With Chaos Increasing, the Fed Must Cease Targeting
On March 15, 2013, just before news out of Cyprus would capture nearly all financial attention, the US Treasury issued a request for a "large position report" from institutions holding at least $2 billion par value of 2% US Treasury Notes maturing February 2023. These 10-year instruments had been the subject of market consternation as a large spike in repo fails occurred that week, reported to be more than $80 billion or five times normal levels. The 10-year was of particular interest because repo rates on that tenor had turned decidedly negative and were fast approaching the -3% fail penalty.
Repo fails had been a particularly troublesome, but still, seemingly, misunderstood facet of the 2008 panic. The huge increase in repo fails after the Lehman bankruptcy in September 2008 crashed the daisy chain of rehypothecated "liquidity", reinforcing the feedback loop of liquidity draining at that time. Any obvious threat to repos then should be treated with extreme caution and care.
The overnight repo rate on that February 2023 bond had been about 20 basis points in the week prior, but fell negative on Monday, March 11. By the close on Wednesday, March 13, the overnight repo rate was -2.95%. That day the Treasury sold $21 billion in a 10-year reopened auction that seems to have abated the negative repo after settlement. We still don't have a good idea as to why that issue became so scarce in the days before the reopening.
This was a bit of an outlier, important nonetheless, in mid-March amidst the obvious investor preference for "risk". Such a negative repo rate indicated a sharp and acute shortage of collateral. It was odd that it was isolated to that part of the curve, thus nudging the Treasury Department to its blanket inquiry.
While Treasury fumbles about in gathering witnesses, there are two more obvious processes at play that may lead to these types of negative repo situations in the first place. The first is QE 3 & 4, as the Federal Reserve purchases both MBS and US Treasury securities from "the market". In the course of doing this, the Fed also purchases and takes out of collateral circulation on-the-run issues, reducing the available pool in the aggregate. We saw this notably in t-bills toward the end of QE 2.
And like that episode in 2011, there was a political element that contributed. In both cases the FDIC appears to be the culprit. With regard to this latest repo query, on November 9, 2010, the FDIC issued a Final Rule that implemented section 343 of the Dodd-Frank Wall Street Reform and Consumer Protection Act. That meant that the FDIC, as of December 31, 2012, would no longer provide "unlimited" insurance coverage for non-interest bearing transaction accounts. Post December 31, these deposit accounts would revert to aggregation with interest-bearing deposit accounts and be subject to the legal limit for deposit protection of $250,000.
That means that in addition to removing usable collateral through QE 3 & 4, the Fed is also creating deposit balances (the trade in QE is the removal of bonds in exchange for "reserves" that show up as deposits with banks) that are no longer covered under the "unlimited" FDIC umbrella. For some financial institutions (including money market funds) the need to find a "safe" alternative to these deposit accounts was suddenly thrust upon them come January 1, 2013. In aggregate terms, the available supply of "guaranteed" or "safe" asset classes shrunk considerably, particularly since there were more than $1.6 trillion in insured non-interest bearing deposits before December 31. In response, we would expect to see demand for remaining safety classes rise noticeably.
DTCC's overnight weighted average repo index rate had averaged about 25 bp in the fourth quarter of 2012. In the first quarter of 2013, the repo index has averaged only 15 bp, meaning a significant drop in the average general collateral rate across the FDIC insurance change. We see the same 10 bp average reduction in repo rates in both MBS and agency securities as well.
For MBS and agency repos, there is no 300 bp penalty for failures, meaning that a 0% repo rate acts as a floor on repo trades in terms of any special issue. If repo rates drop from an average of 29.6 bp to 19.1 bp as they did in MBS during the referenced time periods, it limits significantly the "specialness" of individual MBS securities. At 29.6 bp, the MBS can trade special up to 29.6 bp, where at 19.1 bp there is significantly less room.
As the spread of the GC rate to the floor is reduced, the spread the bondholder might earn on a repo trade is likewise reduced. Without a fails penalty, those bondholders at small GC to specials spreads simply keep their bonds, reducing the availability of collateral and leading to a potential increase or spike in the amount of repo fails.
According to DTCC's volume statistics, average volume in both the MBS (-5.98%) and agency (-27.2%) has fallen after the FDIC change. Treasury repo volume, on the other hand, has gained (+1.9%).
If we put these movements together, the FDIC change and QE 3 & 4, we would expect to see higher demand for US Treasury repo and a smaller general supply pool. That would mean lower rates and occasionally a very real scramble for any issue in real short supply, like the February 2023 10-year note. What this may mean monetarily is that liquidity conditions just may be tighter than other indications are showing and that QE is counterproductive, again in perhaps a replay of Spring 2011.
This collateral shortage is itself an ongoing problem that first manifested in August 2007. For its part, the Federal Reserve as "lender of last resort" has mostly stayed out of collateral and contained itself and its policies to the traditional unsecured wholesale markets. In many respects (but not limited to this) this has been one of the single biggest "liquidity" failures of the entire crisis affair. Central bankers complain about transmission mechanisms, but the largest piece of that failure is the ongoing shortage of usable collateral.
In September 2011, authors Elizabeth Klee and Viktors Stebunovs proposed to the Federal Reserve Board of Governors an idea to change the very mechanism for monetary policy in a paper titled A Target Treasury General Collateral Repo Rate: Is A Target Repo Rate A Viable Alternative To The Target Federal Funds Rate? Currently, the Federal Reserve targets and conducts monetary policy through the Federal Funds marketplace, meaning the unsecured wholesale markets. As the panic in 2008 and European sovereign problems have shown conclusively, unsecured wholesale markets have been fully closed off to large proportions of the global banking system. That leaves secured lending, ie, repos, as the sole means to flow liquidity into the system.
What Klee and Stebunovs suggested essentially recognizes that transformation. Instead of conducting empty QE's that only fill primary dealer ledger balances with largely useless and unusable "reserves", a repo target would be much more sensitive to the amount of collateral inside the system that intends to turn reserves (money stock) into true liquidity (money flow). Without any attention to collateral, traditional monetary means that depend on unsecured wholesale interbank lending have little chance to truly liquefy the entire system.
In other words, Klee and Stebunovs propose not a change in monetary philosophy, but in mechanics and operations that more closely align with how the system has oriented the past several decades. While that would seem to more closely harmonize intent with reality, it really represents, in the bigger picture, nothing more than the status quo. Interest rate targeting through these various means, regardless of which rate is ultimately pegged or controlled by monetary planners, is nothing more than the same kind of "sterilization" we have seen since the 1920's.
There still seems to be little appreciation for the regime change in monetary operations in the late 1980's. The Federal Reserve's monetary policy used to be entirely based on the method of "reserve targeting". The FOMC would target a level or quantity of bank reserves and only "fund" the liquidity needs of the banking system up to that level. Once that level was reached, market forces would translate the level of reserves against the demand for them into short-term interest rates. That meant there was a direct and beautifully symmetrical (inverse) relationship between the quantity of "money" and interest rates.
In changing from reserve targeting to interest rate targeting, the Federal Reserve upended that natural relation of interest and quantity. The FOMC now sets a target for the Fed Funds market and promises, through two-week "maintenance periods", to issue or "fund" a potentially unlimited amount of "reserves" to maintain a specified interest rate target.
There is a need to clarify here exactly what is being funded and by whom. By saying that the Federal Reserve "funds" the wholesale markets, it really means that the Fed is providing a liquidity backstop. In almost every case, the Fed provides little actual funding into wholesale markets - the banks do that themselves by expanding their own balance sheets. What allows them to do so is this implicit backstop, knowing that the Federal Reserve will, if it is forced, create "money" to maintain whatever it is targeting (interest rates or reserves). The Federal Reserve only "fine tunes" money markets through open market operations, either adding or draining minor quantities of reserves in pursuit of whichever target level.
But the overall transition from reserve targeting to interest rate targeting was a fundamentally different approach to monetary engineering and banking in general. From the perspective of the banking system, in a reserve target environment there is an explicit limit to credit production, a marginal dollar if you will, which will trigger a rise in the cost of funding. At that marginal dollar, there is no guarantee or reliable estimate as to how much that marginal dollar will actually cost, and certainly less idea about a second marginal dollar. This is scarcity at work.
In a scarce dollar environment, banks view attaining "risky" positions far differently than if they have certainty about any marginal dollars. This imposes the process of intermediation upon banks; they must actively manage risk due to very imprecise knowledge about how much marginal funding will cost to add additional risk - they have to choose between risky propositions rather than opt for all of them. This is the beauty of monetary symmetry between quantity and cost.
By contrast, under the interest rate targeting scheme marginal dollars all cost the same. If the Federal Reserve targets a Fed funds rate of 1%, as Alan Greenspan did following the dot-com bust, then banks are assured of a 1% cost of marginal dollar funding. They know they can add risk to any level with only a minimal funding cost at inception. The only risk/funding parameter they have to model out is when that interest rate target is moved upward, and by how much. That provided little challenge to banks intent on adding to credit production or risky positions since they were explicitly guided by FOMC policy statements about what to expect - Greenspan telegraphed that rates would stay low and only gradually rise once the FOMC was assured of whatever economic variables they wanted to see moving in the "right" direction. It reduces the link between funding uncertainly and risk appetite to essentially nothing.
In a Working Paper (#916) for the OECD, author Patrick Slovik defines some of the changing risk characteristics of global banking. In his study, Mr. Slovik quantifies changes in the top 15 "systemically" important banks across the United States, Europe, UK and Switzerland. In 1990, these 15 banks had an aggregate risk-weighted asset (RWA) to total asset (TA) ratio of 66%. That meant that, through Basel rules calculations, these banks were reducing their count of risk-weighted assets by about a third. Doing so allowed them to increase their overall count of total assets on their balance sheet per proportion of actual bank capital. In other words, the lower the RWA/TA ratio, the greater banks can leverage themselves overall.
By 2008, these banks had cut their RWA/TA ratio in half, all the way down to 33%. For capital rules, it looked as if there was no difference between 1990 and 2008. Indeed, overall capital ratios (defined as Tier1 Capital/RWA) actually improved from 7% to 8%, the key was innovating (read: manipulating) risk weightings and calculations. But true risk was contained in leverage ratios, not the accounting fiction of risk weightings. This banking group started 1990 with an aggregate leverage ratio of 23. By 2007, their collective leverage ratio was 35!
While shadow banking and securitization gets the blame for this overall trend, deservedly, it does not account for how this would get funded. I think it cannot be understated as to how big a role interest rate targeting, and the certainty provided to that marginal dollar cost, that allowed such monetary expansion. And the fingerprints of the Federal Reserve are not directly imprinted because the banks themselves are the direct source of that expansion - the Fed was only providing the liquidity backstop.
Without a symmetrical link between the cost of funding and the amount of funding there was no true restraint on risky activities - there was no link between the size of balance sheets and the expense for liquidity. As long as banks believed in the liquidity backstop provided by interest rate targeting in the Fed funds markets, they could reasonably and cheaply, with near-certainty, fund their leveraged expansions. There was no additional, marginal cost between increasing a bank's balance sheet to X+$1 or X+$10 billion. In a reserve-targeting system, there is a huge cost difference between those levels, and thus how risk is viewed between the next marginal dollar of expansion and the 10 billionth.
Leverage ratios exploded not only in the United States, but globally as eurodollar usage soared throughout the period, and the Federal Reserve stopped limiting eurodollar transfers and implicitly linked that market to the Fed funds market, and thus a potentially unlimited supply of dollars globally. So banks fully participated in leveraging themselves in US dollar-denominated assets, precipitating the asset bubbles that seem to be an inseparable feature (not a bug) under the interest rate targeting regime (where they were largely absent under reserve targeting).
The breakdown in eurodollars in August 2007 was the first test of the liquidity backstop. It would come to fruition in December 2007 with the advent of dollar swaps between the Federal Reserve and the various central banks across the world. By October 2008, the Fed was forced to make good on its backstop to the tune of nearly $600 billion in global dollar swaps, on top of domestic liquidity programs through many other channels. Altogether, total Fed "emergency" liquidity means have been tallied in the trillions, which is about proportional to the amount of unrequited dollars created by the banks themselves based on the Fed's promised backstop.
The repo markets have been trying to reduce that swell of leverage in the banking system since 2007. The dramatic decline in the collateral pool is a direct appeal toward retroactively enforcing a quantity constraint on banking. In other words, the repo markets since 2007 have been trying to re-impose reserve targeting upon the financial economy, where the limitation of "money" growth is defined solely by the quantity of usable collateral (to direct flow), fully ignoring any expansions of traditional bank "reserves" by QE or other means (increasing stock).
Unfortunately, if we take the Klee and Stebunovs paper as a serious attempt to understand the short-comings and incongruence of interest rate targeting as it relates to how the modern banking system has evolved away from unsecured lending systems post-2007, then we can anticipate that central banks are seeking to again remove quantitative restraints throughout banking. We know the ECB has done so by loosening or removing collateral eligibility requirements, seeking to alleviate the collateral shortage through largely nefarious means. We also know where this leads.
Central banks have been seeking dominance over banking since William Jennings Bryan's "cross of gold" speech. It has become settled monetary practice to not allow recession to remove imbalance upon the monetary system, rather to allow financial imbalances to build further in forbearance of creative destruction. Inevitably, it is the public/workers that bear the brunt of rebalancing, so the appeal is undeniable. It used to be that banks would also find themselves in the front place of such an unwelcome development, but then they realized the power of such a populist monetary stance as it would relate to perceptions of banks and bankers. Today we hear the trope of "saving banks means saving the economy" - in other words, avoid bank failures at all costs so that the real economy doesn't bear the brunt of monetary rebalancing. That formulation isn't exactly calibrated correctly, as we have seen. Banks have certainly succeeded in removing themselves from the pain of counterbalance, but for some reason the economy remains well within the crosshairs.
In practice, however, this monetary rebalancing is the cost of previous monetary imbalances. It is not the creative destruction that is to blame; it is the removal of quantity restraints and self-correction. It has been the same throughout the past century, where lack of some restraint on banking has always led to disaster. And central banks and economists have been near uniform in decrying their lack of "flexibility" in imposing their monetary engineering and searching for methods to solve the quantity restrictions where they inevitably appear. After all, they claim plausible deniability since they are not directly responsible for bank irresponsibility, they just provided the backstop cover.
The ongoing shortage of collateral is a very durable testament to the true size of what the financial economy should actually be. From there we can extrapolate the size of rebalancing and adjustments yet to be done. What the Great Inflation of the 1970's showed was that reserve targeting was flawed. Unfortunately, central banks took that to mean a "better" target regime was required, and thus interest rate targeting was conjured. Now we have the idea of repo rate targeting (and NGDP targeting), but perhaps it would be best if central banks give up their fetish for targeting altogether in the evidence of such escalating and debilitating disorder. Sure, they can easily achieve their targets but in so doing disastrously disrupt the natural economic and market processes into painfully resilient imbalances, eventually creating the very need for economic adjustments they use to justify all this targeting in the first place.