NGDPLT

 The currency banker always wants to see two percent a year in money loses, for NGDP growth, loosing as much money as needed.

How does the currency banker lose money? The pit boss takes out loans against customer deposits.   The pit boss losing money always aiming for more loans/deposits in its market account and fewer loans to deposits for customers.. 

The thinking is reversed in models that say lower interest charges cause spending to pull forward.  Lower interest charges mean deposits are in excess. If a pit boss is putting in excess deposits then it is earning money from the market, deflationary.

Central banking is different since Treasury takes gains and absorbs losses. Right now the Treasury is soon taking more money from the private sector, consumer prices should be dropping, government prices going up.

Go back to S/L banker with a loss function, It will exhaust all the higher level productivty investments, and suddenly be over charging NGDP, then it switches states, and adds more deposits, lowering charges for cheaper goods.  The price index splits in two, which is fine, I guess.

 The sandbox S/L with neutrality is a pure productivity norm. An externally caused shift in the loans/deposits skew which suddenly appears to the pit boss. Consumer are will to take more credit to buy the new and better goods. Suppliers pay off their loans early, and the pit boss will lose money in restoring skew back to zero. That should put prices back to normal.

How did traditional theory get it backwards? They assume the currency banker is always ex ante with the deposits. When a loan is made at market price, the currency theory assumes the banker immediately makes the equivalent deposit, at market rates.  But the borrower has inside information on productivity and will, ex ante, have already set up its loans/deposits to reflect this productivity gain. The borrower is making the hedged bet. That additional hedge is lost, and must be absorbed later by the currency banker.

In central banking that lost hedge gets transferred to treasury, it is the currency risk. We end up with a sever Wile E Coyote. The S/L banker is the second one to see the skew, it sees the board after every bet is placed.  The pit boss thus seeing both the loan bet and the currency risk separately. It takes the loan risk and splits it across the pit to reduce skew, taking its share of price risk.  

The currency banker can either lose or win. In the neutral /L banks the inventory risk, as a ratio, is equal to about 1.5 times the pit boss bounds. The pit boss bounds are always limited to constant currency risk. Price is the delivery volatility and always yield neutral inflation. So the neural pit boss accounts for entry and exit of products and assumes the 'rank', or complexity level is the same over the long run.   If the S/L is wrong it will step through bankruptcy and a more accurate /L takes its place. This appears as super position along the Markov <1,y,z> chain.  If the S/L banker has a loss function it appears on the <2,y,z> chain, it will have two modes, losses and gains, partitioning the price system.

Central bankers are stuck, they cannot be really neutral S/Ls. But they can reduce volatility and have a loss function with a partial partition of the Treasury inflation tax. They end up making the inflation bet, then reverberate about an inflation axis for a few yer, then make a new inflation bet. The Law makes it impossible for Congress to have an ongoing inflation betting pit because of the built in skew.