How is the default probability implied from market implied CDS spreads for CVA/DVA calculation?

From point 38 on P.17 the default probability can be implied from market implied CDS spreads. "Macro Surface" method is mentioned, but I cannot get any clue of what it is? Where do I get the acedemic reference for that?

Also what is the commonly used methodology to imply default probability for CVA/DVA calculation?

The article "Credit and Debit Valuation Adjustment" can be seen in http://www.ivsc.org/sites/default/files/IVSC%20CVA%20-DVA%20%20ED_0.pdf

There are quite a few methods to calculate default probabilities from CDS data. Simply you start at the shortest tenor, assume constant hazard rate. Then for the next tenor, you assume the previous hazard rate is still valid till the previous tenor, and the hazard rate between the previous tenor and new tenor is calibrated so that CDS PV matches the market price.

The Macro Surface method is explained in the document. Actually we use this proxy method but never seen it named as this. Simply you map your illiquid CDS to liquid index CDS that have similar rating, industry, country. In the mapping you can assign some scaling, that is calibrated historically to from the liquid single names to liquid indices. The issue usually is to figure out the liquid single name CDSs. You might need some trader input.

Your reference says "This method derives implied CDS spreads for unobservable issuers through the interpolation or extrapolation of observable CDS. It is a factor model that constructs CDS spread surface as a function of credit rating and maturity."

So this is for issuers which do not have any CDS contracts priced (there are no CDS spreads to bootstrap). I've never heard of it and I'm pretty sure that this is just their own phrase for a simple linear regression model where the betas are {1} or {0}, {1} indicating membership of a factor group {AAA,AA,BBB} {3yr,5yr,7yr}.

For my own part, if you wish to do this I would suggest using also Country, Sector, and Leverage Factors.

A methodology for estimating rating/ region/ sector proxies for ACVA calculations can be found here: http://www.nomura.com/resources/europe/pdfs/cva-cross-section.pdf Please let me know if you need anything to be clarified (caveat: I am one of the authors). The methodology assigns a CDS mark to counterparties that either have no CDS marks, or their marks are unreliable due to poor liquidity.

This sounds similar to what they call 'macro method', although I have never heard the term before. They could refer to a more direct approach where they map a name to a traded market-wide index.

Given the CDS curve, the Survival Probability Curve is derived by bootstrapping and under an assumption for recovery.