An older description of MMT

Monetary policy has no effect on AD. Fiscal policy can be used, and must be used, because this model, with its vertical AD curve, has no inherent tendency towards "full employment" output. The price level is indeterminate, unless active fiscal policy makes it determinate.Since monetary policy has no role to play in determining AD, the central bank can set any interest rate it feels like setting. Indeed, it might as well set a nominal interest rate near zero, since this reduces the transactions costs of people converting between currency and bonds to try to avoid the opportunity costs of holding zero interest currency. 

Not quite, rates are unknown in the ISLM model, actually, they are instead current interest charges as the central bank.   Changing interest rates means changing interest charges, a greater flow from borrower to depositor.

Thus, if government changes the interest charges it is really moving up and down the curve.  It can feed the poor one day, then suddenly decide to start a moon space project.  It is moving itself in units of 'relative term length'.

The underlying model is corridor.  The recession is an unexpected increase in S/L. The currency banker actuarial gains are deflationary, interest charges last time were too high, and the currency banker becomes instant borrower of current resort, attempting to lose, actuarially.  The currency banker always dealing with its own market making accumulation, keeping that within a bound.

The corridor makes the Hicks stick figures curve. There is a maximum liquidity point between the aggregate of deposits and loans. The left curve on ISLM slopes down to horizontal, and the second slopes up from  horizontal.  When both are parallel then elasticity is maximum, the velocity equations of money work, the hologram effect. The equilibrium is actually impossible, it is an irrational ratio of S/L, and one can see that measuring relative slope when both curves are nearly horizontal is hard.

There, a chart of ISLM, however they  define the axis. But maximum elasticity is hard to identity at the bottom, where the indifference curves are parallel.  The currency banker has some finite set of choices in bouncing around that bottom trying to keep market making error bound.

The MMTers simply set the currency banker error bound to some humongous value. Like I say, it works, but the poor have to instantly become rocket engineers.


In reality:

Solutions at maximum liquidity are algebraic in the sense that demand and supply are a semi-repeatable sequence.
The Keynesian attempt, at the aggregate, is an attempt to reset the sequence, make folks remember back to the last time government intervened at this level.

But at maximum liquidity, the shovel ready projects are visible, they are most transactions. Maximum liquidity, like maximum entropy, is a fundamental assumption. If the currency banker has the best estimate, the there is spectral containment, agents will use the currency banker as an accounting standard. The currency banker has the best estimate when the Walmart checkout manger algorithm is used to stabilize deposit and borrow queues.

When the velocity equation is working

Then the firm can scale up a profit estimate, with some extra risk, it can multiple the price of eggs by a million, determine the size of the egg industry.  It can scale up the number of chicken coops, size of delivery trucks. Price approximately works, because the quotient ring conditions are approximately met. Better yer, the aggregate pricing error is bounded.

But we are dealing with finite ratios and parallel indifference curves. So under maximum entropy, that optimum liquidity point must be the most irrational number, the number that lets us sub-divide the 'packing error' or quantization error. That is, as the sequence count tends to infinity, the maximum liquidity ratio S/L should tend to 1/Phi. One can see this in the egg industry, absent packaging costs, one could build more and more intermediate points, each point being another packaging of eggs, and I think you get the proper ratio up to the precision of one egg.

Automatic stabilizers vs stimulus

Error correcting terms.  That is the function of market maker in the general case, they carry the error correcting term, like those physicist who utilize error correction as a fundamental property.

The innovative, aggregate stimulus mainly error corrects for past imbalances that suddenly come due, but it does so with an increased government liability to be worked off later. This stimulus resets the sequence clock, restarts the new recession cycle.

Balanced queues a maximum liquidity

For deposits or cash in advance, the biding queues are balanced Huffman generators, all paths equal length. At that point, any market making funds have maximum impact, both S/L and delta S/L maximum. The currency banker is most responsive.  On a sudden skewing o the queues, market making corrections are way up and it takes a couple of banker moves to hop back toward maximum liquidity. The key here is to always maintain a representative sample, and add the third color to manage entry and exit congestion. This part is different from a two color Walmart because the currency banker 'profit' is not direct, the market making function does not compute a direct profit except after a number of steps later as it is a deliberately in  error currently.