Outsourcing Government: Accountability

Earlier this year I suggested that outsourcing government should be on the list of ways that signal credible monetary policy:

what's so unthinkable about hiring foreign-nationals as senior government officials? If Russia wants credible monetary policy, it should hire Alan Greenspan

I've previously suggested:

why not force political parties to submit spending plans prior to an election, and make them cover shortfalls or miscalcultions from their own party funds?

In a slight twist according to Harvard's Eric Werker:

Corporations and nonprofit groups, in addition to individuals, should be allowed to run for office in local elections. Firms such as KPMG, Ernst & Young, or McKinsey should all get a shot at becoming candidates to run cities and districts with a history of corruption. Entities that are experts at municipal accounting, consulting to foreign governments, or providing services like procurement support and corruption monitoring might consider becoming candidates... it's a suggestion that the job of mayor itself be awarded by voters to a private firm [via iPienso]

Regardless of the workability of this idea, I think it should be seriously considered. I'm sure many people's instincts would be that this blurs the public/private sectors to an unacceptable level, but first can we all acknowledge that corporations and nonprofits already have huge influence over the public sector. In fact, all I'm proposing is that we formalise and publicly acknowledge what is already taking place, behind the scenes. I think this would have one massive advantage, in that it would provide accountability.

A key problem with how consultant groups operate in the private sector, is that they share little accountability for reforms (i.e the seagull approach - someone flies in, drops something, makes a mess, flies off again, and everyone else has to pick up the pieces). Might transition be more effective if countries appointed foreign experts to ministerial positions, rather than pay them to consult?

I'm not aware of any empirical work that directly applies this theoretical premise, and although it might not be a first best solution, surely it's worth being at the table?