Please, dispel my doubts about POS, miners rewards and too strong incentives to participate in it that could lead to almost total supply of ether locked in stake.
So, we have:
Et - total Ether supply at time t.
St - total Ether at stake at time t.
Rt - reward that goes to stakers at the end of time t (which could be defined as some sort of function F: Rt = F(St), it doesn’t matter now).
IR - inflation rate of Ether due to increasing supply.
IR = Et+1 / Et - 1 and Et+1 = Et + Rt. (big simplification here because of a lag with which the miners would spent their reward on market which should lead to inflation afterwards, but it doesn’t matter here).
IR = Et+1 / Et - 1 = (Et + Rt) / Et - 1 = Rt / Et.
Simultaneously miners interest rate for the same period is:
MR = Rt / St.
Therefore, MR > IR while S < E.
I mean that miners would get constantly higher than inflation rate while ether at stake is less than total supply.
Then, if it is relatively friction-less to stake ether, and if it is really safe to do this, this looks like a perfect bond: everybody could participate and get rate higher than inflation with no risk.
If there were something like this in financial markets, then, no doubts, soon all the money supply would be in such bonds.
If there is not enough risk of losing S, this builds a strong incentive to rise S till it reaches E: S → E. All ether would be at stake which freezes the economics. I mean something like negative deflation effects which leads to major crisis. Nobody wants to spend any wei, because what you lose is not just this wei, but a guaranteed inflation-free income on top of it. You have no motive to make any business and take risks if you could simply put all money at a deposit which is by definition yields you more than inflation.
I have tried to find any answer and excuse me in advance if I missed something. For now I have found only something in this paper: “Incentives in Casper the Friendly Finality Gadget”. From the paper there would several risk types:
Loosing private keys, but this don’t count, this is general risk of using cryptocurrencies, stake it or not.
losses due to network latency and timeouts which could be mitigated by mining in pools.
Presence of malicious miners which leads to NPCP/NCCP (see paper) penalties. But what if there would be no malicious miners? And anyway, in the long run there would be two types of situations for honest miners: either NPCP/NCCP penalties would be such that you constantly lose money and you opt out, either you still get higher than inflation income on average.
So the question is that I see a strong market incentive to put all ether at stake and no force to prevent the situation when S→E and complete disappointment after.
As I see now there could be only two types of such forces: different issuance model or some difficult-to-calculate-in-advance risks to put money at stake.
Of course this is relevant to the time of complete mass adoption of Ethereum. But still.
Update
I have found this paper: https://docs.google.com/document/d/1Xf_iHGI51Lp6nVHPZo0SdUw60-NA6YJUvk2s-eYaxLE
citation 1:
maxDeposit - the maximum deposit allowed to become a proof of stake validator. If you want to deposit more, you must purchase multiple slots. Set to 50000
citation 2:
Alongside this, another question is how to target the minimum entry size and a possible entry fee. One approach is to auction entries, thereby resolving many preference revelation issues and allowing targeting a fixed PoS participation rate, but at the cost of introducing or exacerbating selfish-mining attacks.
If I understand them correctly, there is talk about entry fee to become a validator/validator candidate (thus participate in issuance) which could be defined by market. Looks like this is the thing that will balance the incentives to be full in.