Adapting lending policies in a ‘negative-for-long’ scenario

Miguel García-Posada and Sergio Mayordomo

In February, the Bank hosted its inaugural Bank of England Agenda for Research (BEAR) conference, with the theme of ‘The Monetary Toolkit’. As part of our occasional series of Guest Posts by external presenters at Bank research events, the authors of one paper from the BEAR conference outline their findings on the effect of negative rates on Spanish banks…

Over the past decade, several central banks have cut policy rates below zero. These are unlikely to work in the same fashion as rate cuts in positive territory, because of difficulties with charging negative interest rates on retail deposits, and because of banks’ negative return on excess liquidity. In a recent paper we explore the effect on bank lending by combining data on exposure to negative rates with banks’ balance sheets, the Spanish credit register and firms’ balance sheets. We find it was only after several years of negative rates, when deposit rates reached zero, that affected banks began to decrease credit supply and increased loan rates. This effect was especially strong for undercapitalised banks and lending to risky firms.

Why might negative rates work differently?

Economic theory suggests the effect on bank lending of rate cuts in negative territory could go either way. One the one hand, it is well documented that the existence of cash as an alternative store of value means it may be difficult for banks to set negative rates on deposits (eg Heider et al (2019)Eggertsson et al (2020)). This squeezes the net interest income of banks with a high deposit share: retail deposit rates are stuck at zero, while the rest of liabilities (eg wholesale funding) and assets (eg floating-rate loans) are repriced at a lower level.

On the other hand, because banks with excess liquidity earn a negative return, they have incentives to increase their lending to the private non-financial sector in a bid to reduce their excess liquidity holdings (eg Basten and Mariathasan (2018); Demiralp et al (2019)). This portfolio-rebalancing channel may imply higher risk taking, as risk-free excess liquidity is converted into bank lending.

The analysis of the impact of negative interest rates on banks’ credit supply and risk taking is likely to depend on the time over which policy rates are negative. A ‘negative-for-long’ scenario might look quite different to a shorter spell of rates below zero. As documented by Eggertsson et al (2020), negative rates may have contractionary effects only when retail deposit rates reach the zero lower bound (ZLB).

Estimating the effect of negative rates on Spanish banks

We study the effect of the ECB’s negative deposit facility rate (DFR) on the supply of credit by Spanish banks to non-financial corporations during a protracted time period, 2014–19. Spain provides a good case study here because of exposure to negative policy rates, combined with an attractive dataset which is ideal for identifying the effects of negative rates on bank lending. In addition, deposit rates in Spain were well above the ZLB when negative interest rates were introduced in the euro area and it took them several years to reach the ZLB, which provides an ideal opportunity to study the effects of a ‘negative-for-long’ scenario on credit supply. To do this, we build a unique dataset that comprises the universe of loans granted to Spanish firms from the Credit Register of the Bank of Spain, banks’ and firms’ balance sheets and confidential survey data from the ECB’s Bank Lending Survey (BLS).

The BLS is a quarterly survey through which euro area banks are asked about developments in their respective credit markets since 2003 (for more detailed information see Köhler-Ulbrich, Hempell and Scopel (2016)). Our identification strategy exploits heterogeneous exposure to negative rates across banks. Specifically, we estimate an ‘affected dummy’, which is based on the probability that a bank is adversely affected by the negative interest rates (affected for short), based on the confidential answers to the BLS. We assume that a bank is affected, and our ‘affected dummy’ equals 1, if the probability that it reports that the ECB’s negative DFR contributed to a decline in its net interest income is higher than 75% (ie the median of the distribution of Spanish banks in 2014). Since the literature suggests several channels through which negative interest rates affect banks (ie retail deposits, excess liquidity, floating-rate loans, short-term interbank positions), the BLS provides a summary measure of exposure to them.

We derive the affected dummy from a probit regression where our dependent variable is a dummy which records if the bank reported that the ECB’s negative deposit facility rate contributed to a decrease of the bank’s net interest income (NII) in the past six months, and the regressors include the deposit ratio and the liquidity ratio. In addition, affected banks may have a high share of floating-rate loans or short-term loans, which are repriced at a lower rate following a reduction in the official interest rate. Therefore, we also include the weight of loan overdrafts and loans with a maturity up to one year in the total stock of loans, respectively, and control for other bank characteristics such as solvency, profitability, size and Eurosystem borrowing.

In the second step, we then run regressions using a 3 dimensional (bank-borrower-time) panel, regressing bank lending on the treatment variable, firm-time and bank specific fixed effects.

Banks’ self-assessment of the impact of negative interest rates on their balance sheets may pose an identification challenge, as weak banks may have incentives to strategically misreport their evaluation of the policy in order to ‘blame’ it for their poor performance. However, we obtain similar results in several robustness analyses solely based on hard data, in which we classify banks according to their deposit ratios or their share of credit at floating rates.

Importantly, we allow for different effects in different periods by interacting our key regressor with time dummies, to analyze the dynamic impact of negative interest rates between 2014 and 2019, a period in which deposit rates in Spain exhibited a downward trend until reaching the ZLB. We also address two key identification challenges. First, we disentangle credit supply from credit demand by including firm-time fixed effects à la Khwaja and Mian (2008), which implies comparing lending decisions of multiple banks to the same firm within the same period. Second, we control for the potential effects of other ECB policies on banks’ credit supply, such as the targeted long-term refinancing operations (TLTROs) and the expanded asset purchase programme (APP).

Four key results

First, we find that affected banks contracted their lending supply to firms only during the last sub-sample period (2018–19), while there is no effect during the earlier periods. This result may be explained by the fact that retail deposit rates were high in Spain at the time of the introduction of the negative interest rates, so they had plenty of room to decline before reaching the ZLB in 2018. Nevertheless, since policy rates in the euro area were lowered several times since 2014, we cannot rule out a complementary explanation, namely that policy rates reached the reversal rate (Brunnermeier and Koby (2019)), which is the rate at which accommodative monetary policy ‘reverses’ its intended effect and becomes contractionary for lending.

Second, we also find that the effect of negative interest rates on banks’ credit supply was heterogeneous and depended on the level of banks’ capitalisation. In particular, we observe that affected banks with low capital ratios curtailed their lending supply to firms, but they only did so during the last period 2018–19, when deposit rates reached the ZLB. Hence, our findings suggest that the reversal rate would be bank-specific and dependent on banks’ capitalization levels.

Third, splitting our sample into safe and risky firms, we find that affected low-capitalised banks reduced their credit supply to risky firms in the last two sample periods, 2016–18 and 2018–19, although the effect is much stronger in the latter period. By contrast, there is only a marginally significant effect on safe firms in the last period, and its size is significantly smaller than that for risky firms. Therefore, our findings indicate that affected low-capitalised banks contracted their credit supply to risky firms prior to restricting it to safe firms and in a greater magnitude, arguably because loans to the former consume more regulatory capital than exposures to the latter. This evidence suggests that affected low-capitalised banks took less risk because of their lack of capital buffers to absorb losses and the need to meet capital requirements. Fourth, looking at the impact of the negative interest rates on firms’ total borrowing we find no evidence that companies whose main credit institution was an affected low-capitalised bank experienced a contraction in their total bank credit. This evidence suggests that the lower supply of credit by affected low-capitalised banks was offset by the higher lending supply by non-affected banks, with capacity for taking additional risks thanks to their higher capital buffers. Therefore, while the reversal rate might be reached by some affected undercapitalised banks, there seemed to be no aggregate effect on the supply of lending to firms.

Miguel García-Posada and Sergio Mayordomo work at Banco de España.

If you want to get in touch, please email us at or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.