Guessing ECB liquidity until the first Targeted Longer Term Refinancing Operation

The position of money market rates in the €-area within the corridor recently set by the ECB (-10 to + 15 basis points) depends on the outstanding amount of excess liquidity demanded by banks, and is thus not under the direct control of the ECB. It is thus useful to try and guess what excess liquidity could do between now and the first Targeted Longer Term Refinancing Operation (TLTRO) in September. In principle one could follow either a bottom up or a top down approach to carry out this exercise.
Under the bottom up approach, one would notice that, while the TLTROs effects on liquidity would not have come into effect as yet, the consequence on excess liquidity of the non sterilisation of Securities Market Program purchases will be there. In addition, the usual forces will be at play: drawings under “ordinary operations” (MROs – Main Refinancing Operations and LTROs – Longer Term Refinancing Operations), reimbursements of about to mature Very Long Term Refinancing Operations (VLTROs) and autonomous factors.
Given the uncertainty about all these factors, I find a bottom-up estimate, assessing the individual sources or uses of liquidity in isolation and summing them up, impossible to make. It is easier to make a top down estimate: how much liquidity will banks want to hold, independently of the source?
The cost of holding liquidity is critical in this respect. If you consider that this cost corresponds to the “bid-ask” of ECB funding, which is the effective width of the corridor, this has not changed as the distance between the MRO (which I take as the relevant ceiling of the corridor because of the full allotment-fixed rate auctions) and the deposit rate (the floor) has not changed. There is, however, a sense in which the cost of holding liquidity has changed and this is the negative deposit rate, which should disincentivate liquidity demand.
The prospect of TLTROs should exercise, instead, a positive effect on liquidity demand from banks. Before the announcement by the ECB of its package of measures on June 5th, banks had to prepare for the risk of not being able to rollover, al least with sufficient maturity, the 2 three years VLTROs, which will expire at the turn of the year. In a way, the progressively shortening of the funding drawn under the VLTROs was having a negative effect on the demand for ECB liquidity. Now they have the prospect of rolling over much of that funding through the TLTROs in September and December, which are really 2 year VLTROs, capped but without conditions, therefore they can regard the maturity of ECB funding as being, in practice, extended.
The net effect of the two factors will probably be positive on the demand for liquidity obtained from the ECB.
What has happened since the latest ECB move in terms of excess liquidity and EONIA can also shed some light on  future developments.
Chart 1. ECB excess liquidity 
Source: ECB
As can be seen from chart 1, the trend of ever lower excess liquidity seems to have been broken since the beginning of June: I hesitate to identify a new positively sloped trend, but I think one sees a possible break of the previous downward trend.
On EONIA itself, after the ECB move this has approached zero, as in chart 2.
Chart 2: EONIA developments
Source: ECB
However, the OIS (Overnight Index Swap) yield curve (Chart 3) indicates that EONIA should settle around a level of 4-5 basis points for maturities between 2 months and 1 year.
Chart 3: Euro OIS curve (25th of June)
Source: Bloomberg
We are talking about very small amounts, as the difference between the close to zero EONIA spot in recent days and 4 to 5 bp for OIS maturities from 2 months to 1 year is tiny and probably irrelevant. Still I think it is rather the level just above 0 that is, on average, broadly consistent with an amount of excess reserves of 100-150 billion prevailing since the ECB move, with the new effective floor of -3 bp (taking into account the theoretical floor of -10 and the “Sparkassen premium” of 6 to 8 bp) and with the prospective demand for liquidity. I am not sure, however, we can exclude that there would continue to be occurrences of the “cobweb pattern”, whereby banks behaviour will introduce waves in excess liquidity demand and therefore in EONIA [1]. This can be generated by the fact that, unless liquidity is very abundant (say higher than 150 billion), when market rates are low, banks borrow from the market rather than from the ECB but, by not drawing liquidity from the central bank, banks reduce excess liquidity and this will tend to raise money market rates, but then, when market rates are high banks will again borrow more from the ECB and thus inject liquidity and thus lower rates and so on…..
In conclusion, my best guess for EONIA before the first TLTRO in September is that this could hover close to zero and a little lower than implicit in the 4-5 bp of OIS curve. I am not confident enough, however, to exclude a risk of continued instability of the cobweb type, even if the probability of this is lower than before the ECB move. A level of EONIA hovering close to zero is somewhat higher than my -8 bp expectation before the ECB move [2]: this is due, first, to the fact that the ECB reduced rates  by only 10 bp while I was expecting 15, and, second, to the fact that the LTRO is not as “enticing” as I thought/hoped, since it is limited and short for the first two operations and conditional for the following ones.
[***] Madalina Norocea provided research assistance