The Furnace

#21: Shrinking the Balance Sheet: The Central Bank's Dilemma

In the middle of my grandmother’s house stands a cast-iron wood stove. For years, the stove provided heat through the blistering Western New York winters. First, nightly fires offset the evening chill. Then, as autumn gives way to bitter winter and banks of snow grow, the wood stove runs around the clock.

The fire, when tended properly, can last months. Sometimes it is fanned into flames that radiate heat and light across the room. Other times it is barely visible, smoldering in a bed of embers. But the fire burns constantly, consuming piles of dried wood, stacked meticulously throughout the summer.

Tending the stove requires both constant attention and an artful hand with seasons of experience. Wood is the base fuel, while a tweak of the flue adjusts the flow of oxygen and the pace of the burn. A skilled hand must anticipate and counteract the loss of heat through windows and walls with a properly calibrated flame, while minimizing the consumption of wood.

Growing up, I looked on longingly at the seemingly simple task. Building a fire is fun, and as a kid there is only one way to do it - bigger - pile up the logs and open up the intake. But my unskilled touch would throw off the delicate balance that kept the house just right.

For the last year I have found myself repeatedly returning to the idea of the wood stove, or perhaps a furnace, as a lens to understand the purpose, mechanisms, and challenge of monetary policy. Bear with me.

The Furnace

Fire is a well-worn analog for economic and monetary policy. As in, the economy is overheating, or, the Fed must fan the economic flames.

However, the analogy of a furnace illuminates a more nuanced lesson about the mechanisms of modern monetary policy as well and the challenges that arise.

The central bank is in charge of tending the fire, with the goal of price stability and maximum employment. The challenge, however, is that it has little direct interface with most of the economy. It doesn’t underwrite loans, approve mortgages, or provide financing for the everyday businesses.

Instead, this economic combustion occurs in the private sector, between commercial banks and their customers. The Fed, with limited direct connection with the rest of the economy, must instead work to create the optimal conditions in the banking sector to generate the right heat.

The traditional tool is interest rates. By setting short-term interest rates, the Fed can encourage or discourage demand for credit. Think of this as the airflow. Cutting the oxygen supply (via rising interest rates) will reliably kill the fire, while opening up the flue can coax flames from coals.

Meanwhile, the Fed also controls the amount of coal in the furnace, by giving commercial banks more capital to distribute through the economy and transform into real-world spending.

Prior to the global financial crisis (GFC), the Fed took a delicate approach to each variable, setting headline interest rates to counterbalance the business cycle while using minor open market operations to ensure interbank markets remained in line with their rate targets.

In 2008 the fire went out.

To counteract the dramatic economic downturn, the Fed lowered interest rates to zero - opening the airflow all the way. But there was another problem. After years of lax regulation that allowed commercial banks to become ever more levered on a cash basis (the U.S. banks were levered 33:1), the furnace was desperately low on fuel as well.

To address this emergency, the Fed began conducting large scale asset purchases to recapitalize the banking sector, refilling the furnace with coal in the process.

While the Fed saved the banking sector in 2008, the economic recovery remained lackluster for years. With rates already at zero, there was little the Fed could do to fan flames.

So it opened up the door and started shoveling in more coal. The intent of QE from 2011 - 2014 was not to add fuel, but rather about juicing airflow by opening the front door of the furnace - attempting to lower long-term interest rates. Banks accumulating mass piles of cash was a secondary effect.

The risks of enormous balance sheet expansion were written off, especially after initial warnings of inflation proved wrong.

It became conventional wisdom to assume that loading banks with capital has little impact if there is no demand for credit. As anyone who has tried unsuccessfully to build a fire knows, throwing more wood on the stack doesn’t help if the conditions aren’t right.

So from 2008 to 2014, the Fed’s balance sheet expanded five fold. Policy makers must have had extraordinary confidence that adding so much coal to the furnace would have little consequence and could be quickly reversed, otherwise their actions would appear quite reckless. As easily as the Fed’s balance sheet grew, they assumed it could be shrunk.

The trouble, though, is that shoveling coals into the furnace is very easy to do. Getting it out the furnace once the fire gets going is an altogether different challenge.

Uphill Task

In August, a team from the University of Chicago and New York University including the former head of the Reserve Bank of India, Raghuram Rajan, published a report that caught my eye: Liquidity Dependence: Why Shrinking Central Bank Balance Sheets is an Uphill Task.

In the Abstract, they summarize their concerns:

The Federal Reserve expanded its balance sheet via large-scale asset purchases (quantitative easing) in recent years […] However, when it halted the balance-sheet expansion in 2014 and even reversed it during quantitative tightening starting in 2017, there was no commensurate shrinkage of these claims on liquidity. Consequently, the past expansion of the Fed’s balance sheet left the financial sector more sensitive to potential liquidity shocks when the Fed started shrinking it, necessitating Fed liquidity provision in September 2019 and again in March 2020. If the past repeats, the shrinkage of the central bank balance sheet is not likely to be an entirely benign process.

In simpler terms, the paper suggests that shoveling coal out of a furnace is a tough task. You don’t say!

The ability of a central bank to shrink its balance sheet rests on the notion that there are truly excess reserves in the banking sector that can be removed without consequence. But what the paper discovered¹ is that the trillions of dollars that were pumped into commercial banks over the last decade were not idle. Rather, an increasing amount of claims were written against them over time, even after the Fed began shrinking its balance sheet via QT.

It was only then that the Fed discovered exactly how much of the supposedly excess coal had caught on fire - and it was much more than it expected.

In other words, the Fed discovered that when given capital, banks will do exactly what they are supposed to do, and converting the fuel to economic flames, spreading dollars through the real-world economy.

Cash vs. Deposits

Consider the cash balances of the banking sector (both currency and bank reserves²) compared to the banks' demand deposit liabilities. Demand deposits, such as checking accounts, are available for immediate withdrawal and are the most “spendable” form of money in the real world other than physical currency.

In other words, how much cash do banks hold compared to how much cash their customers could theoretically demand from them today.

Prior to the financial crisis, the two balances remained very close to each other. However, with the advent of QE, the cash holdings of commercial banks exploded much higher than demand deposits. At its peak in 2014, the banking sector had twice as much cash on hand than customers could withdraw. The banking sector was awash with excess reserves and excess liquidity.

As the economic recovery continued through the decade, the Fed pivoted from easing to tightening, raising interest rates and removing cash from the banking sector via QT in 2017.

Yet, even as the central bank drained commercial banks of their excess cash, demand deposits continued to grow. By September 2019, the two series nearly converged for the first time in the QE era.

Then, on September 17, 2019, money markets froze and interbank offer rates exploded, forcing banks and dealers to borrow from the Fed’s repo window for the first time since 2009.

After taking extraordinary measures to prevent another financial blowup, including increasing bank cash holdings by 10x between 2008 and 2014, the Fed was forced to fix a liquidity crisis, this time of its own making.

The Repo Crisis was an awkward moment for the Fed, which led to headlines such as “Fed Unveils Plan to Expand Balance Sheet but Insists It’s Not Q.E.”. But even more worrying was how little the Fed seemed to understand its own liquidity policy, and the ability to reverse the extraordinary monetary accommodation of the past decade.

The experience also raises questions about the long term effect of QE, that runs contrary to the orthodoxy adopted by central banks globally. Namely, that QE does not create inert reserves, but instead the cash is digested by banks and trickles through the economy with a lagging but meaningful effect on real world money creation³.

Even QT could not stop the growing fire, powered by coal that was shoveled into the furnace to influence interest rates years earlier. Attempting to slow the spread resulted in a liquidity crisis.

Then came COVID.

Tightening Today

Any concerns around the 2019 Repo Crisis would quickly be sidelined by the onset of the COVID-19 pandemic, which brought with it a monetary expansion that dwarfed the prior interventions. But this time, QE came with an important twist.

While QE provided liquidity directly into the US Treasury and Mortgage Backed Security markets (and cash to banks in the process), it also provided the funding for the federal government to embark on the historic pandemic stimulus.

Unlike in the prior decade, where QE dollars trickled down to spendable cash over time, QE dollars were sent to checking accounts via direct deposit from the government.

The marriage of monetary expansion with fiscal handouts meant that the cash increase of the banking sector was mirrored almost exactly by an increase in demand deposits. In other words, despite the massive increase in cash, banks are not awash with excess liquidity in the same way as they were in 2014.

Today, U.S. commercial banks hold $3.3 trillion in aggregate, twice as much as they held at the start of the pandemic. But the collective customer base of those banks now has $4.9 trillion of demand deposits - far more cash than the banks hold.

This is a simplistic comparison, and so far we have not seen a repeat of the 2019 episode⁴, but the setup is worse. Since its peak a year ago, banking cash has decreased by $750 billion⁵, while at the same time demand deposits have grown by nearly $500 billion.

Here’s the kicker: Headline QT has just begun.

In September, Quantitative Tightening has now reached its run-rate of $95 billion per month, or $1.14 trillion per year. The Fed expects QT to “run in the background” draining far more cash out of the banking sector over the coming year. But while the Fed pulls cash from banks, it has far less ability to reduce the demand deposits of banking customers.

This poses two immediate problems for the Fed.

• First, tackling inflation would be easier if the Fed could pull money out of people’s checking accounts (reversing the effect of fiscal deposits). Instead, it can only reduce banking liquidity which has a lagging and much less direct effect on consumer spending.

• Second, the more aggressively the Fed tightens to fight inflation, the more likely it is to create a liquidity problem which would force yet another balance sheet expansion - another round of “Not Q.E.”

The current path ensures there will be a liquidity shortfall somewhere as QT continues that will require yet another emergency corrective action to fix.

Shortfall

Where might the next liquidity crisis occur?

Post-GFC liquidity and leverage regulations ensure that no U.S. bank will ever run out of money. But tighter regulations may force banks to stop lending to important non-bank entities and can actually cause a financial blowup on their own as was the case in 2019.

Money market funds - long seen as a source of systemic risk - have been effectively backstopped by the Fed’s reverse-repo facility (RRP). The Fed’s Standing Repo Facility (SRF) also provides an emergency backstop to a wider range of participants. But each new tool the Fed creates to mitigate risk merely transfers risk down to the next link in the chain.

While it remains speculative, an under appreciated risk is the US Treasury (UST) market itself.

While USTs will never actually default, they can still drop in value rapidly in a scramble for liquidity, causing chaos in the process.

The liquidity in the US Treasury market is terrible and worsening⁶. The largest buyer of USTs for the past two years has been the Fed. Now, the Treasury must find a huge new marginal buyer for a growing wave of issuance just as its largest buyer steps away and liquidity in the financial system dwindles.

On the short end, government money market funds meanwhile have reduced their holdings of USTs in favor the RRP. On the long end, banks have been buyers of USTs over the past several years, but may have limited capacity or willingness to step into duration as the Fed hikes rates.

While it may sounds outlandish, it has happened twice in recent history. The UST market broke down both in 2008 and 2020. In each case, the market was saved by direct intervention of the Fed, with its balance sheet exploding in the process. With 8% inflation, will the Fed be forced to rev up the printer yet again?

The longer-term problem for the Fed is existential. Remember the furnace?

Conclusion

Policy makers have assumed that shoveling coal into the furnace would have little negative consequence. Excess liquidity was always assumed to be reversible if inflation took hold. If this theory breaks down - if reserves are permanent - it re-casts the entire monetary policy of the QE era in an ominous light.

Today, inflation is here and sticky. If balance sheet expansion proves to be a self-perpetuating trap, leading to ever expanding supply of dollars, the challenge of inflation may prove intractable.

To quote from the conclusions of Liquidity Dependence:

Central bank balance sheet expansion may be harder to reverse than earlier thought and a part of it may be irreversible due to hysteresis and financial stability considerations, our work suggests careful reconsideration of the merits of quantitative easing.

For long time readers, this all sounds familiar.

---

As always, thank you for reading. If you enjoy The Last Bear Standing, tell a friend! And please, let me know your thoughts in the comments - I respond to all of them.

-TLBS

1. And has been conceptually and observably obvious for years.

2. For clarity, “cash” does not mean “currency”, it means US Dollars. A bank can hold US Dollars in the form of physical currency or in bank reserves at the Fed. This is not a simplification on my part. The H.8 commercial banking report published weekly by the Fed includes just one line that includes both currency and reserves - “Cash Assets”. The attempt to draw some sort of distinction between bank reserves and currency has caused a lot of undue confusion.

3. As discussed at length in QE for Dummies.

4. For example, prior to 2000, demand deposits exceeded bank cash on a fairly regular basis. However, looking in percent terms, the level of demand deposits relative to cash is higher than it has been since the mid-1990’s, and perhaps more importantly, has accelerated upwards at an unprecedented rate and will continue to so as QT commences. Further, liquidity regulations are tighter than ever.

The reality is there still has yet to be a satisfying explanation that has pinpointing the exact cause of the September 2019 Repo Crisis. The comparison between the demand deposits and cash balances demonstrates an important point about the liquidity mismatch of QT, but it certainly is not the only factor impacting banks’ willingness to lend.

Indeed other constraints (namely, leverage regulations) could have been the binding factor. Removing cash also increases a banks leverage. Cash leverage levels at banks ran from under 5x in 2014, and accelerated to 10x by the time of the blowup. Today, leverage sits between 7-8x. As QT commences, these leverage levels will rise.

5. Due to reverse-repo (RRP) uptake sucking up dollar liquidity from non-bank entities as well as the net build in the Treasury General Account (TGA).

6. A trouble that the Fed has acknowledged to some consternation.