climate adaptation around the world (Dell, Jones and Olken 2014; Mendelsohn, Nordhaus and Shaw 1994; Deschênes and Greenstone 2007; Kaiser et al. 1993). A key ingredient in adaptation is the initial understanding that a change needs to be made. Researchers have shown that first-hand experience with factors like excessive heat and natural disasters can promote belief in climate change (Deryugina 2013; Konisky, Hughes and Kaylor 2016) and spur adaptation in the context of agricultural investments or flood insurance takeup (Mase, Gramig and Prokopy 2017; Gallagher 2014). However, there is still relatively little empirical evidence on how exposure to climate change affects real estate purchases (Kahn 2016; Bunten and Kahn 2014; Boustan, Kahn and Rhode 2012). Any such response has major implications for coastal areas. Adverse demand shocks concentrated in these markets would destroy personal and commercial wealth, thereby shrinking municipal tax bases and reducing funding for schools and other public goods. In this study, I use the timing and geographic extent of flooding events as an exogenous informational and psychological treatment on the people living nearby, and measure the effects of this “disaster shock” on subsequent participation in non-local real estate markets. I aim to shed light on two important questions: how do people translate experience with disasters into adaptive behavior, and how destructive are these adaptations for coastal real estate markets? The answers to these questions bear on whether coastal markets are overbuilt with properties that will likely be submerged within a few generations. To analyze adaptive behavior after disasters, I pair data on over 4 million property sales from a database of Florida real estate transactions with information on over 400 flood disaster events across the United States. By focusing on the behavior of buyers who participate in Florida markets but hail from other states, I disentangle the psychological impact of disaster exposure from the direct physical damage impacts that can roil local real estate markets (Graff Zivin, Liao and Panassie 2020). Florida is a natural candidate for study as the state with the most real estate at risk from sea level rise over the next century (Union of Concerned Scientists 2018), as well as the most popular destination for out-of-state home buyers (Kerns and Locklear 2019). I use the random variation in the timing of flood disaster declarations to identify the causal effect of recent flood exposure on home purchases in coastal Florida markets, and purchases of homes near the water in particular. I find significant evidence of retreat from the coast among buyers coming from counties with recent flood disasters, 1 2 CHAPTER 1. HOW DO HOMEBUYERS ADAPT AFTER EXPERIENCING A NATURAL DISASTER? with shocked buyers purchasing 20-30% fewer homes within 1 km of the water (relative to farther inland) than under the counterfactual where they are not shocked. I use a random-utility sorting model to estimate how much shocked buyers devalue flood-prone properties, and find suggestive evidence that shocked buyers value properties in flood zones $500 -$3,000 less than comparable non-shocked buyers. I then employ hedonic price models to quantify the impact of disaster-shocked buyers on the marginal implicit price of coastal proximity. I find that a large influx of shocked buyers to a Florida market can significantly erode the amenity value of access to the water, reducing property values in the most affected parts of the market by 1-2%. In Section 1.2, I survey the fast-growing literature on questions of climate belief, adaptation, and real estate markets. In Section 1.3, I describe ZTRAX, the real estate database, as well as other public data sources that allow me to complete this analysis. I then proceed to attack the question of adaptive behavior in this context from three directions. In Section 1.4, I introduce simple transaction count models that allow me to identify retreat behavior, and show significant and robust evidence of retreat in Florida markets among shocked buyers. In Section 1.5, I introduce a residential sorting model that I estimate in an attempt to value the different preferences of shocked buyers in dollars terms, and measure the welfare gain to shocked buyers of the information imparted by their past storm experience. In Section 1.6, I introduce a “meta” hedonic price approach that I argue allows me to estimate the threat that retreat behavior poses to the value of the coastal amenity currently capitalized into Florida homes. In Section 1.7, I discuss the implications of my findings and propose avenues for future research.