Adaptive actions, once popularly considered the “lazy” alternative to climate change mitigation, are now a primary focus of the Intergovernmental Panel on Climate Change's (IPCC) strategy (Pielke et al., 2007; Prins and Rayner, 2007). Mitigation to limit carbon emissions or enhance carbon sequestration and storage will not produce noticeable results for decades (Pielke et al., 2007). Therefore, adaptive actions are vital components of any climate change policy, regardless of mitigation efforts. Although both coordinated “bottom up” and “top down” strategies will be crucial in climate change adaptation (Hill and Engle, 2013; Bierbaum et al., 2014), adaptive efforts are often more successful at smaller scales (Brooks and Adger, 2005; Estrada et al., 2017), with cities now widely recognized as important actors in responding to climate change (Rosenzweig, 2010). Cities, which are often hubs of innovation and economic activity, allow for a more comprehensive and context-specific response to climate change (Carter et al., 2015). However, climate change in cities is expected to affect municipal energy systems, water demand and supply, wastewater treatment, transportation, and public health (Rosenzweig et al., 2007; Rosenzweig, 2010). Poor urban households are particularly susceptible to these impacts as they are often in the locations most vulnerable to climate change effects (Rygel et al., 2006; Rosenzweig, 2010).

Controlling the cost of adaptive actions through the use of strategic and targeted initiatives is a priority for cities managing limited resources. Because the manner in which people adapt to stress can affect the costs and benefits of public policy (Kane and Shogren, 2000; Head et al., 2013), the adaptive capacity of individual households should be understood before municipal resources are allocated (Bierbaum et al., 2014; Araya-Muñoz et al., 2016). Large-scale governmental efforts to combat climate change that do not consider household-level actions and resources can undermine the inherent adaptive capacity of households (Dietz et al., 2009) or de-incentivize further protective measures taken by households (Barrett, 2006; Toole et al., 2016). Because the decisions of individuals can have cumulative effects on communities and policy outcomes (Kane and Shogren, 2000; Head et al., 2013; Elrick-Barr et al., 2016), households are increasingly viewed as foundational social units for observing adaptation and resilience (Toole et. al., 2016). However, the factors that determine the adaptability of urban households are poorly understood (Nhuan et al., 2016; Toole et al., 2016) and most studies that attempt to quantify adaptive capacity are performed at spatial scales larger than households (Adger and Vincent, 2005; Brooks et al., 2005; Araya-Muñoz et al., 2016) and in rural locations (Wall and Marzall, 2006; Below et al., 2012; Lockwood et al., 2015).

Developing instruments to measure adaptive capacity among households is a critical first step toward allowing cities to evaluate how local policy affects community resilience. To address this, we used Raleigh, NC, as a case study to develop a scale of autonomous adaptive capacity of urban households. Autonomous adaptation is defined here as the reactive adoption of certain perceptions and actions that allow households to better cope with emergencies, including those events resulting from climate change (Fenton et al., 2017). Because there is no well-established method for measuring autonomous adaptation, we utilized the extensively researched rural livelihoods framework and the determinants of adaptive capacity described by the IPCC to develop our survey instrument. The IPCC summarized the main determinants of adaptive capacity as economic resources, technology, information and skills, infrastructure, institutions, and equity (Smit et al., 2001; Engle, 2011). These attributes of adaptive capacity are included in the five capitals that comprise the rural livelihoods framework used by researchers to capture the combination of resources utilized to reduce vulnerability (Farrington et al., 1999; Hammill et al., 2005; Park et al., 2012). Economic resources are considered financial capital; technology and infrastructure are products of economic activity and are considered physical capital; institutions and equity are considered social capital; and information and skills that contribute to well-being and productivity are attributes of human capital (Ellis, 2000; Brown et al., 2010; Tinch et al., 2015). This instrument for assessing autonomous adaptation of households, with further refinement and testing, may be used in urban areas to evaluate programs designed to impact resilience to climate change. Results of this study may also aid in the development of indicators of autonomous adaptation for households in other similar urban environments.

The 14 survey questions included in the scale exhibited internal consistency (α = 0.83). The scale's consistency was also reinforced by its criterion validity, whereby it was positively and significantly correlated with income (r = 0.37), education (r = 0.38), and having renters' insurance (r = 0.42). These results suggest face validity, and that with further refinement the 14 survey questions utilized for this scale may be used to measure autonomous adaptive capacity of urban households.

Results of the principal components analysis indicate three major factors underlying the scale—financial capital, political awareness, and access to resources (Table 2). These three factors met the Kaiser criterion and together explained around 56% of total variance. The first factor had the largest eigenvalue (4.6), explained the most variance (33%), and related to financial capital. The items with the most influence on the scale (loadings ≥0.3) were those regarding financial resources and assets. Together, these items had a Cronbach's alpha of 0.76, indicating their validity as a financial capital subscale. The second factor had an eigenvalue of 1.9, explained 13% of the variance, and highlighted items related to political awareness. Cronbach's alpha for the second factor also met the internal consistency criteria with a rather large alpha (α = 0.92). The third factor, with an eigenvalue of 1.3, explained 10% of the variance, and highlighted resource access. The resource access items did not have as high internal consistency, with a borderline Cronbach's alpha of 0.60.

The increasing role of the private sector in modern society and economies around the world may help explain why financial capital was the strongest dimension within our adaptive capacity measure. Nhuan et al. (2016) had similar results when they quantified household-level adaptive capacity in Vietnam. Taken together, these results suggest unequal wealth may play a role in adaptive capacity across a range of social, economic, and cultural contexts globally (McKenzie, 2005). The high internal consistency of our financial capital subscale (α = 0.76) could be partially explained by the more uniform and directly observable nature of financial resources, as compared to the more qualitative concepts found within human, physical, and social capitals (Tinch et al., 2015).

Political awareness, the second-strongest dimension identified, may illustrate the importance of public awareness of governance structures in the measurement of autonomous adaptation in urban households. Governance indicators have been identified as important determinants of adaptive capacity at nearly all other spatial scales and locations: nationally (Brooks et al., 2005), in rural agricultural communities (Lockwood et al., 2015), and in several countries around the globe (Engle and Lemos, 2010; Nhuan et al., 2016). Municipal governance is critical in the global response to climate change, as cities often do not face the same political barriers to mitigation and adaptation as national governments (Estrada et al., 2017). Without the divisiveness of national politics, political awareness at the local level may influence the adaptive capacity of individuals by increasing concern and possibly engagement with the political process (Smith and Leiserowitz, 2014; Shi et al., 2016). This is particularly meaningful to the urban poor who often live in places more at risk from climate change effects and are excluded from the policy making process (Adger, 2003). Therefore, policy makers and stakeholders should engage the public to contribute to a higher overall adaptive capacity.

Access to resources comprised items from human, physical, and financial capitals, which suggests that these three capitals may be interrelated. Similar relationships among these capitals have been found in rural studies conducted at refined spatial scales (Ellis, 2000; Lockwood et al., 2015). While the capitals are often treated as discretely measureable items at the national level (Tinch et al., 2015), these findings indicate that different combinations of capitals may be required to measure adaptation at the household level. Traditional livelihoods assessments at large spatial scales that cast household actions as voluntary and calculated do so without gathering information on how people gain and utilize the capitals, and may not reflect the interdependencies of people's needs and arrangements (Dijk, 2011).

Social capital was not influential to the measurement of autonomous household adaptation in this case study, the questions we developed may not have been sufficiently comprehensive. For instance, social capital has been shown to include many distinct dimensions that may be difficult to quantify (Van Beuningen and Schmeets, 2013). Alternatively, social capital can contribute less to adaptive capacity in populations that are more educated, as well as among those with a more individualistic culture (Guiso et al., 2004). Raleigh has both of these characteristics, with a high percentage of people with secondary degrees (U. S. Census Bureau, 2015) and, as part of the United States, traditionally considered a more individualistic society (Han and Shavitt, 1994).

Our sample overrepresented the economic diversity of Raleigh, NC and was also limited in scope to that region, these limitations likely impacted the dimensions of autonomous adaptation identified with the PCA. Thus, future analyses with larger samples, representing larger geographic areas, are likely to improve both the instrument and understanding of dimensions underlying autonomous adaptation. Future research should also explore the role of natural capital in urban locations, as the current methodology and indicators are limited. Lastly, a more comprehensive measurement of social capital should be incorporated in future instruments.

With further refinement, our tool may help identify economic disparity, refine areas of interest for city planners, and direct municipal investments to the most vulnerable residents. Investments in the economic welfare (e.g., employment availability and sustainability) and resource access (e.g., healthcare and public transportation) of households could improve the overall adaptive capacity of cities. Although long-term city plans often incorporate these types of welfare and infrastructure projects, citywide socio-economic exclusion and income inequality could continue to increase without a targeted approach (Lee, 2011). Our instrument could help provide this targeted approach because it incorporates multiple measures of household welfare and economy that are rarely incorporated into publicly available data sets. While adaptation planning by cities is on the rise (Cruce, 2009), implementation is in its infancy because cities lack scientific guidance to justify expenses (Carlson and McCormick, 2015; Lehmann et al., 2015). Deeper and more context-specific understanding of household adaptation may help justify expenses as well as identify approaches that increase resiliency and climate adaptation without high-cost infrastructure, such as through initiatives to increase political awareness and engagement.

This study was carried out in accordance with the recommendations of the NC State University Institutional Review Board for the Use of Human Subjects in Research, with written informed consent from all subjects. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the NC State University Institutional Review Board for the Use of Human Subjects in Research (Protocol Number 4087).

KS, GH, MP, SB, and MM all provided substantial contributions to the conception and design of the work, the interpretation of the data, the drafting of the manuscript, and the final approval of the manuscript. KS was responsible for the acquisition of the data.