Jump to ContentJump to Main Navigation
The glass consumerLife in a surveillance society$

Dr Susanne Lace

Print publication date: 2005

Print ISBN-13: 9781861347350

Published to Policy Press Scholarship Online: March 2012

DOI: 10.1332/policypress/9781861347350.001.0001

Show Summary Details

Data use in credit and insurance: controlling unfair outcomes

Data use in credit and insurance: controlling unfair outcomes

Chapter:
(p.157) 6 Data use in credit and insurance: controlling unfair outcomes
Source:
The glass consumer
Author(s):

Harriet Hall

Publisher:
Policy Press
DOI:10.1332/policypress/9781861347350.003.0007

Abstract and Keywords

This chapter demonstrates how the use of personal information can expand overall access to credit and insurance. It focuses on impacts that are potentially exclusionary, to understand better the distribution of the risks and benefits of personal information use — and to consider the scope for creative solutions to new forms of exclusion. It also discusses what is known about the adverse effects data use can have on access to credit and insurance and then examines the success of policy responses.

Keywords:   credit, insurance, personal information, information use, data use, policy

Introduction

This chapter considers how information on consumers is used in credit and insurance. It is not a chapter on privacy concerns, but on how data use can affect the availability of these services and how this may produce unfair outcomes.

Before considering the negative effects, it is worth pointing out that data manipulation is fundamental to modern, mass market credit and insurance. Credit and insurance share the fact that data are collected on millions of individuals to define the characteristics of those who are creditworthy or those who are an acceptable insurance risk. Data belonging to an individual are then set against these definitions, in order to decide whether and on what terms to grant credit or insurance. Before electronic collection and manipulation of data developed, many consumers were not able to buy credit or insurance at all (NCC, 2004).

Nevertheless, increasing sophistication in data manipulation means that lenders and insurers can segment their market and this may result in some consumers being excluded or only offered services on unfavourable terms1.

This chapter examines categories of problems outlined by 6 and Jupp (2001). They point out that whereas in the past concern has been for people excluded from information, there is an additional concern that the use of information technology might itself be an engine of exclusion. They set out three categories of risk.

The first is that “organisational pressures … may lead organisations (p.158) to identify and focus on only the most profitable customers” (6 and Jupp, 2001, p 42), thus depriving less profitable customers of services that are essential to everyday life. They use geographical illustrations for this, citing the closure of bank branches in areas of deprivation and the use of data to site supermarkets where they will be accessible to more profitable customers. Discrimination may not be unjust in individual terms and may be based on significant characteristics of those excluded that are commercially justifiable. But it may have socially undesirable consequences. In some circumstances credit and insurance are seen as quasi ‘essential services’. In these cases, consumers' lack of access can be seen as unfair in the context of needs. I call this unfair exclusion.

The second type of problem that 6 and Jupp identify is that “data systems will be accurate enough to enable companies to target the worst off customers and heavily market services to them that in fact worsen their position” (6 and Jupp, 2001, p 42). In a sense this is a variant of the first problem: it is a question of who the more profitable customers are. 6 and Jupp cite the example of the use of data sets on people in financial difficulty to sell them expensive forms of debt. The companies justify the high fees and rates of interest by pointing out the higher risk of default, claiming that they are actuarially fair. Nevertheless, the loans may increase the problems of those who take them, rather than solving them. I use the term unfair targeting for this type of problem.

Finally there is the concern “that information systems will not be accurate enough to enable this kind of focusing on the more profitable and that there will be problems of unjust discrimination based on crude categorical information” (6 and Jupp, 2001, p 42). In this case it is not the sophistication of the use of data that allows firms to target the more profitable, but the lack of sophistication that places people into undifferentiated or inaccurately differentiated categories. People are denied goods or services on the basis of belonging to a set that does not take into account other data that might make them profitable and suitable purchasers. I use the term unfair discrimination for this.

The chapter examines how data are used in credit and insurance, what problems relating to the categories identified by 6 and Jupp (p.159) are evident, and what attempts have been made to tackle them. It then attempts to draw some conclusions.

Data collection and manipulation

This section sets out how information is collected to make decisions on the granting of credit and insurance.

Data and credit risk assessment

Credit reference agencies

Credit reference agencies collect information on loans to help lenders to assess the credit risk of an applicant. It should be noted that there is no obligation on a lender to consult a credit reference agency.

There are three credit reference agencies in the UK, which collect information on individuals and make it available to lenders for the purpose of deciding whether or not to grant credit to an applicant. I summarise here the information given by Experian (www.experian.co.uk). Experian collects information on individuals that is publicly available. The data come from the electoral roll, from county court judgements held by the Registry Trust2, and from records of bankruptcies and individual voluntary arrangements.

In addition, it holds data in a closed membership group called Credit Account Information Sharing (CAIS). The other credit reference agencies have different closed groups. The group is closed in the sense that the information is only available to members. Members submit information to Experian on their customers' credit accounts, including debts repaid in the recent past and repayments made on current loans. The information includes credit given in the usual course of supplying services such as mobile phones and utilities. The information may be on failure to keep up with payments (negative information), loan repayments that may be delinquent (that is, three months or more in arrears) or in default (where the relationship has broken down and repayments are not being made) and may also include information on accounts where the consumer (p.160) is keeping up with the payment according to the contract (positive information).

A lender that consults an agency is not obliged to submit data to the closed membership groups, but will not be able to consult the data submitted by the members of the closed groups if it does not. In addition, a member firm can choose the extent to which it contributes to the network of information, so it can choose whether to submit and take out only negative information or whether to include positive information.

This system of data sharing through credit reference agencies is governed by a scheme of self-regulation controlled by the Standing Committee on Reciprocity (SCOR), run by trade associations in the credit industry (British Bankers Association, Council of Mortgage Lenders, Finance and Leasing Association, and others) and the three credit reference agencies. The Principles of Reciprocity set out the conditions under which firms submit information on borrowers or credit applicants and take it out from credit reference agencies. The fundamental principle is that data may only be shared in order to promote responsible lending. It may not be shared for marketing purposes, although a firm may ‘wash’ its list of potential clients to exclude bad credit risks.

The following information is also shared under industry-wide schemes for data sharing by companies:

  • Council of Mortgage Lenders' repossessions register;

  • Credit Industry Fraud Avoidance System (CIFAS) – data on fraud committed or attempted by customers or by other persons on customers;

  • Gone Away Information Network (GAIN) – information submitted on customers who have left an address while owing money without leaving a forwarding address.

When someone applies for credit from a lender that uses a credit reference agency, he or she must be told that a check with the agency will be made and asked to give consent to the data on the application form and future data on the management of the account, if credit is granted, being shared with a credit reference agency3.

(p.161) Credit scoring

Credit reference agencies are only one part of the decision to grant credit. A company may consult an agency but, as mentioned above, is under no obligation to do so. If it does so, it will have to interpret the information and almost certainly use additional information. The agency file, for example, will produce no data on age or employment, which may be relevant to a decision to grant credit. Firms use their own scoring systems, since they will have different approaches to risk: some may accept a higher level of risk of default and charge higher interest rates. Others may target good payers with low interest rates. Companies may also have different views on what risk factors are reliable predictors.

Some firms develop their own systems; others buy credit-scoring systems from credit reference agencies. All systems use historical data relating to the personal characteristics of existing customers, such as their age, salary, housing tenure, and analyses their risk potential against how they performed their obligations under loans granted to them. In addition, some companies may make inferences as to how rejected applicants would have performed if they had been granted a loan. This is called reject inference and is done by assuming characteristics that the accepted and rejected share or by using data from a credit reference agency as to how the rejected applicants performed with other lenders.

Scoring systems enable a scorecard to be developed with different weightings for the various distinguishing characteristics and potential customers are judged on the basis of their scores. If lenders are behaving rationally, credit-scoring systems should be tuned so that people are only excluded if there is statistical evidence that they are at risk of defaulting, the level of risk being set by the company's business model.

Data and insurance pools

Insurers collect data from a large number of events, and calculate the likelihood of the event occurring and the likely cost of it occurring if it does. Firms develop their own actuarial systems, using data to assess risk. For reasons of space, I do not propose to go into (p.162) the details of data sharing used by insurers. As an example, Experian holds two relevant databases: the Perils Data Portfolio – postcode sector-level data on building stock, subsidence, flood, wind and crime; and the Claims and Underwriting Exchange (CUE) – comprising databases of Household Personal Lines Claims and Motor Personal Lines claims. The CIFAS register, referred to above under credit reference agencies, also contains information on insurance frauds4.

Faced with an individual seeking insurance, insurers can apply the individual's profile to the model. This may be done without individual underwriting, that is, treating the consumer as if he or she had no particular characteristics that would make the risk greater than for the average person. This is how travel insurance is usually designed. Insurers will attempt to exclude people with such characteristics from cover altogether through the terms and conditions. This can cause problems if the exclusion clauses in the contract are not properly drawn to the consumer's attention.

Individually underwritten insurance involves looking at information belonging to the individual applicant and assessing the extent to which it deviates from a norm and then pricing the product accordingly, or in some cases writing in special terms or refusing to insure.

Difficulties can arise from the defining characteristics of the insured pool. Insurance works on the basis that although some people may suffer the loss insured against, others will not, so those who do not subsidise those who do. Insurers regard it as a matter of actuarial fairness that the level of risk of the event occurring should be about the same for those in the pool of insured. Most people would agree that fairness requires that, for example, a person with a conviction for dangerous driving should pay more in premiums than someone who has no convictions. There are two pools, one for those with convictions and one for those without.

Unfair exclusion

This section considers the first of the potential problems identified by 6 and Jupp (2001), that as companies find it profitable to use data to segment the market, they do so to such an extent that some people or groups of people may be excluded from goods or services. (p.163) To the extent that these services are essential to everyday life, this exclusion may be regarded as unfair. The behaviour of the company may not be unfair, but the outcome for the excluded individual is.

Unfair exclusion in credit

Datamonitor research (2003) shows that about 7.8 million adults are barred from mainstream credit because of credit scoring. The risk of their defaulting on a loan is too great for them to be accepted by mainstream lenders because of their income or credit histories. The sophisticated manipulation of data by mainstream companies excludes them, as it is designed to do. In a climate where companies are criticised for offering too much credit, it would be odd to argue that they should ignore the results of credit scoring to give credit to those who are at risk of defaulting.

Nevertheless, those who live on low incomes may need credit, not for luxuries but to tide them over or to cope with emergencies. To the extent that benefit levels or minimum wage levels are set so low that borrowing is necessary to deal with cashflow problems on a regular basis, it can be argued that access to credit is an essential service5.

Responses to unfair exclusion in credit

What follows summarises a number of responses that have been made to the problem of exclusion from credit. They all have drawbacks, which make them less than ideal as solutions.

Door-to-door credit firms

There are firms that specialise in providing credit for consumers who otherwise would find it hard to get it, through door-to-door credit offerings. Borrowers who use them like them because they are the only kind of credit available to them, loans are available in small amounts and because those selling credit come to the borrower's home and there is a discipline in weekly collections of repayments (NCC, 2004). Customers for this type of loan appear to be targeted by the collectors, who have local knowledge of (p.164) consumers and family connections between borrowers, rather than data manipulation. Kempson and Whyley (1999) point out that some targeting is done by getting round the prohibition in the 1974 Consumer Credit Act on selling credit door to door.

While home credit is an example of a response to exclusion from credit, firms that offer it are the subject of much criticism for high interest rates and for encouraging further borrowing before one loan has been paid off, although Kempson and Whyley make the point that not all loans are at extortionate rates of interest and these firms serve a particular market that would otherwise struggle to get credit. The companies say the interest rates are justified on the grounds of the expense of providing a service that suits the needs of their customers. The National Consumer Council (NCC) has recently been successful in a making a supercomplaint6 to the Office of Fair Trading (OFT) on the uncompetitive aspects of this type of door-to-door service. In December 2004 it was announced that the Competition Commission would conduct an investigation which is expected to take over one year to complete.

Not-for-profit supply

Commercial firms may not be able or willing to grant credit to low-income consumers on other than very high interest rates and other restrictive terms. So, alternative sources of supply have been developed by not-for-profit organisations. Some success has been achieved in this field by organisations such as credit unions or housing associations that stand as surety for loans to their tenants made by local building societies, but by definition these are very local and advantage only a small number of people. The latest figure on the Financial Services Authority's website for credit union members was around 475,000 at June 2004.

The Social Fund

Another source of credit for those who are not served by mainstream credit companies is the Social Fund. The fund provides budgeting loans to those who have been on a number of income-related benefits for more than 26 weeks. The grant of a loan is discretionary. (p.165) Those who apply often do so because they know there is no other source of credit available to them. Users, while on the whole positive about it, complain that the scheme is a bit of a lottery (only 60% of applications are granted) (Kempson et al, 2000). Older users complain about the inflexibility of the repayments required and the length of the term of the loan (Kempson et al, 2002).

Unfair exclusion in insurance

In insurance, analysis of a consumer's data given on an application form in some cases results in refusal, or an offer to insure on terms that are too expensive for the consumer. The risk profile of the applicant does not match the insurer's risk pool. This is inevitable, given the use of data in the insurance industry. Again, as in credit, this is the purpose of the use of data. So, is insurance different from any other commercial product, which consumers may or may not be able to buy? Is there any way in which it can be regarded as an essential service, so that those refused can be seen as unfairly excluded?

I shall look at two cases in which insurance could be seen as a quasi-essential service, in property and home contents insurance. It is not possible to buy a house that is not insurable on a mortgage. If a home becomes uninsurable after purchase, it is arguable that there should be some response to make insurance available. In the case of home contents insurance, people living on low incomes in areas of high deprivation are more prone to burglary than those in other areas. Through low incomes they are likely to be less able to selfinsure and their inability to replace stolen goods is compounded by the fact that they may not easily get credit.

Property insurance in areas liable to flooding

For property insurance, underwriting is generally done partly through assessment of the average risk applying to houses in the same postcode area, although insurers usually use multiple discriminators7. The data on loss from various causes in any area are aggregated to produce a way of predicting how many houses are likely to be affected in any year and at what cost. This has started to (p.166) create problems in postcode areas where houses are so likely to suffer loss that it can no longer be regarded as a matter of uncertainty. Insurers are then unlikely to offer insurance, or may only do so at premium rates that consumers cannot afford. In some cases they would do better to self-insure.

Although I have included this as an example of unfair exclusion, it is not necessarily a matter of insurers concentrating on profitable business to the exclusion of the unprofitable. It is more that the model of insurance, which requires a certain degree of risk, does not match a situation where the risk of the insured-against event happening is too high.

Insurers who assess risk on postcode data are reluctant to provide cover for property in areas that are subject to frequent flooding. Flooding has become more frequent in recent years, perhaps because of global warming, perhaps because builders and local authorities have allowed development in river floodplains. An additional problem may have arisen because of policy decisions to reduce flood prevention barriers.

When this became evident as a problem, the government sought the help of the Association of British Insurers (ABI). The government sought to influence insurers to do what they, commercially, possibly could not justify. In January 2003 the ABI announced that:

  • in urban areas where the risk of flooding is one year in 75 or better, insurers will continue to insure, although premiums will vary according to risk;

  • where the risk is greater than one in 75, but improved flood defences to bring it to this level are planned for completion before 2008, insurers will maintain cover, including for purchasers from the current owner;

  • where no such defences are planned, insurers will not guarantee to maintain cover, but will look at risk on a case-by-case basis;

  • in return, the ABI expects the government to carry out planned expenditure on flood defences, to implement the findings of the review on flood defences, to give new planning guidelines to prevent building in areas of high risk and to improve early flood warnings (ABI, 2003).

(p.167) In effect, what this solution does is to oblige insurers to widen the pool more than they might actuarially prefer. If they could exclude these properties that make frequent claims, they could reduce the premiums for everyone else. The owners of higher-risk properties are subsidised by the owners of lower risk properties. The reason why it works, in as much as it does, is because all insurers, through their trade association, agree to do the same. In theory, assuming all insurers are covering a selection of properties within the areas affected, they are all obliged to increase their premiums above what they could charge if they refused cover. There is no competitive advantage to be had. It does, however, leave out some householders with very high-risk properties. I deal with this more fully below.

Home contents insurance

A report for the Joseph Rowntree Foundation (Whyley et al, 1998) considered access to home contents insurance for low-income households. The lack of availability can be seen as an example of insurers failing to meet the interests of poorer neighbourhoods, finding more profitable business elsewhere. The research found that few were actually refused insurance, but many were uninsured because they could not afford the premiums, did not think their possessions worth insuring, found the minimum insured value too high or could not manage premiums collected in a lump sum once a year. The products of mainstream insurance companies were too inflexible or too expensive, or both.

The Rowntree project looked at a number of solutions to these problems, including offerings by commercial insurers that concentrate on the needs of low-income neighbourhoods and by partnerships between insurers and local authorities and sometimes housing associations, which collect premiums with rent. The latter solves some of the problems by pooling all local authority tenants together, whatever the risk level of the property, thus reducing premiums for those living in homes with a high risk of burglary, and allowing weekly collected premiums and a lower minimum value insured. Again, some subsidy of higher-risk tenants by lowerrisk tenants is involved.

(p.168) Adequacy of responses

If insurance in these cases is an essential service, then we need to examine whether the responses that have been developed meet the needs of those excluded. In the case of property insurance, if insurance against damage by flooding meets an essential need, what kind of protection will be provided for those who remain uninsurable, despite the ABI agreement? And if flooding is merely the first sign of problems arising from changing weather patterns due to global warming, what does a policy of intervening through self-regulation of insurance companies mean for damage caused by other natural causes? Recently, the ABI published a report on climate change and the need for insurers and government to act together to control damage (ABI, 2004). Is there an argument for a government fund for damage from these causes rather than seeking to make adaptations to normal insurance behaviour to meet the losses?

Unlike the problem with insurance and flooding, pressure to create a solution to a lack of home contents insurance has not come from central government – which felt no need to act to produce a countrywide solution – but solutions have been developed by authorities with a concern for their tenants or constituents. These solutions to a lack of access to an essential service are, consequently, patchy. They do not meet the needs of all those affected.

Unfair targeting

The second problem identified by 6 and Jupp (2001, p 42) is that “data systems will be accurate enough to enable companies to target the worst off customers and heavily market services to them that in fact worsen their position”. Targeting with data use in insurance is likely to be a benefit to consumers. (Insurance companies, for example, use data to identify and target members of affinity groups who suffer a particular disease to offer them access to travel insurance, which might not be available on standard travel policies.) I shall, therefore, consider only credit in this section.

In credit, are firms deliberately using data to target those who are over indebted, whether or not the loans they offer are overpriced, (p.169) identifying them by manipulating data? 6 and Jupp (2001) quote this as happening, without giving any hard evidence. A report on extortionate credit by the Personal Finance Institute at Bristol University (Kempson and Whyley, 1999) makes the statement that some sub-prime lenders buy lists of those refused credit by mainstream lenders, although no details are given8. Those who use door-to-door credit (see ‘Unfair exclusion’ above) are often thought of as unfairly targeted, although it is not clear whether borrowers are primarily targeted by use of their data.

There is a serious concern about those with multiple debts who are encouraged to consolidate their debts in one loan, sometimes secured on their home (Kempson and Whyley, 1999). The loans are sometimes at punitive rates of interest and often with severe penalties for late payment. It may, however, not be necessary to use data to target customers in difficulties for loans, as advertising produces applicants who choose to respond. Watching daytime television is instructive. Advertising credit at a non-peak time, when those out of work are likely to be watching, appears to pay off.

A scheme which involves targeting consumers using their data was set up in 2003 by Barclays Bank. Under this pilot scheme, the bank referred customers who had been refused loans on standard terms to Welcome Financial Services, a subsidiary of Cattles plc. Cattles is a company with the stated aims of serving the needs of those who cannot or do not want to access mainstream financial services. The loans given by Cattles were at rates from 20.9% to 35.9%, higher than mainstream loans by Barclays. If the loan was granted by Welcome, Cattles provided the processing, administration and monitoring, while Barclays provided the funds.

Recognising that it lacked the expertise in serving customers in this section of the market, Barclays used a different business model to give access to credit to those who were not creditworthy according to their usual scoring system. Was targeting these using data providing a service, albeit at a higher price to reflect the higher risk, or unscrupulously taking advantage? Barclays argued that it was allowing those with poor or limited credit ratings to show themselves worthy of a better score and thus more mainstream loans. Until there is further analysis of this experiment, including whether it was sustainable and profitable for Barclays, looking at any default (p.170) by borrowers and at the terms of the loans to determine how easily borrowers could switch once they had shown themselves to be less risky, it is difficult to decide what, if any, control should be placed on firms seeking to identify borrowers by means of data on those refused ‘mainstream’ credit9. Targeting may not always be unfair. And as mentioned above, it may not always involve targeting using data.

Control of unfair targeting: credit

Statutory regulation

In theory at least, the 1998 Data Protection Act should provide some protection for those who find themselves on a list of the over-indebted, which might result in targeting. When applying for credit, applicants have to be told the purposes for which information is collected and consent to it. If it is to be collected to collate a list of those over-indebted so that additional loans can be marketed to them, they should know. However, even if the information is given in a sufficiently explicit form for them to understand, consumers are very unlikely to read that far, or to reject a loan on that basis. A database of those on publicly available lists, such as the list of county court judgements, would in any case be outside this notional protection.

The regulatory response to unfair targeting has tended to concentrate not on controlling the use of data but on examining the events around an application for a loan and seeking to prevent lenders granting a loan to those who will struggle to repay. This may reflect the fact that targeting using data is not seen to be the problem, or the main problem. However, some additional control on using data does exist.

The OFT has a code of conduct (1997) to which lenders who give secured credit to non-status borrowers (those with impaired credit histories or who find it difficult to borrow on ‘normal’ terms) should adhere, as a condition of their credit licence. Much of the code relates to lenders being open with borrowers and requiring them to assure themselves of the prospective borrowers' ability to repay. The only part of the code that might deal with questions of (p.171) targeting using data collected from those with poor credit histories relates to cold calling. Lenders and brokers must avoid engaging in marketing credit by telephone to customers known to be or likely to be non-status, if the credit offered would be secured. There is no similar restriction on marketing secured credit to non-status borrowers using databases of those with poor credit histories by other means, for example by targeted mailshots.

The OFT announced in June 200310 that it would undertake a report on the debt consolidation industry, including advertising and marketing. It also announced a review of the 1997 non-status lending code. Both could offer an opportunity to control targeting through mailshots or other uses of databases consisting only of names of people who are known to be struggling with debt or who have been repeatedly refused credit. But the exact terms of the regulation would have to be carefully considered, since a company that offered reasonably priced credit on acceptable terms should not be prevented from identifying those who would benefit from transferring their debts.

The Department of Trade and Industry's (DTI) White Paper on consumer credit (2003) proposed to overhaul consumer credit legislation. There was, however, very little that related specifically to unfair targeting in the proposals, beyond a requirement to show annual percentage rates (APRs) in advertisements in circumstances where they would otherwise not be required, if the advertisements were targeted at those who might be credit restricted. Unless advertisements refer here to direct mailshots, this does not directly relate to data use.

The DTI paper states that 7% of the population fall within the criteria for those likely to be over-indebted, that is those who:

  • have four or more current credit commitments;

  • spend more than 25% of gross income on consumer credit; or o spend 50% or more of gross income on consumer credit and a mortgage.

The DTI should consider some kind of codification of these triggers into regulation and self-regulation, to require firms to be able to (p.172) justify either targeting for marketing or lending to consumers who are in any of these situations11.

The role of SCOR

As described above, data sharing in credit is regulated under the Principles of Reciprocity by SCOR for those lenders who choose to join the closed member groups. While the self-regulatory system owes its existence to the fact that firms need to be able to trust each other and to prevent poaching of each other's customers, a fundamental principle is to promote responsible lending. The Principles of Reciprocity that govern data sharing should have an effect on unfair targeting.

Research carried out for the Financial Services Authority by B. & W. Deloitte in 2002, reported in the Financial Risk Outlook for 2003 by the Financial Services Authority, revealed that 6.1 million households found it moderately difficult or difficult to meet their debt obligations, with 1.8 million households falling in the difficult category – owing in aggregate £46 billion. This figure includes both secured and unsecured lending.

With such widespread difficulties reported, the emphasis in government circles has recently been on preventing further lending to those already over-indebted. This could, in theory, be achieved by a requirement for responsible lending, which in turn could be achieved by a requirement to consider data on existing debt and not lend to those who look as if they will struggle if granted further credit. This is the stated purpose of data sharing under SCOR. Unfair targeting should be prevented if the data are only used in accordance with Principles of Reciprocity. However, there are some difficulties:

  • Membership is voluntary and firms can decide whether to contribute positive and negative data or negative data only.

  • Data shared does not contain information on how well consumers keep up with payments other than loan repayments; thus rent, council tax and so on are not included. Arrears on these commitments will not show up in the credit reference agency file until a judgement has been obtained or in cases where consumers have given consent to the information being shared.

  • (p.173) While data concentrates on how borrowers keep up with payments, this does not take account of the fact that borrowers are not shown in default provided they are keeping up with minimum payments; sometimes these are very low and paying off a debt while meeting them would take a considerable time; but while they are met, borrowers can get another loan, as their credit status is unimpaired.

Arguments have been made that to encourage responsible lending, greater data sharing should be encouraged, or indeed required, to identify those who should not be offered additional credit. There are, however, arguments against this. It is disputed that information on consumers' commitments is predictive of the likelihood of default on loans (DTI, 2001). In addition, small, local not-for-profit or even commercial organisations, which wish to develop schemes for affordable credit, find the cost of checks at a credit reference agency expensive. Compulsion to share non-credit and positive credit data could make this worse. A requirement to consult a credit reference agency before offering a loan could have a significant effect on the availability of credit to some consumers.

The role of SCOR has wider ramifications than that of simple self-regulation to protect the interests of members. Some of the ways of improving credit risk assessment outlined above could only be definitively solved by a legal requirement to share data both positive and negative, and non-credit data. Whether this is desirable or not requires more detailed research into consumer behaviour once granted credit, the predictiveness of data on outgoings other than loan repayments and the effect on competition.

Given the interest of the government in promoting responsible lending, it is surprising that this aspect of data sharing remains in the hands of a self-regulatory body, particularly since there is little evidence of how the rule that requires data to be used only to ensure responsible lending is enforced. Unless it is enforced, companies can manipulate data on both good and bad payers to target consumers who will be a good risk (and thus make credit more expensive generally for bad payers) or target bad payers, who could be profitable because of the terms of the loan. The Principles of Reciprocity are not particularly transparent. There is no central (p.174) secretariat and there is certainly no non-industry representative on the board. Enforcement is not open to scrutiny12.

A report for the Centre for Studies in Economics and Finance at Salerno University (Jappelli and Pagano, 2000) shows that in all EU member states except the UK there is a public credit reference agency and in some countries there are no private agencies. The report also comments on the drift towards consolidation in the industry. A state-funded reference agency could ideally make an assessment of what is in the interests of consumers in relation to what kind of data are gathered and how much should be shared and at what cost. Whether state-funded bodies abroad do so was unfortunately not the subject of the report. The NCC might wish to consider whether any improvement in the sharing of data under SCOR should be made.

Unfair discrimination

Unfair discrimination: credit

The third and final problem indentified by 6 and Jupp (2001) is unjust discrimination based on crude categorical formation. In the area of credit, a number of differentiators are used that may exclude people from mainstream credit under the system of credit referencing and credit scoring. Many are excluded because of county court judgements or because they have no bank account. The selfemployed, who number 3.2 million, are frequently refused mainstream terms for credit unless they can show three years' employment history and can produce records to be scrutinised by an accountant. In addition, one of the factors frequently taken into account is type of tenure of the home. Home ownership increases your score even if you are not seeking a secured loan.

I have no evidence to show that risk factors identified in this way are irrelevant. As I said above, if the market is working well there should be no incentive for a firm to refuse credit where there is no statistical evidence of the likelihood of default. Bridges and Disney (2001) quote research by Japelli in the US showing that the probability of a household being credit-constrained declines with an increase in family disposable income, age and savings. Crook, (p.175) also quoted by Bridges and Disney, finds that the probability that a household is credit-constrained is negatively related to the number of years at the same address, ownership of the main residence, and the number of years of schooling of the head of the household. These seem fairly plausible risk factors. However, Bridges and Disney say that information on credit use in the UK is hard to come by, so that evidence of credit scoring producing unfair results for some would be difficult to find.

Some people may not get credit because of a personal history that is seen as a risk factor. Others are put in the category of being uncreditworthy because they have no financial record to show that they can keep up with repayments. They are discriminated against by the fact that they have not borrowed before, or have not borrowed from those who share data with other lenders, even though they may be able to keep up with repayments. As seen above, membership of closed interest groups in credit reference agencies is voluntary. The effect is that firms, such as pawnbrokers and home credit companies, which make a practice of lending to ‘poor risks’, do not have to share their data and no mainstream lender will be able to check on how well a consumer has kept up with a loan. Borrowers will not be able to improve their credit scores and so may have to rely on their existing lenders13.

Bridges and Disney (2001) point out that there is also the potential for discrimination against low-income applicants based on something other than the obvious discriminators such as income, household tenure and age. This is because the credit-scoring card can be designed on a population of people who have been granted credit and the scores are set according to how they perform as good or bad risks. Since there is a population of those who either do not apply, expecting to be refused credit, or apply and are rejected, the score does not take into account the possibility that people on very constrained incomes nevertheless might manage to keep up with loan repayments. Some companies use reject inference to overcome this problem, but the accuracy of this is the subject of some debate ‘in the scoring literature’ (Bridges and Disney, 2001). Clearly this is another area that needs developing if unfair discrimination is to be avoided.

(p.176) Control of unfair discrimination: credit

Statutory regulation

At the level of the individual, where data are recorded incorrectly or inaccurately, an applicant may be discriminated against and refused credit. The 1998 Data Protection Act addresses the question of the accuracy of information and its use. It allows people to look at their files in credit reference agencies and to correct records that are incorrect or to add statements to explain records that are correct but misleading.

The 1975 Sex Discrimination Act, the 1976 Race Relations Act and the 1995 Disability Discrimination Act prohibit discrimination in the offer and provision of services, which include banking and credit, on the grounds of gender, race and disability. Legislators have decided that in these cases fairness forbids firms taking data relating to them into account. However, since other discriminators, such as income or employment status, which may disproportionately affect women, people from minority ethnic groups or disabled people, are not outlawed, it is hard to see this as anything other than a statement of principle.

Self-regulation

There is a scheme of self-regulation that applies to credit scoring and that is enforced (at least in theory) by the licensing system to which lenders are subject under the 1974 Consumer Credit Act. The rules are contained in the Guide to Credit Scoring. The latest version is dated 2000 and was developed by the industry in conjunction with the Office of the Information Commissioner and the OFT14. It is not possible here to go into a detailed description of the rules. But, for example, a rule that might protect those who might be unfairly discriminated against is that credit grantors will check from time to time that the predictiveness of the scoring system is comparable with expectations. There is also a rule prohibiting redlining, that is, refusing credit simply on address only, although taking postal address into consideration, properly weighted, is permitted.

(p.177) It is difficult to assess whether the guide is effective as selfregulation since there seems to be no central mechanism by which it is enforced, except where the OFT is considering licensing questions. I have seen no evidence that the two rules have given rise to any enforcement action. The British Bankers Association's Banking Code Guidance for Subscribers, March 200315, says that subscribers should comply with the Guide to Credit Scoring, but also points out that the Banking Code Standards Board, which monitors compliance, does not monitor compliance with the guide (see www.bba.org.uk/public/consumers/40883/1974).

A reflection of the provision in the guide on checking predictiveness appears in a leaflet published by the Finance and Leasing Association (FLA), a trade association in the credit industry. The leaflet is for member firms to hand out to applicants and is called ‘Your Credit Decision Explained’. In this document, under the question ‘Is credit scoring fair?’, the following statement appears: “We test our credit scoring methods regularly to make sure they continue to be fair and unbiased”. However, this is prefaced by a statement that a decision to grant credit is determined by policy and will reflect commercial experience and requirements.

This is not part of the FLA consumer code, so it cannot be tested by a challenge. But in any case, it is clear that fairness can be overridden by commercial needs.

Unfair discrimination: insurance and genetic testing

In the late 1980s, the development of genetic tests for certain diseases appeared to be leading towards a crisis in insurance. There was a concern that those seeking life insurance (and to a lesser extent critical illness and income replacement insurance) who might be at risk of a genetically transmitted disease, would be required to take a test and reveal the result to the insurer. Much concern was raised about the emergence of an uninsurable underclass as a result of those carrying certain genes being unable to get life insurance (see, for example, NCC, 2001). It was also said that they would be excluded from buying their own homes. It was felt that those carrying genetic markers for a disease, which they might never develop, could be unfairly discriminated against, because insurers, unfamiliar with the (p.178) science, would assess the risk as if everyone with the gene would develop the disease. It seemed a matter of unfair discrimination that those with genetic markers should be excluded from normal life insurance pooling. As Onora O'Neill (2002) points out, the practice of life insurers until now has been to allocate individuals to inclusive pools, 95% of applicants being offered standard terms, even though the risk between applicants must vary considerably, even allowing for different premium rates related to age.

Control of unfair discrimination: insurance

Statutory regulation

Anti-discrimination legislation applies, at least in theory, to insurance. The 1976 Race Relations Act prohibits discrimination on grounds of race and ethnic origin in insurance. There is no exception. The 1975 Sex and 1995 Disability Discrimination Acts, however, prohibit discrimination but allow it where justified on statistical or actuarial information on which it is reasonable to rely. Since gender and disability often bring with them risks related to health and life expectancy, the effect of the Acts is considerably neutralised16. The 1995 Disability Discrimination Act could apply to the genetic testing issue, if it could be shown that those with genetic markers are treated less favourably than those with a similar level of risk. I have not come across any suggestion that it has been used in this way.

Self-regulation

Following previous initiatives, in March 2003 the Association of British Insurers agreed to a moratorium until November 2006, under which, for life insurance under £500,000, and health insurance under £300,000, insurers will not require an applicant to take any genetic test available or to reveal the results of a test if taken. Above these levels, insurers may ask for the result of the test if it has been taken, but only where the test has been certified by an independent government committee (the Genetics and Insurance Committee) as properly predictive of the likelihood of getting the disease. This was to deal with the question of the lack of experience of insurers (p.179) in interpreting data from new tests. To date17 only one test has been certified, that for Huntingdon's Chorea.

At a recent meeting of interested parties18 to discuss what might happen when the moratorium comes to an end, the following points emerged:

  • life insurance is rarely demanded as a condition of getting a mortgage;

  • susceptibility to genetic disorders may show up in family history, which insurers have long asked about;

  • there are not likely to be many cases where there is a single mutation in a cell carrying the likelihood of getting a disease;

  • there may be inter-reaction between genes, which may increase or decrease the likelihood of getting some diseases, but these will not show up in a predictive single test, so an explosion of genetic tests relevant to insurance is unlikely;

  • families with a member who has a genetic disorder, which would be the reason for taking the test in the first place, are often hard pressed, through other calls on their income, to buy life insurance.

So while the approach of the ABI is to be commended, it appears that the problem may not have been as acute as it at first appears.

Genetic testing got attention because it was a new problem. But if it is thought that life insurance is so desirable (although not essential) that it merits intervention on behalf of those who are unfairly discriminated against as a result of genetic disorders, surely this should be the case for anyone unfairly discriminated against, for whatever reason?19 If some diseases are seen to require the insurance industry to show that risk pooling is fair, why should it be limited to those presently selected for this treatment? A principle of fairness in data use is needed to make sure that high-profile problems do not get attention while others are ignored.

Conclusion

This chapter has looked at three areas of risk in an expansion of data use in credit and insurance: of unfair exclusion, unfair targeting and unfair discrimination. The unfairness in this information use (p.180) does not lie in privacy considerations. Indeed if privacy considerations were taken too far, modern credit and insurance would be severely restricted since they depend heavily on data use for assessing the risks involved in the products. Too much privacy could result in less accessibility.

I have tried to show the extent to which data use is causing unfair outcomes and the variety of approaches there are to mitigating unfairness. Some involve regulatory or self-regulatory controls on the use of data, while others attempt to improve accessibility for those excluded from mainstream products, through alternative, noncommercial, provision. The fact that there is such a variety of approaches indicates that government, regulators, trade bodies and providers acknowledge a concern. It is, however, very difficult to assess how effective they are, especially as the self-regulatory schemes are not properly open to scrutiny. Some solutions only give benefit to a limited number of people, while others rely on individuals who are affected taking court or other action to enforce rights, which may be beyond their means.

More needs to be known about: whether making firms alter their behaviour under self-regulation works better than making noncommercial provision; whether prohibiting discrimination has any effect if it has to be challenged by an individual enforcing rights, rather than a regulator in a policing role; and whether in countries with central regulators for data sharing some of the problems around responsible lending are tackled. The effect on competition of compelling further data sharing needs to be explored. Privacy considerations also have a role to play.

The label ‘unfair’ in the three categories examined above reflects a wider societal view of what is just, that is:

  • people should have access to goods and services essential to everyday life;

  • without some protection20, people should not be sold goods or services that are harmful to them;

  • people should not be refused goods or services as a result of discrimination on the grounds of personal characteristics that are irrelevant to the goods or services in question.

(p.181) In the case of essential services, policy considerations on how access should be achieved cover much wider ground than merely the question of data use. The NCC (2002) has examined these and I do not propose to explore this further, except to say that data use could have a role to play where it is thought that consumer cross-subsidy is an appropriate way of funding access.

In the case of unfair targeting and unfair discrimination, however, there are principles relating to data use that would mitigate some of the problems and that are implicit in the existing self-regulatory and regulatory responses.

I want to argue here for consideration of a more coherent approach to fair data use, beyond the privacy-based approach of the 1998 Data Protection Act. The first data protection principle requires data to be fairly and lawfully processed. But fairness is interpreted in relation to processing and use as it affects individuals and privacy, rather than other outcomes. For example, in deciding fairness, “regard is to be had to the method by which they are obtained, including in particular whether any person from whom they are obtained is deceived or misled as to the purpose or purposes for which they are to be processed”(1998 Data Protection Act, sch 1).

The solutions and responses to perceived problems described above reveal an underlying concern with fairness, rather than merely privacy, whether or not they are successful in their aims. The Principles of Reciprocity state that the purpose of sharing data is to improve responsible lending, thus preventing targeting those to whom further debt might be harmful. As we have seen, the OFT Guide to Credit Scoring requires credit grantors to check from time to time that the predictiveness of the scoring system is comparable with development expectations. Discrimination on the grounds of race, gender and disability is prohibited. So is discrimination on raw postcode data. The response to the problems of those with markers for genetic diseases requires scientific proof of the relevance of those markers.

But fairness in data use is likely to have wider implications than the categories of problems set out above and certainly a wider application than in financial services. Much work is being done at present on unfair commercial practices. The EU is close to enacting a duty ‘not to trade unfairly’ in the Unfair Commercial Practices (p.182) Directive21. The Financial Services Authority requires the firms it regulates to pay due regard to the interests of their customers and to treat them fairly22. This thinking should be applied explicitly to data use, not just by commercial users but by public bodies as well.

Notes

(1) In some cases, regulatory pressure is adding to this. The effect of the Basel II Capital Accord on regulatory requirements for all international banks is likely to increase the risk-based pricing of credit, as banks will have to show regulators that their capital is adequate for the risks of their credit portfolios.

(2) The Registry Trust is a not-for-profit organisation, which maintains a register of county court judgements for England and Wales, including administration orders, for the Lord Chancellor. It also keeps registers for Scotland and Northern Ireland. The register is public. It is governed by a Board drawn from the consumer credit trade sector, with a consumer representative.

(3) These requirements have been developed by the Information Commissioner as a result of the 1998 Data Protection Act. The Information Commissioner works with the industry to make the consent form transparent, so that applicants know what information will be shared.

(4) Consumers taking out insurance may have to consent to insurance companies sharing data on claims in order to prevent fraud.

(5) Recent research commissioned by the DTI found that low-income households have an irreducible need for credit and households with constrained resources the greatest need for credit (Policis, 2004). This might be seen as an argument for more generous benefits or a higher minimum wage, but this point is not argued here.

(6) A super-complaint (defined by Section 11 of the 2002 Enterprise Act) is a complaint to the OFT that features of a market for goods or services harm the interests of consumers.

(p.183) (7) Interestingly, in the US, credit checks are sometimes made on those applying for insurance. In the UK it is, according to the rules of the Principles of Reciprocity described, not possible to check with closed user groups in credit reference agencies for the purposes of deciding on granting insurance, although the industry-wide database CIFAS is open to insurers.

(8) Since writing this chapter, a news story has featured Vanquis, a subsidiary company of Provident Financial, which offers credit cards to those refused credit, at rates of interest between 50 and 70%. It was originally stated that the company obtained data on customers to target from Experian, but a spokesperson for Experian said that they were not informed when credit was granted and so had no list of customers refused credit. It would be interesting to know whether, without breaching the ‘responsible lending’ principle of SCOR, it would be possible to obtain a list of those with poor credit histories (rather than those refused credit) from a credit reference agency.

(9) Cattles has now announced in its annual report for the year ending December 2003 that it has ended the pilot as not enough new customers have been referred.

(10) In March 2004 the OFT announced the result of the study. Better financial awareness among consumers, and clear, accurate and relevant information from credit providers are required to make the use of debt consolidation fairer and more transparent. The study will inform the OFT's enforcement of the 1974 Consumer Credit Act.

(11) The Consumer Credit Bill, which resulted from the White Paper, is currently (March 2005) before Parliament. There are no proposals for controlling data collection or use, but there is provision for a general control by the court of unfair relationships between lenders and borrowers. A relationship may be unfair because of things done or not done by the lender before or after the making of an agreement. Some control of unfair targeting might be achieved, if the DTI codified the triggers for overindebtedness, for use in court challenges using this clause.

(p.184) (12) The principles are, however, subject to scrutiny by the OFT for potential breaches of competition law and the most recent version has been signed off by the OFT. The rules provide for new entrants accessing data for a period before they are in a position to contribute their own, which allows competitors to enter the market.

(13) Under the 1998 Data Protection Act, a borrower could demand a copy of his or her file with this kind of lender, to show to an alternative loan company, which would not otherwise be able to get access to it.

(14) A copy of the Guide to Credit Scoring can be found on the Experian website at www.experian.co.uk/corporate/compliance/creditscoring. It is referred to frequently on other websites of the trade organisations which agreed it, but a copy cannot always be accessed on these websites. At the time of originally writing this chapter, it was available on the OFT website, but this is no longer the case (March 2005). The difficulty in obtaining a copy gives some indication as to how inadequate the Guide is as a self-regulatory scheme.

(15) The Guidance is available on the Banking Code Standards Board website at www.bankingcode.org.uk/home.htm

(16) Although the original text proposed that discrimination in insurance on the grounds of gender should be prohibited, the final text of the EU Gender Directive (http://europa.eu.int/scadplus/leg/en/cha/c10935.htm), adopted in December 2004, provides that gender may be used as a basis to assess risk in insurance, if objective data can justify the difference.

(17) No other tests have been approved, but on 28 March 2005 the government announced that the agreement with the ABI will continue until 2011.

(18) ‘Insurance, genetics and fairness’, joint reading of the Genetics and Insurance Committee and the Human Genetics Commission, 22 September 2003.

(p.185) (19) A similar response from the ABI to discrimination on the grounds of tests and other data that might reveal HIV status is beyond the scope of this chapter. A change in the original position taken by the ABI was published for consultation in September 2003. The final draft version can be seen at www.abi.org.uk/Display/File/86/response_to_responses_FINAL.pdf

(20) For example by warnings or information that enables them to take precautionary measures. In some cases there is a complete prohibition on harmful/unsafe goods.

(21) Proposal for Unfair Commercial Practices Directive, Com (2003) 3J6 final.

(22) Principles for businesses, Financial Services Authority. Available at http://fsahandbook.info/FSA/handbook.jsp?doc=/handbook/PRIN/2/1

References

Bibliography references:

6, P. and Jupp, B.(2001) Divided by information?: The ‘digital divide’ and the implications of the new meritocracy, London: Demos.

ABI (Association of British Insurers) (2003) Statement of principles on the provision of flooding insurance, London: ABI.

ABI (2004) A Changing climate for insurance, London: ABI.

Bridges, S. and Disney, R. (2001) Modelling consumer credit and default: The research agenda, Nottingham: Experian Centre for Economic Modelling, University of Nottingham.

Datamonitor (2003) ‘UK non-standard and sub-prime lending’, London: Datamonitor DMFS 1506.

DTI (Department of Trade and Industry) (2001) Report on the Task Force on Tackling Overindebtedness, London: DTI.

DTI (2003) Fair, clear and competitive: The consumer credit market in the 21st century, Cm 6040.

Jappelli, T. and Pagano, M. (2000) Information sharing in credit markets: The European experience, Salerno: Centre for Studies in Economics and Finance (CSEF), University of Salerno.

(p.186) Kempson, E. and Whyley, C. (1999) Extortionate credit in the UK, London: DTI.

Kempson, E., Collard, S. and Taylor, S. (2002) Social Fund use among older people, London: DWP.

Kempson, E., Collard, S. and Whyley, C. (2000) Saving and borrowing, London: DSS.

NCC (National Consumer Council) (2001) Whose hands on your genes?, London: NCC.

NCC (2002) Meeting basic needs, London: NCC.

NCC (2004) Home credit, London: NCC.

OFT (Office of Fair Trading (1997) Non-status lending: Guidelines for lenders and brokers, London: OFT.

O'Neill, O. (2002) Autonomy and trust in bioethics, Cambridge: Cambridge University Press.

Policis (2004) The effects of interest rate controls in other countries, London: DTI.

Whyley, C., McCormick, J. and Kempson, E. (1998) Paying for peace of mind, London: Policy Studies Institute.

Notes:

(1) In some cases, regulatory pressure is adding to this. The effect of the Basel II Capital Accord on regulatory requirements for all international banks is likely to increase the risk-based pricing of credit, as banks will have to show regulators that their capital is adequate for the risks of their credit portfolios.

(2) The Registry Trust is a not-for-profit organisation, which maintains a register of county court judgements for England and Wales, including administration orders, for the Lord Chancellor. It also keeps registers for Scotland and Northern Ireland. The register is public. It is governed by a Board drawn from the consumer credit trade sector, with a consumer representative.

(3) These requirements have been developed by the Information Commissioner as a result of the 1998 Data Protection Act. The Information Commissioner works with the industry to make the consent form transparent, so that applicants know what information will be shared.

(4) Consumers taking out insurance may have to consent to insurance companies sharing data on claims in order to prevent fraud.

(5) Recent research commissioned by the DTI found that low-income households have an irreducible need for credit and households with constrained resources the greatest need for credit (Policis, 2004). This might be seen as an argument for more generous benefits or a higher minimum wage, but this point is not argued here.

(6) A super-complaint (defined by Section 11 of the 2002 Enterprise Act) is a complaint to the OFT that features of a market for goods or services harm the interests of consumers.

(p.183) (7) Interestingly, in the US, credit checks are sometimes made on those applying for insurance. In the UK it is, according to the rules of the Principles of Reciprocity described, not possible to check with closed user groups in credit reference agencies for the purposes of deciding on granting insurance, although the industry-wide database CIFAS is open to insurers.

(8) Since writing this chapter, a news story has featured Vanquis, a subsidiary company of Provident Financial, which offers credit cards to those refused credit, at rates of interest between 50 and 70%. It was originally stated that the company obtained data on customers to target from Experian, but a spokesperson for Experian said that they were not informed when credit was granted and so had no list of customers refused credit. It would be interesting to know whether, without breaching the ‘responsible lending’ principle of SCOR, it would be possible to obtain a list of those with poor credit histories (rather than those refused credit) from a credit reference agency.

(9) Cattles has now announced in its annual report for the year ending December 2003 that it has ended the pilot as not enough new customers have been referred.

(10) In March 2004 the OFT announced the result of the study. Better financial awareness among consumers, and clear, accurate and relevant information from credit providers are required to make the use of debt consolidation fairer and more transparent. The study will inform the OFT's enforcement of the 1974 Consumer Credit Act.

(11) The Consumer Credit Bill, which resulted from the White Paper, is currently (March 2005) before Parliament. There are no proposals for controlling data collection or use, but there is provision for a general control by the court of unfair relationships between lenders and borrowers. A relationship may be unfair because of things done or not done by the lender before or after the making of an agreement. Some control of unfair targeting might be achieved, if the DTI codified the triggers for overindebtedness, for use in court challenges using this clause.

(p.184) (12) The principles are, however, subject to scrutiny by the OFT for potential breaches of competition law and the most recent version has been signed off by the OFT. The rules provide for new entrants accessing data for a period before they are in a position to contribute their own, which allows competitors to enter the market.

(13) Under the 1998 Data Protection Act, a borrower could demand a copy of his or her file with this kind of lender, to show to an alternative loan company, which would not otherwise be able to get access to it.

(14) A copy of the Guide to Credit Scoring can be found on the Experian website at www.experian.co.uk/corporate/compliance/creditscoring. It is referred to frequently on other websites of the trade organisations which agreed it, but a copy cannot always be accessed on these websites. At the time of originally writing this chapter, it was available on the OFT website, but this is no longer the case (March 2005). The difficulty in obtaining a copy gives some indication as to how inadequate the Guide is as a self-regulatory scheme.

(15) The Guidance is available on the Banking Code Standards Board website at www.bankingcode.org.uk/home.htm

(16) Although the original text proposed that discrimination in insurance on the grounds of gender should be prohibited, the final text of the EU Gender Directive (http://europa.eu.int/scadplus/leg/en/cha/c10935.htm), adopted in December 2004, provides that gender may be used as a basis to assess risk in insurance, if objective data can justify the difference.

(17) No other tests have been approved, but on 28 March 2005 the government announced that the agreement with the ABI will continue until 2011.

(18) ‘Insurance, genetics and fairness’, joint reading of the Genetics and Insurance Committee and the Human Genetics Commission, 22 September 2003.

(p.185) (19) A similar response from the ABI to discrimination on the grounds of tests and other data that might reveal HIV status is beyond the scope of this chapter. A change in the original position taken by the ABI was published for consultation in September 2003. The final draft version can be seen at www.abi.org.uk/Display/File/86/response_to_responses_FINAL.pdf

(20) For example by warnings or information that enables them to take precautionary measures. In some cases there is a complete prohibition on harmful/unsafe goods.

(21) Proposal for Unfair Commercial Practices Directive, Com (2003) 3J6 final.

(22) Principles for businesses, Financial Services Authority. Available at http://fsahandbook.info/FSA/handbook.jsp?doc=/handbook/PRIN/2/1