SAMBA no seminário (wonkish)

Since FK is asleep, let me say something more about the Seminario de Metas do BC.

First, Silvio e Fabia gave a quick view of the new SAMBA, now with reserve requirements as an alternative monetary policy tool (actually, they call it macroprudential…). Interestingly, this happens even if the reserves yield Selic rates. Basic idea is to suppose banks have a cost which is proportional to the difference between the return on loans and the Selic. (This only makes sense if it is implicitly assumed banks are risk neutral, or if there is some form of transaction cost that makes banks prefer to stick to an optimal level of Gov’t bonds). Needless to say, I’m not particularly happy with their microfoundations. But I believe it is a useful addition to the original model.

Second, a gringo called Wolter, or something like that, discussed how to perform conditional forecasting in DSGE models. (You know, the typical question of how much would be inflation if the interest rate path is this or that.) The issue is if one should use “stochastic shocks” (keep the same Taylor rule and add a monetary shocks), or “deterministic shocks” (change the Taylor rule). In the second case, agents anticipate the monetary policy change and its effects get hugely magnified.

The gringo schmuck didn’t know which way was better (and neither did FK). I believe even though deterministic shocks make more theoretical sense, stochastic shocks are better, because people (in general) do not anticipate.

Something with this flavor is happening now in the US: Bernanke is kinda suggesting QE tapering would not be deterministic or monotonic. You know, when they start buying less treasuries and MBS, they may increase or reduce the pace depending on the state of the economy. By saying this, Bernanke hopes the economic agents will not react strongly, and Fed “great exit” maybe runs more smoothly.