Today Tim Geithner opines in the Times about the great recession, the policy response, and the recovery. He cites a recent study by Mark Zandi and Alan Blinder that uses a macromodel of the U.S. economy maintained by Moody's Analytics to divvy up the effects of the fiscal stimulus and the various monetary/financial policies (TARP, quantitative easing).
Their counterfactuals are entirely model-derived and thus subject to the usual criticisms. What's interesting is that they separately parcel out the effects of the fiscal stimulus and the effects of the financial policies, and they find that (1) the sum total effect of both is greater than their parts, and (2) the partial effects of the financial policies were larger than the partial effects of the fiscal policies.
Blinder and Zandi argue that the reinforcing effects of the two policies produce a sum greater than its parts; for example, they offer, housing credits in the fiscal stimulus helped raise housing demand and thus prices, which may have improved the efficacy of financial policies.
If economics-of-scale in policy interventions are indeed operative, then a cost-benefit comparison of fiscal and financial policies is a little messier. Just comparing their individual partial effects to their costs seems to strongly suggest that one was more cost effective. Blinder and Zandi don't seem to highlight this, but I think their results show that financial policies (mostly the TARP) may have raised 2010 GDP by 2.65% (Table 6) at a cost of about $101 billion (Table 9), while fiscal stimulus may have raised 2010 GDP by 1.9% (Table 8) at a cost of about $1 trillion (Table 10).