The Event Horizon

When is an event ?

Yesterday Janet Yellen said

Event studies can therefore be helpful in gauging the financial market effects of such communications. [skip]

Last August, the FOMC announced that it would begin reinvesting principal payments on agency MBS and agency debt into longer-term Treasury securities, and over the subsequent couple of months or so, the public remarks of Federal Reserve officials made note of the possibility of a further expansion of the portfolio. Consequently, when the Committee announced in early November that it intended to purchase an additional $600 billion in longer-term Treasury securities, that decision was largely anticipated by financial market participants, and it occasioned only minimal market response.

So it seems that the event took place over months from August through November. If Yellen is using the words “Event” and “communications” consistently, she must be asserting that the FOMC just kept talking about it for four months and the message gradually sunk in.

More generally, if one is willing to argue that announcements are anticipated and, so, there is no event even though there is a communication, then one must conclude that event studies are not useful. A methodological approach in which one can decide ex poste which communications are events is one in which one can reach any conclusion one wishes based on any data.

Yellen notes two communications related to quantitative Easing II — an announcement in August and an announcement in November. The responses of asset prices to both announcements provide minimal support for her conclusion. So she decides that she can reinterpret the timing of the event as she pleases.

If this sort of argument is treated with respect, then there is no point in looking at data at all.

Yellen’s argument is reduced to the assumption that if the first $ Trillion and change had an effect, then the next $ 500 billion must have an effect on the same order — that something is roughly linear over a genuinely huge range. She must also assume that market conditions are similar in November 2010 and in 2008 and 2009 — that willingness to bear risk and demand for liquid assets was fairly normal at the height of the crisis. Finally, she must assume that RMBS are about like Treasury notes.

I’d say that the facts are that there was a small then a tiny response to the announcement of gigantic purchases.

If Yellen remains convinced that non standard monetary policy has an important role to play, what evidence could possible confince her that it doesn’t ?