9
First, the dynamics of financial markets and risk premia are likely to be richer than captured
in a frictionless complete-markets model. To be sure, the standard expectations hypothesis
remains a very relevant benchmark, especially since alternatives are rather difficult to
model.
5
However, it certainly appears conceivable that actual demand curves for assets are
downward-sloping, so that changes in asset supply can affect risk premia or, in the case of
“supply overhang”, even lead to a complete shutdown of market activity. The underlying
frictions may be negligible in most financial markets in normal times, but they arguably
become more relevant during times of crisis, precisely when unconventional monetary policy
enters the scene. Indeed, the mere fact that there is a very large number of assets with
nontrivial risk premia indicates the limitations of the most basic liquidity trap scenario: a zero
interest rate on short-term government bonds may constrain the effects of monetary
expansion, but it is very unlikely to be the only market price that is relevant for money
demand. As long as other market rates remain above the lower bound, there may thus be
some room for affecting aggregate demand through changes in these other rates.
Second, the above-mentioned assumption about the central bank’s optimization problem
probably overstates the robustness of actual policy frameworks, or at least agents’ perception
thereof. In particular, it seems hard to believe that very aggressive monetary easing today
would do nothing at all to affect agents’ anticipation of future policies.
6
Even where
monetary policy is delegated to an independent central bank, it ultimately hinges on the
broader political support for low inflation. This support, in turn, may be weaker than the
standard model posits, for example, because of existing large debt stocks. Yet, this argument
cuts both ways. If indeed agents’ expectations about future policy are less firmly anchored
than the standard model assumes, conservative central banks may be wary of very aggressive
easing to begin with, lest inflation expectations could “overreact” and become unhinged.
Hence, unconventional monetary policy can be effective to the extent that (i) it convinces the
public of a looser-than-expected future policy stance (relative to inflation developments) or
(ii) it directly reduces risk premia or outright quantitative restrictions in financial markets.
With these general considerations in mind, we now turn to specific policy options.
5
One notable attempt is the paper by Andres, Lopez-Salido, and Nelson (2004), which introduces time-varying,
stochastic transaction costs in the bond market as a way to model the illiquidity of long-term bonds. It also
assumes that households holding long-term bonds compensate for their illiquidity by holding more currency, as
a self-imposed “reserve requirement” of sorts. Lastly, the model features heterogeneous agents, some of which
only trade long-term bonds. Together, these assumptions give rise to deviations from the expectations theory of
the term structure, generating asset supply effects on bond spreads and aggregate demand.
6
In fact, Eggertsson and Woodford (2004) allow for this possibility to the extent that a sharp monetary easing
today contains a credible signal about future policy intentions. The signal could be made credible, in particular,
by unconventional operations that affect the policymaker’s future optimization problem. In this vein,
Eggertsson (2006) and Jeanne and Svensson (2007) argue that central bank purchases of real or foreign-
currency assets, along with higher domestic public debt, can incentivize policymakers to accept higher future
inflation, as this would imply capital gains and/or a reduced need for distortionary taxation to pay down debt.