Assessment of vulnerability reduction policies: Integration of economic and cognitive models of decision-making

Highlights

• A novel decision-making model based on economic and behavioral aspects is proposed.

• The decision-making model is integrated into an agent-based model.

• The ensuing model assesses the impact of policies and interactions on community loss.

• It facilitates ex-ante hazard mitigation policy design and fine-tuning.

• Risk-aversion and social interactions affect response to hazard mitigation policies.

Abstract

Earthquakes can cause significant damage to vulnerable residential buildings and cause irreversible adverse economic, social, and socioeconomic consequences. Implementing hazard mitigation policies can enhance homeowners' willingness to adopt hazard mitigation measures such as seismic retrofits or insurance. Several past studies have proposed models that aimed to explain individuals' hazard mitigation behavior. Despite their advantages, these decision-making models are subject to limitations. This manuscript proposes a new decision-making model that addresses the shortcomings of previous models. The proposed decision-making model is then incorporated at the core of an agent-based model to evaluate owners' responses to candidate hazard mitigation policies. The proposed decision-making model considers the characteristics of the buildings, the owners' attitude, and their condition-dependent response to proposed incentives. Hence, it can effectively explain homeowners' behavior when presented with various alternatives for mitigating the seismic risk. It can be applied to measure the impact of marginal changes in government subsidies to minimize community losses, as demonstrated in the illustrative example. Practitioners and policymakers can use the outcomes of the proposed model to select appropriate policies to enhance communities' resilience based on physical infrastructure characteristics.

Introduction

In disaster-prone societies, natural hazards can result in significant adverse consequences [1]. Among these consequences are the financial and life losses resulted from damage to residential and commercial buildings [2], transportation networks [3], and healthcare facilities [4], [5], [6]. Policymakers often attempt to mitigate the negative consequences of probable future hazards by implementing various policies. They can use existing models such as the recently presented reliability-based and Bayesian network models [7], [8], [9], [10], [11], [12], [13] to assess the vulnerability of civil infrastructure systems at system or component levels and, based on the outcome of these assessments, design appropriate policies. The adopted policies may aim to promote the adoption of hazard mitigation measures such as seismic retrofits that can enhance the robustness of buildings and reduce the probability of death, injury, and economic losses. Alternatively, they may promote disaster insurance, which transfers losses sustained by the residents to other parties [14, 15]. Examples of these programs that could be supported by policies are enhancing the public awareness of the seismic risk and its adverse outcomes, providing long-term and low-interest loans, tax credits, and penalties, as well as subsidies on insurance premiums [16], [17], [18], [19], [20], [21], [22], [23].

Despite the past efforts made by governments, evidence suggests that, in many cases, the implementation of hazard mitigation policies has not led to the widespread adoption of protective measures [24], [25], [26]. Several past studies [27], [28], [29] have highlighted the need for a multi-disciplinary perspective toward enhancing communities' resilience. In other words, the design and implementation of policies based on the application of technical models such as the existing reliability and resilience models while ignoring the human response may lead to unsatisfactory results [30]. Therefore, to optimally use the available government resources and achieve the desired outcomes, policymakers should consider the community's probable feedback to the proposed policies [31], [32], [33]. Critical to the prediction of the response of the community to a given policy is the understanding of the decision-making process and the behavior of individual owners under their social settings.

In the past, the expected utility theory [34, 35] has been used in several analytical models [36, 37] to model how people decide about the adoption of hazard mitigation measures. The expected utility theory argues that individuals decide based upon their level of risk-aversion, wealth, and uncertain financial benefits and losses. Despite its strength, the expected utility theory is subject to a critical limitation. This limitation is stemmed from the underlying assumption that individuals make rational decisions solely based on objective financial losses and benefits. It does not take into account how individuals perceive low-probability, high-consequence events such as earthquakes in a region [38, 39]. Due to their lack of knowledge and misunderstanding about the probability and consequences of events, individuals frequently under- or over-estimate the risks [40]. Due to its omission of individuals' perception of low-probability, high-consequence events, the expected utility theory cannot explain sub-optimal choices that individuals make about adopting seismic retrofit measures even when presented with objective risk analysis results [31].

An expanding body of literature (see, e.g., Mulilis and Duval (1995) [41], Lindell and Perry (2012) [42], Paton (2006) [43], Egbelakin et al., (2011a) [44]) has attempted to model individuals' hazard mitigation behavior as a complex phenomenon. They propose behavioral models that characterize the building owners' hazard mitigation decision and behavior based on several social and behavioral factors [43]. Unlike the expected utility theory, these models consider how individuals' biases influence their decision-making and behavior. Nevertheless, these models cannot characterize the impact of incremental changes in financial gains and losses associated with investments in seismic risk mitigation measures on individuals’ behavior. Therefore, there is a need for a new generation of models that considers the role of social and behavioral factors as well as incremental changes in the financial gains and losses associated with a given measure in driving individuals' mitigation behavior against low-probability, high-consequence natural hazards such as devastating earthquakes.

Furthermore, past studies (see, e.g., Terpstra and Lindell (2013) [45]) often approach individuals' risk mitigation decision making as a static, one-step, and now-or-never problem; whereas the owners’ choice of risk mitigation measures is inherently dynamic (e.g., an owner that currently finds a given measure inappropriate, may decide to adopt it at a later time if situation becomes favorable). Considering individuals hazard mitigation behavior as a static problem can lead to lead to over- or underestimation of building owners' response since humans respond adaptively to risk communication [46], risk reduction incentives [47] and interactions with neighbors and friends [48]. There is a need for appropriate models that characterize the dynamic and adaptive behavior of individuals while taking into account the role of social interactions, the diffusion pattern of the impact of government subsidies, and the changes in the level and nature of government interventions [49, 50]. Policymakers are most concerned with the impact of candidate policies and programs as characterized by the consequent change in the vulnerability of the community. Accordingly, they need tools that consider the behavior of individuals to determine the overall response of the community to vulnerability reduction policies and programs and characterize the vulnerability of the community at a given time using theoretically sound and well-founded models.

This paper contributes to the body of knowledge by addressing these two needs. First, it presents a new generation of models that characterize the homeowners' response to vulnerability reduction policies and programs. The proposed model addresses the limitations of existing models by simultaneously considering several social and behavioral factors as well as the utility individuals assign to the financial outcome of each vulnerability reduction measure (e.g., seismic retrofit). Secondly, an agent-based simulation model is developed to characterize the dynamic and emergent risk mitigation behavior of households under various exogenous and endogenous factors (e.g., seismic risk mitigation programs and social interactions among individuals). At the core of this agent-based model is the proposed decision-making model. By characterizing the time variant risk mitigation behavior of individuals, this agent-based model can determine the overall response of the community to a given program. A novel feature of this simulation model is the characterization of the community's vulnerability using a reliability-based seismic performance assessment model. Practitioners and policymakers can use the results and ensuing insights obtained from applying the developed simulation tool in the design and fine-tuning of seismic risk mitigation programs before their implementation.

The remainder of this manuscript is organized as follows. First, the foundations of the utility theory and the social-behavioral models are discussed. Next, the proposed decision-making model is presented. Then, the integration of the proposed decision-making model into an agent-based model is discussed. This is followed by an illustrative application of the proposed decision-making model to assess the impact of candidate seismic risk mitigation programs on the individuals' behavior and the community's overall vulnerability. The validation and testing for the developed model are subsequently presented.