In the past I haven't thought about prepayments much. They happen. We put up with them. In recent months tho, something is changing. Prepayments are increasing. (For the record, worrying about this was Rob L's idea.)
The chart below shows prepayments in the first few months of a loan. Each curve shows how prepayment by certain "age" of the loans varies with vintage. The curves are cumulative. The red "month 1" curve shows the fraction of loans that have prepaid by the end of month 1, so it includes the month 0 prepays. The horizontal axis shows the quarter in which the loan was issued, more or less (1).
Look at the right side of the chart. You can see that in the past nobody ever prepaid during the first month (month 0), but recently 1.5% of borrowers are doing just that. Furthermore, that rate is still climbing. Interestingly, the increase is all in month 0, which is the month that in the past never had any prepays.

I wonder what is going on here. Why do people prepay before the first payment is due? And whatever the reason is, why is it happening now and not in prior years? I have a theory.
My idea is that this is a sign of competition among online lenders. There are now a lot of different online lenders. Suppose you pick one and borrow, but during your investigation you discover all these other guys, and you check the interest rate they offer. Seems likely that quite often you would find another guy who offers you a lower rate. At that point a logical person might borrow from the 2nd guy and pay off the 1st guy.
If I'm right, then the proliferation of online lending is generating competition whose impact we can see. I'm thinking that this is an explicit example of pressure on LC to keep rates low. We lenders tend to ascribe the "bad" rate reductions during the past 2 years to LC & Prosper's lust for growth. Growing competition was likely part of the picture.
There have been a few articles recently about "stacking" by online borrowers. The credit bureaus have noticed more folks borrowing from more than one online lender in rapid succession. The thrust of these articles is always that this is a sign of fraud. Maybe it is. However, credit bureaus, who want to sell new services, seed such articles with the intent to frighten their customers, so we should stay skeptical. Some of these stackers are apparently just repaying their loans with cheaper ones.
(1) footnote: I've recently learned that when LC provides data by "vintage", that they assign the vintages by the date of the first payment, not the date of origination. I learned this only by processing the payment history file. In fact, this probably occurs because some fellow at LC is generating the prepayment data from the payment history file. The payment history file only contains information about payments, not origination. Because LC allows borrowers to set the payment day, and then censors the payment day from the payment history file, there is no way to reconstruct the issue date. All of which is to say that while we think of the vintage quarters (like "15Q4") as being the quarter of loan origination, actually there can be up to 1 month of slop. This may only matter to data nerds who expect consistency between data from multiple files provided by the same company, but don't find it. For example, if you add up the number of loans in a given quarter from the loan history files, they don't match the number of loans in a quarter from the payment history file, because of this slop. Personally, I prefer consistency. Its something you can test for to help ensure that your software isn't fouling up. Of course when the guy upstream from you is completely undisciplined and issues massive data files that are inconsistent by design, then this doesn't work. 