Tuesday, November 12, 2019

lendingclub diversification benefit and required yield

This is a continuation of previous post.  I am trying to measure and put a price on concentrated loans.
In a nutshell, I'll put the price based on Value At Risk (VAR).  The idea is to require extra yield so that VAR of my portfolio after buying a large loan is the same of buying same principal amount yet more individual loans.
Suppose I have 1000 active loans of various grades.  From historical ROIs of different ratings, I can simulate 1000 trials of potential outcomes.  Now I face the choice of investing 10 times of my average loan size amount to one loan vs. investing 10 equal sized loans.  I can again simulate 1000 trails of the two choices.  Combining these two set of trails with my current portfolio trails, I get distributions of my two choices.  Then, I compare my 5, 10 percentiles etc.  In my example, investing 10 equal sized loans will produce a larger 5 percentile; the extra discount I require to invest one loan would be the difference.
An example of my simulation:
for single A grade loan size of 5, 10, 20, 50, 100 and 1000 size of my average loan, the 5 percentile (95% VAR) differences would be 9.7%, 2.5%, 4.8%, 6.7%, 1.3% and 18.6%.  The extra discounts should increase with increasing size, but in this case, 1000 simulation is not enough and the fluctuation is due to statistic noise.

There are many assumptions in this simulation
  • future return distribution is similar to historical ones
  • individual loans are uncorrelated
  • The outcome is statistic based. one loan may not produce the predicted outcome
  • I haven't consider the extra potential upside of concentration

No comments:

Post a Comment