Negative rates and the transmission of monetary policy

Prepared by Miguel Boucinha and Lorenzo Burlon^[1]

Published as part of the ECB Economic Bulletin, Issue 3/2020.

1 Introduction

As structural and cyclical factors have brought nominal interest rates closer to zero, the need to ease financing conditions further has prompted the adoption of a negative interest rate policy (NIRP). The introduction of negative policy rates has been part of a comprehensive policy strategy adopted by the ECB since mid-2014 in order to stave off the unprecedented disinflationary forces that arose in the aftermath of the global and sovereign debt crises. The ECB has cut its deposit facility rate (DFR) into negative territory five times since 2014. The latest lowering of the DFR in September 2019 and the associated market expectations of a longer period of negative rates have reignited the question of how negative rates are transmitted to the economy, especially through banks, and whether they may have counter-productive effects by impinging on banks’ intermediation capacity.

Negative rates are transmitted via different channels. Negative rates soften the expectation of markets that current and future short-term rates cannot be negative. By lowering the perceived lower bound of central bank rates, negative rates allow the monetary accommodation to propagate through the entire yield curve. Moreover, investor demand for longer-dated assets increases more than when rates are positive, exerting further downward pressure on the term premium, i.e. the compensation that investors demand for the uncertainty regarding the future path of interest rates. Finally, commercial banks are incentivised to expand lending so as to avoid the negative rate applied to their excess holdings of reserves with the central bank (excess liquidity) in a situation in which the cost of liabilities is partially constrained.

The transmission of NIRP via banks could in principle be hindered by potential large-scale shifts into cash and downward pressure on bank profitability. The transmission of monetary policy could be diluted if investors hoard cash rather than rebalancing their portfolios towards longer-term or riskier assets. So far there are no signs of large-scale liquidity “leakages” of this type, mainly owing to the costs of forgoing the services provided by central bank reserves or commercial bank deposits as a means of conducting payments and storing value. This is partly due to the limited transmission of negative deposit rates to retail deposits, especially of households, which can, however, dent bank profitability and ultimately hamper their ability to provide lending to the real economy. Potential factors that may hinder the transmission of monetary policy in the event of any further extension of the policy or deeper cuts into negative rate territory must therefore be monitored closely.

Overall, negative interest rates have supported economic activity and ultimately contributed to price stability. As a result of NIRP, lending volumes have expanded and the creditworthiness of borrowers has improved, thereby mitigating the impact of lower interest margins on overall bank profitability. While NIRP, and more generally the low level of interest rates, may contribute to the build-up of debt and spur over-pricing of financial assets or exuberance in housing markets, when such phenomena are identified they are generally best addressed by targeted macroprudential policies. Meanwhile, the first-order and tangible effect of NIRP on financial stability has been that it has enhanced it by improving the sustainability of outstanding debt.

This article is organised as follows. Section 2 explains how negative rates are transmitted via banks and financial markets. Box 1 illustrates the transmission of NIRP to financial market prices. Section 3 then turns to the impact of negative rates on bank profitability and risk-taking. Box 2 describes the transmission of negative rates in money markets. Section 4 reports empirical evidence on how negative rates affect the broader economy. Section 5 concludes.

2 The transmission of negative interest rates

In June 2014 the ECB introduced NIRP as part of a broad credit easing package, leading to lower bank deposit and lending rates. NIRP shares some of the basic easing features of a standard interest rate cut, as is evident in the reaction of bank funding and lending rates to the original cut in the DFR into negative territory as well as to the subsequent reductions (see Chart 1). However, when rates are below zero, additional channels are activated. Some enhance the stimulus provided by the cut, while others can dampen transmission via banks.

NIRP reduces bank funding costs, stimulating loan supply. Composite funding costs for euro area banks have been compressed since 2014, helping to maintain intermediation margins for banks with an average funding structure. Moreover, there is evidence that, as negative rates persist, even banks that rely predominantly on deposit funding eventually pass the interest rate reduction on to at least part of their deposit base, namely corporate deposits and large retail deposits. However, some degree of downward rigidity in retail deposit rates remains, limiting the scope for further improvements in funding costs without changes to the funding structure.

Moreover, there are additional channels that strengthen the transmission of negative rates compared to a standard rate cut, especially when the latter occurs close to the perceived “lower bound”.^[2]

First, negative interest rates remove the non-negativity restriction on current and future expected short-term rates and, therefore, monetary accommodation can propagate throughout the yield curve. Before June 2014 the distribution of future expected short-term rates was effectively truncated at zero, as market participants were not assigning significant probability to future rates being negative. The introduction of negative policy rates allowed the constellation of rates to expand into negative territory. As the expansion of the interest rate distribution on the negative side affects not only current rates but also expected future rates, the stimulus also propagates to longer maturities (see Box 1).

Second, NIRP increases investor demand for longer-dated assets, exerting further downward pressure on the term premium. The negative remuneration of excess liquidity and its transmission to short-term interest rates create incentives for investors to rebalance their portfolios towards assets with longer maturities, increasing the demand for these securities relative to their supply. While this compression in yields of longer-dated assets also occurs following policy rate cuts in positive territory, it becomes more prominent when rates are below zero, as some investors are particularly averse to accepting negative nominal returns, for instance because they are committed to providing positive nominal returns to their final beneficiaries. This ultimately exerts extra downward pressure on the term premium.

Third, commercial banks are encouraged to expand lending to avoid negative interest on their holdings of excess liquidity. Banks face a direct charge on their excess liquidity, which exerts pressure on their profitability, particularly as the remuneration of their liabilities is, in part, bounded at zero. They can avoid this charge by rebalancing their portfolios in favour of credit expansion or by purchasing securities.^[3] In fact, returns on all liquid assets in banks’ portfolios drop relative to loans, providing them with a strong incentive to rebalance in favour of credit origination.^[4]

At the same time, negative rates also entail frictions that may hinder the transmission of monetary policy via banks. Banknotes yield a zero nominal return, so, if deposits carry negative nominal interest rates, depositors may at some point resort to cash hoarding as a way to avoid them. This implies that retail deposit rates, which are quite responsive downwards when policy rates adjust in positive territory, become much stickier. As the returns on the assets held by banks are not subject to the same downward stickiness, this results in a compression of banks’ interest margins, which applies pressure on bank profitability. Moreover, banks tend to hold assets with relatively long durations, which reprice more slowly than their liabilities, so protracted periods of negative rates may dent banks’ net worth and thereby impair the bank-based transmission mechanism of monetary policy. Moreover, the pressure to maintain intermediation margins and to move away from less risky liquid assets provides banks with an incentive to increase their risk tolerance, which may eventually lead to excessive risk-taking.^[5]

More generally, investors (banks and non-banks) might hoard cash rather than rebalance their portfolios towards longer-term or riskier assets, although there are no strong signals of such a “leakage”. To the extent that cash and deposits are perfect substitutes, agents would respond to negative interest rates on deposits by holding cash, implying an effective lower bound of zero for interest rates. However, in practice they are not perfect substitutes. In fact, deposits have several advantages, such as being a more efficient means of storing value and conducting payments. This convenience has an intrinsic monetary value. So far, there is no evidence of large-scale shifts into cash among depositors such as households, corporates or non-bank financial institutions (see Chart 2, panel a). Banks themselves may, in principle, also use their holdings of excess reserves with the central bank to acquire and hold banknotes as a way to eschew the negative rate on those reserves. While an increase in banks’ holdings of banknotes has indeed been observed, this has not occurred to an extent that would provide strong signals of a leakage of liquidity away from the banking sector and into cash, as there is very little correlation between excess liquidity and actual vault cash (see Chart 2, panel b).

Retail deposits tend not to carry negative rates, either because of legislative hurdles and litigation risks or concerns about deposit withdrawals. Legislators and courts have so far maintained a legal framework that poses a series of legal constraints and litigation risks related to the application of negative deposit rates. These vary from country to country and according to the sector of the depositor and are also dependent on whether they would apply to new or existing contracts. However, in the majority of euro area countries, the transmission of negative rates to corporate deposits is not subject to legal constraints and is indeed a relatively widespread phenomenon which has been increasing over time (see Chart 3).^[6] Moreover, fees and commissions can lead to an effective negative nominal return on deposits, even if they do not carry a negative interest rate. Nonetheless, there is no evidence that, as the remuneration of retail deposits passed the zero line, banks experienced significant outflows of deposits. This suggests that, until it reaches a level that exceeds the cost of hoarding cash, a negative nominal rate would only generate a change in behaviour among retail depositors as a result of some form of money illusion or other behavioural bias.^[7]

Other instruments in the current policy toolbox, in particular asset purchases, forward guidance, targeted longer-term refinancing operations (TLTROs) and the two-tier system for reserve remuneration, complement and reinforce the transmission of NIRP. Asset purchases increase the volume of excess liquidity in the banking system, strengthening the pressure on individual banks to shift their excess liquidity to other banks by acquiring assets. Forward guidance reduces uncertainty about the future path of interest rates. TLTROs ensure that banks can obtain funding at very low interest rates (as low as the DFR) and help to channel the resulting reduction in funding costs towards new lending to firms and households (other than for house purchases). The two-tier system for reserve remuneration allows the accommodative effects of NIRP to be maintained while mitigating the downward pressure on bank profitability stemming from the negative remuneration of excess liquidity holdings.

3 Impact on bank profitability and risk-taking

Euro area bank profitability has gradually recovered from the significant decline that followed the crisis, but remains low by historical standards (see Chart 4). Bank profitability showed an upward trend in the run-up to the financial crisis, followed by a decline, which was largely driven by a sharp increase in loan loss provisions. Since 2012, bank profitability has gradually recovered, as loan loss provisions have decreased and net interest income has remained broadly stable. Nevertheless, the return on equity stood at around 6% in the third quarter of 2019, which is below most estimates of the cost of equity – for example, the majority of banks participating in the European Banking Authority’s Risk Assessment Questionnaire report an estimated cost of equity of between 8% and 10%.^[14] Beyond the cyclical environment, structural challenges afflicting the euro area banking sector played a significant role in depressing the overall level of bank profitability.^[15] Low profits limit the scope for organic recapitalisation of banks through retained earnings, which is necessary to build and retain the capacity to intermediate and provide credit to firms and households. At the same time, market valuations of bank equities are low, also compared to non-financial or other financial firms. This, in turn, can inhibit banks from tapping markets to raise capital, as doing so when valuations are low would heavily dilute the holdings of current shareholders.

Negative interest rates have both adverse and beneficial effects on bank profitability, and the net effect is ultimately an empirical question. In addition to the constraints on lowering retail deposit rates below zero, NIRP leads to a more significant reduction in rates at the longer end of the term structure than a conventional policy rate cut (see Box 1). Since banks typically engage in maturity transformation, obtaining funding at shorter maturities and durations than those at which they lend, this contributes to an additional squeeze on net interest margins compared to a rate cut in a positive interest rate environment. However, there are also channels through which NIRP supports bank profits, in particular through its impact on asset valuations and, more importantly, on the macroeconomy.^[16] The positive impact of the policy on macroeconomic conditions leads to higher intermediation volumes, supporting net interest income. Moreover, the improved economic outlook and the lower level of interest rates boost borrower creditworthiness, reducing costs stemming from loan loss provisions. At the same time, decreases in yields driven by NIRP are reflected in an increase in the value of the securities held by banks, thereby leading to (temporary) capital gains for banks.

The main component of bank profitability, net interest income, has so far remained rather resilient, as larger intermediation volumes have offset the lower margins. Chart 5 shows that the decrease in income on loans and securities has been offset by lower interest expenses on deposits and debt securities issued, based on actual developments in interest rates and volumes. In the chart, these movements are represented by the red circles. The developments are further decomposed into price and quantity effects (yellow and blue bars). These show that the negative impact of lower interest rates on the income from loans has been partly offset by higher intermediation volumes. The negative quantity effect on debt securities reflects banks’ portfolio rebalancing towards loans. Analogously, on the liability side, the liquidity injected by the APP led to an increase in deposits by the money holding sector, thereby increasing the relevance of this funding source for banks. At the same time, lower interest rates on deposits and debt securities issued supported net interest income.

To comprehensively assess the impact of NIRP on bank profitability, it is crucial to look at components beyond net interest income and to take into account the impact of the policy on the broader economy. As discussed above, one would expect, all else being equal, that lower and, in particular, negative interest rates would lead to a decline in banks’ net interest margins, and indeed several empirical studies highlight this relationship.^[17] However, the overall impact on bank profitability depends not only on developments in net interest margins but also on the macroeconomic outlook, which affects credit demand and borrower creditworthiness and therefore intermediation volumes and loan loss provisions. Therefore, a comprehensive assessment of the impact of low or negative interest rates on bank profitability requires consideration not only of developments in interest rates but also of the impact of monetary policy on economic activity.

Negative interest rates have had a broadly neutral impact on bank profitability so far, as their negative effect on net interest income has been offset by a positive effect on borrower creditworthiness. Chart 6 reports the results of a comprehensive assessment of the impact of NIRP on bank profitability. The exercise was conducted by comparing actual developments in the main components of bank profitability in the period during which NIRP has been in place with those under a counterfactual scenario in which NIRP was not implemented. This scenario is obtained from the simulation of a BVAR model under the assumption that the zero lower bound would be enforced at all times, thereby preventing the term structure of interest rates from assuming negative values across all maturities.^[18] Under this scenario, higher interest rates would have been reflected in weaker loan growth and lower borrower creditworthiness and therefore in higher loan loss provisions. Actual developments in bank profitability components (red circles) can therefore be decomposed into a part that represents the estimated impact of NIRP (yellow bars) and one which would have been observed under the counterfactual scenario (blue bars). In line with the frictions highlighted above, the estimated impact of NIRP on net interest income is negative.^[19] The direct cost of remunerating banks’ holdings of excess liquidity at the negative DFR is shown as a separate component and, overall, is limited. The chart also illustrates the positive impact of negative interest rates on non-interest income. This effect is relatively small and short-lived, as it mainly reflects the impact of decreases in interest rates on the value of the securities held by banks. More importantly, NIRP is estimated to have accounted for a significant share of the observed decline in loan loss provisions. Overall, taking all effects into account, NIRP is estimated to have had a negligible impact on bank profitability up to now.

While some banks are more exposed to the low interest rate environment, so far there is no evidence of an overall negative impact of NIRP on bank profits across bank business models. This conclusion is based on an assessment of the impact of NIRP on bank profitability across bank business models using a dynamic macroeconomic model similar to that described above. The results indicate that, while NIRP has a more negative impact on the net interest margins of retail banks – those relying more on retail deposits – the same banks also benefit more from the improved credit quality and intermediation volumes that are linked to the positive impact of monetary accommodation on the macroeconomic outlook.

Looking ahead, the balance between costs and benefits may evolve if interest rates remain low for a long period of time. The detrimental impact on net interest margins is likely to be more significant as rates remain low for longer.^[20] This is because banks are still benefiting from income on fixed-rate loans originated when interest rates were higher, and this income will progressively decrease as legacy loans mature.^[21] Moreover, the scope for further savings in provisions and funding costs is limited, in particular for banks whose loan portfolio is concentrated on low-risk borrowers. The cost for banks of holding excess reserves also increased following the cut in the DFR in September 2019 and the increase in excess liquidity driven by the reactivation of the APP, but these effects are mitigated by the introduction of the two-tier system for the remuneration of excess reserves (see also Box 2).

While the compression of intermediation margins can lead to an increase in the riskiness of bank portfolios, so far there is limited evidence that such an increase has been excessive. Banks may be more willing to raise the risk profile of their investments in the aftermath of NIRP in a search for higher yields.^[22] Whether this riskiness can be deemed excessive depends on whether it is adequately priced in in the terms and conditions of new loans and, even more importantly, on whether it is factored in in credit standards. Evidence from the euro area bank lending survey (BLS) shows that the easing of margins on riskier loans has been much more contained than that on average loans, and, over the last year, margins on riskier loans have actually been tightening.^[23] Moreover, model-based evidence suggests that the increase in the risk profile of portfolios has been adequately reflected in lending rates, although some sectors, like consumer lending, have started to show signs of exuberance.^[24] Regulatory pressures and supervisory oversight have played, and continue to play, a crucial role in fostering a balanced coexistence between negative rates and prudent loan pricing practices.^[25]

While the broader low interest rate environment can pose financial stability risks, targeted macroprudential action is best placed to address specific fragilities. Excessive debt, overpricing of certain financial assets or exuberance in some local housing markets can be supported by the broader low interest rate environment and may threaten financial stability. For example, the January 2020 BLS has highlighted that, even amid growing concerns about the macroeconomic outlook that have curtailed borrowing by firms, net demand for housing loans is continuing to increase at above the historical average rate.^[26] While the overall rate of growth of housing loans in the euro area is not very high compared to past episodes, in some countries it may pose financial stability concerns, as it may generate bubbles in local house prices that may warrant targeted macroprudential action by national authorities.^[27] Indeed, the European Systemic Risk Board (ESRB) has recently issued a series of warnings and recommendations on vulnerabilities in the residential real estate sector.^[28] Moreover, macroprudential authorities in different jurisdictions have activated a series of measures in the real estate domain, regarding the maturity and loan-to-value ratio of loans and the debt service-to-income ratio of borrowers.

4 Impact on the broader economy

The overall impact of NIRP on loan growth derives from both the funding cost relief granted by the policy rate cut and the non-standard channels activated when rates move into negative territory. As is also the case with standard rate cuts, NIRP exerts an easing effect which operates through the bank lending channel of monetary policy transmission and stems primarily from a reduction in bank funding costs. This first-order effect, however, may differ in magnitude from that of a policy rate cut in positive territory since, as explained above, negative rates give rise to some specific frictions, while also activating additional channels, which can result in a multiplication of the stimulus provided by the monetary policy accommodation, especially as regards loan creation.

The compression in funding costs not related to retail deposits activated a bank lending channel that accounted for an expansion of aggregate corporate loan growth by around 0.4 percentage points. Funding options for banks include not only deposits of households and firms but also debt securities, Eurosystem funding, deposits of other financial institutions and governments, money market funds and external liabilities. Even under the conservative assumption that all deposits of the non-financial private sector, irrespective of their size, are bound at zero, the sum of all these funds still leaves around two-thirds of funding sources that can be remunerated at rates below zero. According to past regularities, this funding cost relief would have stimulated loan growth by around 0.4 percentage points on average, as envisaged under a standard bank lending channel of propagation to the broader economy.^[37] It is important to note that this is a conservative estimate, not only because deposit rates actually declined by 0.6 percentage points but also because it abstracts from the compression of risk and term premia.^[38]

A review of available empirical studies suggests that non-standard channels support aggregate loan supply, leading to an additional increase in annual corporate loan growth of around 0.3 percentage points (see Chart 7). Negative interest rates exert pressure on the intermediation margins of banks that rely more on retail deposits, leading healthier banks to expand their loan supply or to try and maintain profitability via search-for-yield.^[39] Conversely, weaker banks with limited capacity to expand credit supply or increase risk may keep lending rates unchanged or even be forced to increase them. This has been referred to in the economic literature as the “reversal rate”.^[40] Other studies focus on the weight of liquid assets in banks’ balance sheets or the self-reported impact of the policy on banks.^[41] For each study and each key characteristic, we compute the growth of bank loans to firms that would have emerged if the ECB had not adopted NIRP in 2014. We then order these counterfactuals depending on the resulting loan growth, and report their median (the dashed black line in Chart 7) and their overall range (grey area), spanning values from just above the actual loan growth (blue line) to well below it. In the absence of NIRP, loan growth for banks negatively affected by the policy (related to the income loss implied by the hard bound on retail deposit rates) would have been 0.3 percentage points higher than the observed loan growth. At the same time, loan growth would have been substantially lower for banks whose lending was positively affected by the policy (-0.5 percentage points each year being the median across studies, with estimates as low as -1.6 percentage points at the bottom of the range). Taking into account the whole set of available empirical analyses, the counterfactual loan growth in the absence of NIRP would have been lower by around 0.3 percentage points each year.

Overall, empirical evidence points to a positive impact of NIRP on loan growth (of around 0.7 percentage points each year), an assessment which is corroborated by survey evidence from the euro area BLS (see Chart 8). A consistently positive net percentage of euro area banks have been reporting that the negative DFR led to higher loan volumes (panel a). On average, 10% of banks in net percentage terms reported that the negative DFR led to an increase in loan volumes to enterprises, while the share was 15% for loans to households for house purchases.^[42] Consistent with the expansion in credit supply, the decrease in lending rates was even more widespread, with around 40% of banks reporting a compression of lending rates in net terms for both enterprises and households for house purchases.

Negative rates, in conjunction with the other policy measures, have contributed to the euro area expansion and supported inflation expectations. According to estimates carried out by Eurosystem staff, at the end of 2019 the level of real GDP was between 2.5 and 3.0 percentage points higher than it would have been in the absence of the unconventional measures that the ECB has taken since the middle of 2014. The policy contribution to the euro area inflation rate is estimated to have been on average between one-third and one-half of a percentage point per year until 2019. Identifying the contribution of each individual instrument is difficult because of the interaction among these policy tools (a feature that has often prompted the ECB to combine them in a policy package). However, one tentative way to isolate the contribution of NIRP is to run its estimated yield impact, as documented in Box 1, through macroeconomic models. These estimates suggest that around one-sixth of the overall cumulated impact on GDP growth is attributable to NIRP as a stand-alone instrument. This represents a tangible contribution, considering the overall limited rate adjustments in negative territory.^[43] NIRP is found to explain a similar share of the overall policy effect on inflation. As acknowledged above, these estimates provide a conservative assessment of the contribution of NIRP, which is reinforced by its complementarity with other monetary policy tools.

There are reasons to believe that the impact on the broader economy goes beyond the transmission via bank lending. Additional channels emerge or are strengthened as the policy rate moves deeper into negative territory. One example is the effect of NIRP on the remuneration of liquid assets in which the corporate sector usually invests. NIRP compresses the remuneration of liquid assets, and is passed on to corporate deposits as policy rates move deeper into negative territory and expectations of a future rate hike move further into the future. Using information from more than 300 banks and 3 million firms operating in 19 euro area countries, Chart 9 shows different patterns in the non-financial investment of firms exposed to negative deposit rates, distinguishing them according to their holdings of liquid assets.^[44] Firms with large holdings of liquid assets subject to negative deposit rates have an incentive to reduce these liquid assets by increasing non-financial investment (yellow line). By contrast, firms that have low liquid asset holdings and are therefore not particularly affected by negative deposit rates did not show such an acceleration (blue line). Formal econometric analysis of this corporate channel suggests that it is capable of adding 1 percentage point to aggregate business investment. In line with this empirical finding, a private sector survey of 500 German firms conducted in July-August 2019 found that, in the face of negative deposit rates, 32% of firms invested more in their own business.^[45]

5 Concluding remarks

The stimulus to the broader economy provided by NIRP has been effective in inducing an easing of financing conditions and thereby, ultimately, contributing to price stability. The interest rate cut has been channelled through both standard and non-standard transmission mechanisms, generating a reduction in bank funding costs and spurring loan creation. Effects were heterogeneous across bank characteristics. However, the dispersion in some bank responses does not challenge the overall positive first-order effect of NIRP on financing conditions. Notwithstanding the potential headwinds to transmission discussed above, as yet there is no sign that the stimulus provided by the measure has been exhausted, as new channels emerge while others fade away. Ultimately, the macroeconomic response has been sizeable and has helped to bring inflation closer to the ECB’s aim.

Protracted periods of negative rates do, however, have the potential to hinder the transmission of monetary policy. Prolonged periods of negative rates are qualitatively different from brief, “experimental” periods. If negative policy rates had proved to be short-lived, they might have become a mere footnote in central banking history. As negative rates persist, however, banks react to them to avoid the negative effects on profitability, although their leeway to do so is eroded over time. In the current euro area monetary policy environment, the effects of a long period of negative rates require continuous and careful monitoring as we venture further into uncharted territory.

Overall, NIRP still largely benefits the macroeconomic outlook and price stability. In order to support the bank-based transmission of monetary policy, the ECB has adopted a two-tier system for reserve remuneration. In parallel, specific risks to financial stability are addressed by action from other policy areas, which are specifically mandated to tackle phenomena and behaviours associated with such risks. Microprudential supervision monitors banks’ risk-taking behaviour, and so far it has provided an adequate set of incentives for intermediaries to calibrate their risk attitude to the macroeconomic circumstances. Moreover, national and supranational macroprudential authorities can effectively monitor and respond to localised house price bubbles and other threats to financial stability. Such mitigating action allows the euro area economy to continue to benefit from the significant and necessary stimulative impact of NIRP, which has proven to be an integral and effective part of the ECB’s policy response to past and current challenges.