Should we apply a transactions tax to REPO?

A number of readers wrote me this morning and asked what I thought of Paul Krugman's column today.  Krugman writes:

As Gary Gorton and Andrew Metrick of Yale have shown, by 2007 the United States banking system had become crucially dependent on “repo” transactions, in which financial institutions sell assets to investors while promising to buy them back after a short period – often a single day. Losses in subprime and other assets triggered a banking crisis because they undermined this system – there was a “run on repo.”

And a financial transactions tax, by discouraging reliance on ultra-short-run financing, would have made such a run much less likely. So contrary to what the skeptics say, such a tax would have helped prevent the current crisis – and could help us avoid a future replay.

My view is different.  I do favor such a tax in the following sense: capital requirements on financial institutions should rise with our best estimate of risky behavior and of course these capital requirements serve as a tax.  From this follows a few points:

1. The net tax applied, whether through capital requirements or a demand for a revenue payment, can only be so high.  I would prefer to "convert" all of the applied tax into the form of capital requirements.  Krugman doesn't call his idea "a proposal for lower capital requirements," but relative to the better and more politically feasible reform of higher capital requirements, that's what it is.  At the margin we have to choose between various forms of tax.

2. Is the quantity of short-term REPO, or other financial transactions, really the best measure of risky behavior?  I doubt it.  So we're taxing the wrong thing, relative to what we might do.

3. Taxing short-term credit means that firms will resort to long-term credit.  In most models longer-term credit increases moral hazard and excess risk, even if it does limit the short-term "runs" effect.  

A few broader points are relevant:

4. In many jurisdictions, REPOs are already taxed, as the "borrowed" security is counted for tax purposes as "owned" and the income from it is subject to tax.  Of course the tax could be higher, but is there evidence that the taxes to date have had beneficial effects on financial stability?

5. There are many substitutes for REPO — virtually any transaction can include an implicit short-term credit component — and I am skeptical of the ability of the regulators to catch them all and prevent destabilizing regulatory arbitrage.

6. There was a transactions tax on sale and transfers of stock before and during the Great Depression; it did not obviously help matters.

7. There will always be some untaxed network for speculators to trade in and a transactions tax would push them into such a network, probably in destabilizing fashion.  The fact that trading is relatively centralized now does not refute this effect.

Of these points, only #6 and #7 apply to the standard Tobin "currency tax."  No matter what you think of the Tobin tax, overall there is not sufficient logic or empirics to justify its extension to REPO and related short-term transactions.

Addendum: I am more in agreement with Krugman's post on Dubai.


Comments for this post are closed