2) By focusing our attention on bank lending we can see a couple of things:
A liquidity trap is better described as no worthwhile lending opportunities rather than animal spirits causing a drop in borrowing. This is because the risk premium is so high. This could be because of real factors or it could be because bank balance sheets are all torn up and they cannot afford to take risks.
Government borrowing changes the game not because of marginal propensities to consume but because the government is a different type of borrower. The government is always a good credit risk. Indeed, in a world where reserves are swapped for government bonds the government can’t not be a good credit risk. Thus a rise in government borrowing suddenly makes overall lending safer and the BL [IS] curve moves out.
Governments which may directly default (rather than inflate) loose traction on the BL [IS] curve. It is not at all clear that Greece can move the BL curve.
Banks have a special role in the economy such that a trillion dollar by banking institutions is a much different thing then a trillion dollar loss by households. Dot-Com vs Subprime.
3) Raising inflation expectations works by making lending safer and pushing out the BL [IS] curve. Holding the Fed Funds rate constant higher inflation means that each borrower is more likely to be able to pay back any loan at any given interest rate for any give project.
I like this because it helps nominalize our thinking. Too much thinking – in my opinion – is done in real terms. What’s the real interest rate? What’s the real return on investment?
Does this actually matter in the world we live in? The opportunity cost of funds simply is the federal funds rate. As long as the overnight lending market is functioning, nothing in the real economy can change that. Thus an expanding nominal economy – what ever the source of the expansion – makes it a better idea to lend.
 With the nominal interest rate on the vertical axis, banking policy--government guarantees of or equity investments in banks--moves the IS curve out and looks like a (perhaps highly effective) form of fiscal policy. It is, I think, a defect in the IS-LM model that it leaves you with only two curves to shift, and thus everything that affects the economy has to be either "fiscal policy" or "monetary policy". Similarly, a change in expected inflation looks like a form of fiscal policy. And taking risk onto the Federal Reserve's balance sheet at the zero lower bound and thus lowering the risk premium looks like fiscal policy too. There is something screwy here--but you can't descrew it without a more complicated model.
 The distinction between "because of marginal propensities to consume" and "because the government is a different kind of borrower" seems to me to be impossible to maintain. This is a general-equilibrium model, after all, with a goods flow equilibrium condition, a bond holding equilibrium condition, and a money holding equilibrium condition but only two degrees of freedom in the system. It's "both… and", not "not… but".
 Could the Greek government right now raise spending in Greece by increasing its spending and borrowing more? That is, I think, a very interesting question. I will procrastinate on other projects by thinking some more about it this afternoon.