Euro Area Policies: Selected Issues
Author:
International Monetary Fund
Search for other papers by International Monetary Fund in
Current site
Google Scholar Close
Publication Date:
19 Jul 2011
eISBN:
9781462338542
Language:
English
Keywords:
ISCR; CR; state; TFP growth; policy measure; ECB bank lending; bank equity price; GIP bank; Piraeus bank SA; Systemic risk; Bank credit; Distressed institutions; Stocks
Download PDF (932.2 KB)

The European Union’s (EU) financial stability framework is being markedly strengthened. This is taking place on the heels of a severe financial crisis owing to weaknesses in the banking system interrelated with sovereign difficulties in the euro area periphery. Important progress has been made in designing an institutional framework to secure microeconomic and macroprudential supervision at the EU level, but this new set-up faces a number of challenges. Developments regarding the financial stability may assist in the continuing evolution of the European financial stability architecture.

Abstract

The European Union’s (EU) financial stability framework is being markedly strengthened. This is taking place on the heels of a severe financial crisis owing to weaknesses in the banking system interrelated with sovereign difficulties in the euro area periphery. Important progress has been made in designing an institutional framework to secure microeconomic and macroprudential supervision at the EU level, but this new set-up faces a number of challenges. Developments regarding the financial stability may assist in the continuing evolution of the European financial stability architecture.

II. ECB Policy Measures and Euro Area Banks4

A. introduction

14. Since the start of the current financial crisis in mid-2007, governments and central banks across advanced countries have implemented multiple standard and nonstandard measures to safeguard their financial systems. In the initial phase of the crisis, risks were confined to U.S. subprime asset, banks’ exposures, and funding profiles. In order to offset increased liquidity, counterparty, and credit risks, central banks started lowering interest rates (i.e., standard policy measures), but also introduced a number of nonstandard policy measures, such as broader access to liquidity (credit easing) and temporary FX swap lines. Following the failure of Lehman in September 2008, governments in advanced countries injected capital in banks, guaranteed liabilities and purchased/guaranteed impaired assets, in order to avoid widespread bank defaults. At the same time, central banks beefed up lender of last resort operations and started quantitative easing (IMF 2009a). In the euro area, the ECB reacted with a series of interest rate reductions, the expansion of liquidity provision to banks at longer maturities and under fixed rate full allotment (FRFA), expanded the collateral framework and set up a covered bond purchase program (CBPP) for 60 billion euros.

15. The intensification of the financial crisis in the euro area in 2010 forced European authorities to take additional measures. In late 2009, concerns started to emerge about a number of euro area governments’ ability to continue to support their banking sector as well as about the size and sustainability of their debt and deficits. Government bond spreads reached historic levels in a number of periphery countries and in May 2010, Greece received financial assistance from other euro area countries and from the IMF. The European Financial Stability Facility (EFSF) was created to cover the future needs of member states with solvency problems and was accessed by Ireland in November 2010 and by Portugal in May 2011.

16. The ECB contributed to the stabilization of the euro area by introducing additional nonstandard measures. In May–June 2010, it established the securities market program (SMP) to purchase government bonds in secondary markets, provided additional FRFA long-term refinancing operations (LTROs) and reactivated USD swap lines with the Federal Reserve. The ECB’s balance sheet expanded from 1 to 1.8 trillion euros and long-term refinancing increased from around 20 percent in 2005 to over 90 percent by early 2010. As to the standard measures, the ECB interest rate was maintained at a 1 percent lower bound between May 2009 and April 2011. The ECB considered its nonstandard measures necessary for sustaining financial intermediation in the euro area and maintaining the link between the official ECB policy rate, money markets and the market for longer-term securities (ECB, 2010).

17. While previous research examined the effectiveness of crisis-related policy measures, none examined directly their impact on banks. A number of papers have previously examined the effects of standalone government rescue measures on bank equity prices, CDS spreads and other market indicators (see, e.g., IMF (2009b), BIS (2009), King (2008)). Others analyzed the impact of central bank measures on selected money and capital markets (see, e.g., Gagnon and others (2010) for the U.S., Joyce and others (2010) for the U.K., Beirne and others (2011) for the ECB’s CBPP) and on credit and GDP (Borio and Disyatat (2009), Gambacorta and Marqués-Ibañez (2011), Peersman (2010), Fahr and others (2010)). Also related is a whole strand of research on the effects of monetary policy announcements on bank stock prices and interest rates (see, e.g., Blinder and others (2008)). However, until now, no studies seem to have specifically analyzed the impact of central bank policies on individual banks.

18. This paper aims at examining the effects of ECB policy measures on individual euro area banks and on systemic risk. A key question is whether ECB policy measures have affected banks uniformly, or whether their effectiveness depends on banks’ initial condition (for instance, more strongly affecting weak banks). It is also interesting to know whether the impact is different for standard (interest rate) versus nonstandard policy measures, and whether the central bank’s measures helped reduce systemic risk. These findings have relevance for the discussion of the ECB’s exit strategy, including issues of speed, sequencing, and whether a special facility for weak banks would be desirable.

19. The main findings are as follows. The ECB policy measures affect a large number of banks, as measured by the reaction of bank stock prices to policy announcements on a high frequency basis. This is also confirmed by analysis of systemic risk indicators, which shows a significant impact of various nonstandard policy measures, especially for the ECB’s move to fixed-rate full allotment (FRFA). For the other measures, however, the effects vary with the state of financial markets, location and the risk indicators at hand. At a quarterly frequency, the analysis shows a significant impact of standard and nonstandard policy measures on bank profitability. Here, the impact is shown to differ across banks depending on their strength in terms of capital, liquidity, funding and loan-deposit mix. Weak banks—defined as those scoring low on these dimensions—are negatively affected by rising interest rates, while most banks would benefit from withdrawal of nonstandard measures (except in the periphery, in case of reduction in liquidity provision), although the latter probably reflects the contemporaneous nature of the relation between nonstandard measures and banking health. The remainder of the paper discusses the methodology and data, presents and analyzes the results and ends with some policy conclusions.

B. Methodology and Data

Methodology

20. The paper examines the effects of ECB policy measures on banks according to three complementary approaches. The paper analyzes the effects of ECB measures, defined below, on: a) individual bank equity prices; b) indicators of systemic bank risk; and c) euro area banks’ quarterly return on assets (ROA). Choosing for three different approaches helps overcome model-dependence and allows for assessing the impact of ECB measures both in the short and longer run, as well as at an individual bank and systemic level.

21. First, an event study is undertaken on euro area banks’ equity prices. Bank equity prices are used to determine the short-term impact of various ECB policy measures around specific event days.5 In line with the traditional event study approach for equities, the capital asset pricing model (CAPM) is used to obtain estimates of expected returns and to construct abnormal returns (MacKinlay 1997):

R i t = α i + β i R t + ɛ i t ( 1 )

where Rit denotes bank i’s daily equity return, εit a bank-specific news factor, Rt the market return, and α and β are parameters estimated over a pre-specified estimation window. From equation (1), abnormal returns (AR) are computed as ARit = Rit – (ai + bi Rt) and cumulative abnormal returns (CAR) around the event as CARi,t = Σr ARit. These (cumulative) abnormal returns are further aggregated across events to obtain average bank-specific reactions to a set of policy measures. In addition to this parametric approach, a sign test was used, exploiting the sign of (cumulative) abnormal returns. Here the assumption is that the expected proportion of positive abnormal returns under the null hypothesis is 0.5 and CARs have equal probability of being positive or negative. Results for the latter test are not reported but were qualitatively very similar to those based on ARs and CARs.

22. Second, the paper estimates the impact of ECB policy measures on bank systemic risk indicators. Two types of systemic risk indicators are calculated: the first one is the normalized score from a principal component decomposition of bank CDS spreads distinguishing between core and periphery banks (first chart).6 The second one is based on cumulated bank equity returns (second chart). A vector autoregression (VAR) is run with these systemic risk indices, using daily data from late 2005 onwards, allowing for possible spillovers between core and periphery banking sectors, and between the different proxies for systemic risk. The EONIA rate is taken as proxy for the ECB standard policy measure. ECB nonstandard policy measures are introduced as zero-one dummy variables and in the case of the CBPP, FRFA and for the period with 1-year refinancing, as step variables equaling one during their existence and zero elsewhere. The VIX, sovereign bond spreads and money market spreads are used as exogenous variables in the VAR to control for broad changes in risk perception. Alternatively, a variant with refinancing volumes, SMP purchases and CBPP purchases as quantity variables is estimated.

uA03fig01

Euro Area Banking Risk Indices: 2007-2011

CDS and Equity Market-Based Risk Measures

Citation: IMF Staff Country Reports 2011, 186; 10.5089/9781462338542.002.A003

Source: Bloomberg, Datastream and staff calculations.1 Normalized score from a principal component analysis on 5-year senior bank credit default swap spreads, estimated using daily data (1 Jan. 2005-3 June 2011). The core risk index comprises CDS spreads of 35 banks and the GIP risk index 11 banks (from GRC, IRL and PRT). The first principal component captures 85.4% of the common variation across core country banks and 83.1% across GIP country banks. 2/ Based on country indices of weekly banking sector equity returns, cumulated since January 2007 (equally weighted returns; inverted scale). Periphery consists of GRC, IRL and PRT (GIP). Core euro area consists of other euro area countries except SVK and EST (which do not have a banking equity index).

23. Third, the impact of ECB policy measures on bank performance is estimated using panel regressions for quarterly bank return on assets (ROA). For a panel of banks from the event study which report on a quarterly basis, baseline and extended models for ROA are estimated. The baseline model includes bank- and country-specific variables. As for bank-specific factors, lagged ROA, loan loss provisions, the tangible common equity ratio (TCE), the liquid asset ratio and the share of wholesale in total funding are included. The panel also controls for country-level GDP growth, GDP growth volatility and cross-section fixed effects. In a next step, this specification was augmented by standard and nonstandard ECB policy measures. Changes in the Euribor are used as standard policy measure and the share of LTROs in total ECB refinancing and the ratio of ECB bank lending to GDP are included as alternative proxies for nonstandard measures (see also in Data section below). These policy measures are interacted with various bank-specific factors which allow for a differentiation across strong and weak banks. Ceteris paribus, we expect that banks in a weaker state (less liquid, more reliant on wholesale funding, with higher loan-deposit ratio and with lower TCE capital) should benefit more from the various ECB policy measures, although it can probably not be excluded that also strong banks benefit from some measures (e.g. from taking up liquidity from the ECB for a long period at a very low rate).

Data

24. The ECB’s policy instruments comprise both standard and nonstandard policy measures:

  • Standard policy measures relate to changes in the key ECB policy interest rate. When examining quarterly bank profitability, we use the 3-month Euribor rate, which is closely related to the ECB policy rate and has a greater impact on bank performance.

  • Nonstandard policy measures associated with the financial crisis include changes in the collateral framework, changes in the procedure and profile of refinancing operations (fixed rate full allotment and additional longer-term operations), and purchases under the covered bond and securities market purchase programs, as well as the introduction and usage of USD swaps (see Table II.1A). The effects are differentiated for each measure according to whether they imply a loosening or a tightening (normalization), or whether purchases are high or low in case of the SMP. For the CBPP, only news related to its announcement and reactions to the monthly progress reports is examined. For the SMP, some discretion was necessary to determine whether interventions were ‘high’ or ‘low’. Here, as well as in the case of standard measures, we analyze both event-day and broader event windows. When examining quarterly bank profitability, several proxies for nonstandard measures (NSM) are used: the share of long-term refinancing operations (LTROs) in total ECB operations, the size of the ECB balance sheet and the amount of ECB lending to banks as a percent of GDP or bank assets, as well as a news count based on the cumulative (net) number of measures announced during the quarter (see text figure). Interestingly, based on the share of LTROs and ECB lending to banks, some withdrawal of nonstandard policies is already underway. The same is true when considering the net versus gross count of nonstandard measures.

Table II.1.

Contribution of ECB Policy Measures in Systemic Risk VAR

article image
Source: Bloomberg L.P., Datastream and staff computations. Top panel reports VAR coefficients for core and GIP (GRC, IRL, PRT) bank credit default and bank equity risk indices, estimated with daily data from November 2005-May 2011. Bottom panel reports marginal impact of ECB policy measures on systemic risk indicators, at various levels of the interaction terms. The VAR controls also for the level and change in the VIX, changes in the average GIP 10-year bond spread vis-à-vis Germany (ΔGIP sovereign spread) and the Euribor-EONIA spread. ECB policy measures are captured by dummy and step variables (equal to one for the period when active and zero otherwise). Insignificant policy measures were deleted from the VAR. The VAR was estimated in first differences to ensure stationarity and with 3 lags. To save space, estimates of the endogenous variables are not reported. Statistically significant at a: 1, b:5 and c: 10 percent. In bold are significant marginal effects based on the significance of the parameters (at 10 percent level).
Table II.2.

Panel Regression Estimates of Banks’ ROA and Marginal Effects of ECB Policy Measures

article image
Source: Bloomberg, Haver Analytics, ECB, IMF staff computations Notes: Upper part of the table shows panel estimates for a cross-section of 60 euro area stock-listed banks from the wider sample of 86 banks used in the event study for which quarterly financial statement data are available. Estimation period is 2005:Q1-2011:Q1. Cross-section fixed effects are included but not reported here. Bank-specific variables are lagged one quarter in order to avoid strong endogeneity. LLP: loan loss provisions (percent of assets), TCE: tangible common equity ratio, LIQ: liquid asset ratio, WHOLE: share of wholesale in total funding, LDEP: loan-deposit ratio. Besides a baseline model, several variants are presented, interacting standard and non-standard ECB policy measures with bank-specific state variables (STATE). The standard policy measure is the change in the 3-month Euribor rate (ΔEuribor), while the non-standard measure (NSM) is either the percentage of long-term refinancing (LTRO) in total bank refinancing or total ECB bank lending as a percent of GDP. Lower part of the table computes marginal effects on bank ROA of a one percentage point change in interest rates, ECB operations/GDP and of a 10 percentage point change in the share of LTROs. Statistically significant at a: 1, b: 5 and c: 10 percent level. In bold are significant marginal effects based on the 10 percent significance level of the parameter estimates.
Table II.1A:

Set of ECB Policy Measures

article image
uA03fig02

Non-standard ECB policy measures (NSM)

Citation: IMF Staff Country Reports 2011, 186; 10.5089/9781462338542.002.A003

25. The event study and quarterly ROA panel relies on a set of listed euro area banks. For the event study, bank stock prices and company data are readily available from Bloomberg L.P. for a set of 86 euro area banks going back at least to early 2005 (the starting point for our study; see Table II.2A for coverage).7 For the analysis of quarterly bank ROA, a subset of 60 banks with available income and balance sheet information is examined.

Table II.2A.

Banks in the Sample for the Event Study and Panel Regression

article image

C. Results

Event study results

26. As expected, the ECB’s conventional interest rate policy affects a significant number of banks. On the event day, abnormal returns are significantly different from zero for 38 out of 86 banks (16 positive, 22 negative) in case of easing (excluding the October 15, 2008 rate decrease, which corresponded with other measures) and for 18 banks (13 positive, 5 negative) during the tightening cycle of 2005 (Figure II.1 Similarly, for CARs, results are similar in both easing (31 significant CARs: 14 positive, 17 negative) and tightening (13 significant CARs: 3 positive, 10 negative) cycles. The correlation between ARs and CARs is positive and significant: those banks that exhibit stronger one-day abnormal returns also tend to have more persistent multi-day abnormal returns (simple correlation 0.593, Kendall’s tau 0.366). The stronger and more significant reaction of banks during the easing cycle probably reflects the fact that this took place during the financial crisis when bank stock prices reacted more strongly to policy decisions than before. Moreover, official interest rate reductions typically are more quickly translated into banks’ lending rates and involve a compression in banks’ net interest margins, which tends to negatively affect profitability and stock prices. This effect appears especially important in countries with predominantly floating rate loans, such as Ireland and Portugal.

Figure II.1.
Figure II.1.

Euro Area Listed Banks: Abnormal and Cumulative Abnormal Returns

(January 2005–May 2011)

Citation: IMF Staff Country Reports 2011, 186; 10.5089/9781462338542.002.A003

Source: Bloomberg L.P.Figures show standardized abnormal returns (AR: left side) and cumulative abnormal returns (CAR: right side) for events listed in Table A.1. Values in excess of +/-1.96 are statistically significant at the 5 percent level.

27. Different nonstandard measures also affect a large number of banks. The expansion of refinancing operations and the change to FRFA, purchases under the CBPP and SMP, and the broadening of the collateral base were all associated with significant ARs for one third of the banks (see also Figure II.1). News on USD swaps affected nearly half of the bank ARs significantly. The majority of significant reactions are attributable to negative ARs, except for the news on the CBPP and USD swaps, where mainly positive ARs are found. The subsequent and still ongoing normalization—including the expiration of one-year LTROs, the (temporary) move back to variable-rate tenders, the tightening of the collateral framework and low SMP activity—affect between 10 and 30 percent of all banks, with mostly negative ARs for refinancing operations, suspension of USD swaps and SMP purchases. Similar results hold for longer event windows, judging from the similar size and patterns in the CARs. This suggests that announcements of unconventional measures are mostly perceived as reflective of the high uncertainty in the market and the relative scarcity of fundamental information in the midst of the crisis. Furthermore, it appears that bank stock returns react broadly similar to positive and negative news about the ECB’s unconventional policies, i.e., banks that underperformed on the day of credit easing measures tend to respond in the same direction to tightening or withdrawal of the same measures (correlations between 0.118 and 0.242).

28. These results lead to two questions which are answered by subsequent analysis: 1) did ECB policy measures significantly reduce risks at a broader systemic level, and 2) do ECB policy measures have distributional effects, i.e., do they affect weak and strong banks, or core versus periphery, differently?

Impact on systemic risk

29. Various nonstandard ECB policy measures appear to have a significant impact on systemic bank risk indices, but standard interest rate measures did not. In the 4-variable VAR with daily data, announcements of standard policy measures (dummies for interest rate changes) did not impact the systemic risk indicators. Also actual interest rates (or changes thereof) did not affect bank systemic risk indicators and hence were not included in the final specification (Table II.1). This probably reflects the fact that these measures are generally well anticipated by market participants and hence have no impact on market prices or risk.8 Various nonstandard measures, on the other hand, seem to significantly influence bank systemic risk across both core and periphery countries. Their effect is both direct and indirect, through the interaction with some market risk variable, which means that their effect on systemic risk tends to be state-dependent. More specifically, the results can be summarized as follows:

  • Bank credit default versus equity risk indicators. Nonstandard measures appear to have overall a more significant impact on bank credit default than on equity risk indices, with 12 versus 8 (out of 18) significant impact effects.

  • Core versus periphery. For periphery banks, nonstandard measures have a strong impact on credit default risk indices but almost none on equity risk indices. For core banks, the impact is divided roughly equal across both types of risk indicators.

  • Median impact. The policy of fixed-rate full allotment seems to result in lower credit risk and higher equity returns, both in periphery and core euro area countries and hence appears to bring down bank systemic risk. This is in line with market intelligence which suggests that the ECB’s FRFA policy is important for banks. Apart from that, there is no consistent pattern as regards the impact of the other nonstandard measures on the bank risk indices, evaluated at the median of the market risk variables: sometimes they are positive and sometimes negative. For instance, easing of liquidity provision and CBPP tend to reduce bank credit default risk, while during the 1-year LTRO window, there was an increase in bank credit default risk. News on high SMP purchases tends to reduce credit default risk for core banks, while raising it for periphery banks. Hence, when evaluated at the median market condition, the systemic risk impact of nonstandard measures is very instrument-specific, without a clear overall picture emerging.

  • State-dependency. The impact of nonstandard measures varies across financial market conditions, judging from differences in signs of the marginal effects at the two ends of the distribution. For instance, measures that ease liquidity provision reduce default risk and increase equity returns when euribor-eonia spreads are low (at the 10th percentile), but vice versa when they are high (at the 90th percentile of the spread distribution). With news on SMP purchases, equity returns are boosted when the VIX index is low but fall when VIX is high.

Impact on banks’ quarterly performance

30. Both standard and nonstandard measures affect bank performance. When estimating a quarterly bank ROA panel with LTRO as proxy for nonstandard policies and TCE as state variable, both standard and nonstandard policy measures significantly affect bank performance (Table II.2, top panel). However, with wholesale funding or the loan-deposit ratio as state variable, only standard policy (interest rate changes) appears to affect banks’ ROA, while with liquidity, only nonstandard measures do. Hence, the significance of ECB policy measures on bank performance depends somewhat on the state variable used, although it is not entirely unreasonable that liquidity overtakes the significance of interest rate effects, while more structural bank-specific factors such as wholesale and loan-deposit ratios subsume the effect of the more structural, nonstandard measures. When estimating the panel with ECB lending as a percent of GDP as proxy for nonstandard policies, the significance of nonstandard measures changes: they lose significance in the models with TCE and liquidity as state variable, but gain significance in the model with wholesale funding. The fact the wholesale funding plays an important role in the model with ECB lending as proxy for nonstandard policies is not entirely implausible, given the substitution of market funding (wholesale and interbank) and ECB lending during the crisis.

31. Differentiating across banks, higher interest rates appear to benefit stronger banks, while during periods with heightened nonstandard policies, all banks seem to suffer. In order to capture the full effect of policy measures on ROA, one needs to account for both their direct and indirect impact, evaluated at various levels of the interaction term (Table II.2, bottom panel). This shows that stronger banks (i.e., banks with TCE and liquidity at the upper 10th or wholesale funding and loan-deposit ratios at the lowest 10th percentile) achieve higher ROA throughout this period in response to standard policy measures (tightening of interest rates). This is in line with expectations, since weak banks—defined as those in the bottom 10 percent of the distribution of the aforementioned bank-specific state variables—may be more vulnerable to increases in the cost of capital and funding, because they are perceived as more risky by market participants or have a smaller deposit base (knowing that bank revenues improve as deposits adjust less fully to interest rate increases; see, e.g., Gropp and others (2007)). Hence, according to these results, weak banks should raise capital in order to reduce the negative impact from withdrawal of standard measures. However, in response to an increase in nonstandard measures (a raise in either the LTRO share or ECB bank lending to GDP), virtually all banks—both weak and strong—appear to perform worse. This may reflect the contemporaneous nature of the relation between nonstandard measures and banking health: when banking performance deteriorated across the board as a result of the financial crisis, nonstandard measures were enacted to offset the risks to financial stability. Conversely, one may expect that these measures are withdrawn only when banks have sufficiently recovered from the crisis and would thereby not have negative repercussions on bank performance.9

32. Simulations confirm that higher interest rates would adversely affect weak banks, but not necessarily banks in the periphery, and improvements in ROA from exiting nonstandard measures are broad-based. A simulation exercise is undertaken, based on the above estimations and ROA from 2010:Q4, and assumes a 50 basis points increase in interest rates. Separately, a withdrawal of nonstandard measures is assumed—either reducing the LTRO share by 25 percentage points (in the specification with TCE as state variable) or reducing ECB lending to financial institutions in the euro area by 1 percent of GDP (in the specification with wholesale funding as state variable). In both cases, this would bring these nonstandard measures back to pre-crisis levels. Furthermore, another set of simulations is reported, distinguishing banks by location, based on similar panel estimations as in Table II.2, but interacting ECB policy measures with a location dummy instead of bank-specific state variables. The results are as follows:

  • Interest rate increase. Taking 2010:Q4 as a starting point, a 50 bps increase in interest rates would have relatively small overall effects on banking sector ROA, although there appear to be distributional effects: weak banks would see their profits substantially reduced, while strong banks would gain (text figure).10 Also location appears to matter: periphery banks, although starting from overall losses in 2010:Q4, would experience a substantial improvement in ROA, while other banks would see little change on average. One plausible explanation may be the difference in sensitivity of assets and liabilities to interest rate changes: new loan volumes at GIP banks are extended at very short-term maturities, which allows for a faster repricing in case of interest rate increases. GIP banks also have TCE ratios slightly above the average of the distribution. Furthermore, strong banks (with TCE capital buffers at the top of the distribution) generally have higher deposit funding and higher loan-deposit ratios, which makes them less sensitive to repricing of liabilities and more to asset repricing, relative to weak banks (i.e., with TCE ratios in the bottom of the distribution). A similar pattern emerges from a model with wholesale funding as the state variable (second figure): overall, there is very little impact on ROA from an interest rate increase, but banks with high reliance on wholesale funding seem to suffer, although they start from a very high ROA, while those with low wholesale funding benefit, because they are less sensitive to a repricing of liabilities and have less leverage. Also here, GIP banks (with wholesale funding ratios around the average of the distribution) stand to cut back their losses significantly.

  • Withdrawal of nonstandard measures. A reduction in the share of LTROs would lead to gains across the board, and across different types of banks. This possibly reflects the fact that nonstandard measures were expanded as the financial crisis evolved, and a withdrawal would likely coincide with a recovery of the financial sector. In other words, markets may expect nonstandard measures to be in place for as long as needed and they may be withdrawn when banks are able to record sound operating results. Results using wholesale funding as state variable and ECB lending as a percent of GDP as nonstandard measure are broadly similar. An exception appears to be GIP banks, which see their ROA decline following a reduction in ECB lending (as opposed to the shift in long versus short refinancing using the LTRO variable as nonstandard policy measure in the first text figure). This may reflect the sensitivity of their banking sector to outside financing and their difficulty to access market financing.

uA03fig03

Simulation of Euro Area Banks’ ROA for Increase in Euribor Rate and Withdrawal of Nonstandard Measures

Citation: IMF Staff Country Reports 2011, 186; 10.5089/9781462338542.002.A003

Source: Bloomberg, staff computations.Note: Shown are bank return on assets (ROA) in 2010Q4, after a 50 basis points increase in the euribor interest rate, a decrease in the share of LTROs by 25 percent (left figure) and in ECB lending by 1 percent of GDP (right figure), for all banks and differentiated by levels of capital strength (left), wholesale funding (right) and location. Overall, weak/strong TCE and high/low wholesale compute ROA impact using the average, 10th and 90th percentile from the respective Q3 2010 ratios. GIP is for banks located in GRC, IRL and PRT.

D. Conclusions and Policy Implications

33. Four broad messages can be taken away from the above findings:

  • Both standard and nonstandard ECB policies matter for a large number of banks, judging from the cross-section of abnormal stock returns. Reactions to interest rate changes appear to have been asymmetric: more banks responded significantly and negatively to declines than to rate increases, which may be due to greater sensitivity of assets than liabilities to repricing risk. Hence, from this perspective, future gradual interest rate increases may not affect euro area bank equity prices too negatively, especially if the ECB policy is well communicated to markets. Furthermore, nonstandard measures appear to have generated positive and significant abnormal returns for a large number of banks.

  • At a systemic level, the change in operating procedures to fixed-rate full allotment and—to some extent, the CBPP and measures to expand liquidity provision—helped to reduce systemic risks. However, for the rest, the evidence on the effects of nonstandard policy measures on system risk indicators is mixed and appears to be dependent on the state of financial markets. In terms of exit from nonstandard measures, withdrawal from liquidity measures and low purchases under the SMP seem to have been market-neutral, as they did not impact systemic risk indicators. Going forward, it may be necessary to keep nonstandard measures in place as long as systemic risk indicators remain elevated.

  • At a quarterly frequency, there is evidence of significant effects of both standard and nonstandard ECB policy measures on euro area banks. Their impact also seems to vary according to various dimensions: weak banks (with low capital, high wholesale reliance, high loan-deposit and low liquidity) appear to suffer from a policy of higher interest rates, while strong banks and banks in the periphery benefit. In response to nonstandard measures, most banks would benefit from an eventual withdrawal, except banks in the periphery dependent on access to ECB lending, although this may more reflect comovement rather than causality between nonstandard measures and bank health. These findings are confirmed by simulations on 2010:Q4 profits.

  • In sum, the results suggest care should be taken not to adversely affect weak banks and banks in the periphery. Raising interest rates and withdrawal of nonstandard measures should be properly timed, done in a proper sequence, and communicated consistently and well in advance to markets, as appears to be the case. This way, banks have sufficient time to adjust and look for alternative funding and capital to support or improve their performance. The findings also somehow support the ECB’s separation principle, allowing decisions on interest rates to be independent from those on nonstandard measures, as the impact on bank performance appears to be different across these policy measures.

References

  • Beirne, John, and others, 2011, “The Impact of the Eurosystem’s Covered Bond Purchase Program on the Primary and Secondary Markets,” ECB Occasional Paper 122, January (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

  • Bank for International Settlements, 2009, “An Assessment of Financial Sector Rescue Programmes,” BIS Papers 48, July (Basel: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation

  • Blinder, Alan S., and others, 2008, “Central Bank Communication and Monetary Policy: A Survey of Theory and Evidence,” ECB Working Paper 898, May (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

  • Borio, Claudio and P Disyatat, 2009, “Unconventional Monetary Policies: An Appraisal,” BIS Working Paper 292 (Basel: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation

  • European Central Bank, 2010, “The ECB’s Response to the Financial Crisis,” Monthly Bulletin (October).

  • Fahr, Stephan, and others, 2011, “A Monetary Policy Strategy in Good and Bad Times: Lessons from the Recent Past,” ECB Working Paper 1336, May (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

  • Fama, Eugene F., 1970, “Efficient Capital Markets: A Review of Theory and Empirical Work,” Journal of Finance, Vol. 25, pp. 383–417.

    • Search Google Scholar
    • Export Citation

  • Gagnon, Joseph, and others, 2010, “Large-Scale Asset Purchases by the Federal Reserve: Did They Work?” Federal Reserve Bank of New York Staff Reports 441, March (New York: Federal Reserve Bank).

    • Search Google Scholar
    • Export Citation

  • Gambacorta, Leonardo and David Marques-Ibanez, 2011, “The Bank Lending Channel: Lessons from the Crisis,” ECB Working Paper 1335, May (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

  • Gropp, Reint, Christoffer Kok Sørensen, and Jung-Duk Lichtenberger, 2007, “The Dynamics of Bank Spreads and Financial Structure,” ECB Working Paper 714, January (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation

  • International Monetary Fund, 2009a, Initial Lessons of the Crisis, February (Washington).

  • International Monetary Fund, 2009b, Market Interventions during the Financial Crisis: How Effective and How to Disengage? Global Financial Stability Report, Chapter 3, October, 117-152 (Washington).

    • Search Google Scholar
    • Export Citation

  • Joyce, Michael, and others, 2010, “The Financial Market Impact of Quantitative Easing,” Bank of England Working Paper 393, August (London: Bank of England).

    • Search Google Scholar
    • Export Citation

  • King, Michael R, 2009, “Time to buy or just buying time? The Market Reaction to Bank Rescue Packages,” BIS Working Papers 288, (Basel: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation

  • MacKinlay, Graig A., 1997, “Event Studies in Economics and Finance,” Journal of Economic Literature, March, pp. 13-39.

  • Peersman, Gert, 2010, “Macroeconomic Effects of Unconventional Monetary Policy in the Euro Area,” CEPR Discussion Paper No. 8348, April (Washington: Center for Economic and Policy Research).

    • Search Google Scholar
    • Export Citation

  • Rodriguez-Moreno, Maria and J. Ignacio Peña, 2011, “Systemic Risk Measures: The Simpler the Better?” Universidad Carlos III de Madrid Working Paper, February (Getafe: Universidad Carlos III de Madrid).

    • Search Google Scholar
    • Export Citation
4

Prepared by Nico Valckx.

5

Bank stock prices express market-based expectations of future discounted profits, and any news which impacts market expectations should be reflected instantaneously in prices, according to the efficient market hypothesis (Fama, 1970).

6

The first principal component of single-name CDSs are among the best indicators of systemic risk, according to Rodriguez-Moreno and Peña (2011).

7

Alternatively, event studies can also be set up with bank credit default swap (CDS) spreads, but these are available for a much smaller subset of banks and the series remain at quasi-constant levels until late 2007. For the analysis of interest rate policy effects, this is too short. However, as indicated above, bank CDS spreads are used for an impact analysis of the ECB’s policy measures under the first approach (as a systemic risk indicator).

8

One could even add that interest rate measures also generally are not intended to impact systemic risk.

9

To verify whether this contemporaneous relation is sample-specific, the panel models were estimated over a shorter time span, ending before the onset of the crisis (although this becomes then a very small sample, 2005:Q1-2007:Q3). Results remained qualitatively similar.

10

The overall impact is evaluated at the average TCE ratio in 2010:Q3, while the impact on weak (strong) banks is evaluated at the 10th (90th) percentile of the TCE ratio, taking into account both the direct effect on ROA of interest rates and the interaction with the TCE ratio level.

Same Series

Other IMF Content

Other Publishers

  • Collapse
  • Expand
Cover IMF Staff Country Reports
Euro Area Policies: 2011 Article IV Consultation—Lessons from the European Financial Stability Framework Exercise; and Selected Issues Paper
Author:
International Monetary Fund
Volume/Issue:
Volume 2011: Issue 186
Publisher:
International Monetary Fund
ISBN:
9781462338542
ISSN:
1934-7685
Pages:
145
DOI:
https://doi.org/10.5089/9781462338542.002
  • View in gallery

    Euro Area Banking Risk Indices: 2007-2011

    CDS and Equity Market-Based Risk Measures

  • View in gallery

    Non-standard ECB policy measures (NSM)

  • Figure II.1.
    View in gallery
    Figure II.1.

    Euro Area Listed Banks: Abnormal and Cumulative Abnormal Returns

    (January 2005–May 2011)

  • View in gallery

    Simulation of Euro Area Banks’ ROA for Increase in Euribor Rate and Withdrawal of Nonstandard Measures

Euro Area Banking Risk Indices: 2007-2011

CDS and Equity Market-Based Risk Measures

Non-standard ECB policy measures (NSM)
Figure II.1.

Euro Area Listed Banks: Abnormal and Cumulative Abnormal Returns

(January 2005–May 2011)

Simulation of Euro Area Banks’ ROA for Increase in Euribor Rate and Withdrawal of Nonstandard Measures