This article uses data on CHAPS Sterling transactions to describe the segment of the unsecured overnight loan market that settles within CHAPS. It assesses the size, timing and importance of these transactions for the underlying payments infrastructure. Advances and repayments of overnight loans are estimated to have accounted for around 20% of CHAPS Sterling activity by value over our sample period; four CHAPS Sterling members send and receive virtually all payments corresponding to these loans; and, finally, the value of CHAPS Sterling payments associated with this market rises towards the end of the CHAPS day.

Introduction

This article uses data on transactions processed in the Clearing House Automated Payment System (CHAPS Sterling) to provide a description of the segment of the sterling interbank overnight unsecured loan market that settles within CHAPS. In particular, we examine the size of this market, the costs of these loans and the Liming of settlement of these loans. In addition, we consider the implications of these payments for the system through which they are made.

Whether a loan is brokered or is the result of a direct deal between two financial institutions, it will eventually result in a payment from one bank to another. Unless the two involved financial institutions are customers of the same settlement bank (in which case the transaction may be settled on the books of the settlement bank) this will be settled in CHAPS Sterling. The Bank of England operates the CHAPS system and keeps a record of all transactions among the settlement banks for surveillance and research purposes.(1) Here, we use this information to match the two legs of an unsecured overnight loan as the payment and repayment are made across CHAPS Sterling.

We find that around Â£22 billion of overnight interbank loans are processed across the CHAPS Sterling system every day. This represents a large proportion of total CHAPS Sterling activity; in particular, we estimate that 22% of all CHAPS Sterling transactions by value are advances or repayments of overnight loans. Although there are 13 settlement banks (including the Bank of England) in CHAPS Sterling, we find that four members send and receive virtually all payments corresponding to overnight loans. Finally, we find that the value of CHAPS Sterling payments associated with this market increases as the day progresses.

Data

The Bank of England keeps track of all CHAPS Sterling transactions that occur among the settlement banks.(2) We use data from 4 March 2002 Lo 4 March 2003. Removing weekends, holidays and the one day on which the system encountered operational problems leaves 252 days of data. The average total daily CHAPS Sterling value was around Â£200 billion over the sample period. Though CHAPS is designed to handle large-value payments, it is common to find small-value transactions as well. We consider only payments of value larger than Â£1 million. We also exclude those payments where one of the sides is known to be a non-bank customer of a bank and consider only transactions that occur among banks.(3) Finally, CHAPS Sterling involves a total of 13 banks but we ignore transactions involving the Bank of England and, though NatWest and RBS still run separate accounts in CHAPS, we merge their accounts and consider them as a single group.

With these criteria in place, we were left with CHAPS Sterling payments averaging Â£155 billion a day (about 7,900 payments). Chart 1 depicts the value and volume of the selected payments while Table A reports the mean and standard deviation for total volumes and total values.

Method

An overnight interbank unsecured loan involves one bank entering into an agreement to borrow from another bank a sum, K, with the promise to repay the following working day an amount equal to this sum plus interest, K(1 + r), where r is the overnight interest rate. Provided that the two banks are not customers of the same settlement bank and the trade is not settled on its books, both legs of the transaction will appear as CHAPS Sterling payments. So it should be possible to see both the loan advance and the repayment within data on CHAPS payments. In what follows, we apply the method developed by Furfine (1999) in order to identify pairs of payments made in CHAPS Sterling on consecutive days that are associated with overnight loan advances and their repayment.(1)

We select a band of between 3% and 6% as it encompasses the repo rate-which was 4% for most of the sample period-and ensures that we consider only overnight loans as opposed Lo loans made for a longer duration.(3) To see this, consider a loan for Â£1 million made on day t over two days at an annualised rate of 4%. The repayment would be Â£1,000,219 on day t + 2. Now if there were a payment from the lending bank to the borrowing bank of Â£1 million on day t + 1, our algorithm would consider it as a possible overnight loan made on day t + 1. But the implied interest rate of 7.99% would be outside our band and so the pair of payments would be rejected. This would be the case for any loan of maturity longer than one day whose interest rate was between 3% and 6%.

More specifically, the precise algorithm used to identify the payments (at date t - 1) and the repayments (at dale t) works as follows:

(1) At date t, round all payments down to the nearest hundred thousand figure. For example, Â£251,345,891.54 is approximated by Â£251,300,000.00. In other words, we start by assuming that all non round valued payments on day t are potential repayments of overnight loans and we calculate the values of the advances that we would wish to look for on day t - 1 that would correspond to such repayments.

(2) Compute the implied interest rate, r, using the simple rate rule. In our example, r = (Â£251,345,891.54/Â£251,300,000.00 - 1) *365 = 0.0667. We do this in order to eliminate some of the payments identified as possible repayments of overnight loans in Step (1) on the grounds that such repayments would imply an interest rate that was either too high or too low Lo be 'reasonable'.

(3) If the implied rate, r, lies between 3% and 6%, select the rounded payment and the associated rate, otherwise exclude the payment. Having excluded these payments we are left with a set of possible repayments and a set of associated advances (the rounded payments) that we now wish to look for on day t - 1.

(4) Check if the selected advances left after Step (3) can be found among the payments at day t - 1. If the answer is yes, then this payment is considered to be the advance of an overnight loan and its associated payment on day t the repayment of this loan, otherwise it is discarded. Given this approach, we will only pick up overnight loans as opposed to loans of two or more days' maturity, since, unless the payment can be matched to one made the previous day, it is dropped.

(5) Repeal Steps (1)-(4) for all payments for each pair of banks and for each pair of consecutive business days.

The algorithm assumes that the principal and the interest are repaid together as one CHAPS Sterling transaction. Furfine suggests that in Fedwire it is possible Lo pay the principal and the interest separately, that is, as two payments.(1) Also, we use a first in, first out (FIFO) rule for the timing of the payments and repayments. This means that the first loan is assumed to be the first one repaid the day after. Finally, our algorithm will catch overnight loans that are negotiated on previous days, ie forwards. But this will not affect any of our conclusions with respect to the volumes and timings of payments associated with such loans.

Have we identified overnight unsecured loans?

In order to identify overnight unsecured loans, we have not explicitly used the intraday quoted overnight rates. So we can evaluate how good our method is by comparing the rates of interest charged on what we identify as loans with quoted overnight rates. Of course, even if we perfectly identify unsecured overnight loans, our rates will still differ slightly from quoted rates because quoted rates are only indicative and may differ from the actual rates applied to the transactions. In addition, there may be a significant time lag between when the loan is agreed and when the payment is transferred over the CHAPS Sterling system.