Efficient-market Hypothesis & P2P Lending

Introduction

When I was younger I would always talk about how Apple’s stock was going sky rocket on the day of a release event.  After a few of Apple’s annual product release events I noticed a disappointing trend. The stock never really did anything that day, in fact, sometimes it dropped! This baffled me.  One day I was talking to my friend, let’s say his name is Jeff (although his real name is Erik).  He told me anticipation of the release of some new awesome product was “already priced in” into the stock price.

Jeff would go on about how every piece of public information and even speculation was already built into a stock price, and that price is exactly how it should be priced.  Interesting stuff.

From Wiki:

“In finance, the efficient-market hypothesis (EMH) asserts that financial markets are “informationally efficient”. That is, one cannot consistently achieve returns in excess of average market returns on a risk-adjusted basis, given the information available at the time the investment is made”.

So basically we should stop trying to beat the market because it’ll be a waste of time. In fact there is an entire investment company predicated on this idea, and all they they do is basically buy stocks like VTI that mimic the entire global market.

EMH Example For P2P Loans

I’m going to use a simple example, inquiries in the last six months. For some reason the ROI is almost perfectly negatively correlated. As the inquires go up, the ROI goes down.   For the best returns you should always set this value to 0 on your filter… For now…

What if because of blogs the word gets out that you should always set inquiries to 0. Let’s get a little crazier with the example and say they even start to talk about this on the 10 o’clock news to the point everyone knows this. What would happen? No one would buy notes that have a borrower in excess of 0 inquires. Obviously there will be people with more than 0 inquiries that need money.  What would happen?  The P2P platforms would have to raise interest rates on these to a point where lenders would start to fund them again. You can see here how the market in essence re-priced the loans correctly. Pretty efficient stuff!

Emotion Introduces Inefficiencies

I personally believe EMH contains only half of the equation. The other side is our emotions. Emotions compound the volatility of the markets. How many times did you or someone you know pay too much for a house, an item on eBay or a car.  This is identical to how the stock market works with bid and ask prices. In the last 5 years we’ve also seen swings in excess of 10% as a result. Prosper 1.0 was based on an auction system and did not provide good returns overall.  Personally, I don’t think there was anything wrong with the Prosper 1.0 model, it was simply too small and ended to soon to lower variance enough to be more “efficient”. 

The (Huge) Fixed Rate Advantage

Because both platforms have removed the emotional/auction aspect by setting the rates P2P lending turns out be much more efficient than the stock markets, which shows up in the stable returns.  The the only place inefficiencies show up in masses is on the secondary market. Let’s say my same friend Jeff thinks that P2P market is going to collapse, he can “dump” all of his notes onto the secondary market for a bargain price. Some lucky people will pick up some deals and perhaps beat the “market average” for those loan grades.

There are actually a lot of parallels between stocks and notes. You can think of the P/E ratios, earnings yields, market caps of the Stock Market as being the inquiries, loan grades and sizes of the P2P market (maybe in that exact correlation). All of the filter options we have when picking notes are effectively valuations. That’s where the similarities end.

Prosper and Lending Club set rates putting lenders at a huge advantage over investing in the stock market. This allows us to see how filters impact ROI in an objective manner.  What’s also neat is the LendStats model factors in every loan’s performance so it reflects every available piece of information available to us. How efficient is that!?

Michael

7 thoughts on “Efficient-market Hypothesis & P2P Lending

  1. Prosper 1.0 was failing not because of it’s auction model, but because Prosper loans were defaulting at a rate that was much higher then Prosper’s projections. Since the loans originated at rates that factored in the projected default rates at the time of the bidding that left most early investors with (large) losses when years later the real default rate was know.

  2. Excellent article.  There are varying forms of EMH, but most “weaker” forms tend to boil down to the idea that easy profits are hard to find or likely fleeting due to the pressures of competition.  It’s not to say that there aren’t occasional opportunities.  

    Rather than arguing that markets are perfectly informed, frictionless, rational, liquid, tax-free, universally accessible, etc, we instead can observe that most opportunities to exploit market inefficiencies are themselves scarce competitive.  (E.g., you can pay for low-latency, direct market access to some securities exchanges in order to profit from more recent information and faster trade execution, but there rather few seats at that level and the cost of access is generally quite high).

    I agree that we’ll see greater competition on the lending platforms. However, the P2P lending sites are really competing with other institutional lenders.  P2P lending would have to grow by at least a few orders of magnitude before you have retail banks truly worried about “disintermediation risks” (being cut out as the middle-man between lenders and borrowers).

    I don’t think the profitability is simply from the P2P sites keeping rates high. It’s also from the fact that even such high rates are still competitive with retail banking rates for many borrowers.  We’re far from saturating the market.

    If the P2P sites don’t allow for auction rate loans at some point, then you’ll have competition to be first in line to buy out the preferable loans and then resell them in the secondary market for a competitive rate (like a ticket scalper).  In that case, the lenders with faster/better/cheaper analytics, transaction costs, costs of capital, and market access will tend to dominate.

    Two things that worry me:

    1. If they don’t return to auction rate loans, then the most fearsome competitor in these markets are the P2P systems themselves.  Do they participate in their own markets?  If so, what’s to keep them from consistently picking off the best loans at a high rate before they get to market?

    2.  As the whole ecosystem scales, you’ll probably see some consolidation among lenders with better loan selection processes and many borrowers lending indirectly through them to mitigate their risks (i.e., banks).  This tends to be a natural pattern due to its efficiency.  If we start to see large systematized lenders on P2P sites, how long will it be before they are regulated, forcing greater consolidation, then regulated more due to the consolidation, and so on until we’re back to where we started before P2P lending?

    1. Brett,
      Thanks for the thoughtful response, you really covered a lot.  I’m not sure the auction system is required for P2P to survive.  In auction based system is there is always “the winner’s curse”.  If you won, you most likely paid too much.

      I prefer the fixed rate, the flip side is that I need to trust the p2p lenders are setting rates in a way that protect me and reward the proportional amount of risk associated.

      This comment of your intrigus me “you can pay for low-latency, direct market access to some securities
      exchanges in order to profit from more recent information and faster
      trade execution, but there rather few seats at that level and the cost
      of access is generally quite high”.  I think you are absolutely right. If all stock data was public knowledge, i.e. I could download a database every night with 30 years of financial history and all financial data for that day I could make much better choices on stocks that just reading the headlines to find a solid company or just plugging away one symbol at a time looking at ROC and earning yields, etc….

      1. And if that were cheaply available, then the more of the market would be competitively trading upon it… right?

        Historical Stock Prices: finance.yahoo.com .  Just use a data downloader to pull the .CSV files programatically.  They are pretty nice about it too.

        Historical Financial Statements: The EDGAR database has all filings available, but they are hard to pull out.  Most sites won’t give more than 4 qtrs (or charge >$1000 a month if you want a deeper data subscription). If you’re feeling cheap/lazy, ADVFN.com has one of the most comprehensive free financial data collections and it’s nicely organized for scraping.

        Economic Data: The St. Louis Fed’s FRED database is free and fantastic.

        Again though, you can just pay a few thousand $ a month for a Bloomberg terminal or Reutrers’ data API…  or you can engage in some home-built financial data integration work. The data from these sources can be used profitably, but the cost to every firm to hire a staff of professionals to design, maintain, and operate rigorous and reliable models  can be astronomical.

        Competition allows for efficiency so long as people allow for true competition.  For example, OpenTick.com had provided streaming access to low-level tick-by-tick stock data for a few years until they were shut down doing so without the proper licenses.  Is the market inefficient due to the costs and restrictions on the flow of market information, or is it efficient in that the value of such information is priced accordingly?

        It’s all bit of a zero-sum game; Wherever financial markets are inefficient, someone making an excess profit has an interest in keeping them that way — whether through regulation, contract, or crook.  Competing with entrenched institutions can be challenging.

        P2P lending is one of the few true financial innovations that’s competitive, potentially revolutionary, and fundamentally valuable in its ability to cut out the layers of middlemen  (“financial intermediaries”) and salesmen (financial planners are often paid on commission for selling investments in funds). 

        More importantly, transparent debt servicing, open data, and open market access like we see in P2P markets is completely unprecedented. No bank would share such information — it’s too valuable.  Right now you have better access to data on credit and loan standings of hundreds of thousands of people via Prosper and Lending Club.  The corporate bond market is many orders of magnitude large and has been around for some time, but good luck getting comparable bond data for less than a few thousand $ a month.

        -Brett

    2. “If the P2P sites don’t allow for auction rate loans at some point, then you’ll have competition to be first in line to buy out the preferable loans and then resell them in the secondary market for a competitive rate (like a ticket scalper).”
      I doubt we’ll ever get to a point where you’ll be able to predict a default down to this level with the data available. A slice of data from a point in time plus monthly Fico can’t predict divorces, medical issues, etc. 3 or4 years down the road.

      “In that case, the lenders with faster/better/cheaper analytics, transaction costs, costs of capital, and market access will tend to dominate.” I think that would mean a lot of cash had to be coming in to LC to corner the market like that as LC would be adjusting rates to compensate.
      Lou

Leave a Reply