Implicit Sustainable Development Theories Obscure Disproportionate Impacts from Climate-related Extreme Events: Example from Hurricane Michael and Housing Losses on Florida’s Forgotten Coast

8 A central challenge for sustainable development (SD) is how societies are to avoid, minimize or address 9 impacts from anthropogenic climate change. However, competing perspective on “what should be 10 sustained” lead to widely different understandings of what mitigation, adaptation and loss and damage 11 entail and how best to approach them. We provide a novel conceptual and empirical comparison of two 12 contrasting SD-based theoretical approaches to the study of impacts from climate-related extreme events: 13 Capital Theory and Human Development. We use our analysis of immediate residential property value and 14 housing capacity impacts caused by Hurricane Michael in Gulf County, Florida, to demonstrate how the 15 theory used to assess and interpret impacts greatly affects the identification of whom and where is 16 considered to be objectively “ most impacted ” . From our comparative analysis and discussion, we conclude 17 that, while currently underutilized, Human Development is the more advanced approach to SD oriented 18 climate-impact research and policy when compared to Capital Theory. 19


22
Evidence is mounting that anthropogenic climate change (CC) is already causing geographically uneven 23 and socially disproportionate harms in different places around the world (IPCC, 2018). However, the kinds 24 of impacts considered relevant to CC policy and practice, e.g., economic vs. non-economic (Serdeczny et  25 al., 2018), tangible vs. intangible (Tschakert et al., 2019), and at what scale remains contested terrain. The 26 way evidence of impacts is interpreted by researchers and policy makers will have important implications 27 for where and whom is considered "most impacted", and thus for what is considered a prudent policy 28 response to address climate-related impacts when they occur (Thomas et al., 2020). In the context of 29 Sustainable Development (SD), addressing this question means clarifying "what should be sustained" when 30 attempting to avoid, minimize or address impacts from CC. 31 Sustainable Development (SD) offers a fully comprehensive, integrative and coherent approach to 32 understanding and addressing anthropogenic climate change-driven impacts . Multiple 33 theories of SD are (implicitly) informing research and practice in this area, each with a unique set of core 34 concepts, evaluative criteria, informational requirements and related policy prescriptions (Boda et al., 35 2021). Two of the main approaches include: 1) Capital Theory, a utilitarian economic theory that prioritizes 36 maintaining a society's aggregate productivity through its capital stocks; and 2) the capabilities approach 37 to Human Development, a non-utilitarian welfare theory that prioritizes improving the substantive 38 freedom (i.e., capabilities) available to individuals in society, starting with the least-well-off. Outside,or 39 rather, between, these two approaches, a third unique body of work exists that emphasizes numerous 40 heterogeneous ways in which CC impacts may manifest, which has been labelled the "wish-list of valid 41 concerns" (Boda et al., 2021). While the spectrum of ideas that make up this body of work are not as 42 internally coherent as the main two approaches, its most notable and consistent characteristic is a concern 43 with the place-and cultural-specificity of many impacts from CC (see, e.g., Tschakert et al. 2017; Barnett 44 et al, 2016). This leads some of its proponents to develop extensive lists of different types of impacts from 45 CC that are supposedly incommensurable (e.g., Tschakert et al., 2019). How best to handle measurement 46 and monitoring of heterogeneous impacts from CC remains an important topic of contention in loss and 47 damage circles, though it is essential if a catalogue of climate-related harm is ever to materialize (Otto et 48 al., 2020). The current study, as explained below, maintains a focus on comparing the two main 49 approaches; however, when discussing the advantages of the capabilities approach to Human 50 Development in particular, we argue that it also provides a satisfactory resolution to the on-going "wish-51 list" debate over how best to conceptualize, measure and ultimately address the heterogeneous impacts 52 from CC. 53 It is well-known that there are differences in how competing theories conceptualize, measure and seek to 54 address development related challenges (see, e.g., Greig et al., 2007, Ch. 11 differ and how this could impact post-event response priorities (for exceptions see, e.g., Gardoni and 75 Murphy, 2010). More evidence of the advantages and disadvantages of competing SD approaches to CC-76 impact studies will furthermore help clarify the consequences for CC-impact researchers, and the policies 77 that flow from this research, of adopting, implicitly or explicitly, a particular approach. 78 We here empirically assess the impacts of Hurricane Michael to Gulf County's (Florida, USA) residential  79  properties through three different analytical impact indicators (total property value losses, proportional  80 property value losses, and loss of residential units or newly vacant residential properties), which we apply 81 to parcel-level property data collected by the Gulf County Property Appraiser before and after the 82 hurricane. We then interpret the results from the perspective of two contrasting approaches to SD, namely 83 Capital Theory and the Human Development approach. We discuss how these competing perspectives 84 lead to different appraisals of what defines a "highly impacted" area, as well as how they have the 85 propensity to skew towards certain types of information and thus the properties and people represented 86 by this information. We conclude with a reasoned and comparative appraisal of the two approaches, 87 favoring the capabilities approach to Human Development. We argue that the capabilities approach to 88 Human also the only category 5 storm on record to make landfall along this region of Florida, affectionately known 96 as the "Forgotten Coast". More than two years later, many communities impacted by the storm are only 97 beginning the process of long-term recovery. 98 The best science available suggests that the scale and type of impacts seen in the wake of Hurricane 99 Michael are indicative of what will become more likely in a climate-changed world (Patricola and Wehner,  100 2018; IPCC, 2018). By this, we mean impacts that exceed a variety of local hard and soft "limits" to 101 adaptation (Barnett et al, 2015), the culmination of which then lead to widespread damages. Post-102 hurricane analysis showed that storm surge reach 9-14 feet (2.7-4.3 m) above ground level in the hardest 103 hit areas along the coast, easily topping the natural height of the local barrier fore-dune system that 104 normally acts as a protective barrier against the impacts of coastal storms (Beven et al., 2018.). A rapid 105 damage survey conducted by Prevatt and Roueche (2019) showed that the breaching of this height limit 106 lead to catastrophic flooding and extensive damage to coastal infrastructure and residences. Hurricane 107 Michael wind speeds also breached other important hard limits, leading to extensive damage. Prevatt and 108 Roueche (2019) also found that wind speeds exceeded the physical limits of many structures in the 109 impacted area, though with important divergences. Homes with certain building characteristics, e.g., 110 construction year and material types, were correlated with particular degrees and types of damage, with 111 older homes generally fairing worse than newer houses in terms of wind damage, the former being less 112 likely to have adopted current construction standards. These same wind speeds also exceeded the 113 physiological limits of regional crops and forest species, e.g. longleaf pines (Pinus palustris) (

Study area -Gulf County 132
Gulf County is located in the Northwestern "Panhandle" region of Florida in the southeastern United States 133 ( Figure 1). Gulf County is a predominantly rural county with a population of just over 13,500 residents. Per 134 capita income is just over $21,000 while median household income is around $44,000. Roughly, 20% of 135 Gulf County's population lives in poverty (U.S. Census Bureau, n.d.). As previously mentioned, it was among 136 the most heavily impacted counties on a per-capita basis by Hurricane Michael (Beven et al., 2018). 137

Analysis of housing impacts 138
We focus on housing impacts for three main reasons. First, research has consistently shown that damage 139 to housing is both a common and deeply significant impact occurring as a result of tropical cyclones and 140 other natural disasters (Comerio, 1997;Zhang and Peacock, 2009). Second, in the context of the United 141 States, housing is commonly a major source of wealth for many households, and losses in its value can 142 have significant implications for household financial security, even inter-generationally (Wolff, 2016 and Human Development approaches when it comes to impact of climate-related extreme events. Our 153 claim is more modest but still important. We show, using housing as an example, how the concepts and 154 metrics in these competing approaches promote either exclusion or inclusion of impacts on particular 155 portions of the population, which clearly has implications for inter alia recovery policy at a variety of scales. 156 We analyze housing impacts using three different impact indicators. First, we analyze total residential 157 property value losses (i.e., aggregate monetary damages). This is taken as an indication of the quantitative 158 severity of residential property damages; that is, the higher the total monetary damages, the higher the 159 impact severity. Second, we analyze proportional residential property value losses (i.e., monetary damages 160 as a percentage of total property value). This is taken an indication of the qualitative severity of residential 161 property damages; that is, the higher the proportion of total property value lost, the higher the impact 162 severity. Third, we analyze the loss of residential units (i.e., newly vacant or lost residential properties). 163 This is taken as an indication of the severity of impacts to housing capacity; that is, the higher the 164 percentage of residential units lost, the higher the impact severity. We note here that just because an 165 indicator is itself numerical, does not imply it cannot represent qualitative characteristics (see, e.g., 166 Tabandeh et al., 2017). We evaluated these indicators at the parcel level, then analyzed them at two scales: 167 the county level and the intermediate "neighborhood" level. Parcels are the smallest spatial units of land 168 delineated in the study area ( Figure 1), with more than 18,000 discrete parcels across the entirety of Gulf 169 County in 2019. Our second spatial scale of analysis, the neighborhood level (Figure 1), is a grouping of 170 census blocks around six neighborhoods of high-density residential properties. These neighborhoods were 171 identified using a combination of 1) county zoning maps to narrow the geographic focus to residential 172 zonings only, 2) visual identification of higher density residential areas using parcel data in ArcMap, and 3) 173 the author's pre-knowledge of Gulf County and its distinct residential neighborhoods. 174 Insert Figure 1  175 Housing losses were evaluated using the parcel-level property appraisal data from the Florida Tax  176 Authority. Property appraisal data for 2018 and 2019 provide records of parcel value and land use and 177 property type before and after Hurricane Michael. We confirmed with the state property appraiser that a 178 parcel-by-parcel damage survey was conducted and recorded after the storm, and that tax information 179 contained in the property appraisal data is the most comprehensive measure of Hurricane Michael's 180 immediate impact on residential properties. It is still quite possible the survey under-valued some property 181 damages due to the practical limitations of the post-storm valuation process (e.g., external observation 182 vs. internal damages). The property appraisal data were made spatially-explicit by joining them to the 2019 183 shapefile of county-wide parcel boundaries. 184 We analysed all parcels in Gulf County categorized as residential in 2018. We included all parcels with any 185 of the three residential base strata used by the Tax Authority (Table 1). We used the base strata to 186 identifying the specific zoning of parcels as residential. We also observed the active strata to determine if 187 parcels were actively being used for residential purposes. We then analysed for 2019 the same parcels 188 that were categorized as residential according to their 2018 base strata, which were identified in the 2019 189 data by the unique parcel identifier. Only those parcels whose unique identifier matched between the 190 2018 and 2019 appraisal data, as well as the 2019 shapefile, were retained. A total of 6731 residential 191 parcels were analysed for the entire county, but two were removed for block-and neighborhood-level 192 analyses because of non-matching spatial data. 193 Insert Table 1 194 Monetary values of all residential parcels in 2018 and 2019 were obtained from the "just value" recorded 195 in the property appraisal. We adjusted "just value" by the "just value change" also recorded in the 196 appraisal data, which reflects any adjustment made to an initial property valuation upon a subsequent 197 valuation. Monetary losses (or gains) were calculated for each parcel as the change in adjusted just value 198 from 2018 to 2019. We then calculated these parcel-level losses (or gains) as a proportion of the 2018 199 property value. One outlier was removed (parcel ID 03178-110R) whose residential, mixed commercial-residential or municipal. Within these zones, the neighborhoods were 228 spatially bounded based on geographic proximity rather than formal municipal boundaries to maximize 229 the number of parcels captured in the neighborhood analysis. All statistical analyses were performed in R 230 and all spatial data processing performed in ArcGIS. Additional informational sources were also collected 231 to complement the primary analysis of housing data, including damage surveys conducted by academic, 232 state and private institutions. 233 converted to a single-unit property with greater value). However, our qualitative assessment suggests the 250 actual impact to housing capacity is much higher when considering some residential units are still occupied 251 but in squalid condition due to storm damage. 252

Results
Our analysis of the 5668 residential parcels that suffered monetary loss in value revealed that many 253 properties sustaining very high proportional damages remained active residences ( Figure 3A). Of the 413 254 parcels that lost more than half of their value from 2018 to 2019, 84 remained active residential in 2019, 255 including 10 properties that lost more than 75% of their value ( Figure 3A

Neighborhood-level impacts 267
Neighborhoods differed greatly in terms of total value, proportional value and housing capacity impacts 268 (

Differences among indicators 277
Our results show that the three indicators used give very different pictures as to the distribution and 278 magnitude of impacts. If the lenses were to give equivalent evaluations of impacts, one would expect the 279 rank-order to be similar when assessed through each indicator, which is not the case in Gulf County ( Figure  280 4). Large monetary losses do not necessarily imply large proportional losses or losses in housing capacity 281 (i.e., percent of units lost within a block), and vice versa. 282 The uneven distribution of impacts throughout Gulf County become even clearer when comparatively 283 ranking the six neighborhoods along the three lenses used to assess losses and damages (Figure 4) e.g. Cape San Blas, they still experienced devastating impacts to property, as reflected in high proportional 290 loss rankings ( Figure 4C), as well as in direct housing capacity, as reflected in high percent unit losses 291 ( Figure 4A)

Housing impacts from the perspective of Capital Theory 299
Capital Theory approaches SD from a utilitarian perspective and aims to sustain aggregate utility over time, 300 generally indicated by the level of per capita income. It thus emphasizes that SD is development that 301 maintains a society's overall productive capacity (i.e., stock of productive capital), as this is considered the 302 driver of economic growth (Solow, 1991). Monetary metrics and cost-benefit analysis are necessary tools 303 used to monitor and evaluate capital growth over time. It should be noted that, while Capital Theory is not 304 concerned with the precise distribution of wealth in society, it is not completely negligent of the issue. 305 Rather, Capital Theory relies on the assumption that a well-functioning market economy will provide the 306 most efficient (and least coercive) mechanism for distributing aggregate social wealth within society 307 (Solow, 1989). 308 From this perspective, housing is understood primarily as a "stock of capital" that can be invested in or 309 divested from depending on the rate of return on investment. Housing stock in this view is substitutable 310 with other productive industries, and investing in housing stock is (economically) rational when it leads to 311 growth in overall economic production (e.g., GDP). In this way, the most relevant information for 312 understanding Hurricane Michael's impacts to housing in Gulf County is to focus on the more than $250 313 million in county-wide aggregate property value losses. When it comes to the disproportionate distribution 314 of these monetary damages, and thus the identification of "most impacted" areas within the county, the 315 focus will be on those areas with the highest monetary losses. In other words, the most impacted areas 316 from the perspective of Capital Theory are those with the largest quantity of property value lost, which 317 are highly affected by the pre-existing relative value of a given property. From this perspective, housing is understood primarily as a basic necessity (often called a "conversion 349 factor") for a wide variety of essential capabilities, including maintaining health and employment. 350 Interpreting housing impacts in this perspective implies understanding them as leading to the deprivation 351 of individual capabilities. The most relevant information, then, is not the aggregate monetary losses as in 352 Capital Theory, but the disaggregated impacts on individual capabilities to continue to lead valued (e.g., 353 healthy) lives, which clearly includes being adequately housed. Thus, the most pertinent information is 354 regarding the qualitative severity of damage to a particular residential property/neighborhood (e.g., 355 proportional value losses), including the over-all capacity to accommodate citizens with residential units 356 (e.g., housing capacity losses). An interest in qualitative disproportionality, however, does not preclude 357 the possibility of assessing these impacts using quantitative data, as we have done here via proportional 358 value losses. When it comes to the disproportionate distribution of impacts, the capabilities approach aims 359 to focus on those residents who are the least well off, recognizing that there are qualitatively different 360 kinds of capability deprivation connected to pre-existing inequalities in capability sets and functioning 361 achievements ), including, for example, differences in housing quality. 362 The difference between proportional and total losses that we have shown is instructive. Our results reveal 363 that impacts viewed through the lens of proportional monetary losses (an indication of the qualitative 364 severity of impact to a property) highlight different highly-impacted areas than those under Capital Theory 365 (which emphasizes quantitative severity). North Port St. Joe, for example, sustained very high levels of 366 proportional property value loss, even though it ranks relatively low in terms of total damage levels. This 367 is, first, an indication that many houses in North Port St. Joe were severely damaged in the storm (even if 368 they did not become vacant) and, second, that the high proportional losses represent the potential for a 369 major hit to intergenerational wealth in an already low-income community. The importance of recognizing 370 the qualitative difference between total and proportional damages, and their implications for low-income 371 households, has been noted by other climate-impact researchers as well (van der Geest, 2018). 372 Comparing high value losses and high housing capacity losses brings out further important differences. 373 Many places with lower levels of monetary damage (i.e., low property values to begin with) experienced 374 high levels of housing units loss, for example North Port St. Joe and Highland View, implying a significant 375 loss in the ability to house residents. From this capabilities perspective, areas with high proportional losses 376 and high vacancy rates, as seen for example in St. Joe Beach as well as North Port St. Joe and Highland 377 View, could be considered potential deprivation hotspots. 378

Identifying the "high impact" areas: which SD approach is best? 379
Interpretation of the different impact indicators leads to very divergent assessments of which 380 neighborhoods in Gulf County were "most impacted" by Hurricane Michael (Table 3). Reviewing this 381 provides a good opportunity for discussing the comparative advantages and disadvantages of the two 382 competing SD approaches. 383 Capital Theory comes with strong technical advantages. One of the most immediate is that it is the 384 appropriate fit for much of the current practice in the areas of DRR, CC adaptation and loss and damage, 385 where economic-based assessment and policy rules the game (see, e.g., McNamara and Jackson, 2019; 386 Boda et al. 2021). Its focus on economic concepts, metrics and financial risk reduction policies results in a 387 degree of relative decisiveness and comprehensiveness that some argue may be essential for 388 operationalizing climate-impact research in the existing political climate (Roberts et al., 2017), or including 389 "stakeholders" such as the private sector (Surminski and Eldridge, 2015). Historically, the kind of strict 390 reductionism inherent in Capital Theory has proven appealing to policy makers in relation to a wide variety 391 of environmental and development concerns (Porter, 1996). A serious downside of Capital Theory, 392 however, is that its focus on aggregate monetary losses has the potential to draw attention away from the 393 worst off areas. For example, it is practically impossible for low property value neighborhoods like North 394 Port St. Joe (average parcel loss of $4,677) to be identified as the "most impacted" areas under this 395 perspective, as they simply do not have the property wealth to compare quantitatively with places like 396 Cape San Blas (average parcel loss of $74,872). 397 Insert Table 3  398 On the other hand, the Human Development approach, as we see it, has important distinct substantive 399 advantages over Capital Theory. While Capital Theory is primarily concerned with aggregate monetary 400 losses and must convert all relevant impacts into this unitary metric, the Human Development approach 401 does not ignore the importance of monetary losses, but incorporates it as one relative factor potentially 402 affecting capabilities. That is to say, one should take note of monetary losses not in its absolute magnitude 403 (á la Capital Theory), but in terms of the contribution that this magnitude makes relative to the overall 404 human capabilities of its proprietors. Relative figures such as proportional losses are instead preferred as 405 they allow some meaningful comparability between lower property value areas like North Port St. Joe and 406 high property value ones like Cape San Blas in terms of qualitative severity of impacts. This is because the 407 Human Development approach is concerned with the lost use-value of property, not only the exchange 408 value. The Human Development approach's interest in the qualities of impacts facilitates the inclusion of 409 all properties, irrespective of value, that experienced severe immediate impacts. This is reflected in for 410 example the ranking of St. Joe Beach (largely high property values) as highly impacted from both 411 perspectives, as it was ground zero of the hurricane and experienced near complete devastation (Prevatt 412 and Roueche, 2019) and thus high proportional and housing capacity losses, in addition to high total value 413 impacts. At the same time, North Port St. Joe and Highland View (low to middle value properties) take the 414 place of Cape San Blas as "most impacted" under the capabilities approach, due to higher levels of 415 proportional and capacity impacts. This attention to the qualitative heterogeneity of impacts points 416 towards another important advantaged afforded by the Human Development approach to climate-related 417 impact studies, which we elaborate in the following section. 418

Opportunities and challenges for a Human Development approach to climate-impact research and 419
policy 420 There are many important capabilities not well captured by the kind of value-focused housing impacts we 421 analyzed here, which might reasonably be expected in the aftermath of any extreme event. These include, 422 for example, access to adequate nutrition, environmental quality concerns, or loss of community 423 belonging, which affect many of the most vulnerable people including homeless, people with disabilities, 424 children, elderly and low-income renters. These kinds of so-called "non-economic" impacts have been 425 much discussed in research on the impacts from anthropogenic climate change, particularly in loss and 426 damage circles. One of the main bones of contention regards what metrics are appropriate when 427 accounting for non-monetary impacts. Some argue non-economic or intangible impacts from CC cannot 428 or should not be quantified or made comparable due to their being derived from particular cultural and 429 geographical contexts, which renders them incommensurable (Tschakert et al., 2017). The answer to this 430 question from within Capital Theory, of course, is to require all relevant impacts be converted into 431 monetary metrics, thus rendering them commensurable (Dilley and Grasso, 2016), for which there are 432 many standard tools (Preston, 2017 process of open and reasoned public deliberation (see Sen, 1999). While his attempt to operationalize the 448 capabilities approach lead Sen into the realm of ethics and theories of justice (see Sen, 2011), there are 449 unquestionably other avenues open to exploration for its operationalization, which points towards fruitful 450 further research. 451

452
In this article, we have provided empirical measurements of impacts to residential properties in Gulf 453 County, Florida occurring from Hurricane Michael, a limits-breaching climate-related extreme event. In the 454 aggregate, we reported widespread and devastating impacts, with nearly nine in ten of county-wide 455 residential properties sustaining immediate value losses, and nearly 1 in 5 of total residential units 456 rendered vacant or lost between 2018-2019. Our interpretation of these impacts through competing 457 theories of SD show how they lead to dramatically different appraisals of where and whom is considered 458 "most impacted". While Capital Theory is a highly operational approach compatible with much current 459 practice, its emphasis on aggregate monetary losses has the potential to draw attention away from 460 properties that experienced severe qualitative damages and towards high-value properties, with the 461 possibility of overlooking those most in need.   The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 2
Percent frequency distributions of (A) total and (B) percentage changes in value from 2018-2019 for 6731 Gulf County residential parcels. Darker colour shades indicate parcels with net losses. NB horizontal axes are truncated and actual ranges were $-2.15 million to $0.42 million for (A) and -126% to 795% for (B), although only a tiny fraction of parcels had a change in value outside the ranges shown.