If a retailer like Woolworths goes bust and the company is
liquidated, there has, since the Bankruptcy Act of 1542 been a
hierarchy of claims on the fruits of liquidating that comany's
assets. These days the order is as follows. First the
"secured" creditors (e.g. those that had lent money secured against a
particular sort of machinery or a car) take their cut, then salaries, the
the creditors termed "senior unsecured" come next, then "junior
unsecured" creditors, and whatever is left over (if anything) goes to the
shareholders. At any point in the hiearchy, you have no claim until more
senior claims have been satisfied in full, and if you take a haircut everyone
that ranks equal with you takes the same haircut.

In a number of countries in the past, if a bank went bust,
depositors ranked ahead of (were "senior to") bondholders. So no
depositor would lose anything until bondholders had been completely wiped
out. But in Britain, as of 2008, legally depositors and bondholders ranked
equally. One of the recommendations of the Vickers Commission was that
depositors should have what is called "preference" over bondholders –
i.e. that in future they should be senior to bondholders, as in certain
other countries. This concept has been taken up at EU level, and is
currently under debate.

One sticking point in negotiations over depositor preference
concerns the status of insured depositors. The British, represented
by Chancellor George Osborne in recent negotiations on this
point, wanted insured depositors to rank ahead of uninsured
depositors. A number of other European leaders wanted insured and uninsured
depositors to rank equally.  In my view the British position is wrong, and
that of other EU leaders correct.

To see why, let us consider two cases: the treatment of
depositors in Cyprus in the recent "bailins" of the banks there, and
the level of deposit insurance in the UK in 2007. In theory, in Cyprus the
way EU deposit insurance should have worked was as follows. Insured
depositors would share in the haircuts experienced by other depositors (and, as
it happens, bondholders), but then be compensated by the national deposit insurance
scheme. It doesn't really matter for our purposes whether the insured
depositor would be aware of being compensated – the issue isn't whether anyone
has to "claim back" after initially being out of pocket, or anything
like that. We can assume that the national deposit insurance scheme
seamlessly and instantly transfered the compensation into everyone's
account. Neither does it especially matter for our purposes whether the
national deposit insurance scheme is funded by other banks or by the government.

What does matter is this.  In Cyprus "insured"
deposits weren't insured as such at all – partly because the Cypriot government
didn't have enough money to cover them. Instead, they were made
senior to other deposits, as per the British proposal for what should apply
across the EU. The consequence of that was that other depositors
experienced much larger haircuts than they would have experienced had the
insured depositors also taken a haircut and then been compensated – perhaps
twice as high a haircut.

So far so straightforward – that is obviously a consequence
of making insured deposits senior to uninsured deposits. But now grasp
this: the amount of the extra haircut experienced by the uninsured deposits
depended on the level of deposit insurance. If instead of being €100,000
the deposit insurance level were €50,000, uninsured deposits would have taken a
much smaller haircut, whereas if the insurance level were €200,000 the
uninsured deposits would have taken a much larger haircut.

Again, that's obvious, but it's crucial, because governments
change deposit insurance thresholds quickly and arbitrarily, as we can see by
considering the UK in 2007. Up to September 2007, only £2,000 of deposits
were 100% insured. That suddenly went to all deposits placed in any bank
after an arbitrary date, then settled back to £50,000 in October 2008, then
rose to £85,000 in 2010. If insured deposits were not preferential to
uninsured deposits, rises in the deposit insurance threshold would affect
only the insured depositors – making them better off. But if insured
deposits are preferential to uninsured deposits, then rises in the deposit
insurance threshold make uninsured depositors worse off. So you could
deposit £1 million in a bank believing the interest rate it offered you
provided a reasonable compensation for the risk of any plausible haircut you
might experience if the bank went bust, but then find that, because the
government raised the deposit insurance threshold, what had seemed like a fair
risk suddenly turns into a very bad deal, not because you have mis-assessed the
riskiness of the bank but instead because you were the victim of a regulatory
policy change.

That would be bad for the use of market forces to maintain
banking stability. It would mean that when a bank started to become
riskier, it could not compensate by attracting deposits by raising interest
rates, because large depositors would be concerned that the government would
respond to the bank's distress by raising deposit insurance thresholds after
the large deposit was made, confiscating funds ex post from uninsured
depositors. (And raising interest rates would make no difference to
insured depositors, since they would be insured anyway.)

It would be better not to insure the capital value of deposits
in fractional reserve banks at all. But if they are insured, and if we are
to have depositor preference, all depositors should rank
equally. Otherwise uninsured depositors become subject to confiscation of
funds through arbitrary (and plausible) regulatory decisions, which is not good
for healthy market processes in banking nor financial stability.