Market Reaction to Bank Liquidity Regulation

We measure market reactions to announcements concerning liquidity regulation, a key innovation in the Basel framework. Our initial results show that liquidity regulation attracts negative abnormal returns. However, the price responses are less pronounced when coinciding announcements concerning capital regulation are backed out, suggesting that markets do not consider liquidity regulation to be binding. Bank- and country-specific characteristics also matter. Liquid balance sheets and high charter values increase abnormal returns while smaller long-term funding mismatches reduce abnormal returns. Banks located in countries with large government debt and tight interbank conditions or with prior domestic liquidity regulation display lower abnormal returns.


I. Introduction
We present the first empirical analysis of how stock markets respond to the new global standards of bank liquidity regulation which are gradually adopted between 2015 and 2018.
Unlike capital regulation, which has been a cornerstone of the regulatory framework for years, regulating bank liquidity was typically not a major concern prior to the recent crisis in most countries. The regulatory package (known as Basel III) that the Basel Committee on Banking Supervision (BCBS) developed in response to the recent financial crisis 1 constitutes a fundamental change in that respect. For the first time, the BCBS introduces global standards for bank liquidity to improve liquidity risk management and banks' ability to withstand liquidity shocks (Calomiris, Heider, and Hoerova (2015)). Policy makers view these new standards as a major improvement in regulation after the recent crisis, as the following quote illustrates: [The Basel III Framework is] a landmark achievement that will help protect financial stability and promote sustainable economic growth. The […] global liquidity framework will significantly reduce the probability and severity of banking crises in the future 1 During the crisis, several banks faced substantial liquidity outflows and shortages owing to overreliance on volatile funding sources, improper asset-liability management, and off-balance-sheet positions which gave rise to liquidity risk (Strahan (2012)). Consequently, banks hoarded liquidity (Acharya and Merrouche (2013)), and reduced liquidity provision to other banks and to the real sector (Cornett, McNutt, Strahan, and Tehranian (2011)).

Nout Wellink Chairman of the BCBS and President of the Netherlands Bank 2
The new rules address different aspects of liquidity risk. The liquidity coverage ratio (LCR) aims to ensure that banks have high quality liquid assets available to accommodate short-term cash outflows. The net stable funding ratio (NSFR) requires a minimum volume of stable funding sources given a bank's asset structure to mitigate the risk of funding mismatches over a one year time horizon. These two ratios aim to achieve different objectives and they are, at least from a theoretical perspective, equally important for financial stability.
An empirical question that we seek to answer in the paper is whether the market perceives the LCR as more important than the NSFR, or vice versa.
By requiring banks to switch to higher quality and more liquid assets, and towards more stable funding sources, the new standards are likely to affect banks' operations, in particular the structure of the balance sheet in terms of maturity structure and asset and funding choices. Therefore, complying with these rules has potential to affect bank profitability and valuation.
Industry representatives expressed disquiet about the costs arising from the new rules which would reduce profits (Institute of International Finance (IIF) (2012)). The final version of Basel III, released after a five year period of intensive discussions and lobbying, entailed several amendments to the original proposal and a weakening of the initial guidelines.
To examine how market participants react to liquidity regulation, we use event study methodology and exploit seven official announcements by the BCBS regarding the introduction of global liquidity standards. We turn to this methodology because the new regulation is gradually implemented over several years from 2015 onwards and an examination of the long-term impact of the LCR and the NSFR on bank profitability is infeasible due to the lack of historical information for the two ratios and the impossibility to compute them with the current level of disclosure in bank balance sheets. Therefore, only an event study approach and the combination of market and accounting data can be used to examine how markets expect bank profits to be affected.
We study European banks because their funding metrics compared unfavorably during and after the crisis with international peers, making them less prepared than U.S. and Japanese banks to meet the requirements in Basel III (European Banking Authority (EBA) (2012)).
Moreover, the sovereign debt crisis amplified between 2010 and 2012 the funding problems in Europe where banks are subject to different national liquidity regimes, making convergence a desirable, yet challenging, objective to level the playing field. Finally, our cross-country comparison that focuses on large listed banks that display bank-and country-specific heterogeneities allows examining how banks headquartered in similar yet different banking systems respond to the new rules.
Our empirical strategy proceeds in three steps. First, we examine cumulative abnormal returns (CARs), and cumulative market-adjusted returns (CMARs) over 3-day event windows to estimate aggregate and average market reactions to seven announcements by the BCBS relating to proposals, amendments, and revisions of rules concerning bank liquidity regulation for the period Feb. 2008-Jan. 2013. Second, we disentangle stock price reactions that also refer to confounding events such as announcements concerning capital regulation from those that exclusively address liquidity regulation. To this end, we calculate CARs and CMARs over event dates that only relate to statements regarding liquidity regulation. Third, we analyze whether banks respond heterogeneously. We correlate the cross-sectional variation in CARs and CMARs with bank-specific and country-specific characteristics. These tests illustrate how CARs and CMARs increase or decrease as an effect of the pre-existing liquidity conditions, funding mismatch, charter values, and country-specific characteristics that affect the two-way feedback between the sovereign's creditworthiness and refinancing conditions in the interbank market.
Several arguments suggest that bank stock prices may respond to liquidity regulation.
Even if the regulation is not implemented yet, bank stock prices can reflect the market perception regarding the possible consequences of liquidity regulation.
One view is that regulation serves the public interest to promote welfare at the expense of the regulated industry (Needham (1983)). Clearly, achieving and maintaining safe and sound banking, and mitigating systemic risk serves the public interest. Indeed, these arguments are put forward by the policy community for regulating bank liquidity. On the other hand is the capture theory. Stigler (1971) and Peltzman (1976) posit that regulation is desired by the regulated industry to reap benefits at the expense of opposing parties. Consistent with this view, lobbying by the banking industry against this regulation resulted in amendments that watered down the initial proposal and led to an eventually less restrictive regime.
First, the public debates about regulation during and after the crisis made investors anticipate tighter regulation, suggesting that markets may have expected regulatory changes prior to the announcements of the BCBS. We deal with this problem using a 3-day event window centred on the announcement day. Moreover, we test for anticipation effects using placebo events we pretend to occur prior to the actual events. Second, liquidity regulation is only one component of the regulatory framework. Some announcements relating to liquidity regulation coincide with announcements on capital regulation or other aspects of Basel III. Establishing a causal effect of market reactions to liquidity regulation requires identifying such confounding news.
A key contribution of our study is to disentangle the effects of announcements that also contain information concerning other components of Basel III (specifically, capital adequacy regulation) from the effects that are solely attributable to liquidity regulation. We refer to those tests as the Liquidity only bucket. These tests shed light on the question of whether market participants view liquidity regulation as binding and help establish their value relevance. Last, share prices may also respond to corporate events, such as changes in corporate governance or dividend cuts. We also show tests that omit bank-specific events that coincide with the BCBS announcements.
Our initial set of tests points towards negative stock price reactions to the introduction of liquidity regulation. The aggregate effect on shareholders of European banks based on the CARs for all seven events is equivalent to an average decrease in market capitalization between 269 and 354 million Euros, depending on the proxy for the market portfolio (MSCI Europe and MSCI World, respectively). These figures represent a decline in market capitalization of -5.27%, and -6.95%, respectively. However, our inferences are more nuanced when we separate out the announcements relating to capital regulation and other components of Basel III. Tests based on the Liquidity only bucket highlight that the average stock price reaction is either only marginally significant or insignificant. These analyses suggest that markets do not consider these new rules to be binding. Therefore, their value relevance is more limited than our initial results suggest. Our interpretation is plausible because the standards were not implemented yet. Precisely, while a smooth and gradual process of implementation was finalised for the LCR, the NSFR will only become a binding requirement by 2018 and observers anticipate further modifications to be made (Santos and Elliot (2012)).
Further tests illustrate that bank-specific variables explain the cross-sectional variation in stock price reactions to test the motivation behind the LCR and the NFSR. Banks with a greater LIQUID_ASSETS_RATIO (defined as the ratio of liquid assets to customer and shortterm funding) which serves as a proxy for the LCR, display less negative CARs and CMARs. This is in line with the idea that banks with more liquid balance sheets react better to the new rules. Our proxy for the NSFR, the CORE_FUNDING_RATIO, enters our tests with a negative sign, suggesting more negative CARs and CMARs for higher levels of CORE_FUNDING_RATIO. As explained in detail below, this result reflects that the CORE_FUNDING_RATIO is largely driven by Tier 1 capital, an expensive source of funds. 3 Country-specific factors also matter: banks from countries with more government debt experience more negative CARs and CMARs, and the negative wealth effects are also more pronounced if banks are headquartered in countries that already had national provisions for liquidity risk in place prior to the new rules.
Our research complements the evolving literature on Basel III. These studies address the effect of new regulation on economic growth and the cost of lending (Banerjee and Mio (2015), Duijm and Wierts (2016)). Our work is also related to the broader literature on 3 Including off-balance sheet items (such as guarantees and other contingent liabilities) and excluding liquid assets in the denominator of the CORE_FUNDING_RATIO does not alter our results. banking regulation and its impact on banking sector development, performance and stability (Barth, Caprio, and Levine (2004)), and the effect of regulation compliance on soundness (Demirgüç-Kunt, Detragiache, and Tressel (2008)). Our work differs from those studies by focusing specifically on the reaction by market participants to the introduction of a novel component of the Basel III framework: the regulation of liquidity risk.
We proceed as follows. Section II describes the institutional background, and Section III introduces testable predictions. Section IV describes the data and Section V the methodology.
The results are presented in Section VI. Section VII concludes. To strengthen banks' ability to withstand liquidity shocks over a short-term period, the BCBS developed the LCR. The standard requires banks to hold sufficient high quality liquid assets (HQLA), such as cash or government bonds, to meet a severe cash outflow for 30 days.

II. Institutional
The standard was introduced in 2015.
ii) To promote long-term resilience and reduce maturity mismatch, the BCBS proposed the NSFR. It incentivizes the use of stable funding sources by limiting short-term wholesale funding. Precisely, it requires banks to hold equity and liability financing expected to be reliable sources of funds over a one-year time horizon. The amount of stable funds required is conditional on liquidity characteristics of assets on and off the balance-sheet. The standard will be introduced in 2018.
Appendix A presents a detailed description of the two requirements. 4 The initial reaction to the announcements about regulating liquidity was negative. 5 Next, the BCBS decided to gradually introduce the standards to avoid detrimental effects on lending. 6 Consequently, the initial proposal was amended, and in two cases (Events #5 and #7) these adjustments resulted in a loosening of the liquidity requirements relative to previous announcements. We briefly discuss the key features of these events.
A first set of amendments was set out in the Annex to the BCBS press release on July 26, 4 Both measures are included in the Dec. 2010 Basel III Agreement (Event #6). From this point in time, supervisors monitored the ratios. However, there was a great deal of objection by the industry and the BCBS indicated the likelihood of further refinements in the formulas prior to the start of the mandatory period. Further, data gathered during the observation period may lead to further changes (Santos and Elliott (2012) 2010 (Event #5). The objective of the revisions was "to achieve a calibration and definition that penalises imprudent liquidity profiles, while minimising system level distortions" (p. 5).
In particular, the LCR was relaxed by widening the range of qualifying liquid assets, and the NSFR was modified to favor retail over wholesale banking, mainly by loosening requirements for customer deposits and residential mortgages.
In response to these objections and to take into account problems in the Eurozone, the BCBS modified the short-term liquidity standard (Event #7 We summarize the seven key events examined in this research in the following timeline. The LCR was introduced on Jan. 1, 2015 with a minimum requirement of 60%. It will rise in annual steps of 10% to reach 100% on Jan. 1, 2019. in greater detail. [

III. Testable Predictions
We develop predictions for how bank stock prices respond to liquidity regulation. Next, we discuss how bank-and country-level features amplify or mitigate the stock price returns.

A. Hypotheses on Market Reactions to Liquidity Regulation
Several arguments suggest that liquidity regulation attracts negative stock price reactions.
Introducing quantitative liquidity requirements aims to correct banks' suboptimal liquidity choices to restore optimal liquidity. A priori, it appears plausible to expect that such regulation interferes with banks' business operations which are geared towards profit maximization, and shareholders are likely to respond negatively.
The negative view on holding liquidity is driven by the fact that liquidity facilitates managerial moral hazard (Jensen (1986)). Moreover, liquidity holdings are considered to be costly for several reasons. Myers and Rajan (1998) highlight the 'dark side' of liquidity. By holding more liquid assets, firms are less likely to commit to a specific investment strategy that protects creditors because more liquid firms have a higher value in liquidation. A manager will therefore use the more liquid assets to alter the implicit property rights she has to maximize her own benefit. Indeed, Glaeser and Shleifer (2001) and Ratnovski (2009) propose that quantity regulation is costly: a precise definition of what constitutes a liquid asset in the context of bank liquidity regulation is difficult. While assets that are unconditionally eligible for central bank repo operations may have guaranteed liquidity, they are restricted to a subset of high-quality government bonds. Thus, these assets are expensive to hold. Other assets such as highly-rated corporate bonds may be liquid during tranquil times but their liquidity may dry up during market stress. Ex ante, it is therefore difficult to judge which assets constitute liquid assets that serve as a building block for liquidity regulation.
Moreover, mandating banks to hold liquid assets gives rise to opportunity costs because it interferes with asset choices in a similar vein as do minimum reserve requirements. Slovin et al. (1990) show that increases in unremunerated reserve requirements reduce shareholder wealth. Such requirements are an excise tax on banking activities that lower future cash flows.
While liquidity regulation does not compromise banks' asset choices and profitability to the same extent as do reserve requirements, the direction of the effect is similar. In absence of such regulations, banks may deploy the funds to finance loans that attract higher yields than government bonds. Liquidity regulation, similar to reserve requirements, is therefore likely to reduce income and, concomitantly, bank profits.
We also acknowledge an alternative hypothesis that predicts positive shareholder wealth effects. The point of departure for this view is that the new rules reduce systemic risk and limit contagion (Cifuentes, Ferrucci, and Shin (2005)). By forestalling contagious failures and improving soundness, liquidity regulation increases the probability of survival of banks, and simultaneously increases charter values. Banerjee and Mio (2015) illustrate the contagionreducing effect of liquidity regulation. They show that tighter liquidity regulation reduces interconnectedness of the banking sector. Finally, share prices may respond positively due to convergence benefits towards a global standard of liquidity regulation. Such convergence benefits arise from a common standard of regulating liquidity. This is beneficial because Beck, Todorov, and Wagner (2013) demonstrate in the context of resolving cross-border bank distress that national regulators may have biased incentives to intervene into such institutions.

B. What Explains the Cross-Sectional Variation in Stock Price Returns?
Beyond understanding how markets respond to liquidity regulation, we investigate whether bank-and country-specific characteristics amplify or mitigate the shareholder wealth effects. Economic intuition and prior work on liquidity risk suggest that bank-specific characteristics warrant further investigation. This analysis is also relevant from a policy perspective to enable regulators to gauge how banks are likely to respond to the new rules.
We focus on characteristics that capture the motivation behind liquidity regulation: funding mismatches and funding mix. Ideally, any such analysis pays attention to the LCR and the NSFR, and these ratios' components. However, the data required to compute both ratios are not yet disclosed in the balance sheets. To overcome this challenge, we rely on plausible approximations of the LCR and the NFSR using data from BankScope.
Subsequently, we also discuss the role of charter values because holding liquidity increases the likelihood of preserving the bank as a going concern and preserves the value of the charter.

Long-Run Funding Mix: Liability Composition
The mix of funding sources and the extent to which banks are exposed to funding mismatch, reflected in the composition of the liability side of the balance sheet, is likely to increase or decrease stock price reactions. The rules for liquidity regulation tighten the funding conditions by placing restrictions on what constitutes long-term stable funding sources. In particular, poorly capitalized banks as well as those relying on wholesale shortterm funds may be forced to issue equity, and move towards long-term borrowing to comply with the NSFR.
We argue that tightening funding for poorly capitalized banks may attract greater negative stock price reactions as these banks are more likely to suffer from an increase in funding costs. In contrast, institutions whose balance sheet characteristics signal that assets are largely supported by long-term stable funds are less pressed to adjust the asset-liability mix relative to banks far off the required standards. We therefore anticipate the negative shareholder wealth effects to be smaller in magnitude for banks with a relatively limited funding mismatch. This will be reflected in a higher coverage of assets financed by long-term funds (e.g., equity). This hypothesis matches the intuition behind the NSFR.
Hypothesis 1. A higher funding mismatch has a negative impact on the price reaction to bank liquidity regulation.
We test this hypothesis using the CORE_FUNDING_RATIO (the sum of Tier 1 capital plus hybrid debt and customer deposits divided by total assets) as a proxy for the NSFR.
Because this ratio is inversely related to funding mismatch, Hypothesis 1 is consistent with a positive coefficient.

Liquidity of Bank Balance Sheets: Asset Composition
Banks whose balance sheet characteristics suggest that the volume of liquid assets can absorb large and sudden cash outflows are in a better position to accommodate liquidity requirements, and are less pressed to adjust the asset and liability mix. 8 Thus, we expect the 8 Ideally, an empirical proxy for the NFSR also considers asset quality in terms of risk-weights, ratings, and issuers. However, in a cross-country study like ours such a granular level of data is not available.
adverse shareholder wealth effects that we anticipate to be less pronounced for banks with more liquid assets. Such a hypothesis directly tests the intuition behind the LCR.
Hypothesis 2. More liquid balance sheets will have a positive impact on the share price reaction.
To test this hypothesis, we use the ratio of liquid assets to customer and short-term funding (LIQUID_ASSETS_RATIO) as a proxy for the LCR. 9

Charter Values
Recent work by Ratnovski (2009) establishes a link between banks' liquidity choices and charter values. The basic premise is that maintaining liquidity insures the charter value. In a two-bank model, he shows that banks reduce liquidity if charter values are low or when banks expect shocks that will reduce charter value. These outcomes are due to inter-bank strategic complementarities with respect to liquidity choices. 10 When banks expect their competitors to 9 Customer funding consists of current accounts, savings accounts, and time deposits from customers. Note that meeting the standards of the LCR may also require banks to move towards more high quality assets. In other words, banks that have ex ante poor asset quality may have to adjust the asset side to comply with the LCR. We therefore examine the effect of a key indicator of asset quality, the ratio of non-performing loans to total assets, on CARs and CMARs. However, the t-statistic on this variable remains indistinguishable from zero and our other inferences are unaffected. The results are available from the authors. 10 Ratnovski (2009) focuses on crises. He shows that banks herd in equilibrium. This results in low levels of liquidity when they expect other banks to display suboptimal levels of liquidity. This is driven by banks' anticipation of rents that arise from bailouts that distort liquidity choices. The choice trades off the preservation be illiquid (e.g., during crises or when other banks suffer liquidity shocks) and bailouts are likely, a bank will choose high levels of illiquidity. In contrast, when other banks are liquid or when other banks are likely to survive liquidity shocks, banks have incentives to be liquid.
While the two-bank model takes a banking-system perspective, Ratnovski's (2009)  Our empirical strategy relies on two alternative measures of charter values: the Market to-book ratio (MARKET_TO_BOOK_RATIO), and the ratio of customer deposits to total assets (CUS_DEP_TO_TOT_ASSETS, a proxy for core deposits). These proxies find motivation in Keeley (1990) and Goyal (2005).
of charter values versus bailout rents. The intuition is that liquid banks have a higher probability to survive and realize long-term returns. In contrast, illiquid banks are likely to fail, and this attracts bailout rents. 11 In unreported tests, we run auxiliary regressions of the proxies for charter values (MARKET_TO_BOOK_RATIO and CUS_DEP_TO_TOT_ASSETS) on the LIQUID_ASSETS_RATIO. We find positive and significant associations, supporting Ratnovski's (2009) intuition in his model. The results are available from the authors.

Two-Way Feedback Loops between Banking Sector Conditions and Sovereign Debt
We also investigate country-specific factors to understand how they affect the market reaction to announcements related to the new liquidity regulation.
The interbank market conditions between core Eurozone members and countries located at the periphery such as Greece, Ireland, Italy, Portugal, and Spain (henceforth referred to as GIIPS) differ considerably. Figure 1 shows these countries on a map, illustrated in dark grey.

[FIGURE 1]
These characteristics may play a role for the way equity markets respond to liquidity regulation. Acharya, Drechsler, and Schnabl (2014) suggest that this divergence of interbank market conditions is largely due to sovereign problems in these peripheral countries, reflecting two-way feedback effects from sovereign credit risk to the domestic banking industry and vice versa. Le Leslé (2012) also attributes differences in bank funding conditions between core and peripheral countries to concerns about the fragility of the economies in stressed Eurozone countries. To illustrate, deposit rates in interbank markets at short maturities below one year for core countries in the Euro area declined in line with the Euribor spread in the aftermath of European Central Bank's actions in 2011 but deposit rates for banks located at the periphery remained around 150 basis points higher at the end of 2012. Given that tighter liquidity regulation implies higher cost of capital, we expect tighter interbank funding conditions in countries that experienced sovereign debt problems to trigger greater negative stock price reactions. Therefore, for countries whose banks are on average net lenders in the interbank market, we anticipate larger stock price reactions to liquidity regulation than for countries whose banks are, on average, net borrowers (and therefore have tighter funding conditions).

Hypothesis 4.
Higher average interbank ratios on the country level have positive effects on the price reaction to liquidity regulation.
Related to the previous argument is the link between a country's fiscal position, government debt, and domestic interbank market conditions. Budget deficits signal that governments spend more than they earn per fiscal year. Unless such deficits are exclusively financed by minting funds, they result in increases in national debt which further undermines the sovereign's creditworthiness, aggravating the adverse feedback effects between sovereign risk and the banks' refinancing conditions. Importantly, the recent bailouts in the form of recapitalizations are closely associated with increases in government debt. For instance, Lane (2012) highlights that Ireland and Portugal witnessed increases in debt to gross domestic product (GDP) ratios during the crisis, and Acharya et al. (2014) show that increases in financial sector distress correlate positively with increases in the public debt to GDP ratios.
Via the two-way feedback loop, the higher sovereign credit risk triggers further increases in the refinancing costs of the banking sector. For these reasons, we expect worse price reactions for banks domiciled in countries with high debt to GDP ratios, and in particular for those in the periphery countries, as they were most affected by the Eurozone crisis.
Hypothesis 5. Greater government indebtedness has a negative effect on the price reaction to liquidity regulation. Being domiciled in a GIIPS country has a negative impact on the price reaction to liquidity regulation.
Finally, we hypothesize that banks headquartered in the Eurozone may display more negative stock price reactions than banks located elsewhere. Figure 2 highlights in dark grey the countries that are not members of the Eurozone.

[FIGURE 2]
This differential effect is reflective of concerns about contagion in the Eurozone that may ultimately also affect the refinancing conditions in the banking sector and liquidity risk. Such concerns were repeatedly raised in the media, and also find support in academic work.  (0   otherwise). We use a dummy variable, GIIPS, to identify banks located in any of the five GIIPS countries that experienced sovereign debt problems.

Pre-Existing Domestic Liquidity Regulation
Several countries had some form of domestic liquidity regulation in place prior to the announcements by the Basel Committee.  (2012)).
Using several waves ensures obtaining information about guidelines on liquidity risk prior to supervisors this may also explain why new liquidity regulations have a relatively large negative impact on stock prices of banks from the Eurozone. Section IV.A highlights that many of the banks in our sample are indeed subject to the SSM.
the crisis, and we can also trace the evolution of these provisions during the sample period.
We focus on that section in the database which contains details about guidelines for liquidity risk. We then verify this information by cross-checking the details with information from the websites of the countries' central banks and the Bank for International Settlements (BIS). 14 Denmark, Germany, the Netherlands, Switzerland, and the United Kingdom had provisions in place that aim to bolster banks' liquidity holdings. Figure 3 highlights these countries in dark grey.

[FIGURE 3]
The Danish rules stipulate that banks are required to have adequate liquidity, and are based on ratios that resemble the spirit of the LCR. Similar provisions exist for Germany.
Since 2003, the Dutch guidelines also contain quantitative liquidity requirements that resemble the new proposals. Switzerland introduced quantitative liquidity requirements in 1988, and the United Kingdom also had minimum liquidity requirements in place with the objective to avoid maturity mismatches. We expect shareholders of institutions headquartered in these countries to respond more negatively to the new rules because they are potentially better able than shareholders elsewhere to gauge the effect of new regulation for bank profitability.
However, the effect of prior legislation may also go in the opposite direction. For banks that are already subject to national liquidity requirements any new liquidity requirement as part of Basel III may not lead to an observable reduction in bank profitability. Hence, it is also possible that shareholders of institutions headquartered in these countries will respond less negatively to the new liquidity regulation. We therefore offer two alternative hypotheses on the effect of pre-existing domestic liquidity regulation: Hypothesis 7A. Being headquartered in a country with pre-existing domestic regulations on liquidity has a negative impact on the price reaction to liquidity regulation.
Hypothesis 7B. Being headquartered in a country with pre-existing domestic regulations on liquidity has a positive impact on the price reaction to liquidity regulation.
Our tests use a dummy, LIQUIDITY_REGULATION, that takes on the value of 1 in instances when a country has pre-existing domestic liquidity regulation in place (0 otherwise). 15 A negative coefficient on this dummy variable would indicate that Hypothesis 7A is applicable, while a positive coefficient would support Hypothesis 7B. If the two effects offset each other, the coefficient may be insignificant.

IV. Data, Representativeness, and Choice of Event Dates
We now discuss our sample. While our empirical work only considers listed banks, liquidity regulation extends to all banks, irrespective of their listing. Therefore, we also discuss sample representativeness. Subsequently, we elaborate on the choice of our event dates.

A. Data and Representativeness
The starting point for the sample selection is the population of commercial banks and bank holding companies (BHCs) in the European Union (EU) and Switzerland, as reported in BankScope. BankScope is our source for the bank-specific variables used below to establish heterogeneous responses to liquidity regulation. We include Switzerland because of the vast size of its banking system and the linkages of Swiss banks with banks in the EU.
BankScope contains data for 142 banks and BHCs in Europe listed on a stock exchange.
We filter out banks without deposits to ensure that the sample banks engage in financial intermediation. This screen reduces the sample by 14 banks for which BankScope records zero deposits. We also omit seven observations for five banks due to negative common equity, resulting in a final sample of 128 banks and BHCs. Appendix C presents more details.
For these banks, we retrieve daily closing prices from DataStream for the period Feb. 21, 2007 to Jan. 7, 2013 for our event study.

B. Event Dates
Our event dates refer exclusively to official announcements by the BCBS that result in proposed or actual changes of liquidity regulation. This choice is a restrictive, yet plausible, criterion. Any other news and debates regarding the introduction of liquidity regulation are based on and influenced by the debates within the BCBS and its representatives. Third, we conduct a media search to ascertain that the events we focus on indeed convey significant information to the market. This media search helps rule out anticipation effects, a key concern for event study analysis. This condition ensures that the new information is available to all stock exchanges.

V. Econometric Methodology
Modeling market reactions to the announcements of the BCBS presents econometric challenges. The literature lacks consensus on the choice of the estimation method for abnormal returns. Moreover, Brown and Warner (1980) stress that short event-window models based on estimation of the security's beta (such as the market model or the CAPM) do not lead to significantly more precise estimates for the abnormal returns. For these reasons, we estimate both abnormal returns (AR) and market-adjusted returns (MAR) for the event window (-1,1), followed by calculation of the CARs and CMARs for each event (Focarelli, Pozzolo, and Casolaro (2008)). 17 We employ a variety of keyword searches to assess the international press coverage of the Basel Committee's initiatives included in our analysis. In particular, we use the following keywords: bank liquidity, liquidity proposals, Basel Committee, BIS, Bank for International Settlements, liquidity risk, Basel 3, Basel III, bank supervisors, bank supervision, and liquidity management. 18 When the hour of the press release is unavailable we screen the international press to check whether any European bank stock reaction is reported, on the date of the event, in response to the Basel Committee's announcement.
We estimate the AR using the market model extended to include day-of-the-week dummies (Kaplanski and Levy (2010)): (1) Consistent with prior work (e.g., Armour, Mayer, and Polo (forthcoming)), we employ an estimation window of 260 trading days (-261,-2) for the market model. Because we consider a long sample period, using different estimation windows for each event instead of only one estimation window for all events allows for potential parameter instability over time. We also adjust for first-order autocorrelation in the error term of the market model regression using the Prais-Winsten method (Allen and Wilhelm (1988)).
We compute the market-adjusted return (MAR) as the difference between the log return of the security (R i,t ) and the log return of the proxy for the market portfolio (R m,t ) The literature highlights that the MAR is free of bias resulting from significant events in the estimation period, which undermines estimation of the beta (Fuller, Netter, and Stegemoller (2002)). Since our sample period covers part of the 2007-2009 financial crisis and the Eurozone crisis, it is likely that significant events affected estimation of the beta.
We then compute the corresponding CAR and CMAR for the 3-day event window (−1, 1 For our regressions we rely on 3-day event windows, where t 1 is the trading day before the event and t 2 the trading day after the event.
We adopt the MSCI World and the MSCI Europe as proxies for the market portfolio.
These proxies, especially the first one, are less subject to bias than national indices because of potential effects of liquidity regulation on the stock prices of non-financial firms of a country.
The first step of our analysis focuses on the market-wide reaction to liquidity regulation using equal-weighted and market-weighted portfolios of the bank stocks in our sample. We compute the aggregate effect of the regulation by considering the average CARs and CMARs over all seven events, followed by tests for the four Liquidity only events that back out the effects of other aspects of Basel III.
To correctly gauge the market reaction to stricter liquidity rules even though Events #5 and #7 (see Table 1) are associated with loosening of the initial proposals, we follow the standard approach in the literature pioneered by Armstrong, Barth, Jagolinzer, and Riedl (2010), and multiply the corresponding CAR or CMAR by minus one for these two events.
For example, a positive price reaction for Events #5 and #7 suggests that bank shareholders benefit from looser liquidity requirements, and therefore it would not make sense to sum the raw CARs or CMARs for these events to the CARs and CMARs for the other five events. (3) After multiplying the CARs and CMARs for these events by minus one, the sum of CARs and CMARs for all events measures the market reaction to stricter liquidity rules appropriately.
Because we have only seven and four events for the total effects and the Liquidity only events, calculating the significance of the CARs and CMARs based on the assumption of Normality may lead to unreliable t-statistics. For this reason, we rely on bootstrap simulations to evaluate the significance of the cumulative effect of all seven events and the effect of the four Liquidity only events, similar to Armstrong et al. (2010).
The bootstrap simulations are performed as follows: we exclude days which fall in the 3day window for the seven events to consider only non-event trading days. Next, we randomly choose seven (or four, for the results pertaining to the Liquidity only bucket) non-overlapping placebo events for the period Feb. 1, 2008 to Feb. 28, 2013. This step is repeated 800 times.
Finally, we compute the sum of the CAR and CMAR over all seven (and four) events for each of the 800 samples of placebo events. These steps ensure that the simulated data represent the distribution of CAR and CMAR under the null hypothesis because they are estimated for nonevent trading days (for which announcements related to liquidity regulation did not occur).
The p-values are computed on the basis of the number of cases for which the CAR or CMAR is larger or smaller than the estimated value based on 2-tail tests.
For the second stage of our analysis, we run regressions of the CARs and CMARs on both a vector of bank-level (B i,t ) and a vector of country-level (C i,t ) characteristics: 19 19 Investigating the cross-sectional determinants of CARs using such a two-stage approach is common in banking.

VI. Results
We first provide a visual inspection of the behavior of CARs and CMARs around the event date, and then present the empirical analysis for the market reactions. Subsequently, we show tests that relate bank-specific and country-specific characteristics to the CARs and CMARs. 20 All coefficients in the tables on the cross-sectional regressions should be read as follows: A negative coefficient signals that an increase in variable B i,t generates a decrease in the CAR/CMAR (i.e., if the CAR/CMAR is negative, this amplifies the reduction). A positive coefficient signals that an increase in variable B i,t generates an increase in the CAR/CMAR (i.e., if the CAR/CMAR is negative, this mitigates the reduction and potentially leads to a positive CAR/CMAR).

A. Market Reaction to Bank Liquidity Regulation: Aggregate and Individual Effects
Graphs A-D in Figure 4 illustrate the behavior of equal-weighted and market-weighted CARs and CMARs around a 3-day event window centred on the announcement date (represented with a vertical solid bold line). The solid line represents CARs (Graphs A and B) or CMARs (Graphs C and D) estimated using the MSCI World as a proxy for the market portfolio, while the dashed line represents CARs (Graphs A and B) or CMARs (Graphs C and D) estimated using the MSCI Europe as a proxy for the market portfolio. Graphs A and C use a market-weighted portfolio of banks, while Graphs B and D use an equal-weighted portfolio of the bank stocks in our sample. The four graphs consistently indicate declines in CARs and CMARs around the announcement dates. A slight decline on the day prior to the announcements suggests that there may have been information leakages, this supports our choice to focus on the event window (-1,1). Table 3 presents our first set of analyses. We compute the total and the average effect for all seven events. Table 3 reproduces the findings for both equal-weighted and marketweighted portfolios. We also present bootstrapped p-values for the average CAR or CMAR, based on 800 bootstrap simulations for the period Feb. 1, 2008-Feb. 28, 2013, as discussed above. Importantly, we also present the results for the Liquidity only bucket based on randomly selected trading days. The latter setup is the most rigorous way of establishing the causal effects of liquidity regulation on bank stock prices. Table 3 uses the MSCI World as a proxy for the market index, and Panel B relies on the MSCI Europe. Panel A points towards negative wealth effects arising from the announcements. The t-statistics for the total effect are statistically significant in three cases out of four. The average CAR(-1,1) and CMAR(-1,1) ranges between -0.0095 and -0.0221, and the price reaction is stronger for the market-weighted portfolios (with a CAR of -0.0188 and a CMAR -0.0221) than for the equal-weighted portfolio (-0.0099 and -0.0095), suggesting that bigger banks react more negatively to the regulation than smaller banks. Panel B shows that using the MSCI Europe yields very similar inferences. The t-statistics are significant in three out of four cases for the total effect. Likewise, the average CAR and CMAR considering all seven events is larger in magnitude for the market-weighted portfolio (-0.0143 and -0.0159) than for the equal-weighted portfolio (-0.0075 and -0.0033).

Panel A of
The economic magnitude of these aggregate effects appears sizeable. Considering the sum of the CAR(-1,1) for all seven events for the market-weighted portfolios, the overall loss in market capitalization amounts to a staggering 354 million Euros in Panel A (equivalent to a reduction in market capitalization for all listed banks in Europe by -6.95%), and 269 million Euros in Panel B (a decline in market capitalization by -5.27%).

Individual Effects
Clearly, we focus on the aggregate effect of announcements concerning liquidity regulation. For completeness, however, we also briefly discuss the market reactions to the seven individual announcements. These tests are useful to establish if the market responses are consistent with the hypothesised direction of the effects in terms of tightening or loosening of the initial proposals concerning liquidity regulation. To preserve space, these tests are relegated to Table A.1 in our Internet Appendix (available at www.jfqa.org) which reports the CARs and CMARs.
Three observations stand out. First, several of the individual announcements attract negative market reactions. This is consistent with both the visual inspection of the data and the aggregate effect documented so far. Second, consistent with our expectations, Events #5 and #7 result in positive stock price reactions due to the loosening of the initial proposals (consistent with Table 3, we multiply these CARs and CMARs by minus one in Table A.1).
Third, the largest price reaction is for Event #7 which is the final and definitive announcement for the LCR which again relaxed previously proposed liquidity rules. The magnitude of the market reaction suggests that the extent of the relaxation surprised the market more than the previous announcements.

Liquidity only Bucket
However, these tests still leave open the possibility that confounding announcements from the BCBS on capital regulation or any other component of Basel III drive the results.
Concluding that liquidity regulation causally drives reductions in market capitalization would be inappropriate. Indeed, our inspection of the tests based on the Liquidity only bucket highlights that the t-statistics are only weakly significant or insignificant when we back out confounding effects. In line with the decline of statistical significance, the economic magnitude of the adverse price movements is less pronounced. The loss related to the events 1, 2, 3, and 7 amounts to 206 million Euros in Panel A, and 154 million Euros in Panel B.
Which factors are behind these weaker inferences? First, the findings for the Liquidity only bucket are likely to due to a loss of power of the tests when only four events are considered. Second, the events we exclude from the Liquidity only bucket are important for market participants in the sense that these announcements received extensive media coverage because they cover not only liquidity regulation but also other characteristics of Basel III. 21 The decline both in terms of statistical significance and in terms of economic magnitude is therefore not surprising. Liquidity regulation is only one component of the revised Basel framework. The findings for the Liquidity only bucket are therefore bound to be weaker.
Third, and most importantly, the revisions and the lobbying by the industry to mitigate the effects of tight liquidity regulation suggests that markets did not consider the new rules to be binding at the time of the announcement. The successful lobbying fuelled expectations about further modifications (Santos and Elliott (2012)). For instance, the LCR has been amended twice since its initial proposal, to eventually allow additional, less liquid instruments to be considered as high quality liquid assets (Fullenkamp and Rochon (2016)). This is consequently reflected in limited effects for bank valuation.
[  Table 3 also contains tests based on placebo events that assume the events considering the CARs and CMARs occur five trading days prior to the each one of the actual events.

Placebo Tests
These tests help understand whether the results are driven by a downward trend in returns of bank stocks relative to the market in the days surrounding the events. Moreover, these analyses also allow ruling out anticipation effects that may have occurred prior to the event windows.  Table 3 remain insignificant at the 5% level. These tests suggest the absence of anticipation effects and alleviate the concern that the significance of the results for the actual events is due to short-run trends in the CARs and CMARs. Table 4 shows how bank-and country-specific variables correlate with the stock price reactions. We again present results with both the MSCI World (Panel A), and the MSCI Europe (Panel B) index as a proxy for the market portfolio.

[TABLE 4: BANK-AND COUNTRY-LEVEL CHARACTERISTICS]
Importantly, these tests demonstrate that bank-specific variables explain much of the variation in CARs and CMARs. Our analysis that focuses on the intuition behind the LCR using the LIQUID_ASSETS_RATIO as a proxy for the LCR supports the idea that banks with liquid balance sheets display positive CARs with t-statistics between 3.494 and 4.268.
The proxy for the NFSR, the CORE_FUNDING_RATIO, enters the regressions with a negative sign. 22 This result empirically refutes Hypothesis 1. The pecking order theory provides a plausible explanation for this result. Myers and Majluf (1984) posit that issuing equity is expensive, a prediction that finds ample support in empirical work (e.g., Cornett and Tehranian (1994)). As the CORE_FUNDING_RATIO is considerably influenced by Tier 1 capital, which increases banks' weighted average cost of capital (Baker and Wurgler (2015)), well capitalised banks may react negatively to the new standards because they are likely to be 22 We obtain qualitatively very similar results when we substitute the CORE_FUNDING_RATIO with the Tier 1 capital ratio. These tests are available from the authors. more resilient to liquidity shocks. 23 Thus, they may be less willing to bear the additional costs of adjusting their asset/liability composition to comply with the new standards.
Shareholders of banks with more high quality capital seem to view liquidity regulation as particularly onerous and undesirable. To verify this statement empirically, we use an interaction between a CORE_FUNDING_RATIO_DUMMY (which takes on the value of 1 if the CORE_FUNDING_RATIO is above the 75th percentile, 0 otherwise) and our proxy for the LCR, the LIQUID_ASSETS_RATIO. In line with our interpretation, this interaction term enters significantly negatively, suggesting that banks with a large stable funding base are less exposed to liquidity shocks that may arise from a sudden dry ups of short-term funds.
Therefore, those banks are also less pressed to maintain large liquidity holdings to buffer such a shock.
The magnitude of the effect is considerable. The values from the second column in Table   4 for the coefficient on the CORE_FUNDING_RATIO indicate that moving from the sample's 25th percentile to the 75th percentile of the distribution of the CORE_FUNDING_RATIO is associated with a decline in the (-1,1) CAR of 1.98% and 2.15% for the models using the MSCI World and the MSCI Europe, respectively. The effects for the ratio of CUST_DEP_TO_TOT_ASSETS suggest increases in the (-1,1) CAR of 2.88% for the MSCI World (2.91% for the MSCI Europe). For the LIQUID_ASSETS_RATIO these values are 0.35% for the MSCI World, and 0.38% for the MSCI Europe. These results suggest that regulation on the LCR (as proxied by the LIQUID_ASSETS_RATIO) has had a relatively smaller impact on bank share prices than regulation on the NSFR (as proxied by the CORE_FUNDING_RATIO). A plausible 23 The correlation between the Tier 1 capital ratio and the CORE_FUNDING_RATIO is 0.59. interpretation for this result is that the regulation on LCR was severely dampened over time (especially in association with Event #7).
We find some support for the hypothesis that charter values matter. While the coefficients on MARKET_TO_BOOK_RATIO remain insignificant, the coefficients on the ratio of Customer deposits to total assets are positive and significant, consistent with Hypothesis 3.
Next, we discuss country-level characteristics. Location at the periphery, approximated by the GIIPS dummy variable, and Eurozone membership do not affect the CARs or CMARs. 24 However, pre-existing domestic liquidity regulation, government indebtedness, and the interbank market conditions play a role. The INTERBANK_RATIO is significantly positively related to the CARs and CMARs, suggesting that being domiciled in a country with 24 It may be argued that banks located in highly-indebted countries, the GIIPS, may not be able to use their governments' debts as collateral to borrow from the European Central Bank at all, or only with large haircuts, to obtain the liquidity they need to satisfy liquidity requirements. To address this, we run our baseline regressions again after replacing the dummy variable GIIPS with GIIPS\timesGovernment bonds, where Government bonds is the ratio of government bonds to total assets. As explained by Gennaioli, Martin, and Rossi (2014), while BankScope does not provide a breakdown of government bonds by nationality, more than 75% of banks' bondholdings correspond to domestic bonds. Therefore, the interaction between GIIPS and GOVERNMENT_BONDS captures the effect of holding a large portion of assets in bonds issued by GIIPS countries. The coefficients on GIIPS × GOVERNMENT_BONDS are large and significantly negative. Results are available from the authors. In unreported tests, we also examine if Eurozone banks are more strongly affected during the Eurozone crisis to further examine the two-way feedback loop. We use an interaction term between EUROZONE and a dummy for CRISIS years that takes on the value of one in the years from 2010 onwards. The magnitude of the coefficient on the interaction term is larger than for the EUROZONE dummy in the main specification, reinforcing the view that the nexus between banking conditions and sovereign debt amplifies the stock price reactions. tighter funding conditions in the interbank market decreases the CARs and CMARs. This result is consistent with Hypothesis 4. There is also some evidence of negative effects from pre-existing domestic liquidity regulation which supports Hypothesis 7A. The coefficients indicate that shareholders of banks in countries that already had some form of liquidity regulation in place experience more negative wealth effects. The fact that bank-specific variables are more robustly associated than country-level variables with abnormal returns is not surprising. Bank-level characteristics are likely to play a greater role for the crosssectional variation in stock price returns than country-level characteristics. Table 5 replicates these regressions but excludes the Events #4, #5, and #6 to focus on the Liquidity only bucket. Our findings are similar.

C. Further Robustness and Falsification Tests
A potential concern using event study methodology is that the news released by the BCBS coincide with bank-specific events such as earnings announcements, ratings downgrades, and changes in corporate governance structures which also affect share prices.
To rule out such concerns, we screen the international press via LEXIS/NEXIS and use the following keywords: dividends, earnings, CEO, losses, write-downs, restatement, downgrade, rating, fraud, annual report, manipulate, inspection, restructuring, M&A, merger, acquisition, stock split, dilution, fired, restructuring, issue, takeover. We then replicate the regressions from Tables 4 and 5 but exclude banks for which our news search suggests the presence of confounding events over a 3-day window, centred on the event day. Our findings remain unchanged. These tests are relegated to the Internet Appendix, Tables A.2 and A.3.
While our main tests already correct for heteroskedasticity, equation (1) may not sufficiently capture the impact of volatility changes on the standard errors. The phenomenon that volatility tends to cluster can undermine the assumption of constant variance. If volatility increases around the announcement days we study and we ignore such volatility clustering, we may unintentionally over-reject the null hypothesis (Boehmer (1991)). To address this issue, Internet Appendix, Table A.4, shows tests based on GARCH modelling. 25 Our inferences remain unchanged when all seven events are considered. Likewise, the results for the cross-sectional determinants of the CARs and CMARs remain unaltered when we exclude the announcements associated with a negative effect on the probability to impose stricter liquidity rules (i.e., Event #5 and #7).

Falsification Tests
Prior to offering some concluding remarks, we embark upon a falsification exercise. This analysis raises the question of whether the effects we capture are limited to bank stocks.
We run the tests from Table 3 but use the national indices (see Appendix D) instead of the portfolios of bank stocks to calculate CARs and CMARs. The idea is to analyze whether other firms, i.e., non-financial corporations, are affected by the announcements by the BCBS. 25 We estimate the market model similar to equation (1), but according to an AR(1)-GARCH(1,1) model instead of a Prais-Winsten model. The conditional mean equation is estimated as follows: The conditional variance equation is estimated according to: The daily abnormal returns are the residuals of the conditional mean equation.
The results are shown in Table A.5 of the Internet Appendix. Out of 18 tests in Panel A that focus on CARs, only the analyses for Belgium, Luxembourg, and the United Kingdom enter weakly significantly, all other tests remain insignificant. Focusing on the Liquidity only bucket further strengthens these inferences: we only find significant effects for the United Kingdom. Panel B reports on CMARs. We only find one significant association for the full sample (Luxembourg), and one for the Liquidity only bucket (United Kingdom). Thus, for most of the countries in the sample, we can rule out any simultaneous effects for other listed companies arising from liquidity regulation.

VII. Concluding Remarks
In this paper, we use event study methodology to present the first analysis of the effect of bank liquidity regulation, a major innovation of the Basel III framework, on bank stock prices.
Central to this landmark event in banking regulation are two new ratios that address liquidity risk, the NSFR and the LCR.
The regulatory process leading to the introduction of liquidity regulation consists of seven separate but related announcements by the BCBS. Over a time horizon of approximately five years, the BCBS made several amendments and revisions to the initial proposal in response to comments received by the banking industry. We exploit this gradual release of new information about details of the new regulation to establish the effects on shareholder wealth in terms of CARs and CMARs for listed banks in Europe.
While proponents of liquidity regulation place great emphasis on pointing out the role of global liquidity standards for banking system soundness, the controversial debate and criticism highlighted by the banking industry in the run-up to the implementation of the new rules suggests that regulating liquidity is considered to be costly.
Our first set of results indeed suggests that market participants respond negatively to announcements about bank liquidity regulation. This finding is consistent with the view that complying with liquidity regulations interferes with banks' asset and liability choices and ultimately reduces profits. However, further tests that separate out stock price reactions to liquidity regulation from responses to announcements to capital regulation and other aspects of Basel III which occur at the same time and are contained in the same press release, weaken these inferences and the average stock price reaction is at best weakly significant. Two reasons may be responsible for the weaker inferences: First, the long time horizon until the rules are fully enforced may have led markets to believe that the new regulation is not (yet) binding, and therefore only of marginal value relevance at the timing of the press release. This may be further amplified by the challenges posed by any new regulation: banks can only gradually build the competencies and procedures to comply with the rules and may not yet fully comprehend the ramifications that arise from liquidity regulation. Second, and more importantly, while liquidity regulation is likely to gradually become an important tool of the regulatory framework, capital regulation currently remains the dominant component of Basel II, both in terms of what is considered by markets to be binding and in terms of value relevance. What our tests for the Liquidity only bucket indicate is that liquidity regulation currently has some, yet limited potential to affect bank conduct above and beyond capital regulation. Therefore, these findings can be viewed to constitute suggestive evidence that regulating bank liquidity is a complement to capital regulation.
In addition, our data support the view on regulation according to which not all regulated firms are equally affected. Bank-specific characteristics such as the liquidity of balance sheets and the exposure to funding mismatch decrease negative stock price reactions. We also find some support for theories about the role of charter values which posit that holding liquidity insures charter values. Country-specific characteristics also play some role. The two-way feedback loop between the domestic banking industry and the sovereign's creditworthiness affects shareholder wealth effects. High government indebtedness and strained funding conditions in the domestic interbank market reinforce each other, and banks from countries with these characteristics display bigger share price reactions than elsewhere in Europe. The cumulative abnormal returns are lower in countries if banks were subject to domestic liquidity regulation that was in place prior to the announcements by the BCBS.
Clearly, our initial exploration of the market reaction towards the introduction of quantity regulation of bank liquidity can only be viewed as a starting point for a more comprehensive research agenda that explores the effects of liquidity regulation. Ultimately, more work is required to better understand how the new rules will affect bank conduct, and asset and liability composition. Additional work is also needed to establish how the new rules affect the soundness of individual banks and the financial system on the whole. We leave these questions for future research.

Appendix A. The Liquidity Standards
The liquidity coverage ratio aims to ensure that a bank has an adequate stock of unencumbered high quality liquid assets which consists of cash or assets that can be converted into cash at little or no loss of value in private markets to meet its liquidity needs for a 30 calendar-day liquidity stress scenario:

LCR =
Stock of HQLA Total Net Cash Outflows over the Next 30 Calendar Days ≥ 100% As in its final version (Jan. 7, 2013), in order to qualify as HQLA (the numerator of the ratio), assets should be liquid in markets during a time of stress and, in most cases, be eligible for use in central bank operations. HQLA are comprised of Level 1 and Level 2 assets. Level 2 are subject to limits and a range of haircuts conditional on their market liquidity.
The denominator of the LCR is the total net cash outflows, i.e. total expected cash outflows, minus total expected cash inflows. Expected cash outflows (inflows) are calculated by multiplying the outstanding balances of various categories or types of liabilities and off-balance sheet commitments with the rates at which they are expected to run off or be drawn down. 26 Banks are expected to meet this requirement on an on-going basis. However, during a period of financial stress, banks are allowed to use their stock of HQLA, thereby falling below 100%. 27 The standard was introduced on Jan. 1, 2015, but a gradual approach will be followed as the minimum requirement will be initially set at 60% in order to rise in equal annual steps to reach 100% on Jan. 1, 2019, as reported below: The net stable funding ratio (NSFR) establishes a minimum acceptable amount of stable funding based on the liquidity characteristics of an institution's assets and activities over a one year horizon. The NSFR aims to limit over-reliance on wholesale funding during times of buoyant market liquidity and encourage better assessment of liquidity risk across all on-and off-balance sheet items. In addition, the NSFR approach would help to counterbalance incentives for institutions to fund their stock of liquid assets with short-term funds that mature just outside the supervisory defined horizon for that metric. The standard is expressed as the ratio:

NSFR = Available Amount of Stable Funding Required Amount of Stable Funding ≥ 100%
As for the numerator, "Stable Funding" are those types and amounts of equity and liability financing expected to be reliable sources of funds over a one-year time horizon under conditions of extended stress. The amount of such funding required for a specific institution (the denominator of the ratio) is a function of the liquidity characteristics of various types of assets, off-balance sheet contingent exposures, and/or the activities pursued by the institution.
Liabilities and assets are weighted according to their stability and liquidity characteristics, respectively. The standard will be introduced by Jan. 1, 2018. years include an expansion in the range of assets eligible as HQLA and some refinements to the assumed inflow and outflow rates to better reflect actual experience in times of stress.

Event Description
The main measures are summarized as follows: i) the expansion of the list of HQLA by the introduction of Level 2B assets (subject to higher haircuts and a limit of 15% of total HQLA), including corporate debt securities rated A+ to BBB-and certain unencumbered equities (both subject to a 50% haircut), certain residential mortgage-backed securities rated AA or higher (with a 25% haircut); ii) a more favourable treatment of: insured deposits, by a lower outflow on certain types of fully insured retail deposits (from 5% to 3%); of fully insured non-operational deposits from non-financial corporates, sovereigns, central banks and public sector entities (from 40% to 20%), "non-operational" deposits provided by non-financial corporates, sovereigns, central banks, and public sector entities (from 75% to 40% ); iii) a more favourable treatment of committed liquidity facilities to non-financial corporates with the reduction of the drawdown rate on the unused portion of committed liquidity facilities to non-financial corporates, sovereigns central banks, and public sector entities (from 100% to 30%). Similarly, a better treatment has been applied to interbank credit and liquidity facilities (distinguished from inter-financial credit facilities), in order to reduce the outflow rate on the former from 100% to 40%; iv) a better treatment of central bank operations by reducing the outflow rate on maturing secured funding transactions with central banks from 25% to 0%; trade finance, including guidance to indicate that a low outflow rate (0%-5%) is expected to apply; v) a new and standardized treatment for derivatives positions, comprising additional derivatives risks included in the LCR with a 100% outflow (relates to collateral substitution, and excess collateral that the bank is contractually obligated to return/provide if required by a counterparty); a standardised approach for liquidity risk related to market value changes in derivatives positions; net outflow of 0% for derivatives (and commitments) that are contractually secured/collateralised by HQLA.
In addition, the BCBS has agreed a revised timetable to phase in the standard and to give effect to the BCBS intention for the stock of liquid assets to be used.  Table 1 shows the event dates, and we also provide a brief description for each event. The final column gives an overview about whether the event increased or decreased the probability to introduce stricter rules for liquidity regulation.      and CMAR over all seven events for each of the 800 samples of placebo events. These steps ensure that the simulated data represent the distribution of CAR and CMAR under the null hypothesis, because they have been estimated for non-events trading days (for which, therefore, announcements related to bank liquidity regulation did not occur). The p-values are computed on the basis of the number of cases for which the CAR or CMAR is larger or smaller than the estimated value (2-tail tests). *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.    (1)-(4). We use random effects regressions with robust standard errors clustered at the country level. CORE_FUNDING_RATIO_DUMMY is a dummy that takes the value 1 if the CORE_FUNDING_RATIO is above the 75th percentile, and 0 otherwise.

Events and Predicted Effects on Probability of Stricter Rules after Each Event
All bank-level variables, CARs and CMARs are winsorized at the 1st and 99th percentile.
Constant included but not reported. t-statistics in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.   CMARs. These tests are identical to those shown in Table 4, except for the fact that we focus on the Liquidity only bucket. Panel A uses the MSCI World as a proxy for the market portfolio and Panel B uses the MSCI Europe as a proxy for the market portfolio. The CARs and CMARs are estimated according to equations (1)-(4). We use random effects regressions with robust standard errors clustered at the country level. All bank-level variables, CARs and CMARs are winsorized at the 1st and 99th percentile. t-statistics in parentheses. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Core and Periphery Countries in Europe Figure 1 plots a map with all countries in our sample. Greece, Ireland, Italy, Portugal, and Spain are considered periphery countries in the public debate about the Eurozone crisis.
These countries experienced sovereign debt problems; they are highlighted in dark grey.

FIGURE 2
Non-Eurozone Countries Figure 2 highlights the countries that are not members of the Eurozone (Sweden, Denmark, Switzerland, and the United Kingdom). These countries are highlighted in dark grey.

FIGURE 3
Countries in Europe with Pre-Existing Domestic Liquidity Regulation Figure 3 shows which countries had pre-existing domestic rules for the regulation of bank liquidity in place. Switzerland, the United Kingdom, Germany, and Denmark had such legislation prior to the announcements of the BCBS in place. These countries are highlighted in dark grey.

CARs and CMARs around the Event Dates
Graphs A-D of Figure 4 illustrate the behavior of equal-weighted and market-weighted CARs and CMARs around a 3-day event window centred on the announcement date (represented with a vertical solid bold line). The solid line represents CARs (Graphs A and B) or CMARs (Graphs C and D) estimated using the MSCI World as a proxy for the market portfolio, while the dashed line represents CARs (Graphs A and B) or CMARs (Graphs C and D) estimated using the MSCI Europe as a proxy for the market portfolio. Graphs A and C use a marketweighted portfolio of banks, while Graphs B and D use an equal-weighted portfolio of the bank stocks in our sample.