At the height of the last crisis, the market value of the assets of insurance companies fell sharply while the present value of their liabilities remained essentially unchanged. Under recently proposed insurance regulations, similar events might result in insurance firms ending up in breach of regulations, thus requiring them to increase capital quickly to avoid official interventions.
Since the worst time to increase capital is during crises, the insurance industry and policymakers have been searching for ways to minimise the impact of the future shortfall of this nature, without either needing increases in insurance premiums or reducing profitability.
The problem arises because of a time-inconsistency problem between how assets are valued and liabilities are discounted. Requiring the industry to only hold highly liquid, safe assets could eliminate this problem. However, because such assets have lower expected returns than assets currently in the portfolios of insurance firms, the result would be a substantial increase in insurance premiums. Similarly, addressing the problem by forcing the industry to increase capital during a crisis event is equally ill-advised.
It is the search for alternative solutions to the problem that has led the industry to proposing discounting liabilities at a higher rate than currently is done by means of what it terms the “liquidity premium”. This approach has recently been endorsed by the committee of European insurance and occupational pension supervisors (CEIOPS 2011).
We argue in this column that the liquidity premium is not an appropriate solution as it is based on a misunderstanding of the nature of liquidity and a lack of understanding of the structure of the liabilities of the insurance industry.
The liabilities of insurance companies consist of uncertain future payments to clients. Other than for policies that may be surrendered or redeemed, insurance policies are not liquid in the same sense as bank demand deposits.
To ensure insurance companies can meet future payments as they occur, they are required to keep certain assets as reserves, typically in the form of fixed-income assets like bonds. Over time, competitive pressures in insurance markets have led to increasing holdings of higher-yielding but illiquid (or maturity mismatched) assets by insurance companies.
This arrangement has worked more or less well for over a hundred years. However, the recent crisis demonstrated a problem; the spreads on corporate bonds increased significantly – in many cases by hundreds of basis points – causing a sharp fall in the value of insurance assets. Some bond markets became thin or non-existent. By contrast, the present value of liabilities remained unchanged or even rose. Consequently, an insurance firm could find itself in serious regulatory difficulties even though no insured event has occurred, or credit loss been experienced.
In such situations, the ongoing process for reforming European insurance regulations, Solvency 2, allows the use of mark-to-model valuation of assets. It is appropriate for these models to include liquidity premiums, to the extent that these assets may need to be realised to meet insurance commitments. However, as it is practically impossible to reliably decompose bond spreads into credit and liquidity spreads, there is considerable room for abuse when using such models.
What about the liabilities?
The liabilities of insurance companies are future claims by policyholders. The industry can sometimes predict future claims relatively accurately, but in other cases there is much more uncertainty as to either or both the amount of claims and their timing. Direct calls for an insurance company to supply liquidity to a counterparty are a central part of their business. This may happen in any of three ways.
- Claims under the terms of a policy arising from the occurrence of an insured event.
- Claims under rights of policy surrender. Since valuations are usually lower than fair value, such events are not common.
- Claims arising from collateral requirements under policy or derivative credit support agreements, such as credit default swaps.
The final case is determined by the terms of a credit support agreement, and claims under it may be substantial as we saw with AIG. However, these claims can usually be met by providing assets (subject to haircuts) rather than simply money liquidity. Note that the provision of assets as collateral security does lower the insurer’s ability to borrow funds on a secured basis (such as repo). Though it is sometimes a condition of regulatory admissibility to a particular national market, the provision of collateral security as credit enhancement to their policies by insurers is the exception rather than the rule
Present value calculations
The risk-free discount rate applicable to derive present values for the liabilities of an insurance company is not the asset market derived risk-free, or zero coupon, government rate for the term of the liability. Instead, the relevant discount rate is the prospective risk-free cost of production of those liabilities. Such calculations are made under long-established actuarial methods.
The discount function used in present value calculations sets a schedule for the recognition of the amount of a liability over its entire lifetime; it can be thought of as an amortisation function. Two consequences follow from this simple observation.
- First, that the transfer or sale of a liability depends upon both the current price and also the future performance of the “buyer”. The insured has rights with respect to this future performance. This limits the transferability of liabilities and thus constitutes a barrier to exit for the insurer.
- Second, using market based discounted present values for liabilities and market values for assets means that any profit recognition prior to the full discharge of the liability is arbitrary.
In contrast to insurance assets, insurance policies are relatively unaffected by the presence of crises. One might see a higher incidence of fraudulent claims, but the potential for the creation of runs on the firm is limited. Consequently, discounting of liabilities is unaffected by crises or market turmoil.
This suggests that uncertainties in the amount and timing of liability cash-flows should be reflected in the estimates of those cash-flows, not in the discount rate used to reduce those liabilities to their present values.
Asset and liability mismatch
Insurance companies can and do maintain maturity mismatches between their policy contingent liabilities and the assets held as provisions for these. This is a form of credit creation.
The problem exercising insurers here stems from the mixed attribute nature of the accounting for assets and liabilities; discount functions versus market prices. When the markets lower the value of assets, the insurance company may face funding and regulatory difficulties if the present value of liabilities remains unchanged.
Liquidity is one of those terms that, in use, has a myriad of meanings. However, it is obvious that liquidity has a cost; if it did not all assets would be liquid. Both the academic and the practitioner literatures have made significant recent advances in providing workable definitions of liquidity. Perhaps the most common is due to the BIS (2006), which identifies liquidity as having two relevant and interrelated aspects: market liquidity and funding liquidity.
Markets reveal opportunities for exchange; reflecting the presence of buyers and sellers as well as their willingness and ability to supply or demand market liquidity. In turn, market liquidity is directly related to the risk tolerance of these other market participants, and if their willingness or ability is impaired, the market may become illiquid.
Some financial instruments serve well as stores of liquidity. Their defining characteristic is that they exhibit price-insensitivity to information, with government bonds a prime example. Markets for assets price-insensitive to information constitute the principal markets for liquidity. The best examples are government bond markets.
In this context, a liquidity premium can be thought of as an insurance premium for an asset’s market liquidity. By buying a lower yielding liquid asset rather than a high yielding but illiquid asset, the investor is paying a liquidity premium. Therefore, liquidity is an option not exercised by the hold-to-maturity long-term investor. Similarly, market liquidity risk is the possibility that liquidity premiums in a market may rise.
Uncertainty in payments, unless directly related to the possibility of the disappearance of market access, is not a feature of liquidity. Instead, such uncertainty is credit risk.
Liabilities and liquidity
The notions of market liquidity, as discussed above, relate to liquidity of assets for which there is a market. If there is no market, it doesn't make sense to talk about market liquidity. Insurance products are, by definition, illiquid. Indeed, if an insurance policy had a liquid secondary market, there would be no need for an insurance company.
The assets of insurance companies can typically be traded on financial markets, while the liabilities cannot, because the liabilities are the property or assets of those insured. These property rights are defined by the terms of the policy and include restrictions upon both the insurer and insured, such as the maintenance of an insurable interest. There is little market for the trading of insurance liabilities.
Of course, an individual firm or a business line can be sold by private treaty, but there are well-documented problems with the sale of whole firms and business lines. For example, because of the loss of value in senior debt obligations of firms to investors when these firms have been acquired by private equity and leverages buy-out houses, it is now standard for the contractual terms of these debt securities to contain restrictive covenants.
Consequently, the impact of liquidity on insurance liabilities is relatively minor, the main exception being certain types of insurance that the subject to runs.
For these reasons, one cannot think about the liquidity of insurance liabilities in the same way as financial assets, since the owner of a financial asset is usually free to dispose of this property at their discretion, which is not usually possible for an insurance company with their insurance liabilities.
The liquidity premium for liabilities does not make sense
The problem facing the insurance industry is that the value of their assets is subject to securities’ market forces, while the value of liabilities is not. The basis of the industry’s proposal appears to be that because their bond assets are subject to market forces via government rates, as well as credit and liquidity spreads, their liabilities must also be. If this were true, it would make sense to include a liquidity premium in the discount rate applied to liabilities. Unfortunately the premise is false.
The accounting identity that assets equal liabilities applies within the books of a firm, an inside sense, but it is not applicable to the assets and liabilities of any firm in any external sense.
Considering that the risk in insurance products is readily modeled by established actuarial methods, and the fact that such liabilities do not have an accessible market, it does not make sense to incorporate liquidity in the discounting of insurance liabilities. The liquidity premium for liabilities does not make sense.
The solution to the industry’s problem is not to inappropriately apply methods relevant to market assets for non-market insurance liabilities. By understating liabilities, it would misrepresent the shareholder position and hence render the value of their securities market prices suspect, which would undermine the use of market prices for assets in valuation more generally.
It also undermines the integrity of the regulations governing the industry; applying inappropriate methodologies signals a lack of credible commitment on the part of supervisors and the industry, and a lack of understanding by the supervisors.
The insurance industry is vulnerable to the problem of a shortfall in the value of assets relative to liabilities during times of crises. To solve this problem, the industry is lobbying heavily to be allowed to discount its liabilities at a higher rate during crises than currently allowed with the so-called liquidity premium.
This is inappropriate. For non-marketable assets like insurance liabilities, assets whose risk is well understood and easily modelled, liquidity risk, as commonly understood, is not applicable. If it was, there would be no purpose for insurance companies, since insurance could be directly and efficiently obtained in the financial markets
The liquidity premium solution to the crisis shortfall is not founded on correct theoretical and practical foundations and hence misunderstands the nature of liquidity. Consequently it undermines the integrity and effectiveness of the regulations governing insurance firms. For example, if adopted, one unintended consequence of discounting liabilities with the liquidity premium would be a further reduction in the value of market discipline in the regulatory regime. Other solutions should be sought for the problem with a few alternatives listed below.
For example, the use of mark-to-market accounting for assets supporting liabilities could only be applied in cases when these assets might be liquidated to meet claims. The remaining assets can be treated as held to maturity, negating the impact of liquidity risk.
Liquidity risk could even be ignored for all assets if expected current claims were covered by sufficient committed letters of credit, hence enabling all assets to be held to maturity.
We further could give the regulator discretion over the choice of applicability of mark-to-market, mark-to-model, and hold to maturity accounting regimes.
Finally, the regulators could allow insurance companies to discount of liabilities at the insurer’s prospective rate of return on capital – this is what determines the security of any insurance liability since it is their cost of production. Changes in this rate have meaning and do not distort the insurer’s financial statements.
BIS (2006), The management of liquidity risk in financial groups. Joint Forum of the Basel Committee on Banking Supervision.
CEIOPS (2011) Task Force Report on the Liquidity Premium.