Risk Based Pricing: Is It Fair and Sensible?

It is not uncommon for organisations to resist the implementation of risk based pricing because they either fear that it may  lead to higher defaults among the high risk groups in the long run or they feel that it may unfairly punish certain customer groups.

Three mistaken assumption usually underlie this resistance: the first of which is that a higher price necessarily leads to a higher repayment burden; the second assumption is that risk based pricing strategies are unfairly biased against poorer customers; and the third assumption is that when setting a risk based price, a lender’s only choice is between accepting a loan at a low price and accepting that loan at a high price.

Does Risk Based Pricing Accentuate Risk?

The first common reason that organisations may resist the implementation of risk based pricing strategies is that they feel it will result in a higher monthly repayment burden for customers who are already less likely to pay and so cause an even worse situation characterised by a vicious cycle of ever-increasing risk.

This need not be the case for two reasons.  Firstly, The relationship between price and the eventual monthly instalment is governed by an annuity curve and the nature of that annuity curve means that an impact on instalment of an increase in price is significantly diluted.  Secondly, with an annuity curve the interest rate charged on a loan is not the only factor that determines a loan’s monthly instalment.  The size of the loan and its term are the other two.  An increase in the interest rate can be offset by a change in one of these other factors – either a decrease in loan size of an extension of loan term.

Consider an example of a portfolio of unsecured loans, each of €15 000 and paid in monthly instalments over five years.  The lender charges everyone 10% and pays 5% cost of funds.  In this base case, a borrower will pay a monthly instalment of €319 of which €36 goes to the lender as gross margin earned before risk.  Now, assume that a group of customers has been identified with significantly higher risk.  Increasing the rate charged to this group from ten percent to twelve percent will increase the pre-risk margin earned on this group by 41% (€15) while only resulting in an increase in the monthly instalment of 5% (the same €15).  If the lender wished to avoid this increase altogether, they could simultaneously increase the term of the loan by four months.

The key is timing: a lender should always determine the price of a loan before determining its size and term.  The price should be based on risk; the size and term on affordability.  If the order of these decisions is reversed, higher risk customers will always be paying higher monthly instalments and this may inadvertently make the new loan unaffordable.  However, if price is a fixed input to the annuity curve, the monthly instalment can easily be managed.

When done correctly, the effect of risk based pricing should actually be a lowering of overall risk in a portfolio.  A well implemented risk based pricing strategy will use prices to discourage high risk customers and encourage low risk customer and thereby improve the risk make-up of the portfolio over time.

Is Risk Based Pricing Fair?

The second source of resistance stems from a sense of fairness.  No one wants to charge a customer more just because they are poor.  However, to make this assumption is to incorrectly substitute the concept of ‘high risk’ with the concept of ‘low income’.

Risk based pricing strategies, as the name suggests, assign prices to loans at a level congruent with the perceived risk.  This is done because risk is a key driver of cost differences between otherwise similar loans.  While there are of course times when there will be an inverse relationship between income and risk, there are also many times when low-income customers are also be low risk customers.

The risk of a loan should be defined as the probability that a customer with a given set of characteristics will have both the willingness and ability to pay down a loan that, given their income, was affordable at the time the loan was issued.

I have previously discussed the issue of risk versus affordability but in summary again here, the risk of a loan is usually measured by looking backwards at a customer’s previous behaviour and assumes that the past behaviour can be used to predict future behaviour by also assuming no significant changes to circumstances.  The affordability of a loan is measured looking forward, testing that assumption of ceteris paribus.

Or put another way, a customer is of low risk when their previous behaviour suggest they will continue to pay back their existing loans and that customer is able to afford a new loan when the addition of said loan to the existing debt burden does not change this fact.

So, the payment behaviour of a customer will determine risk – and thus price – while available income will determine affordability and thus the size of the loan.  Risk based pricing will therefore not necessarily lead to higher prices for poorer customers.  Rather, higher prices are charged where higher costs – in terms of capital provisions, operational costs and write-offs – are expected.

Is A Low Price The Only Alternative To A High Price?

The third common but mistaken assumption is that, with risk based pricing, a lender is making a conscious choice between granting a loan to a high risk customer at a low rate and granting the same loan to the same customer at a higher rate.  This assumption, like the previous one, creates the impression that a lender is unfairly seeking to maximise their profit at the customer’s expense.  However, this is not the case.

However, the lender is really making a choice between granting a loan to a high risk customer at the higher rate and declining the loan.

The higher rate charged is to compensate for higher costs and so, not charging this premium would lead to the lender suffering a loss on each high risk loan granted.  In our previous example, if the monthly risk and non-risk costs of the higher risk group were €37 then it each loan would have been loss-making unless the price was increased.

Where a lender has a sufficiently rich understanding of the cost and revenue of the loans they grant, risk based pricing strategies become as much about increasing acceptance rates as they do about increasing revenue.  This is true of pure risk based pricing but may become blurred when dealing with profit based pricing although here it can be argued that the extra profit is being generated on those with the least price sensitivity not on those with the highest risk.

 

Summary

All loans are not equal in terms of risk and since risk is a key driver of costs, all loans do not have the same cost structures.  If costs vary but revenue remains steady, profitability will vary with cost.  Risk based pricing is simply a means of better assigning price to compensate for these cost differentials and thereby to ensure profitability across all loans.  Provided it is well implemented, risk based pricing is neither unfair nor likely to increase the risk of a portfolio.

Read Full Post »

Profit Based Pricing

In a previous article I discussed the basics of risk based pricing: using an understanding of risk to assign a price to each loan based on the cost of risk that is expected to accompany it.  But risk based pricing in this form is not the summit of strategic pricing; it is one of the foundations. 

Once an organisation has mastered risk based pricing it should look to implement a profit based pricing strategy.  That is, a pricing strategy that takes into account both long-term risk and long-term reward.

In its most basic form, risk based pricing will group accounts along a single axis showing their relative risk according to some internal risk measure.  In most cases this can be easily improved by adding a second axis showing an external view of risk – usually in the form of a bureau score.  The result then is a matrix which can be split into price segment and will something like the one below:

 

In the matrix above, a two dimesnional matrix has been used to identify four specific pricing bands – each represented by a different colour block.  The key change when one moves to profit based pricing is the addition of third dimension – representing price sensitivity or some other revenue proxy – turning the matrix into a cube. 

Customers do not respond equally to changes in price.  So while some low risk customers might be persuaded to sign-up by a 0.5% discount, others may require a 1.5% discount before taking action; likewise on the other side, some high risk customers might accept a 3% price premium while others may leave if they have to pay any more than a 1% premium.

Customer reaction is governed by customer-unique factors as well as by external market conditions.  A very price-competitive market with many different lenders it will force bigger discounts and smaller premiums while a less price-competitive market will allow smaller discounts and larger premiums to be charged.  However, under any one of these prevailing market conditions there will still be some individuals who are more swayed by price changes than others.

Adding the third dimension allows you to fine tune your pricing strategy, turning some of the deep discounts into smaller discounts and some of the small premiums into larger ones as well as potentially offering more pricing options.

 

Price sensitivity probably impacts the marketing response rate more than any other metric.  In a marketing campaign with a fixed cost, a larger the response rate results in lower marketing cost per account signed and thus increased marketing efficiency.  But this increase in efficiency comes at a cost, since each customer will now generate less individual profit.   

A given discount offered to two groups will therefore have  two different impacts on profit.  A discount is value-adding when it leads to an increase in response rate that is sufficiently large to offset the drop in per-customer profit.  A discount that fails to do this is value destroying.  So, response rates on their own can’t be taken as a measure of success.  They need to be measured against a benchmark to ensure the increase in response rates has compensated for the decrease in per-customer revenue.

On the other side of the coin the profit mechanisms remain the same, albeit reversed.  A premium rate increases the profit per account but decreases the number of profit generating customers.  Overall profit is increased so long as the premium charged does not decrease the total response rate too far. 

However, there is one caveat here.  In a portfolio without risk based pricing it is likely that the highest risk customer group will be value destroying.  Among this group, even a price premium that results in a response rate of zero will add value.  This is because, although it has not directly created value it has prevent value destroying accounts from the entering portfolio.  However, it would not be optimised as there is likely to have been a smaller premium that would have resulted in a non-zero response rate at a price point that was still profitable.  A too large discount offered to the lowest group has no such in-built safety net and can result in a large decrease in total profit if it does not lead to sufficiently increased response rates.  This is why it is always safer to start any differentiated pricing project with the charging of price premiums to high risk customers rather than with providing discounts to low risk customers. 

Of course measuring price sensitivity per account – or per group of accounts – is easier said than done.  The best way to do this from scratch is through a thorough series of test-and-learn campaigns designed to measure the relative response rates to offers at a variety of prices.  Over time, this will allow you to collect enough data to build a price sensitivity model.  This requires a level of sophistication within your organisation and a certain data analysis and data storage capabilities.  This internal data can be supplemented with external marketing data – the sophistication and availability of which will vary on a market-by-market basis – for even better results.  Where a shorter term solution is required, I would suggest structuring a long-term test-and-learn campaign as above but using the first results to build a basic model which can then be optimised and upgraded on a dynamic basis as further results come in.

Read Full Post »

Risk Based Pricing

There are two routes an organisation can take when setting a price for its product: they can calculate the cost of producing that product, add a margin and hope that the market will pay the resulting price or they can start by taking their price from the market and then find ways to produce the product for a lower cost.  In reality though, any business operating in a competitive environment must follow the second route.  This is particularly true for lenders whose product, money, is essentially a commodity.

However, although money is a commodity that does not mean there can only be one price.  In fact, the commodity nature of money only applies to each individual customer, not to a group of customers.  In other words, though one individual may see a loan from any bank as equally valuable customers are sufficiently different from each other that as a group they will all assign significantly different values on the same loan.  These differences are what allow risk based pricing to be a viable business strategy.

Risk based pricing strategies seek to assign a fair price to each customer – or, more accurately, to each group of customers.  The measured difference in risk between each customer is used to assign differentiated prices.  The logic is that operationally every customer costs the same to service, it is only the cost of risk that fluctuates – or at least, it is the cost of risk that fluctuates the most.

One way to think about this is to imagine a company that produced two version of its product, both essentially the same except that one comes fitted with an extra resilient outer shell to protect it if it is dropped.  Customers who only use the product at home would have little need for the costly outer shell and could therefore buy the cheaper version while customers who use it regularly outside might buy the more expensive but safer model.  Although each group of customers are buying essentially the same product, they have paid a different price and that difference in price is based on their risk.  It is the same with loans.  High risk customers must pay higher prices in order to cover the higher cost of potential write-offs.  The only difference really between the imagined scenario and the real one is that in the imagined scenario it would be the purchaser of the product who would carry the cost of the risk while in the latter it is the lender who carries it.  So, in the first case the purchaser would decide which option was cheaper in the long run and voluntarily pick the more expensive option where applicable while in the second case that premium must be enforced.

Unless risk based pricing is implemented, lower risk customers will be subsidising the cost of higher risk customers.  This situation is doubly problematic as it doesn’t only lead to sub-optimal profit but also to a sub-optimal portfolio structure.  The lowest risk customers (who are being over-charged) can be tempted to leave for cheaper competitor offers while the higher risk customers (who are being under-charged) will gravitate towards your product; leading to a riskier portfolio on average.  Risk based pricing addresses this by lowering the rates charged to low risk (cheaper to serve) customers and raising the rates charged to high risk (expensive to serve) customers.

The key to successful risk based pricing is, not surprisingly, a good understanding of risk.  So scorecards should once again be at the heart of any strategy.  Scorecards should be built off historical data and used to split customers into groups of similar risk – the better the scorecard’s discriminatory power, the more groups will be possible.  The groups should represent a fluid progression of risk in a portfolio.  Unlike a basic originations scorecard, the goal here is simply to order risk and not to make a cut-off.  Once customers have been assigned to groups it is best to add a second dimension using any available external risk data to create a risk matrix. 

Thus assigned to groups based on internal and external risk measures, the risk of each group should be measured and compared to the average risk of the portfolio.  In complex strategies all costs can be added together to get an accurate price but in most cases it is sufficient to use a relative pricing approach.  This means that the average risk group is assigned a baseline price – usually the price everyone was being charged before risk based pricing was implemented – and all other risk groups are assigned a premium or discount based on their risk relative to that of the ‘average’ group.  In a simple case where all customers had been paying 10% interest, you might be able to identify three distinct risk groups and, based on their risk differentials, charge the average risk group the baseline rate, the lower risk group a 1% discount and the higher risk group a 2% premium: or 10%, 9% and 12% respectively.  This approach assumes that the baseline price is accurate and sufficient to cover all costs and profit expectations, an assumption that should be tested periodically.

 When implementing risk based pricing for the first time it is worth following a phased approach, starting by charging premiums to higher risk customers first.  High risk customers that are unprofitable at the current rate are value destroying and so the downside risk of a price premium is minimal – either they pay the premium and increase the portfolio’s profitability or they leave and, in so doing, increase the portfolio’s profitability.  The downside risk of a price discount is higher.  For a price discount to increase the profitability of a portfolio it needs to lead to sufficiently larger total balances to compensate for the lower revenue per dollar of balance now received.  If too large a discount is offered relative to the resultant growth in balances, portfolio profitability can be destroyed.

The best way to limit these risks during a roll-out of risk based pricing is to do thorough data analysis.  All sources of revenue and cost should be considered, including knock-on effects to other products, before it is decided whether a customer is worth keeping or losing.  Small discounts and premiums should then be offered through test-and-learn process which can be slowly increased over time as experience and confidence in the models increases.

 *** Risk based pricing is only the first step however and once this has been mastered it is important to take the next step by considering more general marketing concepts and, in particular, price sensitivity.  A profit-maximising organisation should seek to provide as small a discount as possible to low risk customers – just large enough to attract them – and to charge as large a premium as possible to high risk customers – just small enough not to cause attrition.  Price sensitivity can be measured through the intelligent application of test-and-learn experiments or, less optimally, by a subjective analysis of market trends.

*   *   *

But is it fair to implement risk based pricing?  Read this article to see why risk based pricing is both fair and sensible in the long run

Read Full Post »