Problem credit card debt

Back to top

Problem credit card debt

We are concerned about the scale of potentially problematic debt in this market. We have used a series of indicators to assess the scale  of potential problems:

  • We estimate that 6.9% of cardholders (2.1 million) are in arrears or have been charged-off. The financial and non-financial implications in these cases are likely to be significant. We found that firms take steps to avoid lending to consumers who cannot ultimately repay and cardholders who default are not profitable. We found firms were generally proactive in contacting consumers when they began to miss payments.
  • We estimate that a further 6.6% of cardholders (2.1 million) have persistent high levels of credit card debt which they may be struggling to repay. This cycle may begin with relatively minor incidents, but the cumulative welfare implications that follow may be large. These consumers are profitable for firms and there was little evidence to suggest firms intervene to help consumers address persistent debt burdens where they do not miss payments.
  • We estimate that a further 5.2% of cardholders (1.6 million) make systematic minimum repayments. We recognise that this is a weaker indicator of problem credit card debt and that this group includes consumers who are not struggling. Nonetheless, it indicates that consumers are taking longer to repay their credit card debt (with associated costs) than perhaps they need to. Again, these consumers are profitable for firms.

We recognise that some bad debt is a feature of all credit activity − borrowing is never risk-free, as the ability to repay is affected by major negative life events (such as divorce, redundancy or long term illness).

However, there are known patterns of consumer behaviour (‘behavioural biases’) that will tend to lead to over-borrowing and under-repayment. These include optimism bias (consumers typically over-estimate their ability to repay), framing effects (consumers perceive costs expressed in % terms as being smaller than the same costs expressed in £), and the anchoring effect of minimum repayment amounts (consumers will tend to pay what their lender suggests).

Balance transfers do not currently appear to be materially contributing to problem credit card debt. However, if wider economic conditions were to change significantly then, as with any credit product, the proportion of consumers unable to pay is likely to increase.

This chapter outlines our interim findings on problem credit card debt, in particular whether some consumers are over-borrowing/under-repaying on their balances, and whether firms have incentives to provide unaffordable credit that results in consumer harm. The emerging findings are based on our review of firm data and strategy documents submitted by firms, discussions with firms, and our account-level data analysis, as well as the literature reviews on affordability, consumer behaviour and behavioural biases.

Why we are concerned about problem credit card debt

Like other credit products, credit cards can provide a valuable service to consumers by enabling them to substitute future consumption for current consumption, smooth consumption over time, and provide flexibility to manage income and expenditure shocks.

A major difference between credit cards and many other credit products is that both the amount borrowed (subject to credit limits) and repayment schedule are flexible. Subject to meeting the minimum repayment, the consumer can decide how much to repay each month. This allows consumers to opt for very low repayments over a very long period, which may be necessary to tide them over in the short term. Over a longer period, however, lower monthly payments imply a longer time to repay and, in turn, a higher total cost. This  can have implications for their wider financial situation, including their ability to meet other credit or non-credit commitments and basic household expenditure. 

As set out in our terms of reference72, there may be instances where consumers are taking on such levels of credit card debt that they are unable to repay the full credit card balance within a reasonable time-frame (over-borrowing). Or, it may be the case that some consumers are repaying their debt slower than they could, incurring avoidable costs (under-repayment).

Consequences of problem credit card debt

We are concerned that certain consumer segments may be worse off as a result of their usage of one or more credit cards, a situation we refer to here as ‘problem credit card debt’. This harm may be in financial or non-financial terms.

Financial harm includes paying high interest rates or fees. Later in this chapter we estimate the total cost consumers pay in debt servicing per account relative to the amount borrowed as an indication of the direct financial impact of credit card debt.

The harm from credit card debt can also be non-financial, such as forgoing essential expenditure or experiencing personal distress.

A survey by Step Change73, a debt charity, found that problem debt can have a negative impact on consumers’ physical and mental health as well as resulting in their relationship with family and/or friends getting worse.

Another report74 showed that debt worries have an impact on people’s ability to work, by affecting their attendance or concentration. Consumers might also lose access to cars, telephones or the internet, and may find it more difficult to work or seek employment.75

Later in this chapter we also estimate the length of time it would take consumers to pay off all their credit card debt on these accounts. This gives some indication of the non-financial impacts (e.g. because consumers experience the debt for longer).

In launching this market study we considered that two factors may contribute to these outcomes:

  • Over-borrowing consumers may be over-confident in their ability to repay or not judge future outcomes appropriately when using their credit card and suffer adverse consequences at a later stage. These outcomes constitute consumer harm, and could potentially have been avoided.
  • Over-lending firms may not face a sufficient incentive to avoid or limit lending to certain consumers who experience harm as a result of the lending.

Credit risk and affordability definitions

In their responses, firms typically distinguished between ‘creditworthiness’ and ‘affordability’. However, as both are elements of ‘creditworthiness’ as defined in our rules, we use the alternative terms ‘credit risk’ and ‘affordability’ instead. Credit risk refers to the likelihood an individual will default and affordability to the ability of an individual to meet future repayments without an adverse impact on their financial situation and being able to make repayments as they fall due within a reasonable period. We refer to credit card debt that is or has become unaffordable as problem credit card debt for convenience.

Causes of unaffordable borrowing by consumers

There are two main potential causes of unaffordable borrowing, namely:

  • life-events, such as divorce or redundancy
  • consumer decision-making

Life-events

Major life-events, such as divorce or redundancy, can lead to:

  • a need for additional borrowing, or
  • what was originally affordable borrowing quickly becoming unaffordable

Some bad debt is a feature of all credit activity – borrowing is never risk-free, as the ability to repay can be affected by such life events. Both borrowers and lenders take a degree of risk entering into any kind of credit agreement, and especially where this involves open-end running-account credit such as a credit card.

While credit cards’ flexibility on repayments can be helpful in managing short-term implications of life-events, they can also mean that affordability issues may take longer to emerge or be addressed.

Step Change reports that unemployment continues to be the reason given most often by their clients for seeking debt advice, with just under a quarter citing it as the primary cause of their problem debt.76 45% of Christians Against Poverty’s clients reported the primary reason for their debt as relationship breakdown, unemployment or long-term illness.77

Consumer decision-making and behavioural biases

Consumer harm may arise from consumers making poor borrowing or repayment decisions, potentially caused by a lack of information and understanding or deep-rooted behavioural issues (for example, being overly optimistic). This harm might take the form of higher costs, sacrificed consumption, the stress of meeting repayments, or the consequences of default.78

If these higher costs are incurred because of behavioural biases, such as over-optimism, then this could lead to consumers taking longer to repay their debt than would be optimal. This type of consumer harm is less likely with other debt products where repayments are less flexible (e.g. fixed at the point of borrowing).

UK Cards’ recent Credit Card market report79 found evidence of behavioural biases, including anchoring to minimum payments and being overly optimistic about the future availability of funds.

The academic literature also points to a range of consumer behavioural biases in relation to credit cards.80 This includes:

  • Present bias (when a decision is unduly influenced by the present at the expense of the future and can lead to regretful purchases). Evidence of this includes consumers making initial borrowing decisions that are inconsistent with their subsequent borrowing behaviour and consumers exhibiting higher levels of present bias borrowing larger amounts.
  • Optimism bias (where consumers may be overconfident when assessing the likelihood of future events). Consumers with more moderate levels of optimism have been shown to be more likely to pay off their credit card balances while extreme optimists have been shown to have preferences for credit card features that are inconsistent with their subsequent borrowing behaviour.
  • Presentation and framing of information about repayment. One particular area of focus has been on how repayment behaviour may be influenced by the presentation and framing of information about repayment, particularly the minimum repayment level. One study showed that most repayments are either in full or the minimum. A group of studies have then shown that removing minimum payment information or accompanying it with additional information (e.g. regarding alternative courses of action) has increased levels of repayment.
    • Our consumer survey also found that 29% of respondents claimed to spend more than they had budgeted for on their main credit card and that around half of consumers with outstanding balances were very or slightly concerned about it.

Drivers of lending by firms that may be unaffordable for consumers

Firms' lending decisions and criteria are key to mitigating the risk of unaffordable borrowing. These decisions include which consumers to lend to, how much to lend, and how to manage open accounts, including when consumers get into difficulty.

From the perspective of the firm, the incentives to lend will be driven in large part by profitability. This means that:

  • Incentives of the firm may well be aligned with the consumer, as both have an incentive to avoid default (‘credit risk’).
  • There may be instances when incentives may be less well aligned, as the firm is concerned to get its money back, with interest, but may be less concerned with whether the consumer can do so only by suffering an adverse impact on their financial situation for example by not being able to meet other reasonable financial commitments (‘affordability’).

Creditworthiness and forbearance

In this section we consider how firms are making key decisions in the following areas:

  • firms’ credit risk assessments
  • firms’ affordability assessments
  • firms’ forbearance policies

The first two of these issues relate to lending decisions. FCA rules81 require an assessment of creditworthiness before a credit agreement is entered into and before any significant increase in the amount of credit or the credit limit. This must include an assessment of the potential for the commitments under the agreement to adversely impact the customer’s financial situation, and the customer’s ability to make repayments as they fall due or (in the case of an open-end agreement) within a reasonable period. This is usually referred to as ‘affordability’ (in contrast to an assessment of credit risk, i.e. the probability of default).

The third issue relates to how firms’ treat customers that are failing to meet repayments. FCA rules82 require firms to treat customers in default or arrears difficulties with forbearance and due consideration. They also83 require firms to monitor repayments and take appropriate action where there are signs of actual or possible repayment difficulties.

Our review of these issues draws primarily on firms’ submissions.

Firms’ credit risk assessments

Firms assess credit risk using empirical, data-driven models that predict the probability of default. The assessment process is at the core of firms’ credit card business and they invest in it heavily.

For the assessment, firms tend to use a combination of data which can include data from credit reference agencies, application form data, such as employment status and income, and, for existing customers84 or once an account becomes active, transaction data from firms’ own records.

Each firm has its own scorecard and policy rules. A scorecard is a set of weights that are attached to the pieces of information held on each consumer and form part of the calculation of the consumer’s credit score. Higher weights are associated with information that is a stronger predictor of default. Each firm designs and reviews its own scorecards to try and improve the predictive power of the model. Policy rules in contrast are ‘hard stops’ or ‘cut offs’ which may lead applications to be rejected irrespective of credit score or before a scorecard is applied.

Once an account is active, firms routinely update their credit score estimates for each consumer and review how consumers have managed their accounts. The frequency of these updates varies between firms but is typically between monthly and quarterly. Updated credit scores are monitored and feed into decisions over changes to consumers’ credit limits (which can be increased or decreased as a result) and potentially changes in interest rates. These updates can also affect the promotional offerings and, in extreme cases, lead to the suspension of credit facilities.

For all credit card firms these processes are largely automated with very little role for manual processes. Even in cases where firms told us that there was a role for human intervention this was said to very rarely lead to a change in outcome.

Firms’ affordability assessments

Firms are required, in assessing creditworthiness, to consider affordability, that is, making an assessment of the potential for commitments to adversely impact a customer’s financial situation and the customer’s ability to make repayments as they fall due within a reasonable period.

Depending on the particular firm, affordability assessments might influence a credit card application in different ways. In some cases it is part of an integrated assessment alongside credit risk, while in others it is a separate assessment, typically applied only if the applicant first passes a credit risk assessment. In either case it can directly influence on the credit limit that is granted.

The practical steps involved in an affordability assessment are broadly similar across firms, and typically involve a comparison between a consumer’s income, expenditure and prospective debt repayments. This may use actual information from the consumer, or (in the case of expenditure) proxy information such as data from the Office of National Statistics (ONS) on average household expenditure. Some firms also rely quite heavily on affordability or indebtedness indices provided by credit reference agencies.

Income data plays a key role in affordability assessments yet it is rarely verified. In most cases it is information that is self-reported by the applicant or obtained from credit reference agency data. Income data from the latter is often based on historic credit applications (self-reported) or estimated.85 We also noted that the definition of ‘income’ used by firms can vary significantly.

Firms do not typically conduct regular affordability assessments once an account is active. The firms that do undertake some form of later review request updated income data, though again this is usually not verified.

We are currently wider on our expectations as regards creditworthiness across consumer credit markets and we will take forward our thinking through that process.86

Firms’ forbearance policies

Forbearance can be short-term or long-term and come in the form of, among other things, freezing interest and/or charges or offering a different payment arrangement.

Each firm has its own forbearance policy, with one or more of the following options available to its customers:

  • A period without any collections activity (‘breathing space’) that usually lasts for 30 to 60 days. This does not necessarily involve suspending or reducing interest and charges, thus interest may continue to accrue during this period.
  • A reduction or suspension of interest and charges during the forbearance period.
  • Refinancing, when monthly payments or the rate of interest are reduced on a permanent basis. Some firms also offer temporary payment arrangements that allow for reduced payments for a short period.
  • Re-aging of an account. This involves writing off arrears following a number (usually three) of consecutive missed payments by the consumer.
  • Part-settlements may occasionally be negotiated, where less than 100% of the outstanding debt is accepted by the firm.
  • Write-off of debts in response to major unsettling life events of their consumers, such as terminal illness.

Debt is normally charged-off after an account has been in arrears for a certain period of time. When this occurs varies by firm, but several firms charge off debt after between 120 to 180 days of arrears. This may be followed by internal recoveries, placing the debt with debt collection agencies, or the sale to debt purchasers.

Overall, it appears that firms are generally proactive in contacting consumers to initiate these forbearance procedures when consumers begin to miss payments. However, there is limited evidence of any significant efforts by firms to initiate these types of procedures for consumers currently meeting repayments but exhibiting other signs of potentially problematic debt (e.g. maintaining a high credit limit utilisation). We believe there is more that could be done to intervene earlier with consumers exhibiting signs of potentially problematic debt.

At the end of this chapter we turn to the issue of firm incentives. Firm decisions are likely to be made largely on the basis of profit incentives and we have therefore analysed the relationship between profitability and our indicators of potential problem credit card debt.

Scale and nature of problem credit card debt

In this section we set out a series of indicators that we have used to assess the scale and nature of potentially problematic credit card debt.

Context

Identifying problem credit card debt can be challenging. For example, levels of debt that appear affordable one period may become unaffordable the next and, similarly, what is affordable for one person may not be for another.

Past work has emphasised these challenges and there is no general consensus on what constitutes problem debt. Most studies rely on one or more indicators based on subjective thresholds.

In the context of credit cards, this can be particularly difficult. The flexible drawdown and repayment options of these products make it difficult to distinguish between consumers actively choosing to use these features to their advantage and those struggling with problem credit card debt. This distinction is clearly important as the welfare implications are opposed for these groups, with the flexibility offered by credit cards being a key feature that is likely to be valued by many consumers.

Our approach to these issues has been as follows. We first identified our high-level concerns regarding affordability for credit card products. Using these as a guide, we then specified a number of quantitative indicators that map to these concerns.

At a conceptual level, we identified three main areas of concern:

  • The first and most clear concern is when consumers default or miss payments. The financial and non-financial implications in these cases are likely to be significant.
  • The second concern is persistent and long-term debt. Consumers in these cases may be able to meet repayments but have reached a level of debt that they are unlikely to be able to recover from, even over a sustained period of time. This cycle may begin with relatively minor incidents, but the cumulative welfare implications that follow may be large.
  • The third concern is when consumers are making minimum repayments while incurring interest charges. The low minimum repayment requirement of credit cards means that these consumers may not be struggling to meet the repayments but, over time, may be incurring high interest costs as a result of their repayment profile.

The indicators we chose cover a range of measures that map to these concerns. These include these three areas of concern as well as the cost that consumers pay for credit and the length of time it will take them to repay their debt. Each of these measures is described in more detail below.

To implement each indicator it was necessary to specify a number of thresholds. For presentational purposes, we focus on certain baseline thresholds to present our results. However, we also tested a number of alternative thresholds and definitions, which are presented as sensitivity checks. The choice of thresholds necessarily involves subjective judgement and, for this reason, the indicators are best interpreted collectively.

There are also a number of limitations to each indicator.

  • In some cases they will capture some consumers that do not have problem credit card debt issues (for example, some consumers making minimum repayments will be doing so only for a temporary period before resuming higher repayments).
  • In other cases, individual indicators will neglect aspects of unaffordable credit card debt (for example some consumers making above minimum repayments are doing so with difficulty).

The tension between these two possibilities motivated our approach of using several indicators and again highlights the importance of interpreting the results together.

More generally, it is important to recognise that some degree of problem debt will always be present in this market. With any risk-based product, by definition there must be some losses. It is unfortunate but inevitable that some consumers will, for example, fall into financial difficulty because of major adverse life events (e.g. job losses, divorce). Many of the outcomes captured by our indicators may be due to these unavoidable causes. However, at least some of the outcomes will be due to other factors, for example because some consumers may make uninformed or poor borrowing decisions and firms find it profitable to lend to these consumers.

indicators are grouped into three areas as follows:

  • indicators of potential problem credit card debt
  • balance transfers and future problem credit card debt
  • cost of credit and repayment term

We discuss each set of indicators in turn.

Potentially problem credit card debt

We use quantitative indicators to provide an indication of the likely scale and nature of problem credit card debt.

We have used four indicators, each based on 12 months of data to December 2014. The indicators are defined as follows:

  1. Severe arrears: consumers that have been charged-off87 or have been at least six months in arrears.88
  2. Serious arrears: consumers who having missed three or more repayments, and are either in or have been in arrears.89
  3. Persistent debt: consumers that have an average credit limit utilisation of 90% or more while also incurring interest charges.90
  4. Systematic minimum repayments: consumers that have made nine or more minimum repayments, while also incurring interest charges.

We chose these indicators based on a review of the academic literature, existing research and consumer surveys, and our own analysis and understanding of the market.

As explained above, there is necessarily an element of subjectivity in defining the indicators and their thresholds. In recognition of this we have conducted a number of sensitivity checks to assess the impact of our choices on the results, for example by considering alternative thresholds (e.g. 75% rather than 90% for persistent debt, or six months instead of nine for systematic minimum repayment) or alternative definitions (e.g. using an absolute level of debt rather than credit limit utilisation for persistent debt). We describe these sensitivity checks in more detail in Annex 6.

For each consumer, we checked which indicators applied to the accounts they held. In some cases, more than one indicator applied to a single account or to multiple accounts held by the same consumer. We assigned each consumer a single indicator by selecting the ‘worst’ arrears status across all of their accounts, with indicator 1 being the most severe and indicator 4 being the least severe.91

We estimate (see Figure 18 below) that most credit card consumers do not appear to have problem credit card debt. The results of the baseline indicators show that:

  • Approximately 1.9% of consumers (600,000 consumers) were in severe arrears. These consumers were either charged-off or were at least six months in arrears. In addition, approximately 4.9% of consumers (1.5 million consumers) were in serious arrears. These consumers missed three or more repayments and were in arrears at some point. These are more prevalent amongst consumers with higher credit risk and in the more deprived demographic segments.
  • Approximately 6.6% of consumers (2.1 million consumers) were in persistent debt. These consumers were maintaining a credit limit utilisation of 90% or more over one year while incurring interest.
  • Approximately 5.2% of consumers (1.6 million consumers) made systematic minimum repayments. These consumers are repeatedly making minimum payments while incurring interest.
  • For both persistent debt and systematic minimum repayments, the repayment behaviour appears across all credit risk groups and demographic segments. This, in part, suggests that a number of low risk individuals may be struggling with credit card debt or not paying down their debt for other reasons. For example, they may be unaware that they are making low repayments or that this leads to higher costs. It may also reflect that our indicators are capturing some consumers that may not be struggling.

We conducted sensitivity checks on the thresholds used in the definitions of persistent debt and systematic minimum repayments. This increased the proportion of consumers identified by each indicator, to 12.0% for persistent debt and 6.2% for systematic minimum repayments.

Figure 18: Indicators of potential problem credit card debt

Figure 18: Indicators of potential problem credit card debt

Source: FCA analysis of account-level data

Balance transfers and future problem credit card debt

We have also considered whether balance transfer products may be masking future unaffordable credit card debt, in that consumers who are currently able to shift balances around without incurring interest for long periods may in future be unable to do so. These problems might only be realised once consumers’ existing promotional balance transfer period ends and they are unable to transfer their outstanding balance to a new 0% deal.

We estimate that 22% of consumers who acquired a new card in 2014 chose a balance transfer card, and such transfers account for about £14 billion (out of £60 billion) of outstanding balances. Most of those taking out balance transfer cards were low risk consumers. Around one in five of those who acquired a new card in 2014 and chose a balance transfer card had previously taken out a balance transfer between 2012 and 2013, that is, appeared to be shifting balances repeatedly.

The repeated use of balance transfers was most prevalent amongst lower risk consumers with only around 2% of those cards taken up by higher risk consumers, and the evidence suggests that consumers struggling to pay any of their debts are unlikely to be offered repeated balance transfer deals. Balance transfers do not currently appear to be materially contributing to problem credit card debt. However, if wider economic conditions were to change significantly then, as with any credit product, the proportion of consumers unable to pay is likely to increase.

Cost of credit and repayment term

To provide further insight into levels of problem credit card debt, and to assess some of the consequences of problem debt, we computed estimates of the cost of credit and repayment term. The former provides an indication of the direct financial impact of credit card debt while the latter may be related to the non-financial impacts (e.g. because consumers experience the debt for longer).

We have defined the two indicators as follows:

  • Estimated cost of credit: the total interest and charges on an account divided by the total value of transactions. The rationale behind this measure is that it provides a simple estimate of the actual costs incurred by the account holder as a proportion of each £1 of transactions.92 We calculated this measure for all accounts opened after January 2010 with total transactions above £50.
  • Estimated repayment term: the length of time taken for a consumer to repay their outstanding credit card debts as of January 2015 − i.e. the last observation in our sample − with a rate of repayment equal to their average over the last six months.93 It was assumed that no future transactions are made and all future interest charges are applied to the outstanding balance at the current purchase rate.94

While the two measures are clearly linked − lower repayments will lead to a longer repayment term and more interest charges − the cost of credit measure also takes into account any charges that the consumer may have incurred. The cost measure also takes into account actual payments while the repayment term measure is an estimate based on recent behaviour.

We examine these indicators for accounts identified by the previous four indicators and the remainder of accounts. The results for consumers not identified as potentially having problem debt provides not only a useful comparison, but also an alternative view of whether some of these consumers may be struggling despite not being picked up by our earlier four indicators.

We found that consumers identified by our potential problem credit card debt indicators pay a significantly higher cost of credit and will take significantly longer to repay their debt than those without problem credit card debt:

  • We estimate that 1% of credit cards opened after January 2010 (360,000 accounts) paid debt service costs over five years exceeding the amount borrowed. Those in arrears or with persistent debt incurred the highest cost.
  • We estimate that 8.9% of credit cards active in January 2015 (5.1 million accounts) will – on current repayment patterns and assuming no further borrowing – take more than 10 years to pay off their balance. Again those in arrears or with persistent debt take the longest to repay.

Figure 19 shows that for those consumers not identified by our problem credit card debt indicators, around three-quarters pay an estimated cost of credit that is below 5% and around 94% paid an estimated cost of 20% or less.95 For those in arrears or with persistent debt, over 15% of consumers paid an effective cost of 50% or more. A non-negligible proportion of consumers in each category also paid an effective cost of 100% or more—10% of those in severe arrears, 7% of those in serious arrears, 4% of those in persistent debt and 3% of those making systematic minimum repayments.

Figure 19: Distribution of the estimated cost of credit over five years, by problem credit card debt indicator

Figure 19: Distribution of the estimated cost of credit over five years, by problem credit card debt indicator

Source: FCA analysis of account-level data

Figure 20 shows that around 86% of accounts not identified by our problem credit card debt indicators are estimated to repay their credit card debt within five years. The comparable proportion for accounts identified as being in serious arrears, persistent debt or making minimum repayments is much lower at 55% to 65%.

Figure 20: Distribution of the estimated repayment term, by problem credit card debt indicator96

Figure 20: Distribution of the estimated repayment term, by problem credit card debt indicator

Source: FCA analysis of account-level data

The sensitivity tests show that these results are robust to minor changes in the methodology for calculating effective cost and repayment term.

We believe that the picture our three sets of indicators show comprises a mixture of:

  • People struggling under a debt burden that is or has become problematic.
  • People paying more in debt service cost and taking longer to pay off debt than they need to. While these people may not regard their debt as problematic, it is more expensive than it needs to be and there is some risk that it becomes problematic in the future.

Firms’ lending incentives

As mentioned earlier in the chapter, firms’ decisions are likely to be made primarily on the basis of profit incentives and therefore in order to determine firms’ incentives to lend we need to look at profitability.

In Chapter 5 we showed that firms’ financial forecasts focus on the average profitability of different cohorts. This is a useful model for understanding products’ competitive dynamics but to understand firms’ profitability incentives requires an analysis of individual account-level profitability.

The focus of our analysis has been on where firms’ and consumers’ interests may not fully align. In some instances, firms’ and consumers’ incentives align, for example, firms generally wish to minimise the likelihood and cost of default, as do consumers. In other instances there may be a misalignment of incentives. We identified three areas to investigate:

  • Distribution of profitability: The first area relates to the distribution of profitability, and whether this is heavily influenced by a small number of very profitable consumers. To explore this issue we calculated the profitability of each account in a cohort of new accounts and examined their contribution to overall profits made by each account. Our concern was that if profits were heavily skewed to a small number of accounts, firms could be incentivised to target, and possibly exploit, these consumers.
  • Expected profitability: The second area we investigated is the profitability of consumer risk segments. In principle, firm and consumer incentives should be aligned from a risk and default perspective, simply because both firms and consumers wish to avoid this outcome. However, since firms cannot distinguish at the point of lending between consumers that will default and those that will not, firms may find it profitable to lend to some risk groups that contain a very high proportion of consumers that will go on to default. This might happen if the positive profit contributions of some consumers (e.g. those that do not default, or those that default but are highly profitable prior to default) are sufficient to outweigh the negative contributions of default. To investigate this possibility, we have examined the distribution of profits across risk segments.
  • Profitability and problem credit card debt: The third area we investigated is how the profitability of consumers relates to the problem debt outcomes defined in paragraph 6.58. If consumers with problem debt are profitable for firms, then this will not create incentives for firms to assist these consumers, even if a firm can identify the consumers that are struggling by their ongoing borrowing and repayment behaviour (e.g. after an initial lending decision). To evaluate this issue, we computed profitability for consumers that were identified, after 12 months of borrowing, by our problem credit card debt indicators.

In the following three subsections, we describe each of these pieces of analysis in more detail and present our findings.

Distribution of profitability

We begin our analysis of firms’ profitability incentives by considering the actual marginal profits on a sample of accounts. Some firms’ accounting systems were unable to provide details of variable operational costs at the account-level and so we use a sample cohort of consumers each coming on book in January 2010 across six credit card products, three products targeted at lower risk segment and three from the higher risk segment. This sample represents approximately 10% of accounts opened in January 2010. See Annex 6 for further details on our sampling methodology and sensitivity analysis.

In order to look at the profitability distribution, we order the five year profitability of accounts for each of the six products in our sample. We then split the data into one hundred groups based on their profitability and take averages for each percentile to give a distribution of profit and loss for each product. Figure 21 shows an average of these distributions with each product’s distribution split into two groups based on whether the product under examination is targeted at a higher or lower risk segment of the market.

This allows us to look at the five year profitability for all accounts in our sample from those making the highest losses through accounts that are breaking-even to accounts making the highest profits. The steeper the slope of the right hand side of the distribution, the smaller the proportion of accounts on which firms are making profits. The smaller the group making profits, the greater firms’ incentives to target this group and the greater the possibility that firms’ incentives could be misaligned with consumer welfare.

Figure 21: Distribution of five year profitability in the high and low risk segments

Figure 21: Distribution of five year profitability in the high and low risk segments

The lower risk products tend to produce more extreme tails. We find that a small proportion of accounts, approximately 30% of accounts, are responsible for 90% of profits but another small group of consumers, around 5%, are responsible for almost 90% of losses.

The large flat section of the low risk segment distribution curve is caused by a higher number of consumers in the lower risk sample using their account as purely a payment card. We know that the majority of transactors are typically close to break-even for the firm. We find that transactors can make a positive contribution to profits due to interchange fees but do not typically generate large profits, since they do not make interest payments. Firms can also have other financial or strategic reasons for attracting and retaining these accounts.

The tails of the distribution are more extreme for the lower risk segment because these accounts have higher average credit limits. Consumers taking out a lower risk card are typically less likely to default and tend to have higher credit limits. As a result, when compared with the higher risk segment, lower risk consumers can borrow more using these cards. This makes profitability higher but leads to larger losses when an account defaults, though this happens less frequently than in the higher risk segment.

Higher risk accounts tend to exhibit more borrowing behaviour, and although lower credit limits mean lower levels of profitability on individual accounts, a higher proportion of accounts are profitable as a result. It can also be seen that the rate of default is higher leading to more loss-making accounts. This is consistent with our understanding of firms’ product design.

A tension exists between firms’ incentive to maximise consumer borrowing versus its incentives to avoid default and hence losses, because those who borrow and earn profits are the same consumers who are more likely to default.

Expected profitability distribution

It appears that firms and consumer incentives are aligned to some extent, in so far as firms are incentivised to lower the instance of default as much as possible. Firms that can do so reduce the cost of default and improve their products’ financial performance.

Firms are not able to identify perfectly which accounts will default and which will be profitable. Firms are able to calculate a default rate and hence forecast default costs for groups of accounts. Firms then make their decisions based on the group as a whole. This means that, for the highest risk group, although by definition more accounts will default, if firms think that the profit generated from the other consumers in the group will outweigh these losses, they will be inclined to lend to this group.

Therefore, to fully investigate firms’ incentives we also need to consider the ‘in-expectation’ profitability of different risk groups. If the expected profitability of a very risky group was disproportionally high, firms would be incentivised to lend to that group, even though it would tend to inflate the occurrence of default in the short run. This could lead to the kind of misalignment of incentives that we are concerned about.

To conduct this analysis we used the same sample as before and report our findings based on the average results of the three products in the lower risk segment and the three products in the higher risk segment. However, where before we looked purely at five year profitability, here we used rates of return on lending as a further measure of financial performance for products offered by different credit card firms. This allowed us to take into account not just profits but the amount of money being lent to generate that profit, allowing us to better assess firms’ incentives.

We calculated the return on lending by dividing the total profits made by all accounts with the same product (the same five year profitability measure used in the profit distribution analysis above) by the average outstanding balances of that cohort. We then annualised this monthly return estimate.97

Figure 21 shows that products targeted at the higher risk segment tend to generate higher returns on lending than in the lower risk segment. However, Figure 22 allows us to look at how rates change as risk increases, with risk category 1 being the lowest harmonised risk category (see Annex 6) and risk category 15 being the highest.

This approach to analysing the data allows us to look at how average annualised return changes based on firms’ initial assessment of risk. If the data shows that very high risk groups generate the greatest returns, firms would tend to be incentivised to target these groups, even if default was more probable, potentially leading to a misalignment of incentives.

Figure 22: Return on lending rates by initial risk category

Figure 22: Return on lending rates by initial risk category

Our analysis demonstrates that, as expected, lower risk individuals do not take out higher risk cards and higher risk individuals do not have access to lower risk cards. By targeting a specific range of risk, firms are able to forecast default cost and match product characteristics (such as interest rate, fees, credit limit) to deliver profits in line with internal targets.

Typically, as risk increases firms compensate for increasing default costs by offering terms that are less generous; credit limits fall and the interest rate increases. Despite lower credit limits, utilisation and revolve rates tend to rise as risk increases. Higher risk groups can therefore generate higher revenue, though default costs and other organisational costs such as capital requirement costs will also rise. Firm’s incentives will be to target the point where profitability and returns are maximised.

Figure 22 shows for the lower risk segment the optimal profitability point tends to occur at relatively low levels of risk. Whereas the higher risk segment tends to maximise returns in the mid-to-high groups. At very high levels of risk, profits decrease on higher risk products because default rates increase faster than revenue. At lower levels of risk the reduced cost of bad debt is not enough to compensate for loss of revenue caused by improved terms.

Based on this analysis, we conclude that higher risk, and potentially more vulnerable, consumers are not the most profitable group of consumers and that firms do not seem to be incentivised to target these groups. Rather, firms make greatest returns from consumers for whom the product was designed. Consumers at the edge of products’ risk appetite are typically less profitable.

Profitability and problem credit card debt

Consumer behaviour may be unknown when an account is first created. However after a period of time, firms can determine with reasonable accuracy what behaviour an account is exhibiting.

If accounts exhibiting behaviour that could lead to problem debt are highly profitable this could indicate firms have incentives to target these consumers once they are identified, or to encourage existing consumers to engage in behaviour that could lead to problem credit card debt.

To assess the relationship between profitability and the problem credit card debt groups detailed above we split the account-level data into five groups based on our problem credit card debt indicators.98

To assess the financial performance of these groups we used both average profitability figures and the return on lending measure from the previous section. We also looked at these performance metrics in conjunction with the numbers of accounts in each category, credit limits and other account data stratified by the groups defined above.

Figure 23: Return on lending rates for problem debt groups in the lower and higher risk segments

Figure 23: Return on lending rates for problem debt groups in the lower and higher risk segments

Severe arrears

Our analysis shows that accounts initially classified as being in severe arrears are, on average, heavily loss-making.

Losses are driven by lower credit limits (indicating firms had already identified these accounts as higher risk) which leads to lower interest income. These accounts typically have a much shorter period than other accounts before default occurs which would further reduce revenue. Lower credit limits and short active durations also lower costs on these accounts. However default costs are higher than in any other category to such a degree that accounts are loss-making on average.

Firms’ credit risk models are designed to screen out accounts that would default within 12 months. Given our definition of severe arrears identifies accounts that have either already defaulted or are close to default within 12 months it seems likely that the majority of accounts in this category are attributable to the error margin of firms’ models. With respect to these accounts, the incentives of the firm seem to be largely aligned with the consumer, as both have an incentive to avoid default.

Serious arrears

Firms have less incentive to eliminate accounts in the serious arrears category. Taken as a whole this group tends to be in expectation profitable in both the higher and lower risk segments, though the low risk segment is very close to break-even.99 Profitability in the higher risk segment is driven by the fact that interest income in the serious arrears category is far higher than for those in severe arrears whilst charge-off cost is, on average, far lower. The serious arrears category typically has lower credit limits, lower interest income and higher default rates, than the persistent debt, minimum payers or accounts with no problem debt categories. This makes this category less profitable than all categories except for the severe debt. However, the longer duration before default in the serious debt category makes it a profitable proposition.

We note that while firms are unlikely to prefer serious arrears accounts, since the rate of return is lower than for all groups other than severe arrears, firms have limited incentives to screen out this type of account, assuming this were possible.

We acknowledge that accounts falling into severe and serious arrears groups could be similar types of consumer separated only by the timing of their default. If these two groups are linked to the extent that firms could only exclude accounts in the severe debt category at the expense of excluding serious debt accounts, they may be incentivised not to do so. Firms’ incentives would depend on the number of serious and severe accounts excluded by the new scorecard, as well as the relative profitability to the firm of each group.

Systematic minimum repayments

Systematic minimum payers have slightly higher credit limits than the serious arrears category but pay much higher levels of interest income on their accounts. This drives higher levels of profitability and return.

Systematic minimum payers and those not identified by our problem credit card debt indicators tend to have similar credit limits on average. Figure 23 shows that in the higher risk segment, the rate of return on lending is higher for the no problem category than for minimum payers. However, in the higher risk segment the average profit for a systematic minimum payer is higher than for the no problem category but a higher active balance tends to reduce the return on lending ratio. In other words, generating slightly lower profits on no problem accounts requires proportionally far less lending by firms. In the lower risk segment returns from minimum payers are, on average, the same as returns from the no problem category.

Persistent debt

Similarly, the persistent debt group returns, on average, a higher rate of return than all other groups in the higher risk segment. In the lower risk segment returns are comparable with the minimum payment and no problem categories. This might indicate that there is some incentive to encourage high utilisation behaviour in the high risk segment. This is in line with the incentives discussed in Chapter 5 especially around the standard products typically offered in the higher risk segment.

Summary

Our analysis shows that accounts that in their first year are in serious arrears, systematically minimum paying or having a very high persistent level of debt on average all generate positive returns over the product life-cycle.

We conclude that firms in the lower risk segment have no particular incentives regarding accounts in the no problem, systematic minimum payer or persistent debt groups, though they have clear incentives to minimise default across their portfolio.

For firms in the higher risk segment firms would also prefer accounts that are in the no problem, minimum payer or persistent debt groups. However they may also be incentivised to push up utilisation. Depending on the profitability and number of accounts in serious and severe arrears firms may also find that adopting score cards designed to minimise these groups may not be profit-maximising.

Based on this, our view at this stage is that firms make losses on consumers that default so they have a strong incentive not to lend to these consumers. However, firms have fewer incentives to avoid lending to consumers who have persistent debt or make systematic minimum repayments because they are profitable to the firm


72. FCA, Credit card market study terms of reference (2014) MS14/6.1

73. Step Change, Personal Debt 2014; Statistics Yearbook Findings (2014)

74. Step Change, Life on the Edge: Towards more resilient family finances (2014)

75. CMA Problem Debt – a report commissioned by the Consumer Protection Partnership (2014)

76. Step Change, Personal Debt 2014; Statistics Yearbook Findings (2014)

77. Christians Against Poverty official response to the FCA’s credit card market study terms of reference.

78. These issues were considered when the mortgage market was reviewed. In that case, borrowing was deemed to be affordable (i.e. not problematic) if repayments could be met with disposable income.

79. Research submitted to the FCA from UK Cards, March 2015.

80. See Annex 6 for summary of relevant academic literature.

81. CONC 5.2.1R and 6.2.1R

82. CONC 7.3.4R

83. CONC 6.7.2R

84. This covers existing credit card customers, as well as customers who use other types of products with the provider, such as current accounts, mortgages, savings accounts, and personal loans.

85. Some firms told us that they do make some efforts to verify data where possible. For example, one firm told us that it will check an applicant’s stated income against current account turnover data. Another firm told us that they verify an applicant’s stated income where there is a mismatch between salary, age and employment type. Validation may require the customer to send in proof of income.

86. FCA Business Plan (2015/16)

87. Charged-off refers to debt that issuers have deemed unlikely to be collected and that they have written off on their balance sheet. Consumers whose accounts have been charged-off have not been relieved of their repayment requirement, and charged-off accounts are often pursued via collection processes.

88. We noted that the distinction between this category and the serious arrears category is partially driven by firm practices and their decision of when to charge-off a consumer. Some firms will do this sooner than others. This will mean that there is a degree of overlap between these two categories that the data does not reflect.

89. We chose three repayments as the threshold rather than one or two repayments as we considered that the latter two thresholds may capture a number of consumers that simply missed repayments due to inattention.

90. This was calculated by finding the credit limit utilisation each month and then taking a simple average across months. We also considered an alternative measure of persistent debt which was based on the actual value of debt over time rather than credit limit utilisation. In particular, we were interested to see how sustained levels of credit card debt are, and whether there was issue with consumers being unable pay-down their outstanding balance once it had reached a particular level. This is discussed in more detail in Annex 6.

91. This ordering is supported by analysis presented later in this chapter regarding cost of credit and estimated length of repayment.

92. An alternative measure of cost would be to calculate the Annual Percentage Rate (APR). This measure is usually applied when there is a fixed repayment schedule, which does not apply in the case of credit cards. As a result, applying the APR measure to credit cards raises a number of methodological challenges. An APR can be calculated each month, but then it must be decided how best to weight the monthly APR to give a single number. One solution would be to introduce a weighting scheme based on closing balance. However, such a solution creates new challenges, such as the treatment of fees and charges for consumers who are characterised by transacting behaviour, where this important component of costs would be ignored. Given these difficulties, we opted for the simple measure described above.

93. As an indication of how representative average repayments over the last six months are as predictor of future repayments, we estimated the average repayments made on accounts over the six months prior December 2012 and measured how well they forecasted repayments in 2013. We found that the actual level of repayments predicted by this measure were within 30% of the average actual repayments for 52% of accounts.

94. This analysis has been conducted at the account-level as undertaking it at the consumer-level would be more complex and require making additional assumptions in order to capture the repayment behaviour of consumers across multiple cards.

95. Many of these consumers will be transactors, paying little or no interest and charges.

96. The cost of credit estimates reported reflect the observed costs that have accrued on accounts to date. No allowance is taken for the size of the outstanding balance at the end of the sample or the length of time it is expected to pay off this balance. See Annex 6.

97. The cohort’s average monthly balance is the average of all accounts’ average outstanding balance over five years (see Annex 6 for further details).

98. See paragraph 6.58.

99. We discuss in both chapters 3 and 5 firms’ wider strategic incentives regarding account which are close to break-even.