New Perspectives on Insurance

Abstract

This special issue originates from a dual submission conference with the NBER Insurance Working Group and the Corporate Finance Program in 2020. It brings a broader perspective on important frictions in insurance markets, including trust between insurers and policyholders, conflicts of interest among brokers, suboptimal policyholder behavior, and risk-based capital regulation. Several developments in the economy and the academic literature have provided an impetus for new perspectives, including the growth of savings products with minimum return guarantees, the global financial crisis, and intermediary asset pricing. We conclude with an overview of research questions that are promising for further exploration.

1. Impetus for New Perspectives on Insurance

The traditional role of insurers is to facilitate the diversification of idiosyncratic risks across households and firms. Insurers collect premiums from policyholders and invest in financial assets. Thus, insurers are one of the largest categories of institutional investors, especially in fixed income markets. Because life insurers are the largest category of institutional investors in the corporate bond market, their portfolio choice affects real investment decisions and economic activity.

The insurance sector also plays an important role in some of the most important economic issues today. Climate change has made property and casualty insurance more challenging and essential. Socially responsible investing has fundamentally changed the asset management industry, in which insurers play an important part. An aging society places a greater need on longevity and long-term care insurance, whereas the pressures on public finances necessitate private insurance solutions. The digital revolution has created cyber risk that also requires insurance.

Since Rothschild and Stiglitz (1976), a large literature focuses on informational frictions on the demand side that lead to adverse selection and moral hazard. A smaller literature on property-casualty insurance focuses on capital market frictions that cause supply shocks following weather-related catastrophes (Gron 1990; Froot 2007). This special issue brings a broader perspective on important frictions in insurance markets, including trust between insurers and policyholders, conflicts of interest among brokers, suboptimal policyholder behavior, and risk-based capital regulation. We believe that several developments in the economy and the academic literature have provided an impetus for new perspectives over the last decade.

First, with the secular decline of private defined benefit plans and government pension plans around the world, the primary business of life insurers has evolved from life insurance to savings products with minimum return guarantees (Koijen and Yogo 2021). In the United States, these products are called variable annuities, which are essentially mutual funds with long-dated put options that guarantee their returns. Savings products with minimum return guarantees are the largest category of liabilities in other countries such as France (Hombert and Lyonnet 2022). Consequently, the fundamental nature of risk has evolved from idiosyncratic risks to aggregate market and interest rate risks. The complexity of variable annuities also raises important issues about the role of brokers for financial advice (Bhattacharya, Illanes, and Padi 2020; Barbu 2021; Egan, Ge, and Tang 2022), and the quality of financial advice could ultimately affect households’ trust in insurance markets (Gennaioli et al. 2022).

Second, the global financial crisis exposed the fragility of the life insurance sector, partly because of gaps in financial reporting and regulation. Lax reporting standards allowed securities lending to build up, and AIG lost at least $21 billion through securities lending during the global financial crisis (McDonald and Paulson 2015). In addition to AIG, Hartford was bailed out through the Troubled Asset Relief Program due to losses on their variable annuity business. Life insurers have had a challenging time managing aggregate market and interest rate risks through the global financial crisis, the subsequent low interest rate environment, and the COVID-19 crisis (Hartley, Paulson, and Rosen 2017; Koijen and Yogo 2022c). Consequently, life insurers have reduced risk exposure through safer contract design (Koijen and Yogo 2022b) and portfolio choice (Ellul et al. 2022).

Third, the global financial crisis has stimulated a literature on intermediary asset pricing. This literature shows that the portfolio choice of institutional investors has an important impact on asset prices. The insurance sector provides particularly clean examples because of the availability of security-level holdings data and quasi-random variation in the risk-based capital regulation. Life insurers’ portfolio choice has an important impact on corporate bonds (Ellul, Jotikasthira, and Lundblad 2011; Becker and Ivashina 2015) and nonagency mortgage-backed securities (MBSs) (Becker, Opp, and Saidi 2022).

So far, insurers have received less attention from financial economists than other institutions, such as banks, broker-dealers, mutual funds, and hedge funds. Yet, the abundance of important questions, as we will discuss in Section 4, and the availability of relatively unexplored data make insurance a ripe field for research.¹ We hope that this special issue will provide new perspectives and stimulate more research on insurance. For researchers new to insurance, Koijen and Yogo (2022a) is a graduate-level textbook that provides a unified framework to study all important functions of the insurance sector, including pricing, contract design, reinsurance, portfolio choice, and risk management.

2. Dual Submission Conference

Together with Ben Handel of the University of California at Berkeley, we co-organized a joint meeting of the NBER Insurance Working Group and the Corporate Finance Program on April 24, 2020. We widely distributed a call for papers in January 2020 with a dual submission option at the Review of Financial Studies. The conference was moved to a separate week from the Corporate Finance Program Meeting and held virtually because of COVID-19. The program is available at https://www.nber.org/conferences/financial-economics-insurance-spring-2020.

We received 93 submissions to the conference, of which 49 (53|$\%$|⁠) were dual submissions to the Review of Financial Studies. We were thrilled by the quantity and quality of submissions. We prescreened the papers for thematic fit and quality and selected 13 papers for further review. In addition, we invited two papers that were not part of the initial submission pool. All 15 papers went through the regular review process. In the end, six papers were accepted and are published in this special issue. Two papers are still under review and, if accepted, will be published in a future issue of the Review of Financial Studies.

3. A Summary of the Papers

3.1 Trust, conflicts of interest, and investor sophistication

Gennaioli et al. (2022) study the role of trust in insurance markets, building on earlier work (Guiso, Sapienza, and Zingales 2008; Guiso 2012). Trust is particularly important when an insurance claim is too small for legal action. The authors obtain unique data from a multinational insurer that sells homeowners insurance in 28 countries. They find that in high trust countries, insurers are less likely to reject claims and pay a higher share of the initial claims.

To understand how trust affects contracting outcomes, the authors develop a model in which both the insurer and policyholders can act opportunistically by denying claims or by filing false claims. The cost of acting opportunistically is higher in higher trust countries. The model has three predictions. First, higher trust countries report fewer disputes over claims. Second, the cost of insurance is lower in higher trust countries because less resources are wasted on verifying claims and paying false claims. Third, insurance markets operate more efficiently in higher trust countries by lowering markups and raising profits. The authors find broad empirical support for these predictions.

An important theme that emerges from the paper is that trust affects the entire contracting process and shapes all aspects of insurance markets, including pricing, contract design, equilibrium market size, and the degree of market incompleteness. The authors map out a research agenda to further explore the role of trust in insurance markets across business lines and countries.

Trust also plays an important role in the market for financial advice. Brokers play a central role in the insurance industry by selling most types of policies and accounting for the largest share of sales. Thus, their financial incentives are potentially important for market outcomes. Egan, Ge, and Tang (2022) use new data on broker commissions to study the impact of financial incentives on the variable annuity market.

The authors report three main findings. First, variable annuity sales are strongly positively related to broker commissions, and sales are more sensitive to broker commissions than to contract expenses. Second, broker commissions are higher for variable annuities with higher contract expenses and inferior investment options. Consistent with the presence of a conflict of interest, higher broker commissions predict higher cases of policyholder complaints and broker misconduct. Third, the Department of Labor issued a fiduciary rule to reduce conflicts of interest in 2016. Despite being vacated in 2018, the fiduciary rule decreased the sales of high-fee products and created new lines of low-fee products. Finally, the authors estimate a structural model and show that the fiduciary rule increased consumer surplus.

Financial advice plays an important role partly because customers have limited financial knowledge to evaluate complex insurance products. Hombert and Lyonnet (2022) show the importance of limited investor sophistication in achieving intercohort risk sharing. The authors study euro contracts in France, which are popular savings products sold by life insurers. Life insurers collect premiums from various cohorts of investors, invest in risky assets, and pay out smooth returns by adjusting reserves. The authors show that the average annual intercohort transfer is 1.4|$\%$| of account value or 0.8|$\%$| of gross domestic product.

In a competitive market with fully optimizing investors, intercohort risk sharing cannot be sustained in equilibrium (Allen and Gale 1997). Because reserves predict subsequent returns, investors have an incentive to switch from insurers with lower reserves to those with higher reserves. Yet, the size of the euro contract market suggests that intercohort risk sharing is possible in reality. The authors develop a model in which investors could have a low demand elasticity to expected returns (conditional on reserves). In equilibrium, the scope for risk sharing is inversely related to the demand elasticity. An interesting implication is that if demand were more elastic, due to higher investor sophistication or better financial advice, intercohort risk sharing would be more limited.

The authors find that demand is insensitive to reserves even though reserves predict returns across insurers. Thus, the authors estimate a low demand elasticity. Furthermore, the authors find supporting evidence that the low demand elasticity is due to limited investor sophistication. The authors estimate an annual welfare gain of 90 basis points from investing in euro contracts.

3.2 Regulatory frictions and portfolio choice

The traditional view is that life insurers are safe financial institutions. However, this view changed when the global financial crisis exposed risks that had been building up in the life insurance sector. Life insurers were more exposed to systematic risk because their liabilities had shifted from life insurance and fixed annuities to variable annuities. In addition, leverage was building up through the use of modern risk management tools, such as securities lending and captive reinsurance (Koijen and Yogo 2016). Binding risk-based capital constraints caused extraordinary pricing of fixed annuities and life insurance during the global financial crisis (Koijen and Yogo 2015).

The next two papers in this special issue contribute to this literature by showing that life insurers’ portfolio choice is sensitive to risk-based capital regulation. In related work, Sen (2022) shows that their hedging behavior through derivatives is also sensitive to risk-based capital regulation. These papers highlight potential sources of fragility in the life insurance sector.

Becker, Opp, and Saidi (2022) find that life insurers’ demand for nonagency MBSs responded to a 2009 reform in risk-based capital regulation. Rating agencies downgraded a large share of MBSs as a consequence of the global financial crisis. Under the existing regulation, risk-based capital was a decreasing function of ratings. Therefore, life insurers would have had to sell downgraded MBSs to avoid an adverse impact on risk-based capital, which would have caused a fire sale.

However, regulators removed the risk-based capital regulation for MBSs, so that risk-based capital was no longer a function of ratings. This reform only applied to MBSs. The authors find that life insurers were less likely to sell downgraded MBSs, compared with other downgraded fixed income securities. This effect is stronger for life insurers with weaker capital positions.

Life insurers were not only more likely to hold onto downgraded MBSs but were also more likely to purchase MBSs in the primary market. Moreover, their demand for MBSs, particularly in the high-yield market, crowded out other institutional investors, such as pension funds and mutual funds. Finally, the authors estimate a causal effect of the reform on MBS demand, by exploiting quasi-random variation in the risk-based capital regulation.

These results show that risk-based capital regulation affects not only insurers’ asset demand but also the asset demand of other institutional investors by market clearing. Understanding how capital regulation in one sector spills over to other sectors through financial markets (e.g., from insurers to pension funds or from banks to insurers) is an important topic of future research. This research has policy relevance because regulators typically focus on their own sector without considering spillover effects in other parts of the financial system.

Ellul et al. (2022) study how the liability structure of life insurers could affect their portfolio choice and risk management through risk-based capital regulation. The authors build a model of life insurers whose liabilities consist of traditional life insurance and variable annuities, where the latter are sensitive to stock market risk. Life insurers can allocate their portfolio between safe-liquid bonds, risky-illiquid bonds, and stocks. Life insurers perceive higher risk-adjusted returns on illiquid bonds than liquid bonds (e.g., because of limited liability). Life insurers use stocks to imperfectly hedge the stock market exposure of variable annuities.

Under the assumption that variable annuities are more profitable than traditional insurance, an exogenous increase in variable annuity liabilities increases retained earnings and relaxes the risk-based capital constraint. Consequently, life insurers increase their allocation to illiquid bonds. The authors find support for this prediction using insurer-level data on variable annuities and portfolio holdings.

However, a stock market downturn or a low interest rate environment can stress risk-based capital because of imperfect hedging. Thus, insurers simultaneously sell illiquid bonds, leading to a fire sale. Based on an extended version of the model, the authors quantify the resilience of the life insurance sector to various shocks. They conclude that a policy intervention may be required in an adverse scenario in which the stock market falls, the value of minimum return guarantees rises (because of an increase in stock market volatility or unexpectedly low interest rates), and a large share of fixed income securities are downgraded. This paper is an example of how regulatory filings and financial modeling could be used for stress testing of the insurance sector.

3.3 Public insurance and real economic activity

The previous papers that we summarized are about private insurance markets. Doornik et al. (2022) is an example of work on public insurance markets, such as health care, public pensions, and unemployment insurance. The main questions in this literature are whether public and private insurance are substitutes or complements, and how public insurance could create incentives that affect labor supply and real economic activity.

Doornik et al. (2022) study an unexpected unemployment insurance reform in Brazil that tightened eligibility criteria for two-thirds of workers in 2015. This natural experiment allows the authors to observe changes in labor supply within the same firm for some workers who lost eligibility and other workers who remained eligible. The authors report three main findings. First, labor supply decreases and wages increase for workers who lost eligibility. Second, these effects are stronger for riskier firms, which is partly explained by workers moving from riskier to safer firms. These results imply that unemployment insurance subsidizes employment at riskier firms. Third, the reform reduces entrepreneurial activity, which is the net effect of two opposing forces. On the one hand, the reform makes formal employment less attractive, so that workers are more likely to start their own businesses. On the other hand, the reform increases labor costs, so that starting a business is less profitable. The authors find that the latter effect dominates, reducing entrepreneurial activity overall.

4. Open Research Questions

The abundance of important questions and the availability of relatively unexplored data make insurance a ripe field for research. We conclude with an overview of research questions that are promising for further exploration.

4.1 Climate risk

A growing literature in finance attempts to estimate the relation between climate risk and asset prices, including stocks (Hong, Li, and Xu 2019; Engle et al. 2020; Bolton and Kacperczyk 2021), bonds (Painter 2020; Goldsmith-Pinkham et al. 2021; Acharya et al. 2022), and options (Ilhan, Sautner, and Vilkov 2021). We refer readers to Giglio, Kelly, and Stroebel (2021) for a recent review. Climate is a low-frequency risk factor that is difficult to measure, and the presence of other risk factors makes inference of climate risk from asset prices difficult. In this context, insurance, reinsurance, and catastrophe bonds that have relatively pure exposures to climate risk are promising (Tomunen 2021).

Another important issue is how insurers adapt to climate risk by changing pricing, contract design, and participation in different business lines. Regulators also play an important role in determining market outcomes by standardizing contract terms and approving rate changes for homeowners insurance (Oh, Sen, and Tenekedjieva 2021).²

Finally, banking and insurance regulators are designing and implementing climate stress tests, which address direct physical risk and indirect financial risk from the transition to cleaner energy. Understanding how insurers are exposed to such risks is an important area of future research.

4.2 Cyber risk

The digital revolution, particularly big data and cloud technology, has created cyber risk as a new business line for insurers. Despite the growing demand (Eisenbach, Kovner, and Lee 2020), insurers are reluctant to underwrite cyber risk insurance due to an uncertain loss distribution, especially with respect to correlated tail events (e.g., a cyberattack on cloud computing services used by many firms). Loss experience data are difficult to collect because firms that have been attacked are reluctant to share such information publicly.

Similar to climate risk, cyber risk is difficult to measure. However, recent papers attempt to measure firms’ exposure to cyber risk (Florackis et al. 2020; Jamilov, Rey, and Tahoun 2021). As more data on cyber risk insurance becomes available, it would be interesting to test whether pricing and contract design follow the same principles as other markets with tail risk, such as catastrophe insurance, variable annuities, and long-term care insurance.

4.3 Big data and artificial intelligence

Following Rothschild and Stiglitz (1976), the insurance literature assumes that customers have private information about their risk type. However, big data and machine learning tools may give insurers superior information about the loss distribution. For example, life insurers have vast amounts of underwriting data and medical records that allow them to better estimate mortality, lapsation, or surrender risks. Auto insurers collect vast amounts of driving data from policyholders that agree to install tracking devices in exchange for lower premiums (Jin and Vasserman 2021).

This modern information environment raises new questions. Theoretically, insurance markets may achieve a different equilibrium when insurers have information that customers do not have (Abrardi, Colombo, and Tedeschi 2020; Brunnermeier, Lamba, and Segura-Rodriguez 2020). Empirical methodology to test this theory is yet to be designed, and whether this theory better explains insurance markets remains to be seen.

4.4 InsurTech

Big data and artificial intelligence have affected every part of the financial sector, including banking, brokerage services, and mortgage lending. As summarized by Goldstein, Jiang, and Karolyi (2019), the Review of Financial Studies published a special issue on FinTech in May 2019.

InsurTech refers to similar developments in insurance. Big data and machine learning could improve risk pricing, and artificial intelligence could simplify the traditional underwriting process that involves brokers. In addition, blockchain technology could help automate payments and claims, leading to greater trust between insurers and policyholders. This technology could be particularly valuable in emerging economies, where insurance markets are less developed and legal enforcement is weaker.

4.5 Insurers’ portfolio choice and asset prices

Insurers are the largest category of institutional investors in the corporate bond market. Given the importance of corporate bond prices for investment and real economic activity (Philippon 2009; Gilchrist and Zakrajsek 2012), insurers’ portfolio choice and its impact on asset prices is an important topic of research (Ellul, Jotikasthira, and Lundblad 2011; Chodorow-Reich, Ghent, and Haddad 2021; Coppola 2021). However, we can only draw conclusions about asset prices if we study insurers’ portfolio choices together with those of other investors. For example, insurers hold corporate bonds in a leverage-constrained equilibrium (Black 1972) because they have access to relatively cheap leverage through their underwriting activity (Koijen and Yogo 2022d).

Demand system asset pricing is a new approach to studying asset prices by modeling the portfolio choice of all institutional investors and households (Koijen and Yogo 2019). Bretscher et al. (2021) apply demand system asset pricing to the U.S. corporate bond market, using security-level holdings data for insurers and mutual funds. An important agenda is to micro-found and improve models of asset demand with realistic features of the insurance sector, such as liability hedging demand and risk-based capital regulation. Then demand system asset pricing could be used for counterfactuals, such as the impact of changes in risk-based capital regulation on corporate bond prices.

4.6 Household finance

Household finance is a large and vibrant field that studies all aspects of household decisions, including precautionary and retirement saving, portfolio choice, housing, mortgages, and credit cards. From a welfare perspective, mistakes over insurance choice have a welfare loss that is an order of magnitude greater than portfolio choice (Koijen, Van Nieuwerburgh, and Yogo 2016). The annuity participation puzzle, which is the fact that many households do not annuitize their wealth in retirement, has dominated insurance research within household finance. However, we find interesting other puzzling facts that are less researched.

In Koijen, Van Nieuwerburgh, and Yogo (2016), a life-cycle model of insurance choice implies that households should rebalance from life insurance to annuities as they age. However, actual households do not rebalance due to apparent inertia. Moreover, heterogeneity in the strength of bequest motives that would justify insurance choices do not correlate sufficiently with observed household characteristics. However, wealth is an important household characteristic that correlates with demand for life and homeowners insurance, confirmed in administrative data (Gropper and Kuhnen 2022). This correlation is consistent with decreasing relative risk aversion or bequest motives that strengthen with wealth. Further research that compares life-cycle models of insurance choice with actual household behavior would be valuable for how the insurance industry designs and sells contracts.

Beliefs are an important aspect of why households may not purchase enough insurance or display inertia in their choices. The behavioral economics literature shows that households form beliefs based on past experience and underestimate rare events. These insights could lead to a more realistic model of insurance demand that narrows the gap between life-cycle models of insurance choice and actual household behavior. For example, Gottlieb and Smetters (2021) show that patterns in life insurance lapsation are consistent with a model in which consumers underestimate the likelihood of lapsing their policies.

Acknowledgement

This introduction is written for a special issue of the Review of Financial Studies on insurance. We thank Jim Poterba, Rob Shannon, Ben Handel, Daniel Paravisini, Manju Puri, Antoinette Schoar, and Amir Sufi for help on organizing a joint meeting of the NBER Insurance Working Group and the Corporate Finance Program. We thank Alexandru Barbu, Andrew Ellul, Johan Hombert, and Ishita Sen for comments. Koijen acknowledges financial support from the Center for Research in Security Prices at the University of Chicago and the Fama Research Fund at the University of Chicago Booth School of Business.

Footnotes

References