Select Committee on Treasury Appendices to the Minutes of Evidence


Memorandum submitted by Shelagh A Heffernan, Professor of Banking and Finance, City University Business School

  This submission is a summary of a report by Heffernan (1999), which was funded by and prepared for the Cruickshank Review of British Banking. Some of the findings appear in chapter 4 of the final report by Dan Cruickshank (2000), though the actual study was not referred to in his report[6]. It subsequently appeared as a working paper, which can be obtained from the website:

  A shorter, more technical version of the paper will shortly appear in the Journal of Banking and Finance—see Heffernan (2002).


  1.  This study used monthly deposit and loan rates to analyse the competitive behaviour of financial firms offering five generic retail banking products from 1993-99.[7] The products tested were:

    —  Savings Accounts: For 90[8] day deposits, at two representative amounts, "High" and "Low"[9].

    —  Higher Interest Cheque Accounts: Again, two representative balances, "High" and "Low" are used. These accounts offer all the features of a current account, paying variable interest rates on the deposit.

    —  Unsecured Personal Loans: annual percentage rates.

    —  Mortgages: Financial institutions quote basic annual percentage rates for new and existing borrowers.

    —  Credit Cards: The study included institutions offering Visa, Mastercard or both to their clients. In addition to the annual percentage rate, information on annual fees levied by the institution was also collected.

  2.  The financial institutions examined in the study included 40 banks, 42 building societies, seven building societies which converted to banks during the period of study, 15 community building societies and 16 other financial institutions[10].

  3.  The analysis focuses on the difference between deposit (and loan) rates and a bench mark for a perfectly competitive rate, Libor, the London Interbank Offered Rate. Libor is the rate banks quote each other for overnight deposits and loans. Libor represents the opportunity cost of all of a bank's assets; for a bank that aims to maximise expected profit, Libor is the basis for determining the marginal revenue for all assets, and the marginal cost of all liabilities. It is an international rate, to which all financial firms have access, and therefore, is representative of a competitive rate. Since retail rates are unlikely to respond to changes in current Libor immediately, the rate was lagged by one, two and three months. The study used four different measures to examine the state of competition in retail banking market over the last decade[11]. These include spread analysis, looking for evidence of smoothing, identifying the extent to which deposit and loan rates deviate from the competitive rate, and ranking financial institutions according to whether they offer customers a relatively good or bad bargain. Each is discussed under separate subtitles, below, followed by a conclusion.


  4.  The spread on a deposit is the gap between Libor and the deposit rate. Consumers benefit from lower spreads because it means they are getting a more competitive rate. The spread was found to vary with the size and type of deposit; the smaller the deposit, the greater the spread. Spreads were higher for chequing accounts than for savings. Over the period 1993-99, the general trend is a slight fall on most deposit products. However, this observation does not hold for all products or institutions. Low level savings accounts are another exception, however, with spreads rising until 1998, and banks enjoy wider spreads than building societies. In the case of higher level savings accounts, building societies converting to bank status display sharply widening spreads.

  5.  For a loan, the spread is the amount by which the loan interest rate exceeds Libor. This, too, displays a slightly downward overall trend in the past decade, at least for personal loans and mortgages.

  6.  Spreads on mortgages are much lower than spreads on personal loans[12], though the latter narrowed over the period. There is some evidence of discrimination between new (offered keener rates) and existing borrowers. The latter are trapped by redemption penalties and switching costs, and hence charged more.

7.  There are differences in the way spreads react to rising, as opposed to falling, Libor. When Libor falls, financial institutions cut interest rates quite smartly but take longer to raise them when Libor is climbing. Again, the reaction to rising and falling Libor differ by product and institution. For example, the converted building societies do not pass on the benefit of a sustained, falling Libor, while the traditional banks do the opposite. During a period of rising Libor, the converted firms pass on the effects more than the established banks. For mortgages, an asset with a long maturity, all firms absorbed most of the effects of a falling or rising Libor.

  8.  In the period 1985-89 (see Heffernan,1993), rising spreads were found on virtually all the products studied. In the 1990s, they fell, though when broken down by type institution, some rose[13]. All of these findings point to some degree of market imperfection: in a highly competitive market, firms would be expected to behave the same way.


  9.  Smoothing means a less than one for one response. Some delay may be explained by the presence of "menu costs", that is, administrative and other costs involved in changing the rates. Also, some building societies announce they will not change rates when market rates are falling to "protect our savers", or to shield mortgagees when rates are rising. There was strong evidence of significant degrees of smoothing for all five banking products.


  10.  Table 1 below summarises the extent to which deposit/mortgage/loan/and credit card rates deviate from a benchmark competitive rate, Libor. Column (1) shows that on the deposit side, the savings accounts are closer to the competitive rate than the interest cheque accounts.

  11.  Compared to the 1985-89 study by Heffernan (1993), column (2) shows the savings accounts and mortgages have become more competitive; the interest cheque accounts, less. The deviation for personal loans has hardly changed. There is supporting evidence of a mark-up added to the existing borrower rate that is 2.5 times greater than the new borrower rate.

  12.  Column (3) of table 1 shows how long it is taking these products to respond to a change in the competitive rate. Interest cheque accounts, credit cards, and personal loans all show a slow response of three months; the other products respond with a lag of one to two months.

Table 1


Deviation from
Rate 1993-99
Deviation from
the Competitive
Rate 1985-89
Length of time it takes to respond to a change in the competitive rate
Savings Accounts
1-2 months
Interest Cheque Accounts
3 months
Mortgage Rates—Existing Borrowers
1-2 months
Mortgage—New Borrowers
1-2 months
Credit Cards
3 months
Unsecured Personal Loans
3 months

  Na: not possible to report a figure because credit cards and new borrowers were not part of the 1985-89 study.


  13.  This test is based on an idea developed by Salop and Stiglitz (1977). With unequally informed consumers, there can be two equally profitable prices a firm can charge, a high one to a few, uninformed customers, or a low one with a higher volume. Thus, relative bargains and rip-offs co-exist in the same market.

  14.  For example, consider the London restaurant market. There is general agreement this sector is highly competitive, because there are so many firms. However, if the amount of information about the quality and price of these establishments varies among consumers, then it is possible for some restaurants to offer bad deals to customers and stay in business. The ill-informed diner who samples London restaurants on an occasional basis might choose a highly priced, low quality restaurant. He or she will never return again, but given a sufficient number of ill-informed consumers, the restaurant will remain a profitable business. Well-informed diners chose value for money restaurants which, likewise, stay in business, depending on regular clientele.

  15.  In the retail banking market, a similar situation exists. Some consumers are well informed, others are not. The econometric test for the presence of relatively good and bad buys is reported in Heffernan (1999, 2002). Table 2 (see end of submission) shows large differences in the ranges between the relative best buys and worst buys, depending on the product. The first or "product" column in table 2 ranks the mortgage, chequing, savings, unsecured personal loans, and credit card offerings according to the size of the margin between the relative best bargain and worst rip-off. For example, consider personal loans. The second column of table 2 shows a difference of 8.7 per cent between the best and worst loans.

16.  According to table 2, mortgages have the smallest margins, not only because of the number of players, but because the size of a mortgage (the largest investment the majority of individuals make in a lifetime) and its duration mean most potential customers will shop around for the best deal. Furthermore, these mortgages are considered "safe as houses" by banks because of the tradition for real property market values to rise[14]. Once locked in however, there appear to be opportunities for firms to practice price discrimination on existing borrowers.

  17.  From table 2, there appears to be persistent (but not large) relative bargains for existing and new mortgages being offered by Derbyshire, First Direct, Nationwide, and the Yorkshire Building Society.

  18.  The margin of less than 1 per cent between the best and worst low chequing accounts is explained by the very low market interest rates throughout the period. Since interest rates cannot fall below zero, banks encounter a floor, with very little room to offer relatively bad buys, i.e. low deposit rates. In the 1985-89 study, when nominal interest rates were comparatively high, there was a difference of 3.74 per cent between the best bargain and worst rip-off.

  19.  One possible interpretation of table 2 is the use of loss leaders by financial firms. One product might be priced below cost to attract new accounts, with losses defrayed by higher prices on other products. However, firms typically engage in this type of behaviour for a short period. The results in table 2 are from estimating equations over a relatively long period, 1993-99, indicating the sustained persistence of relative bargain and rip-offs.

  20.  Firms may also price products below cost or cross-subsidise to attract customers into a form of relationship banking, expecting to obtain services at reasonable (if not the most competitive) prices. This might help to explain the findings in table 2, though British banks, in general, are not known for this type of banking in the mass retail market. A more likely explanation is the presence of switching costs, causing consumer inertia that can be profitably exploited sometime after the customer has opened an account.

  21.  Notable also from table 2 is the absence of any consistency in the position of financial institutions. For example, Northern Rock, which converted to a bank, offers the best buy for unsecured personal loans and the interest cheque account for the low amount, but is the worst deal for mortgages.

  22.  Apart from the odd exception, the big four clearing banks usually offer relatively bad bargains. It is the smaller players that offer the best deals for the interest cheque account and mortgages: the relative best and worst buy places are occupied by comparatively small banks or building societies.


  23.  The results of this study suggest competition has increased in some retail banking markets in the 1990s, compared to the later half of the 1980s. The mortgage market, in particular, seems to have become more competitive[15]. However, the price comparison did show existing borrower rates to be less keen than those for new customers.

  24.  Unlike the mortgage market for new borrowers, there are indications of substantially less competition for savings accounts, interest earning cheque accounts, personal loans, and especially, the credit card market.

  25.  Imperfect competition continues to prevail in the retail banking market, but no single "model" describes the way firms behave in the pricing of their products. However, for many of the products, the "good buy/bad buy" model fits well with the results. Some firms remain profitable by offering relatively bad buys to the poorly informed consumers. An increase in the number of firms entering the market (because of, say, a fall in fixed costs) could actually increase the number of relative rip-off bank products.

  26.  Cruickshank (2000) argued that additional regulation in personal banking is unwarranted[16] because increased entry will increase competition. This conclusion is correct, but only in the absence of consumer inertia and in an environment of full or more complete information about the products on offer.

  27.  Policy should be aimed at encouraging banks to supply consumers with comparable information, so it is more difficult for the relatively bad buys to survive in the market. Also, financial institutions should be required to make it easy for customers to change banks. In the meantime, consumers should search out the best buys on a product by product basis, and then stick with that firm (or firms), but be on guard for any decline in the competitive pricing of a product, and/or the introduction of new, close substitutes by that financial institution.

* - subsidiary of Robert Fleming Bank ** - subsidiary of National Westminster Bank.


  British Bankers' Association (1998,1999), Banking Business: An Abstract of Banking Statistics, vols 15,16, London: BBA Enterprises Limited.

  Cruickshank, D (2000), Competition in UK Banking: A Report to the Chancellor of the Exchequer, London: HSMO.

  Heffernan, S A (1993), "Competition in British Retail Banking", Journal of Financial Services Research, 7, 309-322.

  Heffernan, S A (1999), "Competition in British Retail Banking, 1993-1999"—Report for the Cruickshank Review of British Banking, September, 1999. Subsequently appeared as a working paper, which can be obtained from the following website:

  Heffernan, S A (2002), "How do UK Financial Institutions Really Price their Banking Products?", Journal of Banking and Finance, October.

  Salop, S and J Stiglitz (1977), "Bargains and Ripoffs: A Model of Monopolistically Competitive Price Dispersion", Review of Economic Studies, 4,493-510.

6   Many background papers were not cited by Cruickshank, and it would be helpful for researchers if a full list of references could be provided. Back

7   The author is grateful to MONEYFACTS for supplying the data, especially the Chairman of the company, Mr. John Woods, and Mr. Mike High for their generous help. Back

8   Financial institutions offer a large range of deposit products. In early tests done for Heffernan (1993), firms demonstrated similar pricing behaviour across deposit products that required notice of withdrawal (to avoid interest penalties) and do not have current account facilities such as a cheque/debit card or a chequebook. For this reason, the 90 day account was chosen as the representative deposit product. Back

9   Data from the Abstract of Banking Statistics (1998,1999) were used to obtain the representative amounts. See appendix 1 of Heffernan (2000) for an explanation of how these representative amounts were computed. Back

10   Community building societies limit their customer base by geographical residency requirements, and other financial institutions are all firms that fall outside the other four categories but offer a limited range of financial products, such as the various mortgage corporations. Given space constraints, the community building societies are not discussed in this summary. Back

11   Where possible, the results are compared to those reported by Heffernan (1993) for the period 1985-89. Back

12   The lower mortgage spreads are partly explained by overhead costs being spread over larger sums, longer time spans, and a lower risk premium because the property acts as security. Back

13   No comparison is possible with credit cards because they were not part of the earlier study. Back

14   The exception to the historical rule was the negative equity crisis from 1990-93, and a high repossession rate from 1991 to 1995 that led managers to raise risk premiums. However the action was temporary, and the episode increasingly forgotten or treated as a non-recurrent risk by 1997. Back

15   This observation is subject to the proviso that non-price features, such as redemption clauses, were not part of the study. Back

16   Cruickshank (2000) made numerous policy recommendations but only those directly relevant to this study are discussed. Back

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2002
Prepared 4 September 2002