Liquidity management and overnight rate calendar effects: Evidence from German banks

Available online at www.sciencedirect.com

North American Journal of Economics and Finance 19 (2008) 7–21

Liquidity management and overnight rate calendar effects: Evidence from German banks Falko Fecht a , Kjell G. Nyborg b,c,∗ , J¨org Rocholl d b

a Deutsche Bundesbank, Wilhelm-Epstein-Strasse 14, 60431 Frankfurt am Main, Germany Norwegian School of Economics and Business Administration, Helleveien 30, 5045 Bergen, Norway c CEPR, United Kingdom d ESMT European School of Management and Technology, Schlossplatz 1, 10178 Berlin, Germany

Available online 14 September 2007

Abstract We document a general pattern in the euro area overnight interbank rate (EONIA) and analyze how German banks compared to other EMU banks respond to these predictable changes in the price for reserve holdings. At the beginning of the maintenance period, when the EONIA is typically above average, we observe that German banks hold substantially less reserves than their daily average required reserves. Thus in contrast to other EMU banks, German banks back load the fulfillment of their reserve requirements over the reserve maintenance period and thereby benefit from the general pattern in the EONIA. Looking at the disaggregate data we find than this is particularly the case for the Landesbanks. We argue that the end of the calender month effect in the EONIA may be driven by a temporary shortage of liquidity, relative to reserve requirements, at the beginning of the maintenance period (which coincides with the end of the calendar month). © 2007 Elsevier Inc. All rights reserved. Keywords: Reserve requirements; Liquidity; Overnight rates; Banking

1. Introduction It is well documented that there are calendar effects in overnight rates in the euro area Hartmann, Manna, and Manzanares (2001), Perez-Quiros and Rodriguez (2006), Bindeseil, Weller, and Wuertz (2003), Nautz and Offermanns (2006) as well as in the US Hamilton (1996), Bartolini, ∗ Corresponding author at: Norwegian School of Economics and Business Administration, Helleveien 30, 5045 Bergen, Norway. Tel.: +47 55 95 92 87; fax: +47 55 95 96 50. E-mail addresses: [email protected] (F. Fecht), [email protected] (K.G. Nyborg), [email protected] (J. Rocholl).

1062-9408/$ – see front matter © 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.najef.2007.09.003

8

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

Bertola, and Prati (2001), and Furfine (2000). For example, the EONIA (euro overnight index average) is systematically higher towards the end of the month than at other times Bindseil et al. (2003). This raises two issues. First, to what extent do banks adapt their liquidity management to these systematic calendar effects? Second, why does the overnight rate not follow a martingale; why are the higher end of month rates in the euro area not arbitraged away? In this paper, we shed light on these issues by studying the time pattern of reserve holdings of German banks within the reserve maintenance period, i.e. within the period in which banks have to fulfill their minimum reserve requirements. The existence of calendar effects in the EONIA is arguably surprising in light of the principles behind monetary policy implementation in the euro area. First, reserve requirements in the euro area are set at levels which are substantially larger than the demand for working balances European Central Bank (2002). Second, reserve holdings at different days within the reserve maintenance period contribute equally to the fulfillment of reserves requirements, since banks need only to meet reserve requirements as a daily average over the reserve maintenance period. Third, the ECB has a liquidity neutral policy, that is, it aims to inject, through its refinancing operations, exactly the amount of central bank money that banks need in aggregate to meet their liquidity needs, including satisfying reserve requirements. Finally, remuneration on reserves held with the ESCB (European system of central banks) is the same for all days within the same maintenance period. Thus there is no apparent reason why the EONIA, or the demand for liquidity, should follow a systematic pattern over the maintenance period.1 To examine how banks respond to time patterns in the EONIA, we use two sets of data relating to the fulfillment of reserves. First, we have the aggregate reserve holdings of German and all European Monetary Union (EMU) banks on a daily basis from 06/2000 to 12/2001. This permits us to contrast the liquidity management of German banks with the rest of Europe. Second, over the same time period, we have the marginal and cumulative reserve holdings of each individual German bank at each day before one of the ECB’s weekly main refinancing operation. Thus, we can study differences in liquidity management across different bank types. German banks are particularly interesting to study since they account for the largest share of the euro area banking sector, and they are commonly viewed as acting as interbank intermediaries in the market for liquidity in the euro area.2 Since the EONIA typically increases towards the end of the calendar month, efficient liquidity management should involve relatively low reserve fulfillment during this time. Since the reserve maintenance period during the sample period runs from the 24th of a month until the 23rd of the next, the end of month increase in the EONIA coincides with the first week of the maintenance period. Thus, typically, efficient liquidity management involves the holding of less reserves than the average daily required amount during the beginning of the maintenance period. Using the aggregate reserve fulfillment dataset, we find that German banks have on average lower fulfillment ratios (cumulative reserve holdings within the maintenance period as a

1 The ECB operates with a system of lending and deposit facilities that are 100 basis points above and below, respectively, the minimum bid rate in the ECB’s main refinancing operations. Thus, assuming that there are no expectations of rate changes, if the probabilities that the ECB injects too much or too little liquidity are equal, this means that the overnight rate should be the average of the lending and deposit rates throughout the reserve maintenance period. More generally, this system should assure that the overnight rate follows a martingale, assuming no imperfections in the market for overnight money (see, e.g., Bindseil et al. (2003)). 2 See Deutsche Bundesbank (2004).

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

9

percentage of cumulative reserve requirements) than other EMU banks in the first week of the maintenance period. Essentially, German banks seem to supply liquidity to the rest of Europe at the beginning of the maintenance period, when liquidity is expensive, and then increase their reserve holdings around the second week of the maintenance period when liquidity starts to get cheaper. Thus the collective benefit to German banks from their ability to adapt to changes in the cost of liquidity is mirrored by a collective loss to the rest of the euro area banking sector.3 We suggest that this apparent superior liquidity management performance of German banks may be related to their ability to obtain large quantities of liquidity without having to pay a premium. For example, German banks are collateral rich, giving them a potential advantage in the ECB’s refinancing operations, which are the main source of euro liquidity. Furthermore, German Landesbanks have had government guarantees in the period under consideration, thus reducing credit risk, which may be relevant since overnight credit is largely unsecured. They also generate liquidity by virtue of being saving bank head institutions. Indeed, among German banks, we find that Landesbanks have by far the lowest fulfillment ratios the first week of the maintenance period. Why are there time patterns in overnight rates? Several explanations have been put forward and tested for why calendar effects are not arbitraged away. Hamilton (1996) and Bartolini et al. (2001) suggest that transaction costs might play an important role. Furfine (2000) shows that specific liquidity needs on days with large payment volumes determine the intra maintenance period demand for reserves, while Bindseil et al. (2003) argue that “window dressing” may be an important motive for demanding liquidity at the end of the calender month, leading to an increase in the overnight rate at the end of the calendar month. Our analysis suggests an alternative explanation of the EONIA end of month effect. We find that the aggregate reserve fulfillments of all EMU banks (including German ones) typically fall below the required daily average during the first week of the maintenance period. In other words, there is insufficient liquidity in the whole euro area banking sector for banks to keep up with reserve requirements during this time. Essentially, injection of liquidity from the ECB is such that there is a deficit relative to aggregate reserve requirements during the beginning of the maintenance period (end of the calendar month) and a surplus during the remainder of the maintenance period. If markets were perfect, this should not induce a rise in the EONIA during the beginning of the maintenance period given the ECB’s policy of being liquidity neutral in aggregate over the course of the maintenance period. We suggest a reason for this beginning of maintenance period shortage of liquidity below. For now, our point is that this may contribute to the beginning of maintenance period/end of month increase in the EONIA because of a lack of depth in the interbank market. Banks may be wary of being short with respect to their reserve fulfillments because they may face an upward sloping supply curve later on. As a result, the temporary shortage of available reserves at the beginning of the maintenance period may drive up the EONIA. The rest of the paper is organized as follows. Section 2 provides an institutional background to the German banking sector and reserve requirements and management in the euro area. Section 3 reports on time patterns in the EONIA over the sample period for which we have reserve fulfillment data. Section 4 uses the aggregate reserve data to compare the liquidity management

3 In related work, Angelini and Vacca (2004) find that among Italian banks, only large banks’ reserve holdings are sensitive to the overnight rate. Large Italian banks reduce reserve holdings when rates increase.

10

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

of German banks versus those in the rest of the euro area. Section 5 studies liquidity management of individual German bank, categorized by bank type. Section 6 concludes. 2. Institutional background Compared to the FED system, the European System of Central Banks (ESCB) applies higher reserve requirements to a larger class of liabilities. The basis for the calculation of each bank’s reserve requirement is its end-of-month short-term liabilities4 owed to non-banks or banks outside the euro area two month before the maintenance period. A ratio of 2% of those liabilities are the applied minimum reserve rations.5 During the sample period, the reserve maintenance period starts at the 24th of a month and ends on the following 23rd. In contrast to the US, required reserve holdings of euro area banks are remunerated at the average marginal rate of the Eurosystem’s main refinancing operations during the respective maintenance period. The ECB’s main refinancing operations are repo auctions with a two week maturity. Through these auctions the ECB tries to allocate precisely the amount of reserves needed by the banking sector to fulfill their reserve requirements. However, changes in the banknotes in circulation and changes in the deposits held by EMU governments with the ESCB affect the reserve positions of the banking sector. The ECB tries to forecast the development of these autonomous liquidity factors and correct the amounts allocated in the auctions for these factors. But nevertheless these effects can create a temporary shortage of reserves. The German banking sector is by far the largest in the euro area. At the end of 2001 German banks accounted for roughly 36% of the euro area banking sector in terms of total asset as well as in terms of number of banks. German banks dominate the primary money market – their share in the allotment in the money market auctions of the ECB amounts to more than 55%. Thus German banks play a key role in the euro money markets and the distribution of liquidity to other financial institutions. The German banking sector has a three pillar structure, consisting of private, public, and cooperative banks. The private commercial bank sector is comprised of four large banks and a host of regional and smaller commercial banks as well as German branches of foreign banks. This pillar accounted for roughly a third of the German banking sector in terms of balance sheet total by the end of 2000. The second pillar – the public banks – also made up around a third of the German banking sector. This group is comprised of the savings banks and their regional head institutes – the Landesbanks – which are jointly owned by the respective state and the regional association of savings banks. The cooperative banking sector with the credit cooperatives and the cooperative central banks which are primarily owned by the regional credit cooperatives constitute the third pillar. They had around 12% of the German banking sector’s asset under management by the end of 2000. Besides those three pillars mortgage banks together with the buildings societies (Bausparkassen) and foreign banks accounted for 17% and 4% of the banking sector, respectively.6 This three pillar structure affects the way liquidity is reallocated in the banking sector. The public banks and the cooperative banking sector form two separate and relatively closed giro 4 More precisely, these are the overnight deposits, deposits with an agreed maturity up to two years, deposits redeemable at notice up to two years, and issued debt securities with agreed maturity up to two years. 5 For a more detailed description of the Eurosystem’s minimum reserve system and the penalties imposed by the ESCB if a bank fails to meet the requirements see European Central Bank (2005b). 6 For a more detailed description of the German banking sector see, for example, Hackethal (2004).

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

11

Fig. 1. Development of the EONIA (in %) from 01/1999 to 12/2001. Dotted lines mark the last day of a month. Solid lines indicate the end of a maintenance period.

systems. The second-tier institutes – the savings banks and the credit cooperatives – typically achieve a significant liquidity surplus due to their retail business structure. Within the giro systems they pass this excess liquidity on to the respective (regional) head institute.7 Thus a significant fraction of German banks in the public and cooperative banking sector does actually not participate directly in the overnight interbank market. The second tier institutes in the public as well as in the cooperative banking sector often hold interbank relations only to their head institute and use this to reallocate their liquidity.8 3. Time patterns of the EONIA Fig. 1 graphs the daily levels of the EONIA over the period 1 January 1999 to 31 December 2001. The EONIA is a volume weighted daily average of overnight interbank transaction rates initiated within the euro area by a set of panel banks.9 The solid vertical lines indicate the last day of a maintenance period, while the dotted vertical lines mark the last day of a month. The figure illustrates a general pattern of the EONIA within the reserve maintenance period. First, the EONIA tends to spike on the last day of the maintenance period. Over the sample period, this can be seen to be more often than not a downward spike. This is also shown in Table 1, which tabulates changes in the EONIA relative to a 22 day moving average as well as from day to day. On the 23rd day of the month, the EONIA is on average 11 basis points (bp) below the average of the prior 22 trading days. The EONIA on the 23rd is on average about 9 bp below its level on the 21st and the 22nd. The table reveals an average decrease in the EONIA from the 20th to the 7 For a broader discussion of the interbank linkages in the German banking sector in general and within the three pillars in particular see Deutsche Bundesbank (2000) and Upper and Worms (2004). 8 Ehrmann and Worms (2004) analyze the implications of this peculiarity of the German banking sector for the monetary transmission process. 9 See www.euribor.org.

12

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

Table 1 Pattern of EONIA Day of the month

Mean deviation

Mean change

Times up

Times down

Observations

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

3.9 2.8 1.1 0.5 0.3 0.1 −0.2 2.8 3.1 2.0 3.8 6.2 0.6 −3.4 −1.7 −1.0 6.0 −0.1 −5.6 −9.1 −10.6 −12.9 −11.3 2.6 3.0 0.0 2.3 6.4 7.2 9.5 9.6

−6.4 −4.2 −5.8 −0.5 −0.1 0.3 0.5 −0.7 0.5 1.2 2.1 1.4 −1.9 −2.2 −0.9 0.5 1.6 −5.8 −4.4 −3.2 −1.4 −8.8 −1.7 16.8 8.5 2.9 0.4 2.3 3.8 8.1 4.6

1 3 5 4 6 8 9 6 5 8 12 9 6 5 5 6 7 9 9 10 10 9 14 15 13 10 8 12 11 12 12

18 17 17 15 13 8 8 8 11 7 6 11 14 14 12 9 10 15 16 11 12 16 11 9 7 8 9 8 6 5 1

22 25 25 25 27 26 26 26 25 25 25 27 25 26 26 24 25 25 27 26 25 26 25 24 23 25 26 26 24 22 14

Mean deviation reports for each day of the month the mean deviation of the EONIA from its average over the previous 22 trading days. Mean change reports the average change of the EONIA at a specific day of a month over the previous day of the month. Times up (down) reports the number of times that the EONIA increased (decreased) compared to the previous day at the specific day of the month in the sample period.

23rd. However, although the EONIA drops on average towards the end of the maintenance period, the table shows a large number of cases where the EONIA actually increases. Under the ECB’s liquidity neutral policy, an absence of market imperfections would suggest that there should be an equal number of decreases and increases and that the average change should be zero. Our focus is on the second feature of the EONIA’s time pattern. Specifically, both Fig. 1 and Table 1 show that the EONIA typically increases towards the end of the month, as previously documented by, for example, Bindseil et al. (2003). Between January 1999 and December 2001 the EONIA exceed its monthly average on the last trading day of a month by approximately 9 bp. The last day of the month, the EONIA increases relative to the previous day’s level more than twice as often as it drops. The table also shows that over the sample period, the upward trend in the EONIA tends to start several days before the end of the month. Furthermore, the increase towards the end of the month is reversed at the beginning of the month. On the 1st

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

13

day of a month, the table reveals 18 negative changes in the EONIA, versus only 1 increase. This apparent time pattern in the overnight rate suggests that optimal liquidity management involves being relatively short at the end of the calendar month/beginning of the maintenance period, and subsequently making up for this deficit by borrowing later in the maintenance period. 4. Liquidity management: evidence from aggregate reserves data In this section, we use aggregate reserve data to report on the extent to which banks adapt to the time pattern in the EONIA discussed above. Our data allows us to separate German banks from other EMU banks. We track banks’ liquidity management by the fulfillment ratio, defined for a bank (or sector) as its cumulative reserve holding from the beginning of the maintenance period to the respective day in the maintenance period divided by the cumulative required daily reserves up the respective day. So the fulfillment ratio of all German banks at day t of maintenance period p is given by: t 

fulfillment ratiot;p =

1 t 

daily reserve holdingst;p × 100

(1)

daily required reservest;p

1

where multiplication by 100 means that we express the fulfillment ratio as a percentage. So a fulfillment ratio of 100% means that the bank (or sector) has satisfied reserve requirements in full up the relevant date. A fulfillment ratio of less (more) than 100 means that it is running a reserve deficit (surplus). Fig. 2 reports the aggregate fulfillment ratio of German banks for the period starting at the beginning of the maintenance period in May 2000 and ending at the end of the maintenance period in January 2002. The figure suggests that there is a time pattern in the aggregate fulfillment ratio of German banks. In particular, the fulfillment ratio tends to spike down at the beginning of a new maintenance period. There appears to be a reverse at the beginning of the calendar month, i.e., around the second week of the maintenance period. Fig. 3 cuts the data differently, presenting the average aggregate fulfillment ratio of German banks for each day within the reserve maintenance period (averaged day by day over the sample period). This shows clearly that German banks run reserve deficits during the first month of the maintenance period. They end the maintenance period with a fulfillment ratio of slightly more than 100%, on average. This finding suggests that German banks adapt well to the patterns in the EONIA discussed above. They sell liquidity at the beginning of the maintenance period, when it is dear, and accumulate it in the last three weeks of the maintenance period, when it is relatively cheap. We have argued that the time pattern of the aggregate fulfillment ratio of German banks is driven by active intertemporal liquidity management. Let us consider an alternative view, namely that it is driven by exogenous shocks to the entire banking sector (the autonomous liquidity factors). These liquidity shocks may result from changes in the deposits that the Governments hold with the European System of Central Banks (ESCB), from fluctuations in the value of banknotes in circulation, and from changes in the net foreign asset position of the ESCB. The largest fluctuations

14

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

Fig. 2. Development of the fulfillment ratios of German banks in total. Dotted lines mark the last day of a month. Solid lines indicate the end of a maintenance period.

in the autonomous factors are due to changes in the deposits held by governments with the respective national central bank. However, the German government holds its deposits with the commercial banks and not with the Bundesbank. Therefore, the autonomous factors affecting directly the liquidity position of the German banking sector do not fluctuate significantly within a maintenance period and, more importantly, they do not show a specific pattern in their fluctuations

Fig. 3. Average fulfillment ratios of German banks at the different days of the maintenance period.

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

15

Fig. 4. Average fulfillment ratios of EMU banks at the different days of the maintenance period.

over the maintenance period. Thus, the time pattern of fulfillment ratios is unlikely to be the result of exogenous liquidity shocks. On an EMU wide basis the autonomous factors spike somewhat in the last week of the calender month in the time period under consideration. However, the ECB forecasts these autonomous factors and aims to offset their effect by adapting the allotted amount in the refinancing operations. Fig. 4 shows the average aggregate fulfillment ratio of all EMU banks (including German ones) day by day over the reserve maintenance period. The pattern is similar to that of German banks only, but less pronounced. Even though the ECB follows a liquidity neutral policy and adjusts the reserve supply to changes in the autonomous liquidity factors, it appears that the supply of liquidity in the system is too low during the first week of the maintenance period to allow banks, in aggregate, to fulfill their reserve requirements. Fig. 5 shows the corresponding figure for non-German EMU banks. In contrast to German banks, these banks are not running reserve deficits during the first week of the maintenance period. Instead, during this period when liquidity is expensive, non-German banks are actually accumulating a slight surplus of reserves. To summarize, German banks are offloading liquidity to banks in the rest of the euro area during the first week of the maintenance period, when the EONIA is at a premium relative to overnight rates in the remainder of the maintenance period. German banks subsequently accumulate reserves when rates are lower, during the second week of the maintenance period (the first week of the calendar month). Interestingly, the first week of the maintenance period is a period of insufficient liquidity in the banking sector, as illustrated in Fig. 4, as well as a relatively high level of the EONIA, as discussed above. That a shortage of liquidity (relative to reserve requirements) in the banking sector coincides with a period of high overnight rates may well be a coincidence.10 An alternative 10 We recognize that there may be statistical issues here since our sample period here covers only 18 reserve maintenance periods.

16

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

Fig. 5. Average fulfillment ratios of non-German EMU banks at the different days of the maintenance period.

explanation is that the EONIA is high precisely because of the liquidity shortage. If so, there must be imperfections in the market for liquidity (given the way the ECB implements monetary policy, as discussed above). Such imperfections could manifest themselves by limited depth in the interbank market, implying that banks that need “large” quantities of liquidity face an upward sloping supply curve. An extreme example of this would be that a bank that runs a reserve deficit runs the risk of being squeezed towards the end of the maintenance period as it attempts to make up the deficit. This “limited depth” view is supported by evidence from bidding behavior in the ECB’s main refinancing operations [Nyborg, Bindseil, and Strebulaev (2002) and Craig and Fecht (2007)]. An implication is that banks that are not so fearful of facing an upward sloping supply curve of liquidity later in the maintenance period would lend to banks that are more fearful of this event. This suggests that German banks advance liquidity to non-German banks during the first week of the maintenance period because they have better access to liquidity later in the maintenance period, for example because they are liquidity generators or because they have an advantage in the ECB’s refinancing operations (e.g. due to the large pool of eligible collateral among German banks). The question that remains is why there is a liquidity shortage during the beginning of the maintenance period. Our explanation is related to the increase in reserve requirements over time, as illustrated in Fig. 6. The ECB injects liquidity, to allow banks to satisfy reserve requirements, mainly through weekly refinancing operations. During the sample period, these operations have maturities of two weeks. However, the reserve maintenance period is one calendar month. Thus the first auction of a reserve maintenance period rarely falls on the first day of the period (the 24th). Thus, the amount of liquidity in the system at the beginning of a maintenance period is determined by the two last auctions in the previous reserve maintenance period. But since the previous period’s amount of reservable funds is intended to allow banks to exactly fulfill reserve requirements in that period, it will be too low an amount to allow banks to fulfill reserve requirements in the current period, since reserve requirements are rising.

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

17

Fig. 6. Aggregate reserve requirements (in millions of euros) of EMU area banks.

Note that in addition to the weekly refinancing operations, the ECB also holds longer term operations (maturity of three months). These are held once a month, typically during the first week of the maintenance period. However, over the sample period, they rarely fall on the 24th. They tend to occur closer to the end of the month, and in some cases even fall during the beginning of the month. Thus, the longer term operations generally occur too late to neutralize the lagged liquidity-stock effect discussed in the previous paragraph. An alternative to the story we have advanced above is that differences in fulfillment ratios at the beginning of the maintenance period relate to transactions costs in the interbank market. Such costs are suggested as an explanation behind calendar effects in overnight rates by, for instance, Hamilton (1996) and Bartolini et al. (2001). Banks might want to save on these transactions costs in the first week of the maintenance period. Instead of offsetting liquidity outflows by borrowing in the interbank market, they wait – hoping for a compensating liquidity inflow during the remainder of the maintenance period. One might reasonable expect such transactions costs to be relatively more significant for small banks, for example if there is a fixed cost of transacting in the interbank market. Thus, the hypothesis would be that the time pattern of fulfillment ratios we observe is driven by German banks being smaller than average EMU banks. However, on average German banks have about the same size as other EMU banks European Central Bank (2005a). Thus, we do not find the transactions costs argument very plausible as an explanation for the differences in fulfillment ratios between German and other EMU banks during the first week of the maintenance period. 5. Liquidity management: evidence from individual bank reserves data In this section, we take a closer look at the time pattern of reserve fulfillments among German bank by studying fulfillment ratios at the individual bank level. We possess reserve holdings for all individual German banks for the day prior to each main refinancing operation run by the ECB

18

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

between June 2000 and December 2001. There is one main refinancing operation weekly, typically on a Tuesday. Thus we have individual banks’ reserve data for Mondays. Given that auction x in our sample is on day j + 1 of the maintenance period, bank i’s fulfillment ratio before the auction is given by j 

fulfillment ratioi;x =

1 j 

daily reserve holdingsj;i × 100.

(2)

daily required reservesj;i

1

We categorize the auctions by their chronological position within the reserve maintenance period. Since the auctions are held weekly, there are up to five auctions in each reserve maintenance period. We use the following classification: Auction 1 is the first one in the maintenance period. Auction 2 is the second auction and also the third one if there are five auctions in the period. Auction 3 is the penultimate auction. Auction 4 is the final auction in the maintenance period. Recall that the reserve maintenance period starts on the 24th of each month. Thus Auction 1 tends to fall around the end of the month. The results are presented in Table 2. We categorize banks by their type, as described above. For each bank type and for each auction position, the table reports the equally weighted average fulfillment ratio across banks. The first row presents the average fulfillment ratios at the different auction positions across all German banks. We see the same pattern as the daily aggregate numbers discussed above. German banks have an equally weighted average fulfillment ratio of 95.41% at the time of the first auction of the maintenance period. This is significantly below 100%, both economically and statistically. At auctions later in the maintenance period, the average fulfillment ratio is around 104%. This increasing time pattern of fulfillment ratios is repeated across nearly all bank categories. Table 2 Fulfillment over time Bank type

Auction 1

Auction 2

Auction 3

Auction 4

Number of banks

All Small private Big private Savings Cooperative Foreign Landesbanks Cooperative central banks Bausparkassen Special purpose banks

95.41 (0.11) 92.16 (0.57) 96.24 (3.88) 93.56 (0.21) 96.90 (0.13) 91.06 (1.15) 75.37 (2.35) 91.25 (31.46) 91.86 (4.24) 83.53 (3.11)

103.44 (0.10) 101.04 (0.53) 101.73 (3.67) 106.03 (0.19) 103.23 (0.10) 99.82 (1.04) 76.12 (1.71) 93.44 (29.83) 104.39 (3.62) 93.34 (3.20)

104.61 (0.07) 103.50 (0.40) 94.49 (1.88) 106.05 (0.11) 104.54 (0.07) 103.06 (0.82) 84.45 (1.13) 101.52 (15.57) 105.62 (2.78) 94.94 (2.65)

103.16 (0.04) 104.24 (0.29) 97.65 (0.94) 102.51 (0.05) 103.29 (0.05) 104.81 (0.56) 95.26 (0.67) 97.61 (7.63) 107.73 (2.45) 100.22 (1.71)

2453 193 4 549 1598 69 12 4 13 13

The table reports on the average fulfillment ratio across German banks, by category, over time within the reserve maintenance period. The fulfillment ratio is a bank’s cumulative reserve holdings as a percentage of its cumulative reserve requirements to a specific date in the maintenance period. These are calculated for every bank the day before the ECB’s main refinancing operations (repo auctions). Auction 1 is the first auction in the maintenance period. Auction 2 is the second auction and also the third auction if there are five auctions in the period. Auction 3 is the penultimate auction. Auction 4 is the final auction in the maintenance period. The table reports pooled mean fulfillment ratios, by bank category, for each auction position. Standard errors are in parentheses.

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

19

The category of banks which has the most striking time pattern of reserve fulfillment is the Landesbank category. Their average fulfillment ratio at the beginning of the maintenance period (Auction 1) is only 75.37%. It rises to 95.26% by the time of the last auction. This is consistent with the hypothesis that we advanced in the previous section that the observed time pattern of reserve fulfillment is related to the risk of facing an upward supply curve of liquidity later in the maintenance period. This is so because Landesbanks are liquidity “collectors”; liquidity is passed to them from the savings banks. Furthermore, government guarantees means that their credit ratings are high, which should help them obtain quite large quantities of overnight credit on favorable terms. Landesbanks should therefore be the category of banks with the least risk of facing an upward sloping supply curve for liquidity. As a result, they seem to be best equipped to exploit the temporary rise in the EONIA at the beginning of each maintenance period by lending a substantial quantity of liquidity during this time. Among private banks, we see that the four large domestic banks do not have particularly low fulfillment ratios at Auction 1. They have an average of 96.24%, as compared with 92.16% for small private banks. This less aggressive behavior of the large domestic banks is consistent with the limited depth in the interbank market hypothesis, since this is a potentially bigger issue the larger the amount a bank needs to borrow. 6. Conclusion In this paper, we have studied time patterns of German banks’ reserve fulfillment ratios within the reserve maintenance period and compared this with that of other EMU banks. We have documented that German banks run reserve deficits during the beginning of the maintenance period while other EMU banks run slight surpluses. We have also seen that this coincides with a temporary increase in the price of liquidity. In other words, German banks seem to adapt their liquidity management well to changes in overnight rates, while non-German bank’ reserve holdings are less sensitive to overnight rate fluctuations. Interestingly, there also seems to be a general liquidity deficit during the beginning of the maintenance period; in aggregate, EMU banks (including German ones) run deficits during this time. We suggest that this may be related to a lagged liquidity-stock effect, namely that the amount of liquidity in the banking sector is adapted to reserve requirements from the previous period, which are generally lower since, empirically, reserve requirements rise over time. This effect cannot be neutralized before the first repo auction (refinancing operation) of the maintenance period, which typically occurs a few days after the start of the maintenance period. We hypothesize that this liquidity shortage contributes to the increase in the EONIA during the beginning of the maintenance period. In line with previous authors, we have called this an end of the calendar month effect in this paper, but it may well be that calling it a beginning of the reserve maintenance period effect is more appropriate. An explanation for German banks’ apparent superior liquidity management may be that they can exploit the higher interest rates during the beginning of the maintenance period because they face a less steep upward sloping liquidity supply curve later on in the period than do non-German banks. This could be the result of German banks being collateral rich relative to other banks and thus having an advantage in the ECB refinancing operations, where liquidity is injected into the banking sector. The category of banks which exploits the time pattern in the EONIA to the largest extent are the Landesbanks. These banks function as liquidity collection buckets for the savings bank sector. Landesbanks are thus arguably the banks that face the smallest risk of having to buy expensive liquidity in the interbank market. Furthermore, because of government guarantees,

20

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

raising larger amounts of liquidity in the interbank market might in any case be less costly for Landesbanks than many other banks. This therefore fits with the story that German banks’ apparent superior liquidity management performance relates to competitive advantages in the market for liquidity. An alternative story is that German banks systematically experience an exogenous negative liquidity shock at the beginning of the maintenance period, with the shock being larger for the Landesbanks. German banks might respond by borrowing heavily in the interbank market in order to offset at least a part of the shock. This might in turn lead to an increase in overnight rates. Without further investigation, we cannot say for sure which of these stories is closer to the truth. Another important avenue for further research is extending the time period of the study. Since the end of the sample period, there have been several institutional changes which affect the market for liquidity. For example, in July 2005 the government guarantees of the Landesbanks were abandoned for new claims against the Landesbanks. This may have diminished their advantage in the market for liquidity. Also, in March 2004, the maturity of the ECB’s main refinancing operations was reduced from two to one week. The maintenance period was also changed to coincide with the meetings of the ECB’s Governing Council. This might reduce, but not necessarily eliminate, the structural liquidity shortage at the beginning of each maintenance period that we have seen during the sample period. Thus including more recent maintenance periods in the sample would help in determining to what extent the operational framework is contributing to the end of the calendar month effect. Acknowledgements We wish to thank the Deutsche Bundesbank for supplying data. We also would like to thank Michael Schroeder and participants at the Deutsche Bundesbank and ZEW conference on Monetary Policy and Financial Markets, Mannheim, Germany, November 2006 for comments. The views expressed here are those of the authors and not necessarily those of the Deutsche Bundesbank. References Angelini, P., & Vacca, V. (2004). What determines banks’ sensitivity to money market interest rates. Mimeo. Bartolini, L., Bertola, G., & Prati, A. (2001). Banks’ reserve management, transaction costs, and the timing of Federal Reserve interventions. Journal of Banking and Finance, 25, 1287–1317. Bindseil, U., Weller, B., & Wuertz, F. (2003). Central bank and commercial banks’ liquidity management: What is the relationship. Economic Notes, 32, 37–66. Craig, B., & Fecht, F. (2007). The eurosystem money market auctions: A banking perspective. Journal of Banking and Finance, 31, 2925–2944. Deutsche Bundesbank. (2000, January). Longer-term trend in German credit institutions’ interbank operations (Monthly Report), pp. 49–68. Deutsche Bundesbank. (2004, July). Initial experience with the new monetary policy framework and the Bundesbank’s contribution to liquidity management by the Eurosystem (Monthly Report), pp. 49–66. Ehrmann, M., & Worms, A. (2004). Bank networks and monetary policy transmission. Journal of the European Economic Association, 2, 1148–1171. European Central Bank. (2002, May). The liquidity management of the ECB (Monthly Report), pp. 41–54. European Central Bank. (2005a, October). EU Banking Structure Report. European Central Bank. (2005b, February). The Implementation of Monetary Policy in the Euro Area. Furfine, C. H. (2000). Interbank payments and the daily federal funds rate. Journal of Monetary Economics, 46, 535–553. Hackethal, A. (2004). German Banks and Banking Structure. In H. Jan Pieter Krahnen, Reinhard, & Schmidt (Eds.), The German Financial System (pp. 71–105). Oxford University Press.

F. Fecht et al. / North American Journal of Economics and Finance 19 (2008) 7–21

21

Hamilton, J. D. (1996). The daily market for federal funds. Journal of Political Economy, 104, 26–56. Hartmann, P., Manna, M., & Manzanares, A. (2001). The microstructure of the euro money market. Journal of International Money and Finance, 20, 895–948. Nautz, D., & Offermanns, C. J. (2006). EONIA dynamics and volatility transmission in the European money market. Mimeo. Nyborg, K. G., Bindseil, U., & Strebulaev, I. A. (2002). Bidding and performance in repo auctions: Evidence from ECB open market operations (ECB Working paper 157). Perez-Quiros, G., & Rodriguez, H. (2006). Market for funds in Europe: What has changed with the EMU? Journal of Money, Credit and Banking, 38, 91–118. Upper, C., & Worms, A. (2004). Estimating bilateral exposures in the German interbank market: Is there a danger of contagion? European Economic Review, 48, 827–849.