Ignore supply and demand, look at the queues

Imagine two areas: Gotham and Pleasantville. Say the demand to live in Pleasantville increases a little while the demand to live in Gotham soars. And say that due to differences in land use restrictions, housing supply responds dramatically in Pleasantville and very little in Gotham. Then what we'll observe in Pleasantville is a rapid increase in population and slower growth in prices, and what we'll observe in Gotham is rapid growth in prices and slower growth in population. The Economist

Its a channel. People step toward the city, and the queue separates into two groups, one in the suburbs and one in the urban area. The money flows and the people flows are coherent, so they queue up together and show as relative prices. Why does it matter? Because channel theory tells us more, it tells us the queue sizes, for both money and prices. Channel theory should also tell us the arrival rates, which should go as Nlog(N) where N is the arrival rate and log(N) the queue length. There are two Ns in this analyisis, and they should be within k of each other, where k is the household size. (When Ni is larger than 1)

(k)(Ni)log(Ni) = constant to the nearest integer. The N1 -> N2 is the expansion of the network out from the urban core, and N2/N1 goes like the golden ratio. The total arrival rate should be N1 + N2.

There are two steps to get here. 1) From minimum redundancy, exp(sum(N1LogN1) ) = constant = 1+ SNR. by Shannon. But 2, there is a competitive aspect, all parties are compete for construction and land space which they minimize as variance inventory. That parties force into this two stage model  the golden ratio between stages, the Canadian interpretation.. When you move up the line from suburban to urban, you can only do so when you accumulate, within the standard variance,  one set of urban inventory, land, crew, government permits, and  So the transition liits your choices of move to the Finonacci sequences or equivalent. prmies to the future, that limits the range of future steps to discrete points. It is the economy of scale effect.

I think I have most of that right.  
So take something like a Windowing Huffman encoder and it will fid the proportional bunches of minimum entropy sub sequencies. Get estimates of the construction quants needed for a minimum step move.