Deciding “What Should We Do?” in the Risk Business

Risk models can be used primarily to answer two very important questions for an enterprise whose primary activity is the risk business.

1. How did we do?
2. What should we do?

The “how did we do” question looks backwards on the past, usually for 90 days or a full year.  For answering that question properly for a firm in the risk business it is absolutely necessary to have information about the amount of risk that the firm is exposed to during that period.

The “what should we do” question looks forward on the future.  The proper time period for looking forward is the same as the length of the shadow into the future of the decision.  Most decisions that are important enough to be brought to the attention of top management or the board of a company in the risk business have a shadow that extends past one year.

That means that the standard capital model with its one year time frame should NOT be the basis for making WHAT SHOULD WE DO? decisions.  That is, unless you plan on selling the company at the end of the year.

Let’s think about it just a little bit.

Suppose the decision is to buy a laptop computer for the business use of one of the employees of an insurer.  You can use two streams of analysis for that decision.  You can assume that the only use of that computer is what utility that can be had from the computer during the calendar year of purchase and then you plan to sell the computer, along with the rest of the company, at the end of the calendar year.  The computer is valued at the end of the year at a fair market value.  Or you can project forward, the utility that you will get from that employee having a computer over its useful life, perhaps three years.

The first calculation is useful.  It tells us “HOW DID WE DO?” at the end of the calendar year.  But it not a sensible basis to make the decision about whether to buy the computer or not.  The reason for that is not because there is anything wrong with the calendar year calculation.  In theory, you could even run your company by deciding at the end of each calendar year, whether you wanted to continue running the company or not.  And then if you decide to continue, you then must decide whether to sell every laptop or not, and similarly to sell every part of your business or not.

Most companies will automatically make the decision to continue, will not consider selling every part of their company, even if they have gone through the trouble of doing a “for sale” valuation of everything.  That approach fits better with Herbert Simon’s “Satisficing” idea than with the theory of maximizing value of the enterprise.

But from a less theoretical point of view, putting absolutely everything on the table for a decision could be very time consuming.  So what most companies is to imagine a set of conditions for the future when a decision is made and then as the future unfolds, it it does not deviate significantly from those assumptions, decisions are not reopened.  But unfortunately, at many companies, this process is not an explicit conscious process.  It is more vague and ad hoc.

Moving away from laptops to risk.  For a risk decision, first notice that almost all risk decisions made by insurers will have an effect for multiple years.  But decision makers will often look forward one year at financial statement impact.  They look forward one year at a projection of the answer to the “How DID WE DO? question. This will only produce a full indication of the merit of a proposal if the forward looking parts of the statement are set to reflect the full future of the activity.

The idea of using fair value for liabilities is one attempt to put the liability values on a basis that can be used for both the “How did we do?” and the “What should we do?” decisions.

But it is unclear whether there is an equivalent adjustment that can be made to the risk capital.  To answer “How did we do?” the risk capital needed has been defined to be the capital needed right now.  But to determine “What should we do?”, the capital effect that is needed is the effect over the entire future.  There is a current year cost of capital effect that is easily calculated.

But there is also the effect of the future capital that will be tied up because of the actions taken today.

The argument is made that by using the right current year values, the decisions can really be looked at as a series of one year decisions.  But that fails to be accurate for at least two reasons:

• Friction in selling or closing out of a long term position.  The values posted, even though they are called fair value rarely reflect the true value less transaction costs that could be received or would need to be paid to close out of a position.  It is another one of those theoretical fictions like a frictionless surface.  Such values might be a good starting point for negotiating a sale, but anyone who has ever been involved in an actual transaction knows that the actual closing price is usually different.  Even the values recorded for liquid assets like common equity are not really the amounts that can be achieved at sale tomorrow for anyone’s actual holdings.  If the risk that you want to shed is traded like stocks AND your position is not material to the amounts normally traded, then you might get more or less than the recorded fair value.  However, most risk positions that are of concern are not traded in a liquid market and in fact are usually totally one of a kind risks that are expensive to evaluate.  A potential counterparty will seek through a hearty negotiation process to find your walk away price and try to get just a litle bit more than that.
• Capital Availability – the series of one year decisions idea also depends on the assumption that capital will always be available in the future at the same cost as it is currently.  That is not always the case.  In late 2008 and 2009, capital was scarce or not available.  Companies who made commitments that required future capital funding were really scrambling.  Many ended up needing to change their commitments and others who could not had to enter into unfavorable deals to raise the capital that they needed, sometimes needing to take on new partners on terms that were tilted against their existing owners.  In other time, cheap capital suddenly becomes dear.  That happened when letters of credit that had been used to fulfill offshore reinsurer collateral requirements suddenly counted when determining bank capital which resulted in a 300% increase in cost.

RISKVIEWS says that the one year decision model is also just a bad idea because it makes no sense for a business that does only multi year transactions to pretend that they are in a one year business.  It is a part of the general thrust in financial reporting and risk management to try to treat everything like a bank trading desk.  And also part of a movement led by CFOs of the largest international insurers to seek to only have one set of numbers used for all financial decision-making.  The trading desk approach gave a theoretical basis for a one set of numbers financial statement.  However, like much of financial economics, the theory ignores a number of major practicalities.  That is, it doesn’t work in the real world at all times.

So RISKVIEWS proposes  that the solution is to acknowledge that the two decisions require different information.

Advertisement

Capital Allocation – Different Questions

RISKVIEWS has been confused by the vehemence of some people about the topic of capital allocation.

Some people feel that capital MUST be allocated to facilitate proper management.

Other feel that capital MUST NEVER be allocated because it leads to incorrect decisions.

But RISKVIEWS suspect that they may be talking about two different questions.

Those who think that they MUST allocate capital are trying to answer the question “How DID we do?”

Those who think that they MUST NEVER allocate capital are focused on the question “What SHOULD we do?”

Of course, the two questions often get mixed up.  But one is about the past and the other one is about the future.  The problem that folks who object to capital allocation are afraid of is that if capital is allocated for the purposes of answering the “How DID we do?” question, then the same sort of allocation will be used to answer the “What SHOULD we do?” question.

And that IS a problem.  The “What SHOULD we do?” question needs to be answered with projections of the future.  Many decisions that are worth worrying about do not settle within a single year, so the projections need to be multi year.

But the problem that they are worried about is the problem of making a multi year decision with a single year projection.  Whether capital is allocated or not, that is a poor way to go.

Multi year decisions need multi year projections.  The multi year capital impact needs to be included.  That can be done with a cost of capital factor or be a carefully constructed model that reflects capital inflows and outflows and then implicitly charges a cost for capital held.   The multi year calculation usually needs to be discounted at an appropriate risk adjusted discount rate.

RISKVIEWS rule of thumb for selecting a discount rate is that all risks should be included ONCE and only ONCE in the entire calculation.  So if the calculation is a stochastic one that includes scenarios that reflect the possible adverse effects of a risk, then the discount rate should not also include a charge for that risk.  If your projection includes ALL possible risks, then a risk free rate is an appropriate discount.  Remember that the market charges a risk premium for its perception of emerging risks.  And for the risk of strategic failure.

So RISKVIEWS concludes that there is no harm from allocating capital.  There is a harm from making multi year decisions with a one year projection.  Whether or not capital is allocated.  And multi year decisions need to include the effect of capital usage.

Are you using Enron Accounting?

Over 10 years ago the high flying ENRON corporation came crashing down.

One of the things that was responsible for their high flying and for the delayed recognition of their actual underlying troubles was their accounting. Arthur Andersen, their external auditors, went up in smoke 10 years ago also over related issues.

At its heart, Enron’s problems were caused by greed and a 65 multiplier.  Their stock was valued at 65 times earnings at its peak.  So a $1 increase in earnings meant a $65 increase in stock price.

And earnings were determined by accounting rules.

In Enron’s case, they were taking an approach to long term projects that fronted profits to an early stage in the life of the long term contract.  They often reported present value of all future profits from a new project when the project was contracted.  A long term energy delivery contract would have 5 or 10 years of future projected profits reported when the deal was closed.

This was called a mark to market approach.  But in fact it was almost always a mark to model.

But both systems can lead to risk management problems.  The fronted profits may or may not emerge. Some of these contract’s value were subsequently marketed down as market conditions changed in future years.  And in Enron’s case, the later adjustments became more and more on the down side because of optimistic original booking.  This then put pressure on management to find a larger and larger quantity of “new” deals where they could report large up front and possibly inflated pv of future profits.  Eventually it crashed like the ponzi scheme that it was.  They got too big.  They just couldn’t find enough new deals to paper over the losses from write downs adjustments to previously deals.

There are two other choices to accounting for long term transactions.  The profits could be recognized over the life of the deal or the profits can be recognized when the deal is completed.  That later approach seems so, so dated.  How 19th century. Let’s call these three types of accounting Type E, Type L and Type C.

But the risk manager needs to understand the accounting treatment of every deal that is not closed in the same accounting period that it is opened.

• Type E accounting means that the company probably put up an asset for future profits.  The risk capital that should be associated with that asset may well be very high.  The firm is at high risk that that asset will not be worth what the books say.  Proponents of this approach say that all it takes to get this right is the correct discount rate.  With the right discount rate, the E approach is more or less the same as the L approach.  That discount rate is the rate that makes the activity break even at inception.
• Type L accounting means that in some cases, the company is still setting up an asset, albeit small, at the inception of the deal, since in many cases there are either more expenses or less revenues at the outset of a long term project than there are over the life o the deal on the average.  That is the usual objective of Type L accounting, to make every year look like an “average” year on an expected basis.  This smoothing also carries some accounting risk, but much less than Type E.
• Type C accounting means that profits are realized when cumulative revenues exceed cumulative expenses on the deal.  Deals that are expected to be profitable (and what company enters into long term deals that are not?) are usually reported as losers at the outset.  You could argue that this is just another discount rate applied to the future cashflows.  But this discount rate is infinite, making future cashflows worthless until they happen.

RISKVIEWS opinion is that one approach is not necessarily superior to another.  However, if the financial statement contains various projects that are not treating the future consistently, then there will be major distortions of perception about results.  That distortion is itself a major risk.

Marking Risks to Market

If financial statements are set to mark to market, why aren’t they marking uninsured risks to market?
Under all accounting systems, a business that buys no fire insurance will show a better result then a similar company who is buying insurance. Except in the year when they have a claim. The market price for their risk is an insurance premium.  But for some reason, risk has never been treated in this way.

If risk was market to market, then a firm that buys no insurance, or does not hedge a risk would not report a gain, they would need to put aside an amount at least equal to the insurance premium. That amount could be put into a fund and released when they have an event that would have generated an insurance claim.

Of course, to be mathematically correct, they would need to make adjustments to the insurance premiums. One to remove the profit margin/risk charge in the premium and another to reflect the fact that they are in effect creating an insurance pool with one participant which appropriately replaces the risk charge.
An insurance pool with one participant? That doesn’t make any sense. But that is what a business who is not buying insurance is doing. What then would be the correct premium, not loaded for profits, for an insurance pool of one? The pool would have to bare the cost of holding capital (or a contingent capital facility) for the entire maximum claim amount to the extent that amount exceeds the reserves (or the amount in the pool).
So if the cost of capital is 3%, and the claims rate is 1%, then the mark to market cost would be about 400% of expected claims at first, declining as the fund builds up.
Pretty expensive. But that would make the financial statement make sense on a mark to market basis for risk.
This approach could be applied to unhedged risks as well. The mark to market accounting is actually much too lenient on hedgable risks that are unhedged. The MTM accounting in effect allows those companies to reflect the cost of hedging even if they are not hedging. In fact, when they do not hedge, they are self insuring and need to reflect a much higher cost as described above.

Not managing risk is expensive, particularly to investors.  Investors deserve appropriate information on risk.  The longstanding accounting paradigm that ignors risk gives investors the exact wrong information and needs to be immediately corrected.

One of the main reasons that risk management is not already completely embedded in all firms is that they can get away with this scam on their investors, supported by their accounting statement.

Risk needs to be accounted for properly, especially when it is not managed.

Reliance on Risk Management

Many life insurance firms may not really be aware of the degree to which they are exposed to risk.

When these firms write a life insurance policy, they are immediately exposed to a significant amount of gross risk.  Looking at the entire liability book, the risk is immense.  Many multiples of capital.

I  am not talking about the fact that face amounts of insurance far exceed premiums.  What I am trying to point out is that there is a very large amount of risk created by accepting premiums with the guarantee of certain surrender values.  (There is somewhat more mortality risk there than many insurers may realize, but it is not significant on a gross basis compared to the interest rate risk on the cash values.)

Insurers tend to forget about this because there is a very longstanding practice of offsetting that risk by investing funds (called the assets) of the life insurer.

The folks who are insisting on market value accounting for insurance liabilities are trying to point out this fact of life.

In many markets, the insurer will then take investment risk – credit or market – with the investments and finally they will do something further that deeply offends the market value folks.

They will split some of the money that they are paid in risk premium with their policyholder/customer.

This practice can probably be traced back to the time when the predominant form of life insurance was mutual life insurance.  Under that structure, the policyholder is thought to share the risk of the insurance company, and it therefore makes sense that they would share in the risk premium.

Non-mutual firms found that they could not compete with this because most customers did not understand that they had the choice of one level of return within their insurance policy at a certain level of risk and a lower level of return with a lower amount of risk. The customers usually just saw the net return.  Risk was not communicated well.  Usually risk was communicated very vaguely while return seemed to be really tangibly conveyed.

So what the market value folks are trying to accomplish is to overcome hundreds of years of confusion about the actual level of risk of an insurer.

You see, risk premiums are usually collected in advance of losses.  If an insurer is paying some fraction of its risk premiums to its customers, and it does not have a loss sharing mechanism as is fundamental to a mutual insurance scheme, then it is acting similarly to a leveraged hedge fund.

The resources of the insurer to absorb losses is the capital, but the exposure to losses extends to a much larger pool of insured funds.

So the market valuing of insurance liabilities is really a risk recognition exercize.  It is trying to make a point, that point being that the practices of insurers have evolved to become much riskier than what they had been in the past.  And the mark to market system would force insurers to acknowledge that additional risk at the point at which they decide to tak on the risk.

Now, it appears that IFRS accounting is heading a different direction.  The IASB seems to be backing away from a full mark to market system for assets.  This will wreck havoc on the balance sheets and income statements of the insurers who will be marking their liabilities but not their assets to market.

Sort of like the mess that has existed in the other direction for some time not, were insurers in many situations have been marking assets, but not liabilities to market.

Insurance has a reputation for totally opaque financial reporting.  It seems that this reputation will continue to be well deserved.