Volume 26 Issue 2 Article 3 June 2024 Implementing the Single Supervisory Mechanism in the Euro Area: Implementing the Single Supervisory Mechanism in the Euro Area: Effects on Deposit Structure of Banks Effects on Deposit Structure of Banks Emilija Popovska University of Ljubljana, School of Economics and Business, PhD Student, Ljubljana, Slovenia, emilija.popovska@gmail.com Marko Koš ak University of Ljubljana, School of Economics and Business, Ljubljana, Slovenia Follow this and additional works at: https://www.ebrjournal.net/home Part of the Finance and Financial Management Commons Recommended Citation Recommended Citation Popovska, E., & Koš ak, M. (2024). Implementing the Single Supervisory Mechanism in the Euro Area: Effects on Deposit Structure of Banks. Economic and Business Review, 26(2), 104-129. https://doi.org/ 10.15458/2335-4216.1337 This Original Article is brought to you for free and open access by Economic and Business Review. It has been accepted for inclusion in Economic and Business Review by an authorized editor of Economic and Business Review. ORIGINAL ARTICLE Implementing the Single Supervisory Mechanism in the Euro Area: Effects on Deposit Structure of Banks EmilijaPopovska a, * ,MarkoKošak b a University of Ljubljana, School of Economics and Business, PhD Student, Ljubljana, Slovenia b University of Ljubljana, School of Economics and Business, Ljubljana, Slovenia Abstract In this paper we investigate whether the banks which fall under direct supervision by the European Central Bank (ECB) are more likely to be considered more stable and trustworthy by the depositors due to the stricter supervisory activities performed since the implementation of the Single Supervisory Mechanism (SSM). Under the SSM, signicant banks switched from national supervisors to ECB, whereas the remaining banks remained under national supervisory authorities (NSAs). Using the difference-in-difference (DID) method, we have found evidence of increased depositors’ trust in signicant banks after the SSM implementation. Additionally, in anticipation of the SSM launch and the comprehensive assessment, we have found evidence of increased depositors’ trust in the banks which were expected to be supervised by the ECB. Keywords: Deposits, Interbank deposits, Banking supervision, Single Supervisory Mechanism JEL classication: G2 Introduction T he SSM is a system of banking supervision which was implemented in 2014 in the euro area as the rst pillar of the European Banking Union. The second and third pillars of the European Banking Union are the Single Resolution Mechanism and the common deposit guarantee scheme (ECB, 2018, p. 3). Before the SSM, the supervision of the banks in the euro area was performed inconsistently by national institutions in each country. The heterogeneity of bank regulation and supervisory practices across the countries caused difculties for implementing mea- sures to respond to the crisis in 2009 (Barth et al., 2013; Financial Crisis Inquiry Commission, 2011). There- fore, in 2012, the European Commission proposed the implementation of the SSM, aiming for consistent su- pervisory practices, increased safety and stability of the banks and restored trust in the banking sector (ECB, 2014b, 2016). The proposal for implementing the SSM was approved in 2013 by the Council of the European Union (Council of the European Union, 2013b; European Commission, 2012; ECA, 2014). Un- der the SSM, the ECB as the main decision body developed criteria for classifying banks as signicant or less signicant. The most signicant banks, which represent more than 80% of the total assets in the euro area, switched from national supervisors to the ECB, whereas the remaining banks remained under NSAs (ECB, 2014b). In October 2014, as a preparatory step for the banking supervision activities under the SSM, a comprehensive assessment was performed by the ECB on 130 banks which comprised 81.6% of the total assets in the euro area (ECB, 2014a). The compre- hensive assessment involved an asset quality review and stress test exercise. The asset quality review was aimed to assess the value of the banks’ assets, and the stress test to examine bank resilience (ECB, 2014a). The aim of this study is to inspect the effect of SSM on depositors’ trust measured via the total-deposits- to-total-assets and interbank-deposits-to-total-assets ratios. This is because improved bank stability is the Received 2 October 2023; accepted 28 March 2024. Available online 5 June 2024 * Corresponding author. E-mail address: emilija.popovska@gmail.com (E. Popovska). https://doi.org/10.15458/2335-4216.1337 2335-4216/© 2024 School of Economics and Business University of Ljubljana. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/). ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 105 primary goal of the SSM (ECB, 2014b). Therefore, we expect increased depositors’ trust in signicant banks, which switched from national supervisors to the ECB, in comparison to less signicant banks, which re- mained under national supervisors. We expect this trust effect to be more pronounced in short-term interbank deposits and less pronounced in insured deposits. Additionally, we investigate the immedi- ate trust effect of depositors in banks which were expected to be supervised by the ECB and to be eval- uated under the comprehensive assessment. For that purpose, we address the following research questions in this study: 1) whether the SSM implementation affected depositors’ trust in signicant banks com- pared to less signicant banks; and 2) whether the anticipation of the SSM launch and the comprehen- sive assessment affected depositors’ trust in the banks which were expected to be supervised by ECB, com- pared to the banks which were expected to remain under supervision by their NSAs. We provide empirical evidence of increased de- positors’ trust in signicant banks after the SSM implementation. The trust effect of the depositors is strongly demonstrated with the short-term interbank deposits, which are based on trust and are not collat- eralized. Long-term deposits are collateralized more frequently and are therefore considered safer. These results imply that the SSM has improved the credibil- ity of signicant banks, which in turn implies that the SSM is fullling its main priority, which is increased safety and stability of banks (ECB, 2014b). Addi- tionally, we provide empirical evidence of increased depositors’ trust in signicant banks in anticipation of the SSM and the expected comprehensive assessment. The trust effect of the depositors is strongly demon- strated with the short-term interbank deposits with a maturity of up to 3 months. These results imply that the ECB is perceived as a stricter supervisory authority compared to the NSAs and are consistent with the literature that investigates supervision archi- tecture (Colliard, 2020; Fiordelisi et al., 2017). We performed various robustness checks to assess the validity of our results. First, we applied placebo tests where, by creating a ctional time dummy variable, we assumed that the SSM had been imple- mented in 2012. We did not nd any differences in the interbank-deposits-to-total-assets ratio between sig- nicant and less signicant banks in 2012, the year before the announcement of the SSM. These results imply that the change of the interbank deposits struc- ture of signicant banks is associated with the SSM implementation rather than any other past events. Second, we found no evidence of changes in the share of interbank deposits in the period of SSM imple- mentation (2014) in banks in the European countries which are not part of the euro area and do not partici- pate in the SSM. These results imply that there are no other factors that could have affected the interbank deposits structure of the euro area banks apart from the implementation of the SSM. Third, we tested our results for sample selection bias, by removing France from our dataset. Our results are consistent with our main ndings and conrm the absence of sam- ple selection bias. Fourth, on this subsample without France, we applied placebo tests where, by creating a ctional time dummy variable, we assumed that the SSM had been implemented in 2012. We conrmed our results that the change of the interbank deposits structure of signicant banks is associated with the SSM implementation. Fifth, we inspected the effect of the xed interest rate on main renancing opera- tions on our DID coefcient, by adding it as a control variable in the model. We conrmed our results that increased depositors’ trust in the signicant banks is associated with SSM implementation. Moreover, we performed an additional analysis to inspect both the impact of SSM implementation and SSM anticipation on banks’ interbank deposits structure by different maturities. In this analysis, we applied placebo tests where, by creating a ctional time dummy variable, we assumed that the SSM had been implemented in 2012. We found no differences in the share of interbank deposits of any maturity between signicant and less signicant banks in 2012, the year before the announcement of the SSM. This implies that the changes of the interbank deposits structure by maturity are associated with the SSM implementation and not with other events. We did observe differences in the portion of total assets funded with total deposits between signicant and less signicant banks in the period of SSM imple- mentation as well as in the period of SSM anticipation. However, our robustness checks warned about the ex- istence of other factors that might have affected those differences rather than the SSM. Therefore, we cannot conrm that the implementation of the SSM and the anticipation of the SSM launch affected those differ- ences. A possible explanation for this is the presence of customer deposits and saving accounts in the total deposits of the banks, which cannot be affected by institutional changes such as the implementation of the SSM. This paper contributes to the recent literature stream on the SSM and to the established literature on supervision. In this respect, this study is related to Altunba¸ s et al. (2022); Alves et al. (2023); Avgeri et al. (2021); Avignone et al. (2021); Cuadros-Solas et al. (2023); Fiordelisi et al. (2017). These works provide evidence of reduced credit risk exposure (Avignone et al., 2021), lower sovereign risk (Cuadros-Solas et al., 106 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 2023), increased protability (Avgeri et al., 2021), worsened risk disclosure practices (Altunba¸ s et al., 2022), and adjusted lending behaviour (Fiordelisi et al., 2017) of signicant banks compared to less signicant banks, as well as evidence of improved asset quality of SSM-supervised banks in terms of de- creased nonperforming loans and loan loss reserves (Alves et al., 2023). Our study contributes to this lit- erature by providing empirical evidence of increased depositors’ trust in the signicant banks due to the SSM implementation and SSM anticipation. Our em- pirical results also have important policy implications for policy makers and supervisory authorities as they conrm that the SSM has improved the credibility of signicant banks and is fullling its main priority, which is increased safety and stability of banks and the overall banking sector (ECB, 2014b). The remainder of this paper is structured as follows: Section 1 reviews the literature, Section 2 describes the data and the methodology, Section 3 covers the empirical results along with the robustness checks, and Section 4 concludes the article. 1 Literature review and theoretical framework 1.1 Liquidity funding Liquidity funding is important for the riskiness of the overall banking sector, as it directly affects the risk prole of each individual bank and its probability of failure (Bologna, 2011). A diversied funding struc- ture leads to stable banks and a stable banking sector (Oura et al., 2013). Deposits are the optimal form of bank funding. First, this is because they help banks transform illiq- uid assets in their balance sheets (Diamond & Dybvig, 2000), and second, because they are a cheaper source of funding compared to equity capital (Allen et al., 2015). Moreover, deposits are an important source of liquid liabilities, because with them, banks provide liquidity in the economy (Kundu et al., 2021). On the other hand, deposits can be a source of bank vulner- ability. For instance, multinational banks are exposed to vulnerabilities by transmitting shocks as they col- lect deposits from countries in which they operate and allocate them as loans in other countries (Kundu et al., 2021). Deposit withdrawals, which happen in case of absence of depositor’s trust in the banking sector, also represent a source of bank vulnerability (Martin et al., 2018). Interbank deposits are another form of bank fund- ing, which happens on the interbank market and is based on trust. To participate on the interbank market, which has a crucial economic role for the movement of savings (Bruche & Suarez, 2010), banks need to demonstrate their creditworthiness (Allen et al., 2020). Their risk taking is monitored (Dinger & Hagen, 2009) and reected in the interbank inter- est rates (Furne, 2001). Interbank lenders tend to be better capitalized (Angelini et al., 2011), whereas interbank borrowers tend to engage in less risky activ- ities (Dinger & Hagen, 2009). Both interbank lending and interbank borrowing can decrease bank riskiness via diversication (Dietrich & Hauck, 2020) and via extending maturity periods (Dinger & Hagen, 2009), respectively. On the other hand, the interbank market is highly contagious and can be a source of instability and systemic risk (Allen et al., 2020; Bernard & Bisig- nano, 2000; Furne, 2001). 1.2 Banking supervision and SSM A signicant amount of research has been done to investigate banking supervisory architecture. Con- icting opinions exist regarding centralized and de- centralized supervisory practices. Decentralized su- pervisory practices, where banks are supervised by different authorities in different countries, lead to increased liquidity risk and lower capital ratios of banks (Barth et al., 2002). On the other hand, cen- tralized supervisory frameworks, which have one main decision-making body, promote incentives for moral hazard (Barth et al., 2004). Moreover, cen- tralized supervision where central banks have the role of supervisory authority can lead to increased riskiness in the banking sector due to higher nonper- forming loans (Barth et al., 2002). On the contrary, centralized supervision reduces possibilities for in- formation asymmetry and arbitrage (Ampudia et al., 2019). Moreover, centralized supervision is claimed to be the better option for multinational banks as decentralized supervision can lead to accumulated risk and bank failures. According to Calzolari et al. (2019), multinational banks supervised under decen- tralized supervisory frameworks tend to adjust their organizational structure by converting subsidiaries to branches in order to decrease their supervisory monitoring. Moreover, centralized supervision can accelerate cross-border activities of the supervised banks, as centrally supervised banks have lower funding costs and can easily obtain foreign funds (Colliard, 2020). Dual supervisory systems, on the other hand, are based on both centralized and de- centralized supervisory practices. In dual supervisory frameworks inconsistent implementation of identical rules is possible due to differences in the institutional design and the incentives of supervisory authorities (Agarwal et al., 2014). Moreover, in dual supervisory ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 107 frameworks, aligned incentives and goals between the supervisory authorities are needed for achieving effective supervision and consequently a stable and resilient banking sector (Carletti et al., 2021). The su- pervisory architecture affects the regulatory powers of supervisory authorities (Näther & Vollmer, 2019). However, regardless of the institutional design, all supervisory practices must be based on timely in- formation disclosure and absence of moral hazard. Moreover, they should ensure proper implementation of rules and regulations in order to increase stability and trust in the banking sector (Barth et al., 2004; De Larosière, 2009). The SSM is a dual supervisory framework based on centralized and decentralized supervision. It is centralized because the ECB is the main body for su- pervising signicant banks and decentralized because NSAs supervise less signicant banks by perform- ing supervisory tasks over which the ECB has no direct hierarchical control (Zeitlin, 2023). Recent liter- ature analyses its institutional design and the division of supervisory tasks and actions of the ECB and NSAs (Gortsos, 2023; Quaglia & Verdun, 2023). The SSM implementation is not anachronistic (Mansson, 2014). The literature provides little evidence regard- ing the effectiveness of its implementation. The SSM has proved to be effective in reducing the riskiness of the overall banking sector measured in terms of credit risk (Avignone et al., 2021) and sovereign risk (Cuadros-Solas et al., 2023). Moreover, it has proved to be effective in improving banks’ performance mea- sured in terms of increased protability (Avgeri et al., 2021) and improved banks’ asset quality (Alves et al., 2023). Additionally, it has proved to be effective in increasing the competitiveness of SSM banks located in weak economies such as Portugal and Greece (Sig- mund & Raunig, 2023). On the other hand, the SSM has negatively affected banks’ efciency in the early years of its implementation (Moura et al., 2023). Other weaknesses of the SSM are the possibility of agency problems due to the inefcient information ow be- tween the ECB and NSAs, regulatory weaknesses due to separation of supervision and regulation, and worsened risk disclosure practices of signicant banks (Altunba¸ s et al., 2022; Ferrarini, 2015). The SSM also affects banks’ stock returns (Loipersberger, 2018), investors’ response, which signals fear of regu- latory inconsistencies (Abad et al., 2020; Carboni et al., 2017), and third parties, such as creditors and clients (Möslein, 2015). Strong banking supervision reduces the overall risk in the banking sector (Buch & DeLong, 2008; Delis & Staikouras, 2011). Closer supervision of banks’ fund- ing structure leads to reduced riskiness of the banks and improved safety and resilience of the overall banking sector (Bologna, 2011). One of the main prior- ities of the SSM is increased safety and stability of the banks (ECB, 2014b). Therefore, banks which are su- pervised by the ECB are likely to be considered safer due to the stricter supervision, compared to banks which have remained under supervision by their NSAs. Consequently, we expect an increase of depos- itors’ trust in banks supervised by the ECB, compared to banks supervised by NSAs. This kind of deposi- tors’ perception should be especially identiable in case of non-protected deposits such as interbank de- posits, which are based on trust, and less pronounced in case of deposits covered by deposit insurance schemes. Therefore, we expect a positive effect of the SSM implementation on depositors’ trust and have developed the following hypothesis: The implemen- tation of the SSM has led to increased depositors’ trust in signicant banks compared to less signicant banks. A preparatory step for the supervision activities under the SSM included a comprehensive assess- ment. The comprehensive assessment involved an asset quality review and stress test exercise performed on 130 banks which comprised 81.6% of the total as- sets in the euro area (ECB, 2014a). It was publicly announced for the rst time in February 2013 and was performed in October 2014 on banks’ balance sheet data as of 31 December 2013 (Constancio, 2012, 2013; ECB, 2014a). The criteria with which ECB was going to select the banks on which it was going to perform the comprehensive assessment were publicly known in December 2012 (European Commission, 2012). Therefore, we argue that it was possible to identify the banks which were going to be assessed with the comprehensive assessment and which were going to be supervised by the ECB. Moreover, the aim of the comprehensive assessment was to evalu- ate the asset quality of the most signicant banks in the euro area and to check if they had an adequate capital buffer for withstanding shocks (ECB, 2014a). The literature provides evidence that banks which take part in stress test exercises decrease their credit risk exposure (Kok et al., 2023) and adjust their lend- ing behaviour (Fiordelisi et al., 2017). Therefore, we argue that banks which were going to be assessed with the comprehensive assessment and which were going to be supervised by the ECB were considered safer due to stricter supervision. Consequently, we expect an increase in depositors’ trust in signicant banks compared to less signicant banks. Therefore, we have developed the following hypothesis: The anticipation of the SSM launch and the comprehen- sive assessment has led to increased depositors’ trust 108 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 in the banks which were expected to be supervised by the ECB and to participate in the comprehensive assessment. 2 Data and methodology 2.1 Data The empirical research is built on a sample of panel data for 290 euro area banks which fall under the scope of the SSM, covering the period from 2011– 2018. Effects from changes in supervisory architecture are visible in the medium to long run (Fiordelisi et al., 2017). Therefore, in line with recent and ex- panding literature which investigates the effects of the SSM on the banking sector (Altunba¸ s et al., 2022; Avgeri et al., 2021; Avignone et al., 2021), we chose a narrow timeframe concentrated around the years of the SSM implementation, which allowed us to capture the impact of the SSM on intrabank and to- tal deposits. An additional reason why we excluded the years 2019–2021 from the analysis is the possi- ble effects of the pandemic-related economic crisis. In that period, the ECB implemented both supervi- sory and monetary policy measures aimed to restore banks’ safety and resilience (Quaglia & Verdun, 2023). The sample covers universal commercial banks, retail and wholesale, located in: Austria, Belgium, Cyprus, Estonia, Finland, France, Germany, Greece, Ireland, Italy, Latvia, Luxembourg, Malta, Netherlands, Por- tugal, Slovakia, Slovenia, and Spain. Banks located in Lithuania and in Croatia and Bulgaria were excluded from the analysis because these countries had only become members of the SSM and the euro area in January 2015 and in October 2022 (ECB, 2015, 2022), respectively. Moreover, central banks and investment banks were excluded from the analysis due to dif- ferences in the business model. Banks with missing observations of the dependent variables were also ex- cluded from the analysis. Consolidated bank-specic data (prepared under the IFRS reporting standard, on a yearly basis), were retrieved from the Fitch Connect database, and macroeconomic data from the World Economic Outlook database (International Monetary Fund); both were applied on constructed models. According to the ECB, banks are classied as signicant if they full one of the signicance cri- teria 1 (Council of the European Union, 2013a; ECB, 2014b; European Commission, 2012). In the sample, banks were classied as signicant or less signicant according to the ECB signicance criteria (ECB, 2014c). Most of the banks were classied as signi- cant or less signicant by considering the total assets criteria, with the exception of banks located in smaller economies (Malta, Slovenia, Slovakia, Estonia, Latvia, and Cyprus), which were classied by considering other signicance criteria (total assets above 20% of GDP , signicant cross-border activities, and being one of the three largest credit institutions in the country) because they are smaller (ECB, 2014b). The classica- tion of banks in the sample resulted in 121 signicant banks, which fall under direct supervision of the ECB (treatment group), and 169 less signicant banks, which fall under supervision of their NSAs (control group) 2 . Table A1 in the Appendix lists banks lo- cated in smaller economies and the criteria for their classication. Table 1 displays all variables used in our analy- sis, and Table 2 displays the summary statistics. The ratio of total deposits to total assets (DA) indicates the share of total assets which are funded with total deposits. As visible from Table 2, the mean and me- dian values of the ratio of both signicant and less signicant banks were higher after the SSM imple- mentation. This indicates increased overall reliance on total deposits for funding bank assets after the SSM implementation. The maximum value of the ratio increased for both groups of banks in the pe- riod after the SSM implementation, which indicates increased reliance of individual banks on total de- posits for funding bank assets. The minimum value of the ratio of signicant banks increased, whereas that of less signicant banks decreased in the period after the SSM implementation. This implies that there are individual banks within the group of signicant banks that decreased their reliance on total deposits for funding bank assets after the SSM implementa- tion. Furthermore, it implies that there are individual banks among the less signicant banks that did not use deposits as a funding source in the period after the SSM implementation. The ratio of interbank deposits to total assets (BDTA) indicates the share of total assets which are funded with interbank deposits. As visible from Ta- ble 2, both the mean and median values of the ratio for both groups of banks, signicant and less sig- nicant, were slightly lower in the period after the SSM implementation. However, the maximum value of the ratio in the period after the SSM implementa- tion slightly increased for both group of banks, while 1 The ECB developed the following signicance criteria for classifying banks as signicant: 1) bank size: total assets exceeding EUR 30 billion; 2) ratio of total assets to gross domestic product of the country in which the bank operates exceeding 20%; 3) the economic importance of the bank for the economy—one of the three largest banks in the domestic economy; 4) possible direct public nancial assistance; and 5) cross-border activities. 2 Banks in the sample kept their signicance status stable during the analysed period. ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 109 Table 1. Variables and sources of data. Variable Abbreviation Source Dependent variables Total deposits to total assets DA Fitch Connect database Interbank deposits to total assets BDTA Fitch Connect database Interbank deposits>5 years to total assets IDmore5yTA Fitch Connect database Interbank deposits 1–5 years to total assets ID1to5yTA Fitch Connect database Interbank deposits 3–12 months to total assets ID3to12mTA Fitch Connect database Interbank deposits<3 months to total assets IDless3mTA Fitch Connect database Independent variables Bank-specic variables Return on average assets ROAA Fitch Connect database Equity to total assets ETA Fitch Connect database Liquid assets to total assets LATA Fitch Connect database Macroeconomic variables Growth of the gross domestic product GDP World Economic Outlook database Unemployment UNE World Economic Outlook database Ination INF World Economic Outlook database Dummy variables Treated treated Dummy variable Signicant D 1 and Less Signicant D 0 Time time Dummy variable Before SSMD 0 and After SSMD 1 DID did Composite variable of the two dummy variables treated and time Source: Authors’ calculations. the minimum value remained unchanged (zero). This indicates that while the overall reliance on interbank deposits for funding decreased for both groups of banks in the period after the SSM implementation, there are individual banks within each group that substantially increased their reliance on interbank de- posits, and that there are individual banks that did not use interbank deposits at all in both periods. As visible from Table 2, both groups of banks had higher mean values of the ratio of return on average assets and the equity-to-total-assets ratio in the pe- riod after the SSM implementation. This indicates improved protability and bank capitalization of both group of banks after the SSM implementation. The mean value of the ratio of liquid assets to total assets of signicant banks slightly increased in the period after the SSM implementation, which implies im- proved liquidity of signicant banks. On the contrary, the mean value of the liquidity ratio of less signi- cant banks decreased after the SSM implementation, Table 2. Summary statistics of control and treatment group, before and after the implementation of the SSM. Before the SSM implementation After the SSM implementation Variables n mean SD median min max n mean SD median min max Treatment group (signicant banks) DA 374 42.79 21.17 44.05 0.17 91.65 591 50.28 21.85 53.66 0.25 94.59 BDTA 374 0.14 0.15 0.10 0.00 0.83 589 0.12 0.13 0.09 0.00 0.84 LnTA 374 11.64 1.46 11.45 8.09 14.75 591 11.55 1.44 11.37 7.96 14.74 ROAA 372 0.12 1.71 0.36 14.31 3.06 591 0.65 0.74 0.59 3.71 5.88 ETA 374 6.19 3.33 5.76 0.86 21.18 591 7.37 3.46 6.62 1.73 25.26 LATA 374 18.80 13.76 15.82 1.10 85.53 591 18.88 13.85 15.88 1.09 86.73 GDP 374 0.51 2.03 0.46 6.55 7.26 591 2.28 2.45 1.84 1.83 25.18 INF 374 2.25 0.89 2.29 0.01 5.08 591 0.90 0.85 0.71 1.54 3.65 UNE 374 10.56 5.55 9.22 4.88 26.09 591 9.52 4.49 9.44 3.40 24.44 Control group (less signicant banks) DA 476 46.88 22.85 49.91 0.03 95.23 846 50.98 23.64 55.16 0.00 96.71 BDTA 474 0.25 0.21 0.19 0.00 0.84 842 0.22 0.21 0.15 0.00 0.89 LnTA 476 8.92 1.10 9.29 4.77 10.29 846 8.92 1.05 9.21 5.32 10.31 ROAA 461 0.75 1.70 0.74 9.07 10.38 846 0.86 1.50 0.76 16.20 12.49 ETA 476 10.03 4.74 9.61 0.61 41.79 846 11.08 6.83 9.97 1.54 84.37 LATA 476 16.02 14.86 11.08 0.69 82.76 846 15.61 14.72 10.73 0.74 95.78 GDP 476 0.42 1.93 0.62 10.15 7.26 846 1.94 1.77 1.67 0.49 25.18 INF 476 2.20 0.86 2.29 0.85 5.08 846 0.86 0.82 0.62 1.39 3.65 UNE 476 10.81 5.24 9.77 4.88 27.48 846 10.33 4.30 10.05 3.40 26.50 Source: Authors’ calculations. 110 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 implying worsened liquidity of the banks. The macroeconomic variables indicate improved eco- nomic conditions in the period after the SSM imple- mentation, reected in increased economic growth and decreased ination and unemployment. 2.2 Methodology To investigate the effect of implementing the SSM on banks’ deposit structure we used the DID method, which is widely used in the literature for inspect- ing the effects of SSM on bank behaviour (Altunba¸ s et al., 2022; Alves et al., 2023; Avgeri et al., 2021; Avignone et al., 2021; Fiordelisi et al., 2017), as well as for inspecting effects of directives and regulations (Li & Marinˇ c, 2018; Pancotto et al., 2018). The DID estimator evaluates the impact of a treatment on out- come Y over a population. It requires a control group of population—the population that has not received the treatment—and treatment group—the population that has received the treatment. In our case, the treat- ment group is the signicant banks, which switched from national supervisors to the ECB, and the con- trol group is the less signicant banks, which have remained under supervision by their NSAs. We have used the following econometric model: Y it Da 0 Ca 1 time it Ca 2 treated it Ca 3 (time it treated it ) C a 4 B it Ca 5 M it C+ it (1) where the dependent variable Y it is one of the follow- ing variables measured at time t for bank i: 1) total deposits to total assets (DA) and 2) interbank deposits to total assets (BDTA). The DA and BDTA ratios rep- resent comparative metrics of the relative dependence of banks on deposits and interbank deposits as fund- ing sources. The DA ratio (DA) reects the portion of total assets which are funded with total deposits and indicates depositors’ trust (Koroleva et al., 2021). The BDTA ratio reects the portion of total assets which are funded with interbank deposits and indicates (in- terbank) depositors’ trust. The dummy variable time it indicates the period when the SSM was implemented (from 2014 on- wards), by taking values 1 for the period after the implementation of the SSM and 0 for the period before the implementation of the SSM. The dummy variable treated it takes the value of 1 for signicant banks that fall under direct supervision of the ECB or 0 for less signicant banks that fall under supervision by their NSAs. The coefcient of our interest is the composite variable time it treated it , which takes the value of 1 for directly supervised banks in the period after the implementation of SSM or 0 for the period before implementing the SSM regardless of the signicance of the banks. The slope of this composite variable in- dicates the effect of implementing the SSM on bank behaviours. If the slope is positive, the causal effect in our dependent variable will be positive, and vice versa. We ran each model three times. In the rst run we did not use any control variables, we ran the DID model on our dependent variables using the two dummy variables and the composite variable (did). In the second run, besides the two dummy variables and the composite variable (did), we added bank-specic variables to control for bank differences, consistent with the recent and expanding literature stream which analyses the effects of the SSM on the banking sector (Altunba¸ s et al., 2022; Avgeri et al., 2021; Fiordelisi et al., 2017). Signicant banks located in smaller economies are smaller and have total as- sets bellow EUR 30 billion, compared to signicant banks located in bigger economies. Therefore, we controlled for size differences in the sample by includ- ing the variable of natural logarithm of total assets (LnTA) in the model. We expected a positive rela- tion between the coefcient of the bank size variable and the dependent variable of total deposits to total assets. Larger banks hold more deposits compared to smaller banks (Kaufman, 1972; Valahzagharda & Kashb, 2014). Moreover, we used the variable of re- turn on average assets (ROAA) as a measure of banks’ protability, the ratio of equity to total assets (ETA) as a measure for banks’ capitalization, and the ratio of liquid assets to total assets (LATA) as a measure for banks’ liquidity. The protability of the banks is affected by the wideness of the loan–deposit interest spread (Chang et al., 2011). Therefore, the relation of the coefcient of the protability variable with the dependent variable is ambiguous. We expected a neg- ative relation of the coefcient of the capitalization variable with the dependent variable. This is because well capitalized banks are less dependent on external funding (Oura et al., 2013). We expected a positive relation of the coefcient of the liquidity variable with the dependent variable. This is because banks with more demand deposits should have more liquid as- sets relative to total assets (Kashyap et al., 2002). We expected a positive relation between the coef- cient of the bank size variable and the dependent vari- able of interbank deposits to total assets. Banks which are nanced with interbank deposits are monitored, therefore tending to engage in less risky lending ac- tivities and consequently being less risky (Dinger & Hagen, 2009). On the other hand, interbank deposits are not insured and pose a risk for the lender be- cause in case of bank bankruptcy, they are most likely to be lost (Furne, 2001). However, according to the “too big to fail” theory, large banks are systemically ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 111 important and their failure would inict serious dam- age to the overall economy and the banking sector. Consequently, governments assist them in times of difculties to prevent their default (Stern & Feldman, 2004). Therefore, we expected larger banks to be more engaged on the interbank market and to have more interbank deposits. In order to participate on the in- terbank market, banks should establish themselves as creditworthy institutions (Acharya et al., 2012; Allen et al., 2020). Since protability, capitalization, and liq- uidity positively affect the creditworthiness of banks, we expected a positive relation of the coefcients of these variables with the dependent variable. In the third run, besides the two dummy variables, the composite variable (did), and the bank-specic variables, we added the following macroeconomic variables to control for macroeconomic and country differences: ination (INF), growth of gross domes- tic product (GDP), and unemployment (UNE). A country’s economic and macroeconomic factors can affect depositors’ tendency to place money in the banking system. Growth of gross domestic product positively affects bank deposits—increase of income boosts savings and investment (Thao & Thanh, 2021; Valahzagharda & Kashb, 2014). Ination can have an adverse effect on deposits. When ination rises, deposits become less attractive due to a drop of real interest rates (Valahzagharda & Kashb, 2014). There- fore, we expected a positive relation of the coefcient of the ination variable and the coefcient of the variable of growth of gross domestic product with the dependent variable of total deposits to total as- sets. Unemployment can have an adverse effect on deposits—drops in income decrease savings (Thao & Thanh, 2021). Therefore, we expected a negative rela- tion of the coefcient of the unemployment variable with the dependent variable. Although the interbank market differs across coun- tries, the main reason for those differences is the country-specic trust in the banking sector (Allen et al., 2020). The economic cycle affects the level of economic activity and consequently the interbank transactions. Growth of gross domestic product can positively affect interbank deposits, since the in- crease of economic activity can boost demand for nancial services including interbank transactions. Unemployment can have an adverse effect on inter- bank deposits due to the decreased economic activity. Ination can positively affect interbank deposits—to counter ination, central banks increase interest rates, and consequently, interbank deposits become more attractive (Grandi & Guillet, 2021; Md-Yusuf & Md- Zain, 2020). Therefore, we expected a positive sign of the coefcients of ination and growth of gross do- mestic product and a negative sign of the coefcient of the unemployment variable with the dependent variable. In Equation (1) B it refers to a vector of bank-specic control variables, and M it refers to a vector of macroe- conomic variables. Each model was tested with the Hausman test to check whether a xed or random effects model was appropriate for the panel data. 3 Empirical results 3.1 Preliminary data inspection The DID model must satisfy the parallel trend assumption. This assumption requires that, in the absence of the treatment, the unobserved difference between the treatment and control groups be con- stant over time. If this assumption is not fullled, the results of the DID model might be biased. There is no statistical test for the parallel trend assumption. A visual inspection is the best way for verifying this assumption (Bertrand et al., 2004; Hill et al., 2018). Fig. 1 displays the visual inspection of the parallel trend of the dependent variables: 1) total deposits to total assets and 2) interbank deposits to total assets. Fig. 1 conrms that there is no differential trend be- tween the total deposits and the interbank deposits in the period before the implementation of the SSM. 3.2 Main results Table 3 and Table 4 display the results of the empir- ical analysis from estimating Equation (1) for the two dependent variables. We ran each model three times. In the rst run we did not use any control variables, we ran the DID model on our dependent variables using the two dummy variables time it and treated it and the composite variable did. Please note that the dummy variable time it indicates the period when the SSM was implemented (from 2014 onwards), by tak- ing values 1 for the period after the implementation of the SSM and 0 for the period before. The dummy vari- able treated it takes a value of 1 for signicant banks or 0 for less signicant banks. The composite variable did takes a value of 1 for directly supervised banks in the period after the implementation of SSM or 0 for the period before implementing the SSM regardless of the signicance of the banks. In the second run, besides the two dummy variables and the composite variable did, we controlled for bank differences by adding bank-specic control variables in the models. In the third run, besides the two dummy variables and the composite variable did, we controlled for both bank and country and macroeconomic differ- ences by adding bank-specic and macroeconomic control variables. Each model was tested with the 112 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Fig. 1. Visual inspection of the parallel trend assumption. Hausman test to check whether a xed or random effects model was appropriate for the panel data. We were interested in the statistical signicance of the coefcient a 3 , which represents the average difference in the dependent variable between the signicant banks (treatment group), which are supervised by the ECB, and less signicant banks (control group), which are supervised by NSAs. The coefcient of the variable treated it is the estimated mean difference in the dependent variable between the treatment and Table 3. Impact of SSM implementation on depositors’ trust. Model 1a Model 2a Model 3a Model 4a Model 5a Model 6a DA DA DA BDTA BDTA BDTA (Intercept) 46.351 (1.354) treated 2.388 2.106 1.855 C 0.004 0.025 0.026 (0.880) (0.993) (0.982) (0.008) (0.009) (0.009) time 4.838 4.996 4.387 0.031 0.025 0.022 (0.333) (0.333) (0.422) (0.003) (0.003) (0.004) did 1.453 1.158 0.916 C 0.016 0.019 0.019 (0.512) (0.513) (0.508) (0.005) (0.005) (0.005) LnTA 3.736 3.717 0.015 0.015 (0.752) (0.750) (0.007) (0.007) LATA 0.061 0.056 0.0005 0.0005 (0.024) (0.023) (0.0002) (0.0002) ROAA 0.395 0.232 C 0.004 0.003 (0.124) (0.126) (0.001) (0.001) ETA 0.408 0.403 0.003 0.003 (0.051) (0.051) (0.0005) (0.0005) GDP 0.157 0.0001 (0.076) (0.0007) UNE 0.482 0.002 C (0.085) (0.0008) INF 0.084 0.002 (0.168) (0.002) n 2287 2270 2270 2279 2262 2262 R 2 .181 .226 .247 .060 .115 .117 Hausman test $ 2 0.87854 40.533 698.5 27.34 111.37 33.529 df 3 7 10 3 7 10 p value .8386 9.95e 07 2.2e 16 4.99e 06 2.2e 16 .0002219 Note. This table displays the results from the difference-in-difference model. The dependent variables are total deposits to total assets (DA) and interbank deposits to total assets (BDTA). Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 113 Table 4. Impact of comprehensive assessment and SSM launch on depositors’ trust. Model 1b Model 2b Model 3b Model 4b Model 5b Model 6b DA DA DA BDTA BDTA BDTA (Intercept) 46.256 98.976 (1.370) (5.605) treated 3.978 2.560 1.718 C 0.002 0.025 0.025 (0.946) (1.059) (1.007) (0.009) (0.009) (0.009) time 4.579 5.045 5.523 0.026 0.022 0.016 (0.382) (0.382) (0.491) (0.003) (0.003) (0.005) did 2.284 1.620 1.539 0.010 0.015 0.015 (0.581) (0.580) (0.562) (0.005) (0.005) (0.005) LnTA 5.271 4.694 0.020 0.016 (0.759) (0.537) (0.007) (0.007) LATA 0.054 0.039 C 0.0005 0.0005 (0.024) (0.023) (0.0002) (0.0002) ROAA 0.517 0.248 0.004 0.003 (0.125) (0.125) (0.001) (0.001) ETA 0.460 0.463 0.003 0.003 (0.052) (0.047) (0.0005) (0.0005) GDP 0.418 0.0008 (0.073) (0.0007) UNE 0.483 0.002 (0.078) (0.0008) INF 0.502 0.003 C (0.183) (0.002) n 2287 2270 2270 2279 2262 2262 R 2 .154 .208 .241 .036 .101 .107 Hausman test $ 2 0.38477 48.813 0.47018 29.141 198.111 32.22 df 3 7 10 3 7 10 p value .9434 2.469e 08 1 2.091e 06 2.22e 16 .000368 Note. This table displays the results from the difference-in-difference model. The dependent variables are total deposits to total assets (DA) and interbank deposits to total assets (BDTA). Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. control groups prior the treatment. It shows the dif- ferences that existed between the groups of signicant and less signicant banks before the treatment period (SSM implementation). In the group of less signicant banks (control group), the expected mean change in the dependent variable after the implementation of the SSM corresponds to the coefcient of the variable time it . In the group of signicant banks (treatment group), the expected mean change in the dependent variable after the implementation of the SSM is the sum of the coefcients of time it and did. 3.2.1 Impact of SSM implementation on depositors’ trust Table 3 displays the results of the empirical analysis from estimating Equation (1) for inspecting the im- pact of the SSM implementation on deposit structure. In Models 1a, 2a, and 3a, we inspected the impact of the SSM implementation on depositors’ trust mea- sured via the dependent variable of total deposits to total assets. In Model 1a we observed a positive and statistically signicant effect on depositors’ trust in the signicant banks compared to the less signicant banks. The coefcient of the composite variable did shows that the expected mean change in deposits to total assets from before to after the implementation of the SSM is different in the control and treatment groups. The statistical signicance of the coefcient of the variable did was not affected by adding control variables; however, we observed small differences in the estimated coefcients between the models. Model 2a reports the results of the DID model using the two dummy variables, time and treated, and the composite variable did together with bank- specic variables. We have observed a positive and statistically signicant effect on depositors’ trust and statistically signicant bank-specic variables. Specif- ically, we have observed a positive relation of the coefcients of liquidity and protability with the de- pendent variable, and an inverse relation of bank size and bank capitalization with the dependent 114 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 variable. In Model 3a we show the results of the DID model using the two dummy variables, time and treated, and the composite variable did together with bank-specic and macroeconomic variables. Again, we have observed a positive and statistically sig- nicant effect on depositors’ trust (at a statistical level of 10%), statistically signicant bank-specic variables and statistically signicant macroeconomic variables, with the exception of the coefcient of ination, which is statistically insignicant. Specif- ically, we have observed a positive relation of the coefcient of growth of gross domestic product with the dependent variable and an inverse relation of the coefcient of unemployment with the dependent variable. In Models 4a, 5a, and 6a, we have inspected the impact of the SSM implementation on (interbank) depositors’ trust measured via the dependent vari- able of interbank deposits to total assets. In Model 4a we have observed a positive and statistically sig- nicant effect on (interbank) depositors’ trust in the signicant banks compared to the less signicant banks. The coefcient of the composite variable did shows that the expected mean change in interbank deposits to total assets from before to after the im- plementation of the SSM is different in the control and treatment groups. The coefcient of the variable treated corresponds to the estimated mean difference in interbank deposits to total assets between the treatment and control groups prior the treatment. It shows the differences that existed between the groups before implementing the SSM. It is statistically in- signicant, showing no difference existed between the signicant and less signicant banks before the SSM implementation. The statistical signicance of the coefcient of the variable did is not affected by adding control variables; however, we have observed small differences in the estimated coefcients be- tween the models. Model 5a reports the results of the DID model using the two dummy variables, time and treated, and the composite variable did together with bank-specic variables. We have observed a pos- itive and statistically signicant effect on (interbank) depositors’ trust and statistically signicant bank- specic variables. Specically, we have observed an inverse relation of the coefcients of liquidity and protability and bank capitalization with the depen- dent variable and a positive relation of bank size with the dependent variable. In Model 6a we show the results of the DID model using the two dummy variables, time and treated, the composite variable did together with bank-specic and macroeconomic vari- ables. In this model, we have observed a positive and statistically signicant effect on (interbank) deposi- tors’ trust, statistically signicant bank-specic vari- ables and statistically insignicant macroeconomic variables. The positive and statistically signicant effect on both the total-deposits-to-total-assets and interbank- deposits-to-total-assets ratios of the signicant banks compared to the less signicant banks implies that changes in the deposit structure are associated with the SSM implementation. These results indicate that the SSM implementation has positively affected de- positors’ trust in the signicant banks, measured via both the total-deposits-to-total-assets and interbank- deposits-to-total-assets ratios. This implies improved trustworthiness and credibility of the banks super- vised by the ECB. 3.2.2 Impact of comprehensive assessment and SSM launch on depositors’ trust Table 4 displays the results of the empirical anal- ysis from estimating Equation (1) for inspecting the impact of the comprehensive assessment and SSM launch on deposit structure. In Models 1b, 2b, and 3b, we have inspected the impact of the comprehensive assessment and SSM launch on depositors’ trust mea- sured via the dependent variable of total deposits to total assets. In Model 1b we have observed a positive and statistically signicant effect on depositors’ trust in the signicant banks, which were expected to be supervised by the ECB, compared to the less signif- icant banks, which were expected to remain under NSAs’ supervision. The coefcient of the composite variable did shows that the expected mean change in total deposits to total assets from before to after the treatment period is different in the control (less signicant banks) and treatment (signicant banks) groups. The statistical signicance of the coefcient of the variable did was not affected by adding control variables; however, we observed small differences in the estimated coefcients between the models. Model 2b reports the results of the DID model using the two dummy variables, time and treated, and the composite variable did together with bank-specic variables. We have observed a positive and statistically signicant effect on depositors’ trust, and statistically signif- icant bank-specic variables. Specically, we have observed a positive relation of the coefcients of liq- uidity and protability with the dependent variable and an inverse relation of bank size and bank capi- talization with the dependent variable. In Model 3b we show the results of the DID model using the two dummy variables, time and treated, the com- posite variable did together with bank-specic and macroeconomic variables. We have observed a posi- tive and statistically signicant effect on depositors’ trust, statistically signicant bank-specic variables and statistically signicant macroeconomic variables. ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 115 Specically, we have observed a positive relation of the coefcients of growth of gross domestic prod- uct and ination with the dependent variable and an inverse relation of the coefcient of unemployment with the dependent variable. In Models 4b, 5b, and 6b, we have inspected the impact of the comprehensive assessment and SSM launch on (interbank) depositors’ trust measured via the dependent variable of interbank deposits to to- tal assets. In Model 4b we have observed a positive and statistically signicant effect on (interbank) de- positors’ trust in the signicant banks, which were expected to be supervised by ECB, compared to the less signicant banks, which were expected to re- main under NSAs’ supervision. The coefcient of the composite variable did shows that the expected mean change in interbank deposits to total assets from be- fore to after the treatment effect is different in the control (less signicant banks) and treatment (signif- icant banks) groups. The coefcient of the variable treated corresponds to the estimated mean difference in interbank deposits to total assets between the treat- ment and control groups prior to the treatment. It is statistically insignicant, showing no difference existed between the signicant and less signicant banks in the period before the comprehensive assess- ment and SSM launch. The statistical signicance of the coefcient of the variable did was not affected by adding control variables; however, we observed small differences in the estimated coefcients between the models. Model 5b reports the results of the DID model using the two dummy variables, time and treated, and the composite variable did together with bank-specic variables. We have observed a positive and statisti- cally signicant effect on (interbank) depositors’ trust, and statistically signicant bank-specic variables. Specically, we have observed an inverse relation of the coefcients of liquidity and protability and bank capitalization with the dependent variable, and a positive relation of bank size with the dependent variable. In Model 6b we show the results of the DID model using the two dummy variables, time and treated, and the composite variable did together with bank-specic and macroeconomic variables. In this model, we have observed a positive and sta- tistically signicant effect on (interbank) depositors’ trust, statistically signicant bank-specic variables and statistically signicant macroeconomic variables, with the exception of the coefcient of gross domestic product, which is statistically insignicant. Speci- cally, we have observed a positive relation of the coefcients of unemployment and ination (statisti- cally signicant at 10%) with the dependent variable. The positive and statistically signicant effect on both the total-deposits-to-total-assets and interbank- deposits-to-total-assets ratios of the signicant banks, which were expected to be supervised by the ECB, compared to the less signicant banks, which were expected to remain under NSAs’ supervision, implies that changes in the deposit structure are associated with the anticipation of the SSM launch and the ex- pected comprehensive assessment. This implies that the ECB was perceived as a stricter supervisory au- thority compared to the NSAs. Consequently, banks which were going to be assessed with the compre- hensive assessment and which were going to be supervised by the ECB were considered safer due to stricter supervision and consequently encountered in- creased depositors’ trust. 3.3 Robustness checks 3.3.1 Placebo test: changing the year of the SSM implementation In order to investigate for other factors that might have affected depositors’ trust in signicant banks compared to less signicant banks, before the im- plementation of SSM, we performed this robustness check, where we created a ctional time dummy vari- able assuming that the SSM had been implemented in 2012. Please note the dummy variable time indicates the ctional period when the SSM was implemented (from 2012 onwards), by taking values 1 for the period after the ctional implementation of the SSM and 0 for the period before. Here we have examined whether in the period up to the SSM implementation, the signi- cant banks, which were expected to be supervised by the ECB, encountered changes in their deposit struc- ture compared to the less signicant banks. Table 5 displays the results of our robustness check with the ctional time dummy variable assuming that the SSM was implemented in 2012. From Table 5 (Models 1c, 2c, and 3c) it is evident that there is a statistically signicant effect on the total-deposits-to- total-assets ratio of signicant banks compared to less signicant banks. This result points to the existence of other factors which affected the total-deposits-to- total-assets ratio of the signicant banks compared to the less signicant banks in the period before the SSM implementation (2012). These results do not support our claim that the increase of depositors’ trust in the signicant banks is associated with the SSM imple- mentation, but rather imply that it is associated with other past events. Table 5 (Models 4c, 5c, and 6c) shows that there is no evidence of a statistically signicant effect on the interbank-deposits-to-total-assets ratio in 2012. This implies that there were no differences in the interbank deposits structure of signicant banks com- pared to less signicant banks, assuming the SSM was 116 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Table 5. Placebo test: changing the year of the implementation of the SSM (2012). Model 1c Model 2c Model 3c Model 4c Model 5c Model 6c DA DA DA BDTA BDTA BDTA (Intercept) 46.339 98.553 (1.421) (5.642) treated 5.112 3.151 2.362 0.009 0.020 C 0.020 C (1.121) (1.231) (1.141) (0.010) (0.011) (0.010) time 4.050 4.604 4.418 0.024 0.022 0.019 (0.528) (0.527) (0.560) (0.005) (0.005) (0.005) did 2.745 2.008 2.104 0.003 0.009 0.008 (0.797) (0.789) (0.745) (0.007) (0.007) (0.007) LnTA 5.925 4.403 0.019 0.014 (0.792) (0.540) (0.007) (0.007) LATA 0.049 0.034 0.0005 0.0004 (0.025) (0.023) (0.0002) (0.0002) ROAA 0.651 0.156 0.004 0.003 (0.131) (0.127) (0.001) (0.001) ETA 0.434 0.417 0.003 0.003 (0.055) (0.047) (0.0005) (0.0005) GDP 0.697 0.001 (0.076) (0.0007) UNE 0.639 0.002 (0.078) (0.0007) INF 0.457 0.003 (0.152) (0.001) n 2287 2270 2270 2279 2262 2262 R 2 .083 .131 .218 .023 .095 .107 Hausman test $ 2 0.10721 56.489 14.072 30.141 40.554 40.175 df 3 7 10 3 7 10 p value .991 7.551e 10 .1697 1.289e 06 9.86e 07 1.579e 05 Note. This table displays the results from the difference-in-difference model. The dependent variables are total deposits to total assets (DA) and interbank deposits to total assets (BDTA). Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. implemented in 2012. These results support our claim that the increase of (interbank) depositors’ trust of signicant banks is associated with the SSM imple- mentation rather than any other previous events. 3.3.2 Robustness check: countries outside the euro area In order to inspect if there had been other factors that affected depositors’ trust in Europe in 2014 be- sides the implementation of the SSM, we performed another robustness check. We chose banks located in European countries which are not part of the euro area and of the SSM (Poland, Denmark, Hungary, Czech Republic, Romania). We divided the banks in two groups, signicant and less signicant, according to their size. If the total assets of a bank exceeded 30 billion EUR, it was classied as signicant. The sample consisted of 46 banks, out of which 11 were classied as signicant and 35 were classied as less signicant. The dataset covers the period 2018–2021 and has the same bank-specic and macroeconomic variables. Table 6 displays the results of this robust- ness check. In Table 6 all three models with the dependent vari- able of total deposits to total assets (Models R1, R2, and R3) have statistically signicant coefcients of the composite variable did. These results do not support our claim that the increase of the depositors’ trust in signicant banks is associated with the SSM im- plementation, but rather imply the existence of other factors that have affected the deposit structure of Eu- ropean banks apart from the implementation of the SSM. From Table 6 it is clear that in all three models with the dependent variable of interbank deposits to to- tal assets (Models R4, R5, and R6), the coefcient of the composite variable did is statistically insignicant. This result supports our claim that the increase of the (interbank) depositors’ trust in the signicant banks is associated with the SSM implementation and that there are no other factors that could have affected the ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 117 Table 6. Robustness check: countries outside the Euro area. Model R1 Model R2 Model R3 Model R4 Model R5 Model R6 DA DA DA BDTA BDTA BDTA (Intercept) 62.813 270.048 0.099 0.104 0.233 (2.152) (25.307) (0.010) (0.157) (0.149) treated 0.170 4.871 3.495 C 0.030 0.044 0.037 (2.178) (2.126) (1.985) (0.015) (0.016) (0.016) time 5.378 7.439 1.910 0.037 0.039 0.005 (0.731) (0.749) (1.251) (0.006) (0.006) (0.010) did 6.134 6.647 6.511 0.017 0.016 0.015 (1.553) (1.430) (1.371) (0.012) (0.012) (0.012) LnTA 8.379 8.224 0.010 0.011 C (1.506) (1.059) (0.007) (0.006) LATA 0.198 0.254 0.001 0.001 (0.048) (0.043) (0.0003) (0.0003) ROAA 0.680 0.591 0.006 0.005 (0.272) (0.264) (0.002) (0.002) ETA 0.168 0.094 0.005 0.004 (0.204) (0.190) (0.002) (0.001) GDP 0.146 0.00005 (0.245) (0.002) UNE 1.425 0.009 (0.249) (0.002) INF 0.572 0.004 (0.233) (0.002) n 361 358 358 361 358 358 R 2 .140 .301 .386 .124 .237 .281 Hausman test $ 2 6.1152 24.281 6.6238 2.7015 11.598 59.855 df 3 7 10 3 7 10 p value .1061 .001017 .7604 .44 .1146 3.861e 09 Note. This table displays the results from the difference-in-difference model. The dependent variables are total deposits to total assets (DA) and interbank deposits to total assets (BDTA). Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. interbank deposits structure of the European banks apart from the implementation of the SSM. 3.3.3 Subsample: removing France from the dataset With this robustness check we tested our results for sample selection bias. Since most of the banks in our dataset were from France (98 banks, out of which 31 are signicant and 67 are less signicant), we removed them from our sample. The subsample without France resulted in 192 banks, out of which 90 are signicant and 102 are less signicant. Table 7 displays the results of this robustness check performed on the dependent variable of total deposits to total assets. In Models 1, 2, and 3 (Table 7), we have inspected the impact of the SSM implementation on depositors’ trust measured via the dependent vari- able of total deposits to total assets. These results, with the exception of Model 1, which has a positive and statistically signicant coefcient of the did variable, do not support our claim that changes in the deposit structure of signicant banks are associated with the SSM implementation. In Models 4, 5, and 6 (Table 7), we have inspected the impact of the comprehensive assessment and SSM launch on depositors’ trust measured via the depen- dent variable of total deposits to total assets. All models have positive and statistically signicant co- efcients of the did variable. These results support our claim about depositors’ trust in the signicant banks compared to the less signicant banks in anticipation of the SSM and the comprehensive assessment and further conrm absence of sample selection bias. In Models 7, 8, and 9 (Table 7) we have performed a placebo test by creating a ctional time dummy vari- able assuming that the SSM was implemented in 2012. Please note the dummy variable time indicates the c- tional period when the SSM was implemented (from 2012 onwards), by taking values 1 for the period after the ctional implementation of the SSM and 0 for the period before. Here we have examined whether in the 118 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Table 7. Subsample: removing France from the dataset. Dependent variable total deposits to total assets. Impact of SSM implementation Impact of CA and SSM launch Placebo test (2012) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 (Intercept) 52.677 53.010 53.801 (1.613) (1.645) (1.735) treated 4.533 2.065 1.712 6.901 3.122 2.671 C 8.970 4.355 4.120 (1.194) (1.343) (1.332) (1.290) (1.431) (1.387) (1.532) (1.662) (1.564) time 5.760 5.666 5.181 5.144 5.557 5.994 3.879 4.634 4.033 (0.474) (0.467) (0.624) (0.548) (0.532) (0.691) (0.762) (0.733) (0.777) did 1.371 0.748 0.630 2.581 1.499 1.459 3.805 2.457 2.677 (0.683) (0.674) (0.668) (0.782) (0.761) (0.737) (1.080) (1.039) (0.976) LnTA 7.878 7.912 9.565 9.335 10.475 8.966 (0.949) (0.941) (0.950) (0.935) (0.992) (0.948) LATA 0.086 0.083 0.091 0.083 0.100 0.089 (0.029) (0.028) (0.029) (0.028) (0.030) (0.029) ROAA 0.530 0.404 0.691 0.436 0.821 0.347 (0.139) (0.141) (0.140) (0.139) (0.146) (0.142) ETA 0.673 0.681 0.752 0.776 0.728 0.727 (0.074) (0.073) (0.075) (0.073) (0.078) (0.074) GDP 0.093 0.313 0.541 (0.086) (0.081) (0.086) UNE 0.413 0.514 0.701 (0.094) (0.088) (0.087) INF 0.079 0.545 0.461 (0.214) (0.228) (0.189) n 1508 1497 1497 1508 1497 1497 1508 1497 1497 R 2 .209 .284 .301 .173 .269 .319 .093 .195 .293 Hausman test $ 2 0.40558 55.73 23.48 0.59059 58.592 78.006 0.4988 71.201 80.917 df 3 7 10 3 7 10 3 7 10 p value .9391 1.068e 09 .009108 .8986 2.882e 10 1.233e 12 .9192 8.445e 13 3.319e 13 Note. This table displays the results from the difference-in-difference model. The dependent variable is total deposits to total assets (DA). CAD comprehensive assessment. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. period up to the SSM implementation, the signicant banks, which were expected to be supervised by the ECB, encountered changes in their deposit structure compared to the less signicant banks. All models have positive and statistically signicant coefcients of the did variable. These results conrm the existence of other factors which affected the deposit structure of signicant banks compared to less signicant banks in the period before the SSM implementation (2012). These results also conrm absence of sample selection bias. Table 8 displays the results of this robustness check performed on the dependent variable of interbank deposits to total assets. In Models 1, 2, and 3 (Table 8), we have inspected the impact of the SSM implemen- tation on (interbank) depositors’ trust measured via the dependent variable of interbank deposits to total assets. All models have positive and statistically sig- nicant coefcients of the did variable. These results further support our claim that the implementation of the SSM has led to increased (interbank) depositors’ trust in signicant banks compared to less signicant banks and conrm the absence of sample selection bias. In Models 4, 5, and 6 (Table 8), we have inspected the impact of the comprehensive assessment and SSM launch on (interbank) depositors’ trust measured via the dependent variable of interbank deposits to total assets. The statistical signicance of the coefcient of the variable did in all models implies that in 2013 in anticipation of the SSM and the expected stress test exercise under the comprehensive assessment, the signicant banks, which were expected to be su- pervised by ECB, encountered increased (interbank) depositors’ trust, compared to the less signicant banks. This result further supports our claim that banks which were going to be assessed with the com- prehensive assessment and which were going to be supervised by the ECB were considered safer due to stricter supervision and consequently encountered ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 119 Table 8. Subsample: removing France from the dataset. Dependent variable: interbank deposits to total assets. Impact of SSM implementation Impact of CA and SSM launch Placebo test (2012) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 (Intercept) 0.151 0.107 0.149 0.087 C 0.145 0.095 (0.009) (0.047) (0.010) (0.046) (0.011) (0.046) treated 0.026 0.033 0.042 0.020 C 0.034 0.042 0.008 0.025 C 0.033 (0.010) (0.012) (0.011) (0.011) (0.013) (0.012) (0.012) (0.014) (0.013) time 0.042 0.034 0.031 0.034 0.030 0.019 0.026 0.026 0.018 (0.004) (0.004) (0.006) (0.005) (0.005) (0.006) (0.007) (0.006) (0.007) did 0.025 0.028 0.027 0.018 0.024 0.022 0.004 0.013 0.010 (0.006) (0.006) (0.006) (0.007) (0.007) (0.007) (0.009) (0.009) (0.009) LnTA 0.016 C 0.007 0.022 0.007 C 0.022 0.007 (0.009) (0.004) (0.009) (0.004) (0.009) (0.004) LATA 0.0008 0.0005 0.0009 0.0005 0.0009 0.0005 (0.0003) (0.0002) (0.0003) (0.0002) (0.0003) (0.0002) ROAA 0.005 0.005 0.005 0.004 0.005 0.004 (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) ETA 0.003 0.003 0.003 0.003 0.003 0.003 (0.0007) (0.0006) (0.0007) (0.0006) (0.0007) (0.0006) GDP 0.0002 0.0006 0.001 (0.0008) (0.0007) (0.0008) UNE 0.001 0.002 0.002 (0.0007) (0.0007) (0.0007) INF 0.002 0.004 0.004 (0.002) (0.002) (0.002) n 1500 1489 1489 1500 1489 1489 1500 1489 1489 R 2 .069 .134 .120 .039 .116 .107 .019 .104 .104 Hausman test $ 2 31.985 14.368 14.257 3.2894 18.021 17.487 0.85822 18.536 16.328 df 3 7 10 3 7 10 3 7 10 p value .362 .045 .1616 .3491 .01187 .06427 .8355 .009774 .09061 Note. This table displays the results from the difference-in-difference model. The dependent variable is interbank deposits to total assets (BDTA). CAD comprehensive assessment. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. increased depositors’ trust. These results also conrm absence of sample selection bias. In Models 7, 8, and 9 (Table 8), we have performed a placebo test by creating a ctional time dummy vari- able assuming that the SSM was implemented in 2012. Here we have examined whether in the period up to the SSM implementation, the signicant banks, which were expected to be supervised by ECB, encountered changes in the share of interbank deposits compared to the less signicant banks. The statistical insigni- cance of the coefcient of the variable did in all models implies that there were no differences in the share of the interbank deposits of signicant banks compared to less signicant banks in 2012. These results fur- ther support our claim that the (interbank) depositors’ trust in the signicant banks is associated with the SSM implementation rather than any other previous events and conrm the absence of sample selection bias. 3.3.4 Robustness check: xed rate on the main renancing operations Changes in the policy rates affect liquidity hold- ings and liquidity transfers on the interbank market (Näther, 2019). One of the three key interest rates used by the ECB for controlling the money supply and the overall liquidity in the banking sector is the rate on the main renancing operations. This rate is mainly used for short-term lending from the ECB at times of temporary liquidity shortages (ECB, 2011). Therefore, to inspect for effects of money supply and overall liquidity on the interbank deposits, we performed a robustness check where we added the xed interest rate on main renancing operations (MROF) as a con- trol variable in the model. Table 9 displays the results of this robustness check. In Model IR we show the estimated coefcients of our model where we included the xed interest rate of main renancing operations (MROF) as a control 120 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Table 9. Robustness check: xed rate on the main renancing operations. Model 4a Model 5a Model 6a Model IR BDTA BDTA BDTA BDTA treated 0.004 0.025 0.026 0.026 (0.008) (0.009) (0.009) (0.009) time 0.031 0.025 0.022 0.014 (0.003) (0.003) (0.004) (0.005) did 0.016 0.019 0.019 0.020 (0.005) (0.005) (0.005) (0.005) LnTA 0.015 0.015 0.018 (0.007) (0.007) (0.007) LATA 0.0005 0.0005 0.000 (0.0002) (0.0002) (0.000) ROAA 0.004 0.003 0.003 (0.001) (0.001) (0.001) ETA 0.003 0.003 0.003 (0.0005) (0.0005) (0.000) GDP 0.0001 0.001 (0.0007) (0.001) UNE 0.002 C 0.001 C (0.0008) (0.001) INF 0.002 0.000 (0.002) (0.002) MROF 0.013 (0.006) n 2279 2262 2262 2262 R 2 .060 .115 .117 .119 Hausman test $ 2 27.34 111.37 33.529 34.244 df 3 7 10 11 p value 4.99e 06 2.2e 16 .0002219 .0003299 Note. This table displays the results from the difference-in-difference model. The dependent variable is interbank deposits to total assets (BDTA). DID coefcient rounded to four decimals is: Model 6a: 0.0192; Model IR: 0.0196. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. variable. For comparison, in the same table we also present the results of our main models (Models 4a, 5a, 6a). As visible from Table 9, adding the xed interest rate of main renancing operations as a con- trol variable in the model has had a minimal effect on some coefcients and their statistical signicance. Moreover, we have observed a positive and statisti- cally signicant relation of the MROF variable with the dependent variable. This indicates that an increase of the xed rate on the main renancing operations is associated with an increase of the interbank-deposits- to-total-assets ratio for both groups of banks, signi- cant and less signicant, all else being equal. In other words, at times of an increased rate on the main re- nancing operations, banks have a higher portion of total assets funded with interbank deposits. As regards the coefcient of the did variable, we have observed a positive and statistically signicant effect. However, there is a minor difference in the estimated coefcient of the variable did in Model IR, where we added MROF as a control variable (0.0196), in com- parison to the did coefcient estimated in our main model (0.0192; Model 6a, Tables 3 and 9), whereas the statistical signicance of the coefcient is not af- fected. These results indicate that adding the MROF as a control variable in the model has a minimal effect on the did coefcient (a change in the did co- efcient of 0.0004), therefore further supporting our claim that the increased (interbank) depositors’ trust in the signicant banks is rather associated with the SSM implementation than changes in the monetary policy. 3.4 Additional analysis: interbank deposits by maturity We performed an additional analysis in order to in- spect both the impact of the SSM implementation and the impact of the comprehensive assessment and SSM launch on the share of the interbank deposits of sig- nicant banks in comparison to less signicant banks, by using alternative interbank deposit variables as de- pendent variables: interbank deposits with a maturity of less than 3 months to total assets (IDless3mTA), interbank deposits with a maturity of 3–12 months to total assets (ID3to12mTA), interbank deposits with a maturity of 1–5 years to total assets (ID1to5yTA), and interbank deposits with a maturity of more than 5 years to total assets (IDmore5yTA). For this anal- ysis, we had to remove from our sample all banks with missing data for the dependent variables (listed above). This subsample is composed of 115 banks, out of which 31 are signicant and 84 are less signicant, and covers the period 2011–2018. The results from this additional analysis are shown in Tables 10, 11, 12 and 13. In each table, Models 1, 2, and 3 display the impact of the SSM implementation on banks’ deposit structure, Models 4, 5, and 6 display the impact of the comprehensive assessment and SSM launch on banks’ deposit structure, and Models 7, 8, and 9 display the results from the placebo test, where, by creating a ctional time dummy variable, we as- sumed that the SSM had been implemented in 2012. Table 10 displays the results of this analysis per- formed on the dependent variable of interbank de- posits with a maturity of less than 3 months to total assets. Models 1, 2, and 3 display the impact of the SSM implementation on the share of interbank deposits. Results show a positive and statistically signicant coefcient of the composite variable did in Model 1, implying that signicant banks had increased the portion of total assets funded with inter- bank deposits with a maturity of less than 3 months after the SSM implementation, compared to the less ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 121 Table 10. Interbank deposits by maturity. Dependent variable: interbank deposits of less than 3 months to total assets. Impact of SSM implementation Impact of CA and SSM launch Placebo test (2012) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 (Intercept) 0.093 0.098 0.135 0.097 (0.005) (0.006) (0.093) (0.006) treated 0.027 0.014 0.015 0.027 0.015 0.030 0.010 0.004 0.001 (0.008) (0.011) (0.011) (0.009) (0.011) (0.010) (0.011) (0.013) (0.013) time 0.029 0.026 0.025 0.030 0.025 0.026 0.024 0.018 0.017 (0.003) (0.003) (0.004) (0.003) (0.004) (0.005) (0.005) (0.005) (0.006) did 0.016 0.017 0.019 0.010 0.013 C 0.012 C 0.011 0.007 0.006 (0.006) (0.006) (0.006) (0.007) (0.007) (0.007) (0.010) (0.010) (0.010) LnTA 0.000 0.001 0.007 0.001 0.018 0.008 (0.008) (0.009) (0.008) (0.004) (0.008) (0.009) LATA 0.000 0.000 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ROAA 0.001 0.001 0.001 0.001 0.001 0.002 (0.002) (0.002) (0.002) (0.002) (0.002) (0.002) ETA 0.003 0.003 0.004 0.003 0.004 0.004 (0.001) (0.001) (0.001) (0.000) (0.001) (0.001) GDP 0.001 0.001 0.004 (0.001) (0.001) (0.002) UNE 0.002 0.002 0.002 (0.002) (0.001) (0.002) INF 0.002 0.001 0.005 (0.002) (0.002) (0.002) n 852 852 852 852 852 852 852 852 852 R 2 .106 .175 .178 .092 .162 .140 .049 .127 .154 Hausman test $ 2 46.339 23.469 26.183 3.724 23.207 17.746 0.71718 30.202 28.528 df 3 7 10 3 7 10 3 7 10 p value .2007 .001412 .003502 .2928 .001568 .05941 .8692 8.72e 05 .001485 Note. This table displays the results from the difference-in-difference model. The dependent variable is interbank deposits of less than 3 months to total assets (IDless3mTA). CAD comprehensive assessment. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. signicant banks. The statistical signicance and the sign of the coefcient of the did variable have not been affected by adding control variables in the model (Models 2 and 3). These results are consistent with our claim that the implementation of the SSM has led to increased (interbank) depositors’ trust in the signicant banks compared to the less signicant banks. Models 4, 5, and 6 (Table 10) display the impact of the comprehensive assessment and SSM launch on the share of the interbank deposits. Results show positive and statistically signicant coefcients (at a signicance level of 10%) of the variable did in Mod- els 5 and 6 (Table 10) and a statistically insignicant coefcient in Model 4 (Table 10). These results im- ply that signicant banks, which were expected to be supervised by the ECB, had increased the portion of total assets funded with interbank deposits with a maturity less than 3 months compared to less sig- nicant banks, which were expected to remain under NSAs’ supervision, in anticipation of the SSM and the comprehensive assessment. Models 7, 8, and 9 (Table 10) display the results from the placebo test where, by creating a ctional time dummy variable, we assumed that the SSM was implemented in 2012. These models in Table 10 show that there is no evidence of a statistically signicant effect on the interbank deposits with a maturity of less than 3 months to total assets. This implies that there were no differences in the portion of total assets funded with interbank deposits with a maturity of less than 3 months between signicant and less sig- nicant banks, assuming the SSM was implemented in 2012. These results imply that the increased portion of total assets funded with interbank deposits with a maturity of less than 3 months in signicant banks is associated with the SSM implementation rather than any other past events. 122 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Table 11. Interbank deposits by maturity. Dependent variable: interbank deposits with a maturity of 3–12 months to total assets. Impact of SSM implementation Impact of CA and SSM launch Placebo test (2012) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 treated 0.002 0.008 0.009 0.002 0.002 0.005 0.003 0.004 0.001 (0.006) (0.007) (0.006) (0.006) (0.007) (0.007) (0.013) (0.013) (0.013) time 0.005 0.005 0.001 0.009 0.010 0.003 0.024 0.018 0.017 (0.002) (0.002) (0.003) (0.002) (0.002) (0.003) (0.005) (0.005) (0.006) did 0.006 0.005 0.005 0.010 0.010 0.009 0.009 0.007 0.006 (0.004) (0.004) (0.004) (0.004) (0.004) (0.004) (0.010) (0.010) (0.010) LnTA 0.007 0.014 0.005 0.011 0.018 0.008 (0.005) (0.005) (0.005) (0.005) (0.008) (0.009) LATA 0.000 0.000 C 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ROAA 0.003 0.003 0.003 0.003 0.001 0.002 (0.001) (0.001) (0.001) (0.001) (0.002) (0.002) ETA 0.001 0.000 0.001 C 0.000 0.004 0.004 (0.000) (0.000) (0.000) (0.000) (0.001) (0.001) GDP 0.000 0.000 0.004 (0.001) (0.001) (0.002) UNE 0.001 0.001 0.002 (0.001) (0.001) (0.002) INF 0.005 0.004 0.005 (0.001) (0.001) (0.002) n 825 825 825 825 825 825 852 852 852 R 2 .011 .041 .076 .027 .059 .079 .049 .127 .154 Hausman test $ 2 10.764 79.152 91.022 10.991 85.661 90.252 11.232 72.944 84.741 df 3 7 10 3 7 10 3 7 10 p value .01308 2.051e 14 3.358e 15 .01177 9.601e 16 4.774e 15 .01053 3.748e 13 5.872e 14 Note. This table displays the results from the difference-in-difference model. The dependent variable is interbank deposits of 3–12 months to total assets (ID3to12mTA). CAD comprehensive assessment. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. Table 11 displays the results of this analysis per- formed on the dependent variable of interbank de- posits with a maturity of 3–12 months to total assets. From Models 1, 2, and 3, which display the impact of the SSM implementation on the share of the in- terbank deposits, it is evident that the coefcient of the variable did is statistically insignicant. This im- plies that there were no differences in the portion of total assets funded with interbank deposits with a maturity of 3–12 months between signicant and less signicant banks which could be associated with the SSM implementation. Models 4, 5, and 6 (Table 11) display the impact of the comprehensive assess- ment and SSM launch on the share of the interbank deposits. Results show a negative and statistically signicant coefcient of the composite variable did in Model 4, implying that signicant banks, which were expected to be supervised by ECB, had de- creased their portions of total assets funded with interbank deposits with a maturity of 3–12 months in anticipation of the comprehensive assessment and the SSM launch, compared to the less signicant banks, which expected to remain under NSAs’ supervision. The sign and the statistical signicance of the co- efcient of the did variable has not been affected by adding control variables in the model (Models 2 and 3). These results are consistent with our claim that in anticipation of the SSM and the comprehen- sive assessment, the banks which were expected to be classied as signicant and to be supervised by the ECB encountered changes in their share of inter- bank deposits compared to the less signicant banks, which were expected to remain under NSAs’ supervi- sion. Models 7, 8, and 9 (Table 11) display the results from the placebo test where, by creating a ctional time dummy variable, we assumed that the SSM had been implemented in 2012. These models show that there is no evidence of a statistically signicant effect on the-interbank-deposits-with-a-maturity-of- 3–12-months-to-total-assets ratio of signicant banks compared to less signicant banks, assuming the SSM was implemented in 2012. These results imply that ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 123 Table 12. Interbank deposits by maturity. Dependent variable: interbank deposits with maturity of 1–5 years to total assets. Impact of SSM implementation Impact of CA and SSM launch Placebo test (2012) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 (Intercept) 0.092 0.093 (0.007) (0.007) treated 0.006 0.004 0.005 0.007 0.004 0.006 0.007 0.005 0.006 (0.007) (0.008) (0.008) (0.007) (0.008) (0.008) (0.009) (0.009) (0.009) time 0.005 0.003 0.009 0.003 0.001 0.011 0.001 0.003 0.009 (0.002) (0.002) (0.003) (0.003) (0.003) (0.004) (0.003) (0.003) (0.004) did 0.017 0.013 0.013 0.016 0.012 0.012 0.002 0.003 0.006 (0.004) (0.004) (0.004) (0.005) (0.005) (0.005) (0.007) (0.007) (0.007) LnTA 0.013 0.006 0.014 0.006 0.015 0.006 (0.006) (0.006) (0.006) (0.006) (0.006) (0.006) LATA 0.000 0.000 0.000 0.000 0.000 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ROAA 0.006 0.007 0.006 0.007 0.007 0.007 (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) ETA 0.000 0.001 0.000 0.001 0.000 0.001 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) GDP 0.001 0.002 C 0.004 (0.001) (0.001) (0.001) UNE 0.000 0.000 0.001 (0.001) (0.001) (0.001) INF 0.005 0.006 0.005 (0.002) (0.002) (0.001) n 823 823 823 823 823 823 823 823 823 R 2 .009 .096 .121 .004 .093 .119 .001 .085 .116 Hausman test $ 2 73.516 18.743 52.46 71.125 16.424 52.791 82.691 16.769 42.097 df 3 7 10 3 7 10 3 7 10 p value .0615 .009031 9.379e 08 .0684 .02151 8.143e 08 .04077 .01895 7.207e 06 Note. This table displays the results from the difference-in-difference model. The dependent variable is interbank deposits of 1–5 years to total assets (ID1to5yTA). CAD comprehensive assessment. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. the decreased portion of total assets funded with interbank deposits with a maturity of 3–12 months in signicant banks is associated with the impact of the comprehensive assessment and the SSM launch, rather than any other past events. Table 12 displays the results of this analysis per- formed on the dependent variable of interbank de- posits with a maturity of 1–5 years to total assets. Models 1, 2, and 3 display the impact of SSM im- plementation on the share of interbank deposits in total assets. Results show a negative and statistically signicant coefcient of the composite variable did in Model 1, implying that signicant banks decreased their portions of total assets funded with interbank deposits with a maturity of 1–5 years after the SSM implementation, compared to less signicant banks. The statistical signicance and the sign of the coef- cient of the did variable have not been affected by adding control variables in the model (Models 2 and 3). These results are consistent with our claim that due to the implementation of the SSM, signicant banks encountered increased (interbank) depositors’ trust compared to less signicant banks. Models 4, 5, and 6 (Table 12) display the impact of the comprehensive assessment and SSM launch on the share of inter- bank deposits in total assets. Results show a negative and statistically signicant coefcient of the compos- ite variable did in Model 4, implying that signicant banks, which were expected to be supervised by ECB, had decreased their portions of total assets funded with interbank deposits with a maturity of 1–5 years in anticipation of the comprehensive assessment and the SSM launch, compared to less signicant banks, which were expected to remain under NSAs’ super- vision. The sign and the statistical signicance of the coefcient of the did variable have not been affected by adding control variables in the model (Models 2 and 3).These results are consistent with our claim that 124 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Table 13. Interbank deposits by maturity. Dependent variable: interbank deposits with a maturity of more than 5 years to total assets. Impact of SSM implementation Impact of CA and SSM launch Placebo test (2012) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 (Intercept) 0.062 0.058 0.064 0.068 0.080 0.042 (0.005) (0.068) (0.005) (0.005) (0.067) (0.068) treated 0.008 0.010 C 0.000 0.009 C 0.001 0.001 0.008 0.010 0.011 (0.005) (0.006) (0.006) (0.005) (0.006) (0.006) (0.006) (0.007) (0.007) time 0.005 0.006 0.007 0.007 0.006 0.009 0.010 0.011 0.015 (0.002) (0.002) (0.002) (0.002) (0.002) (0.003) (0.002) (0.003) (0.003) did 0.001 0.000 0.002 0.001 0.000 0.001 0.001 0.000 0.000 (0.003) (0.003) (0.003) (0.004) (0.004) (0.004) (0.005) (0.005) (0.005) LnTA 0.001 0.007 0.007 0.007 0.000 0.002 (0.003) (0.005) (0.004) (0.005) (0.003) (0.003) LATA 0.001 0.000 0.000 0.000 0.001 0.001 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ROAA 0.001 0.003 0.003 0.003 0.002 0.002 (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) ETA 0.000 0.001 0.001 0.001 0.000 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) GDP 0.002 0.001 C 0.001 C (0.001) (0.001) (0.001) UNE 0.002 0.001 0.001 (0.001) (0.001) (0.001) INF 0.002 C 0.002 C 0.000 (0.001) (0.001) (0.001) n 776 776 776 776 776 776 776 776 776 R 2 .007 .033 .062 .016 .056 .066 .019 .045 .051 Hausman test $ 2 73.357 34.009 41.802 66.855 15.432 31.351 70.515 0.61405 17.552 df 3 7 10 3 7 10 3 7 10 p value .06193 .8456 8.134e 06 .08263 .03084 .0005133 .07027 .9989 .063 Note. This table displays the results from the difference-in-difference model. The dependent variable is interbank deposits of more than 5 years to total assets (IDmore5yTA). CAD comprehensive assessment. Statistical signicance: C p< :1, p< :05, p< :01, p< :001. Source: Authors’ calculations. in anticipation of the SSM and the expected compre- hensive assessment, the banks which were expected to be classied as signicant and to be supervised by the ECB encountered increased (interbank) depos- itors’ trust compared to less signicant banks, which were expected to remain under NSAs’ supervision. Models 7, 8, and 9 (Table 12) display the results from the placebo test where, by creating a ctional time dummy variable, we assumed that the SSM had been implemented in 2012. These models show that there is no evidence of a statistically signicant effect on the interbank-deposits-with-a-maturity-of-1–5-years- to-total-assets ratio of signicant banks compared to less signicant banks, assuming the SSM was imple- mented in 2012. These results imply that the decrease of the portions of total assets funded with interbank deposits with a maturity of 1–5 in signicant banks is associated with the SSM implementation, rather than any other past event. Additionally, these results imply that signicant banks had decreased their portions of total assets funded with interbank deposits with a ma- turity 1–5 years in anticipation of the comprehensive assessment and the SSM launch, rather than any other past event. Table 13 displays the results of this analysis per- formed on the dependent variable of interbank de- posits with a maturity of more than 5 years to total assets. From Models 1, 2, and 3, which display the impact of the SSM implementation on the share of interbank deposits in total assets, measured via the dependent variable of interbank deposits with a ma- turity of more than 5 years to total assets, it is evident that the coefcient of the variable did is statistically in- signicant. This implies that there were no differences in the portion of total assets funded with interbank deposits with a maturity of more than 5 years between signicant and less signicant banks which could be associated with the SSM implementation. From ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 125 Models 4, 5, and 6 (Table 13), which display the impact of the comprehensive assessment and SSM launch on the share of interbank deposits in total assets, it is evident that the coefcient of the variable did is statistically insignicant. This implies that there were no differences in the portions of total assets funded with interbank deposits with a maturity of more than 5 years between signicant and less signicant banks which could be associated with the anticipa- tion of the comprehensive assessment and of the SSM. Models 7, 8, and 9 (Table 13) display the results from the placebo test, where, by creating a ctional time dummy variable, we assumed that the SSM had been implemented in 2012. These models show that there is no evidence of a statistically signicant effect on the interbank-deposits-with-a-maturity-of-more- than-5-years-to-total-assets ratio of signicant banks compared to less signicant banks, assuming the SSM was implemented in 2012. These results imply that the interbank-deposits-with-a-maturity-of-more- than-5-years-to-total-assets ratio was affected neither by the SSM implementation, nor by the banks’ antic- ipation of the comprehensive assessment and of the SSM, nor by any other past events. To sum up, results from this analysis show that due to the implementation of the SSM, signicant banks had increased their portions of total assets funded with short-term interbank deposits with a maturity of less than 3 months and decreased their portions of to- tal assets funded with long-term interbank deposits, with a maturity of 1–5 years. No differences be- tween the signicant and less signicant banks were identied in the portion of total assets funded with interbank deposits with a maturity of 3–12 months and interbank deposits with a maturity of more than 5 years. These results imply that the trust effect of the interbank depositors is strongly demonstrated with the short-term interbank deposits, which are based on trust and are not collateralized. Long-term deposits are collateralized more frequently and therefore are considered safer. These results indicate that (inter- bank) depositors’ trust in signicant banks, which are supervised by the ECB, increased signicantly after the SSM implementation. These ndings further sup- port our hypothesis that the SSM implementation in fact improved the credibility of signicant banks. Additionally, we have found evidence that in 2013 in anticipation of the SSM and the comprehensive assessment, the banks which were expected to be classied as signicant and to be supervised by ECB increased their portions of total assets funded with interbank deposits with a maturity of up to 3 months (at a statistical signicance of 10%), compared to less signicant banks, which were expected to remain un- der NSAs’ supervision. Moreover, results show that signicant banks decreased their portions of total assets funded with interbank deposits with maturi- ties of 3–12 months and 1–5 years in anticipation of the SSM and the comprehensive assessment. No dif- ferences between the signicant and less signicant banks were identied in the interbank-deposits-with- a-maturity-of-more-than-5-years-to-total-assets ratio. These results are consistent with our claim that banks which were going to be assessed with the compre- hensive assessment and which were going to be supervised by the ECB were considered safer due to stricter supervision and consequently encountered increased (interbank) depositors’ trust. On this sub- sample we applied placebo tests, where, by creating a ctional time dummy variable, we assumed that the SSM had been implemented in 2012. We did not nd any differences in the portion of total assets funded with interbank deposits of any maturity between the signicant and less signicant banks in 2012, the year before the announcement of the SSM. This implies that the changes in the interbank deposits structure by maturity we have identied arise from the SSM implementation and not from any other past event. 4 Conclusion The SSM as a supervisory framework was imple- mented in 2014, when the most signicant banks in the euro area, which comprise 80% of the total banking assets, switched from national supervisors to the ECB and the remaining banks remained un- der national supervisors (NSAs). A preparatory step to the SSM implementation was the comprehensive assessment of the banks which fullled the criteria for signicance and expected to switch to the ECB as the main supervisory body. Since the ultimate goal of the SSM is increased bank safety and resilience (ECB, 2014b), the signicant banks which are super- vised by the ECB are likely to be considered safer. Consequently, we have examined changes in depos- itors’ trust in the signicant banks compared to less signicant banks, which remained under supervision by their NSAs. The effects of institutional changes such as the SSM implementation can be visible in the medium to long run (Fiordelisi et al., 2017). However, we have also inspected the immediate trust effect of depositors caused by the SSM anticipation and the expected comprehensive assessment. With this paper we contribute to the recent lit- erature stream on the SSM and to the established literature on supervision and bank regulation. The main contribution is in providing insight into how de- positors’ trust changed due to the implementation of the SSM and due to the anticipation of the SSM launch and the comprehensive assessment. Our main nding 126 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 is that depositors’ trust in signicant banks increased signicantly after the SSM implementation. We have provided empirical evidence that signicant banks, which are supervised by the ECB, increased their shares of interbank deposits in their total assets after the SSM implementation, compared to less signicant banks, which are supervised by NSAs. More specif- ically, the signicant banks increased their shares of short-term interbank deposits with a maturity of less than 3 months and decreased their shares of long- term interbank deposits, with a maturity of 1–5 years, compared to the less signicant banks. These results provide empirical evidence of the effectiveness of SSM as a supervisory framework to improve the cred- ibility and trustworthiness of signicant banks and in turn to achieve its primary objective—improved bank stability (ECB, 2014b). Additionally, we have provided empirical evidence that in 2013, in antici- pation of the SSM and the expected comprehensive assessment, signicant banks, which were expected to be supervised by the ECB, increased their shares of interbank deposits in their total assets compared to less signicant banks. More specically, the banks which were expected to be classied as signicant and to be supervised by the ECB increased their shares of interbank deposits with a maturity of up to 3 months (at a statistical signicance of 10%) in their total assets and decreased their shares of interbank deposits with maturities of 3–12 months and 1–5 years in their total assets compared to the less signicant banks. This implies that the ECB was perceived as a stricter super- visory authority compared to the NSAs. This nding is in line with the existing literature that explores su- pervisory frameworks and their structure (Colliard, 2020; Fiordelisi et al., 2017). We have found no evidence that the SSM imple- mentation and the SSM launch affected the share of total deposits in total assets of signicant banks. A possible explanation for this could be that institu- tional changes such as the SSM implementation do not impact customer deposits and saving accounts, which are included in banks’ total deposits. References Abad, P ., García-Olalla, M., & Robles, M. D. (2020). Does the Single Supervisory Mechanism reduce overall risk in the European stock market? Global Policy, 11(S1), 39–51. https://doi.org/10 .1111/1758-5899.12755 Acharya, V . V ., Gromb, D., & Yorulmzer, T. (2012). Imperfect com- petition in the interbank market for liquidity as a rationale for central banking. American Economic Journal: Macroeconomics, 4(2), 184–217. https://doi.org/10.1257/mac.4.2.184 Agarwal, S., Lucca, D., Seru, A., & Trebbi, F. (2014). Inconsistent regulators: Evidence from banking. The Quarterly Journal of Eco- nomics, 129(2), 889–938. https://doi.org/10.1093/qje/qju003 Allen, F., Carletti, E., & Marquez, R. (2015). Deposits and bank capital structure. Journal of Financial Economics, 118(3), 601–619. https://doi.org/10.1016/j.jneco.2014.11.003 Allen, F., Covi, G., Gu, X., Kowalewski, O., & Montagna, M. (2020). The interbank market puzzle (Staff Working Paper No. 862). Bank of England. http://doi.org/10.2139/ssrn.3601848 Altunba¸ s, Y., Polizzi, S., Scannella, E., & Thornto, J. (2022). Euro- pean Banking Union and bank risk disclosure: The effects of the Single Supervisory Mechanism. Review of Quantitative Finance and Accounting, 58(2), 649–683. https://doi.org/10.1007/s11156 -021-01005-z Alves, C. F., Marques, B. P ., & Silva, J. C. (2023). Transnational banking supervision, distance-to-distress and credit risk: The SSM case. Applied Economics Letters, 30(15), 2079–2085. https:// doi.org/10.1080/13504851.2022.2092590 Ampudia, M., Beck, T., Beyer, A., Colliard, J. E., Leonello, A., Maddaloni, A., & Marques-Ibanez, D. (2019). The architecture of supervision (Working Paper No 2287). European Central Bank. https://doi.org/10.2866/355828 Angelini, P ., Nobili, A., & Picillo, C. (2011). The interbank mar- ket after August 2007: What has changed, and why? Journal of Money, Credit and Banking, 43(5), 923–958. https://doi.org/ 10.1111/j.1538-4616.2011.00402.x Avgeri, I., Dendramis, Y., & Louria H. (2021). The Single Super- visory Mechanism and its implications for the protability of European banks. Journal of International Financial Markets, Insti- tutions and Money, 74, Article 101382. https://doi.org/10.1016/ j.intn.2021.101382 Avignone, G., Altunbast, Y., Polizzi, S., & Reghezza, A. (2021). Cen- tralised or decentralised banking supervision? Evidence from European banks. Journal of International Money and Finance, 110, Article 102264. https://doi.org/10.1016/j.jimonn.2020.102264 Barth, J. R., Caprio, G., Jr., & Levine, R. (2004). Regulation and su- pervision: What works best? Journal of Financial Intermediation, 13(2), 205–248. https://doi.org/10.1016/j.j.2003.06.002 Barth, J. R., Caprio, G., Jr., & Levine, R. (2013). Bank regulation and supervision in 180 countries from 1999 to 2011. Journal of Fi- nancial Economic Policy, 5(2), 111–219. https://doi.org/10.1108/ 17576381311329661 Barth, J. R., Dopico, L. G., Nolle, D. E., & Wilcox, J. A. (2002). Bank safety and soundness and the structure of bank supervision: A cross-country analysis. International Review of Finance, 3(3–4), 163–188. https://doi.org/10.1111/j.1369-412X.2002.00037.x Bernard, H., & Bisignano, J. (2000). Information, liquidity and risk in the international interbank market: Implicit guarantees and private credit market failure (Working Paper 86). Bank for International Settlements. http://doi.org/10.2139/ssrn.849014 Bertrand, M., Duo, E., & Mullainathan, S. (2004). How much should we trust differences-in-differences estimates? The Quar- terly Journal of Economics, 119(1), 249–275. https://doi.org/10 .1162/003355304772839588 Bologna, P . (2011). Is there a role for funding in explaining recent US bank failures? (IMF Working Papers Volume 2011, Issue 180). International Monetary Fund. https://doi.org/10.5089/ 9781462311286.001 Bruche, M., & Suarez, J. (2010). Deposit insurance and money mar- ket freezes. Journal of Monetary Economics, 57(1), 45–61. https:// doi.org/10.1016/j.jmoneco.2009.11.007 Buch, C. M., & DeLong, G. (2008). Do weak supervisory systems encourage bank risk-taking? Journal of Financial Stability, 4(1), 23–39. https://doi.org/10.1016/j.jfs.2007.12.002 Calzolari, G., Colliard J. E., & Loranth, G. (2019). Multinational banks and supranational supervision. The Review of Financial Studies, 32(8), 2997–3035. https://doi.org/10.1093/rfs/hhy116 Carletti, E., Dell’Ariccia, G., & Marquez, R. (2021). Supervisory in- centives in a banking union. Management Science, 67(1), 455–470. https://doi.org/10.1287/mnsc.2019.3448 Carboni, M., Fiordelisi, F., Ricci, O., & Lopes, F. S. S. (2017). Surprised or not surprised? The investors’ reaction to the com- prehensive assessment preceding the launch of the banking union. Journal of Banking & Finance, 74, 122–132. https://doi .org/10.1016/j.jbankn.2016.11.004 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 127 Chang, C., M., Nieh, C., C., & Peng, Y. H. (2011). Are bigger banks more protable than smaller banks? Journal of Applied Finance & Banking, 1(3), 59–71. https://www.scienpress.com/journal _focus.asp?main_id=56&Sub_id=IV&Issue=70 Colliard, J. E. (2020). Optimal supervisory architecture and nancial integration in a banking union. Review of Finance, 24(1), 129–161. https://doi.org/10.1093/rof/rfz004 Constancio, V . (2012, September 7). Towards a European Banking Union [Lecture transcript]. European Central Bank. https://www.ecb.europa.eu/press/key/date/2012/html/ sp120907.en.html Constancio, V . (2013, February 12). Towards the Banking Union [Speech]. 2nd FIN-FSA Conference on EU Regulation and Supervision: Banking and Supervision under Transforma- tion, Helsinki, Finland. https://www.ecb.europa.eu/press/ key/date/2013/html/sp130212.en.html Council of the European Union. (2013a). Council Regulation No 1024/2013 conferring specic tasks on the European Central Bank concerning policies relating to the prudential supervision of credit institutions. Ofcial Journal of the European Union. https://eur-lex.europa.eu/eli/reg/2013/1024/oj Council of the European Union. (2013b, October 15). Council approves the single supervisory mechanism for banking union [Press release]. https://data.consilium.europa.eu/doc/document/ ST140442013INIT/EN/pdf Cuadros-Solas, P ., Salvador, C., and Suárez, N. (2023). Banking su- pervisory architecture and sovereign risk. SSRN. http://doi.org/ 10.2139/ssrn.4503434 De Larosière, J. (2009). Report of the High-Level Group on Financial Supervision in the EU. European Commission. https://ec.europa.eu/economy_nance/publications/pages/ publication14527_en.pdf Delis, M. D., & Staikouras, P . (2011). Supervisory effectiveness and bank risk. Review of Finance, 15(3), 511–543. http://doi.org/10 .1093/rof/rfq035 Diamond, D., & Dybvig, P . H. (2000). Bank runs, deposit insurance, and liquidity. Federal Reserve Bank of Minneapolis Quarterly Re- view, 24(1), 13–23. https://doi.org/10.1086/261155 Dietrich, D., & Hauck, A. (2020). Interbank borrowing and lending between nancially constrained banks. Economic Theory, 70(2), 347–385. https://doi.org/10.1007/s00199-019-01220-9 Dinger, V ., & Hagen, J. (2009). Does interbank borrowing reduce bank risk? Journal of Money, Credit and Banking, 41(2–3), 491–506. https://doi.org/10.1111/j.1538-4616.2009.00217.x European Central Bank. (2011). The monetary policy of the ECB 2011. https://doi.org/10.2866/99667 European Central Bank. (2014a). Aggregate report on the comprehen- sive assessment. https://doi.org/10.2866/26036 European Central Bank. (2014b). Guide to banking supervision. https://doi.org/10.2866/17199 European Central Bank. (2014c). The list of signicant supervised entities and the list of less signicant insti- tutions. https://www.ecb.europa.eu/pub/pdf/other/ ssm-listofsupervisedentities1409en.pdf European Central Bank. (2015). The list of signicant super- vised entities and the list of less signicant institutions . https://www.bankingsupervision.europa.eu/ecb/pub/ pdf/ssm-listofsupervisedentities1502en.pdf European Central Bank. (2016). ECB annual report on supervisory activities 2015. https://doi.org/10.2866/053085 European Central Bank. (2018). SSM supervisory manual. https:// doi.org/10.2866/891872 European Central Bank. (2022). List of supervised entities. https://www.bankingsupervision.europa.eu/ecb/pub/pdf/ ssm.listofsupervisedentities202202.en.pdf European Commission. (2012). Proposal for a Council Regulation conferring specic tasks on the European Central Bank concerning policies relating to the prudential supervision of credit institutions. https://op.europa.eu/s/zgHH European Court of Auditors. (2014). European banking supervision taking shape—EBA and its changing context (Special Report No 5/2014). https://doi.org/10.2865/9459 Ferrarini, G. (2015). Single supervision and the governance of banking markets: Will the SSM deliver the expected bene- ts? European Business Organization Law Review, 16(3), 513–537. https://doi.org/10.1007/s40804-015-0018-7 Financial Crisis Inquiry Commission. (2011). The nancial cri- sis inquiry report: The nal report of the National Commission on the Causes of the Financial and Economic Crisis in the United States. https://www.govinfo.gov/content/pkg/GPO -FCIC/pdf/GPO-FCIC.pdf Fiordelisi, F., Ricci, O., & Lopez, F. S. S. (2017). The unintended con- sequences of the launch of the Single Supervisory Mechanism in Europe. Journal of Financial and Quantitative Analysis, 52(6), 2809–2836. https://doi.org/10.1017/S0022109017000886 Furne, C. (2001). Banks as monitors of other banks: Evidence from the overnight federal funds market. Journal of Business, 74(1), 33– 57. https://doi.org/10.1086/209662 Gortsos, C. V . (2023). The Single Supervisory Mechanism (SSM). In C. V . Gortsos (Ed.), The European Banking Regulation Handbook (Vol. I). 437–498. https://doi.org/10.1007/978-3-031-32859-6_8 Grandi, P ., & Guillet, M. (2021). The upside down: Banks, deposits and negative rates. SSRN. https://doi.org/10.2139/ssrn.3363743 Hill, R. C., Grifths, W. E., & Lim, G. C. (2018). Principles of econo- metrics (5th ed.). Wiley. Kashyap, A. K., Rajan, R., & Stein, J. C. (2002). Banks as liquidity providers: An explanation for the coexistence of lending and de- posits taking. Journal of Finance, 57(1), 33–73. https://doi.org/ 10.1111/1540-6261.00415 Kaufman, G. G. (1972). Deposit variability and bank size. Journal of Financial and Quantitative Analysis, 7(5), 2087–2096. https:// doi.org/10.2307/2329956 Kundu, S., Park, S., & Vats, N. (2021). The geography of bank deposits and the origins of aggregate uctuations . SSRN. http://doi.org/10 .2139/ssrn.3883605 Kok, C., Müller, C., Ongena, S., & Pancaro, C. (2023). The disci- plining effect of supervisory scrutiny in the EU-wide stress test. Journal of Financial Intermediation, 53, Article 101015. https:// doi.org/10.1016/j.j.2022.101015 Koroleva, E., Jigeer, S., Miao, A., & Skhvediani, A. (2021). Determi- nants affecting protability of state-owned commercial banks: Case study of China. Risks, 9(8), Article 150. https://doi.org/ 10.3390/risks9080150 Li, S., & Marinˇ c, M. (2018). Derivatives clearing and systemic risk. SSRN. https://doi.org/10.2139/ssrn.2734134 Loipersberger, F. (2018). The effect of supranational banking su- pervision on the nancial sector: Event study evidence from Europe. Journal of Banking & Finance, 91, 34–48. https://doi.org/ 10.1016/j.jbankn.2018.04.003 Mansson, E. J. L. (2014). A single European Banking Supervisor— An anachronistic move? [Master’s thesis]. University of Lund. https://www.lunduniversity.lu.se/lup/publication/4628753 Martin, C., Puri, M., & Uer, A. (2018). Deposit inows and outows in failing banks: The role of deposit insurance (NBER Working Pa- per 24589). National Bureau of Economic Research. https:// doi.org/10.3386/w24589 Md-Yusuf, M., & Md-Zain, M. S. (2020). Macroeconomic determi- nants of interbank money market rate in Malaysia. Asian Journal of Accounting and Finance, 2(2), 42–51. https://myjms.mohe.gov .my/index.php/ajan/article/view/9804 Möslein, F. (2015). Third parties in the European Banking Union: Regulatory and supervisory effects on private law relationships between banks and their clients or creditors. European Busi- ness Organization Law Review, 16(3), 547–574. https://doi.org/ 10.1007/s40804-015-0024-9 Moura, P ., Barbosa, F., Alves, C., & Camanho, A. (2023). The impact of the single supervisory mechanism on Eurozone banking: The assessment of trends in efciency and frontier position. Ap- plied Economics. Advance online publication. https://doi.org/ 10.1080/00036846.2023.2290598 Näther, M. (2019). The effect of the central bank’s standing facilities on interbank lending and bank liquidity holding. Economic Theory, 68(3), 537–577. https://doi.org/10.1007/s00199-018 -1134-8 128 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 Näther, M., & Vollmer, U. (2019). National versus supranational bank regulation: Gains and losses of joining a banking union. Economic Systems, 43(1), 1–18. https://doi.org/10.1016/j.ecosys .2018.05.004 Oura, H., González-Hermosillo, B., Chan-Lau, J. A., Gudmundsson, T., & Valckx, N. (2013). Changes in bank funding patterns and nancial stability risks. In Global Financial Stability Report: Transition Challenges to Stability (pp. 105–148). International Monetary Fund. https://www.imf.org/en/Publications/GFSR/Issues/2016/ 12/31/~/media/Websites/IMF/imported-agship-issues/ external/pubs/ft/GFSR/2013/02/pdf/_c3pdf.ashx Pancotto, L., Gwilym, O., & Williams, J. (2018). The European Bank Recovery and Resolution Directive: A market assessment. Jour- nal of Financial Stability, 44, Article 100689. https://doi.org/10 .1016/j.jfs.2019.100689 Quaglia, L., & Verdun, A. (2023). The European Central Bank, the Single Supervisory Mechanism and the COVID-19 related eco- nomic crisis: A neofunctionalist analysis. Journal of European Integration, 45(1), 139–156. https://doi.org/10.1080/07036337 .2022.2155640 Sigmund, M., & Raunig, B. (2023). Watching over 21,000 billion euros: Does the ECB Single Supervisory Mechanism affect bank competition in the euro area? SSRN. http://doi.org/10.2139/ssrn.4662381 Stern, G. H., & Feldman, R. J. (2004). Too big to fail: The hazards of bank bailouts. Brookings Institution Press. https://www.jstor .org/stable/10.7864/j.ctt1gpcc87 Thao, V . T. P ., & Thanh, L. T. (2021). An empirical analysis of macroeconomic and bank-specic factors affecting bank de- posits in Vietnam. International Journal of Financial Research, 12(2), 172–183. https://doi.org/10.5430/ijfr.v12n2p172 Valahzagharda, M. K., & Kashb, S. M. (2014). Investigating the effects of various factors on bank deposits. Management Science Letters, 4, 707–712. https://doi.org/10.5267/j.msl.2014.2.021 Zeitlin, J. (2023). Hierarchy, polyarchy, and experimentalism in EU banking regulation: the Single Supervisory Mechanism in ac- tion. Journal of European Integration, 45(1), 79–101. https://doi .org/10.1080/07036337.2022.2144279 ECONOMIC AND BUSINESS REVIEW 2024;26:104–129 129 Appendix Table A1. List of banks located in smaller economies and criteria used for classication. Bank Country Criteria for classication Nova Ljubljanska banka d.d. Slovenia Total assets (TA) above 20% of GDP Nova Kreditna banka Maribor d.d. Slovenia Among the three largest credit institutions in the country Vseobecna Uverova Banka, a.s. Slovakia Among the three largest credit institutions in the country Slovenska sporitelna, a.s. Slovakia Among the three largest credit institutions in the country Tatra Banka Slovakia Among the three largest credit institutions in the country AS SEB Pank Estonia TA above 20% of GDP Swedbank AS Estonia TA above 20% of GDP AS SEB Banka Latvia Among the three largest credit institutions in the country Swedbank AS (Latvia) Latvia TA above 20% of GDP Hellenic Bank Public Company Limited Cyprus TA above 20% of GDP Bank of Cyprus Public Company Limited Cyprus TA above 20% of GDP ABLV Bank AS Latvia Among the three largest credit institutions in the country Sberbank Europe AG Austria Signicant cross-border activities HSBC Bank Malta p.l.c. Malta TA above 20% of GDP Bank of Valletta p.l.c. Malta TA above 20% of GDP Banque Internationale à Luxembourg Luxembourg Part of Precision Capital S.A. (Size of TA EUR 30–50 bil.) Source: Authors’ calculations.