1 of 7

Optimal quantitative easing and tightening�by Richard Harrison

Discussion by Dominik Thaler, ECB

​

Views are my own and do not represent the ECB or Eurosystem.

2 of 7

This paper: The model

    • Introduces a portfolio rebalancing channel into the 3 equation NKM
    • Assumption: private sector has preferences for maturity (composition + stability)
    • Linearizes to

​

​

​

​

​

​

    • 2nd order approximation of welfare

​

Phillipscurve

Policy stance

IS curve

Nominal rate

QE

QE stock QE flow

3 of 7

This paper: Optimal Policy

    • Away from the ZLB:
      • Only stock effects: no QE
      • Stock and flow effects: no QE in the long run� if initial BS>0: reduce gradually (trade off negative flow and stock effects)

​

    • Optimal response to ZLB shock:
      • Rapid increase in BS.
        • Don’t increase all at once to reduce flow costs.
      • Start QT while still at the ZLB to reduce stock costs.
      • Liftoff before completing QT to minimize flow costs.
    • When starting from maximal balance sheet:
      • Similar (larger stock effects substitute for lower flow effects)

4 of 7

This paper

    • Great paper
      • Elegant, Simple, Intuitive

​

    • Speaks directly to many extremely policy relevant questions
      • Exit strategy ✓
      • QT response to cost push shocks !

5 of 7

Comment 1: Stocks and flows

    • Flow effects are larger then stock effects in baseline calibration
      • Theory: Contradicts macroeconomist’s intuition (Gertler and Karadi 2011, Vayanos and Vila 2021)
      • Empirics: Flow effects order of magnitude smaller (D’Amico and King 2013)

​

​

​

6 of 7

Comment 2: The cards are stacked against a large balance sheet

    • The private sector has preferences over the maturity structure of debt
    • The treasury provides the optimal maturity structure
      • No term premium, flat yield curve in SS
      • SS is first best
      • QE can only make things worse
    • Does empirical YC point to inefficiencies?
      • Greenwood, Hansen, Stein (2016)
    • The treasury does not respond to QE

(neither quantity nor maturity of debt)

      • Bigio, Nuno, Passadore (2023)
    • Large balance sheet may have other advantages:
      • generates fiscal revenues
      • increases liquidity
      • better operational control of short term rate
      • stabilizes risky sovereign debt

7 of 7

Minor comments

    • Unlike nominal interest rates, QE has real effects even with flex prices. How does this shape optimal policy in the context of a flat Phillipscurve?
    • ZLB constrains short rate, but no ZLB LT rate or on period return of LT bond. With QE, HH holds LT debt even though period return is negative.
    • Is QE like a tax on savings? Both make savings less attractive, but tax would have no welfare effects.
    • If solved globally, why solve the linearized model in the numerical part?
    • QE introduces a state variable because of flow effects
      • Section 3.2.3 claims that unbounded q makes q stop being a state, but that’s not formally true, because it appears also in the objective
    • Britain is a SOE. MP has FX effects. Model is closed economy, but calibrated to match British QE effects.
    • What explains the 2nd kink in the black policy rate in figure 5 (maybe colors are switched?)
    • Explanation why the two FIT policies are not equivalent could be clearer.
    • Editing:
      • Specify what you mean by welfare losses (relative to what) in intro.
      • Specify what z is on p. 9
      • Define Phi on p. 14
      • Explain D-notation on p. 17