Bank Mergers and Deposit Rate Rigidity

Bank Mergers and Deposit Rate Rigidity

Valeriya Dinger

Received: 22 December 2011 /Revised: 24 September 2013 /Accepted: 1 October 2013# Springer Science+Business Media New York 2013

Abstract In this paper, I empirically explore the relation between bank mergers and therigidity of banks’ deposit rates. I find that merging banks are more likely to change theirdeposit rates in the first months following a merger. However, in the long term, mergingbanks change their deposit rates less frequently than non-merging banks. This finding isparticularly true after mergers with large target banks and after mergers with a substantialgeographical expansion of the bank’s operations. The documented post-merger deposit raterigidity has important implications for the evaluation of merger effects in the context of anti-trust and monetary policies.

Keywords Bankmergers . Bank market structure . Interest rate dynamics . Hazard rate

JEL Classifications . L11 . L13 . D43 . G21

1 Introduction

The literature on interest rates extensively reviews the effect of bank mergers. In particular,this literature analyzes the effects of mergers on the behavior of banks when setting interestrates (see, e.g., Hannan and Prager 1998; Focarelli and Panetta 2003; Craig and Dinger2009; Erel 2011). However, this research exclusively focuses on the magnitude of theinterest-rate changes around and after the completion of a merger but does not explorehow the frequency of changing these rates relates to the mergers. While earlier studies onthis topic (Hannan and Prager 1998; Focarelli and Panetta 2003) do not address the rigidityin interest rates, a study by Craig and Dinger (2009) highlights this rigidity and argues thatignoring it can generate inconsistent estimates of the effect of mergers on the changes ininterest rates. These authors present a method for the estimation of the magnitude of the

J Financ Serv ResDOI 10.1007/s10693-013-0182-2

The empirical analysis presented in this paper was performed during my visit at the Federal Reserve Bank ofCleveland. I thank Ben R. Craig for sharing insights and data. I also thank James Thomson and Jürgen vonHagen for useful comments on earlier versions. Financial support by the Deutsche Forschungsgemeinschaft(Research Grant DI 1426/2-1) is gratefully acknowledged.

V. Dinger (*)University of Osnabrueck, Rolandstr. 8, 49069 Osnabrueck, Germanye-mail: [email protected]

V. DingerLeeds University Business School, Leeds, UK

changes that explicitly controls for their rigidity. However, they do not explore the effect ofthe merger on the frequency of the changes.

By leaving the frequency of changing interest rates unexplored the literature on bankmergers leaves several important questions unanswered. How does the documented differ-ence between the interest rates of merging and non-merging banks arise (Hannan and Prager1998; Focarelli and Panetta 2003; Craig and Dinger 2009)? Do merging banks tend toperform numerous small changes in their interest rates, or do they perform a few with greatermagnitudes? Is there a transition period in which the merging banks synchronize theirpreviously independent behaviors for setting interest rates? Is the difference between thelevels of retail rates of merging and non-merging banks a result of merging banks activelychanging their retail rates or rather of merging banks staying passive? Are the merging bankspassive in order to not adjust their retail rates to monetary policy rate shocks for longerperiods than their non-merging peers? All these questions are relevant to an understanding ofsome of the important aspects of post-merger deposit-rate dynamics. These include theability of merging banks to extract profits from their changed market position; the potentialattempts by merging banks to disguise the ultimate effect of the merger on the deposit-ratelevels by numerous deposit-rate changes; as well as the effect of bank mergers on the speedof monetary policy transmission. Overlooking these aspects of post-merger deposit ratedynamics limits the ability to evaluate the effect of mergers on depositors’ welfare andmonetary policy efficiency.

In this paper, I address these open questions by presenting empirical evidence on how thefrequency of bank deposit rate changes evolves around bank mergers. By exploring the linkbetween bank mergers and the deposit rate’s rigidity, I also address some of the shortcom-ings in the literature that analyzes retail interest-rate rigidity (Haubrich 1992; Neumark andSharpe 1992; Hannan and Berger 1991; Hofmann and Mizen 2004; De Graeve et al. 2007;Driscoll and Judson 2009; Craig and Dinger 2010). This literature finds a significant relationbetween the rigidity of interest rates and the market’s structure but has failed to establish alink between this rigidity and bank mergers as a major source of the dynamics in themarket’s structure. By explicitly focusing on bank mergers, I particularly refer to two gapsin the research on the rigidity of interest rates. First, the empirical models that explore thelink between the market’s structure and rigidity potentially suffer from the weak identifica-tion caused by the presence of omitted variable biases. These biases can arise because somevariables that are likely to affect the frequency of interest-rate changes are not observable.For example, researchers are typically unable to perfectly measure either the complexity ofthe decision to change the interest rate or the degree of market power at the bank level. Thefocus on mergers allows me to strengthen the identification by isolating the effect of theobservable changes in the bank’s and the market’s structures that result from the mergerwhile holding the effect of all other relevant and potentially unobservable variables fixed.The second gap is related to the fact that most studies ignore the possibility that the short-term effect of the mergers on the retail interest rate’s rigidity could differ from the long-termeffects of the merger that arise through the changes in these structures. If this difference isthe case, then the estimation of the effect of the market’s structure on the rigidity in sampleswith a substantial number of mergers can be inconsistent. I address this issue by explicitlydisentangling the short-term post-merger dynamics from the long-term relation betweenmergers generated by changes in the market’s structure and the frequency of interest-ratechanges.

The empirical analysis of the relation between bank mergers and the frequency ofdeposit-rate changes proceeds in several steps. First, I visualize the post- merger changesin deposit-rate rigidity by comparing the standard Kaplan-Meier nonparametric hazard

J Financ Serv Res

functions of changing the deposit rates between banks that have recently undergone a mergerand the rest of the sample banks. Next, I focus on exploring the relation between the timedistance from the merger, the merger features—such as the change in bank size, the changein the number of markets, and the change in market share—and the conditional probabilityof a bank changing its deposit rates. To estimate this relation, I use a semi-parametric Coxhazard model. By including these features as covariates, the model allows me to estimate thefrequency of deposit-rate changes that follow bank mergers while controlling for thedynamics of monetary policy and the market’s interest rates. In addition, I also analyze thereaction of the direct competitors of the merging banks by exploring whether the frequencyof deposit rate changes in banks not directly involved in a merger but operating in the samelocal markets is also affected. Furthermore, I relate the results of the frequency of thedeposit-rate changes around bank mergers to the discussion on the changes in the level ofthe deposit rate by presenting the distributions of the magnitude of the rate changes both indifferent time periods after a merger and for markets with different structures.

The estimations use a comprehensive data set that combines weekly information aboutthe deposit rates from roughly 600 US banks for the period of 1997 to 2006 with data aboutthe corresponding bank and local market characteristics. A complete list of bank mergers inand around this time period is matched to the deposit-rate data. The resulting sample coversbanks with a wide range of variation in size and market share that reflects their deposit-ratepolicy. These banks represent more than 160 local markets defined as metropolitan statisticalareas (MSAs). The focus on deposit rather than loan interest rates is driven by the betteravailability of deposit-rate data. Deposit rates also are more homogenous in the banks andless affected by credit risk considerations.

The results of the analysis are as follows. Bank mergers are associated with substantialchanges in the durations of the banks’ deposit rates. However, deposit rate rigidity in theshort-term after a merger substantially differs from deposit rate rigidity in the longer-termafter a merger. For example, the conditional probability for merging banks in the first yearfollowing the merger of changing their deposit rates is higher than that of non-mergingbanks. This finding is consistent with the existence of a period when the process of settingthe deposit rate is synchronized between the merging banks. The finding also suggests thatmerging banks consider their pre-merger deposit rates as outdated given the changes inmarket power and the market’s structure generated by the merger. Therefore, they are eagerto begin with the search for a new optimal deposit rate shortly after the merger. This fact,which to my knowledge has not been empirically highlighted so far, underlines the impor-tance of the exploitation of market power in pricing as a major motive for bank mergers. Thefact that the high frequency of changes is sustained throughout the first year after the mergeralso suggests that banks are either not confident about the exact magnitude of the optimaldeposit rate, or they prefer not to move to that rate at once, for example, in order to disguisethe effect of the merger on deposit-rate levels. About two years following the merger, thefrequency of the changes becomes substantially lower relative to non-merging banks. Thisresult suggests that once the merging banks establish the deposit rates that correspond totheir increased market power, they become less likely to revise their deposit rates. Amongthe characteristics of the merger, the frequency of changing deposit rates is strongly relatedto the size of the target bank as well as to the change in the number of local markets wherethe bank operates. Larger banks and banks with a broad geographical scope change theirdeposit rates less frequently and react more slowly to shocks in the monetary policy rate. In adynamic setting, this result confirms the finding in the literature that larger and geograph-ically diversified banks react particularly slowly to monetary policy rate dynamics. Theeffect of an increase in the market share of the bank following a merger depends on the

J Financ Serv Res

market concentration in the affected local markets. While in more concentrated markets amerger generates a rise in the bank’s market share that is negatively related to the frequencyof changing the deposit rates, the opposite is true in less concentrated markets. Thedistributions of the size and direction of the post-merger rate changes show that negativedeposit-rate changes dominate in the first year after the merger. This result is particularlystrong for concentrated markets.

The major contribution of the paper is the presentation of robust empirical evidence thatbank mergers are associated with significant modifications in the frequency of deposit-ratechanges. This evidence sheds new light on how mergers affect interest-rate dynamics thatgoes beyond the effect of mergers on the levels of interest rates. While Craig and Dinger(2009) show that if merging banks change their retail rates at all, they do so in a negativedirection; in this paper, I endogenize the reaction of the interest rate’s rigidity to the bankmerger. By doing so, I find that merging banks in the months immediately after a merger aremore likely than their non-merging peers to change deposit rates, but they become less likelyto adjust their rates in the long run. Thus, this finding explains the observed difference indeposit rates between merging and non-merging banks. But in the long run, the differencesbetween the levels of deposit rates offered by merging and non-merging banks potentiallyarise from the slow adjustment of merging banks’ rates to shocks in the market’s interest-ratelevel.

In the framework of anti-trust policy, these findings suggest that the estimatedeffect of the merger on interest rates can differ substantially depending on when theevaluation takes place. In particular, my results suggest that an exclusive focus on theshort-term effect can produce misleading results.1 In the framework of monetarypolicy, this evidence about the long-term changes in the deposit rate’s rigidityfollowing a merger underlines the importance of changes in the market’s structurefor the effectiveness of monetary policy and indicates the challenges to monetarypolicy in consolidated markets.

The rest of the paper is structured as follows. Section 2 presents the data and defines themeasures of retail interest rate durations. Section 3 discusses the dynamics of the probabilityof changing deposit rates around bank mergers and compares the Kaplan-Meier hazardfunctions of changing deposit rates between merging and non-merging banks. Section 4presents the results of the Cox estimation. Section 5 explores how bank mergers are relatedto the frequency of deposit-rate changes in the merging banks’ direct competitors. Section 6discusses the robustness of the results, and Section 7 concludes.

2 Data sources and measurement issues

2.1 Data sources

The empirical analysis is based on a data set that combines information on the banks’ depositrates in different local markets with information on the bank’s merger history. The data setcombines this information with information on the key balance sheet characteristics of thebank and the features of each of the local markets.

1 This concern has also been discussed by Focarelli and Panetta (2003). These authors argue that a longer-termtime horizon for the evaluation of post-merger effects on deposit rate levels is more appropriate since it allowscontrolling for potential merger-generated long-term efficiency gains.

J Financ Serv Res

The main restriction in the compilation of the data set is the data’s availability in theweekly reports of the BankRate Monitor Inc.2 The BankRate Monitor Inc. comprises morethan 900 financial institutions (both banks and thrifts). I exclude the thrifts from the analysisbecause I have no information on their merger history. The remaining information comprisesthe deposit rates of 624 banks in 164 local markets (defined as metropolitan statistical areas).The sample covers a period that starts on September 19, 1997, and ends on July 21, 2006.The market exit and mergers lead to substantial changes in the composition of the sample,which start with 440 banks in 1997 and end with 220 banks in 2006.

My focus on deposit rates admittedly limits the scope of the analysis by leaving asideloan-rate dynamics that play a key role in the monetary transmission process. But, depositrates enable the analysis of the price-setting behavior of the banks without concerns forcustomers’ credit risk. Another limitation is the potential sample selection that could arise ifthe banks reporting to the BankRate Monitor’s survey follow systematically differentstrategies from the rest of the banking population when setting interest rates. These differentstrategies could be a concern if the fact that banks report their interest rates every weekgenerates incentives for them to change interest rates more frequently or to window-dresstheir retail rates (especially around a merger). The fact that this paper documents both asubstantial degree of rigidity as well as a negative link between bank mergers and thefrequency of adjusting deposit rates underlines the strength of the observed effects andrelativizes the sample selection concerns.

The Bankrate Monitor Inc. reports on a broad range of rates: checking accounts, moneymarket deposit accounts, and certificates of deposits with a maturity of 2 months to up to5 years. I concentrate on the rates for checking accounts and money market deposit accounts(MMDA) because these are the deposit rates with a substantial degree of rigidity for whichthe duration of the rates is a key determinant of the rate’s dynamics.3 For comparability, Ichoose to include those checking account rates—representing roughly 90 % of the obser-vations for checking-account rates—posted to accounts with a required minimum balance ofbetween 100 and 1,000 USD.4 Similarly, I include the MMDA rates on accounts withrequired minimum balance of between 100 and 2,500 USD, which again account for morethan 90 % of the MMDA rate observations.5 Because of the very low correlation between theminimum balance and the deposit rate (less than 10 %), the sample selection with regard tothe minimum balance restrictions is unlikely to affect the consistency of the estimates.

The reported deposit rates show a substantial variation not only across time but alsoacross banks and local markets. In particular, deposit rates offered by multimarket banks indifferent local markets vary substantially. This variation, consistent with earlier studies(Craig and Dinger 2009; Craig and Dinger 2010), is a signal of the banks’ reaction to thecompetitive conditions in the local market. In order to use both the cross-market and cross-bank variations in the deposit-rate dynamics for the identification of the impact from thebanks’ and local markets’ characteristics, I use the deposit-rate observations reported at the

2 Other studies use the same source of deposit-rate data, for example, Ors and Rice (2007), Driscoll andJudson (2009), Craig and Dinger (2009), and Craig and Dinger (2010).3 See Tables 1 and 2 for illustrations of checking and MMDA rate (relative) rigidities. The fact that these ratesare stickier than alternative deposit rates such as the rates on Certificates of Deposit has also been illustrated inDriscoll and Judson (2009) and Craig and Dinger (2010).4 Replicating the analysis with expanded data including a required minimum balance between 100 and 2,500USD (which cover in more than 99 % of the observations) generates qualitatively the same results.5 For both the checking accounts and the MMDAs the reported deposit rates refer to insured not brokereddeposits.

J Financ Serv Res

bank-market level. The resulting data encompasses a total of roughly 184,000 observationsof 1,738 bank-market groups.

The compiled deposit-rate data set is then matched with the information on the mergerhistory of the banks drawn from the Supervisory Master File of Bank Mergers andAcquisitions. To account for the long-term effects of bank mergers on their policy for settingdeposit rates, I add all of the mergers that took place starting in January 1992. A total of 142of the banks for which I have deposit-rate information were involved in mergers as acquirersduring the covered time period.

Next, information on a broad range of bank-specific variables for individual banks comesfrom the Quarterly Reports of Conditions and Income (call reports). This information ismerged into the data set. Also, I control for a number of variables that describe the localmarket’s structure. Data about these variables stems from the Summary of Deposits that isonly available annually.

2.2 Defining spells and durations

The hazard function analysis requires a measure of the deposit-rate durations. For thispurpose, following Craig and Dinger (2010) for each bank and market, I first track theduration of the deposit rates by setting the definitions of the individual quote lines anddeposit-rate spells. I define the quote linei,j,p as the set of deposit rates offered by bank i inlocal market j for deposit product p. The deposit-rate spell is defined as a subsection of thequote line for which the deposit rate goes unchanged. The definition of the deposit-ratespells assumes that if the same deposit rate is reported in two consecutive weeks, it has notchanged between observations. I define the number of weeks for which the deposit rate goesunchanged as the duration of the deposit-rate spell.

To avoid left censoring, I include only spells for which the exact starting date can beidentified. That is, for each bank-market I exclude all observations before the rate changesfor the first time. A spell ends with either a change in the deposit rate or with an exit of thebank-market unit from the observed sample. In the latter case the issue of right censoringarises. To deal with this issue I only include spells for which the end date is identifiable. Thisapproach is standard in survival analysis studies. Given the long time period of the samplerelative to the duration of the spells, this selection issue is much less pronounced in mysample relative to alternative survival data applications. Furthermore, because the exclusionof the spells is not related to the merger event and only reflects the fact that some spells beginor end outside the observation window, no risk of censoring biases arises.

The Bank Rate Monitor Inc. sometimes reports rates offered by smaller banks (less than10 % of the bank-market observations), but only if the quoted rate deviates from the ratequoted in the preceding week. To control for this deviation, I assume that a deposit-rate spell“survives” through the weeks until the next observation is reported (if the next reported rateis in week t, I assume the rate has “survived” until week t-1). However, a few instances existin the sample in which the bank-market unit exits the sample for a longer period (up to a fewyears) and re-enters the sample again. In this case, the assumption that the observations aremissing only because no change in the deposit rate is observed is too strong. I control for thisby treating an unreported rate as an unchanged rate only if the period of the missingobservations is less than 52 weeks.

An important measurement issue is the treatment of temporary changes in the deposit rate(the equivalent of sales in the price rigidity literature). Because temporary changes are animportant component of a bank’s policy on setting the deposit rate, I consider a temporarydeposit-rate change as a “failure” in the spell. Table 1 presents the summary information on

J Financ Serv Res

the number of spells defined with the procedure described above as well as information onthe number of temporary changes with different durations. The temporary deposit-ratechanges are common. However, the number of temporary deposit-rate changes reversedwithin 1 week is substantially larger than the number of temporary changes with a longerduration. This observation suggests that a substantial portion of the 1 week “temporarychanges” might not reflect changes in the deposit rate but rather misreporting. Because Icannot disentangle the potential reporting errors from the temporary deposit-rate changes,the estimations presented in the following sections are based on a definition of a spell failurethat ignores temporary changes with 1 week duration. As a robustness check, I have rerun allestimations using the full sample of spell failures as well as samples ignoring temporarychanges with durations of 2 or 4 weeks. Although the number of spells affected varies (e.g.,in the case of checking account rates: from 204 spells for which the rate was not observed fora few weeks but reappeared within 52 weeks to 191 spells for rates that reappear within26 weeks and to 311 spells when no restriction on the number of weeks is imposed), theresults that for the sake of parsimonious exposition are not reported here, stay qualitativelyunchanged.

Table 2 presents the summary statistics regarding the duration of the deposit rates in thesample when a spell is considered to end only if the change is sustained for more than1 week. The table uncovers the lumpiness of deposit-rate adjustments illustrated not only bythe low frequency but also by the large magnitude of the observed deposit-rate changes. Thesecond column of this table illustrates that the absolute value of the change in the checkingaccount rate in the occasions when a nonzero change is observed is 0.16 %, which is quitesubstantial given the average magnitude of the checking account rates of only 0.53 %. Given

Table 1 Number of spells and number of temporary deposit rate changes. This table shows the total numberof deposit rates spells, total number of uncensored spells (for which both the beginning and the end can beidentified) as well as the number of temporary deposit rate changes (sales) with durations of one, two, threeand four weeks

Product Totalnumberof spells

Total numberof uncensoredspells

Number of“sales” withone weekduration

Number of“sales” with2 weeksduration



Checking account 8084 5714 628 149 107 70

MMDA 14433 11814 1600 240 257 103

Table 2 Average duration of deposit rate spells and the average change of deposit rates. This table shows theaverage duration of deposit rates in weeks (column (1)) and the average absolute value of the magnitude of thedeposit rate changes in those weeks where the change is non-zero (column(2)). The magnitude of the averagedeposit rate change is set in relation to the average deposit rate level reported in column (3). The resulting ratioof average deposit rate change to the average deposit rate level is shown in column (4)

Product (1) (2) (3) (4)

Average duration(in weeks)

Averagechange (in %)

Averagerate (in %)

Average change relativeto average rate

Checking account 17.71 0.16 0.53 0.30

MMDA 12.76 0.26 1.07 0.24

CD 3 months 7.87 0.33 2.33 0.14

CD 12 months 6.08 0.35 2.96 0.12

J Financ Serv Res

the degree of lumpiness in the deposit-rate adjustment process, the examination of the effectof the mergers on the deposit rate’s duration and its determinants is of key importance tounderstanding deposit-rate dynamics.

3 The hazard of changing bank retail deposit rates in post-merger periods

I start the empirical examination by exploring the difference in the durations of the deposit-rate spells between the banks with mergers and banks without. For this purpose, I comparethe nonparametric Kaplan-Meier estimations of the hazard function of changing the depositrates in the first through fifth years following a merger. To accomplish the estimation, I addtime dummies that reflect the time to the latest merger of the bank.6 These time dummies areconstructed as follows. First, the merger date is defined as the date on which the charter ofthe target bank was withdrawn. This definition of the merger date is standard in the literature(see Hannan and Prager 1998; Focarelli and Panetta 2003).

Next, the time distance to the merger date is computed for each of the observations. Thenthe dummy variable merger 1 year is generated that takes the value of one if the bank hasundergone a merger in the last 12 months and zero otherwise. Similarly, the dummyvariables merger 2 years and merger 3 years are generated taking the value of one if thebank has undergone a merger in the last 13 to 24 months and in the last 25 to 36 monthsrespectively. And, a dummy merger 3 to 5 years is added that takes the value of one if thebank has undergone a merger 36 to 60 months prior to the observation time and zerootherwise.

The results of the estimated Kaplan-Meier hazard functions are presented in Fig. 1 and 2for the checking account and the MMDA rates respectively. For all of the hazard functionestimations, the reference group is the sample of non-merging banks. The estimates of thehazard functions present a graphical illustration of the frequency of changing the depositrates around a merger. In addition to this graphical illustration, a battery of formal statisticaltests on the estimated hazard functions confirm their inequality: the log-rank, the Cox, thePeto (see Peto and Peto 1972), and the Fleming-Harrington tests (see Schoenfeld 1981 andHarrington 2005). Table 3 presents the results of these tests.

This comparison points to two effects present for both the checking account and theMMDA rates. First, the probability of changing the deposit rates is higher for the banks thathave undergone a merger in the last 12 months than for non-merging banks. This differenceis visible from the significantly higher hazard for merging banks in the first 15 weeks of theduration of the deposit rate as illustrated in the upper left panel of both Fig. 1 and Fig. 2.These first panels show that the difference between the hazard of merging and non-mergingbanks for changing relatively new deposit rates is more than 1 %. For example, while theprobability of a spell changing in the fourth week of its “life” is less than 2.5 % for non-merging banks, it is about 3.5 % for a bank that has merged in the preceding year. Thisdifference corresponds to a relative increase in the probability to change a MMDA rate ofmore than 40 % (per week). Roughly the same economic magnitude is observed for any ofthe first 15 weeks of a spell’s duration.

6 The focus on the latest merger substantially reduces the number of mergers that are explored. The limitationis imposed in order to avoid the noise of overlapping time periods affecting the tightness of the estimatedcoefficients. As a robustness check I have rerun the estimations using up to three earlier mergers in theanalysis. Results stay qualitatively the same.

J Financ Serv Res

The evidence of a higher frequency of deposit-rate changes in the first post-merger year isconsistent with the existence of a transition period when the two merging institutions explorenew policies for setting the deposit rate taking into account the potential changes in the poolof depositors. The evidence also suggests that the merging banks change deposit rates innumerous steps.

The second effect starts with the third year after the merger. In the third year the hazard ofchanging both the checking account and the MMDA rates significantly drops below thehazards estimated for non-merging banks. The evidence is again confirmed both visually bythe range of low durations shown in the lower panels of Fig. 1 and 2 and by the statisticalsignificance of the difference in the estimated hazards in Table 3. The difference isparticularly pronounced for the hazards in changing the MMDA rate in the first post-merger year. Again the effect is not only statistically but economically significant because

Fig. 1 Smoothed hazard Kaplan-Meier estimates of checking account rate duration. This figure illustrates thesmoothed Kaplan-Meier hazard estimates for the checking account rate for four different time windowsfollowing a bank merger. The solid line refers to the estimated hazard of banks which have merged in therespective period, the dotted line refers to the estimated hazard for non-merging banks. The light grey linesdefine the respective 95 % confidence intervals. Analysis time is measured in weeks

J Financ Serv Res

the probability of a spell ending is much lower for merging banks in the first weeks of thespell’s life relative to non-merging banks. For example, for the MMDA rate, the hazard formerging banks in changing a deposit rate in the first weeks of its duration is less than 2.5 %,while its almost 4 % for non-merging banks. This difference corresponds to almost a 38 %decline in the conditional probability of changing deposit rates in a given week. This secondobservation is consistent with the argument presented in studies that treat the market’sstructure in a static manner where consolidation—as measured by large bank size and themarket power of a few banks—enhances the deposit rate’s rigidity (Hannan and Berger1991; Craig and Dinger 2010). These hazard-function estimates underline that this effectdoes not materialize immediately but instead only after a substantial period of frequent ratechanges in the first year. Even for the second post-merger year, I do not find a significantreduction in the probability of changing the deposit rates. The graphical illustration showsthat the hazard in changing the checking account rates in the second post-merger year is notsignificantly different from that of non-merging banks. For the MMDA, the graphs show

Fig. 2 Smoothed hazard Kaplan-Meier estimates of MMDA rate duration. This figure illustrates thesmoothed Kaplan-Meier hazard estimates for the checking account rate for four different time windowsfollowing a bank merger. The solid line refers to the estimated hazard of banks which have merged in therespective period, the dotted line refers to the estimated hazard for non-merging banks. The light grey linesdefine the respective 95 % confidence intervals. Analysis time is measured in weeks

J Financ Serv Res

that the hazard in changing the rates in the first weeks is lower during the second post-merger year, but starting from about the twentieth week the hazards are statistically notdistinguishable. Formally, the econometric tests in Table 3 cannot reject the equivalence ofthe estimated hazards for the non-merging banks and the merging banks in the second post-merger year.

However, when merging banks reset their deposit rates more often during the firstyear after the merger, the reset does not necessarily imply that the rates in thistransition period are more competitive. More frequent rate changes could actuallyemerge because the merging bank is testing the “limit” of its new pricing horizon.Indeed, studies like Hannan and Prager (1998) and Craig and Dinger (2009), whichuse the same data set as this study, show that the deposit rates of merging banks dropalmost immediately after the merger.

To shed more light on this issue, I examine the direction and the magnitude of deposit-rate changes in the first year after a merger by presenting the distribution of the non-zero ratechanges in Fig. 3.

These distributions show that for both the checking account and the MMDArates that the negative deposit-rate changes are more likely than positive ones inthe first post-merger year. This result is particularly pronounced when the generaldeposit-rate level as measured by the T-Bill rate is decreasing. However, the result

Table 3 Tests for the equivalence of the estimated hazards. This table presents the estimation results of thelog-rank, the Cox, the Peto and the Fleming-Harrington test for the equivalence of the estimated hazardfunctions. Column (1) reports the type of the test, column (2) lists the compared hazards, columns (3) and (4)report the p-values of the tests for the MMDA and the checking account rates, respectively

(1) (2) (3) (4)

Test Compared hazards p-value forthe MMDArates

p-value forthe checkingaccount rates

Log-rank test First year after the merger to non-mergingbanks

0.0044 0.0001

Second year after the merger tonon-merging banks

0.4038 0.1228

Third year after the merger to non-merging banks 0.0039 0.0000

3–5 years after the merger to non-merging banks 0.0001 0.0000

Cox test First year after the merger to non-merging banks 0.0074 0.0000

Second year after the merger to non-merging banks 0.4282 0.1271



Peto test First year after the merger to non-merging banks 0.0055 0.0000

Second year after the merger to non-merging banks 0.2249 0.1096


Three to 5 years after the merger to non-mergingbanks

0.0002 0.0000

Fleming-Harringtontest

First year after the merger to non-merging banks 0.0193 0.0000

Second year after the merger to non-mergingbanks

0.2739 0.1203



J Financ Serv Res

is also present when the market-deposit rates are increasing. These observationstherefore confirm the findings of earlier studies (Hannan and Prager 1998; Craigand Dinger 2009) that also show negative changes in the deposit-rate levels afterbank mergers.

4 Bank mergers, changes to bank and market structures, and the hazard in changingthe deposit rates

The Kaplan-Meier nonparametric estimates of the hazard functions indicate a significantrelation between mergers and the frequency of changing deposit rates. However, althoughsuitable for illustrating the basic link between bank mergers and the deposit rate’s rigidity,the simple univariate Kaplan-Meier framework is unsuitable for the identification of the

02

46

810

-2 -1 0 1 2

02

46

810

Den

sity

-.3 -.2 -.1 0 .1 .20

24

68

10

Den

sity

-2 -1 0 1 2

0.5

11.

52

-2 -1 0 1 2

0.5

11.

52

Den

sity

-2 -1 0 1 2

0.5

11.

52

Den

sity

-2 -1 0 1 2

Magnitude of checking account ratechanges in the first post-mergeryear:

a) sample including all observationsof merging banks;

b) observations of merging bankswhen the market interest rate (3-months T-Bill rate) changes in anegative direction;

c) observations of merging bankswhen the market interest rate (3-months T-Bill rate) changes in apositive direction

Magnitude of MMDA rate changes

in the first post-merger year:

d) sample including all observations

of merging banks;

e) observations of merging banks

when the market interest rate (3-

months T-Bill rate) changes in a

negative direction;

f) observations of merging banks

when the market interest rate (3-

months T-Bill rate) changes in a

positive direction

a b c

d e f

Fig. 3 Distribution of post-merger rate changes depending on the direction of T-Bill rate changes. This figureillustrates the distribution of checking account (upper panel) and MMDA rate (cpb) changes by mergingbanks in the first year following the merger. For each type of deposit rates different subpanels depending onthe direction of T-Bill rate dynamics are presented

J Financ Serv Res

channels and determinants of a merger’s impact. For this purpose, I estimate a proportionalCox hazard model of the general form7:

h t xijt��

� � ¼ h0 tð Þexp xijtβx

� � ð1Þwhere h0(t) denotes the baseline hazard, xj is the vector of the covariates, and βx are theregression coefficients to be estimated from the data. A major advantage to the Cox model isthat it requires no parameterization of the baseline hazard function. The model solelyassumes that the value of the covariates ceteris paribus proportionally shifts the baselinehazard. However, the standard Cox model can produce biased results because of theomission of the unobserved common component across the observations of different marketsfor the same bank. To this end, I estimate the model with a shared frailty option that controlsfor the existence of a random bank-specific effect.

The choice of variables added to the vector of the covariates xijt builds on the empiricalmodel presented by Craig and Dinger (2010) that examines the role of wholesale ratedynamics and static measures of bank and market characteristics. As in this study, theirempirical model is based upon the argument stemming from the theoretical literature on thesources of price rigidity. This argument claims that the decision to change depends on thetradeoff between the costs of the deviation from an unobservable optimal price level and thecosts of adjusting the price (menu costs) to this optimal level. I approximate the deviationfrom the optimal deposit rate by the cumulated change in the market’s interest rate betweenthe observation time t and the time of the latest change in the deposit rate. This approxima-tion is based on the reasoning that the unobservable optimal deposit rate is a function of themarket’s interest rate. I measure the market’s interest rate by the 3-month T-bill rate. Toaccount for the potential asymmetry of the interest-rate adjustments (see, Hannan and Berger1991; Neumark and Sharpe 1992; Craig and Dinger 2010), I also add a dummy for a positiveT-Bill rate change and the interaction of this dummy with the absolute value of thecumulated T-Bill rate change as a covariate.8

The market’s structure affects the frequency of deposit-rate changes because in theabsence of perfect competition, the reaction of the optimal deposit rate to changes in themarket’s interest rate is modified by the characteristics that indicate the market power of thebank. To account for this, I add bank size as measured by the natural logarithm of the bank’stotal assets; the market share of the bank in the respective market computed as the ratio ofthe bank’s deposits to the total sum of deposits in the MSA; the number of markets given byan integer reflecting the number of local MSAs represented in the sample where the bank hasa branch; and the Herfindahl-Hirshman Index of the MSA that controls for the level of thegeneral concentration in the local market.

Table 4 presents the descriptive statistics of the bank, the local market, and the mergercharacteristics in the sample. These statistics illustrate a large variation in the characteristicsof the sample banks. For example, the sample includes banks with a total asset value of lessthan a billion to more than a trillion USD. The average market share of the sample banks in

7 The Cox models applied throughout the paper cluster standard errors at the bank-market level in order tocontrol for the repeatability of the failure events arising from the fact that for each bank-market unit ofobservation multiple failures (changes in the deposit rate) can be observed.8 The T-Bill rate is used for the sake of comparability with other studies exploring retail rate reactions tomarket interest-rate changes. However, from a monetary policy perspective, the effect of the adjustment speedto changes in the monetary policy could be of key interest. I have rerun the estimations using the averageeffective Fed funds rate as a marginal costs proxy instead of the T-Bill rate. The results of these estimations arequalitatively the same as the one using the T-Bill rates and have not been reported in the text for the sake ofparsimonious exposition.

J Financ Serv Res

the sample MSAs is 14 %, and the average Herfindahl-Hirshman Index value is 0.16;although again large variations across banks and markets are observed.

Next, I add a set of variables that measure the key dimensions of the mergers to thecovariates’ vector. The estimation of the effect of these variables aims at the dynamicidentification of the effect of the market’s structure on the deposit rate’s rigidity but alsoat the identification of those merger characteristics that most substantially affect the fre-quency of the changes.

To start with, a bank merger changes the size of the bank. To this end, I add the change inthe bank’s size as measured by the natural logarithm of the target bank’s total assets(target_size) as a covariate. The change in the bank’s size generated by the merger canaffect the frequency of changing deposit rates by affecting both the adjustment costs and thecosts of deviating from the optimal rate. If the deposit-rate adjustment costs have a fixedcomponent independent of the volume of deposits (such as the costs from reviewing thecompetitor’s rate and the management costs from making the decision to change), then anincrease in the size of the bank might reduce the relative weight of the adjustment costs andresult in more frequent adjustments. On the other hand, the growing size of the bank mightinhibit the frequent changes because of the challenges of coordinating the decisions tochange both across branches and across deposit products. Also, as pointed out by Parkand Pennacchi (2009), larger banks have access to more diversified sources of financing. Forthese banks, interest on retail deposits can represent only a minor share of the bank’s totalexpenses. If this is the case, large banks could avoid the costs of adjusting retail liabilitiessimply because these are of minor importance to their funding costs. Therefore, the incen-tives to adjust deposit rates to competitors’ benchmarks can depreciate when a mergergenerates substantial growth in the bank’s balance sheet.

Next, bank mergers change the market share of the bank in those local markets whereboth the merging and the target banks operated prior to the merger. Market shares affect thepressure to adjust to shocks in the money market rates because they determine market powerand thus the reaction of the optimal deposit rates to changes in the money market rates. Thisis a well-documented effect (Rotemberg and Saloner 1987; Hannan and Berger 1991;Neumark and Sharpe 1992; Craig and Dinger 2010). The effect’s theoretical foundationsare motivated by the argument that the lower the elasticity of the supply of deposits (that isthe case for banks with stronger market power), the lower the probability that a bank willreact to a change in the money market rates by changing its deposit rates.

By adding the change in market share caused by the merger as a covariate, I examinewhether this result also holds in a dynamic sense, that is, does the deposit rate’s rigidity

Table 4 Descriptive statistics of bank and market characteristics. This table reports the descriptive statisticsof the key bank and market characteristics employed in the empirical analysis. Bank size is defined as thevalue of total bank assets in billions of USD; market share is the share of the deposits of a given bank in thetotal volume of deposits in a given local market (defined as MSA); herfindahl is the value of the herfindahlindex of the local deposit market’s (again defined as MSAs) concentration and number of markets is a variableequal to the number of local markets (MSAs) where a bank is operating

Variable Number of observations Mean Standard deviation Minimum Maximum

Bank size (bill USD) 140242 148 237 0 1100

Market share 140242 0.143 0.104 0.000 0.872

Herfindahl 140242 0.161 0.071 0.051 0.773

Number of markets 140242 25.365 26.200 1.000 110.000

J Financ Serv Res

increase after a merger that causes a substantial change in the bank’s market share. Becausethe data on the change in the market share is directly related to the merger for each of theaffected local markets is not available, I can approximate this change with the change ofmarket share realized in the year of the merger. I approximate this change in market share asthe difference between the bank’s market share in the years before and after the mergernormalized by the pre-merger year’s market share.9

Also, a substantial number of bank mergers involve an expansion in the geographicalcoverage of the bank’s operations that can be measured by the change in the number ofmarkets in which the bank operates. The number of markets might significantly affect itspricing behavior, because banks that create competitive pressure in one of the markets byquickly adjusting deposit rates might fear a competitor’s response in numerous other marketswhere the adjustment is not desirable. The industrial organization literature calls thishypothesis the linked-oligopoly hypothesis (Bernheim and Whinston 1990). In addition tothe effects described by this hypothesis, the rise in the number of markets increases thecoordination efforts across different geographical divisions of the bank and can thereforealso lead to less frequent changes in the deposit rates. In order to estimate the effect of themarket-extension dimension of the mergers, I add the change in the number of local marketsdivided by the number of markets prior to the merger as a regressor. As with the change inthe market share, I approximate the change in the number of local markets with thedifference between the number of markets in which a bank operates in the years after andbefore the merger.

The target size as well as the change in the market share and the change in the number oflocal markets summarize the three main dimensions of the dynamics in bank and marketstructures generated by the merger. Table 5 presents the summary statistics for the keymerger features and illustrates how modern bank mergers substantially vary in their nature.

This variation strengthens the identification of the dynamic empirical approach presentedhere and allows me to empirically disentangle the effect of the various changes in thedimensions of the bank and market structures on the deposit rate’s rigidity.

I control for the peculiarities in the frequency of the deposit-rate changes in the yearsfollowing the merger by adding to the covariates a vector of time dummies related to thetime elapsed to and from the merger. This vector contains the dummy variables for the firstand the second years after the merger as well as a dummy for the period of 3 to 5 years afterthe merger.10 The vector also includes a pre-merger dummy taking the value of one if thebank is merging with another bank in the following year and zero otherwise.11

4.1 Empirical results

Tables 6 and 7 present the results of the Cox proportional hazard estimations for thechecking account and the MMDA rates respectively.

These results illustrate that, among the various characteristics of the merger, the targetsize and the change in the number of new local markets are the features that are most closely

9 The Summary of Deposits publishes the market shares as of June 30; therefore, I define the year in this caseas the period of July 1 to June 30.10 As robustness checks I have rerun the estimations using linear splines for the time distance to the latestmergers. Results are qualitatively the same.11 This is to reflect the fact that the merger date reported in the data, which is the date when the charter of thetarget bank is revoked is usually preceded by a period of merger preparations that might be reflected in theinterest rate setting policy of the bank.

J Financ Serv Res

related to deposit rate’s rigidity. Both of these features are associated with a significantlyreduced ceteris paribus frequency of changing the deposit rates.

The economic impact of these estimated coefficients is quite substantial. One standarddeviation in the target size (equal in natural log terms to 1.9 and in levels to USD 82 bil.)reduces the probability of changing the checking account rates by roughly 24.13 % and theprobability of changing the MMDA rate by 15.2 % per week. One standard deviation in thechange in the number of markets (equal to 0.51) corresponds to about a 6 % lowerprobability of changing the checking account rates in each of the weeks of a deposit-ratespell. The corresponding probability of changing the MMDA rate in each of the weeks isreduced by slightly more than 3 %.

These results show that the mergers that increase the size and the geographical scopes ofthe bank are associated with a substantial reduction of the frequency of changes in thedeposit rates. They suggest that the increased complexity of these banking institutionssubstantially affects their rate setting policies and creates more rigid deposit rates. If thisfinding is the case, then the recent wave of bank mergers should have important implicationsnot only for the level of interest rates but also for the speed of adjustment to banks’ depositrates to changes in the monetary policy rate. These results also confirm in a dynamic context,which allows better identification, the findings of earlier studies that are based on a statictreatment of the market’s structure.

The coefficient of the third merger feature, the change in the market share, is statisticallyinsignificant. The interpretation of this result at face value implies that the change in themarket share does not substantially affect the rigidity of the deposit rates. This resultcontradicts the finding of earlier studies on the positive relation between market share andthe deposit rate’s rigidity (e.g., Hannan and Berger 1991; Neumark and Sharpe 1992).

However, the lack of statistical significance could also explain the heterogeneity in thesample. In particular, the change in the market share might have different effects on theduration of the deposit rate in relation to the concentration of the markets where the mergingbanks operate. Another potential explanation for the statistical insignificance of the changein the market share coefficient is that the variation in the change in the market share variableis much smaller than the variation in the other two variables describing the merger.Nevertheless, the insignificance of the change in the market share and the significance ofthe change in the bank’s size and the change in the number of local markets points to thecomplexity of the decision to change rather than to the increased market power as the maindriver of the changes in the deposit rate’s rigidity in response to the merger.

The dummies for the time distance to the merger enter the regressions with coefficientsthat are generally consistent with the pattern in the Kaplan-Meier estimates of the hazard

Table 5 Descriptive statistics of merger characteristics. This table reports the descriptive statistics of the keycharacteristics of bank mergers. Target size is the size of the acquired bank measured by the total assets inbillions USD. The change market share is the increase of the deposit market share of the acquiring bank perlocal market (MSA) generated in the year of the merger normalized by the pre-merger market share of theacuiring bank. The change number of markets is the change (during the year of the merger) in the number oflocal markets (MSAs) where the acquiring bank operates normalized by the pre-merger number of markets ofthe acquiring bank

Merger chracteristics Number of mergers Mean Standard deviation Minimum Maximum

Target size (bill USD) 121 111 148 0 655

Change market share 121 0.001 0.032 0.000 0.177

Change number of market 121 0.343 0.635 −0.250 5.000

J Financ Serv Res

functions. The estimation of the full model (including the features of the merger) confirmsthe finding that both the checking account and the MMDA rates become less flexible a fewyears after the merger. This result is also consistent with the effect of the banks’ post-mergercomplexity on an increased rigidity. This result is, however, not confirmed by the moreparsimonious estimations (that do not include the merger control variables). In this case, therelevant coefficients of the merger 2 years and merger 3 and more years are not statisticallysignificant. The short-run effects of the merger enter the estimation of the full model withonly limited statistical significance. This result suggests that the variation in the variables forthe merger’s features and the market’s structure dominates the effect of the time dummies.

Table 6 Checking account rate duration: Cox proportional hazard estimates. This table shows the results ofsemi-parametric Cox proportional hazard estimates of the hazard of changing the checking account rate.Premerger is a dummy variable taking the value of one if the bank is merging with another bank in thefollowing year. Merger 1 year, merger 2 years, merger 3 and more years are dummy variables equal to one if abank has merged in the last 1–5 years, respectively. Target size is the size of the acquired bank measured bythe total assets in billions USD. The change market share is the increase of the deposit market share of theacquiring bank per local market (MSA) generated in the year of the merger normalized by the pre-mergermarket share of the acuiring bank. The change number of markets is the change (during the year of the merger)in the number of local markets (MSAs) where the acquiring bank operates normalized by the pre-mergernumber of markets of the acquiring bank. Absolute change wholesale rate is the absolute value of the changein the three-month T-Bill rate. Dummy for negative change is equal to one if the 3-month T-Bill rate haschanged in a negative direction and zero otherwise. Negative change*absolute change is the cross-product ofthe absolute change wholesale rate and the dummy for negative change variables. Bank size is defined as thevalue of total bank assets in billions of USD; market share is the share of the deposits of a given bank in thetotal volume of deposits in a given local market (defined as MSA); herfindahl is the value of the herfindahlindex of the local deposit market’s (again defined as MSAs) concentration and number of markets is a variableequal to the number of local markets (MSAs) where a bank is operating. Hazard ratios higher than unitysuggest an increased hazard in changing the deposit rate. Hazard ratios lower than unity indicate a lowerprobability of changing the deposit rate and thus more rigid rates

Without merger controls with merger controls

Hazard ratio Standard error Hazard ratio Standard error

Premerger 0.885 0.069 0.803 *** 0.064

Merger 1 year 1.132 *** 0.053 0.999 0.051

Merger 2 years 0.987 0.055 0.852 *** 0.051

Merger 3 and more years 0.954 0.035 0.756 *** 0.043

Target_size 0.876 *** 0.029

Change in market share 0.921 0.404

Change in number of markets 0.861 *** 0.027

Absolute change wholesale rate 0.814 * 0.094 0.986 0.131

Dummy for negative change 1.580 *** 0.059 1.573 *** 0.065

Negative change*absolute change 1.702 *** 0.096 1.708 *** 0.106

Bank size 1.075 *** 0.013 1.211 *** 0.044

Market share 0.723 ** 0.111 0.758 * 0.124

Herfindahl 1.214 0.260 1.286 0.295

Number of markets 0.988 *** 0.001 0.989 *** 0.001

#Spells 6483 5388

# Failures 4581 3818

LR Chi(2) 897.49 888.86

*, **, *** indicate statistical significance at the 10 %, 5 % and 1 % levels, respectively

J Financ Serv Res

The estimated effect of the control variables for the changes in the market’s interest rate isconsistent with the results of earlier studies that find an asymmetric adjustment pattern(Hannan and Berger 1991; Craig and Dinger 2010). I also find that the bank’s size affects theprobability of a positive change in the deposit rates, but this is a ceteris paribus result thatshould be interpreted jointly with the negative effect of the bank’s market share and thenumber of markets. Quite surprisingly market concentration—as measured by theHerfindahl-Hirschman index—enters all of the Cox regressions with statistically insignifi-cant coefficients.

Table 7 MMDA rate duration: Cox proportional hazard estimates. This table shows the results of semi-parametric Cox proportional hazard estimates of the hazard of changing the MMDA rate. Premerger is adummy variable taking the value of one if the bank is merging with another bank in the following year. Merger1 year, merger 2 years, merger 3 and more years are dummy variables equal to one if a bank has merged in thelast 1–5 years, respectively. Target size is the size of the acquired bank measured by the total assets in billionsUSD. The change market share is the increase of the deposit market share of the acquiring bank per localmarket (MSA) generated in the year of the merger normalized by the pre-merger market share of the acuiringbank. The change number of markets is the change (during the year of the merger) in the number of localmarkets (MSAs) where the acquiring bank operates normalized by the pre-merger number of markets of theacquiring bank. Absolute change wholesale rate is the absolute value of the change in the 3-month T-Bill rate.Dummy for negative change is equal to one if the 3-month T-Bill rate has changed in a negative direction andzero otherwise. Negative change*absolute change is the cross-product of the absolute change wholesale rateand the dummy for negative change variables. Bank size is defined as the value of total bank assets in billionsof USD; market share is the share of the deposits of a given bank in the total volume of deposits in a givenlocal market (defined as MSA); herfindahl is the value of the herfindahl index of the local deposit market’s(again defined as MSAs) concentration and number of markets is a variable equal to the number of localmarkets (MSAs) where a bank is operating. Hazard ratios higher than unity suggest an increased hazard inchanging the deposit rate. Hazard ratios higher than unity suggest an increased hazard in changing the depositrate. Hazard ratios lower than unity indicate a lower probability of changing the deposit rate and thus morerigid rates

Without merger controls with merger controls


Premerger 0.935 0.047 0.941 0.048

Merger 1 year 1.027 0.032 1.010 0.034

Merger 2 years 1.056 0.036 1.027 0.037

Merger 3 and more years 1.005 0.026 0.912 ** 0.035

Target_size 0.916 *** 0.020

Change in market share 0.642 0.198

Change in number of markets 0.911 *** 0.019

Absolute change wholesale rate 1.887 *** 0.186 1.729 *** 0.185



Bank size 1.061 *** 0.009 1.156 *** 0.028

Market share 0.786 *** 0.080 0.659 *** 0.073

Herfindahl 1.017 0.147 1.001 0.159


# Spells 12690 10375

# Failures 10579 8648

LR Chi(2) 1050.82 928.76


J Financ Serv Res

4.2 Subsamples of highly and less concentrated markets

As illustrated in Table 4, the local banking markets in the sample exhibit substantialheterogeneity with regard to concentration and therefore the relation between bank mergersand rigidity is likely to differ substantially from market to market. There are at least tworeasons for this phenomenon. First, according to Hannan and Berger (1991), highly con-centrated markets allow banks to more easily collude by not adjusting their deposit rates tothe monetary policy rate. Further, the relative increase in market power generated throughthe merger depends on the distribution of the market shares in the local market. To addressthis issue, I present another set of regressions in which the sample of bank-market observa-tions is divided into two subsamples (highly concentrated and less concentrated localmarkets) depending on the Herfindahl-Hirschman index of the local market. As a cut-off point for the Herfindahl-Hirschman index, I use the critical value of the marketconcentration used by the U.S. Department of Justice in the evaluation of bank mergers.This value is 0.18.

Tables 8 and 9 report the results of these estimations for the checking account and theMMDA rates respectively. The estimated coefficients for the variables measuring thefeatures of the merger substantially vary between highly and less concentrated markets.While target size does not have any significant impact on the deposit rate’s rigidity in highlyconcentrated markets, it substantially enhances the rigidity of deposit rates in less concen-trated markets. This difference could be explained by the idea that in less concentratedmarkets only large banks can resist adjusting deposit rates because of shocks in themoney market rate. The difference is also consistent with the idea that large banks areless willing to adjust their deposit rates in markets where their market presence is notsubstantial.

Also, the estimated effect of the change in the market share 12 differs significantly acrosshighly and less concentrated markets. For the duration of the checking account rate, theestimated effect of the change in the market share is in opposite directions in the highly andless concentrated markets. While the change in the market share increases the frequency ofresetting checking account rates in less concentrated markets, it decreases this frequencysubstantially in highly concentrated markets. These empirical observations could beexplained by the delayed changes in the development of the market’s interest rate that couldonly be sustained in a highly concentrated market where customers cannot easily switch tocompetitors.

For the duration of the MMDA rate, the change in the market share is not significantlyrelated to a change in the hazard in either of the two subsamples. This result could bebecause the MMDA rates are less dependent on the local market conditions than thechecking account rates (Hannan and Prager 2006).

Further, I find that for highly concentrated markets, the frequency of deposit-rate changessignificantly rises in the first post-merger year. This result is consistent with the idea thatimmediately after a merger, banks change their deposit rates very frequently in search of anew optimal rate. In highly concentrated markets, the likelihood that a merging bank canextract effects from their merger-generated market power is higher. Therefore, mergingbanks not surprisingly more actively adjust their deposit rates in those markets. This effect

12 A Wald test formally proves that the difference in the estimated coefficients of the target size and CMSvariables in the two subsamples is statistically significant.

J Financ Serv Res

is ceteris paribus not present in less concentrated markets. If the new optimal rate reflects thestrengthened market position of the merging banks, then a large share of the deposit-ratechanges immediately after the merger should be negative.

To shed more light on this question, I extend the evidence presented in Section 3 bytracing the difference in the distributions of the deposit-rate changes in less and highlyconcentrated local markets in different subperiods of the first post-merger year. Figure 4 andFig. 5 present these for the checking account and MMDA rates respectively.

Table 8 Duration of checking account rate: Cox proportional hazard estimates for subsamples of high andless concentrated local banking markets. This table shows the estimated semi-parametric Cox proportionalhazards for changing the checking account rate for highly concentrated markets with a Herfindahl-Hirshmanindex higher than 0.18 and for less concentrated markets with a Herfindahl-Hirshman index not exceeding0.18. Premerger is a dummy variable taking the value of one if the bank is merging with another bank in thefollowing year. Merger 1 year, merger 2 years, merger 3 and more years are dummy variables equal to one if abank has merged in the last 1–5 years, respectively. Target size is the size of the acquired bank measured bythe total assets in billions USD. The change market share is the increase of the deposit market share of theacquiring bank per local market (MSA) generated in the year of the merger normalized by the pre-mergermarket share of the acuiring bank. The change number of markets is the change (during the year of the merger)in the number of local markets (MSAs) where the acquiring bank operates normalized by the pre-mergernumber of markets of the acquiring bank. Absolute change wholesale rate is the absolute value of the changein the 3-month T-Bill rate. Dummy for negative change is equal to one if the three-month T-Bill rate haschanged in a negative direction and zero otherwise. Negative change*absolute change is the cross-product ofthe absolute change wholesale rate and the dummy for negative change variables. Bank size is defined as thevalue of total bank assets in billions of USD; market share is the share of the deposits of a given bank in thetotal volume of deposits in a given local market (defined as MSA); herfindahl is the value of the herfindahlindex of the local deposit market’s (again defined as MSAs) concentration and number of markets is a variableequal to the number of local markets (MSAs) where a bank is operating. Hazard ratios higher than unitysuggest an increased hazard in changing the deposit rate. Hazard ratios lower than unity indicate a lowerprobability in changing the deposit rate and thus more rigid deposit rates

Highly concentrated markets Less concentrated markets


Premerger 1.071 0.148 0.707 *** 0.069

Merger 1 year 1.355 *** 0.122 0.881 ** 0.055

Merger 2 years 1.107 0.120 0.768 *** 0.056

Merger 3 and more years 0.956 0.100 0.687 *** 0.048

Target_size 0.950 0.066 0.855 *** 0.032

Change in market share 0.283 * 0.181 2.194 * 1.280

Change in number of markets 0.852 *** 0.052 0.853 *** 0.032

Absolute change wholesale rate 1.234 0.306 0.938 0.150



Bank size 1.152 * 0.088 1.226 *** 0.051

Market share 0.725 0.157 0.725 0.182


# Spells 1648 3939


LR Chi(2) 243.93 670.27


J Financ Serv Res

The distributions of the changes in the checking account rates show that negative deposit-rate changes are particularly likely in highly concentrated markets. This result is stronger inthe period of up to a month after the merger but it also holds for the first post-merger year.Qualitatively the same pattern is also observed for the MMDA rates.

The results concerning the frequency of deposit rate changes after the first post-mergeryear are more ambiguous. While in less concentrated markets the probability of a depositrate change generally decreases starting from the second-post merger year, it does notsignificantly change in highly concentrated markets (with the exception of MMDA rates

Table 9 MMDA rate duration: Cox proportional hazard estimates for subsamples of high and less concen-trated local banking markets. This table shows the estimated semi-parametric Cox proportional hazards forchanging the MMDA rate for highly concentrated markets with a Herfindahl-Hirshman index higher than 0.18and for less concentrated markets with a Herfindahl-Hirshman index not exceeding 0.18. Premerger is adummy variable taking the value of one if the bank is merging with another bank in the following year. Merger1 year, merger 2 years, merger 3 and more years are dummy variables equal to one if a bank has merged in thelast 1–5 years, respectively. Target size is the size of the acquired bank measured by the total assets in billionsUSD. The change market share is the increase of the deposit market share of the acquiring bank per localmarket (MSA) generated in the year of the merger normalized by the pre-merger market share of the acuiringbank. The change number of markets is the change (during the year of the merger) in the number of localmarkets (MSAs) where the acquiring bank operates normalized by the pre-merger number of markets of theacquiring bank. Absolute change wholesale rate is the absolute value of the change in the three-month T-Billrate. Dummy for negative change is equal to one if the 3-month T-Bill rate has changed in a negative directionand zero otherwise. Negative change*absolute change is the cross-product of the absolute change wholesalerate and the dummy for negative change variables. Bank size is defined as the value of total bank assets inbillions of USD; market share is the share of the deposits of a given bank in the total volume of deposits in agiven local market (defined as MSA); number of markets is a variable equal to the number of local markets(MSAs) where a bank is operating. Hazard ratios higher than unity suggest an increased hazard in changingthe deposit rate. Hazard ratios lower than unity indicate a lower probability in changing the deposit rate andthus more rigid deposit rates

Highly concentrated markets Less concentrated markets


Premerger 1.046 0.096 0.895 * 0.055

Merger 1 year 1.115 * 0.069 0.972 0.039

Merger 2 years 1.153 ** 0.080 0.991 0.042

Merger 3 and 0.979 0.069 0.875 *** 0.040

Target_size 0.951 0.041 0.902 *** 0.023

Change in market 0.480 0.224 0.748 0.308

Change in number 0.860 *** 0.037 0.926 *** 0.022

Absolute change 1.727 *** 0.360 1.981 *** 0.259

Dummy for negative 1.528 *** 0.084 1.477 *** 0.050

Negative change* 1.179 * 0.114 1.275 *** 0.077

Bank size 1.095 * 0.053 1.179 *** 0.033

Market share 0.596 *** 0.093 0.763 * 0.123


# Spells 2981 7623


LR Chi(2) 250.04 714.27


J Financ Serv Res

where I even find a positive increase in the probability to change the MMDA rate in thesecond post-merger year).

In sum, these results illustrate how the effects of bank mergers on the frequency andmagnitude of deposit-rate changes differ across markets with different concentration levels.They also suggest that the interpretation of more frequent price changes as a signal of morecompetitive pricing as is typically done in the literature on the interest rate’s rigidity can beinconsistent with observed behavior, at least in the short-term after a merger.

5 Bank mergers and the rigidity of deposit rates offered by neighboring banks

In the preceding section, I show that bank mergers are associated with a significant change inthe duration of deposit rates offered by merging banks. However, bank mergers are poten-tially related not only to changes in the behavior of the merging banks but also to reactionsfrom their direct competitors.

In this section, I explore whether the conditional probability of non-merging bankschanging the deposit rates in a given local market is modified by the fact that other banksoperating in the same market merge.

For this purpose, I estimate the Cox proportional hazard model by introducing a newcovariate to the analysis. This new covariate is a binary variable taking the value of one forobservations of deposit rates of non-merging banks in local markets where a merger hastaken place in the last 12 months and zero otherwise. I assume that a merger has taken placein a market if either the acquirer or the target bank collects deposits in the respective market.This binary variable is added to the set of covariates presented in Section 4.

The results of the model are presented in Tables 10 and 11 for the checking account and theMMDA rates respectively. They indicate that for both rates the conditional probability ofchanging the deposit rates is positively related to the fact that merging banks operate in themarket. In other words, if any of the banks operating in a given local market have been involvedin amerger, the rest of the banks are likely to react to the changed environment bymore frequentchanges in their deposit rates in the first year following the merger. This effect is particularlystrong for the checking account rates, but it is economically insignificant for the MMDA rates.

These results are again consistent with the argument of Hannan and Prager (1998) thatchecking accounts react more strongly to local market conditions than MMDAs. The resultsare also related to the fact that a merger can in the short-run generate more flexible depositrates because all of the banks operating in the affected markets adjust to the changed marketconditions, which is particularly true for highly concentrated markets.

6 Robustness checks

In this section, I examine the robustness of the results presented in Section 4 in two majordirections. First, I explore the robustness of the results to defining the merger date as the dateof the merger announcement rather than the date of the formal completion of the merger.While the use of the completion date has the advantage that the results are more comparableto those of earlier studies, the use of the announcement date accounts for the possibility thatthe coordinated market behavior of the merging banks starts on this date. Second, I extendthe empirical framework by adding additional bank-level control variables: the bank’sliability structure, a proxy for deposit supply and a bank’s risk proxy. These control variablesare not in the baseline regressions presented in Section 4. The bank’s liability is measured by

J Financ Serv Res

the ratio of retail deposits to total assets. The underlying reason for this variable is thepossibility that banks that rely heavily on retail funding have more aggressive rates that arecharacterized by frequent changes. The share of senior population in the MSA is introduced

01

23

4

-.5 0 .5 1

01

23

4

Den

sity

-.5 0 .5 10

12

34

Den

sity

-.5 0 .5 1

01

23

-1 -.5 0 .5 1 1.5

01

23

Den

sity

-1 -.5 0 .5 1 1.5

01

23

Den

sity

-1 -.5 0 .5 1 1.5

01

23

4

-2 -1 0 1 2

01

23

4

Den

sity

-2 -1 0 1 2

01

23

4

Den

sity

-2 -1 0 1 2

Magnitude of checking account rate

change in 0 to 4 weeks after the

merger in:

a) sample including all local markets;

b) local markets with Herfindahl

index of less than 0.18;

c) local markets with Herfindahl

index of at least 0.18



merger in:

d) sample including all local markets;

e) local markets with Herfindahl


f) local markets with Herfindahl




merger in:

g) sample including all local markets;

h) local markets with Herfindahl


i) local markets with Herfindahl


a b c

d e f

g h i

Fig. 4 Distribution of changes in the checking account rate in the first-post merger year. This figure illustratesthe distribution of checking account rate changes by merging banks in different time windows following themerger. The upper panels (a–c) cover the changes in the first 4 weeks after the merger, the panels in the middle(d–f) cover the checking account rate changes observed in 5 to 26 weeks following the merger and the lowestpanels (g–i) illustrate the changes in 27 to 52 weeks after the merger. For each time window differentsubpanels depending on the local market concentration are presented

J Financ Serv Res

as a control for deposit supply. Here I follow Becker 2007 and Acharya and Mora 2012 whoshow that deposit supply is higher in markets with a larger share of senior population. I usethis deposit supply proxy to estimate the relation between deposit supply and the intensity of

01

23

45

-1 -.5 0 .5 1

01

23

45

Den

sity

-1 -.5 0 .5 10

12

34

5

Den

sity

-1 -.5 0 .5 1

0.5

11.

52

2.5

-1.5 -1 -.5 0 .5 1

0.5

11.

52

2.5

Den

sity

-1.5 -1 -.5 0 .5 1

0.5

11.

52

2.5

Den

sity

-1.5 -1 -.5 0 .5 1

0.5

11.

52

2.5

-1.5 -1 -.5 0 .5 1

0.5

11.

52

2.5

Den

sity

-1.5 -1 -.5 0 .5 1

0.5

11.

52

2.5

Den

sity

-1.5 -1 -.5 0 .5 1

Magnitude of MMDA rate change in

0 to 4 weeks after the merger in:

a) sample including all local markets;

b) local markets with Herfindahl


c) local markets with Herfindahl




d) sample including all local

markets;

e) local markets with Herfindahl


f) local markets with Herfindahl




g) sample including all local markets;

h) local markets with Herfindahl


i) local markets with Herfindahl


a b c

d e f

g h i

Fig. 5 Distribution of changes in the MMDA rate in the first-post merger year. This figure illustrates thedistribution of MMDA rate changes by merging banks in different time windows following the merger. Theupper panels (a–c) cover the changes in the first 4 weeks after the merger, the panels in the middle (d–f) coverthe MMDA rate changes observed in 5 to 26 weeks following the merger and the lowest panels (g–i) illustratethe changes in 27 to 52 weeks after the merger. For each time window different subpanels depending on thelocal market concentration are presented

J Financ Serv Res

bank re-pricing strategies. The riskiness of the bank is measured by the ratio of nonper-forming loans to total bank assets. This variable controls for the possibility that riskier bankscould be forced to react with more frequent changes.

Table 12 reports the results of the estimations of the Cox model based on the mergerannouncement date. All of the results regarding the merger characteristics are robust to thischange in the definition of the merger date. Also, the merger is associated to more frequentchanges in the deposit rate in the short-run and less frequent changes in the long-run after themerger. These effects are again more pronounced for checking account rates.

Table 13 presents the results of the estimation using an extended set of bank-controlvariables. These results indicate that the inclusion of the additional controls does not affectthe key results concerning the effect of the merger features. They show that neither theriskiness of the banks as measured by nonperforming loans nor the share of retail fundinghas an effect on the probability of a bank changing its deposit rate.13 The insignificance of

13 The insignificance of the bank risk measure could be related to the fact that I study only insured deposits.Earlier research, however, has found that even for insured deposits the riskiness of the bank is related to thedeposit rate level (Acharya and Mora 2012; Craig and Dinger 2013). I show here that while deposit rate levelsmight react to bank risk, the frequency of changing deposit rates is not likely to be related to bank riskiness.

Table 10 Mergers and the hazard of changing the deposit rates of neighboring banks: checking account rates.This table presents the semi-parametric Cox proportional hazard estimates for changing the checking accountrate of the banks operating in the same MSA as the merging banks. Merger in market in the last year is adummy variable equal to one if either the acquiring or the target bank of a merger that took place in the last12 months were operating in the respective market. Absolute change wholesale rate is the absolute value of thechange in the 3-month T-Bill rate. Dummy for negative change is equal to one if the three-month T-Bill ratehas changed in a negative direction and zero otherwise. Negative change*absolute change is the cross-productof the absolute change wholesale rate and the dummy for negative change variables. Bank size is defined asthe value of total bank assets in billions of USD; market share is the share of the deposits of a given bank in thetotal volume of deposits in a given local market (defined as MSA); herfindahl is the value of the herfindahlindex of the local deposit market’s (again defined as MSAs) concentration; and number of markets is avariable equal to the number of local markets (MSAs) where a bank is operating. Hazard ratios higher thanunity suggest an increased hazard in changing the deposit rate. Hazard ratios lower than unity indicate a lowerprobability in changing the deposit rate and thus more rigid deposit rates

Checking account rates Full sample Highly concentratedmarkets

Less concentratedmarkets

Hazardratio

Standarderror

Hazardratio

Standarderror

Hazardratio

Standarderror

Merger in market in the last year 1.246 *** 0.045 1.332 *** 0.088 1.212 *** 0.052

Absolute change wholesale rate 0.819 * 0.094 0.881 0.178 0.850 0.123

Dummy for negative change 1.575 *** 0.058 1.602 *** 0.110 1.565 *** 0.069

Negative change*absolute change 1.718 *** 0.097 1.540 *** 0.154 1.788 *** 0.122

Bank size 1.071 *** 0.013 1.066 *** 0.024 1.068 *** 0.016

Market share 0.727 ** 0.111 0.938 0.191 0.571 ** 0.132

Herfindahl 1.204 0.257

Number of markets 0.989 *** 0.001 0.990 *** 0.002 0.988 *** 0.001

# Spells 6483 1967 4754

# Failures 4581 1327 3254

LR Chi(2) 919.77 232.56 700.51


J Financ Serv Res

both measures indicates that the frequency of changes in the deposit rates is primarily drivenby organizational features of the bank and the market rather than by the riskiness and theliability structure of the institutions. The share of senior population – although insignificantin the MMDA results - is strongly statistically significant in the results regarding thefrequency of re-pricing checking account rates. This result suggests that banks are morefrequently re-pricing their checking account rates in markets where senior citizens representa large share of the population. This could be explained by a potentially different (relative toother demographic groups) sensitivity of seniors to deposit rate changes.

7 Conclusion

The effect from bank mergers on the level of retail interest rates is well-known. However, theeffect of bankmergers on the frequency of changing these rates has so far been ignored. Yet, thiseffect is important for both the understanding of their dynamics and the speed of monetary

Table 11 Mergers and the hazard of changing the deposit rates of neighboring banks: MMDA rates. Thistable presents the semi-parametric Cox proportional hazard estimates for changing the MMDA rate of thebanks operating in the same MSA as the merging banks. Merger in market in the last year is a dummy variableequal to one if either the acquiring or the target bank of a merger that took place in the last 12 months wereoperating in the respective market. Absolute change wholesale rate is the absolute value of the change in the 3-month T-Bill rate. Dummy for negative change is equal to one if the 3-month T-Bill rate has changed in anegative direction and zero otherwise. Negative change*absolute change is the cross-product of the absolutechange wholesale rate and the dummy for negative change variables. Bank size is defined as the value of totalbank assets in billions of USD; market share is the share of the deposits of a given bank in the total volume ofdeposits in a given local market (defined as MSA); herfindahl is the value of the herfindahl index of the localdeposit market’s (again defined as MSAs) concentration and number of markets is a variable equal to thenumber of local markets (MSAs) where a bank is operating. Hazard ratios higher than unity suggest anincreased hazard in changing the deposit rate. Hazard ratios lower than unity indicate a lower probability inchanging the deposit rate and thus more rigid deposit rates

MMDA rates Full sample Highly concentratedmarkets

Less concentratedmarkets

Hazardratio

Standarderror

Hazardratio

Standarderror

Hazardratio

Standarderror

Merger in market in the lastyear

1.044 ** 0.021 1.045 * 0.030 1.043 * 0.025

Absolute change wholesalerate

1.891 *** 0.185 1.753 *** 0.325 2.154 *** 0.259

Dummy for negative change 1.487 *** 0.039 1.473 *** 0.073 1.489 *** 0.046

Negative change*absolutechange

1.222 *** 0.057 1.189 ** 0.103 1.224 *** 0.068

Bank size 1.060 *** 0.009 1.049 *** 0.015 1.062 *** 0.010

Market share 0.791 ** 0.080 0.661 *** 0.095 0.964 0.141

Herfindahl 1.011 0.146

Number of markets 0.993 *** 0.001 0.994 *** 0.001 0.992 *** 0.001

# Spells 12690 3690 9271

# Failures 10579 2983 7596

LR Chi(2) 1050.03 255.64 822.64


J Financ Serv Res

policy transmission. In this paper, I show that the changes in the bank and the marketcharacteristics generated by the mergers are related to substantial modifications of the frequencyof deposit-rate changes. The analysis is based on a comprehensive data set that reflects thedeposit-rate dynamics with a weekly frequency. The high frequency of the data allows me toexplore the deposit-rate dynamics and their determinants by using a duration analysis.

The results of the duration analysis imply an economically and statistically significantrelation between bank mergers and the probability of a bank changing its deposit rates. In

Table 12 Cox proportional hazard estimates: merger announcement date. This table presents the semi-parametric Cox proportional hazard estimates for changing the checking account and the MMDA rate whenthe merger date is defined as the date of the merger announcement. Premerger announcement is a dummyvariable taking the value of one if the bank announces a merger with another bank within the following year.Merger 1 year announcement, merger 2 years announcement, merger 3 and more years announcement aredummy variables equal to one if a bank has announced a merger in the last 1–5 years, respectively. Target sizeis the size of the acquired bank measured by the total assets in billions USD. The change market share is theincrease of the deposit market share of the acquiring bank per local market (MSA) generated in the year of themerger normalized by the pre-merger market share of the acuiring bank. The change number of markets is thechange (during the year of the merger) in the number of local markets (MSAs) where the acquiring bankoperates normalized by the pre-merger number of markets of the acquiring bank. Absolute change wholesalerate is the absolute value of the change in the three-month T-Bill rate. Dummy for negative change is equal toone if the three-month T-Bill rate has changed in a negative direction and zero otherwise. Negativechange*absolute change is the cross-product of the absolute change wholesale rate and the dummy fornegative change variables. Bank size is defined as the value of total bank assets in billions of USD; marketshare is the share of the deposits of a given bank in the total volume of deposits in a given local market(defined as MSA); herfindahl is the value of the herfindahl index of the local deposit market’s (again definedas MSAs) concentration and number of markets is a variable equal to the number of local markets (MSAs)where a bank is operating. Hazard ratios higher than unity suggest an increased hazard in changing the depositrate. Hazard ratios lower than unity indicate a lower probability in changing the deposit rate and thus morerigid deposit rates

Checking account rates MMDA rates


Premerger announcement 0.823 0.102 0.852 ** 0.059

Merger 1 year announcement 1.323 *** 0.078 1.090 ** 0.042

Merger 2 years announcement 1.188 *** 0.071 1.153 *** 0.042

Merger 3 and more years announcement 0.899 ** 0.067 0.942 ** 0.046

Target_size 0.976 0.025 0.948 *** 0.015

Change in market share 0.796 0.381 0.639 0.198


Absolute change wholesale rate 1.025 0.141 1.693 *** 0.180



Bank size 1.091 *** 0.035 1.117 *** 0.022

Market share 0.715 ** 0.122 0.659 *** 0.073

Herfindahl 1.260 0.296 0.982 0.156


# Spells 5388 10375


LR Chi(2) 810.30 943.60


J Financ Serv Res

particular, I find that merging banks change deposit rates relatively often for a period ofroughly 1 year after the merger. Only after about 2 years does the duration of the depositrates offered by merging banks become significantly longer than the duration of the depositrates offered by non-merging banks. This effect is particularly strong for banks that havesubstantially expanded their size and the geographical scope of their operations through the

Table 13 Cox proportional hazard estimates: additional bank-level controls. This table presents the semi-parametric Cox proportional hazard estimates for changing the checking account and the MMDA rate whenthe set of bank-level control variables is expanded to include the ratio of deposits to total assets, the share ofsenior population in the total population of the MSA and bank risk as measured by the ratio of non-performingloans to total assets. Premerger is a dummy variable taking the value of one if the bank is merging with anotherbank in the following year. Merger 1 year, merger 2 years, merger 3 and more years are dummy variablesequal to one if a bank has merged in the last 1–5 years, respectively. Target size is the size of the acquired bankmeasured by the total assets in billions USD. The change market share is the increase of the deposit marketshare of the acquiring bank per local market (MSA) generated in the year of the merger normalized by the pre-merger market share of the acuiring bank. The change number of markets is the change (during the year of themerger) in the number of local markets (MSAs) where the acquiring bank operates normalized by the pre-merger number of markets of the acquiring bank. Absolute change wholesale rate is the absolute value of thechange in the 3-month T-Bill rate. Dummy for negative change is equal to one if the 3-month T-Bill rate haschanged in a negative direction and zero otherwise. Negative change*absolute change is the cross-product ofthe absolute change wholesale rate and the dummy for negative change variables. Bank size is defined as thevalue of total bank assets in billions of USD; market share is the share of the deposits of a given bank in thetotal volume of deposits in a given local market (defined as MSA); herfindahl is the value of the herfindahlindex of the local deposit market’s (again defined as MSAs) concentration and number of markets is a variableequal to the number of local markets (MSAs) where a bank is operating. Hazard ratios higher than unitysuggest an increased hazard in changing the deposit rate. Hazard ratios lower than unity indicate a lowerprobability in changing the deposit rate and thus more rigid deposit rates

Checking account rates MMDA rates


Premerger 0.802 *** 0.064 0.941 0.048

Merger 1 year 1.001 0.051 1.010 0.034

Merger 2 years 0.854 *** 0.052 1.024 0.037

Merger 3 and more years 0.755 *** 0.043 0.914 ** 0.035

Target_size 0.876 *** 0.029 0.916 *** 0.020

Change in market share 0.922 0.405 0.647 0.200


Absolute change T-Bill rate 0.980 0.131 1.729 *** 0.185

Dummy for negative changes 1.571 *** 0.065 1.491 *** 0.043


Bank size 1.222 *** 0.048 1.162 *** 0.030

Market share 0.762 * 0.125 0.658 *** 0.073

Herfindahl 1.293 0.297 1.005 0.160


Deposits to total assets 1.106 0.187 1.070 0.123

Non-performing loans 0.000 0.000 0.000 0.000

# Spells 5388 10375


LR Chi(2) 890.27 929.67


J Financ Serv Res

merger. The change in market share inhibits the flexibility in the deposit rate mostly inhighly concentrated local markets. These are also the markets where the competitors respondto the merger with particularly high frequencies of rate changes.

I also show that the frequency of deposit-rate changes is not necessarily associatedwith changes that are beneficial to the consumers. In most cases the opposite is truewith banks changing their retail rates after the merger mostly in a negative directionespecially in the more concentrated markets where they operate. In the long-run, thenegative effect of the merger on the level of deposit rates emerges by the fact that thebanks are particularly slow to adjust their rates to positive shocks to the market’sinterest rate.

This evidence strengthens the validity of the previous results in the literature on retailinterest-rate dynamics by presenting evidence based a dynamically identified estimation.The evidence also sheds new light on the discussion of the effects of the recent (postfinancial-crisis) wave of bank-market consolidation on interest-rate dynamics. This evidencesuggests that the speed of adjustment to changes in the monetary policy might be affectedsubstantially by the mergers especially in those markets where the market concentration wasalready high prior to the merger.

References

Acharya V, Mora N (2012) Are Banks Passive Liquidity Backstops? Deposit Rates and Flows during 2007–2009 Crisis. NBERWorking Paper 17838

Becker B (2007) Geographical Segmentation of US Capital Markets. J Financ Econ 85(1):151–178Bernheim D, Whinston MD (1990) Multimarket contact and collusive behavior. RAND Journal of Econ

21(1):1–26Craig B, Dinger V (2009) Bank mergers and the dynamics of deposit interest rates. J Financ Serv Res 36(2–

3):111–133Craig, B, Dinger V (2010) A microeconometric investigation into bank interest rate rigidity. Federal Reserve

Bank of Cleveland Working Paper 2010–01Craig B, Dinger V (2013) Deposit market competition, wholesale funding, and bank risk. J Bank & Financ

37(9):3605–3622De Graeve F, De Jongh O, Vennet RV (2007) Competition, transmission and bank pricing policies: Evidence

from Belgian loan and deposit markets. J Bank & Finance 31(1):259–278Driscoll J C, Judson A R (2009) Sticky deposit rates: Data and implications for models of price adjustment,

mimeo, Federal Reserve BoardErel I (2011) The effect of bank mergers on loan prices: Evidence from the United States. Rev Financ Stud

24(4):1068–1101Focarelli D, Panetta F (2003) Are mergers beneficial to consumers? Evidence from the Italian market for bank

deposits. Amer Econ Rev 93:1152–1172Hannan T, Berger A (1991) The rigidity of prices: Evidence from the banking industry. Amer Econ Rev

81(4):938–945Hannan T, Prager R (1998) Do substantial horizontal mergers generate significant price effects? Evidence

from the banking industry. J Indus Econ 46:433–452Hannan T, Prager, R (2006) Multimarket bank pricing: An empirical investigation of deposit interest rates. J

Econ Bus 58(3):256–272Harrington D (2005) Linear rank tests in survival analysis. Encyclopedia of Biostatistics, Wiley Interscience,

TorontoHaubrich J (1992) Sluggish deposit rates: Endogenous institutions and aggregate fluctuations. Working Paper,

Federal Reserve Bank of ClevelandHofmann B, Mizen P (2004) Interest rate pass through in the monetary transmission mechanism: UK Banks’

and building societies’ retail rates. Economica 71:99–125Neumark D, Sharpe S (1992) Market structure and the nature of price rigidity: Evidence from the market for

consumer deposits. Quar J Econ 107(2):657–680

J Financ Serv Res

Ors E, Rice T (2007) Bank imputed interest rates: Unbiased estimates of offered rates? CEPR DiscussionPapers 6036, Centre for Economic Policy Research

Park K, Pennacchi G (2009) Harming depositors and helping borrowers: the disparate impact of bankconsolidation. Rev of Fin Stud 22(1):1–40

Peto R, Peto J (1972) Asymptotically efficient rank invariant test procedures. J Royal Stat Soc Ser A (General)135(2):185–207

Rotemberg J, Saloner G (1987) The relative rigidity of monopoly pricing. Amer Econ Rev 77(5):917–926Schoenfeld D (1981) The asymptotic properties of nonparametric tests for comparing survival distributions.

Biometrika 68:316–319

J Financ Serv Res

Documents

Bank Mergers and Deposit Rate Rigidity