In a few weeks we will know the results of the stress tests. Rightly, the US Treasury and the Fed have pledged that these tests will not be used to liquidate or nationalise banks, but to determine the degree of support that each institution may need. Despite this pledge, the plan has not been favourably received by financial markets for essentially three related reasons:
- First, dilution of existing shareholders could be enormous if all the precautionary capital is to be raised at the early February 2009 prices announced in the plan.
- Second, they fear that as a result of these injections, the share of government ownership of some of these institutions may rise to troublesome levels.
- Third, there is a significant political resistance to using more public funds to support the banks, which may limit the resources and hence the effectiveness of the intervention.
The essence of the Treasury and the Fed’s strategy is right, in that it has the systemic problem as the core priority and relegates other standard incentive and political concerns to a secondary role. This is not the time to get distracted – financial stability should be the only short-run goal. Thus, it is time for us to focus on refining their plan rather than on trying to come up with entirely new policy paradigms. In this light, the goal of the proposal I outline below is to address the specific problems mentioned above. That is, to raise the capital needed to fulfil the requirement of the stress tests without an extreme dilution of existing shareholders and with reduced government participation.
A proposal: Raising capital without diluting shareholders
The proposal is simply to support with an equity guarantee the Capital Assistance Programme’s option of raising private capital for six months after the results of the stress tests are available. More specifically, all new private equity injections would receive a public guarantee for a minimum future price – say, five years hence.1
The reason to set the price-guarantee for several years into the future rather than for the near future is that the crisis is more likely to be resolved by then, and hence there is only a very small probability that the guarantee is exercised. That is, in practice, the government may never have to inject significant funds into the banks, and can leave the recapitalisation process almost entirely in the hands of the private sector.
The guarantee-policy has maximum power when a significant element depressing current equity prices is uncertainty, a condition which almost surely holds at this time. In such a case, the effect of the policy on share prices is many times larger than the direct effect of the guarantee on these prices. The reason is that uncertainty reduction from new capital obtained at non-dilutive prices feeds back into the price of non-insured shares, which in turn increases the price of guaranteed shares (which is equal to the sum of the value of the guarantee and that of the uninsured shares), and so on.
This multiplier effect is also the main reason the policy is ultimately inexpensive for the government, since it creates a large gap between the new equilibrium share price and the guaranteed floor. The low cost of intervention raises the potential to increase the scale of the programme. The guarantees also send a much clearer message that private capital will be protected rather than diluted by policy actions, which is a must for any sustainable recovery based on a return of private capital.
An equity guarantee and its multiplier effect
The policy: Banks that are deemed to need more capital if aggregate conditions worsen (the stress test), can choose to raise private capital supported by a government guarantee. Capital raised within a pre-specified period of time will be guaranteed a minimum equity price five years hence.
The direct effect of this policy works by what economists call backward-induction. If we know that the price will be at least something in the future, and there is an upside potential as well, then the price must be even higher today. Essentially, through the insurance mechanism the government transforms heavily discounted (by uncertainty) equity into a treasury bond plus a call option on the upside of the bank. Alternatively, it protects risky equity with a put option on the downside of the bank.
However, the mechanism is more powerful than just the direct effect of the guarantee on the value of new equity. This is the subtle part of the proposal. It turns out that the price of (non-insured) existing shares rise as well, which means that the price paid for each insured share rises both because of the insurance and because of the rise in the perceived value of the underlying shares. But once those infusing new equity are willing to pay a high price per-share, the private capital injection dissipates the fear of an imminent crisis with much less dilution than would have taken place at the original equilibrium without the guarantee-policy.
Moreover, it can be shown that if the effect of uncertainty in depressing equity prices is significant, a condition that clearly applies to the current juncture, the indirect effect can be much more powerful than the direct effect of the guarantee.
Let me illustrate this situation with a simple example from Caballero and Kurlat (2009).2 Suppose there is a bank that needs a $50 billion capital injection were an extreme scenario to occur in the near future. The share price is depressed by uncertainty and the concern that if the bank does not raise the funds it will be nationalised (or go bankrupt) if the extreme scenario is realised. The current panic-driven share price is $2 (panic driven because essentially, potential investors think that the likelihood of a crisis is twice the real chance). In this context, and under more or less realistic assumptions, if the government offers a minimum future share price guarantee to new equity holders of $2.7, the immediate result of this policy would be to trigger a boom in the existing (non-insured) bank’s share prices to $6.8, while the insured shares price is $7, which is only $0.2 more than the regular shares.3
But how can this guarantee (that is valued so little in equilibrium) generate such a large impact? The reason is that its presence entirely changes the equilibrium. The guarantee is worth little because $2.7 is not much of a guarantee at the new equilibrium price of $6.8, but it is a very large one at the old price of $2.0! Thus, this mechanism is a powerful tool around the depressed price equilibrium, but it is worth very little (and hence it has a small expected cost for the government) in the final high price equilibrium.
Robustness and Incentives
In practice, it is important that the policy prescription be robust, or at least not too harmful, if the underlying basic premise justifying it in a particular model is not the main factor in reality. I based the above conclusion on the notion that panic is a central aspect magnifying an already severe financial problem. But what if the panic factor is small relative to the true structural problem? How harmful would an equity-guarantee policy be in such a case? The answer is, not much. The main impact of overestimating the role of panic in asset and equity prices is that the multiplier will be smaller than anticipated. But, and this hints at an implementation issue which I will return to later on, if the guaranteed price is set at the February 2009 price (plus an interest rate adjustment) used for the preferred shares already announced, then the impact will be similar to the default plan, but with much less government participation in the short run.
Another important concern is whether a “gamble for resurrection” (no skin in the game) type effect may arise in the (unlikely) event that the price of the shares falls below the guaranteed price before year five. However, one should note that a significant fraction of the shares will not be protected by the guarantee, which serves to reduce the severity of this problem should it arise. In any case, it is clearly important that management be excluded from acquiring the guarantees (in fact, they should probably write some of these guarantees themselves!).
There are many different ways to generate a similar structure of payoffs for new capital, some of which, in some contexts, may be better than others. For example, if implementation frictions point toward direct public equity injections rather than pure guarantees, it is still possible to improve on the pure public equity-injection scheme. The government could inject some (limited) amount of common equity in the short run, and give matching new private equity the option to buy it back five years from now at the February 2009 price plus some interest rate. This arrangement still has the effect of partially insuring new equity from the possibility of a bad realisation in the future, and thus serves to encourage private capital to return to the financial sector.
IV. A Practical Implementation Path
After the stress tests have been conducted, it is important for the sake of uncertainty-reduction that the banks (whose results indicate they require more capital) are given the recapitalisation options announced by Treasury Secretary Geithner on 10 February 2009.
However, at this stage there could be an announcement made that new options will be added (but never removed) to the banks in need of recapitalisation. Moreover, the first option could be given to the bank in most need; let’s call this bank C.
Suppose that bank C needs to raise $50 billion to be completely sound in an extreme event, and that its shares are being traded at $2. The Treasury could announce that bank C has $50 billion in government preferred shares at its disposition, but that it will be given the option to raise private funds during the next few months (as the current CAP does). In order to facilitate this fundraising, the Treasury will support new shares with a guarantee for a price equal to that used in the preferred shares, plus some market interest rate, five years from now.
Initially, the Treasury could support up to 20% of the target number of shares. The early bidders determine the price they pay for these shares, and are given the guarantee plus the right to any improvement in guarantees that may take place in the future.
If the auction is a success, the remaining shares are offered with the same terms. If not, the guarantee is sweetened and another 20% of the shares are offered the new terms, and so on.
For example, looking again at bank C, the first guarantee could be offered at $3, but suppose that the price of the shares only rises to $5 rather than to the $7 needed by the bank. Then the next 20 percent of new shares is offered a guarantee of $4, which is automatically granted to the first 20 percent of the shares as well. If this step achieves the desired price of $7, the remaining shares are sold with the warranty of $4. Note that the first entrants always obtain at least as high a return as the late entrants (because they buy the shares at a lower price). This is precisely what makes the mechanism work since there is a strong incentive to jump in as soon as the plan is announced.
Along the way, as the experiment is calibrated, other banks can be brought into this provision, and, in fact, much of the systemic benefit may come from the expectation that all banks in need will eventually be subjected to the guarantees model.
1 See ”Dow Boost and a (Nearly) Private Sector Solution to the Crisis”, Ricardo J. Caballero, VoxEU.org February 2009
2 See ”Public-Private Partnerships for Liquidity Provision” by Ricardo J. Caballero and Pablo Kurlat, March 2009.