Extra, extra: IMF favors raising PDIC insurance limits

Last month, the IMF released a PIN, which included a comment on PDIC legislation.

The reason IMF Directors gave for favoring the proposal to increase bank deposit insurance is a contingent one. It is a means of addressing “financial sector vulnerability.”

Here’s what the IMF actually said:

Directors supported the proposed increase in the Philippines Deposit Insurance Corporation (PDIC) deposit coverage, and noted the importance of allowing for flexibility to temporarily raise it further if the financial sector experiences high stress levels. Recapitalization of the PDIC will be key to support the enhanced deposit protection.”

In the environment where pre-need and rural banks are candidates for Ponzi-type schemes, I considered that raising the PDIC deposit insurance limit was unwise.

ATM to machine the polls – the economics of automation

There is much ado about how to computerize elections in the Philippines.  Here is my attempt at setting out the economics of the matter.

There are two goals: ensure integrity and security of the poll process, and minimize the cost of automation.  In economics, these are the “goods” we wish to buy or produce from our limited budget or resources.  The cost of automation includes not only the cost of training voters but also the foregone opportunity to use what people, such as poll watchers, already know about the election system in the Philippines.

A hybrid approach is likely the only route in the present environment, where perhaps only about half the voters are sufficiently tech savvy or tech “willing,” and it would be too expensive to implement full automation.  Thus, some remnant of the “paper” system needs to remain.  For the automated system, the least-cost route is an existing secure network that a voter already knows how to use.  I suggest that this is simply the ATM network of the local banks.  I estimate that half the voters know how to use ATMs.

How do we ensure integrity?

For the remnant paper system, voting is the same as before, and so is the “watching” of the vote count.  Candidates and other interested parties will still have a right and duty to “watch” each other in the counting.  What would be automated is the precinct-level canvass, since the authentication and input of the data here can be “watched.”  If done well, this eliminates the infamous “dagdag-bawas” in the post-precinct canvassing.  In recent poll contests, it seems that serious candidates have learned to protect themselves at the precinct watch stage.  Training will have to be conducted for the poll inspectors and watchers, but there will be no need to train the voters. 

The ATM is a ready-made vote-counting machine; all it needs is a common interface, set up at election time, which shows the ATM user his candidate choices.  The nice thing about the ATM is that it already has built-in security features: only the voter knows his ATM number, and he can vote in the secure environment of the ATM.  In order not to clog up the ATM system, voting can take place over, say, a two-week period, where the votes are not “final” until the date of the election (the voter can change his mind – which helps prevent vote-buying).  For after-election auditing, a voter can have his vote result emailed to him by the banks, and he will have a chance to compare that with the COMELEC database which will be open to him. An auditing firm can set up a random process to verify integrity, in the same way that such firms track bank transactions to uncover any bank frauds. 

A process will have to be devised to ensure that voters vote only once.  This can be done by assigning each voter a unique voter identification number (VIN).  At the precinct-level count, each ballot’s VIN must be entered into a precinct machine, connected on the web to the central COMELEC server.  The ballot will be rejected (or a blank one not even given to a voter) if the VIN has already been used at an ATM, or if the VIN is invalid.  All poll watchers will be trained to ensure this.  All the VINs actually allowed will have to be entered in the precinct-level canvass, so that an audit process can also randomly verify compliance with the one-vote-only rule. 

As to the cost, the usual answer is “It depends.”  But I would bet that the cost of this approach is smaller than the amounts currently budgeted for automation.  The ATM networks will be compensated for use of the ATMs, and for the special programming needed to convert (temporarily) an ATM machine into a poll machine.  The consortium of banks, in an act of civic-mindedness, can perhaps find a way to allow the security of their network to safeguard the security of the vote.

We often find informal polls offered in many websites, including personal blogs.  We don’t worry about security and integrity because the polls are mostly for entertainment.  Election of public officials is another matter, but it seems possible to have security, integrity, and low cost, with a little help from the banks!  A hybrid approach will also allow those who don’t trust machines to use the old paper ballot.

Nursery rhymery on good banks bad banks, by Joshua M Brown

For a tongue in cheek view on the good bank bad bank scene..   Dr. Seuss lives!

Good bank, bad bank, Woodward and Hall reconsidered

Further to my recent post on the Woodward and Hall proposal, some reconsideration seems in order.

The following further facts and circumstances would seem worthwhile to take into account:

1. While it is desirable to protect depositors in order to prevent bank runs, bond holders have to be induced into the bad bank scheme, since otherwise, they have a proper argument, on equity grounds, to demand equal treatment.

2. The inducement for bond holders can take the form of retaining the current uncertainty on their bond claims on “toxic” assets, while giving them some measure of control over the future. Thus, the bad bank’s equity should be given to bond holders, and not to the current shareholders of the integrated bank.

3. A further important inducement for bond holders is effectively an option on the future profits of the good bank, since the purpose of the bad bank is simply either to avoid insolvency or to limit the “haircut” on them in the event of insolvency. Thus, bondholders can also be given some direct equity in the good bank.

4. There being no free lunch, the whole scheme will likely still require public funds. This can take the form of a take-up of equity by the government in the bad bank. In effect, the government and the bondholders would become partner shareholders in the bad bank.

Taking all these into account, the new balance sheets of the good and bad banks, using the Citi data as an example, would reflect something along the following lines:

(a) The bondholders appear to have a more or less equivalent claim on the long-term assets of Citi, in which case, they could retain half their bondholdings against the good bank. The rest of the bondholdings would be partly converted into equity against the good bank, for example, $250 billion.

(b) The government could put up, say, 5% of the gross capital of the bad bank, in the form of some form of common or preferred stock. Thus, the bad bank would have shareholder equity of that amount, i.e., $30 billion.

(c) The present marked-to-market equity of Citi as a whole ($11 billion) would be assigned to the good bank. This ensures that the current managers of Citi would work to promote normal and expanded lending by the good bank.

(d) Of course, the deposits of customers would remain deposits of the good bank, which would now have an equity base held overwhelmingly by current bond holders of Citi. The bond holders would also have an incentive to see that the good bank is well managed since this represents a bet that they can recover some of any losses they might face if the bad bank were to go into bankruptcy.

The following table illustrates such a revised good-bank, bad-bank proposal. The data are the same as in the Woodward and Hall proposal, except that the $1,144 billion of bond holdings is now split three ways, $610 billion of equity in the bad bank, $250 billion of equity in the good bank, and $284 billion of remaining bond holdings. This could be accomplished by a debt-equity swap in the proper proportions. The other difference is that the government puts in $30 billion of equity into the bad bank, which shows up on both sides of the balance sheet, and the $11 billion of equity of current shareholders of Citi is retained in the equity of the good bank. The capital ratio of the good bank is 20%, instead of 32% in the original Woodward and Hall proposal.

	Citicorp	Good bank	Bad bank
Assets
Short-term	448	448	30
Long-term	1,295	686	610
Other	192	192	–
Equity in other bank	–	–	–
Total assets	1,935	1,325	640
Liabilities
Deposits	780	780	–
Bonds and other non-deposit debt	1,144	284	–
Equity	11	261	640
Capital ratio	1%	20%	100%

Romer on good-bad banks may simply trigger bank runs

Paul Romer suggests that public funds should be used as equity of new “good” banks.  With standard leverage ratios, such banks could go a long way to solving the “credit crunch.” And it looks like a good proposal because it appears to be a good use of public funds.  But are there possible pitfalls?

A possible problem with the Romer proposal is one relating to bank runs.  The new good banks would be well known.  What if depositors “run” from the big existing banks to the new banks? The bank regulators will then have an even bigger problem in their hands: How to contains such bank runs.

In a strange sense, if all big banks were equally at risk, this limits the incentive for bank runs, and yet such a “risky” scenario seems the best.  The old Bagehot rule has always been that there should be a bank of last resort, but there should be uncertainty as to how and when it will act to save a “bad” bank.  This seems to be the practical rule to limit moral hazard in the modern banking system.

Bernanke said it well – Save the economy, and in large part the “banks” problem gets fixed

Over at Arnold Kling’s blog, there is this question of whether to save the economy or the banks.

Bankruptcy and good-bad bank proposals

There seems to be a good proposal on how to split a potentially insolvent bank into a good bank/bad bank.  

It looks good, but there seems to be a real flaw in the proposal.  What Woodward and Hall propose is a de facto way of creating deposit insurance out of thin air.  Why?  This is because the splitting up of the bank (Citi, in their example) is such as to fully protect the depositors, which is not the case in the usual insolvency proceeding. Depositors are also creditors of Citi, and if uninsured, they stand in line, along with other creditors, although of course, they are ahead of shareholders.  With the good bank, the depositors are protected, benefitting de facto from a quasi deposit insurance.

The key question is this: why would the bondholders voluntarily agree to “take the hit” disproportionately with Citi’s depositors?  It is easy to see why we would want to protect the depositors, but there is a zero-sum game here, where protection for depositors has to come from “unprotecting” another group, i.e. the bondholders.  If the hit is involuntary,  the lawyers would have a field day invoking the due process and equal protection clauses on behalf of bondholders.

A similar situation applies to the question of how to separate out the bondholders as between those holding claims against the bad bank (the bulk persumably) and those against the good bank, in which case, the latter would be on the same footing as “fully-protected” depositors, given that the good bank is set up to limit the risk of insolvency to zero.  A perusal of the tabular data in the example given by Woodhard and Hall would easily show this.  They have a small amount of $118 billion of bonds held as liabilities of the good bank. How did they arrive at this figure? They could have set a figure of zero, in which case the equity of the good bank would have been $545 billion, a figure that seems unrealistically high.