Where have all the financial risks gone?
Professor Avinash Persaud
Chart 1: Defaults by debt and by number of issuers
Here is a puzzle. We are in a period of heightened economic risks. At almost $900bn, non-performing loans around the world are at a record high, having climbed 20% this year alone. This is not just the result of a few head-line-grabbing defaults like Argentina and Enron last year. Corporate difficulties are as widespread as they are deep. Last year, the number of issuers defaulting on their debts rose 65% to 220, another record. Equity markets are apoplectic. On average, they are down almost 50% since March 2000 and half of that drop has occurred this year. We have to go back to 1931 to find another consecutive three years of significant losses. It has been a knuckle-whitening ride down with record levels of equity volatility. Yet, despite all these records, despite the spilt blood on boardroom floors, despite a less certain, less secure world, banks in general and in the US in particular, are healthy.
Chart 2: Loans and leases as a percent of US bank equity and loan loss reserves.
In the US bad loans represent just 10% of bank equity, one of the lowest levels of bad debt for decades. Moreover, as a share of bank equity, non-performing loans are significantly less than reserves set aside for bad loans. This compares very favourably with the last two decades when the opposite was the case. With a few exceptions, notably in Japan and Argentina, the dichotomy of economic distress on one hand and a banking system in rude health on the other is common to many developed countries. During this economic and financial cycle, no major bank has yet had to be bailed out.
Bank supervisors and central bankers have put this down to three things: first, the collapse of computing costs which has allowed better collection and manipulation of data on loans and borrowers. Better data should lead to better control of risks; second, the development of new financial instruments that can be used to hedge risks and third, the diligence of supervisors - supported by the Basle 1 Capital Accord in 1988.
These forces have made an impact. Banking has been positively transformed by the collapse of computing costs and bankers are loyal members of the cult of quant. Banks are the biggest employers of mathematicians, modellers and physicists. New financial instruments have enabled banks to pass on risks to others. In a number of ways and in many instances, most notably through credit derivatives, the risks to banks of making loans have been sliced up, securitised and sold on - outside the banking system.
Chart 3: Global Credit Derivatives Market excluding asset swaps
Where have the risks gone?
The mirror image of the decline of risk on banks' balance sheets has been the explosion of the credit derivatives market. According to the September survey by the BBA (British Bankers Association) the global credit derivatives market has grown from a notional value of contracts in 1997 of just under $200bn to just short of $2000bn ($2trn) this year - a ten fold increase in just five years. BBA members are predicting that this rate of growth will be sustained for the next couple of years. This would mean that by 2005, the credit derivatives market would be as large as the annual amount of bank lending in the whole of the United States.
Banks are safer, they have sold on a significant slice of their risks, but this begs the question whether our regulators are too narrowly focused on banks. Drafting the new Capital Accord has consumed vast amounts of time and effort from regulators in recent years and yet it is primarily about banks and not about how risks move around between banks and other financial institutions. Today, with a generally more interconnected, more fluid financial system and with banks deliberately shifting risk elsewhere, this focus is out of date. Worse, it could lead to a misleading view of the robustness of the financial system.
Risks have shifted from banks; but where to? Has this reduced systemic risks or merely moved the location of their trigger point. Are these risks in a better or worse place than before? What are the wider implications for financial markets and economies of this risk transfer? These are the intriguing questions posed to me many months ago by my learned friend, John Olcay. The answers are the focus of this evening's lecture.
Financial innovation has enabled risks to be sliced and diced, and for these slices to be separated and traded on their own, or joined up with other slices and traded together. By itself, the new ability to mix and match risks is a positive development. Those who can best arrange a loan - perhaps because of historical relationships or local expertise - are not always best able to carry the loan on their balance sheets. By selling the slices of a loan they do not want to those that do, originators of a loan can arrange more loans than might otherwise be allowed by their balance sheet, or regulatory capital adequacy ratios, or internal risk management systems. This suggests that there will be greater specialisms in, and a more efficient allocation of, the individual components of risk; and that there will be more risk-taking.
If these risks are better matched and spread, the financial system could end up with more risk-taking and yet lower systemic risks. My father, Professor Bishnudat Persaud, taught me early on in my economics training that individual risk-taking is the life-blood of economic dynamism, but that it can also be quickly killed off by systemic crises. More individual risk-taking, but with more spread risks could lead to less systemic risks and that would clearly serve an economy well. I fear we have missed-out on this particular utopia, partly because the main motivation for the growth of credit derivatives had little to do with the spread of risks.
Chart 4: 5 year rolling average G7, 10-year bond yield.
In the 1990s, the average yield on a G7 government bond was 6%. Insurance companies felt at ease selling financial products like endowments, annuities and life insurance that required them to earn similar investment returns. But, proving that you really can have too much of a good thing, greater international trade, more independent central bankers and the arrival of EMU, all combined to lower government bond yields. By the late 1990s, G7 government bond yields had fallen below 4% and insurance companies could no longer meet their liabilities by investing in these low-risk assets. To obtain higher yields they, collectively and inevitably, had to move up the risk spectrum. From 1998, as this chart shows, their holdings of government bonds fell and their holdings of corporate risk, through bonds, credit derivatives and equities rose.
Chart 5: Equities as a share of financial assets held by US insurance companies.
Chart 6: Breakdown of market participants showing protection bought and sold
I must thank Michael Metcalfe and Brian Garvey for pointing this out to me.
Seen in this light, the explosion of credit derivates looks less like an exercise in sound risk-management and more a yearning for yield by insurance companies. I don't need to tell you that whenever financial institutions go after yield as a group, regulators should sit up.
I am not sure how often an someone has attempted an explanation of credit default swaps in public before, but I would like to go though an example very briefly so that when we come to focus on the key principles we do so with greater understanding. Like most things in life the profession makes them sound more complex than they really are, so that you feel you need to pay them for their advice. Lets take the example of a five year loan from Peabody Bank to Global Telecoms limited and that later Peabody Bank buys protection for this loan from Europe Re, a reinsurance company. Peabody Bank hopes to receive monthly interest payments from Global Telecoms and a final repayment of the loan in five years time. There is of course a risk that Global Telcoms might go bust and fail to repay the loan.
Figure 1: Credit Default Swaps
The yield that Peabody Bank earns on this loan, like most loans, can be broken down into the risk-free yield and an extra premium. This premium principally relates to both the probability of default and Peabody Bank's own risk-aversion. This premium is related to the premium of a credit default swap.
In a credit default swap, Peabody Bank buys protection from Europe Re, against a default by Global Telecoms. In the event of default, Europe Re agrees to pay Peabody Bank an amount equal to the loan. In return for removing the risk of default, Peabody Bank pays Europe Re a premium.
This is neat and clever. But let me focus on a more subtle yet powerful transformation that has occurred at the same time. When it was only on Peabody's balance sheet, the probability of default was assessed by Peabody's loan officers with private data from its long banking relationship with Global Telecoms about developments in the sector and changes to the company and management. This assessment was largely separate from daily movements in Global Telecom's share or bond price.
Europe Re does not have a long-term relationship with Global Telecoms. Its primary interest is in the premium it can earn on Global Telecom's default swap today and on another company in another sector tomorrow. When the credit default swap is on Europe Re's balance sheet it is treated as an asset whose value is estimated using public data. Daily movements in Global Telecom's equity and bond prices feed directly into the estimation of default. As the bond and equity price-falls, the probability of default rises, and Europe Re has to hedge its credit default swap. It could do by selling bonds if a liquid bond market exists, but by driving bond prices down it would directly increase the assessment of the probability of default - a vicious cycle. Selling equities is one step removed and insurers have ended up shorting equities.
But there is really no escape with this approach. As long as we are talking about markets that are not perfectly liquid, if you use publicly available data to assess risks and try to hedge these risks in public markets, you will always be chasing your own tail as markets rise or fall. Selling when the market is falling and buying when the market is rising. This is the portfolio insurance problem that led to such volatility in the 1987 stock market crash. The only way to hedge risks so that the hedge does not act against your own position, is to lay off risks in a distant, unconnected place. That is what insurance is all about. But it is not easy to find offsets for risks that move with the economic cycle. To do so would have limited the demand for credit default swaps and the amount of premium they would have earned.
Insurance companies adopted the more convenient view that credit default swaps were an asset with a toxic slice attached and they valued and hedged this slice using the secondary bond and equity markets. The moving of risks off bank balance sheets on to the assets of insurance companies, led to a greater convergence in the way risks are valued, traded, managed and hedged. Instead of their being two views of risk, one held by the loan officers and one by the equity markets, now one view, that of the markets, forms the other. This convergence resulted in a concentration of risks across time and markets, which in turn led to points of extreme volatility and dislocation. Banks are safer than in previous recessions, but the secondary markets, equities in particular, are far more volatile as this chart of the volatility of the S&P 500 index shows. [Let me take this opportunity to thank Chris McCoy for finding and helping me analyse and present the data I am showing you today.]
Chart 7: VIX
The lesson is that risks are not reduced by being spread between more institutions if they are treated in the same way. Diversity in behaviour, not in name plates, is key to the management and spread of risks across a financial system.
Is more equity volatility a bad thing?
So risks are not better spread, they have just moved. But perhaps they have moved to a better place. What is wrong with a bit of equity volatility? There are a few reasons why we should be worried about this volatility, even in markets that are big and where hedging instruments exist.
First, equity markets have direct economic impact. Their performance has an impact on the cost of capital. When the cost of capital is volatile this is an obstacle to investing at the right time. Second, extreme volatility also adversely impacts business confidence that dents those animal spirits and the desire for investment in the first place. This has played a part in the weakness of business confidence in the Eurozone, where the economy does not have the same imbalances as in the US, but where there is now an equal dearth of business confidence.
Third, the costs of this added volatility are increasingly falling on pensioners, a group not particularly well equipped to deal with financial dislocation. This is not simply because of some poor investment decisions. You can never fully protect people from the misfortunes of their own poor judgement. But given the choice, retail investors will always have too much invested in the equity market.
Chart 8: Given the choice, individuals put too much risk into their pensions
The reason for this is that retail investors are liquidity constrained. By and large, you and I cannot walk into a bank and borrow against our future income without putting up collateral like a charge on our house, unless of course only very small amounts are involved. The only way individuals can borrow against the future is to leverage their pension, in other words to invest in riskier assets in the hope that they would go up to such an extent that they will be able to get the same pension with less contributions. The way to leverage your pension is to put too many equities or low grade bonds in your portfolio, and that is what they do. This is why the asset allocation of professionally managed defined-benefit pension funds has a much lower equity content that that of funds where individuals choose how much to invest and what to invest in. Whenever they are given the choice, retail investors leverage their pension with too much equities.
Given this in-built proclivity and given the trend towards more private choice in pension provision, retail investors will be increasingly exposed to the volatility of the equity markets.
When risks surface today, the banks don't get into trouble, but the employees' pensions do. Is that better?
At the very least, increased volatility and risks in the equity markets will be a disincentive for individuals to save more themselves and to become less dependent on the state.
Let's turn briefly to the wider market implications of what I have discussed. Banks have shifted credit risks to insurance companies, which have hedged themselves by going short the equity markets, which has significantly added to their volatility. There is an important and interesting cross-border element to this story too.
Figure 2: Table A: Bank lending. Table B: Largest life insurers, general insurers and re-insurers globally by revenue at end-2000
The credit derivatives market affects European institutions and markets more than US markets. This is because in the US the equity markets are more developed and larger, and so US companies rely less on bank lending and insurance companies. As the table documents, Euro-area bank lending is twice the size of bank lending in the US, and European institutions are the world's big insurance and re-insurance players. Both the supply and demand sides of this market are bigger in Europe. One of the reasons why US banks look particularly safe despite the concentration of corporate downgrades in the US is that US banks have transferred some of their credit risks to European insurers.
If I am right, that credit derivatives have been hedged, in part, in the equity markets then equity volatility would be markedly higher than normal and if I am right that this is more a European phenomenon with European banks shedding credit risks to European re-insurers, then European equity markets will be weaker and more volatile than US markets, even though the biggest downward revisions to growth expectations have occurred in the US. And this is what we see.
Chart 9: US and German equity market, year-to-date, and VIX and VDAX, year- to-date
The credit derivatives market and its use by insurance companies is the glue which connects the surprising safety of US banks, the surprising volatility of European equities and the surprising weakness of European bancassurance groups.
My analysis suggests there will be some symmetry. If the equity markets anticipate economic recovery and turn around, insurance companies hedges for their corporate default swaps will hurt, be they explicit or implicit short equity positions, and there will be a scramble to remove them, adding further upward pressure on the market. Underperformance on the way down will be met by outperformance on the way up and sustained levels of volatility.
The whole process will sap insurance companies' appetite for credit default swaps. They bought them at tight prices and then their risks appeared to rise as equities fell. Their hedging pushed the market against them and once they are fully hedged, the eventual recovery in the equity markets will cost them again. These losses may refocus market participants on the risk management uses of credit derivatives and not their risk-taking uses. This is probably no bad thing, but it does suggest that the credit derivatives market will not continue to grow as rapidly as those in the BBA survey hoped. Also, less enthusiasm on the part of insurers to own credit risks will reduce the capacity of the banks to lend, unless the banks decide to put more risk onto their balance sheets again. They may do so at some point, but in the near-term, given a weak economy and losses on those credit default swaps held back on their balance sheets, European banks will be in risk-reduction mode and unwilling to increase their risks.
The worrying thing is that this scenario makes Europe look more and more like Japan, with its cautious monetary policy, a constrained fiscal policy and, to top it all, risk-averse bankers. There are a number of key differences, not least that the ECB's focus on broad money growth and inflation may stop it from falling into the deflation trap. But it is certainly the case that the poor experience from the transfer of risks from US and European banks to European insurance companies adds to the risk of deflation in Europe.
Figure 3: Asset prices during deflation
Deflation does not bode well for real assets like equities, but it does support nominal assets like bonds and currency. This chart looks at the performance of Japanese asset markets since inflation turned negative.
Looking ahead, what should regulators do? In the case of pensioners, there needs to be some disincentive to gamble with your pension. It may be that something like the minimum funding requirement made more sense than we thought and should be reintroduced, certainly for those instruments that receive preferential tax treatment.
It is tempting when looking at the folly of what insurance companies have been up to, to consider what regulation could be adopted to avoid this reckless pursuit of yield, but I suspect going forward the best discipline will be provided by the salutary experience of losses which have led to more than a few boardroom shuffles. In insurance as in life, fear of failure provides the strongest motivation to succeed.
There is a danger however, that new regulation of insurance companies actually promotes this behaviour. The essential problem remember is that insurance companies were taking on risks that they were valuing using public data, and trying to hedge any shift in these risks using public markets. The is the portfolio insurance problem because it leads you to sell when markets are falling and buy when markets are rising. The solution is that insurance companies only take on default swaps they can self-insure or where they cannot, they hedge their risks, one step removed from daily market prices. A simple way to do this is to have an approach to risk-management whicht is less sensitive to daily price changes, but instead considers price changes over longer periods of time such as a year or more.
Chart 10: Probability of low within and end-horizon.
This chart looks at a portfolio of Japanese bonds and equities over a ten-year horizon. It asks what is the probability that at the end of ten years the portfolio has lost 10% or what is the probability that at any time during those ten years there was a 10% loss. Arguably an insurance company with a ten year liability is interested in the end-horizon probability, but by using daily price changes in its hedging it is actually responding to the within horizon probability, which means it is selling when the market is falling more often than it needs to. A credit default swap is not quite the same as a long-term liability, it is a contingent liability, but the point of this example is that there is great benefit to the public markets, to insurance companies and those of us paying insurance premiums, if insurance companies treated longer-term liabilities as longer-term liabilities and did not adopt the same risk-management process as someone managing a daily liability like a bank. I labour that point because this is, in fact, the thrust of the current approach to insurance company regulation - that they should manage risks in the way the banks do - an approach which is at the root of the disfunction we have been talking about today.
Despite record corporate bankruptcies, weak economies and a market meltdown, banks are generally safe. Regulators and central bankers are patting themselves on their backs. But risks are not better spread, they have just shifted from banks to insurance companies and from there to equity markets. It is not clear to me that this is a better place for these risks to be. With public pension provision falling, individuals are increasingly exposed to the equity markets. Now when major firms fail, it is not the banks that come crashing down, it's employees pensions. Is that better? Banks may be safer, but the financial system is not, especially from the perspective of individual investors and this background will not help in the struggle to increase individual savings.
There is an interesting cross-border dimension too. London is the trading centre for the global credit derivatives market. European re-insurance companies have been the biggest sellers of protection and US banks some of the biggest buyers. This has made US banks safer than otherwise, European bancassurance groups weaker and European equity markets more volatile. This will undermine the capacity and appetite for bank lending in Europe, which draws some unhappy parallels with developments in Japan. It certainly adds to the risks of deflation in Europe, which will cap the bounce in European equities, but will support European bonds and the euro.
One of the key lessons to be learned from this experience is that in today's fluid financial markets, the spread of risks has less to do with exactly who owns the risk, and more to do with how risks are treated. The more risks are valued, traded and hedged in the same way, in the same markets, the greater are systemic risks. As they turn their attention to insurance markets, the current mantra of regulators that we need high and common standards, could not be more dangerous. We need high and uncommon standards. We need diversity. If the transfer of risks from banks to insurance companies is to make the system safer, we need insurance companies to behave like insurance companies and match their long-term liabilities with long-term assets, with long-run valuation and long-run risk management.
© Avinash Persaud, 3 October 2002