SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
cdery6
Topic Author
Posts: 3
Joined: April 9th, 2015, 9:00 pm

Valuation of illiquid bonds

April 13th, 2015, 8:58 pm

Hi All !I am currently encountering a mathematical issue at work. It may sound very easy to some with quants background but here it is:I am trying to value 50 North-American private corporate bonds that are illiquid. I thought about using comparables or other market data in order to price them.Here?s my failed attempt: I tried to use the yield change of the bond?s benchmark & the duration & convexity of the private bond to estimate the period price change and add to the previous value to get an estimated price. My issue is that a typical yield change of a benchmark is not at all similar to that of a private bond and so I would need to put something that makes sense in my calculation so that it converts the benchmark yield to the probable yield change of the actual security. Then, using that estimated period yield change, the duration & convexity of the private bond, estimate the price the bond (?)Would some sort of a matrix pricing be a good idea? I pulled of a credit sector spread matrix but it only adds to the benchmark, it does not seem to compensate for the larger yield change that risker bonds seem to have when its benchmark moves.Lastly, I have MAX 50bps threshold for price accuracy ? although I really welcome your input even if you are not sure if your method would meet that criterion.Thanks a lot for your input!
 
User avatar
Alan
Posts: 10318
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Valuation of illiquid bonds

April 14th, 2015, 12:00 pm

Since nobody has answered, two obvious questions: 1. since your first attempt was to make adjustments based upon a "previous value", how was the latter done?2. what does "value" mean in a deeply illiquid market? (Intuitively, the best that could be done would result in a huge range like 40-60 or some such,so the midpoint estimated error would be, say, 20% rather than your required 0.5%)
Last edited by Alan on April 13th, 2015, 10:00 pm, edited 1 time in total.
 
User avatar
cdery6
Topic Author
Posts: 3
Joined: April 9th, 2015, 9:00 pm

Valuation of illiquid bonds

April 14th, 2015, 9:27 pm

Hi Alan,Here are my answers:1. since your first attempt was to make adjustments based upon a "previous value", how was the latter done?If you mean price, it was priced using bloomberg2. what does "value" mean in a deeply illiquid market? (Intuitively, the best that could be done would result in a huge range like 40-60 or some such,so the midpoint estimated error would be, say, 20% rather than your required 0.5%)I don't think my bonds will be "deeply illiquid" the 5 bps margin of error would be a daily margin of error
 
User avatar
bearish
Posts: 5790
Joined: February 3rd, 2011, 2:19 pm

Valuation of illiquid bonds

April 14th, 2015, 10:01 pm

I'll continue in Alan's footsteps. When you say "priced using bloomberg", that is not unlike saying priced using a calculator, pencil and paper, or the like. Do you mean you looked up an actual quote on Bloomberg, or a BVAL estimate, or valued a private security with some spread to a particular curve, or some other method that is supported? Also, do you mean 5 or 50 bps? And do you plan to "true up" your estimate now and then?
 
User avatar
cdery6
Topic Author
Posts: 3
Joined: April 9th, 2015, 9:00 pm

Valuation of illiquid bonds

April 15th, 2015, 2:16 pm

QuoteOriginally posted by: bearishI'll continue in Alan's footsteps. When you say "priced using bloomberg", that is not unlike saying priced using a calculator, pencil and paper, or the like. Do you mean you looked up an actual quote on Bloomberg, or a BVAL estimate, or valued a private security with some spread to a particular curve, or some other method that is supported? Also, do you mean 5 or 50 bps? And do you plan to "true up" your estimate now and then?I answered your questions below. I'd also add to the discussion that someone encountered a similar issue as mine and advises using a data panel - is it like a matrix with spreads?I looked up the close in the historical price function in BBG as true price. Whenever a new price comes up, we would adjust our estimate to match bloomberg (and therefore true it up). In my current model, the estimateg price is calculated from the last close price recorded in BBG, after that, I calculate the yield change from that period for its benchmark.Whenever it is not traded or we get a single price source, we would like to be able to validate the price broadly and be able to question our pricing vendor in the event where it is completely out of line.I mean 50 bps as this is the furthest I can go before potentially causing a material impact in our positions.
Last edited by cdery6 on April 14th, 2015, 10:00 pm, edited 1 time in total.
 
User avatar
Sprinter
Posts: 103
Joined: October 21st, 2012, 11:05 am

Valuation of illiquid bonds

April 15th, 2015, 3:34 pm

QuoteOriginally posted by: cdery6Hi All !I am currently encountering a mathematical issue at work. It may sound very easy to some with quants background but here it is:I am trying to value 50 North-American private corporate bonds that are illiquid. I thought about using comparables or other market data in order to price them.Here?s my failed attempt: I tried to use the yield change of the bond?s benchmark & the duration & convexity of the private bond to estimate the period price change and add to the previous value to get an estimated price. My issue is that a typical yield change of a benchmark is not at all similar to that of a private bond and so I would need to put something that makes sense in my calculation so that it converts the benchmark yield to the probable yield change of the actual security. Then, using that estimated period yield change, the duration & convexity of the private bond, estimate the price the bond (?)Would some sort of a matrix pricing be a good idea? I pulled of a credit sector spread matrix but it only adds to the benchmark, it does not seem to compensate for the larger yield change that risker bonds seem to have when its benchmark moves.Lastly, I have MAX 50bps threshold for price accuracy ? although I really welcome your input even if you are not sure if your method would meet that criterion.Thanks a lot for your input!if its not a traded bond how to do you get the previous value. last transaction may be way back irregularly sparsed earlier transactions. I encountered a similar issue sometime back and there were regulator's spreadsheets implementing a matrix method. estimating liquidity premiums of traded bonds and determining the levels of premium for varying liquidity of course in additional to credit spread. add premiums and discount. this is an emerging market experience. I don't know how complicated is what you are trying to do what is the requirement what is value required for.
 
User avatar
wellhellothere
Posts: 1
Joined: July 19th, 2015, 9:30 pm

Valuation of illiquid bonds

July 21st, 2015, 5:48 pm

Hi cdery6!I am facing the same problem. Did you find the appropriate solution?
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On