Rewards: do they work, and for whom?

Sumit Agarwal, Sujit Chakravorti, and Anna Lunn of the Federal Reserve Bank of Chicago published a paper that explores how reward programmes affect consumer behaviour (. 

They find that consumers that receive an average of $25 of rewards over one year.  During the first quarter, these customers increase their spending and debt by $68 and $115 per month respectively.  During the following 3 quarters, average spending and debt rise by $76 and $197 per month. 

Importantly, customers increase their debt by more than their spending, presumably as switch their repayment to other cards in an effort to consolidate their debt to the rewards card.  This is backed up by evidence that overall spending and debt accumulation remain constant or increase slightly, suggesting that cardholders substitute spending and debt from other credit cards.

The authors conclude that only a small incentive is necessary for customers to significantly change their behaviour and therefore an effective tool for banks to attract new customers.  Furthermore they suggest that this is proof of time inconsistency and bounded rationality on the part of the customer.  In other words, consumers do not initially intend to use the credit line, and therefore ignore the interest rate, but change their mind when the bill comes, leading to higher than intended debt levels.

In my mind, this conclusion is incorrect on many levels.  The authors have already shown that consumers’ overall spend and debt-levels stay constant or increase slightly (possibly due to substitution of cash transactions).  As they receive a $25 incentive to switch to a different card, their increased spend and debt on that product therefore seems entirely rational. 

More interestingly; are the authors correct in stating that cash back offers are effective in attracting new customers?  The challenge with cash back offers is that banks benefit only from the incremental spend and debt that customers put on their products, while banks needs to reward total spend, including the spend that they already captured. 

In our case, banks capture $888 of incremental spend during the first year.  If we generously assume that banks have an average margin of 1% on this spend, the incremental spend is worth less than $9.  Against an investment of $25, banks must generate $16 from other revenues in order for the rewards offer break even.  With an average increase in debt of $177 during the 12 months following the cash back reward being introduced; banks require a net spread after credit losses of at least 9% to break even in the first year. 

Banks may be happy to treat the first year as an investment and look to generate profits from the programme during following years.  However, as their customers have already revealed their fickle nature when switching their spend, it appears risky for banks to treat this as a multi-year investment.

Our analysis is of course based on high level assumptions and patchy data, but should cast doubt over the claim that reward programmes are effective (and profitable) marketing strategies for banks.  It is my belief that the rewards space is now so crowded (particularly in more sophisticated markets) that in order for programmes to be effective they must be genuinely differentiated.  These are likely to be offers of products or services, carefully targeted to the specific customer and based on hard-to-replicate technological platforms.