Does anyone have data that can help to quantify the "cons"... the liabilities?

I don't have the data to prove, but I do think more banks are hiring KML to handle the foreclosure.

Initial up front cost to buy the training, time it took to learn/implement, monthly costs (dialer, skip tracing, etc), how you are skip tracing/phone appending large lists and if you are getting accurate landline/cell phone data, etc.?

Piggy-backing off of the last weeks podcast, I would suggest analyzing your data.  

With so many avenues to market these days I would be curious where the agent is listing the property are they doing YouTube videos are the meta-tagging / geo-tagging / are they using keyword search optimization there's a lot of things that they could be doing and instead of just putting it on the MLS.

Once we came to the realization that they bounce we didn't make any changes to our processes as our data & experience has shown that it's got the highest % chance to be a smooth payment.

You have to change a few default values to zero, adjust the closing costs a bit and ignore a few pieces of data, but all of this is very simple.

Thanks to Myka's data I would figure you could conservatively say that it would be rented 45% of the time vs AirDNA's figure of 62%.I'm looking at a 4BR which would rent for $1600-$1700 brand new vs Airbnb for $2700.

It also breaks down data by bedrooms too, this would let you compare yourself to other 2 bedrooms.

Any half way decent property owner will be able to provide annual costs for all past years for anything they pay...but don't automatically trust the data you are provided.