Excellent score but lowest introduction rate

Sandtree

I've never seen the code for underwriting a credit card but have for underwriting the likes of car insurance there are thousands of lines of code and that was before any significant AI was added. What you are offered can have nothing to do with you... the system starts seeing a run on a certain set of characteristics and the price for that automatically increases. 

My industry typically also doesn't explain its decisions to customers either and for similar reasons as banks. We don't want manipulation, our pricing is commercially sensitive and imagine trying to explain to a customer they are being quoted so much because we had an unusually high percent of Golf GTI's buying with us in the hours before the caller asked for a quote or that our propensity modelling shows your cash rich/time poor so we quote higher with all the optional extras included but message the quote with marketing about how easy we make everything?

Credit card underwriting, particularly with these types of promotions you have the script making two decisions... what limit and what duration of offer. Personally I don't know the tension between the two but certainly over the years I have times when given a £15k initial limit but not the full 0% period. Other times the limits have been very modest but had the maximum promotion.

If they did give you pointers it could well be a dozen different ways with a dozen steps in each and many requiring 6 months or so to flow through but in 6 months time the underwriting script may have changed and some of the actions you've now taken are negative as ideas have evolved or new data augmented.

Regulators can, and do, request information to ensure rules are being followed... so card issuers have to confirm the APRs being given in comparison to the stated APR to ensure they are meeting the requirement that at least 50% of successful applicants get the advertised rate or better.

theroninhunter

@Sandtree thanks for the nicely written response, very detailed.

Unfortunately auto decision systems are more common now and I have been on the rejected side of a system but after a manual review they reversed the decision.

Again they didn't provide a specific reason but highlighted the system should have allowed it. Unfortunately it was so difficult to get the company to actually do a manual review and this seems to be a growing issue with any company. The auto decision system is always correct and never wrong. 

Sandtree

theroninhunter said:

@Sandtree thanks for the nicely written response, very detailed.

Unfortunately auto decision systems are more common now and I have been on the rejected side of a system but after a manual review they reversed the decision.

Again they didn't provide a specific reason but highlighted the system should have allowed it. Unfortunately it was so difficult to get the company to actually do a manual review and this seems to be a growing issue with any company. The auto decision system is always correct and never wrong. 

Automated decision making has been common for a very long time now... what's more concerning/interesting now is the AI/ML aspect.

In my consumer days we sat in a room twice a month and the pricing analysts, underwriters (the big wig type not the day to days), commercial manager would talk through the analysis on elasticity and claims since the last meeting and agree to raise the price for 40-45s with a prestige car by 5% because analysis shows only a 1% reduction in sales so quids in, double the price of TPFT because we feel its not a brand fit and we have a small but growing book after another big name had withdrawn from the market then see that the 3% reduction on stay at home mums hadn't caused the uplift in sales we'd anticipated so back that out..

ML makes changes in realtime, it can consider all aspects and not just focus in on 3-4 which is all we had time to in our pricing committee meetings. It allows rating to be much more complex, we'd be able to link a few characteristics (eg sex and age) but ML can test and branch almost unlimited if it starts finding statistically relevant correlations. Clearly people keep an eye on what its doing etc but inevitably as time goes on and the technology proves itself the scope and autonomy will increase.

Ironically the time when they gave me a £15k limit but only 2/3rds of the 0% transfer offer was as the result of an appeal... the initial decision was a straight decline.

ThisnotThat

theroninhunter said:

@Sandtree thanks for the nicely written response, very detailed.

Unfortunately auto decision systems are more common now and I have been on the rejected side of a system but after a manual review they reversed the decision.

Again they didn't provide a specific reason but highlighted the system should have allowed it. Unfortunately it was so difficult to get the company to actually do a manual review and this seems to be a growing issue with any company. The auto decision system is always correct and never wrong. 

It's incredibly rare that the automated systems are wrong.  I speak from experience here, not as a lay person.

I'd add that there is the quasi Dunning-Kruger effect with lending.  Most people massively overestimate their creditworthiness, largely because there are excuses for whatever misdemeanors they've commited, and they expect a lender to think the same thing.

Turns out the lenders do not.

theroninhunter

Sandtree said:

theroninhunter said:

@Sandtree thanks for the nicely written response, very detailed.

Unfortunately auto decision systems are more common now and I have been on the rejected side of a system but after a manual review they reversed the decision.

Again they didn't provide a specific reason but highlighted the system should have allowed it. Unfortunately it was so difficult to get the company to actually do a manual review and this seems to be a growing issue with any company. The auto decision system is always correct and never wrong. 

Automated decision making has been common for a very long time now... what's more concerning/interesting now is the AI/ML aspect.

In my consumer days we sat in a room twice a month and the pricing analysts, underwriters (the big wig type not the day to days), commercial manager would talk through the analysis on elasticity and claims since the last meeting and agree to raise the price for 40-45s with a prestige car by 5% because analysis shows only a 1% reduction in sales so quids in, double the price of TPFT because we feel its not a brand fit and we have a small but growing book after another big name had withdrawn from the market then see that the 3% reduction on stay at home mums hadn't caused the uplift in sales we'd anticipated so back that out..

ML makes changes in realtime, it can consider all aspects and not just focus in on 3-4 which is all we had time to in our pricing committee meetings. It allows rating to be much more complex, we'd be able to link a few characteristics (eg sex and age) but ML can test and branch almost unlimited if it starts finding statistically relevant correlations. Clearly people keep an eye on what its doing etc but inevitably as time goes on and the technology proves itself the scope and autonomy will increase.

Ironically the time when they gave me a £15k limit but only 2/3rds of the 0% transfer offer was as the result of an appeal... the initial decision was a straight decline.

TBH if banks could offer the choice of either a higher limit, longer introduction offer or perhaps a sort of sliding scale (obviously that could only be done after application submission) , I would happily choose the longer introduction offer

ThisnotThat

theroninhunter said:

Sandtree said:

theroninhunter said:

@Sandtree thanks for the nicely written response, very detailed.

Unfortunately auto decision systems are more common now and I have been on the rejected side of a system but after a manual review they reversed the decision.

Again they didn't provide a specific reason but highlighted the system should have allowed it. Unfortunately it was so difficult to get the company to actually do a manual review and this seems to be a growing issue with any company. The auto decision system is always correct and never wrong. 

Automated decision making has been common for a very long time now... what's more concerning/interesting now is the AI/ML aspect.

In my consumer days we sat in a room twice a month and the pricing analysts, underwriters (the big wig type not the day to days), commercial manager would talk through the analysis on elasticity and claims since the last meeting and agree to raise the price for 40-45s with a prestige car by 5% because analysis shows only a 1% reduction in sales so quids in, double the price of TPFT because we feel its not a brand fit and we have a small but growing book after another big name had withdrawn from the market then see that the 3% reduction on stay at home mums hadn't caused the uplift in sales we'd anticipated so back that out..

ML makes changes in realtime, it can consider all aspects and not just focus in on 3-4 which is all we had time to in our pricing committee meetings. It allows rating to be much more complex, we'd be able to link a few characteristics (eg sex and age) but ML can test and branch almost unlimited if it starts finding statistically relevant correlations. Clearly people keep an eye on what its doing etc but inevitably as time goes on and the technology proves itself the scope and autonomy will increase.

Ironically the time when they gave me a £15k limit but only 2/3rds of the 0% transfer offer was as the result of an appeal... the initial decision was a straight decline.

TBH if banks could offer the choice of either a higher limit, longer introduction offer or perhaps a sort of sliding scale (obviously that could only be done after application submission) , I would happily choose the longer introduction offer

They offered you what they were happy with 

It's not your choice to make, it's not your money being lent out.

Sandtree

ThisnotThat said:
They offered you what they were happy with 

It's not your choice to make, it's not your money being lent out.

Fully agree its their decision alone and that I've no right to know their thinking... that doesn't mean I don't remain curious. 

Given you state first hand experience... what is the general thinking of limit -v- promotional period? There appears to be a level of independence/difference of weighting because as mentioned, have had big limit and reduced promotion and small limit but full promotional period. 

theroninhunter

I don't see why they could not ask what your preference would be higher limit or longer introductory offer?

Most companies do ask what credit limit you are looking for albeit its not guaranteed so surely something similar could be designed.

I personally don't need high limit accounts as I don't live beyond my capacity. 

[Deleted User]

theroninhunter said:

I don't see why they could not ask what your preference would be higher limit or longer introductory offer?

Most companies do ask what credit limit you are looking for albeit its not guaranteed so surely something similar could be designed.

I personally don't need high limit accounts as I don't live beyond my capacity. 

It's a dangerous choice as people could game it and then start complaining that they were miss-sold with the information given and start fighting for compo. I understand 100% your frustration with having a perfect record (not score) and not getting what you wanted but the lender will 100% never give you the reason why you didn't get the full rate. This is because that information would eventually make it onto sites like this and lead to people tweaking their applications to get the best chance of the offer, regardless of their real world situation.

A good analogy is car insurance, sites like this encourage you to fiddle around with things like your job title, to lower quotes, which is fine, but leads to quote manipulation where people start playing with mileage, where car is kept etc and suddenly they get a manual intervention and could lead to the policy being voided. With a CC an unscrupulous person might be tempted to inflate your income (say claiming your annual bonus was guaranteed income) or play around with other data in order to "fit" your application to how you think the lender wants it to look.

Thrugelmir

theroninhunter said:

Again they didn't provide a specific reason but highlighted the system should have allowed it. Unfortunately it was so difficult to get the company to actually do a manual review and this seems to be a growing issue with any company. The auto decision system is always correct and never wrong. 

Because lending books at written are the macro level not the micro. In days gone by when you wanted a car loan. You had a face to face interview with a bank manager who grilled you over your finances. These days to make the cost of borrowing cheaper technology is used. Vast arrays of statistical data are used to both manage and price risk.  As a consumer if you don't like what you are offered vote with your feet. No different than going shopping or booking a hotel or a plane ticket. There's always someone us to do business with.