First South Australian State Bank recently has installed a new network for teller operations in its 500 branches throughout the state. Teller machines in each branch are connected to a branch controller. The controllers, in turn, are connected to the head-office machine. Although some processing has been distributed to the controllers, the customer accounts master file still remains centralized.

Unfortunately, the new network is unreliable. Furthermore, the problems are not being solved quickly because the bank's information systems staff are blaming the machine vendor. The machine vendor has responded by arguing that the problem lies with the bank's information systems staff who have not undergone sufficient training to acquire adequate competence with the system that has been installed. On a number of occasions, the machine vendor claims it has had to respond to problems that could have been fixed quickly if the bank's information systems staff had been properly trained to use the system.

One consequence of the unreliable network is that tellers sometimes submit duplicate transactions. The network often goes down for only a few minutes. When the network goes down, tellers are supposed to check whether the last transaction they submitted was posted to the account by submitting an inquiry when the network comes up. Unfortunately, customer lines are long because the network is unreliable, and tellers often simply resubmit the last transaction to save time rather than initiate an inquiry. Thus, the same transaction sometimes is processed twice.

Required: Assuming the unreliability of the network continues, what exposures does the bank face if transactions are processed twice? What controls would you recommend be implemented to prevent duplicate transaction processing?