Payment systems (i.e. the means by which banks send and receive payments) require the use of banks’ liquidity. As such, economists and regulators alike are concerned about how banks might behave in a payment system following a major disruption. In a gross settlement payment system,1 banks can use liquidity from incoming payments to fund outgoing ones. This enables them to make payments without having to use their reserves or to borrow in the interbank money market. Banks maximise the degree of payment recycling when they coordinate their payments. This improves the liquidity efficiency of the system, but leads to interdependencies in banks' behaviour.
For this reason, researchers have been quick to examine the impact of operational or credit events on bank behaviour in payment systems. For example, McAndrews and Potter (2002) looked at the payment patterns of banks in Fedwire (the US large-value payment system) after the terrorist attacks of 11 September 2001. They found that liquidity efficiency in Fedwire dropped as banks’ coordination in sending and receiving payments broke down due to payment delays and disruptions caused by physical damage. They also found that the Fed’s response to provide abundant liquidity by increasing the levels of discount window and intraday lending, helped restore payment coordination across banks.
In recent research (Benos et al. 2012) we examine if and how sterling settlement banks changed their patterns of payments after the collapse of Lehman Brothers and whether there were any risks or costs associated with a potential change in their behaviour. For this research, we used payments data from CHAPS, the UK’s large-value payment system.
Figure 1. Monthly averages of daily aggregate values and volumes for all CHAPS banks, 1/1/2006-30/9/2009. The red line indicates the date of Lehman's default
Source: Bank of England
While average payment values and volumes did not visibly change after Lehman’s default (Figure 1), the data shows that, in the two months following the failure of Lehman Brothers, banks on average made payments at a slower pace during the day than prior to the failure (Figure 2). We show that this delay was partly explained by concerns about bank-specific as well as system-wide risks. The rationale is as follows: suppose that bank A expects a payment from bank B in the afternoon, but itself has a payment to bank B scheduled that morning. Then, if bank A thinks that bank B might default during the day, it may choose to delay its payment to B until after it has received the payment from bank B in the afternoon. That way, bank A may be able in effect to net its exposure to Bank B and in case of a default of bank B may be able to reduce any amounts to be recovered through bankruptcy proceedings.2
Figure 2. Delay in aggregate CHAPS payments, 1/12006-30/9/2009. The plot shows the five-day moving average of the daily delay measure. Positive figures indicate delay and negative ones indicate acceleration of payments relative to the benchmark period. A one percentage point of delay is equivalent to every payment being made 6.2 minutes later. The red part of the line indicates the post-Lehman period
Source: Bank of England
The data also shows that ‘turnover’, a measure which captures the extent to which banks recycle payments to save on liquidity, was reduced by 30% after the default of Lehman (Figure 3). This means that, similar to the US breakdown of coordination in the wake of the 9/11 attacks, coordination among UK settlement banks also became looser initially because of the observed payment delays and later on probably because banks had less of a need to recycle payments as liquidity became abundant due to increased reserves in the system.
Figure 3. Aggregate turnover, 1/12006-30/9/2009. Turnover for a given day is the ratio of total outgoing payments among CHAPS settlement banks on that day over total liquidity used for the same day. The red line indicates the date of Lehman's default
Source: Bank of England
But do payment delays carry any risks and are these risks economically significant? Payment delays exacerbate liquidity risks that can arise when one or more banks experience operational outages, meaning that they are unable to send payments while the outage lasts. Consider a bank that at some time on a given day is a net provider of liquidity to the system; that is, its cumulative payouts are larger than its cumulative receipts. This bank makes up the difference either by using its own reserves (in the case of CHAPS, funds held in a Bank of England reserves account) or by borrowing intraday from the Bank of England against high quality collateral. If this bank were to delay its outgoing payments, it would effectively also be delaying the provision of liquidity to the system. The more the bank delayed its outgoing payments, the more the amount of liquidity yet to be provided at any given point in time would increase.
Now suppose there is an operational outage during the day. The bank suffers a technical failure and is unable to make any more payments for the rest of the day. This means that all the liquidity it would have provided to the system for the rest of the day is irrevocably lost. Other banks in the system that would have benefited from this lost liquidity are now forced to look for liquidity elsewhere. They may respond to this lower level of liquidity either by borrowing in the unsecured overnight money markets or from the central bank, which requires placing collateral. Both of these options are expensive.
This potential liquidity cost is increasing with payment delays, with the probability of an outage and with the cost of overnight borrowing. Since payment delays increased in the wake of the Lehman Brothers default, the potential costs associated with operational risk also rose.
In our research, we use Markov models to quantify all these effects and to estimate the incremental economic cost of foregone liquidity due to payment delays. We consider two risk measures: one which captures the impact of a single outage that occurs at the worst possible time on a given business day, and another which computes the expected impact of a single outage occurring at a random point in time during the day. We calculate the incremental economic cost of foregone liquidity by estimating what a bank would have had to pay to insure itself.
While both measures of risk show a statistically significant increase in the period following the collapse of Lehman Brothers, the incremental economic cost appears to be modest at about £6,700 per bank per day during the month after the Lehman Brothers collapse. It is possible, however, that this low incremental cost could have been significantly higher in the absence of the throughput requirement that settlement banks are expected to observe and oblige them to settle every month on average a minimum proportion of their payments by specific times of the day. Without this requirement, which helps coordinate payment behaviour, payment delays might have been much more significant.
Afonso G, Konver A and A Schoar, 2011, “Stressed, Not Frozen: The Federal Funds Market in the Financial Crisis”, Journal of Finance, No 4, 1109-1139
Benos E, Garratt R and P Zimmerman, 2012, “Bank Behaviour and Risks in CHAPS Following the Collapse of Lehman Brothers”, Bank of England working paper No. 451
McAndrews J and S Potter, 2002, “Liquidity Effects of the Events of September 11, 2001”, FRBNY Economic Policy Review
1 A real-time gross settlement system (RTGS) is a type of payment system whereby transactions are settled as soon as they are processed and without being netted against other transactions. Most modern large-value payment systems are RTGS.
2 Counterparty risk concerns may not only affect the timing of payments but also the amount of unsecured overnight borrowing and therefore the actual amounts sent via a payment system (see Afonso et al. 2012). In our data we do not observe any significant decline in the amount of payments settled.