Impact of Computers in Banking

The empirical studies, however, have shown inconsistency on this hypothesis; some agree with the Solow Paradox, some are against. Since most empirical studies have adopted the production function approach, it is difficult to identify which effect has dominated, hence the reasons attributed have been the difference in econometric methodology and measurement.

This paper attempts to explain the inconsistency by stressing the heterogeneity in banking services; in a differentiated model with network effects, we characterize the conditions to identify these two effects and the conditions for the two seemingly positive effects to turn negative in the equilibrium. The results are tested on a panel of 68 US banks over 20 years, and we find that the bank profits decline due to adoption and diffusion of IT investment, reflecting negative network effects in this industry. 1 1 Introduction

The usage of information technology (IT), broadly referring to computers and peripheral equipment, has seen tremendous growth in service industries in the recent past. The most obvious example is perhaps the banking industry, where through the introduction of IT related products in internet banking, electronic payments, security investments, information exchanges (Berger, 2003), banks now can provide more diverse services to customers with less manpower.

Seeing this pattern of growth, it seems obvious that IT can bring bout equivalent contribution to profits. In general, existing studies have concluded two positive effects regarding the relation between IT and banks’ performance. First, IT can reduce banks’ operational costs (the cost advantage). For example, internet helps banks to conduct standardized, low value-added transactions (e. g. bill payments, balance inquiries, account transfer) through the online channel, while focusing their resources into specialized, high-value added transactions (e. g. mall business lending, personal trust services, investment banking) through branches. Second, IT can facilitate transactions among customers within the same network (the network effect) (see Farrell and Saloner, 1985; Katz and Shapiro, 1985;

Economides and Salop, 1992). Let us consider the case of automated teller machines (ATMs) by banks. If ATMs are largely available over geographically dispersed areas, the benefit from using an ATM will increase since customers will be able to access their bank accounts from any geographic location they want.

This would imply that the value of an ATM network increases with the number of available ATMlocations, and the value of a bank’s network to a customer will be determined in part by the final network size of the bank. Indeed, Saloner and Shepard (1995), using data for United States commercial banks for the period 1971-1979, showed that the concern of network effect is important in the ATM adoption of United States commercial banks (see also Milne, 2006).

In view of these two effects above, it should be surprising to know that the evidence, however, shows some inconsistency in concluding the contribution of IT to banks’ profit. 2 Some studies echo the so called Solow Paradox in concluding that IT will actually decrease productivity. As stated by Solow (1987), “you can see the computer age everywhere these days, except in the productivity statistics”. Shu and Strassmann (2005) studied 12 banks operating in the US for the period of 1989-1997 and found that although IT has been one of the most marginal productive factors among all inputs, it cannot increase banks’ profits.

On the other hand, there are some studies agreeing with the positive influence of IT spending to business value. Kozak (2005) examines the impact of the progress in IT on the profit and cost efficiencies of the US banking sector during the period of 1992-2003. The research shows a positive correlation between the level of implemented IT and both profitability and cost savings. The inconsistency in empirical results can be attributed to differences in measurement1 and econometric methodologies (Berger, 2003; Tam, 1998).

Alternatively, the current paper attempts to provide an interpretation by stressing the heterogeneity in banking services. Indeed, compared to manufacturing industries or agriculture, banking industries present higher diversification in providing customer services. In this case, a differentiated model with network effects would probably describe the market better than the production function approach, which describes each bank’s profit (output) as a specific production function of inputs.

Notice that most empirical studies2 have constructed their testing on productivity or growth. In addition, while most production approaches only present a mixture of IT influences on both demand and supply sides, a differentiated model can distinguish a network effect from the demand side (in banking services) from a cost reduction effect from the 1For example, Berger (2003) pointed out two approaches in measuring productivity: either by the govermment productivity indexes or by a modified form of the Solow (1957) neoclassical growth model (Oliner and Sichel, 2000).

Computers may affect productivity because they are a specific capital input to the production process. This is the approach taken in most existing studies, including both the national and industry-level studies just cited, as well as studies at the plant or firm level, such as Brynjolfsson and Hitt (2000), Dunne et al. (2000), Stolarick (1999), and McGuckin et al. (1998). 3 supply side. Most importantly, a differentiated model can characterize the competition in the industry, which cannot be distinguished from the cost effect in the production function approach.

Specifically, our paper examines the effect of IT in a modified Hotelling model with network effects due to Rohlfs (1974), and the theoretical conclusions are tested on a panel data of 68 US banks for the period 1986-2005. The keypoint to understand the inconsistency is to contemplate IT’s influence to the whole industry, rather than to the individual banks. For individual bank, it is true that both cost and network effects are positive.

When all banks in the industry have the same access to this cost-saving technology, will the cost advantage from adopting IT vanish due to competition (in particular, price competition in banking industries). Will the presence of multiple networks bring determinative benefits to each bank in the industry? By investigating the equilibrium in a Hotelling model with network effects, we are able to explore the overall effect of IT to the whole industry. The main findings are summarized as follows. First, we derive a simple test on the existence of network effect by checking the relation between market share and IT expenditure.

If there is only a cost reduction effect, each bank’s market share will increase with IT; however, if there is also a network effect, the market share does not necessarily increase with IT. This result can be useful if a proxy variable for the size of network is invalid (Saloner and Shepard, 1995, use the number of branches possessed by a bank as a proxy for its expected ATM network size in equilibrium). Our test on the US banks shows that, the market share is positively related to IT expenditure indicating that there is a network effect. Second, we are able to distinguish the cost reduction effect fromthe network effects.

Since the equilibrium price will decrease with IT expenditure, if we could isolate this price effect by treating prices as one of the explanatory variables in the model, Proposition 2 shows that if the overall impact of IT on profits is negative, then the cost reduction effect is negative. Moreover, since in this case the market share still increases with IT, this negative result will indicate Berger’s (2003) observation that banks may have essentially “given away” the 4 benefits from the ATM technology in the 1980s as the industry became more competitive due to deregulation, and rents from market power shifted to consumers (p. 42). Our estimation of the US banks also show that if prices are treated as an explanatory variable, the overall impact of IT on profits is negative.

Finally, in line with both sides of the existing literature, we predict that banks’ profits can be positively or negatively related to IT expenditure. In the equilibrium, each bank’s price will decrease with its IT expenditure, but the impact on the profits will have to depend on whether its market share has increased. The overall effect on the whole industry, however, will depend on the relative sizes of weighted sum of IT and the average of IT.

Here, the weight is measured by each bank’s profit share. For the data of US banks, we conclude that banks’ profits are negatively related to IT expenditure, showing that the weighted sum of IT in the US is less than the average of IT. Overall, a differentiated model not only fits in the banking industry more, but also enables us to distinguish the network effect from demand side and the cost reduction effect from supply side. Our empirical study on the panel data of US banks shows that due to severe competition, each bank has over-invested in IT equipment, while the benefits from networks and cost reductions are competed away.

The remainder of the paper is organized as follows. Section 2 presents the modified Hotelling model with network effects and derives three results concerning the relation between IT and equilibrium behaviors. In Section 3, the theoretical conclusions are tested on a panel data of 68 US banks for the period of 1986-2005. Section 4 concludes the paper. 2 TheModel To cope with the observation that banks provide highly differentiated products, we adopt a simple differentiated model (due to Hotelling, 1929) with two competitive banks and infinitely many heterogenous consumers.

Some modifications are made to take into account 5 the network externality caused by the adoption of IT (see Rohlfs, 1974; Milne, 2006. ). We will characterize the market equilibrium after the adoption of IT and derive three testable conclusions concerning the relation between market performance and IT expenditure. To simplify, consider two competitive banks (A and B) in a banking industry, charging P A and P B respectively for services. There is a continuum of potential consumers indexed by x on the unit interval [0, 1] and let us assume that bank A is ocated at 0, while bank B is located at 1. In addition to price competition, each bank invests ei, i = A, B, in IT equipment.

For the individual bank, the adoption of IT has two effects: reducing the operational cost and creating a network effect to customer service. For the first effect, it is assumed that the adoption of IT will cut the operational cost from ci to ci ? ?(ei), i = A, B; For the second effect, we follow Rohlfs (1974)’s setting in assuming that the valuation of service is positively related to the number of consumers in the same service.