I. Introduction

Since the 1990s interest rates have been declining and remain low across all major advanced economies. In particular, low long-term interest rates reflect the diminishing return on safe assets due to demographic changes, a slowdown in the rate of technological progress, and a high demand for safe assets relative to their supply (Bean and others, 2015). Given the secular decline of inflation and inflation expectations and short-term policy rates approaching zero percent, this has made it more difficult for accommodative monetary policies to reduce real interest rates to a level consistent with stable inflation and output at its potential level (technically, the “natural” rate of interest). For a long time, central bankers believed that the policy rate could not drop below zero, because households and corporates might start converting deposits into cash to avoid devaluation–thus, conflating the nominal lower bound with the “physical lower bound”. However, not going below zero percent meant that with inflation remaining low, real rates could not fall further to help reduce high debt burdens and support aggregate demand.

View Full Size

Major Reserve Currencies: Effective Marginal Policy Rates, 2010-2016

(Percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg, L. P.; and Haver Analytics.Note: Policy rates used for each country are the following: EA: deposit rate; US: Fed funds rate; UK:O/N interbank rate; Japan: deposit rate.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Other Currencies: Effective Marginal Policy Rates, 2010-2016

(Percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg, L. P.; and Haver Analytics.Note: Policy rates used for each countryare the following: SWE: reverse repo rate; DEN: certificates of deposit rate; HUN: deposit rate; BGR: deposit rate; CHE: deposit rate.

• Download Figure
• Download figure as PowerPoint slide

More than two years ago, following the example of the Danmarks Nationalbank (DN), the European Central Bank (ECB) became the first major central bank to effectively move its marginal policy rate into negative territory in response to these macroeconomic challenges. The move in June 2014 strengthened its forward guidance about the expected future path of interest rates and complemented a series of further easing measures aimed at bringing inflation back to the ECB’s price stability objective of below, but close to, two percent over the medium term. Also other central banks–the Sveriges Riksbank (SR) and the Swiss National Bank (SNB)–cut their marginal policy rates to below zero percent over the period from mid-2014 to early 2015 (see text figures).² Central banks in Norway (September 2015), Japan³ (February 2016), and Hungary (March 2016) lowered only their deposit rate for excess bank reserves while keeping the main policy rate in positive territory.⁴ While some central banks have adopted a negative interest rate policy (NIRP) to counter low inflation (ECB, BoJ, SR),⁵ others have focused on mitigating spillover effects from unconventional monetary policy (UMP) measures (Mircheva and others, 2016) and to address currency appreciation pressures (DN, SNB) (Table 1 and Appendix I, Table A1). Most central banks have also introduced a tiered deposit rate to reduce banks’ cost of holding excess reserves while still allowing for a strong pass-through to money markets (Appendix II).

Table 1.

Overview of Central Banks with Negative Policy Rates

Source: National central banks and authors. Note:

^1/effective policy rate are highlighted with a red background, as of end-July 2016;

^2/refers to special rate (liquidity-shortage financing facility) in the case of Switzerland;

^3/Norway has not adopted NIRP, and the negative interest rate on bank deposits at the central bank (“reserve rate”) has had little or no influence on market rates. The reserve rate is one percentage point below the sight deposit rate (key policy rate). On average, NB has kept reserves in the banking system at around NOK 35 billion (and below the aggregate quota of NOK 45 billion). Thus, a bank with reserves in excess of the quota will always be able to deposit reserves with a bank with room on its quota.;

^4/conventional peg (to euro) before January 15, 2015);

^5/in conjunction with the exit from the exchange rate ceiling.

Table 1.

Overview of Central Banks with Negative Policy Rates

			Policy Rates (in basis points) 1/
	FX regime	Objective	Overnight Lending 2/	Open Market Operations	Deposit Facility	Date of Introduction
Denmark	Conventional peg (to euro)	Countering safe-haven inflows and exchange rate pressures	5	0	−65	July 2012-April 2014, Sept. 2014
Euro Area	Free floating, inflation-targeting framework	Price stability and anchoring inflation expectations	25	0	−40	June 11, 2014
Hungary	Floating, inflation-targeting framework	Price stability and countering exchange rate pressures	115	90	−5	March 23, 2014
Japan	Free floating, inflation-targeting framework	Price stability and anchoring inflation expectations	10	0	−10	Feb. 16, 2016
Norway	Free floating, inflation-targeting framework	Price stability 3/	150	50	−50	Sept. 24, 2015
Sweden	Free floating, inflation-targeting framework	Price stability and anchoring inflation expectations	25	−50	−125	Feb. 12, 2015
Switzerland	Free floating 4/	Reducing appreciation and deflationary pressures 5/	50	n.a.	−75	Jan. 15, 2015

Source: National central banks and authors. Note:

^1/effective policy rate are highlighted with a red background, as of end-July 2016;

^2/refers to special rate (liquidity-shortage financing facility) in the case of Switzerland;

^3/Norway has not adopted NIRP, and the negative interest rate on bank deposits at the central bank (“reserve rate”) has had little or no influence on market rates. The reserve rate is one percentage point below the sight deposit rate (key policy rate). On average, NB has kept reserves in the banking system at around NOK 35 billion (and below the aggregate quota of NOK 45 billion). Thus, a bank with reserves in excess of the quota will always be able to deposit reserves with a bank with room on its quota.;

^4/conventional peg (to euro) before January 15, 2015);

^5/in conjunction with the exit from the exchange rate ceiling.

View Table

Table 1.

Overview of Central Banks with Negative Policy Rates

			Policy Rates (in basis points) 1/
	FX regime	Objective	Overnight Lending 2/	Open Market Operations	Deposit Facility	Date of Introduction
Denmark	Conventional peg (to euro)	Countering safe-haven inflows and exchange rate pressures	5	0	−65	July 2012-April 2014, Sept. 2014
Euro Area	Free floating, inflation-targeting framework	Price stability and anchoring inflation expectations	25	0	−40	June 11, 2014
Hungary	Floating, inflation-targeting framework	Price stability and countering exchange rate pressures	115	90	−5	March 23, 2014
Japan	Free floating, inflation-targeting framework	Price stability and anchoring inflation expectations	10	0	−10	Feb. 16, 2016
Norway	Free floating, inflation-targeting framework	Price stability 3/	150	50	−50	Sept. 24, 2015
Sweden	Free floating, inflation-targeting framework	Price stability and anchoring inflation expectations	25	−50	−125	Feb. 12, 2015
Switzerland	Free floating 4/	Reducing appreciation and deflationary pressures 5/	50	n.a.	−75	Jan. 15, 2015

Source: National central banks and authors. Note:

^1/effective policy rate are highlighted with a red background, as of end-July 2016;

^2/refers to special rate (liquidity-shortage financing facility) in the case of Switzerland;

^3/Norway has not adopted NIRP, and the negative interest rate on bank deposits at the central bank (“reserve rate”) has had little or no influence on market rates. The reserve rate is one percentage point below the sight deposit rate (key policy rate). On average, NB has kept reserves in the banking system at around NOK 35 billion (and below the aggregate quota of NOK 45 billion). Thus, a bank with reserves in excess of the quota will always be able to deposit reserves with a bank with room on its quota.;

^4/conventional peg (to euro) before January 15, 2015);

^5/in conjunction with the exit from the exchange rate ceiling.

II. Arguments Surrounding NIRP

In an environment of low inflation and a declining equilibrium real rate of interest, negative rates restore the signaling capacity of the central bank by effectively removing the zero lower bound (ZLB). Moving the marginal policy rate into negative territory can help the real rate adjust downward, compensating for inflation below the inflation target (text figure) and contributing to a significant flattening of the yield curve.⁶ In other words, a decline in the nominal rate could lower its real rate component, allowing inflation expectations to rise and boosting aggregate demand; however, if both nominal and real interest rates are shifted down, a widening gap leads to deflation pressure.⁷ Thus, if banks hold excess reserves, cuts to the central bank deposit rate (as the marginal policy rate) can effectively lower the interbank and other interest rates, encouraging banks to take greater risks and facilitating portfolio rebalancing.

View Full Size

Inflation and Interest Rates, Jan. 2002-April 2016

(In percent, monthly)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg, Haver Analytics, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Assessing the effect of NIRP on exchange rates is difficult since many other factors influence external demand. Negative rates cause exchange rates to depreciate by providing incentives for moving capital to higher-yield jurisdictions. Thus, a widening real term spread differential would put downward pressure on the currency. Higher inflation and inflation expectations in other countries might counter this effect. In addition, the stimulative effect of negative rates on aggregate demand (which is discussed below) and rising asset prices in real terms might offset depreciation pressures on the exchange rate.

The economic lower bound of NIRP is largely determined by the impact of negative rates on financial intermediation. While there is some direct cost pressure from charging interest for excess liquidity that banks hold at the central bank (suggesting limits to negative rates based on the tendency of banks to raise cash balances), some central banks have implemented mitigating policy to limit the incentive to move into cash—such as tiered reserve regimes and penalties for banks making large transfers of reserves into cash. As rates become more negative, the lower nominal bound is increasingly determined by the broader indirect pressure coming from diminishing bank profitability as most lending rates are assumed to fall more than deposit rates.

The downward stickiness of deposit rates could result in a difficult trade-off between effective monetary transmission and bank profitability. If negative policy rates are transmitted to lower lending rates (and term premia),⁸ banks are likely to see their interest earnings decline unless they either impose negative rates (or commensurate fees) on deposits or substitute more wholesale funding (at lower money market rates) for deposits.⁹ But retail deposit rates tend to be downward sticky since (i) households and small businesses do not face the same set-up cost as banks and corporations in storing cash,¹⁰ and (ii) a zero percent interest rate could be a psychological threshold (Alsterlind, 2015).¹¹ The stickiness of deposit rates reflects the avoidance of being penalized to save and is determined by the actual costs of holding cash rather than deposits; under these conditions, demand for cash is likely to be greatest for economic agents with high excess liquidity and increases if negative interest rates are expected to persist for some time.¹² As a result, banks’ net interest margins (NIMs), defined as net interest income relative to average interest-earning assets, compress as lending rates for new loans decline and existing (variable-rate) loans re-price while deposit rates remain sticky. This could reduce bank profitability and impair the pass-through to lending rates.¹³ Thus, banks might also consider increasing lending rates if they face (i) a considerable opportunity cost of accepting deposits at non-negative rates as wholesale funding costs decline in lock-step with the marginal policy rate and/or (ii) corporate deposit rates cannot be cut to potentially subsidize sticky retail deposit rates.

A prolonged period of negative rates could also raise financial stability concerns. In particular, the downward stickiness of deposit rates encourages banks to substitute less stable wholesale funding for deposits. For example, German, Italian, Portuguese, and Spanish banks, whose deposit base is wider than the euro area average, would have stronger incentives to trade off market-based sources of funding against more stable (term) deposit funding (Figure 3). This calls for greater emphasis on the appropriate supervision of bank liquidity, including the transition to the liquidity coverage ratio (LCR) and net stable funding ratio (NSFR) requirements.

As much as negative rates ease financial constraints on borrowers in the short run, they could distort the long-term debt affordability for borrowers if lending rates become negative in real terms.¹⁴ The reduced debt service burden under NIRP could delay the exit of nonviable firms, hurting demand prospects of healthy firms by adding to excess capacity and delaying the efficient allocation of capital and labor (Caballero and others, 2008; Kwon and others, 2015).¹⁵ By effectively removing the profitability constraint of investments as real borrowing rates drop to or even fall below the ZLB, NIRP might also delay corporate restructuring in countries with debt overhangs, especially if inflation does not pick up. In these instances, more assertive supervision and regulatory pressures would be needed to address large amounts of non-performing loans and debt overhang problems (Syed and others, 2009).

Box 1.

Monetary Transmission under NIRP¹

We assess the impact of negative rates on bank profitability and its implications for monetary transmission when deposit rates become sticky using a general equilibrium specification. We adapt the DSGE model by Gerali and others (2010), which was estimated using euro area data. In the model, banks enjoy monopoly powers in intermediating funds between savers and borrowers and setting rates on loans and deposits. The modeled banking sector comprises two retail branches, which are responsible for lending and deposit-taking, while the wholesale unit manages the capital position of the banking group subject to a simple solvency constraint, and, in addition, provides wholesale loans and raises wholesale funding. Banks face different adjustment costs when changing rates. A higher cost implies lower adjustment for a given shock, and, thus, the rates are more “sticky.”

View Full Size

Source: authors. Note: blue line=unchanged pass-through scenario (first scenario, base case); green dotted line=sticky deposit scenario but non-binding solvency constraint for banks (second scenario); red line= sticky deposit scenario and binding solvency constraint for banks (third scenario).

• Download Figure
• Download figure as PowerPoint slide

Sticky deposits under NIRP seem to either weaken bank profitability or diminish monetary transmission. We examine three different scenarios reflecting banks’ response to a policy rate cut assuming that deposit rates are bounded at zero percent (text chart below). Banks can substitute some cheaper wholesale funding for deposit funding but potentially offsetting components of banks’ net operating income are ignored (e.g., capital gains from higher asset prices and lower provisioning cost from higher debt service capacity of borrowers). While higher asset prices boost the investment income, lower funding costs, and decrease provisioning expenses of euro area banks, these benefits weaken over time relative to the adverse effect of compressed interest margins where the pass-through of policy rates is high and credit demand is low.

In the first case (blue line), we assume that the pass-through from the policy rate to deposit rate remains unchanged. Banks reduce the both deposit and lending rates, and their profitability increases over time as output and inflation outturns improve. In the second case (green dotted line), price-setting banks face (artificially) higher adjustment costs in setting deposit rates (i.e., deposits are “sticky”). Banks optimally choose to lower lending rates to increase lending volume at the cost of deviating temporarily from the minimum capital requirement (Angelini and others, 2014). Bank profitability declines significantly as lending volumes are initially insufficient to offset the compression of lending margins due to sticky deposit rates. In the third case (red line), banks’ solvency constraint is strictly enforced for the second scenario of sticky deposits. Here, monetary transmission breaks down as banks (initially) increase lending rates and curtail credit growth given the limited substitution of wholesale funding (due to a large deposit base). However, the impact on output is still positive, although smaller over the short term, as the wealth and substitution effects (from lower discount rates) pushes up loan demand, supporting consumption and investment. In general, the simulation results suggest a positive aggregate impact of NIRP under all three scenarios but rising pressures on bank profitability over the short and medium terms if deposit rates have reached a lower bound and banks are capital-constrained.

¹ Prepared by Jiaqian (Jack) Chen and Andreas (Andy) Jobst.

Negative rates also have distributional implications that are beyond the scope of this paper. For instance, negative rates could increase the re-distributional impact of monetary policy on wealth and income. Any reduction in interest rates makes savers worse off while borrowers benefit and could have important intergenerational implications. Elderly people tend to have accumulated savings, so moving from positive to negative interest rates could increase intergenerational inequality as retirement income declines. However, higher asset prices increase the net worth in present value terms. Lower borrowing rates also support consumption and investment of liquidity constrained households and firms, raising aggregate demand over time and outweighing any adverse impact on savings.

However, several important factors could compensate for the adverse impact of NIRP:

• Stronger credit growth and/or higher non-interest income. The credit supply effects of reduced profitability from lower lending rates can be offset by the credit demand effects if banks increase lending (Box 1)—but this becomes more difficult if credit demand is low, assets re-price quickly, and competition among banks is high. Banks could also supplement declining interest margins with alternative sources of income, such as fees and commissions.¹⁶

• Higher asset prices and lower funding costs. Portfolio rebalancing with negative rates reduces term and credit risk premia, eases financial conditions and ultimately supports credit creation and economic activity. The resulting decline in risk aversion increases asset prices and generates capital gains for banks. Furthermore, higher asset prices (especially in tandem with higher inflation) are likely to raise future income and strengthen borrowers’ repayment capacity, lowering banks’ expected provisioning costs and write-off charges for non-performing loans (NPLs).¹⁷

• Stronger aggregate demand through portfolio rebalancing. Negative rates also increase household consumption and steer portfolio rebalancing towards other investment opportunities, with beneficial effects on aggregate demand.¹⁸ Portfolio rebalancing helps lower firms’ general cost of capital via lower term premia on corporate bond yields.¹⁹ At a lower cost, more investment projects would become profitable, raising investment and credit demand. Higher asset prices and lower interest expenses for indebted households (who tend to have higher marginal propensity to consume) also boost household consumption through wealth effects.²⁰

III. The Impact of Negative Interest Rates

Overall financial conditions have improved as central banks in many advanced economies set policy rates at record low levels, without causing significant swings in exchange rates. Negative rates have been fairly effective thus far in reducing money market rates and have been transmitted to the wider economy through lower lending rates for both corporates and households as lending standards have eased (Elliott and others, 2016; Viñals and others, 2016). At the same time, retail and corporate deposit rates also have declined, allowing most banks to maintain their lending margins and supporting credit growth.²¹ In cases where sticky deposits have compressed lending margins, banks have shifted some of their activities to fee-based services and/or increased their lending volumes to offset declining interest revenues. Negative rates have so far had a muted impact on exchange rates as current disinflationary dynamics in many countries with negative rates prevent real rates from declining further (see text figures).

View Full Size

Exchange Rates relative to Euro

(Index, time t = 100, day before introduction of negative deposit rates)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Haver Analytics. Note: the x-axis shows monthly intervals.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Exchange Rates relative to U.S. Dollar

(Index, time t=100, day before introduction of negative deposit rates)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Haver Analytics. Note: the x-axis shows monthly intervals.

• Download Figure
• Download figure as PowerPoint slide

Money market rates have closely followed an increasingly negative marginal policy rate without disruptions to market functioning (Appendix III, Figures A1 and A2). In the environment of excess liquidity, the observed money market rate will be at or just above the marginal policy rate at which excess reserves are remunerated (or penalized under NIRP).²² In several countries, a tiered central bank deposit rate has facilitated the smooth transmission of the marginal policy rate to money markets, reducing the cost of interbank lending and restoring the signaling capacity of the central bank and strengthened its commitment to keep rates low for an extended period of time (Box 2). However, several factors, in particular related to the design of a tiered reserve system, could keep the money market rate away from the deposit rate as the technical floor of the policy rate corridor (Appendix I, Box A1). These include: (i) the amount of excess liquidity and the fraction that is exempted from the marginal policy rate, (ii) the spread between the marginal and average policy rate for excess reserves, and (iii) banks’ willingness/ability to lend excess liquidity to each other (fragmentation).

Box 2.

Reducing the Direct Cost of NIRP and the Role of Tiering in Monetary Transmission

The implementation of a second effective deposit rate for excess reserves (such as through tiering) would increase the central bank’s general capacity to pursue NIRP while mitigating the direct cost to banks. More specifically, excess reserves can be held in both the current account and the deposit facility of central banks. In a tiered regime, the exemption typically applies to the current account (which satisfies the minimum reserve requirement); thus, banks would have an incentive to shift excess reserves from the deposit facility to the current account to reduce the cost of negative rates up to a certain limit. Currently, euro area banks’ overnight deposits (€297 billion) and current account balances (€613 billion) amount to about €910 billion. The minimum reserve requirement of €116 billion is remunerated at the MRO rate of 0 percent, which leaves excess reserves of €794 billion subject to the negative deposit rate of -0.4 percent as the marginal policy rate—setting the lowest rate at which banks would be prepared to lend to each other (which is reflected in the theoretical EONIA rate in the chart below). Thus, the direct (annual) cost of negative rates (without implementation of a tiered reserve regime) is about €3.2 billion for all euro area banks in aggregate (and 0.1 percent of total consolidated assets). In other words, a 10 basis point reduction in the deposit rate results in a direct cost of about €0.8 billion per year. Given the recent experience in Switzerland, where the direct cost of NIRP did not exceed 0.03 percent of total assets (Barr and others, 2016),¹ the current deposit rate may have already reached its limit (without considering any mitigating effects). However, for a tiered reserve regime excluding 75 percent² of excess reserves from the negative deposit rate (in line with reserve system in Switzerland), the Eurosystem could theoretically tolerate further cuts to the deposit rate based on the direct cost of NIRP only.

View Full Size

Euro Area: Actual and Theoretical EONIA Rate

(percent/EUR billion)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg L.P., ECB, and IMF staff calculations. Note: 1/ The theoretical EONIA rate is calculated as the average of the MRO and deposit rate, weighted by the relative proportion ofthe ECB’s current account balance and excess reserves.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Euro Area: Secured and Unsecured Lending Volumes

(EUR billion)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg L.P., ECB, and IMF staff calculations. Note: 1/ trading volume of overnight contracts 2/ composite outstanding volume of general collateral (GC)repo on German, French, and Italian government debt

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Euro Area: EONIA and Repo Rates

(percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg L.P., ECB, and IMF staff calculations. Note: 1/ Composite of German, French and Italian GC repo.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Euro Area: EONIA and Certificate of Deposit Rate

(percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg L.P., ECB, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

The direct cost of a negative deposit rate on excess reserves is relatively small, even without tiering, but affects banks disproportionately due to significant differences in excess reserve holdings. Since the ECB charges interest only on excess liquidity, the charge is greater in those countries where banks hold large excess reserves. These are generally countries with substantial current account surpluses vis-à-vis other members of the monetary union. Banks in other economies hold much lower excess reserves and thus are much less affected.

In comparison, the indirect effect of NIRP via (potentially) lower bank profitability from lending can be large. Given a total outstanding amount of loans of about €17.6 trillion (end-May 2016) at an average interest spread of 0.8 percent, euro area banks would record aggregate net interest income of about €141 billion (gross of operating expenses, provisioning, and taxes). Based on a (historically conservative) pass-through of 50 percent, a 10 basis point rate cut would reduce lending margins by 5 basis points and result in an indirect cost of about €8.8 billion (or 11 times the estimated direct cost of NIRP above). Valuation gains from investments and trading income might, to some extent, mitigate these costs. However, the aggregate balance sheet of euro area banks suggests that lending is about 6-7 times more important than investments for profitability.

Fine-tuning the exempted portion of excess reserves can alter the effective monetary transmission of negative rates to money markets. In its current reserve regime, the ECB achieves negative short-term money market rates by setting a positive policy rate (MRO at 0 percent) and a negative interest rate on the deposit facility (-0.4 percent) while maintaining excess reserves in the banking system. The money market rate is pushed down towards the lowest marginal policy rate because banks will try to lend their surplus liquidity to other banks in the interbank market to avoid using the central bank’s deposit facility—but only as long as the lending rate exceeds the deposit rate.³ Given the prevailing excess liquidity in the system, the overnight money market rate (EONIA) has converged to the volume-weighted average of (i) the deposit rate (at -0.4 percent) for excess reserves and (ii) the marginal refinancing operations (MRO) rate (of 0 percent), which applies to banks’ minimum reserve requirement held in the ECB’s current account (see text chart above). Thus, the transmission of the marginal policy rate is also affected by the dispersion of the excess liquidity among banks and banks’ willingness/ability to lend excess liquidity to other banks.⁴ This also holds true for secured funding markets (repo), where rates have similarly adjusted downward as the ECB lowered the deposit rate (see text chart). The experience in countries with tiered deposit rates shows that the average rate has a large impact on the fixing; thus, a volume-weighted average of potentially two deposit rates—by excluding a certain share of reserves from the deposit rate and removing some excess liquidity from interbank market—below the marginal policy rate could marginally push up the money market rate.

The effective transmission of the policy rate can also be assessed based on the sensitivity of unsecured term funding of banks via certificates of deposits (CDs). The CD rate reflects the willingness of money market funds to lend to issuing banks over a pre-defined maturity term (see text chart). Since the CD rate continues to track the marginal policy rate very closely even under NIRP (without a change in volumes), this suggests that banks are able to impose lower and negative rates on investors despite counterparty risk given the unsecured nature of CDs.

¹ The assumption of exempting 75 percent of reserves from a negative deposit rate was based on the experience in Switzerland where the share of the overall reserve stock subject to negative deposit rates averaged 23 percent until end-2015. In practice, given the significant heterogeneity of bank business models, banks’ tolerance threshold for the direct cost of negative rates might be different in the euro area than in Switzerland. ² The exemption of a certain amount of reserves can vary over time (and would need to decrease as excess liquidity declines). The opportunity cost of lending can be increased (on average) by calibrating the tiering such that the price of depositing cash with the ECB would be the same (or higher) than the expected net interest margin from lending multiplied by the share of the deposit base funding loans (i.e., the inverse of the aggregate loan-to-deposit ratio of the banking sector). ³ The money market rate could be higher than the lowest marginal policy rate if the exempted portion of excess reserves is too large, leaving banks little incentive to engage in interbank lending; thus, lower supply of liquidity could create potential scarcity in some parts of the system, pushing up money market rates above the technical floor of the ECB deposit rate. ⁴ Given that the average daily quoted turnover underpinning EONIA fixings has only been about €12.6 billion (or 1.5 percent of excess liquidity) since January 2016, the impact of the marginal policy rate on money market rates is quite sensitive to changes in bank behavior and rate setting.

View Full Size

Deposit and Lending Rates (New Lending) for Households

(January 2005-June 2016, percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg L.P., Haver and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Negative policy rates also have been transmitted to the wider economy through lower lending rates for both corporates and households as lending standards have eased. In most countries, banks lowered their lending rates to both households and firms (which continued to decline even as deposit rates had reached the ZLB) while offsetting the negative impact on lending margins (when deposit rates adjusted less) by some small increase in fees and commissions and cost cutting. However, this has occurred, in most cases, as long as deposit rates still had some room to drop to the ZLB (see text figures and Figure 1), allowing banks to transmit lower policy rates without impeding their profitability (Box 1).²³ In some countries (e.g., Denmark and Sweden), banks also passed negative rates to deposits of some large corporations and institutional investors but maintained positive rates for retail depositors. In Switzerland, lending rates adjusted only slowly when policy rates moved into negative territory, helping to raise bank profitability (Appendix III, Figure A3).

View Full Size

Marginal Policy Rate (Central Bank Deposit Rate) and Bank Net Interest Margin

(January 2010–May 2016, percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg, L.P.; Haver Analytics.

• Download Figure
• Download figure as PowerPoint slide

The direct cost imposed on excess bank reserves is modest relative to the size of the overall balance sheet. Negative rates have important implications for banks’ cost of holding central bank liabilities depending on the structure of reserves and their remuneration (Appendix I, Box A1) and the transmission of the marginal policy rate to money markets. However, bank profitability is far less sensitive to negative rates on excess reserves (even under a tiered system) since cash balances of banks represent only a fraction of their asset base (although there are significant differences in the distribution of reserves around the euro area). That said, many banks earn negative returns on many liquid assets, which are required for liquidity risk management and represent a significant portion of their balance sheet. At the same time, rolling over these short-term holdings tends to offset any benefits from valuation gains.

View Full Size

Currency in Circulation at Time of Negative Deposit Rates

(Percent, year-on-year)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Haver Analytics. Note: the x-axis shows monthly intervals.

• Download Figure
• Download figure as PowerPoint slide

In the euro area and countries with an even more negative deposit rate (Denmark, Sweden and Switzerland), there have been no clear signs of cash hoarding (see text figure).²⁴ Current levels of negative rates seem to provide insufficient incentives to build alternative storage capacity for excess reserves. However, banks have been hesitant to pass on negative rates to depositors—with the exception of large corporates (e.g., Denmark and Switzerland)—limiting the incentives for hoarding cash. In fact, most of the recent increase in cash holding in some countries (Denmark and Switzerland) can be explained by the normal relation between currency in circulation and movements in the short-term interest rate, reflecting the reduced opportunity cost of holding cash rather than deposits (Figure 3). Irrespective of whether interest rates are positive or negative, the amount of currency in circulation increases when interest rates decline.²⁵

Despite lower lending rates, there is limited evidence so far of negative rates having directly damaged bank profitability; however, they have contributed to the flattening of the yield curve.

While there is some direct cost pressure coming from the charge on excess reserves, it is dominated by the indirect cost of negative rates coming from the possible decline of net interest income in cases where sticky deposits compress lending margins. However, bank profitability has not worsened because positive effects have so far outweighed these adverse effects. Many euro area banks have been able to more than offset declining interest revenues with higher lending volumes, lower interest expenses, lower risk provisioning and capital gains (Cœuré, 2016a).²⁶ While bank profitability has been a longstanding structural challenge for many euro area countries (Albertazzi and Gambacorta, 2009), the aggregate NIM has even improved slightly towards the end of 2015, after almost 1.5 years of negative rates.²⁷ Moreover, reduced lending margins have also put pressure on banks to consolidate and strengthen operational efficiency.²⁸ Some banks—such as those in Sweden, Denmark and Switzerland²⁹—have been able to maintain overall profitability and benefitted from historically low impairment charges and greater wholesale funding at negative rates (Appendix I, Box A2 and Figure 2).

View Full Size

Currency in Circulation and Pervasiveness of Negative Interest Rates

(2012-2016, percent)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg L.P. and IMF staff calculations. Note: 1/ denotes the extent to which the 1-month interbank offer rate forward curve is in negative territory, calculated as the product of the maturity term and the interest rate.

• Download Figure
• Download figure as PowerPoint slide

IV. Assessment for the Euro Area

So far, negative interest rates have contributed to an improvement in overall financial conditions and a modest expansion of credit. The negative deposit facility rate has been effectively transmitted through the domestic credit channel. According to the ECB’s Bank Lending Survey (ECB, 2016a) negative rates seem to have led to an increase in household lending in the euro area (Figure 4 and Appendix III, Figure A4), and the impact is expected to continue going forward.³⁰ With money market rates tracking the deposit rate in an environment of excess liquidity, the negative rate has also enhanced the ECB’s forward guidance while inflation and inflation expectations have remained subdued. Negative rates have also strengthened portfolio rebalancing (Heider and others, 2016)—an important transmission channel of the ECB’s asset purchase program.

The impact of NIRP on bank profitability so far has been limited but monetary transmission might become less effective as interest rates become more negative. Given that the volume of outstanding loans in the euro area matches the amount of deposits, the pass-through of policy rates to lending and deposit rates matters greatly for the earnings capacity of most banks. Estimates of the impact of the recent decline in policy rates on banks’ NIMs suggest a small effect (7 basis points for a 50-basis point reduction in the policy rate).³¹ Since the adoption of NIRP, however, monetary transmission has become more heterogeneous across the euro area, especially for household lending, with a higher pass-through in countries with a higher share of variable rate loans. While the extent to which deposit rates are sticky at the ZLB remains to be seen, lending rates would likely decline more than deposit rates in the near term if cash demand is highly elastic (see text figure and Rognlie, 2015). This would leave little room for further substantial adjustments to the deposit rate without compromising banks’ net interest earnings. On the other hand, banks might also be reluctant to reduce lending rates unless they can offset lower interest margins by substituting wholesale funding for more expensive deposit funding, which represents a large part of euro area bank liabilities. If lending rates become stickier, monetary transmission could become impaired, reducing the effectiveness of negative rates as a policy measure.

View Full Size

Euro Area: Estimated Change of the Lending Spread to a Reduction in Effective Policy Rate, 2006-2016

(multiple) 1/

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Sources: Bloomberg L.P., Haver, and IMF staff calculations. Note: estimates based on the cumulative response of the lending spread (lending rateminus deposit rate) to a one-percent reduction in the overnight money market rate (EONIA) using a VAR specification with a simple lag structure between July 2006 and May 2016; the lending rates is the volume-weighted average lending rate to both non-financial corporates and households.

• Download Figure
• Download figure as PowerPoint slide

Early evidence suggests that the adverse impact of negative rates on bank profitability may increase non-linearly as the policy rate declines further.³² Indeed, the ECB’s Bank Lending Survey (ECB, 2016a) shows that banks’ profitability has recently declined and is expected to remain depressed (Figure 4). Admittedly, higher aggregate demand and asset quality help raise investment income, lower funding costs and provision expenses, which has mitigated so far the adverse impact on bank profitability in the euro area and support the notion that the economic lower bound to NIRP might be much lower than the ZLB (Cœuré, 2016b). However, these benefits have weakened over time, especially in countries where the pass-through of policy rates is high and low credit demand limit the extent to which banks (can) increase lending to offset the impact of lower lending rates.

Deteriorating investor confidence and declining risk-taking behavior could intensify the potentially adverse impact of negative rates. The prospect of low policy rates for a longer time—amplified by structural challenges to banks especially in countries where the cost of risk remains high due to a (still) large stock of impaired assets—has already worsened the outlook for bank earnings.³³ Further substantial reductions to the deposit rate could further weigh on banks’ equity prices as investors will be likely to revise down their expectations of banks’ future earnings (Figure 5).³⁴ While monetary easing has reduced borrowing costs,³⁵ equity risk premia have risen, and price-to-book ratios have declined during the second half of 2015, with the average cost of equity now exceeding the return on equity.³⁶ As sustainable profitability becomes more difficult to achieve, capital-constrained banks become more likely to reduce lending despite declining rates.

For the euro area, three important adverse implications need to be considered in the way negative rates could diminish the transmission of monetary policy to the real economy:

• NIRP may become less effective in economies most in need of stimulus. Given the wide deposit base in most euro area countries, the extent to which deposit rates are sticky has a direct impact on bank profitability and the effectiveness of NIRP on monetary transmission. Even if banks were to fund themselves increasingly via money markets, the benefit from wholesale funding at negative rates will be limited by the existing deposit base and cannot offset the negative impact of lower rates on existing loans if credit growth is insufficient (Figure 3). Hence, lower profitability from financial intermediation might override possible mitigating benefits from higher asset prices and pricing frictions due to an insufficient reduction in funding costs. In particular, bank profits are likely to decline in countries with large outstanding loan amounts at variable rates if lending growth cannot offset diminishing interest margins as existing loans re-price. In this regard, the ECB’s TLTRO II program could facilitate the transmission to lending rates by mitigating the potentially adverse impact of negative rates on banks’ lending margins (ECB, 2016b). Higher credit demand can offset declining margins, and, in turn, reinforce the impact of TLTRO II on bank profitability (Appendix I, Box A3). Among countries with a high share of variable rate loans, such as Italy, Portugal, and Spain, also high asset impairments amplify concerns about banks’ earnings capacity, and restrict their ability to supply credit to the real economy. For instance, in the case of Italy, NIMs have declined, and credit growth is far below the NIM-preserving threshold (in contrast to high credit growth in Germany and France).

• The direct cost of negative deposit rates is likely to be greater for banks in surplus countries. Since the ECB charges interest only on excess liquidity the charge is greater in those countries where banks hold large excess reserves. Given the imbalances within the euro area, the Target 2 settlement of capital flows generates large amounts of excess liquidity in the banking sectors of those countries with substantial current account surpluses (such as Germany and the Netherlands) vis-à-vis other members of the currency area.³⁷ In addition, the implementation of the Eurosystem’s asset purchase program has generated additional liquidity in other core economies in excess of their national share of asset purchases, such as France (see text figure). Both developments have led to a very uneven distribution of excess liquidity, affecting banks differently across the euro area.³⁸ In principle, tiering of the deposit rate could mitigate the direct cost of NIRP and ensure effective transmission of the marginal policy rate (to short-term rates) even if rates became more negative and excess liquidity increases. However, the heterogeneity of national banking systems within the euro area might complicate the effective implementation of a tiered reserve regime (Appendix I, Box A1).

• Negative interest rates might not necessarily increase the scope of eligible assets for purchases under the current program. Since purchases are limited to eligible government and public agency debt securities trading at yields that are above the ECB’s deposit facility rate, moving interest rates into deeper negative territory would make this price-based exclusion less restrictive as more securities trading at negative yields would become eligible for purchase. However, lowering the deposit rate will also cause yields to fall—albeit at different degrees across the euro area depending on the yield curve flattening—leaving some proportion of the securities trading above the lower (new) deposit rate unchanged.

View Full Size

Euro Area: Rising Financial Fragmentation, January 2015-March 2016

(absolute change, EUR billion)

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Sources: Bloomberg L.P. , ECB, Haver, NCBs, and IMF staff calculations. Note: */The use of ECB liquidity reduces the Target 2 balance and is subtracted; 1/ public sector purchase program (PSPP).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Impact of NIRP on Bank Profitability and Implications for Credit Growth

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Bank Equity Valuation and Credit Growth

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

• Download Figure
• Download figure as PowerPoint slide

V. Conclusion

Negative interest rates so far have had a positive effect on the economy, helping to lower bank funding costs and boost asset prices. In addition, negative rates have significantly enhanced the signaling effect of the ECB’s monetary stance strengthening its forward guidance. Lowering the deposit rate has also supported the portfolio rebalancing channel of the ECB’s asset purchase program by encouraging banks to substitute investment in riskier assets for excess reserves.

With money market rates tracking the deposit rate in an environment of excess liquidity, this has enhanced the ECB’s signaling capacity and strengthened its commitment to keep rates low for an extended period of time until the price stability objective is achieved. In some countries, rate cuts have been passed through to corporate and household borrowers thereby contributing to a modest credit expansion and bolstering the economic recovery. Lower lending rates have encouraged higher credit demand as lending standards continue to ease. Concerns about their negative effect on bank profitability have for the most part not yet materialized.

View Full Size

Equivalency of Government Bond Yield and Marginal Policy Rate

(Years) ¹/

Citation: IMF Working Papers 2016, 172; 10.5089/9781475524475.001.A001

Source: Bloomberg LP and IMF staff calculations. Note: Euro area covers the core economies only;¹/ the “equivalency line” shows the maturity term at which the prevailing government debt yield is no higher than the marginal policy rate in the respective jurisdiction.

• Download Figure
• Download figure as PowerPoint slide

However, further substantial reductions of the deposit rate will likely entail diminishing returns, since the lending channel is crucially influenced by banks’ expected profitability. While most banks have been able to mitigate the squeeze on profitability with higher lending volumes and benefitted from higher asset prices, lower funding costs, and possible cost savings from greater operational efficiency and consolidation, there are clearly limits to such mitigation measures.

The outlook for bank profitability has worsened recently. This is particularly relevant in euro area countries with a high share of variable rate loans (and a high dependence on deposit funding), where concerns about sustainable bank profitability are amplified by low credit growth. Additional rate cuts could weaken monetary transmission if lending rates fail to adjust or customers withdraw cash from banks. Declining bank profitability could then constrain credit expansion and undermine the aim of monetary easing.

NIRP involves a difficult trade-off between implementing unconventional policy measures to support aggregate demand and mitigating adverse effects of declining net interest margins on the bank lending channel. The micro-level risks associated with negative interest rates warrant a balanced use of non-standard policy measures, to help mitigate risks to banks’ profitability and its impact on monetary transmission. Looking ahead, further monetary accommodation could rely more on credit easing measures and a further expansion of the ECB’s balance sheet. Focusing on asset purchases would raise asset valuations and aggregate demand, while also supporting the bank lending channel. More generally, further easing would also benefit from more centralized fiscal policy support and structural reforms to support aggregate demand going forward.