The Role of Contingent Commissions in Property-Liability Insurer
Underwriting Performance
By
Laureen Regan, Ph.D.
Fox School of Business and Management
Temple University
Philadelphia, PA 19122
phone: 215-24-7264
email:
and
Anne Kleffner, Ph.D.
Haskayne School of Business
University of Calgary
2500 University Drive N.W.
Calgary, AB T2N 1N4
Phone: (403) 220-8596
e-mail:
Prepared for the American Risk and Insurance Association conference, Quebec City, Quebec, August, 2007
This is a working draft. Copyright Laureen Regan and Anne Kleffner, 2007. All rights reserved.
Abstract
Recent investigations into insurance compensation practices have drawn attention to the potential conflicts of interest that may be associated with payment of contingent commissions to insurance agents and brokers. However, despite the possible conflicts of interest, contingent commissions have also been recognized as a way to better align agent and insurer incentives. If effective, then contingent commissions should result in better underwriting performance. We investigate whether contingent commissions are associated with improved insurer underwriting performance. We estimate the relationship between the proportion of contingent commissions in insurers’ compensation scheme and loss and combined ratio levels, and the coefficients of variation of loss and combined ratios for the period 2001 through 2005. Using OLS and WLS methods, we find that higher proportions of contingent commissions are associated with lower loss ratios and lower combined ratios, and lower relative variation in both loss and combined ratios over time.
1. Introduction
In late 2004, an investigation into insurance broker compensation practices by (then) New York Attorney General Eliott Spitzer led to allegations of bid rigging and conflicts of interest in the payment of contingent commissions. Contingent commissions are paid in addition to flat percentage of premium commissions with the object of rewarding producers for properly matching risks with insurers, and for meeting volume, profitability, and retention targets. The initial New York lawsuit alleged that the largest broker, Marsh, systematically engaged in bid-rigging to steer clients to insurers that paid relatively higher contingent commissions. Spitzer’s initial investigation was widened by other state attorneys general and other regulators. In January, 2005, Marsh agreed to pay $850 million in restitution.[1] Since the investigation, the four largest insurance brokerage firms have agreed to forego contingent commissions, and a number of property-liability insurers have agreed to withdraw the use of contingent commissions in their compensation formulas (Green, 2007).
Opponents of the contingent commission system argue that it creates conflicts of interest between insurance intermediaries (agents or brokers)[2] and clients. If agents are paid an incentive commission by the insurer for placing business, and if the client is not informed of this payment arrangement, then it is possible that the agent might place the business with the insurer offering the highest compensation package, rather than the one that is most suitable for the client. In this case, agents will have an incentive to hide information about coverage offers, or misrepresent coverage offers. Then, clients might pay more for insurance than they otherwise would bcause they will not be fully informed of the coverage options available to them.
Prior studies have supported the idea that contingent commission payments can provide incentives for agents to place business with insurers so as to match clients with appropriate insurers (Hoyt, et al., 2006; Cummins and Doherty, 2006; Regan and Tennyson, 1996). Because agents provide information to insurers about the risk type of the applicants that insurers would find costly or difficult to verify, agents can reduce problems in pricing and risk classification caused by asymmetric information. Recent studies have measured the market's response to the New York litigation and have found significant negative returns for insurers following the announcement of the investigation, as well as larger discounts for insurers that use relatively more contingent commissions. (Ghosh and Hilliard, 2007; Cheng et al., 2007). This indicates that investors may believe that contingent commission payments add value to insurers, and that eliminating them is expected to reduce insurer profitability.[3]
There have been several studies that examine the performance of insurers using different distribution systems, (see, for example, Berger, Cummins and Weiss, 1997) but we are not aware of any papers that analyze whether contingent commissions achieve the goals they are designed to achieve. Does the use of contingent commissions improve insurer underwriting performance? If so, then eliminating them might have deleterious consequences. If not, then perhaps the incentive conflicts raised by the use of contingent commission outweigh any other value they might have, and thus they should be eliminated.[4]
The purpose of this research is to evaluate the effectiveness of contingent commissions in improving insurer underwriting performance. This is an important question because, as noted above, several insurers have eliminated the use of contingent commissions and the largest brokers have stated that they will not accept contingent commission payments. Further, over twenty states have launched investigations of insurance producer compensation schemes, and both the National Association of Insurance Commissioners (NAIC) and the National Conference of State Insurance Legislators (NCOIL) have developed model laws regulating the use of contingent commissions (Carson, et al., 2007; Fitzpatrick, 2006).
We hypothesize that, if contingent commission payments provide incentives for agents to exercise discretion in underwriting that result in better risk selection and better matching of client risk type with insurer risk appetite, then insurers using relatively more contingent commissions in the compensation scheme should have lower loss ratios on average, and also lower variation in loss ratios across time. One might expect that the payment of contingent commissions would result in higher expense ratios for insurers, so that prices paid by consumers might be higher overall. We test this hypothesis by examining combined ratios across insurers as well. Our findings indicate that both loss ratio levels and loss ratio variability are lower for insurers that use relatively more contingent commissions in the compensation scheme. Further, combined ratios are also significantly lower as contingent commission payments increase. The findings support the use of contingent commissions in aligning incentives for risk assessment between agents and insurers.
The paper proceeds as follows. Section 2 details alternate insurance distribution methods and compensation schemes, and analyzes incentive conflicts that might arise between insurers, agents, and clients in the insurance transaction. We develop our hypothesis about the role of contingent commissions here. In Section 3, our data and empirical methodology are discussed. Section 4 presents results of our estimation, and Section 5 concludes.
2. Insurance Distribution Channels and Compensation Schemes
The majority of property-liability insurance in the U.S. is sold through intermediated channels.[5] Principal-agent problems arise in the insurance transaction because the interests of the three parties to the transaction—intermediaries, insurers and policyholders—are not perfectly aligned. The nature of the principal-agent problem depends on the distribution model that is used. There are a variety of ways that property-liability insurers can distribute their products to clients. Direct channels include mass marketing and employee sales through branch offices. In this case, there is no intermediary between the client and the insurer, and principal-agency problems are minimized. Exclusive agency insurers rely on intermediaries who are autonomous contractors representing the products of a single insurer under an agency contract.[6] The insurer owns the rights to the client expiration list upon severance of the relationship with the exclusive agency.
Independent agency insurers also rely on autonomous contractors, but the agents represent the products of several competing insurers. Insurers may also rely on insurance brokers to distribute products. The chief distinction between independent agents and brokers historically was that brokers had a legal duty to represent the interests of the clients first, while independent agents were the legal representatives of the insurer. In practice, the legal difference is quite blurred, and has become more so since 2004 with the adoption of the NAIC's Producer Licensing Model Act (PLMA) by a majority of states. Under the PLMA, the regulatory distinction between agents and brokers was eliminated, referring to both as insurance "producers". Under either traditional independent agency or brokerage arrangements, the ownership of the client expirations list lies with the agent/broker rather than the insurer. Note that we will refer to each of these as agents or independent agents throughout the remainder of this paper.
The independent agent's role in distribution is to place business with the appropriate insurer, and in some cases, to perform a risk assessment function. From the principal's (insurer's) perspective, there are two types of agency costs that are important. The first arises from asymmetric information between insurers and agents about agent effort in placement, also known as moral hazard.[7] If the insurer cannot perfectly monitor agent effort in business production the agent may have incentives to shirk. This incentive can be mitigated by paying the agent a flat commission based on the volume of business placed with the insurer. Volume based contingent commissions may be paid in addition to flat commissions to reward agents for meeting volume targets. This type of contingent commission payment encourages agents to place enough business with an insurer so that the insurer can capture economies of scale in its agency relationships. These might also allow insurers to benefit from portfolio diversification by adding more risks to the portfolio (Cummins and Doherty, 2006).
The second type of agency cost arises from asymmetric information between the potential insured and the insurer. The underwriting process is designed to determine the risk class of a potential insured and apply the appropriate price, such that actual losses are close to expected losses. Adverse selection can arise when a potential insured has private information about her risk type that would be costly or difficult for an insurer to verify directly. To minimize adverse selection in this case, the insurer may rely on an agent to participate in gathering and verifying risk information. Regan and Tennyson (1996) and Regan (1997) argue that the optimal distribution method depends on both the complexity and underlying risk associated with the exposure. When the risks are relatively standard and easily assessed, direct sales, including exclusive agency will be preferred. Conversely, when the agent's information is important for proper risk-classification, independent agency will be preferred. Commissions that are linked to underwriting profitability can be used to induce independent agents to gather and verify risk information so that risks are priced properly.
In addition, profit-based contingent commissions can be used to prevent agents from moving profitable business to other insurers. Agents possess valuable information regarding policyholder risk type based on the revelation of information over time. The agent who represents more than one insurer is in a position to extract rents from the insurer by using this information and threatening to move business to a competing insurer (D'Arcy and Doherty, 1990). This is a credible threat because the agent is in possession of valuable private information about client risk types. Insurers that want to prevent this can use profit-based contingent commissions to reward the agent for profitable renewal business. This also supports the argument that contingent commissions provide incentives for agents to reveal risk information.
Insurance agent compensation schemes thus consist of two parts. The first is a flat percentage based on premium volume. There may be different flat rates paid for new and renewal business. Contingent commissions are paid in addition to flat commissions, and may be based on profit, volume, retention, or business growth, and are typically some combination of these (IIAA, 1991). Cummins and Doherty (2006) note that the most important component is profit-based. Contingent commissions are not payable on a per risk basis, but are allocated based on the performance of the entire portfolio of business placed with a particular insurer. The contingent commission schedule is known to agents at the beginning of the period, but contingent commissions actually earned are calculated some period after business is placed and loss experience is observed.
Both flat and contingent commission schemes differ across insurers and across lines of business. The majority of compensation to agents is in the form of flat commissions. Commissions and brokerage fees represented just over ten percent of net premiums written in 2005, and 41.4 percent of insurer underwriting expenses. Contingent commissions accounted for almost nine percent of total commission payments, but just 0.92 percent of premium volume in 2005 (AM Best, 2006).
Eliminating contingent commissions may have adverse consequences for insurance prices if it means that insurance agents have less incentive to screen risks and match clients with appropriate insurers.[8] If it is more costly for insurers to verify risk types directly, then insurers may pass this increased cost onto consumers. If the insurer cannot distinguish between good and bad risks, it may set prices too low, resulting in higher than acceptable loss ratios. Prices would then be likely to increase for all exposures in the next period, regardless of true risk type. Alternatively, insurers may simply place all risks into the "high risk" category. In this case the true high risks would be appropriately priced, but the lower risk types will be over-charged. Finally, insurers may restrict availability of coverage in lines of business where the agent's role in risk classification is particularly important, such as complex commercial lines.
If contingent commissions are effective at aligning incentives for risk classification and placement between insurers and agents, then there should be a better match between insurers and client risk types. Pricing should more closely reflect true risk type, and thus the risk of the insurer’s loss portfolio should decline. However, Cummins and Doherty (2006) show that commissions paid by insurers are mostly passed through to policyholders. The combined ratio is a measure of insurer underwriting profit for the current year, and includes both losses and underwriting expenses, including commission expense payments. A combined ratio greater than one indicates that insurers paid out more in losses and expenses in the current period than they collected in premium revenue. One might expect that the payment of contingent commissions would increase insurer underwriting expenses. Thus, even though they might lead to lower loss levels and less variability in losses, contingent commissions that result in higher prices overall would not be a benefit to consumers. A mitigating factor, however, is that insurers must offer attractive price/quality coverage options to maintain a competitive position in the market. Therefore, it might be the case that insurers that use contingent commissions reduce underwriting expenses in other areas so as to reduce overall price levels.