7 ms·

This could be a chicken and egg problem, too. Banks lend on statistics, just as insurance companies have rates based upon the same. And insurance companies, f

This could be a chicken and egg problem, too. Banks lend on statistics, just as insurance companies have rates based upon the same. And insurance companies, for the most part, can assign rates based upon average accident rates, for identifiable traits. Like one's sex. Or where they live (a common way to get statistics without 'race/colour'). So what do I mean by the chicken and the egg, here? Well, imagine that black landowners go bankrupt more often. And sadly, this could be because black landowners go bankrupt more often! If a specific statistical classification indicates more bankruptcy, then it becomes true. For example, bank $a is bold enough to lend to someone whom they view as, statistically, more likely to fail. Then.. a few years later? That person needs a fiscal injection. Were that person not black, it would be easier to get funding at a time of need. However, since that person is? Well, then they get more denials, go bankrupt, and add to the proof that "statistically, more black people go bankrupt". Or as the banks are likely doing it, "the area where people live in". In other words, unless you can make EVERYONE, 100% not deny based upon an statistical models? Race/gender/sexual orientation/sex/where they live, etc, etc? Anyone, like bank $a lending outside of those models, will be lending to people less able to find additional assistance when/if the time comes. Chicken and egg. It's not right, but I view this a little bit like certain regulations, where corporations can't compete against other corporations which 'cut corners', so we have to mandate all corps behave in a certain manner to level the playing field.

macspoofing 6y ago
There is no 'chicken and egg' problem. Banks don't use race as criteria for loans, just as auto insurance companies don't use race as criteria for setting rates. I also find perspectives like yours very perplexing. Debt is not charity. It is not charitable to provide debt to someone who cannot afford it. Not for the debtor, nor the lender. I've seen people suffer for years drowning in debt they could not afford. I've seen people with mortgages (variable rates in foreign currency) that went underwater when the rates and currency exchange shifted the wrong way. Bankruptcy is an incredibly painful process to go through as well. Debt is not charity. Banks also do not just give out big loans (no matter what your credit score is) unless there is collateral. It is clear as day, that the bank valued the collateral to be low or inconsequential (i.e. judged one property at land-value at $1/sqft and $0 for the building). So there was no collateral for those loans, because redevelopment of those properties was expensive, with lots of risk. Why do you think it hasn't happened yet? Why do you think these properties were standing there rotting away? Why didn't anybody take a flyer on those, given how inexpensive these properties probably are? Because there are major risks associated with redevelopment. In our city, a property, seemingly in a prime location, stood abandoned for years .. turns out, the property was cheap, the land was cheap, the environmental cleanup necessary for redevelopment, was not. >Were that person not black, it would be easier to get funding at a time of need Says who!? Have you actually tried? Our company was only extended a line of credit years after we were in business, and it was collateralized against purchase orders and revenue (and we had to provide lots of documentation to prove those). Banks do not want to give out loans with even a little bit of risk associated. They are not like venture capital firms and haven't been for a long time.
- gwd 6y ago
  > Banks don't use race as criteria for loans, just as auto insurance companies don't use race as criteria for setting rates. If this article is correct, then this claim is wrong. Banks may not directly use race as a criteria, but the claim is that they use race as a criteria for property value, and then use property value as a criteria for loans: > Five of his properties were in a decent state of repair and some had sitting tenants, while three needed a complete refurbishment. ...One bank finally sent someone on location to assess his properties 12 weeks later. They valued the buildings at zero dollars and just over a dollar per square foot for the land only. As the rules for most banks say a property must be valued at a minimum of $50,000 (£39,000) to get a line of credit against it, his case was shut. Counting a building with active tenants paying rent as being valued at $0 is ridiculous.
  - MattGaiser 6y ago
    > Counting a building with active tenants paying rent as being valued at $0 is ridiculous. Depends on how much rent they are paying, what other large costs repair costs might exist for the building, and whether anyone is buying in that neighbourhood.
  - macspoofing 6y ago
    >Banks may not directly use race as a criteria, but the claim is that they use race as a criteria for property value, and then use property value as a criteria for loans ... You have to do the leg-work to show that Banks have a policy that is in intent racist because every policy you can ever think of will have a differential impact on any group you create - because groups will never be perfectly uniform. >Counting a building with active tenants paying rent as being valued at $0 is ridiculous. Not necessarily. This is why these articles are so terrible. They didn't do investigative journalism. They didn't do the legwork. I can tell you right now, there is a reason why it is valued at $0 - we just didn't get that reason because the journalist is lazy.
    - bluntfang 6y ago
      Redlining is a real practice done by banks and credit issuers that use race criteria based on where you live to discriminate against people of color. https://en.wikipedia.org/wiki/Redlining https://en.wikipedia.org/wiki/Redlining
chii 6y ago
> imagine that black landowners go bankrupt more often. And sadly, this could be because black landowners go bankrupt more often! But to fix this is "simple" - lend to those higher risk businesses at a higher rate. If those businesses are actually _not_ in reality more likely to fail, then the lender makes more money. As lenders are able to make more money, there should be more competition to lend to those risky-seeming businesses which isn't actually risky. And competition would then drive the rates back down. Therefore, leading to an equilibrium rate.
- bluGill 6y ago
  I assume by higher rate you mean interest rates. (as opposed to lend to more people at the same interest rates) those higher rates themselves cost more and in turn make it more likely that they go bankrupt. The higher interest rates are (if set right) enough to recover the from the bankruptcy that happens.
- DoofusOfDeath 6y ago
  If Rice's risk/reward assessment is correct, it seems strange to not have even one lender offer a loan that aligns with his analysis. I have no clue about how prevalent racism is in the lending institutions that serve that geography. But I'd expect at least one of them to have looked at this carefully, in order to maximize the value of their portfolios.
rocqua 6y ago
I don't think this needs to be such a simple chicken and egg problem. Because, even if black landowners are say 50% more likely to go bankrupt, they could at the very least, still get some loans. Just maybe not as high a loan. It's not like banks have to say 'yes/no' to customers. They have variously strong variants of 'yes' when someone asks for credit. I think there are many other cases (e.g. affluent people choosing a public school / area to live in) where this problem is much more prevalent, because the only options in the decision are 'yes / no'. And since there is a correlation between black neighborhoods / public schools and crime / worse results, it makes sense for these people to avoid the mentioned downsides. Thing is, those correlations aren't causal. It's just that being black is correlated to being poor for historic reasons, being poor causes the downsides. And the above chicken / egg problem exacerbates the problem.
- throwaway0a5e 6y ago
  An interest rate that makes whatever you're doing with the loan not worth doing is effectively a "no" If you're gonna develop properties and the up front investment looks like it will take 10yr to pay off vs 5yr to pay off then that severely constrains the kind of development you can justify doing. You'll have to pick something safer which isn't necessarily the kind of development you want do to do. It's not banks and interest rates that are at fault here. All sorts of local permitting BS, laws about professional certifications, etc, etc, add up and compound each other to create a business environment where two development options are "tear down historic building and build spec luxury condos" or "tear down historic buildings and build a big box store". And this business environment feeds itself as developers, banks, regulators, construction firms, etc, etc. get well practiced at it making them more hesitant to do business with people who want to do any other kind of development thereby raising costs (supply and demand). Frankly I think this is all BS that could largely be solved if government got out of (or at least less involved in) the business of infringing on people's property rights.
bluGill 6y ago
There are insurance companies that specialize in insuring weird one off things. They can figure out risk even if nothing like it has been done before.
op03 6y ago
Well this is why we need more Dumb Bankers. It would work out better than bailing out the mindlessly ambitious pseudosmart ones every 2 years. Nature has provided us a surplus. We must use it.
AnthonyMouse 6y ago
> If a specific statistical classification indicates more bankruptcy, then it becomes true. Fortunately this isn't actually the chicken and egg problem you think it is. Suppose that for defunct historical reasons the model is predicting a higher default rate for some population, so it says to approve 85% of their loan applications rather than 95%. Then half of those people get into trouble that getting a second loan would save them from, but because of the model, only 85% rather than 95% are given the other loan. So then their repayment rate will be 90% rather than 95%, because not as many were saved by the loan, but wait -- 90% is more than 85%, because only half of them needed the second loan. So now the model starts predicting that it should expect a 90% repayment rate for that population, which then increases their actual repayment rate, and so on until it stabilizes at 95%. If you could identify the exact degree to which the prediction is affecting itself then you could take it all out at once, but doing that is hard. If the evidence shows a lower repayment rate in a particular area, is that just the holdover, or is it because that area still has a present-day higher level of drug use and violent crime which negatively affects repayment? If it's the latter and you give them loans at scale under the assumption that it's just an artifact, the bank is going bust. So unless you can disentangle it all, the data to do which is often not reasonably available, the only realistic option is waiting for it to filter through.
cwhiz 6y ago
What you’re suggesting is illegal and has been illegal for quite some time. Credit policies have to be neutral in outcome. You can’t just cook up a clever way to be racist and hide it in some zip code algorithm. If your algorithm turns out to weigh against a certain people, it’s illegal. Even if it happened by accident, it’s illegal.
reportgunner 6y ago
Well why don't black landowners simply not go bankrupt and change the statistic over the years ? Why don't black landowners support each other to make it easier for them to get a fiscal injection when needed ? > He had enough cash to buy eight buildings, all around 100 years old, in his majority black neighbourhood of Ensley, but he needed a banking loan to redevelop them. He thought it would be just a formality. Why did he not buy 4 building instead not needing the loan ? Something is fishy here.
- kiplkipl 6y ago
  Meanwhile the buyer in the white-majority area continues to buy 8, enjoys a higher leverage for the same interest, and makes a higher profit. And so it repeats. Leverage is forgiven when the faceless metaphorical person does it but, when it's given a black face, suddenly the black community as a whole has to answer to you? Why didn't GM simply not go bankrupt?
  - reportgunner 6y ago
    What interest are you talking about? The article said he had money to pay for 8 buildings in cash and he wanted to loan to renovate them. I suggested that if you can't get a loan you should tune down your ambitions a bit. There are many white people who can't get a loan too, tune it down with the racist us vs them thinking. It's not like all white people gain from that one person getting more leverage. I was talking about doing this to remove the statistics that cause the apparent racism, while you want to keep perpetuating the same mentality. Wake up or keep wondering why you are still losing.
    - kiplkipl 6y ago
      I know what the article says. A person who can't get a mortgage on 8 won't be able to get a mortgage on 4 at the same rate as another person who can get a mortgage on 8. >I suggested that if you can't get a loan you should tune down your ambitions a bit. This is not about the ambitions of the individial in the link, it's an issue of structurally unequal differences that make it harder for black areas to see investment. >tune it down with the racist us vs them thinking. It's not like all white people gain from that one person getting more leverage. >I was talking about doing this to remove the statistics that cause the apparent racism, while you want to keep perpetuating the same mentality. Wake up or keep wondering why you are still losing. Please stop telling me what I'm thinking.
vladTheInhaler 6y ago
I saw a comment on HN a while ago that this reminded me of. It made the argument that the primary value of a lot of machine learning models is essentially to "launder" information about protected classes so that it becomes harder to discover when you have been discriminated against, and harder to assign blame to a specific entity when that's the case. For instance, using a complicated nonlinear combination of address (e.g. predominantly black neighborhoods), educational history (e.g. historically black colleges), work history, credit rating, and who knows what else, might simply allow a model to recover a feature (ethnicity) which is correlated with likelihood to repay a loan in historical data. Like you pointed out, that would lock us into a positive feedback loop where people are being denied loans based on "statistical, race-blind models based only on data", but the data only looks that way because of existing racism, which is now being further enshrined and legitimized. But the thorniest issue is that the model reflects an actual, true, relationship in the world, so asking banks to ignore it would be asking them to leave money on the table until such a time as society has eliminated the structural inequalities that cause those correlations. How do we get them to do that, instead of just engaging in more and more sophisticated data laundering like above? Would we have to create an approved list of features that can be used for loan approval, which provably cannot be used to discover race? How would we even know that such a rule was working?

macspoofing 6y ago

There is no 'chicken and egg' problem. Banks don't use race as criteria for loans, just as auto insurance companies don't use race as criteria for setting rates. I also find perspectives like yours very perplexing. Debt is not charity. It is not charitable to provide debt to someone who cannot afford it. Not for the debtor, nor the lender. I've seen people suffer for years drowning in debt they could not afford. I've seen people with mortgages (variable rates in foreign currency) that went underwater when the rates and currency exchange shifted the wrong way. Bankruptcy is an incredibly painful process to go through as well. Debt is not charity. Banks also do not just give out big loans (no matter what your credit score is) unless there is collateral. It is clear as day, that the bank valued the collateral to be low or inconsequential (i.e. judged one property at land-value at $1/sqft and $0 for the building). So there was no collateral for those loans, because redevelopment of those properties was expensive, with lots of risk. Why do you think it hasn't happened yet? Why do you think these properties were standing there rotting away? Why didn't anybody take a flyer on those, given how inexpensive these properties probably are? Because there are major risks associated with redevelopment. In our city, a property, seemingly in a prime location, stood abandoned for years .. turns out, the property was cheap, the land was cheap, the environmental cleanup necessary for redevelopment, was not. >Were that person not black, it would be easier to get funding at a time of need Says who!? Have you actually tried? Our company was only extended a line of credit years after we were in business, and it was collateralized against purchase orders and revenue (and we had to provide lots of documentation to prove those). Banks do not want to give out loans with even a little bit of risk associated. They are not like venture capital firms and haven't been for a long time.

gwd 6y ago
> Banks don't use race as criteria for loans, just as auto insurance companies don't use race as criteria for setting rates. If this article is correct, then this claim is wrong. Banks may not directly use race as a criteria, but the claim is that they use race as a criteria for property value, and then use property value as a criteria for loans: > Five of his properties were in a decent state of repair and some had sitting tenants, while three needed a complete refurbishment. ...One bank finally sent someone on location to assess his properties 12 weeks later. They valued the buildings at zero dollars and just over a dollar per square foot for the land only. As the rules for most banks say a property must be valued at a minimum of $50,000 (£39,000) to get a line of credit against it, his case was shut. Counting a building with active tenants paying rent as being valued at $0 is ridiculous.
- MattGaiser 6y ago
  > Counting a building with active tenants paying rent as being valued at $0 is ridiculous. Depends on how much rent they are paying, what other large costs repair costs might exist for the building, and whether anyone is buying in that neighbourhood.
- macspoofing 6y ago
  >Banks may not directly use race as a criteria, but the claim is that they use race as a criteria for property value, and then use property value as a criteria for loans ... You have to do the leg-work to show that Banks have a policy that is in intent racist because every policy you can ever think of will have a differential impact on any group you create - because groups will never be perfectly uniform. >Counting a building with active tenants paying rent as being valued at $0 is ridiculous. Not necessarily. This is why these articles are so terrible. They didn't do investigative journalism. They didn't do the legwork. I can tell you right now, there is a reason why it is valued at $0 - we just didn't get that reason because the journalist is lazy.
  - bluntfang 6y ago
    Redlining is a real practice done by banks and credit issuers that use race criteria based on where you live to discriminate against people of color. https://en.wikipedia.org/wiki/Redlining https://en.wikipedia.org/wiki/Redlining

chii 6y ago

> imagine that black landowners go bankrupt more often. And sadly, this could be because black landowners go bankrupt more often! But to fix this is "simple" - lend to those higher risk businesses at a higher rate. If those businesses are actually _not_ in reality more likely to fail, then the lender makes more money. As lenders are able to make more money, there should be more competition to lend to those risky-seeming businesses which isn't actually risky. And competition would then drive the rates back down. Therefore, leading to an equilibrium rate.

bluGill 6y ago
I assume by higher rate you mean interest rates. (as opposed to lend to more people at the same interest rates) those higher rates themselves cost more and in turn make it more likely that they go bankrupt. The higher interest rates are (if set right) enough to recover the from the bankruptcy that happens.
DoofusOfDeath 6y ago
If Rice's risk/reward assessment is correct, it seems strange to not have even one lender offer a loan that aligns with his analysis. I have no clue about how prevalent racism is in the lending institutions that serve that geography. But I'd expect at least one of them to have looked at this carefully, in order to maximize the value of their portfolios.

rocqua 6y ago

I don't think this needs to be such a simple chicken and egg problem. Because, even if black landowners are say 50% more likely to go bankrupt, they could at the very least, still get some loans. Just maybe not as high a loan. It's not like banks have to say 'yes/no' to customers. They have variously strong variants of 'yes' when someone asks for credit. I think there are many other cases (e.g. affluent people choosing a public school / area to live in) where this problem is much more prevalent, because the only options in the decision are 'yes / no'. And since there is a correlation between black neighborhoods / public schools and crime / worse results, it makes sense for these people to avoid the mentioned downsides. Thing is, those correlations aren't causal. It's just that being black is correlated to being poor for historic reasons, being poor causes the downsides. And the above chicken / egg problem exacerbates the problem.

throwaway0a5e 6y ago
An interest rate that makes whatever you're doing with the loan not worth doing is effectively a "no" If you're gonna develop properties and the up front investment looks like it will take 10yr to pay off vs 5yr to pay off then that severely constrains the kind of development you can justify doing. You'll have to pick something safer which isn't necessarily the kind of development you want do to do. It's not banks and interest rates that are at fault here. All sorts of local permitting BS, laws about professional certifications, etc, etc, add up and compound each other to create a business environment where two development options are "tear down historic building and build spec luxury condos" or "tear down historic buildings and build a big box store". And this business environment feeds itself as developers, banks, regulators, construction firms, etc, etc. get well practiced at it making them more hesitant to do business with people who want to do any other kind of development thereby raising costs (supply and demand). Frankly I think this is all BS that could largely be solved if government got out of (or at least less involved in) the business of infringing on people's property rights.

bluGill 6y ago

There are insurance companies that specialize in insuring weird one off things. They can figure out risk even if nothing like it has been done before.

op03 6y ago

Well this is why we need more Dumb Bankers. It would work out better than bailing out the mindlessly ambitious pseudosmart ones every 2 years. Nature has provided us a surplus. We must use it.

AnthonyMouse 6y ago

> If a specific statistical classification indicates more bankruptcy, then it becomes true. Fortunately this isn't actually the chicken and egg problem you think it is. Suppose that for defunct historical reasons the model is predicting a higher default rate for some population, so it says to approve 85% of their loan applications rather than 95%. Then half of those people get into trouble that getting a second loan would save them from, but because of the model, only 85% rather than 95% are given the other loan. So then their repayment rate will be 90% rather than 95%, because not as many were saved by the loan, but wait -- 90% is more than 85%, because only half of them needed the second loan. So now the model starts predicting that it should expect a 90% repayment rate for that population, which then increases their actual repayment rate, and so on until it stabilizes at 95%. If you could identify the exact degree to which the prediction is affecting itself then you could take it all out at once, but doing that is hard. If the evidence shows a lower repayment rate in a particular area, is that just the holdover, or is it because that area still has a present-day higher level of drug use and violent crime which negatively affects repayment? If it's the latter and you give them loans at scale under the assumption that it's just an artifact, the bank is going bust. So unless you can disentangle it all, the data to do which is often not reasonably available, the only realistic option is waiting for it to filter through.

cwhiz 6y ago

What you’re suggesting is illegal and has been illegal for quite some time. Credit policies have to be neutral in outcome. You can’t just cook up a clever way to be racist and hide it in some zip code algorithm. If your algorithm turns out to weigh against a certain people, it’s illegal. Even if it happened by accident, it’s illegal.

reportgunner 6y ago

Well why don't black landowners simply not go bankrupt and change the statistic over the years ? Why don't black landowners support each other to make it easier for them to get a fiscal injection when needed ? > He had enough cash to buy eight buildings, all around 100 years old, in his majority black neighbourhood of Ensley, but he needed a banking loan to redevelop them. He thought it would be just a formality. Why did he not buy 4 building instead not needing the loan ? Something is fishy here.

kiplkipl 6y ago
Meanwhile the buyer in the white-majority area continues to buy 8, enjoys a higher leverage for the same interest, and makes a higher profit. And so it repeats. Leverage is forgiven when the faceless metaphorical person does it but, when it's given a black face, suddenly the black community as a whole has to answer to you? Why didn't GM simply not go bankrupt?
- reportgunner 6y ago
  What interest are you talking about? The article said he had money to pay for 8 buildings in cash and he wanted to loan to renovate them. I suggested that if you can't get a loan you should tune down your ambitions a bit. There are many white people who can't get a loan too, tune it down with the racist us vs them thinking. It's not like all white people gain from that one person getting more leverage. I was talking about doing this to remove the statistics that cause the apparent racism, while you want to keep perpetuating the same mentality. Wake up or keep wondering why you are still losing.
  - kiplkipl 6y ago
    I know what the article says. A person who can't get a mortgage on 8 won't be able to get a mortgage on 4 at the same rate as another person who can get a mortgage on 8. >I suggested that if you can't get a loan you should tune down your ambitions a bit. This is not about the ambitions of the individial in the link, it's an issue of structurally unequal differences that make it harder for black areas to see investment. >tune it down with the racist us vs them thinking. It's not like all white people gain from that one person getting more leverage. >I was talking about doing this to remove the statistics that cause the apparent racism, while you want to keep perpetuating the same mentality. Wake up or keep wondering why you are still losing. Please stop telling me what I'm thinking.

vladTheInhaler 6y ago

I saw a comment on HN a while ago that this reminded me of. It made the argument that the primary value of a lot of machine learning models is essentially to "launder" information about protected classes so that it becomes harder to discover when you have been discriminated against, and harder to assign blame to a specific entity when that's the case. For instance, using a complicated nonlinear combination of address (e.g. predominantly black neighborhoods), educational history (e.g. historically black colleges), work history, credit rating, and who knows what else, might simply allow a model to recover a feature (ethnicity) which is correlated with likelihood to repay a loan in historical data. Like you pointed out, that would lock us into a positive feedback loop where people are being denied loans based on "statistical, race-blind models based only on data", but the data only looks that way because of existing racism, which is now being further enshrined and legitimized. But the thorniest issue is that the model reflects an actual, true, relationship in the world, so asking banks to ignore it would be asking them to leave money on the table until such a time as society has eliminated the structural inequalities that cause those correlations. How do we get them to do that, instead of just engaging in more and more sophisticated data laundering like above? Would we have to create an approved list of features that can be used for loan approval, which provably cannot be used to discover race? How would we even know that such a rule was working?