“Sustainability” is not a useful criterion by which to determine the optimum size of the deficit.

Olivier Blanchard, former chief economist at the IMF and Takeshi Tashiro  argue that the deficit should be held to a level such that were it to stay at that level, and all else remained constant, the debt would not continue to grow ad infinitum. That’s in their article entitled “Rethinking Fiscal Policy in Japan” (published by PIIE).

While the word “sustainable” is ultra-fashionable nowadays, the deficit does not actually need to be sustainable in the latter sense.

To illustrate, suppose demand collapses (say because of a collapse of consumer or business confidence), and a deficit bigger than can be sustained long term would solve the problem, do we take it that that deficit should NOT BE implemented: i.e. the unemployed should be left to rot??

I think not. MMT (Mosler’s law in particular) says that the deficit should be WHATEVER deals with unemployment. Keynes said much the same.

Mosler’s law, which used to appear in yellow at the top of Warren Mosler’s site, states: “There is no financial crisis so deep that a sufficiently large tax cut or spending increase cannot deal with it.” (Warren Mosler founded MMT).

Plus the fact that that deficit cannot be sustained long term does not matter, because national debt (as MMTers keep explaining) is a private sector financial asset, and the bigger that asset, the higher will private sector spending be, all else equal.

Ergo, a deficit which is larger than can be sustained long term just isn't going to continue for ever because “Private Sector Net Financial Assets” (an MMT phrase) will eventually rise to a level where demand rises to a point where little or no deficit is needed.

The above is part of a long standing form of schizophrenia that the IMF has had in relation to deficits. That is, it knows that deficits are needed so as to deal with recessions, at the same time as knowing that the resulting national debts cannot be allowed to grow too large. But it does not seem to be able to work out the optimum compromise between those two. E.g. see here and here.