RDP 2011-02: Long-term Interest Rates, Risk Premia and Unconventional Monetary Policy 7. Conclusions

This paper studies two kinds of monetary policies. One takes long-term nominal interest rates as operating instruments of monetary policy. The other considers credible announcements about the future path of short-term nominal interest rates. Within a general equilibrium model in which a component of the risk premium on long-term debt is endogenous, we show that long-term interest rates rules are consistent with unique rational expectations equilibria as much as conventional rules are.

This result is important both in theory and in practice. First, it implies that a unique equilibrium exists if a policy interest rate longer than one period is used in the model. This gives us confidence that results from models which use a policy interest rate that matches the periodicity of the model, say quarterly, are still relevant when central banks in practice use a daily policy interest rate. Second, it implies that long-term interest rates are potential instruments for the conduct of monetary policy. In our framework, long-term interest rate rules give rise to sensible dynamics and depending on the preferences of the monetary authority, they can outperform Taylor rules.

It may seem surprising that a long-term interest rate rule does not necessarily give rise to multiple equilibria. The expectations hypothesis says short rates determine long rates, but it is right to think of long rates as determining short rates too. This is true if the pure expectations hypothesis holds or if more general versions of the expectations hypothesis hold; versions that include a risk premium between interest rates of different maturities.

The idea of monetary policy affecting long-term interest rates is not unprecedented. Friedman's (1968) description of the ‘euthanasia of the rentier’ shows that the central bank has been able to hold long-term interest rates low. Indeed, in many ways, setting a long rate seems less radical than the more conventional policy of setting an exchange rate. In a fixed exchange rate regime, the central bank also sets the price of an asset. But, the central bank's ability to maintain a given exchange rate with market forces that would otherwise depreciate the domestic currency is limited by its stock of foreign reserves. The central bank can buy foreign currency without bounds, but can sell foreign currency within bounds. Not surprisingly, fixed exchange rates often do not last for very long.

To set a long-term interest rate, however, the central bank could use the stock of government debt, of which, in principle, there could always be enough. It could also create its own instruments to set an interest rate of a chosen maturity.

The other policies we study are announcements about the future path of short-term interest rates. Credible announcements can be successful under two conditions: (i) if they entail a return to a monetary policy rule for which the equilibrium is unique, otherwise, every announcement leads to multiple equilibria; and (ii) if the announcement implies a path for the interest rate which is different than what the economy would have produced in any case. The credible promise of lower interest rates, and the actual implementation as the policy is carried out, reduces long-term interest rates through its impact both on expectations and on the risk premium. The first channel is a straightforward consequence of the expectations hypothesis, but the second channel is a consequence of the additional liquidity that is needed to implement a sequence of lower interest rates. This additional liquidity lowers risk premia.

A full explanation of the monetary transmission mechanism, as King (1999) argues, involves understanding the determination of risk premia. We have made progress in this direction. But are the properties of long-term interest rate rules similar in other environments, like those set up by Alvarez, Atkeson and Kehoe (2007)? Are the properties of long-term interest rate rules the same if the zero lower bound on the entire yield curve binds? Do equilibria exist? These are, of course, challenging questions. But the efforts to address these will add to an expanded monetary policy toolkit.