I'm sorry but I think this is a terrible idea. You propose the exact opposite of what you should when the economy is expanding--you are strangling it instead of encouraging it. All to prevent price inflation temporarily if something other than expansion is happening (and cause deflation if it is expansion). Assuming the currency creation base is stable, the value of the money should continue to oscillate around that point. Hell, even without the idea of price stickiness, merchants in a system like Decrits would not need to constantly reprice because of DCR->fiat fluctuations if they could be fairly sure of the stable currency creation base. This will take time and market penetration though.
Messing with account balances or locking coins is going to leave a terrible taste in everyone's mouth, and I don't think it accomplishes anything worthwhile.
The thing that is being curbed is nominal transaction fees. Without a trusted Oracle, we have to rely on within-network indicators. Nominal transaction fees is a within-network indicator, that is relatively secure.
Any indicator that affects further results will be prone to abuse. And sorry, I believe the direction of the feedback you describe is not correct. If there are more fees being spent and the system encourages it with printing more money, then even more fees will be spent and even more money will be printed until the coin collapses due to hyperinflation. There needs to be a written down growth path for some indicator. For bitcoin, it is the monetary base itself. I'm discussing the possibility of using the nominal transaction fees as an indicator. One can try to model a good growth path that is high in the beginning and lower later, that is beside the idea of targeting a level of nominal transaction fees.
Hope I've made things a little more clear.
The way I look at it, there is a component of proof-of-work/mordor coin that has to be incorporated into every coin. The question seems to be how to minimise it and still maintain stability.