Impairment of monetary policy independence by global financial cycles and the mitigating role of macroprudential policies

ABSTRACT

In this paper, we study the impairment in monetary policy caused by different forms of global financial cycles. We find that while both equity inflows and outflows cycles do exert influence over monetary policy, the bond inflows cycle does not have a significant impact. Further, we discuss the role of macroprudential policies in mitigating this impairment by using Difference-in-Difference with heterogeneous treatment effects which is robust to presence of heterogeneity across both time periods and groups. We find that FX-based policies such as Capital Restrictions on Foreign Exchange positions and Limits on Foreign Exposure alongside SIFIs and Loan Loss Provisioning are effective in reducing the impairment.

KEYWORDS:

• Global financial cycles
• monetary policy impairment
• macroprudential policies
• Difference-in-difference with heterogeneous treatment effects

JEL CLASSIFICATION:

• E52
• F38
• F41
• F42

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00036846.2023.2244250

Notes

¹ Han and Wei (2018) study the impact of VIX index on interest rates. Ouyang and Guo (2019) study the impact of GFCy on exchange rate while Kalemli-Ozcan (2019) study the impact of international risk spill-overs on domestic monetary policy transmission.

² 96 is the exhaustive set of countries and the estimation of GFCy for each set is based on data availability. Similarly, for impairment, we consider exhaustive set of 58 countries and estimation is dependent on data availability.

³ We utilize money market rates for Austria, Belgium, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Portugal, Slovak, Slovenia, Spain, Japan, India, Bulgaria, Sweden, Denmark, Uruguay, Ukraine, Croatia. For China, due to non-availability of data on money market rate, we use Deposits rates from IFS, IMF.

⁴ We do not use seasonal adjustment for other variables because the actual series are almost identical to adjusted series. It is only for GDP data, that we encounter some differences.

⁵ Refer to Alam et al. (2019) for more details.

⁶ For estimating output gap, we apply Hodrick Prescott filter (1997) to decompose trend and cyclical components.

⁷ We run two different models for each type of GFCy – one with inflation and another with inflation gap. The coefficients on both are positive and significant and only differ in magnitude. We report results on inflation and not inflation gap as the magnitude on inflation is higher.

⁸ Earlier study considers centre country variables.

⁹ Fuzzy Difference-in-Difference estimator (De Chaisemartin and d’Haultfoeuille 2018) did address these concerns, however, the $\mathrm{DI}{\mathrm{D}}_l$ estimator used in this paper unlike the previous one, gives us dynamic, instantaneous and ATE effects for all time periods at once.

¹⁰ Some observations may have already been treated prior to the first period in the data, and those treatments might still affect some of their period-$1$-to-$\mathit{T}$ outcomes, known as the initial conditions problem. We deal with it by applying restrictions following De Chaisemartin and D’Haultfoeuille (2022b).

¹¹ Following Adarov (2022) and Miranda-Agrippino and Rey (2020), we use reference series to correctly rotate the global factors, as the factors are consistently estimated up to a scale and initial value. While Adarov (2022) utilize US financial cycle, Miranda-Agrippino and Rey (2020) utilize stock market indices to interpret global common factors. We utilize global common factor estimated by Miranda-Agrippino and Rey (2020) as the reference series.

¹² For Capital- FX, we measure 7 effects, for MPI Index, we measure 6 effects, for LLP 5 effects, for LFX 3 effects, For Liquidity, All capital and Other bank restrictions, we measure 2 effects each and for SIFI, we estimate 1 dynamic effect.