> “There is a growing idea in Silicon Valley that there are sources of data on consumer behavior we can use to predict creditworthiness. These will be completely different than the traditional approach to credit ratings, which are tremendously imprecise and ‘laggy.’ PayPal can do a real-time credit score in milliseconds, based on your eBay purchase history — and it turns out that’s a better source of information than the stuff used to generate your FICO score.
Traditional credit ratings are "tremendously imprecise"? That's an insane statement. FICO is a proven model and has weathered multiple business cycles. Virtually all of the alternative models being experimented with today haven't. A lot of these models are going to fall apart when the current environment, which has seen record low credit default rates, changes.
I'd love to see a $50,000 auto loan or $400,000 mortgage approved on the basis of eBay purchase history alone.
> “The hypothesis is that there are many other similar sources of consumer data: credit card bills, social-network behavior, potentially even search history. Lots of people, both in the big Internet companies and at start-ups, are trying to get at these large pools of data and figure out new ways to do scoring. What they all have in common is that they are all being done outside of banks.
Notwithstanding the fact that credit payment history ("credit card bills") is factored in to a FICO score, a couple of things should be pointed out:
1. Many of the companies trying different models are doing so in an attempt to serve thin file borrowers. Not surprisingly, established players like Fair Isaac aren't sitting around twiddling their thumbs. They have their own solutions for these borrowers, like the FICO Expansion Score. Just because startups don't want to pay for somebody else's solution doesn't mean they're the only ones innovating.
2. Credit scoring has never been a core bank function so Andreessen's comment about this taking place outside of banks makes no sense and raises the question: does he even know what he's talking about?
> “The minute any of these new credit vehicles can show any level of repeatability and reliability, the hedge funds come in and provide the funding. Hedge funds are very comfortable with analytic models. If you have sufficient stability, you can get leverage.”
The hedge funds are looking for yield. Putting aside the fact that a lot of this money is going to dry up when the interest rate environment changes, the hedge funds are less interested in analytic models and more interested in demand. In other words, it's about customer acquisition. An underwriting model alone won't cut it; you need to be able to find the borrowers.
Also, Andreessen seems to be viewing hedge fund participation as some sort of meaningful validation. Looking at hedge fund performance it should be obvious that most hedge funds are not "smart money" by any stretch of the imagination. When the current market turns, a lot of them will lose their shirts just as they did in 2008 betting big on junk like subprime mortgage-backed CDOs.
> “Bitcoin is like technology that’s arrived from Mars, and so regulators don’t know what to do with it. That’s a good thing. What a lot of financial technology entrepreneurs will tell you is that if you’re going to innovate in financial services, you want to do something so new and so different that the existing regulatory system doesn’t know how to react to you. That is your window of opportunity.
This is silly and untrue. In early 2013, FinCEN issued guidance for virtual currencies, and the IRS weighed in earlier this year. Not surprisingly, just because Bitcoin is "new" and "different" doesn't mean that the standard rules and regulations don't apply, as some have learned the hard way[1].
"Debt" in name only, perhaps. You don't need to pay a cent in interest to build up a very good FICO score. A credit card used correctly (paid in full at every statement) is plenty.
I would add that if the "FICO score forced you to take on debt" it would give a better score to someone with a 95% debt to credit line ratio than to an otherwise equal individual with only a 5% debt to credit line ratio. In reality, the opposite is true.
Having said that, for you to have a valid FICO score, you do need to have at least six months of credit history (credit card, loan, mortgage or otherwise).
Traditional credit ratings are "tremendously imprecise"? That's an insane statement. FICO is a proven model and has weathered multiple business cycles. Virtually all of the alternative models being experimented with today haven't. A lot of these models are going to fall apart when the current environment, which has seen record low credit default rates, changes.
I'd love to see a $50,000 auto loan or $400,000 mortgage approved on the basis of eBay purchase history alone.
> “The hypothesis is that there are many other similar sources of consumer data: credit card bills, social-network behavior, potentially even search history. Lots of people, both in the big Internet companies and at start-ups, are trying to get at these large pools of data and figure out new ways to do scoring. What they all have in common is that they are all being done outside of banks.
Notwithstanding the fact that credit payment history ("credit card bills") is factored in to a FICO score, a couple of things should be pointed out:
1. Many of the companies trying different models are doing so in an attempt to serve thin file borrowers. Not surprisingly, established players like Fair Isaac aren't sitting around twiddling their thumbs. They have their own solutions for these borrowers, like the FICO Expansion Score. Just because startups don't want to pay for somebody else's solution doesn't mean they're the only ones innovating.
2. Credit scoring has never been a core bank function so Andreessen's comment about this taking place outside of banks makes no sense and raises the question: does he even know what he's talking about?
> “The minute any of these new credit vehicles can show any level of repeatability and reliability, the hedge funds come in and provide the funding. Hedge funds are very comfortable with analytic models. If you have sufficient stability, you can get leverage.”
The hedge funds are looking for yield. Putting aside the fact that a lot of this money is going to dry up when the interest rate environment changes, the hedge funds are less interested in analytic models and more interested in demand. In other words, it's about customer acquisition. An underwriting model alone won't cut it; you need to be able to find the borrowers.
Also, Andreessen seems to be viewing hedge fund participation as some sort of meaningful validation. Looking at hedge fund performance it should be obvious that most hedge funds are not "smart money" by any stretch of the imagination. When the current market turns, a lot of them will lose their shirts just as they did in 2008 betting big on junk like subprime mortgage-backed CDOs.
> “Bitcoin is like technology that’s arrived from Mars, and so regulators don’t know what to do with it. That’s a good thing. What a lot of financial technology entrepreneurs will tell you is that if you’re going to innovate in financial services, you want to do something so new and so different that the existing regulatory system doesn’t know how to react to you. That is your window of opportunity.
This is silly and untrue. In early 2013, FinCEN issued guidance for virtual currencies, and the IRS weighed in earlier this year. Not surprisingly, just because Bitcoin is "new" and "different" doesn't mean that the standard rules and regulations don't apply, as some have learned the hard way[1].
[1] http://www.usatoday.com/story/news/nation/2014/09/04/bitcoin...