For what it ever may have been worth, you have to at least acknowledge the Federal Reserve really did put its (own limited use form of) money where its abundant mouth had been. The entire story of the crisis-era and then post-crisis experience of “abundant reserves” indicated a monetary situation (liquidity, colloquially) where supposedly money was beyond sufficient. Too much, in fact, according to this doctrine.

When the US central bank was forced in 2008 to elevate its ineffective response to the developing disaster, to do so required non-sterilized non-standard policies. In this context, the pertinent term is actually “non-sterilized” whereas popular imagination has been unnecessarily hung-up on “non-standard.”

Basically, the Fed came to do a whole lot of things, and violating longstanding practice in effect up to Lehman Brothers, policymakers weren’t going to focus too much on the systemic level of bank reserves. They would allow these to skyrocket (mostly due to overseas dollar swaps at that point) prioritizing everything else. Further, officials came to believe this “saturation” was responsible for, among other things, depressing the effective federal funds rate (EFF).

In order to address this secondary byproduct of the “rescue”, in order to “soak up” these excessive amounts of excess reserves, the Fed engaged in several seemingly contradictory (in view of the conditions, which were, it somehow has to be pointed out, a gross and global monetary panic in US$s) policies. One of these was IOER, an utterly embarrassing blunder whose long-term purpose will end up being proof that these people have no idea what they are doing.

Another was employing something called the reverse repo program (RRP). These operations are Temporary Open Market Operations, or TOMO’s, therefore typically overnight (O/N). Here, a counterparty who wishes to lend cash on a secured basis activates its electronic bids with FRBNY’s system in order for the Federal Reserve branch to “borrow” funds from the counterparty when FRBNY pledges largely US Treasury assets (agency and even MBS are available, too) as security for this “loan.”

Given the context of the official view of 2008’s “abundant reserves”, you might appreciate why the RRP was inauspiciously inaugurated at that time: the net effect, so far as monetary policies were concerned, was to have removed funds from a marketplace judged to contain “too much.” Instead of further flooding markets, cash holders lured into the RRP (rolled over for as long as needed) with the Fed had essentially taken “money” out of circulation and theoretically leading to the effective federal funds rate moving back upward toward the policy target.

This never happened, of course, simply because the Fed and its top officials haven’t any good idea how the global system really works; EFF remained solidly, distressingly way below its target throughout the worst of the crisis for very different reasons.

But in terms of the RRP itself, these repeated O/N auctions revealed something useful. If you are a cash lender who can’t find a cash borrower with the financial collateral you desire/require, and don’t like the price of T-bills, then the Fed’s UST collateralized operation makes for an emergency substitute in a collateral-constrained environment. And if the RRP rate the Fed offers is sufficient, it might even be an enticing alternative.

In other words, RRP usage could be an indication of collateral constraint regardless of the level of bank reserves.

That said, we have to start by acknowledging in this post-crisis, QE era a relatively close relationship between RRP outstanding and systemic reserves:

As was the case with 2008, though, we can clearly see that over the past fifteen months there must be other factors involved – which dovetails into the FOMC’s recent, belated, and utterly profound observation that other money market dynamics beyond “abundant reserves” can and have (and will) make all the (functional) difference.

Over the past few weeks and even days, RRP use has soared to nearly $433 billion (as of this afternoon). That’s the highest in a very long time, and quite a bit more than during the last few weeks of March 2020.

Everyone now wants to know why.

The first suspect and the one most will point toward, is always bank reserves; after all, I just wrote above how there’d been a close historical relationship between the two. More excess reserves, more counterparties have to lend, and eventually too much so that some has to go back to the Fed (basically a accounting round trip for them). We won’t get into here whether or not prior period usage had any relationship with collateral constraints, focusing instead on more recent developments.

This year has been flooded with talk of the flood of bank reserves first being created by ongoing QE and then further boosted by the release of last year’s reserves created and then locked up in Treasury’s checking account with the Fed (TGA). For debt ceiling reasons as well as “helicopter” payments, the Treasury Department has been drawing down its TGA which has the effect of shifting reserves out from its account and into those of the commercial banking system.

Convention has dictated this would overwhelm money markets and thus we’d expect to see higher and more frequent use of the RRP as one outlet for this reserve flood.

On the Fed’s balance sheet, when it engages in an RRP the asset side doesn’t change; in accordance with GAAP, as effectively a collateralized loan whatever security the central bank puts up it continues to report as if nothing happened. On its liability side, it reports fewer bank reserves but a larger RRP absorption (basically moving the amount from one to the other).

As you can see above, when RRP use increased sharply it began to more seriously deduct from the bank reserve total (the dotted line for the RRP balance is one factor determining the remainder, the blue representing bank reserves). This, however, still doesn’t explain why the sudden shift in mid-April.

Did bank reserves suddenly hitting the magic number of $3.947 trillion trigger automated changes to money market counterparties? Why didn’t the RRP soar back in February (or November to January) when this latest “flood” began its run? What triggered it first mid-March and then a bigger trigger more so mid-April? You can’t answer that question with bank reserves because these have been going up in predictable, steady fashion.

This, then, draws our attention back to March 2020 when RRP had last surged along with the level of bank reserves as the Fed first ramped up QE6.

Curiously, there was at first the same theoretical correlation playing out in RRP use when the world was “flooded” with reserves initially in the second half of last March. However, the RRP fell right out of favor very quickly beginning in early April even though the Fed was still creating even more abundance in bank reserves.

More reserves, less and even zero use of RRP?

But also, you might remember, a hell of a lot more T-bills. Quite simply, with bills rather than reserves becoming usefully abundant there was no reason to pretend to lend to the Fed anymore. The system had quite clearly become more collateral un-constrained leaving the RRP practically unused since cash lenders could more easily and more readily lend to collateral-plenty borrowers in the private market.

This is where we end up in 2021, if, however, in the reverse situation, something we’ve been writing about since January. Around mid-March, that’s when it began to really change before the bigger jump in RRP following mid-April.

Go back to mid-March, what you find is bill yields dropping even more toward zero, compressing all the way out the maturity range to the 52-week (12-month) tenor. As bills get closer and closer to zero, all bills, first the opportunity cost of going to the Fed’s RRP lessens.

More to the point, with such demand for bills, and a sudden outbreak of RRP usage almost certainly unrelated to bank reserves, we’re left to wonder what the marketplace is finding for cash lending purposes. Are those flooding to the RRP doing so because once again they’re having more trouble finding borrowers with the right collateral? 

Bill prices, falling yields and Treasury cutting back on bill supply already has suggested as much.

There must be a shortage of collateral-ready borrowers, especially since mid-April, leaving the Fed’s RRP window/TOMO’s the least-worst alternative for those with spare cash. Again, this isn’t really about bank reserves.

Now, this does not mean there’s some imminent crash or disaster hiding in these details. Like 2017, for example, when RRP use last soared it likely today suggests a heightened state of collateral shortage, a condition the system has lived with and tried to work around (in good part thanks to QE’s true liquidity effects) for a very, very long time.

What this instead would indicate is, like stubborn longer-term bond yields that are only sideways to lower since Feb 24-25, increased risk due to possibly greater fragility. Bank reserves are elevated, sure, but more importantly, so must be the far more effective money collateral shortage. This would mean any smaller potential issue (like Feb 24 Fedwire) has a better potential to spiral into something more meaningful and then dangerous.

Not that it will, but that it could at a meaningfully higher probability today than before mid-April. And, not for nothing, a higher probability of something like that would effectively reduce even more what little chances for full-on inflation and recovery. Thus, one key angle to why the bond market seems to care so little about the US CPI, and so much more about T-bills (with auctions continuing to hit zeroes).

I can sympathize and understand the over-emphasizing of bank reserves as “money” and “money printing”, and even in this more technical context of essentially money market funds sorting out collateral-ized lending options, but even MMF’s are in one crucial sense sorting their options along collateral lines. That’s what the data shows, and what’s driving RRP.

In that sense, as well as the mainstream view of “abundant reserves”, not much has changed since 2008.