The threshold effect of climate risk and the non-linear role of climate policy uncertainty on insurance demand: Evidence from OECD countries

https://doi.org/10.1016/j.frl.2023.103820Get rights and content

Highlights

  • We show CRI and CPU have both linear and non-linear impacts on insurance demand.

  • Climate risk has U-shaped bearing with life insurance demand.

  • Climate risk has inverse S-shaped bearing with non-life insurance demand.

  • Risk consciousness, self-controllability, and insurability lead to non-linearity.

Abstract

This paper investigates whether climate risk and climate policy uncertainty (CPU) have traditional linear or complicated non-linear impacts on insurance demand. We find evidence that both phenomena exist. Linear model shows positive effect of climate risk and negative effect of CPU on insurance demand. However, climate risk and CPU are interdependent, non-linear analysis demonstrates that CPU acts as a threshold. In particular, climate risk has a U-shaped relationship with life insurance demand and an inverse S-shaped relationship with non-life insurance demand, which provides insurance companies with a better understanding of how insurance demand fluctuates under climate risk.

Introduction

Global climate change has contributed to governments and international organizations adopting climate and environmental-related policies in response to the challenges. Insurance is one of the essential economic element to address climate-related risks. Meanwhile, climate-related risks also pose challenges to the development of insurance industry, such as insurance demand, insurance insolvency, and business performance etc. While various effects of climate risk on the insurance sector are well-documented (Tucker, 1997, Tol, 1998, Dlugolecki, 2008, Botzen et al., 2009, Botzen et al., 2010, Ranger and Surminski, 2013, Stechemesser et al., 2015, Crick et al., 2018), little literature considers the impact of climate risk and climate policy uncertainty (CPU) on insurance demand in an integrated manner. Exploring the economic consequences of climate risk and CPU on insurance demand are crucial for the sustainable and healthy development of the insurance sector in the current era when climate change has emerged as a significant issue. It also has practical significance for policymakers to develop climate policies purposefully.

Since the deterioration of global warming in recent years, on the one hand, households and businesses are increasingly exposed to climate risks; on the other hand, the increases in climate risk have significantly accelerated the frequency of climate policy changes and raised the uncertainty around such fluctuations, i.e., climate policy uncertainty. In the context of climate risk, climate change may negatively affect insurance affordability and availability (Mills, 2005), and result in the costs of claims soaring (Tucker, 1997), thereby impeding the industry’s expansion. While the insurance sector confronts climate risk brought by climate change, the insured bear a more significant share of potentially increased natural hazards (Tol, 1998). In terms of climate policy uncertainty, according to real options theory, elevated climate policy uncertainty may cause firms or households to ‘hesitate’ on insurance consumption and wait for more information to be disclosed before consumption decisions. Therefore, climate policy uncertainty could dampen insurance demand. Real options theory is based on the assumption of irreversibility, arguing that only when insurance consumption is irreversible, firms and individuals tend to wait for a more clear-cut investment and consumption environment to avoid uncertainty in investment and consumption now (Bernanke, 1983). In addition, the effects of climate risk on the economy are complicated (Mills, 2005, Ranger and Surminski, 2013). Considering the channel through which climate risk affects climate policy fluctuation and, in turn, the demand for insurance. Intuitively, the presence or absence of the threshold effect of climate risk on insurance demand under the role of climate policy uncertainty comes to mind.

In this paper, we use the novel climate risk score from Germanwatch (Eckstein et al., 2021) and the CPU index of Gavriilidis (2021) to investigate the traditional linear and complex non-linear impacts of climate risk and CPU on insurance demand, and the threshold effect of climate risk on insurance demand when CPU serves as threshold variable. We make two-fold contributions to the literature.

First, we consider life and non-life insurance demand separately in our analysis, distinguishing our work from the extensive literatures that study the catastrophe insurance (e.g. flood insurance) demand affected by climate change. Traditional linear regression results show that both life and non-life insurance demand are positively influenced by climate risk and negatively influenced by CPU.

Second, we apply the non-linear panel smooth transition regression model (PSTR) to explore the threshold effect of climate risk with CPU as the threshold variable. By doing so, we comprehensively connect climate risk and climate policy uncertainty, distinguishing our work from the existing literatures that explore the effect on insurance demand from a single perspective of climate risk. Our results show that there is indeed a threshold effect of climate risk, and climate risk has a U-shaped relationship with life insurance demand and an inverse S-shaped relationship with non-life insurance demand, respectively.

To provide a clear and concise summary of our findings, we present Fig. 1, which shows the patterns and relationships of climate risk and CPU with the life and non-life insurance demand. We organize the rest of the paper as follows: Section 2 introduces methodology and data. Section 3 shows the empirical results and analysis. Section 4 concludes the paper with discussions.

Section snippets

Traditional linear model

A traditional linear regression model is established, as shown in Eq. (1): Insurancei,t=μi+β0+β1CRIi,t+β2CPUt+βzZi,t+δt+ɛi,t,where CRIi,t is climate risk index for country i in year t, and CPUt is climate policy uncertainty in year t. Zi,t represents a vector of control variables. μi serves as unobservable individual fixed effects. δt symbolizes unobservable time-varying effects. ɛi,t represents random error term. Insurancei,t denotes insurance demand.

In traditional linear model, a potential

Traditional linear regression

Table 3 reports the traditional linear regression results. Panel A reports the impact of CRI and CPU against life insurance demand. Panel B shows the impact of CRI and CPU against non-life insurance demand.

In Panel A, results confirm the positive correlation between CRI and life insurance demand, and the negative correlation between CPU and life insurance demand. The coefficient of CRI is 0.1133 at 5% significant level and coefficient of CPU is −0.1432 at 1% significant level in Model 3 (by

Conclusion

In this paper, we investigate whether climate risk and climate policy uncertainty has a traditional linear or complicated non-linear impact on insurance demand. Our evidence shows that both phenomena exist. To sum up, traditional linear regressions show a positive effect of climate risk and a negative effect of climate policy uncertainty on life and non-life insurance demand respectively. Moreover, there is a threshold effect of climate risk on insurance demand, and the non-linear analysis

CRediT authorship contribution statement

Bing Liu: Conceptualization, Methodology, Software, Writing – review & editing. Weijun Yin: Conceptualization, Software, Writing – original draft, Language editing, Visualization. Gang Chen: Conceptualization, Data curation, Methodology, Visualization. Jing Yao: Conceptualization, Methodology, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

We are grateful for the financial support of the National Natural Science Foundation of China (12001267, 11971506, 12101300) and the Natural Science Foundation of Jiangsu Province, China (BK20200833).

References (29)

Cited by (6)

View full text