3 ms·

Their data comes from loan applications. Talk about selection bias ...

sidlls 9y ago
Not that I disagree with the more general point that a loan provider's data isn't necessarily the best, but in what way do you suggest there is selection bias here?
- Simon_says 9y ago
  My thinking is that the population of people who are applying for loans in any given month is going to be drastically different than the population who are not applying for loans in exactly the metric, personal finances, that Pricenomics wants to look at. Kudos on Pricenomics for being so upfront about their data, but I wouldn't have titled it "How Much Are People Making from the Sharing Economy?", because it's not. It's more like "How Much Are Loan Applicants Making from the Sharing Economy?"
  - sidlls 9y ago
    Different in what way? Just for example, during the housing bubble of the 2000s, people in any given slice of "personal finances" one cared to measure were applying for loans (purchases and lines of credit). I suspect the people who are in the gig economy are more likely to be in a class that has lower education, income, and assets. In that sense I'm not sure "personal finances" as a selection bias would apply (that is, it seems likely as a rule gig economy loan applicants might resemble gig economy workers to a high degree).
    - nostrademons 9y ago
      There's a strong inverse correlation between how much you make and how much you borrow. People who are making enough to cover their daily expenses don't need to borrow, by definition, and so they tend to be underrepresented in loan origination statistics. Similarly, people who are making very little money tend to need cash, which makes them more likely to seek out loans, because the very product a loan is selling is cash (today, in exchange for more cash tomorrow). While its true that during the housing bubble, people who took out loans included people from all different slices of the income distribution, the borrowers were disproportionately drawn from poorer income segments. That's why they called it a "subprime" crisis; the percentage of the mortgage market lent to people who wouldn't otherwise qualify for prime loans was much higher than normal, and predictably, these loans ended up defaulting at a higher rate.
- Axsuul 9y ago
  What he means is that typically the people who apply for loans are of the lower income bracket. You won't find higher income generators applying for these types of loans, therefore, selection bias.
cakedoggie 9y ago
There date is from a source that exists, as opposed to not having data. It is a strange complaint to make, unless you know of a better source of data.
- _delirium 9y ago
  The ol' lamp-post theory of data analysis: https://en.wikipedia.org/wiki/Streetlight_effect https://en.wikipedia.org/wiki/Streetlight_effect
- Simon_says 9y ago
  I don't need to have better ideas to point out flaws. But in this case I do know who has a better source of data: AirBnB, DoorDash, Etsy, Fiverr, Getaround, Lyft, Postmates, Taskrabbit, and Uber.

Not that I disagree with the more general point that a loan provider's data isn't necessarily the best, but in what way do you suggest there is selection bias here?

Simon_says 9y ago
My thinking is that the population of people who are applying for loans in any given month is going to be drastically different than the population who are not applying for loans in exactly the metric, personal finances, that Pricenomics wants to look at. Kudos on Pricenomics for being so upfront about their data, but I wouldn't have titled it "How Much Are People Making from the Sharing Economy?", because it's not. It's more like "How Much Are Loan Applicants Making from the Sharing Economy?"
- sidlls 9y ago
  Different in what way? Just for example, during the housing bubble of the 2000s, people in any given slice of "personal finances" one cared to measure were applying for loans (purchases and lines of credit). I suspect the people who are in the gig economy are more likely to be in a class that has lower education, income, and assets. In that sense I'm not sure "personal finances" as a selection bias would apply (that is, it seems likely as a rule gig economy loan applicants might resemble gig economy workers to a high degree).
  - nostrademons 9y ago
    There's a strong inverse correlation between how much you make and how much you borrow. People who are making enough to cover their daily expenses don't need to borrow, by definition, and so they tend to be underrepresented in loan origination statistics. Similarly, people who are making very little money tend to need cash, which makes them more likely to seek out loans, because the very product a loan is selling is cash (today, in exchange for more cash tomorrow). While its true that during the housing bubble, people who took out loans included people from all different slices of the income distribution, the borrowers were disproportionately drawn from poorer income segments. That's why they called it a "subprime" crisis; the percentage of the mortgage market lent to people who wouldn't otherwise qualify for prime loans was much higher than normal, and predictably, these loans ended up defaulting at a higher rate.
Axsuul 9y ago
What he means is that typically the people who apply for loans are of the lower income bracket. You won't find higher income generators applying for these types of loans, therefore, selection bias.

cakedoggie 9y ago

There date is from a source that exists, as opposed to not having data. It is a strange complaint to make, unless you know of a better source of data.

_delirium 9y ago
The ol' lamp-post theory of data analysis: https://en.wikipedia.org/wiki/Streetlight_effect https://en.wikipedia.org/wiki/Streetlight_effect
Simon_says 9y ago
I don't need to have better ideas to point out flaws. But in this case I do know who has a better source of data: AirBnB, DoorDash, Etsy, Fiverr, Getaround, Lyft, Postmates, Taskrabbit, and Uber.