Do low interest rates really hurt banks’ profitability?

Do low interest rates really hurt banks’ profitability?

5 dicembre 2019

P. N. Barbieri, E. Canavesio, A. Cavallo, S. EmilianiL. ZicchinoL. Prosperi

Prometeia finds that the sign of the impact strongly depends on banks' ability to transmit negative rates to deposits

 
Abstract

While it is generally agreed that monetary policy expansionary measures prevented a sharp slowdown in economic activity and reduced deflation risks in the Euro area, there is a large debate on their effects on the banking system. When looking at inflation forecasts by the European Central Bank, the possibility of additional stimulus cannot be disregarded. Could a further cut in interest rates hurt banking profitability?

To answer this question, we quantify the impact on Italian banks’ profit of a 20-basis point cut in the ECB deposit rate. The full impact is estimated by considering both the direct negative impact on lending rates and customer margins and the positive effects from higher asset valuations and lending volumes. We find that the sign of the impact strongly depends on banks' ability to transmit negative rates to deposits.

 
Introduction

The monetary policy measures recently implemented by the ECB have undoubtedly contributed to support the European economy and reduce the risk of a sharp slowdown. However, it is not easy to assess whether these policies can have negative spillover effects on the banking system. What is certainly true is that European banks, and Italian ones in particular, have reported persistently low profitability in recent years. For Italian banks low profitability is the result of a stagnating domestic economy in addition to low interest rates, which also penalise core revenues of all other European credit institutions. Moreover, Italian banks have been burdened by the losses recorded on the NPL disposals and the one-off costs incurred to reduced branches and staff. Our forecasts for the Italian banking sector do not include further cuts in short-term interest rates. However, they cannot be ruled out as long as the central bank's forecast of euro area inflation in two years is 1.5%, as it is the case in their most recent published staff forecasts.

Could further cuts in interest rates hurt banking profitability? And would this make it more difficult for banks to extend credit? Brunnermeier and Kobi (2016) suggest that since banks are not able to fully transmit interest rate cuts on their cost of funding when interest rates are negative, accommodative monetary policy may turn contractionary because banks cease lending amid concerns about their equity positions.

In order to answer to the first question, it is necessary to assess the effects of policy rate cuts on bank profitability. As a first step, therefore, we quantify the impact of lower rates (than in our baseline scenario) on the net interest income and other revenues of Italian banks. We obtain that further cuts in rate would have a negative impact on net interest income. But what if banks could pass the rate cuts to their depositors? We estimate the savings in funding costs that can be achieved by transferring the interest rate cut to customer deposits, at least partially, which means, given the current conditions, charging negative interest rates. The rest of the article is structured as follows. Section 2 briefly illustrates the transmission channels of interest rate cuts on bank profitability. Section 3 describes data and methodology to estimate the impact of lower rates on bank profits. Section 4 shows the results of the exercise. Finally, section 5 shows an estimate of the contribution to net interest income of negative interest rates on deposits of households and non-financial corporations. Conclusions follow.

 
The transmission mechanism of interest rate cut: how does it work?

Figure 1 shows the transmission mechanism of the reduction of monetary policy rates on banks’ profitability. The first effect of lower policy rates is a reduction in lending rates, which should stimulate demand for new loans. If this is not associated with an equivalent reduction in funding rates, the spread between loan rates and funding rates narrows, partially offsetting the positive effect on volumes. This effect is even more pronounced when interest rates are already negative and when deposits, whose interest rate typically doesn’t fall below zero, are a high share of funding. 

Another negative contribution to net interest income comes from the effect of lower policy rates on the interests earned on securities, which will contribute less to net interest income.

On the other side, an expansionary monetary policy shock has a positive effect on the value of financial assets in banks’ balance sheets. The increase in price of the securities measured at fair value will boost the income statement. Finally, lower interest rates decrease the probability of a deterioration in credit quality, and, thus, implies lower loan loss provisions. 

Due to these effects going in opposite directions, it is not obvious what the aggregate impact on the income statement might be. 


 
Fig. 1  Transmission mechanism of the reduction of monetary policy rates on bank profitability
 
Methodology and data

In order to estimate the impact of interest rate cuts on the profitability of the Italian banking sector, we use the Prometeia IBASE (Italian Banking System Scenario Evaluation) model. Data come from Refinitiv and the Bank of Italy and cover the period March 1999-July 2019. The IBASE is an econometric model of the balance sheet and income statement of the Italian banking sector, developed for scenario analysis. In particular, IBASE allows to:

  • obtain the baseline forecast of the endogenous variables, free or conditioned to given paths of one or more variables
  • carry out scenario analysis 
  • conduct impulse-response and sensitivity analyses

The model consists of two blocks: a large Bayesian VAR (LBVAR) estimated on monthly data (Banbura et al., 2010), and a group of satellite models: (1) mixed frequency data sampling models (MIDAS) [1] and (2) Autoregressive Distributed Lag (ARDL) models estimated on monthly data. The first block of the model is used to estimate the relations between macro-financial and bank balance sheet variables. Table 4.1 shows all the variables included in the IBASE.

 
Tab. 1 Endogenous variables in the IBASE balance sheet section 
 

The second block includes income statement variables with annual frequency (Table 2). Each variable is estimated based on key variables observed at a monthly frequency.


Tab. 2 Income statement variables for the IBASE satellite models 
 

In particular, MIDAS models include all the variables for determining customer margin and net interest income, while ARDL models mainly include fees and trading and fair value valuation. 

Then, we compute the net interest income by estimating interests on securities and the net interest on interbank activity and funding from abroad. The former is function of the stock of medium- and long-term government bonds in banks’ balance sheets and the 10-year BTP yield, the latter is a function of the 3-month Euribor rate and total assets. 

Moreover, to estimate fees and commissions income for assets under management and assets under custody, we estimate unit margins and stocks. Instead, for payment services, overnight deposit management, guarantee, and trading and fair value valuation, we estimate revenues directly. The choice of approach depends on the goodness of fit of the model and the significance of the variables, as well as the difficulty of identifying specific drivers for each NFCI component.

 
Results

The alternative scenario: a 20-basis point reduction in the ECB deposit rate 

In order to assess the effects of a decrease in policy rates, we build an alternative scenario over the horizon 2019-2022, where we assume a one-off 20-basis point cut in the ECB deposit rate (the rate is assumed to be cut at the start of the period and then to stay at -0.7% until 2022). The rate cut leads to a reduction in government bond and interbank rates (respectively, -20 basis points for 10-year BTP yields and -14 basis points for 3-month Euribor), an increase in GDP of half a percentage point cumulated over three years and, finally, an appreciation of the Italian stock market index (+13% at the peak). This alternative scenario is obtained from the Prometeia iDREAM model [2], while its impact on banking variables is assessed by using the IBASE [3] model on latest available data [4]. 

Negative effect on net interest income… 

A cut in the interest rate on the deposit facility of the ECB corresponds to a fall of almost 1 billion euros in net interest income due to several concurrent factors. The first one is a reduction of 1.2 billion in interests on loans, due to the fall in lending rates (-10 basis points). This effect more than offsets the increase in credit demand due to lower rates.

Our results are in line with those of the Bank Lending Survey (BLS) [5] (Fig. 2), where banks give their view on the effects of persistent negative policy rates. Among banks, there is a strong agreement that low policy rates are transmitted to lending rates. As far as loan volumes are concerned, banks declare that there are positive effects, but the consensus on this statement is less strong. Overall, 74% of the sample banks claim that negative ECB rates have negative effects on net interest income.

 
Fig. 2 Impact of negative interest rates on ECB deposits on interest rate margin and new loans issuance EMU 
Source: The Euro area bank lending survey – III quarter 2019. Notes: Net percentages are defined as the difference between the sum of percentages of “increased considerably” and “increased somewhat” and the sum of percentages “decreased somewhat” and “decreased considerably”.
 

…if banks cannot transmit negative interest rates on deposits 

In order to assess the impact of a rate cut on banks, we first assume a floor at zero on the rate on bank deposits. Under this assumption, the interest on deposits remains unchanged when the central bank reduces the rate on deposits since it is already negative. However, in our model the rate cut of 20 basis points implies a reduction in the cost of other bank liabilities, namely debt securities, which translates into savings of almost 1 billion (or 4% of the net interest margin in 2021). Finally, the rate cut also affects the yield on the 10-year BTP, with translates into a further negative effect on the net interest margin (approximately -0.3 billion euros). All these combined effects lead to a fall of 0.7 billion in net interest income (Fig. 3) (or 3% of the net interest margin in 2021).

 
 

The effect on net fees and commission income (NFCI)

The reduction in short-term interest rates and government bond yields, combined with an appreciation of the stock market, has a positive effect on NFCI (+0.3 billion euros, around a 1% increase of the NFCI in 2022, Fig. 4), mainly from asset management. On the one hand, the positive performance of asset management is due to the low remuneration of safer assets, like good quality bonds. The demand of investment products with higher returns, as an alternative to safer assets, leads to an increase in the fees and commissions from asset management. On the other hand, NFCI from assets under custody are substantially stable since net outflows are offset by higher unit margins. NFCI related to other services (e.g. payment services, overnight deposits management and guarantees given) are almost unchanged, given the limited effects on new transactions and on unit margins. Instead, fees and commissions from trading and fair value valuation increase (+0.2 billion euros, less than a 1% increase of the NFCI in 2022), mainly due to the appreciation of domestic sovereign bonds measured at fair value (Fair Value Trough Profit or Loss, FVTPL).

Overall, the cut in interest rates leads to a drop in gross income for the Italian banking sector of 0.26 billion euros compared to the baseline scenario. The positive change in profits from services and the effects of the appreciation of sovereign bonds, therefore, do not seem to be able to offset the negative effect on core revenues. 

What is the impact if we did not assume a floor at zero for the remuneration of deposits? 

A key assumption in our simulation is that the remuneration of deposits cannot fall below zero. In other words, banks are not allowed to transfer (even partially) negative ECB policy rate to customers. If we relax this assumption, the results change. A decrease in the Euribor of 14 basis points (following a 20 basis point rate cut in the policy rate) reduces the interest rates on deposits by 5 basis points, while volumes remain stable. In this case, the overall impact on gross income would be approximately 0.2 billion euros (Fig. 4). This effect is probably underestimated since the model coefficients incorporate a low elasticity of the rate on deposits to interbank rates. However, a decrease in interest rates on deposits for all customers is quite an extreme assumption, in light of the choices made by some banks in Europe, for example, to charge negative rates only on corporate clients. We discuss this issue in the next section.

 
Negative interest rates on customer deposits

In Europe some banks started to apply negative interest rates on non-financial corporations’ deposits after the ECB interest rate cut of June 2014. In some cases, banks applied lower interest rates than the rate on the ECB deposit facility. On average, between July 2014 and September 2018, interest rates in the euro area were negative for 5% of total overnight deposits and 20% of non-financial corporation overnight deposits [6].   

Why should customers accept negative interest rates? 

Altavilla et al. (2019) claim the answer to this question depends on the type of customer: non-financial corporations have large cash holdings and need liquidity for their daily operations, therefore cannot respond quickly (and repeatedly) to changes in deposit conditions. Additionally, the extended period of negative interest rates that we are experiencing is characterized by low investments and a high demand for secure assets, that has shifted corporate customers' preferences towards ‘sound’ banks.  A deposit in a ‘sound’ bank can be considered a safe and liquid asset, even when the amount exceeds the deposit guarantee threshold. Instead, households, having smaller deposits, can more easily withdraw and hold them in cash or select banks that apply better conditions without having to worry too much about the soundness of a bank if all the amount deposited falls below the guarantee threshold.

The latest data on interest rates on overnight deposits for non-financial corporations show negative values in the Netherlands (-0.06%, average for the first eight months of 2019), Germany (-0.03%), Latvia (-0.03%) and Luxembourg (-0.01%) (Fig. 4.5). Despite negative rates (Fig. 6), the volume of non-financial corporations’ overnight deposits increased over the period June 2004-September 2018 (+66% in the Eurozone, +46% in the Netherlands, +47% in Germany, +32% in Latvia, +30% in Luxembourg). 

What about households? 

The first cases of negative interest rates in Europe are in Switzerland and Denmark, countries with negative interest rates since more than 5 years: recently UBS and Jyske Bank have announced the introduction of a rate of -0.75% on overnight deposits exceeding, respectively, 2 million Swiss francs (1.8 million euros) and 750,000 Danish kroner (100,000 euros).

In the euro area, banks have just started to consider the possibility of negative interest rates on household deposits. Following the latest cut of the rate on the ECB deposit facility in September, Berliner Volksbank, Germany's second largest cooperative bank, started to charge -0.5% on overnight deposits above 100,000. Although the example has not yet been followed by other banks, many have signalled the intention to apply negative interest rates on deposits above a certain threshold (100,000 euros being the most cited). Significant European banks seem to be waiting for an official ECB opinion, because the move exposes them to legal risk, at least in some countries. There is no specific European banking regulation prohibiting to charge negative rates, but such a measure could come into conflict with national laws on private savings protection.

 
 

What does regulation say in Italy? 

In Italy, the Article 1834 of the Civil Code regulates deposit contracts in a generic way, leaving full freedom to the contracting parties to set specific conditions. The negative interest rate on overnight deposits would potentially come into conflict with this article because: (1) the depository is required to return the same amount received, and (2) the contract of deposit is by its nature considered free of charge and, in the case of negative interest rates, the amount to which the depositor is entitled to would be eroded, infringing the condition (1). In addition, the Article 118 of the Consolidated Law on Banking (Testo Unico Bancario, TUB), which regulates unilateral changes to the contractual conditions, pinpoints the specific case of changes in interest rates consequent to monetary policy decisions: they are allowed as long as “their application does not cause prejudice to the customer”.

How much could Italian bank benefit from negative rates on customer deposits? 

For Italian banks, customer deposits represent quite a relevant cost (around 400 million euros estimated in 2018) as the interest rates are still positive, even if very low (Fig. 7). We performed a simulation to assess the amount of savings that banks could obtain by applying negative interest rates to overnight deposits. Our goal is to quantify the effect on banks’ revenues, regardless what happens to policy rates in the future and all other effects negative rates may cause to banks’ balance sheets. This exercise can give a ballpark estimate of the savings in funding costs for banks if they decide to pass the negative rates to customers.
As a first step, we impose negative interest rates on overnight deposits of non-financial corporations (-0.03%, from 0.06% in June 2019, Fig. 6). This would produce savings of approximately 250 million euros, equivalent to 0.7% of net interest income in 2018. The second hypothesis extends negative interest rates to household deposits over 100,000 euros. The assumption is that households pay 0.03% (the same as corporate customers) on their overnight deposits, instead of receiving 0.05% as they did in June 2019. Under this second assumption, more extreme and less likely than the first one, Italian banks would save more than 400 million euros, 1.3% of the interest margin: deposits would then become a profitable business (Fig. 8). However, this second hypothesis remains very unlikely; recent statements by bankers suggest that negative interest rates would not be charged to customers who are willing to transfer part of their deposits into asset management products.


 
 

The introduction of negative interest rates on overnight deposits of non-financial corporations and households leads to higher banks’ savings (250-400 millions vs 200 millions) than in the exercise of Section 4. Although the amount of deposits with negative rates is smaller than in the previous exercise (only non-financial corporations and households above 100,000 euros vs the entire stock of deposits in the exercise in Section 4), the drop in the deposits rate is larger. This is because in the first exercise estimates of the rate charged on deposits come from an econometric model and are the result of an endogenous response to monetary policy. The second exercise calculates the savings based on unchanged volumes of deposits and an interest rate assumed to be equal to the average rate on non-financial corporation deposits in Germany.

 
Conclusions

This paper tries to estimate the effect of negative interest rates, possibly even lower than the current ones, on Italian banks’ profits. 

Our analysis shows that a further cut in policy rates would hurt the profitability of the Italian banking sector. However, the impact is limited and could change sign if Italian banks were able to transfer negative rates to deposits. In particular, charging small negative interest rates to deposits of non-financial corporations and to households’ deposits above 100,000 euros would neutralize the current negative contribution of the remuneration of deposits to the Italian banks’ net interest income.

 
 
References

Altavilla, C. and Burlon, L., Giannetti, M. and Holton, S., “Is there a Zero Lower Bound? The Effects of Negative Policy Rates on Banks and Firms” (June 7, 2019). ECB Working Paper No. 2289 (2019)
Banbura, M., Giannone, D., Reichlin, L., "Large Bayesian vector auto regressions", Journal of Applied Econometrics, John Wiley & Sons, Ltd., vol. 25(1), 71-92, 2010
Banbura, M., Giannone, D., Lenza, M., "Conditional forecasts and scenario analysis with vector autoregressions for large cross-sections", International Journal of Forecasting 31.3 739-756, 2015
Barbieri, P., Lusignani, G., Prosperi, L., Zicchino, L., “Model-based approach for scenario design: stress test severity and banks’ resiliency”, Prometeia working paper, 2019
Markus K. Brunnermeier & Yann Koby, 2019. "The Reversal Interest Rate", IMES Discussion Paper Series 19-E-06, Institute for Monetary and Economic Studies, Bank of Japan.
De Meo, E., Prosperi, L., Tizzanini, G., & Zicchino, L., “Forecasting Macro-Financial Variables in an International Data-Rich Environment Vector Autoregressive Model (iDREAM)”, Prometeia working paper, 2018 Ghysels, E., Santa-Clara, P., Valkanov R., “The MIDAS touch: Mixed data sampling regression models”, UNC and UCLA Working paper, 2002
Ghysels, E., A. Sinko, and R. Valkanov, “MIDAS regressions: Further results and new directions”. Econometric Reviews 26 (1), 53-90. 2007
Schumacher, C., “A comparison of MIDAS and bridge equations”. International Journal of Forecasting, 32, issue 2, p. 257-270, 2016.

 
 

Appendix

The MIDAS estimation
 
[1] See the appendix for a description of MIDAS models.
[2] De Meo et al. (2018) describes the model.
[3] Following the approach of Banbura et al. (2015), the effects are computed as difference between the main forecast of the model and the forecast conditioned on the alternative scenarios. The application of this approach on the IBASE model can be found in Barbieri et al. (2019).
[4] The macrofinancial data are as of July 2019, whereas the income statement data are those published in December 2018.
[5] October Bank Lending Survey (BLS) “The euro area bank lending survey – Third quarter of 2019”.
[6] Source: Altavilla, Burlon, Giannetti, Holton (2019). At the end of September 2019 (2 weeks after the ECB deposit rate cut) Deutsche Bundesbank surveyed 220 lenders and revealed that almost 60% of German banks apply negative interest rate on non-financial corporation overnight deposits and more than 20% apply negative interest rate on household deposits.
[7] The paper identifies two characteristics for ‘sound’ banks: (1) they are located in non-stressed Euro area countries; (2) they have lower proportion of non-performing loans and CDS spread.
[8] Schumacher C. (2016) and Ghysels, E., A. Sinko, and R. Valkanov (2007) for a more detailed discussion of alternatives polynomial's functional forms. For example, for income on loans the two regressors used are: (1) rates on loans and (2) total net loans.
Leggi tutti gli insight