Menzie says money velocity is unstable

I stole his chart, he is right.  of course, with velocity unstable, then term structure is unstable and the Taylor  rule will only work in a corridor of temporary stability, an equilibrium.  We can see right up to the crash, the wholesale money market had to move large amounts of money at faster rates, it did not have the bandwidth and we crashed, and after the crash the economy adjusted the relative velocities to make a better smooth approximation of a fluid economy.  The chart we want to look at is transaction size vs transact rate for the Mx aggregates because that is the production spectra, the band pass of the economy.    Velocity is the inversion of inventory cycle.

What is a good model here?  The one I want to see is an M2<->M1 distribution network, with queuing mismatches as Ms flow mostly from M2 to  a fanout of M1s.  But this is money, so we expect reverse flow can happen, assume symmetry.  Thus we have two inventory nodes, and our quantum assumption is that mean/sigma for the inventory is constant across both nodes.  So, in the chart we have the M2 and M1 changing the quants, and thus the velocities in such a way as to balance their respective share of the bankers yield curve, actually sharing power spectra in standard frequency form.