Print

[Staneslav]

a

[Staneslav]

a

[ira01]

Mark,

What do we want to set the time periods as? We need a before and after for prosper's lending guidance. I need to make sure I pick appropriate time periods.

The Improved Bidding Guidance and Portfolio Plans started October 30, 2007.

If I had to pick two periods to compare, I would like to see:

• September 1, 2007 through October 29, 2007
• November 1, 2007 through December 31, 2007

These are too recent to get any useful information.  Loans originated in the second group have only had a maximum of 5 payments due, and a maximum of just 4 payments that might be reported as late (since we don't learn about a late until it is 15 days late), while some loans in this group have only had a mere 2 payments due that might be reported as late (i.e., the loans originated in the last week or so of December could only have had the January and February payments reported as late at this time).  Loans in the first group have had between 4 and 6 payments due that could be reported as late.  Aside from the fact that of course a group of loans that could have 4-6 late payments is going to have more lates than an otherwise identical group that could only have 2-4 late payments simply by virtue of being older, there really won't be any useful late data on these loans until they are older.

[Mark12547]

So, would it be fair to state that preliminary data shows that at least one aspect of the lender experienced has improved: a reduction of the default rate, a result of bidding on higher-quality loans (possibly spurred on by Bidding Guidance and Portfolio Plans) ?

As far as being wrong, don't worry about it; you might be right about other credit grades. (ETA: In IRC, bama is telling me only AA rates had gone up.) I remember a line from "Fletch Lives": It takes a big man to admit he is wrong. I am not a big man.

 

[ira01]

Ok, I'm going to have to withdraw my original assertion.

Preliminary analysis on AA interest rates indicates rates are increasing (not decreasing as I had expected). Therefore, as long as default rates remain constant or better (I doubt they are getting worse), lender ROI on average is higher in time period 2 than time period 1 (according to Mark specs).

I'm big enough to admit I'm wrong.

here's some numerics for you:

Sept -Oct 2007:
# of AA loans: 163
Unweighted avg interest rate: 9.69
Weighted avg interest rate: 10.79

Nov-Dec 2007:
# of AA loans: 239
Unweighted avg interest rate: 9.87
Weighted avg interest rate: 11.19

Its not a huge change, but its enough. If there's enough interest, I will do A and B's later.

I don't think you can draw much in the way of a conclusion based on a 20-40 bp change in one two month period for one credit grade.  Unlike lates, for which you need a significant lag time to conduct analysis, interest rates can be analyzed immediately at the end of a month.  To know whether interest rates really are increasing, you should plot the average interest rate for each month from January 2007 through March 2008, with a separate line for each credit grade.  I think that wouldn't be too difficult, and would be quite informative regarding how interest rates have fluctuated over the last 1.25 years.

[Mark12547]

Maybe I should have been more specific: what has estimated, default-adjusted RIO, in aggregate, been doing before and after October 30, 2007?

Yes, data is still quite thin (not many payments for newer loans) and projecting on just three payments having come due (December loans) doesn't give one confidence in the trends.

[Staneslav]

a

[Staneslav]

a

[ira01]

Well time is and isn't a factor. If we pull too many months we run into an issue of seeing interest rates influenced by other factors (do we want to really look at rates from March 2005?). If we pull the months immediately leading up to the change, and immediately after, we get the clearest picture of the effect.

Additionally, as long as you have enough loans to compare, additional months data only increases the sample size. This is a simple comparison of before and after not a month to month series analysis.

There were plenty of loans in the two months after and the two months before to do a comparison/

You also see a huge change in the amount of AA's funded after the change (indicating a tendency towards higher credit grades along with higher rates).

This is crude, but the numbers are adequate to get an idea of the picture.

Although you certainly might be right, I still think there is insufficient analysis here to know one way or the other.  Prosper is a complicated system, with lots of things going on besides the bidding guidance that could be influencing the results.  For example, have you checked to see if the increase in AA's funded was really the result of a flight to quality as you surmise, or perhaps simply a reflection of a greater number of AA listings (or maybe a greater number of "good" AA listings)?  If there were more AA (or a subset thereof) listings during the second time period than the first, then the higher interest rates could simply reflect the greater demand for lenders' money (since lenders might have had more AA loans to bid on, there would have been less bidding down of rates).  The second time period was also "the holidays" -- perhaps a greater number of listings (for holiday spending, although this would presumably be less true for AA than in the low credit grades), coupled with lenders being busy and thus less lender money available raising rates.  The foregoing is admittedly just speculation.  But I'm sure that with thought, other possible explanations would also present themselves.

[zcommodore]

So, would it be fair to state that preliminary data shows that at least one aspect of the lender experienced has improved: a reduction of the default rate, a result of bidding on higher-quality loans (possibly spurred on by Bidding Guidance and Portfolio Plans) ?

As far as being wrong, don't worry about it; you might be right about other credit grades. (ETA: In IRC, bama is telling me only AA rates had gone up.) I remember a line from "Fletch Lives": It takes a big man to admit he is wrong. I am not a big man.

 

Then I'll do A's and B's later as well.

Save yourself some work and just look at the data already available regarding interest rates on erics site: http://www.ericscc.com/index.php?page=rate_history

[Staneslav]

a