EUObserver has a very interesting expose of the role played by a handful of large financial consultancies in shaping Europe's responses to the banking crisis: http://euobserver.com/economic/122415
The article quotes from a number of sources, including myself.
Here is a more in-depth version of my position on the issue:
There are two basic reasons for the Central Banks reliance on external assessment and validation of estimated banks losses. The first one is operational and the second one is reputational.
Operational reason arises from the fact that during the per-crisis boom in credit creation, National Central Banks of countries with rapid credit expansion lost core personnel competencies and skills to staff migration to the private sector financial services providers. As the result, senior staff with skills at professional certification levels (e.g. CFA) and hands-on experience became virtually extinct in the Central Banks and regulatory authorities. The remaining staff largely performed mechanical tasks of collating and repackaging information supplied to the Central Banks by the financial institutions. Forensic analysis and modelling skills were lost. External analysts can supply these skills and provide more up-to-date specialist knowledge, rarely available in the tenured jobs-for-life setting of the Central Banks that was made even more scarce by the staff migrations to private sector. An added operational constraint faced by the Central Banks in crisis-hit countries was the demand for staff time to cover regulatory and policy changes during the crisis and deploying emergency measures. In simple terms, this meant that the Central Banks were short of staff.
Reputational reasons are more complex, spanning a number of areas where Central Banks faced and often continue to face significant deficits. Firstly, reputation ally, Central Banks are not known for possessing specialist forensic analysis skills required to bring together balance sheet analytics and forward business modelling. As such they lack credibility as markets analysts. Secondly, stress testing had two functions: identifying approximate potential losses on banks balance sheets and signalling these losses to the markets. In the case of countries severely hit by the crisis, the latter objective had to be served by supplying a credible signal to the markets. This signal could not rely on the internal assessments by the Central Banks which were at the time seen by the markets as being captive to the incumbent banking institutions. This too required bringing in external validation. Thirdly, as in the case of Greece, there was always concern that more realistic assessment of the banking situation will expose Central Banking authorities to renewed public anger and trigger public retaliations. As the result, a third party often had to step in to provide a public buffer between the losses estimates, the banks and the Central Bank. Fourthly, counterposing potential public backlash, the banks themselves were significantly incentivised (in the context of loss assessment exercises) to attempt influencing the Central Banking authorities to alter the results of the exercise. Perceived objectivity of the estimates, therefore, required more external validation.