Macroprudential policy uncertainty and implications for leaning against the wind

Abstract

I consider whether imperfect macroprudential policy can be ‘improved’ upon by a monetary policy of leaning against the wind. Imperfect macroprudential policy is captured by instrument uncertainty, which leads to tentative and under-responsive macroprudential policy, and model uncertainty, which leads to excessive and over-reactive macroprudential policy. Leaning against the wind by the central bank improves the macroprudential regulator’s policy rule if the impact of model uncertainty is stronger than the impact of instrument uncertainty. Such a policy may therefore be pursued in jurisdictions where the macroprudential regulator has low risk-sensitivity and where the efficacy of macroprudential instruments is more certain.

KEYWORDS:

• Macroprudential policy
• monetary policy
• policy uncertainty
• leaning against the wind

JEL Classifications:

• D80
• E58
• G28

Acknowledgements

I am grateful to Prasanna Gai for helpful comments on previous drafts of this paper. I also thank two anonymous referees and seminar participants at the University of Auckland for their advice and suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The two most recent Remits for the Monetary Policy Committee explicitly allow inflation to deviate from its target level in order to pursue financial stability (Carney, 2014, 2020).

2 Other notable papers in this area include Agur and Demertzis (2013), Gambacorta and Signoretti (2013), Ajello, Laubach, Lopez-Salido, and Nakata (2016), and Gourio et al. (2017), who consider financial stability objectives such as credit, asset prices, and crisis probabilities.

3 This model does not require a strict separation of the central banker and the macroprudential regulator, but rather focuses on the design of policy targets within or between the institution(s).

4 Ueda and Valencia (2014) show the ‘debt-deflation’ of nominal leverage as normalised by real output, demonstrating how surprise inflation erodes the real value of leverage to an economy.

5 This link between macroprudential policy and output is intuitive; an expansionary macroprudential position (for example, as with reduced capital requirements) can stimulate business activity through increased lending to the real economy (Behn, Gross, & Peltonen, 2016).

6 In Roisland (2017), the monetary authority may step in at this point and use monetary policy to ‘lean against’ the residual gap in the financial objective. However, this does not improve the ex-ante or ‘pre-central bank’ policy rule for macroprudential policy, which is the key consideration of my paper.

7 An example of model uncertainty can be seen in the DSGE banking stress-test models that inform regulators on the resilience of financial institutions, and therefore inform the need for macroprudential regulation. Many prominent authors, such as Drehmann, Borio, and Tsatsaronis (2012) and Stiglitz (2018), have heavily critiqued these models for their inability to identify ex-ante imbalances, and for the sensitivity of these models to parameter calibrations.

8 Although $\mathrm{\Omega }$ is treated as exogenous in this model, others (such as Tallarini, 2000) have explored the possibility of an endogenous risk-sensitivity term. In these cases, an endogenous increase in risk-sensitivity increases the accuracy of the regulator’s models, and decreases the extent of model misspecifications.

9 This min-max approach is based on the robust control literature, where the misspecification term, $v$, is more commonly known as ‘entropy’. The maximisation of loss by a malevolent player is common practise in this literature, as it ensures that the interaction between the policymaker and nature reflects the most averse conditions.

10 There is a second case where the regulator’s policy rule can be ‘undistorted’ by model uncertainty: when the regulator has no secondary objective, i.e. the weighting on the output gap is $a=0$. This regulator’s policy rule is equivalent to the single-objective case for instrument uncertainty, ${\delta }^m=(({\theta }^{\ast }-\overline{\theta })/(\overline{d}+{\sigma }_d^2))$. However, this policy rule can still be distorted by instrument uncertainty, ${\sigma }_d^2$. This provides some justification for allowing a secondary macroprudential objective $(a>0)$, as this can counteract the distortion caused by instrument uncertainty.

11 More explicitly, leaning against the wind is not preferred if the regulator’s risk-sensitivity is high, or above the ‘threshold’ value (i.e. $\mathrm{\Omega }>{\mathrm{\Omega }}_{LAW}^{\ast }$). This will be the case if a high level of instrument uncertainty pushes the threshold value of ${\mathrm{\Omega }}_{LAW}^{\ast }$ below 1, as this level of risk-sensitivity is strictly unattainable for the regulator. The level of instrument uncertainty that corresponds to ${\mathrm{\Omega }}_{LAW}^{\ast }<1$ is ${\sigma }_d^2>a{\beta }^2+1-\overline{d}.$