Less Talk, More Action NeededTo Improve Capital Management

Over the past few years, there has been considerable discussionover the issue of capital management.

For several years leading up to 2001, the industry was concernedabout overcapitalization: too much capital chasing too little riskmade for soft pricing and poor returns to shareholders. In the wakeof recent catastrophes, discussions have turned to possibleundercapitalization.

This points out the fact that capital management is not a staticissue. Both risk exposures and available capital differ from firmto firm, and can change quickly over time.

The problem is that there is more talk than action over theissue of capital adequacy. What the industry needs is a moredynamic approach to capital management.

Effective capital management involves measuring required capitalbased on risk, and comparing it to available capital. Dynamiccapital management involves updating calculations of requiredcapital and measures of available capital routinely, and especiallyin the event of significant changes in the market.

An overcapitalized firm will earn sub-par returns forshareholders because the excess capital earns a risk-free rate ofreturn–hardly what an equity investor expects. An undercapitalizedfirm runs too high a risk of bankruptcy in the event of a largerisk event.

Stock companies need to worry about overcapitalization becauseshare prices have suffered as returns-on-equity languished in midsingle-digits during the period of overcapitalization in the late90s. Mutual insurers may seem less affected, but they should worryabout it because their mutual policyholders will figure out, overtime, that the they are getting sub-par returns from the risks theyare taking.

Insurers that are potentially undercapitalized in the wake of alarge catastrophe need capital measurement models just as much.They need to know, on an economic basis, just how much capital theyneed to raise, reflecting both new information about riskexposures, and their diminished amount of available capital.

This article recommends an approach to dynamic capitalmanagement that is based on a simple three-step process:

Assess the capital required to cover a firms risks.

Compare actual capital to required capital.

Choose among a set of alternatives to align actual with requiredcapital.

There is general agreement on the basicrationale by which capital requirements should be determined.Insurance companies hold capital to provide debt-holders and/orpolicyholders with a financial cushion against potential losses.The bigger the risk, the bigger the capital cushion that the firmrequires if it is to maintain its financial strength. Conversely,the higher the firm's desired financial strength, the more capitalthe firm needs for a given level of risk.

There are three basic types of capital measurement models:rating agency models, custom-built models, and vendor models. Let'sconsider these in turn.

Rating agency models and industry rules of thumb areinsufficient for capital management.

In the past, many insurers relied on the National Association ofInsurance Commissioners or rating agencies' models when determiningtheir capital requirements. Since the NAIC sets only minimumstandards for capital adequacy, insurers began to use multiples ofthe NAIC requirement as a target for capital adequacy. For example,an insurer looking for an A-rating might target 2.5 times the NAICrisk-based capital.

Other firms relied on rating agency capital models for settingcapital levels in line with their target debt ratings. These modelswould use capital-to-premium ratios to calculate requiredcapital.

And some firms relied on simple industry rules-of-thumb, such asrequiring a premium to surplus ratio of 2:1.

Over time, however, firms began to realize the drawbacks ofthese approaches:

Rating agency models are too general. These models incorporateindustry-wide data and their results therefore address ahypothetical average firm. These models are necessarily “top-down”models because of their data limitations and so each firm isevaluated against the average risk of the industry, rather thanaccording to its own risk profile.

Rating agency models are not granular enough. Because thesemodels are designed to function in the aggregate, they cannot beused to determine the capital requirements needed for profitabilitymeasurements at the business line, product or customer levels.

Models that use a uniform premium-to-surplus ratio across alllines assume that all risks are priced appropriately.Unfortunately, this is never the case in the real world. Use of aconstant ratio also ignores the fact that the volatility and payoutpatterns of losses vary significantly across different lines ofbusiness. Differences in reinsurance programs and structures alsomake it difficult to apply a standardized premium-to-surplusratio.

Custom models are better, but can be difficult to implementin-house. A number of insurers have tried to build capitalallocation models from scratch, but many have found the developmentprocess to be longer and more costly than they imagined. The reasonis that it is difficult to assemble the right resources to get thejob done, such as:

A multi-disciplinary team.

Building a good enterprise-wide capital model requires financialengineers, actuaries, modelers/programmers and businessmanagers–all costly, scarce resources. There aren't many ways tocut corners, either: a team that lacks a financial engineer willlikely get the asset risks wrong; a team that lacks an actuary willget the liability risks wrong.

Benchmark information.

In practice, every company finds that it is missing some of theinternal data needed to perform an accurate risk assessment. Thepragmatic solution is to use a benchmark/placeholder based onindustry data, and then refine the parameters in the future as dataare gathered.

Validation.

There are so many moving parts in a good capital model that itis difficult to know when you've got it right. To trust theresults, it is important to test the model against an externalbenchmark. For example, some modelers run their models on theindustry as a whole to verify that they come up with reasonableestimates.

Corporate buy-in.

As the name implies, an enterprise-wide capital model impactsall facets of the business. Thus it is essential that commitment tothe model be received from the chief executive officer, chieffinancial officer, chief underwriter and chief actuary in order forthe model to be implemented successfully.

Vendor models are on the rise. In response to the challenges ofdeveloping a custom model, several firms are offering models thatallow insurers to calculate capital requirements. These models arecollectively known as “dynamic financial analysis” models, and theycome in two broad types: accounting-based simulation models andeconomic-based analytical models.

The bottom line is that the accounting-based simulation modelsare more flexible in generating output–the user can choose to lookat many different views of the financial statements. However, thispower brings responsibility: the user must know both which viewsare meaningful and how they can be produced. This meanssimulation-based models can be difficult to implement and use.

Economic-based analytical models are designed to calculate“economic capital.” They provide support for capital-relateddecision-making, such as return-on-equity targets, reinsuranceoptimization and risk-based pricing. (See related story.)

Because these models are designed to calculate capitalrequirements, they are generally easier to implement. ERisk, RiskManagement Solutions and Oliver, Wyman & Company together havedeveloped one such “risk-adjusted return on capital” model, P&CRAROC, in which the methodology for assessing the capital requiredto cover a firm's risks can be described by the four steps outlinedbelow:

Define risk as the distribution of changes in value over aone-year time period.

Identify and characterize risk distributions.

Using proprietary data (or industry benchmarks, if data isinsufficient), model the risk distribution at the desired level ofgranularity (say, lines of business). Each distribution is modeledusing the best available techniques from the banking and insuranceindustries (credit risk analytics from banking, catastrophe riskanalytics from insurance, for example)

Combine stand-alone risks using correlations.

Underwriting uncorrelated or partially correlated risks providessignificant diversification benefit. Thus the risk distributionsmust take into account the correlations of the individualrisks.

Using this combined risk distribution, and a company's solvencytarget (A, AA or AAA), quantify the total capital required tosupport all risks assumed by the firm.

Calculate contributory risk by activity.

The total economic capital required to support the firm is thenallocated back to the individual activities (in this case, lines ofbusiness) based on each activity's relative contribution to theoverall risk.

Having determined the economic capital required to support allof a firm's risks, the next step is to compare this figure with itsavailable capital, as well as the capital needed to satisfy ratingagencies or regulators (say, 2.5 times the NAIC risk-based capitalnumber).

Note that adjustments to GAAP equity are needed to get ameaningful comparison of economic and available capital (such asdiscounting reserves and marking the investment portfolio tomarket). Note also that the rating agency capital requirement maybe quite different from the economic capital requirement (forreasons cited earlier).

Comparisons of economic capital with capital levels required byrating agencies or regulators, and comparisons of economic capitalwith available capital, will bring a p-c insurer to the next andcritical step of the capital management process–deciding what to dowith the excess. (See related story.)

The concept outlined in this article is simple: measure capitalrequired as a function of risk. Compare this to available capital.That's the first part of the process.

The next part of the process seems simple also: If you have toomuch, either give it back to shareholders or invest itproductively. The payoff is dramatically better returns on equitythan the competition.

While the entire process may be simple, putting it into practicehas hitherto been more of a challenge. But now that bothsoftware-based and consulting- based solutions for measuringrisk-based economic capital are becoming widely accepted, insurershave only their own inertia to blame for the industry's capitalmanagement problems.

Peter Nakada is Executive Vice President and head ofconsulting at ERisk in New York. Peter Ulrich is Managing Directorof the Capital Management Group at Risk Management Solutions, inNewark, California. Ugur Koyluoglu is a Director in the RiskManagement Practice at Oliver, Wyman & Company in NewYork.

Reproduced from National Underwriter Property &Casualty/Risk & Benefits Management Edition, November 5, 2001.Copyright 2001 by The National Underwriter Company in the serialpublication. All rights reserved.Copyright in this article as anindependent work may be held by the author.

Contact Webmaster