In Fed Chair Powell's semi-annual testimony to Congress, the former lawyer underscored the fact that the Fed is nearing the end of its balance sheet reduction process. He reiterated that the Fed expects to end balance sheet runoff “later this year” and added that many key details had been worked out. He said: “My guess is we’ll be announcing something fairly soon.”
The announcement was, and was not a surprise.
It was not a surprise because the Minutes of the January 29-30 FOMC meeting had already made it clear that the Fed will end its balance sheet reduction program or Quantitative Tightening (QT) earlier than had been previously envisaged, perhaps as soon as this year, nearly one year ahead of schedule and expectations, which until recently had clustered around 2020.
Where it was a surprise is that initially the Fed expected that its process of balance sheet reduction, which did not involve outright sales of securities and only maturities and redemptions, would have minimal impacts on financial markets (instead of full-blown QT it was more like QTlite). However, as Q4 began and balance sheet reduction ramped up to its maximum permissible pace of $50BN per month on some occasions, the equity market weakness and volatility that unfolded during the period appeared to be more than just a coincidence, particularly in the market’s mind (from its September peak, the S&P 500 had plummeted to a bear market by Christmas Eve).
To be sure both bond and stock markets were delighted by the Fed caving to the demands of "market participants", and quickly priced-in this fact. However, as BMO's Ian Lyngen writes in an extensive analysis overnight, further market movements related to the end of QT will depend on timing and the final details.
To be sure, following the Fed's tacit admission of an early end to its balance sheet runoff has brought renewed risk appetite to markets. The S&P 500 is up 11% YTD. The LEMBI index of local currency emerging market bonds is up 5% and the Treasury-to-BBB spread has narrowed by 25bps. With the markets presumably having priced in at least the confidence aspect of the end of QT, future price responses will depend on the details of what the Fed ultimately announces.
The most crucial detail for markets is the end-date of the program. When the Fed announces the end date and whether or not the Fed will ‘taper’ the end of portfolio runoff are secondary, but noteworthy factors for markets. As we reported last week, Goldman is confident that during its March meeting the Fed will announce that the Fed will end the runoff by the end of Q3. Other details that will have an importance to at least some markets include whether and how fast the Fed transitions its MBS portfolio into Treasury instruments, how the Fed transitions the duration of its portfolio and specifically whether it will move back into purchasing T-bills.
Below, BMO lays out the market impact of four of the most likely scenarios and some of the technical issues confronting the FOMC as it prepares to communicate its final plans. In other words, this is the way QT ends:
Estimating the End Date
Most strategists, FOMC members and BMO itself, estimate the underlying demand for reserves at around $1 trillion. This level (and its subsequent growth over time) reflects total banking system assets and regulatory changes that require banks to hold larger buffers of high-quality liquid assets, of which interest-earning deposits at the Fed (a.k.a. reserves) are a key component (around 30%, on average). In other words, as excess reserves are reduced to around $1 trillion by the end of 2019, put a different way, QE3 and QE2 will have been undone and only the effect of QE1 will be left (see Nordea chart below to the left).
The Fed has also messaged that they would like a cushion over the $1 trillion to avoid active reserve management. That buffer will likely be initially set at $200 billion. This results in a $1.2 trillion "target" with reserves currently running at just above $1.6 trillion. Note that they were $2.8 trillion at their peak in October 2014 and $2.2 trillion when QT started in October 2017.
Given the scheduled maturities of Treasuries (and the current $30 billion monthly cap), estimated MBS redemptions (and the current $20 billion monthly cap), and projected currency growth, BMO estimates the $1.2 trillion reserves target will be hit in September. This assumes that Treasury deposits at the Fed return to recent levels after being temporarily drawn down due to a delay in lifting the debt ceiling (which will have the temporary impact of boosting reserves).
Incidentally, BMO agrees with Goldman in expecting the FOMC to officially call for QT to end by Q4 (i.e. September hard stop), with that announcement coming at the March meeting (or May at the latest). After Powell’s comments in his testimony, there's no point in delaying this specific announcement (not all details have to be released at the same time).
The bank's base case is that the Fed will continue the current SOMA reduction cadence until the $1.2 trillion reserves level is achieved, and that monthly caps will not be lowered (which could potentially push things past year-end anyway). The Fed tends to "taper" the removal of market-friendly policies like QE to ween the market off of them and avoid a potential negative response to their conclusion. QT is a market-unfriendly policy, and concluding it should only have a positive response, so there is no need to ween. Also, the Fed clearly wants to keep MBS holdings shrinking, so why change the maximum permissible pace of $20 billion per month which has been in place for the past five months? This would set up an awkward communications situation of tapering one aspect of SOMA reduction but not the other.
A follow-up announcement could come in March, May, June or July. The follow-up announcement of technical details would cover the Fed’s plans for normalizing the asset mix of its portfolio. In addition to confirming that MBS holdings will continue to shrink after SOMA's size is stabilized, BMO expects the technical note will also state that the Fed’s future purchases will exclude MBS in favor of Treasury instruments and will likely include a growing portion of T-bills.
End of QT Scenarios
Given the multiple dimensions in which the Fed could choose to act (QT end date, announcement dates, taper), there are a myriad of potential scenarios of how things could play out from here. For simplicity, BMO has chosen to reduce the issue down to four scenarios that span the spectrum of what is plausible. Those scenarios are summarized in Figure 1. Note that the table transitions from most dovish on the left to most hawkish on the right.
BMO's base case scenario is the third from the right that we have named ‘September Sudden Stop.’ We have assigned it a 55% probability. The key arguments for and against each scenario follow.
- June Sudden Stop: The principal argument for ending QT as early as June would be to get the issue out of the spotlight so that the Fed can revert back to steering the economy and communicating with markets via its base rate corridor. The argument against this scenario is that ending QT in June could leave the Fed with too big of a reserve cushion, which might reduce its ability to control markets in the future. For that reason, the bank assigns this only a 20% probability.
- Tapered June Stop: The argument for this scenario is that the Fed wants to move on, but it also wants to experiment just a bit. Such experimentation could benefit the Fed in the future and could serve as a favor to other central banks lagging behind the Fed in QE reversal (BoE, ECB, BoJ). But because "taper" is an unnecessary complication, BMO only assigns this scenario a 10% probability.
- September Sudden Stop: The key argument for this approach is that it is the closest to perfect in terms of reducing the excess supply of reserves. Ending in September prevents the Fed from either causing or being blamed for a year-end liquidity event. Also, by holding back the announcement of details and end date for as long as possible, the Fed retains the flexibility to ease policy a bit by pulling forward the QT end date in the event that economic confidence sours again.
- Tapered December Stop: The argument for this scenario is that it gives the Fed the most time and flexibility. By providing a taper, the FOMC might offset disappointing markets with a later end date than has been hope for by market participants. Where Lyngen doesn't think a taper is necessary and since the Fed likely doesn’t want to get anywhere near year’s end with policy action on this front, BMO only gives this scenario a 15% probability.
End of QT Scenario Responses in Markets, US Yields and the USD
The bulk of the market response to the end of QT has probably already occurred. As noted above, the S&P 500 is up 11% YTD, and that is probably the most important price response. The response in interest rate markets has been less dramatic. The yield response has presumably been muted because the impact of the equity market reaction has counteracted the dovish evolution in Fed policy expectations. As a result, BMO does not look for a huge price reaction to any of our scenarios, but it does think that a bit of a response will show up when the end of QT is confirmed by the announcement of a clear termination date.
It is worth a note that the impact on 10-year yields is very indirect. If the FOMC doesn’t formally announce an end date in March, yields would back up about 5bps, but other than that, impact further out the curve should be marginal.
The impact on T-bills will be more flow driven than announcement driven. There is the impact on 3-month bills vs. OIS maxing out at 10bps if QT ends in H1, but at only 5bps if QT extends into H2. Figure 2 shows how Lyngen thinks the US rates market would respond to our four scenarios.
In FX, the response in the USD has been somewhat bifurcated. The announcement effect of QT coming to an end has caused only a slight decline in the USD against G10 currencies, but there has been a more noteworthy USD decline relative to emerging market currencies. The Fed’s ‘AFE’ (Advance Foreign Economies) USD index is down 0.6% YTD. The Fed’s ‘EME’ (Emerging Market Economies) USD index is down 1.4% YTD.
While further impact is likely to be limited, BMO believes that a firm announcement of an end-date to QT will have a second-wave negative impact on the USD, and as a result expect that impact to be greater on local-market currencies than on G10 currencies. See Figure 3.
The MBS vs. Treasuries Issue
As Lyngen notes, the January FOMC minutes indicate that participants agreed that “it would be appropriate once asset redemptions end to reinvest most, if not all, principal payments received from agency MBS in Treasury securities.” As a result, the Fed will likely adopt a policy whereby all MBS redemptions are reinvested in Treasuries unless a sudden increase in MBS redemptions occurs, stemming for example from a wave of mortgage prepayments. The shift of reinvestment from MBS to Treasuries will be marginally bearish for spread products, although the Fed has not been hitting its caps, and thus has not been reinvesting its MBS redemptions since the caps maxed out last October. Further, estimating MBS portfolio evolution using an 8% runoff, significant monthly maturities in excess of $20 billion going forward are not forecast (Figure 4).
Additionally, if the Fed opts to keep a backstop in place for the MBS market to guard against mass redemptions, the current $20 billion cap seems a reasonable level, so it is possible that there is no change in Fed policy as it relates specifically to MBS, and by extension spread products. However, it also seems reasonable that the Fed raises this cap alongside the change in reinvestment policy to something much larger, say $50 billion, so that reinvestment into MBS only occurs following a truly significant wave of prepayments.
On the other hand, the imminent start of a move towards a Fed balance sheet composed primarily of Treasuries would be bullish for Treasuries. The most bullish outcome is a sudden stop in balance sheet normalization beginning in July. Between Treasury and MBS runoff, the Fed would purchase about $298 billion in Treasuries under this scenario. BMO's base case of a sudden stop in September results $203 billion in Treasury purchases by the Fed (Figure 5).
Impact on the US Government’s Net Issuance
An earlier end to Fed balance sheet normalization coupled with Fed reinvestments of MBS runoff into Treasuries results in the Fed buying more Treasuries than previously expected. This decreases Treasury net marketable borrowing from the public, specifically bill issuance, all else equal. The decrease in supply will likely richen Treasury-OIS, with the degree of richening depending on the timing of the sudden stop.
Treasury is likely to hold coupon auction sizes at the current levels as deficits are projected to continue to increase over the next several years. Therefore, the decrease in funding needs translates to a mirror decrease in net bill issuance held by the public. Even if the Fed decides to reinvest some of the proceeds from maturing securities into bills, the impact on net bills available to the public would not change.
BMO's base case is that they simply end runoff, roll MBS maturities into Treasuries, and continue to reinvest as add-ons to the auctions until the SOMA portfolio begins to grow again. The expectation for runoff to end on September 30 is about one year earlier than originally expected. This reduces Treasury net marketable borrowing from the public by about $350 billion, with about $230 billion coming from the halt in Treasury runoff alone and another $120 billion from rolling MBS into Treasuries (assuming 8% in prepays). This decreases Treasury marketable borrowing by 33% from Q4:19 to Q3:20.
In terms of net bills issuance, BMO had expected about $539 billion in issuance during this period and now expects $184 billion as shown in Figure 6. In the bank's base case, Q4:19 net bill issuance totals about $77 billion, down from $166 billion. In the first three quarters of 2020, Lyngen now expects $107 billion in net bill issuance, down from $372 billion with $177 billion coming from SOMA Treasury rollovers and $88 billion from MBS rollovers into Treasuries. Figure 6 also shows updated net bill issuance projections with and without MBS rollovers to put a band around issuance possibilities given the uncertainly in prepayment speeds. In table 8, BMO shows three additional scenarios. Spanning the scenarios, net bill issuance would fall by $41 billion at a minimum and $184 billion at a maximum in 2019.
The Fed’s Portfolio Duration Issue
According to BMO, in designing the end of portfolio normalization, the Fed will need to decide whether to use its holdings as a means by which to influence financial conditions. They could achieve this by structuring holdings longer or shorter versus the WAM of outstanding Treasuries. In essence, if the Fed wanted to contribute to lower borrowing costs, it could disproportionately focus reinvestment purchases further out the curve (and vice versa), creating a (reverse) Operation Twist. This leaves the core question as to whether holdings will converge to passively reflect the distribution of publically available debt or not. For context, current holdings skew longer in duration, partially due to the lack of Treasury bills held in the SOMA, as any maturing proceeds are presently distributed across coupon auctions.
For historical context, pre-crisis SOMA holdings skewed shorter in maturity than the aggregate weighted average maturity of outstanding Treasuries. This was done in order to maintain a highly liquid portfolio, and included 35% to 40% of holdings in Treasury bills. However, since 2008, the component of holdings represented by bills has been completely eliminated. The Fed has concentrated purchases further out the curve in an attempt to suppress longer-tenor yields and bring down borrowing costs for businesses and households. That degree of accommodation is no longer justified at this stage in the cycle, so the Fed will likely rebuild their bill portfolio in order to better reflect the actual Treasury market.
In addition, focusing holdings on a more realistic reflection of the Treasury market allows for another ‘twist’ in future years. Said differently, the impact of extending SOMA duration to provide future accommodation is likely greater when coming from a lower starting point. With neutral rates significantly closer to the effective lower bound than during previous cycles, the central bank may need every tool at its disposal, including adjusting the duration of holdings as required.
On the whole, BMO observes that exchange rate movements have mapped reasonably well with the relative sizes of central bank balance sheets over the last 8 or 9 years (see charts below). But estimating the FX impacts of balance sheet shifts and the timing
of those impacts is not an exact science. In fact, it is extremely imprecise.
Economic theory argues that an increase in the relative supply of a currency – as one result of a QE program – should tend to weaken that currency, while a drop in the relative supply of a currency should do the opposite. However, despite what economic theory regarding QE posits, the Fed’s broad USD index rose about 8% during QE2. QE3 also brought USD strength.
BMO here notes that it does not think that any of the major central banks will change their balance sheet policies in response to the Fed ending QT. The ECB is likely to keep its balance sheet static, while the BoJ is likely to continue slight balance sheet growth as a result of its QQE program, at least for the foreseeable future absent any unexpected spike in inflation (or bond market breakdown).
This means that if the implementation and end of QE didn’t impact major exchange rates in textbook fashion, there is no guarantee that the end of QT would bring a textbook response. But with the red lines in the two figures below flattening out to horizontal (in the case of the Fed/ECB ratio chart) or near horizontal (Fed/BoJ case), the textbook response is flat ranges in major exchange rates.
The risk-positive implications of ending QT indirectly benefit other major economies and their respective central banks because of its loosening effect on global financial conditions. This should offset the slight weakening of the USD and keep everyone in the room happy when G20 central bankers gather.
Meanwhile, emerging central banks that have currencies pegged to the USD or that are tightly aligned with the USD (including the
PBoC) will indirectly benefit from both a rebound in capital inflows as well as slightly weaker trade-weighted currencies. This could lead to the resumption of a modest amount of FX reserve accumulation by the aggregate of EM central banks. That reserve accumulation would lead to modest purchases of sovereign bonds (including US Treasuries), which would further ease global financial conditions.
Finally, the Fed’s clear path to ending QT and policy rates likely being on hold for some time, reinforces that the Bank of Canada is going to stay on the sidelines through at least the first half of 2019. And yet, while BMO's house call is for one more BoC rate hike
in December (Friday's shocking GDP print notwithstanding), it’s going to take a more positive global backdrop and likely a more positive tone from the Fed before the BoC can turn more hawkish. That should keep the front end of the Canada curve range-bound, while longer-term yields are expected to largely follow UST’s in their reaction to the end of QT.