Merger Arb Risk Modeling

rakato · Post by **rakato** » Thu Jan 21, 2010 9:40 pm

Came across this pdf on merger arb risk:

http://www.riskmetrics.com/system/files/private/rmj1q08_ext_merger_arb.pdf

I thought it was pretty intuative and clever. Any thoughts on ways to measure merger arb risk would be appreciated. Difficult to explain to potential investors on a greater "macro/30k feet" level the intuitive "micro/10k feet" sense that risk arb traders get after years of seeing deals on when to hang and when to bug out. Thanks

rmeenaks · Post by **rmeenaks** » Mon Feb 01, 2010 10:59 pm

Here is the original referenced paper:
http://www.riskmetrics.com/system/files/private/rmj1q07_merger_arb_risk_model.pdf

sas · Post by **sas** » Wed Mar 30, 2011 2:38 pm

has anyone implemented this in excel and would be kind enough to share it?

Thanks

Nonius · Post by **Nonius** » Wed Apr 06, 2011 6:33 am

Coincidentally, I was thinking about this just yesterday. Without looking at references, I sort of heuristically figured one could apply something similar to looking at relationships between credit spreads and market implied default rates, where "default" in this case means merger or takeover collapses and success gives rise to earning the spread. At least this seems to make sense for post announcement cases. That should give some market implied probabilities. I also spoke once with a hedge fund that built a historical db of mergers and then built a merger book Var model based on historical success rates of mergers.

Bloke1664 · Post by **Bloke1664** » Wed May 04, 2011 4:25 pm

This is something I am interested in as previously I wrote an extremely simplistic model for calculating VaR on M&A Arbs for an large IB, and then went on to trade the stuff. However, this link just sends me to the MSCI home page. Is it possible to link it to something else I may be able to retrieve it from?

raf · Post by **raf** » Wed May 04, 2011 4:51 pm

Hi,

I tried with the wayback machine wihtout success.

Is it the one you're looking for ?

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1548429

Cheers

Raf

nodoodahs · Post by **nodoodahs** » Wed May 04, 2011 5:40 pm

Might also be worth looking at the ETF 'MNA'. If I recall correctly, the 'index' they created to track was a replicable and systematic grading system for merger arb and the papers they promulgated around the time of the ETF's launch might be interesting to you.

aaron · Post by **aaron** » Fri May 27, 2011 7:36 pm

This is an odd paper. It starts with the problem of estimating VaR of a merger arb portfolio. It proposes a reasonable simple model and does a Kolmogorov-Smirnov test on the residuals on seven deals in 2005. What does this have to do with VaR? Pretty much any seven numbers without a huge outlier will pass a KS test, you need about 30 to have any power. And whether the residuals seem Normal or not doesn't seem relevant. Most importantly, almost all the risk comes from occasional periods where lots of mergers fail and stock prices are down a lot. Testing in a single time period will miss that risk.

There are basically two components to the risk. First is idiosyncratic deal failure. This has to be estimated deal by deal. You could try some generic assumption, like the loss will be twice the difference between the current price and the pre-announcement price adjusted for equity index moves since. That's fine if this is part of a multistrat fund or you have a lot of deals on. But if you're a pure merger arb portfolio, especially if merger activity is limited and you only have a few deals, it's not going to work well.

The second component is anything (including a deal failure) that causes merger arb spreads in general to blow out. If you're aggregating risk to a multistrat fund or prop trading group, this is what will dominate. This is also the component that will have systematic risk. Modeling the merger arb spread isn't particularly difficult compared to other risk spreads.