Beyond the interest rate pass-through: monetary policy and banks interest rates since the effective lower bound

ABSTRACT

We investigate whether ECB monetary policy influences the retail interest rates in the Euro Area when the policy rate reaches the effective lower bound. We estimate a panel Error Correction Model that accounts for potential heterogeneities in the transmission of monetary policy. The analysis disentangles alternative balance-sheet policies implemented by the ECB. We find that unconventional measures have influenced banking interest rates beyond the pass-through of the current and expected policy rate. These effects are driven by liquidity provisions and by the covered bond purchase programmes. Those policies have been effective in the core and in the peripheral countries; however, the effect of purchases of covered-bond has been stronger in the periphery.

KEYWORDS:

• Unconventional measures
• retail interest rate
• heterogeneous panel

JEL CLASSIFICATION:

• E43
• E52
• E58
• G21

Acknowledgment

We thank Jean-Baptiste Bonnier, Raphaël Cardot-Martin, Paul Hubert, François Maréchal and seminar participants at the OFCE, GDRE, LAREFI and ICMAIF conferences for helpful discussion and comments. Any remaining errors are ours.

Disclosure statement

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

Notes

1 See Altavilla et al. (2019a) for a comprehensive analysis of funding costs of banks in the euro area and Darracq Paries et al. (2014).

2 The concept of effective lower bound suggests that the policy rate has reached a lower bound which is close but different from zero and is expected not to be decreased further. It is therefore more appropriate than the zero lower bound even if in practice, the policy rate was reduced later and reached the zero lower bound in March 2016.

3 Liquidity was provided through an extension of the maturity of the LTRO (long-term refinancing operations). In December 2011 and February 2012, the ECB granted liquidity up to a 3-year maturity (VLTRO for very-long term refinancing operations). Finally, in 2014, the ECB conditioned liquidity provision to the credit growth of banks (TLTRO: targeted long term refinancing operations).

4 The expression was first pronounced by Jean-Claude Trichet on 3 February 2011. It was then often used both by Jean-Claude Trichet and Mario Draghi.

5 Hereafter, even if CBPP3 is included in the APP, we have considered that it was more coherent to disentangle the three covered-bond programmes from other asset purchases. The measure of APP will therefore be net of CBPP3.

6 Several papers – focusing on the euro area – assess the effects of unconventional measures in mitigating tensions in the interbank market (Abenassi and Linzert 2012), in reducing sovereign yields and sovereign debt spread (Szczerbowicz 2015; Altavilla, Giannone, and Lenza 2016; Gibson, Hall, and Tavlas 2016; Ghysels et al. 2016; Blot et al. 2020a; Altavilla et al. 2019b; De Santis 2020), in supporting credit activity (Giannone et al. 2012; Martins, Batista, and Ferreira‐lopes 2019), inflation and economic activity (Gambacorta, Hofmann, and Peersman 2014), credit conditions (Ciccarelli, Maddaloni, and Peydro 2013) and asset prices (Rogers, Scotti, and Wright 2014; Alessi and Kerssenfischer 2019; Blot, Hubert, and Labondance 2020b). The literature on the pass-through has been initially developed by Cottarelli and Kourelis (1994), Cecchetti (2001), and Mojon (2000, 2001).

7 For example, the degree of competition explains some heterogeneity in the transmission of the common monetary policy across countries (Leroy and Lucotte 2015).

8 See Sander and Kleimeier (2004), De Bondt (2005), Marotta (2009), Belke, Beckman, and Verheyen (2013) and Andries and Billon (2016) for a survey on the empirical literature devoted to the retail bank interest rate pass-through.

9 In this simplified representation, we do not account for liquidity premia, which refers to the ability to sell an asset quickly. It is mainly related to the volume of transactions, which may be scarce. Here, we do not seek to disentangle pure liquidity risk and counterparty risk. In crises periods, those risks are hard to isolate and fear of insolvency will trigger a liquidity squeeze.

10 Assuming that there is no opportunity of arbitrage, the OIS rates would reflect risk-adjusted financial market participants’ expectations of the average policy rate over the horizon corresponding to the maturity of the swap (see Bauer and Rudebusch 2014).

11 Blot, Hubert, and Labondance (2020b) use the difference between the 10-year OIS rate and the Eonia to capture the term spread.

12 The CISS (Composite Indicator of Systemic Stress) is a real-time composite indicator of systemic risk computed by the ECB. For the EA as a whole, it includes several market indicators. A country-indicator (SovCISS) is also computed by the ECB. It captures the sovereign stress and is available for 11 EA countries.

13 The estimation could also account for heterogeneity in the business cycle. To that end, country industrial production may be introduced in the estimation in the short-term dynamics. However, this variable is never significant and does not improve the estimations.

14 The main adjustments have been realized for Greek interest rates where interest rate data on « total » maturity exhibits missing points. Yet, these series are generally highly correlated with the interest rate for agreed maturity up to one year. Missing values have been replaced with interest rates on loans below EUR 1 M for NFC, interest rates with an agreed maturity up to one year have been used. Data for loans for consumption in the Netherlands are missing from January 2003 to June 2010. Data have then been replaced by data from De Nederlandsche Bank available from January 2003. Data for Italy are also missing for consumption loans before January 2003. Data were also missing for interest from January 2003 to October 2006 on loans over 1 million in Belgium. They have been taken from the National Bank of Belgium but were yet available from March 2003 only. For deposit rates, missing values are taken from the Bank of Greece and De Nederlandsche Bank. The interest rate on deposits rate to households starts in January 2003 in Netherlands.

15 All the variables related to the asset purchase programmes and liquidity provisions are in current euros and considered in log in the estimations.

16 See Blot and Labondance (2013) and Aristei and Gallo (2014) for instance.

17 With the PMG estimator, we suppose that the parameters of the long-term relationship are identical across countries. It is consequently a more restrictive estimation compared to the MG estimator.

18 The robustness of these additional results has also been tested using alternative OIS rates and the PMG estimator. Results are available from the authors upon request.

19 These results are not affected if Greece is removed from the peripheral countries.