Olivier Blanchard explains, (link fixed)

It’s very clear that the traditional monetary and fiscal tools are just not good enough to deal with the very specific problems in the financial system. This has led to the development of macro-prudential tools, which what may or may not become the third leg of macroeconomic policies.

[Macroprudential tools allow a central bank to restrain lending in specific sectors without raising interest rates for the whole economy, such as increasing the minimum down payment required to get a mortgage, which reduces the loan-to-value ratio.] In principle, they can address specific issues in the financial sector. If there is a problem somewhere you can target the tool at the problem and not use the policy interest rate, which basically is kind of an atomic bomb without any precision.

The big question here is: How reliable are these tools? How much can they be used? The answer — from some experiments before the crisis with loan-to-value ratios and during crisis with variations in cyclical bank capital ratios or loan-to-value ratios or capital controls, such as in Brazil — is this: They work but they don’t work great. People and institutions find ways around them. In the process of reducing the problem somewhere you tend to create distortions elsewhere.

Although Blanchard is appropriately cautious about these tools, I do not think that he is sufficiently cynical. I wrote,

Politicians want to make credit allocation decisions. Whatever its nominal purpose, bank regulation is used to enable politicians to undertake credit allocation.