The trajectory of US interest rates depends on changes in expectations over its i) term structure and ii) term premia. We explore what truly matters for rates forecasts/positioning beyond the current dominant reflation narrative. We also discuss often-cited though we deem otherwise less relevant considerations.
Important Considerations
A. Embedding hurdle rates within valuations
Forecasting higher/lower rates than spot curves inevitably need to overcome embedded hurdle rates in valuations across various parts of the curve. We highlight the hurdle rates required for positioning/forecasting for higher spot rates below:
Short-ends: on the extreme short end of the curve (where term premium influence is negligible), Eurodollar futures, Fed Fund futures (FFFs), and OIS forwards have all but priced in a rates liftoff sometime in 2022. ED futures have now priced in 1.2x rate hike by end-2022, while 2Y3M OIS have essentially priced in close to 3 rate hikes by mid-2023 FOMC meeting. FFFs suggest liftoff sometime between Q3-Q422. Current pricing is more ‘hawkish’ than the Fed’s current dot plot (the median voter is guiding for ~two rate hikes by end-2023; although we question the usefulness of the Fed’s dot plot), and embed certain assumptions about the economy by end-2022. During previous FOMC meetings, the FOMC reiterated three criteria that would need to be satisfied prior to any consideration of a rate hike: 1) the labor market would need to reach levels resembling maximum employment; 2) inflation has risen to 2%; and 3) as assessment that inflation is on track to moderately exceed 2% for some time. For short-ends to lead spot curves higher, any rate hiking cycle expectations will need to be more aggressive than currently priced. This implicitly suggest that either (a) the above conditions will be met sometime earlier than 3Q22, (b) these conditions will be met by end-2022, coupled with strong momentum beyond (velocity to neutral quickens), or (c) the Fed changes its reaction function from its current guidance. Sequentially, it is likely that the completion of tapering will precede the first interest rate hike, hence moving rate hike expectations forward will also implicitly assume a quicker/earlier tapering program.
Medium-term Forward Rates: Market expectations of neutral/terminal – proxied via the spot 5y5y forward – are currently priced at around 2.13%. There has been a material re-pricing of neutral rate expectations (5y5y has risen 62bps YTD), with levels now higher than during the pre-pandemic levels seen in 2H19. For context, the peak FFRs post-GFC was similarly at 2.25%-2.50%, which on hindsight, turned out to be short-lived/unsustainable. Neutral/terminal rates expectations will need to be raised further and probably beyond the Fed’s perception of longer-run rates (currently ~2.5%) or past cycle peaks in order for medium-term forward rates to lead spot levels higher.
Breakevens: The YTD increase in optimism reflected in both the near-term liftoff and medium-term neutral rates is juxtaposed against the much more circumspect pricing with regards to medium-term inflation. An interesting aspect of the rates move YTD is the relative subdued performance of medium-term breakevens – 5y5y breakevens (@ 2.19%) are up 21bps YTD, while 5y5y inflation swaps (2.34%) are almost flat at +4bps YTD. The relatively subdued move in medium-term inflation forwards is driven by a decline in the 5s10s breakevens (-22bps YTD) with 5y breakevens (2.52%) up 55bps YTD and 10y breakevens (@2.34%) a lot more subdued at +35bps YTD. The market has priced in cyclically higher inflation (over the next 1-5 years), coupled with more subdued longer-term inflation maintained at a level close to the Fed’s 2% target (adjusted for the core PCE gap). Thus, going forward, markets will have to price in materially higher cyclical inflation >2.5% for inflation expectations to lead a further sell-off in the belly of the spot curve. On the other hand, 5y5y breakevens would need to rise to a level reflecting sustained core inflation above the Fed’s target for inflation expectations to drive a further bear-steepening of the curve on the longer-ends. For context, the US economy has failed to generate core inflation above 2% sustainably despite having much tighter labor markets and narrower output gaps pre-pandemic.
Real Rates: Establishing the hurdle rates for real yields would first require a determination of what real yields reflect in the current era/regime: a) an indicator to balance savings-investments, or b) a mathematical residual (nominal-breakevens) when central bank control over the nominal spot curve is high. For real rates to lead nominal spot levels higher for (a) require revised assumptions about the drivers underpinning the saving/investments imbalances. Proponents of (b) will find it difficult to envisage real yields leading nominal higher from here (with real increasing into deeply positive territory) unless one expects a material shift in monetary policy frameworks – we argued before (see this) that negative real rates essentially becomes an explicit policy objective under the AIT framework (i.e. nominal will lag breakevens in an upward trend; real yields typically rise only after breakevens decline with policy rates near the zero lower bound, hence accompanied by limited moves on the nominal). We estimate 5y5y real rates remaining in negative territory at -19bps, the rise YTD (~+37bps) mirrors the 5s10s flattening in breakevens (for context 5y implied real is largely unchanged YTD).
Term Premium: Establishing hurdle rates for term premia is highly difficult given the challenges in quantifying it with any precision, although we note the tendency for term premia to be negative post-GFC. Negative/low term premia is most probable when two key conditions co-exist: a) market conviction in central bank forward guidance to keep rates low is extremely high, and b) market conviction that the policy mix will fail to stimulate growth/inflation is extremely high. Indeed, episodes of term-premia led (change in ACM TP accounts for >70% of total nominal move) bear-steepening post-GFC were consistently accompanied by shifts in market perception regarding this dual-condition (albeit mostly temporary on hindsight): i) 2013 taper tantrum, ii) post-Trump elections in 2016, and iii) the 2021 YTD sell-off.
Hence, an even more meaningful term-premia led rise in nominal rates will require a more severe shift in market interpretation of CB forward guidance (this could either be a proactive change in reaction function or an unintentional loss in credibility), or a further change in market expectations over the ability of the current policy mix to impact growth/inflation (probably entail more meaningful shifts in ideology/scale/persistence/political constraints in delivering demand-side stimulus – see ‘regime shift’ section for elaboration). We doubt a term-premia led rise in rates can be unilaterally driven by the Fed – it seems premature to deviate from a newly-established framework, considering the importance of policy framework credibility (plus the risk that a hawkish deviation is also likely be interpreted as over-tightening). Risks to this dual-condition is also not one-way – e.g. if underlying conditions force a consideration of introducing YCC to keep long-ends anchored (i.e. a more permanent low/negative term premia state)/ a return to fiscal conservatism. We estimate 5y5y ACM TP at +38bps currently (i.e. positive TP is already embedded in the forwards); and this remains close to the highest of the range seen across 2018-2020 despite its recent correction (peaked at ~1.15% in Q121).
The relevant hurdles rates across various parts of the rates curves will need to be considered in totality, rather than in isolation (e.g. one may establish a view incorporating higher inflation going forward, but only under the condition whereby this is enabled by low nominals via monetary support – a bear-steepening and rise in forwards with short-ends near zero).
The hurdle rates to position for lower levels than the current spot curve will simply be the inverse of the assumptions stated above.
B. The Fed’s Reaction Function
We previously argued that the Fed’s new AIT framework remains very open to interpretation and accommodates a high degree of subjectivity. Key inputs to the framework remain unclear (e.g. the length of the inflation lookback window/what constitutes inclusive employment/the complementary conditions for sustained higher inflation), but reliable rate forecasts will need to lay out probabilistic outcomes about both the “path” (pace/timing of rate hikes, if any) and the “destination” (neutral/terminal rates in the policy cycle). Reflexivity will need to be considered too – e.g. the probability that a likely slower/later “path” vs. the old regime pushes up the “destination” of the new regime (this depends on one’s view of the extent neutral can rise due to supportive policies). Inversely, a reversion back to a Taylor Rule-like framework quickens the “path” but may decrease peak/neutral (i.e. markets pricing in over-tightening). The focus on the dot plots, we believe, is less warranted, considering its poor record as a guide to ex-post policy actions historically, and have complicated forward guidance. Indeed, the current dot plots have even less informational value than usual, considering the high level of uncertainty embedded in FOMC members’ assessments of future growth/UE/inflation (see uncertainty diffusion indices in the latest SEP).
C. Spot Levels &. Forwards Rates
The various connotations above will be captured by key segments of the forwards curve – e.g. views on the rates “path” will be expressed via the shorter-tenor forwards of the short-ends (2y2y/3y2y), while the medium-term forwards (5y5y) reflect the view on peak/neutral. Mathematically, spot curve forecasts will need to correspond to the implied forward rates (i.e. the explicit assumptions/rationale for spot curves forecasts must be consistent with/identical to the inherent assumptions embedded in the implied forward rates based on the same forecasts), else they become irrelevant.
D. Spot Levels vs. Curve Steepness
It thus follows that rates positioning should accommodate differing degrees of conviction for absolute spot levels (absolute directionality), vs. the steepness of curves. For example, a view for lower 2y2y/3y2y forwards (i.e. a later/slower “path” over a 2y-3y horizon) can be expressed via a 2s5s flatteners rather than outright spot levels (similarly a higher 5y5y can be expressed via a 5s10s steepener; not necessarily a higher spot UST10). Focusing on the steepness of the curve (vs. absolute spot levels) can remove an additional layer of complexity when positioning is led by views on forwards (as illustrated above, 5y5y can decline via a bear-flattener when markets price in over-tightening, or simply a decline in medium-term outlook on the economy – in this case potentially a bull-flattener)
D. Reflexivity
Rate forecasts will need to be compatible with the underlying structure of the economy. For example, rising bond yields in a highly-leveraged economy (especially as marginal productivity of debt is declining) may ultimately be self-defeating if financing costs rise. In addition, to the extent that the Fed remains sensitive to financial conditions (one will need to consider the level of the Fed’s implicit put price, which is negatively correlated to the strength of the underlying economy), the unwinding of carry/levered trades can act as a circuit breaker to any sustained rise in rates. Hence, the higher/more crowded the levered positions in both the actual economy and financial markets, the likelier that higher rates is self-defeating, and vice versa.
E. The Unhinging, Oscillation & Entrenchment of Regimes
We expect markets to increasingly shift its focus away from the volatile near-term data (where re-opening supply/demand mismatches in labor markets, and goods & services are likely to subside over time) to evolving probabilistic views over the medium-term outlook. The latter is increasingly dominated by market narratives surrounding ‘regime shifts/change’ – that we are currently at an inflexion point where markets are likely to transition to and reflect a new regime (one which is significantly more inflationary than the post-GFC era). Indeed, we have previously highlighted the underlying faultlines/limits of the current regime (characterized by low growth/inflation, aggregate demand deficiency, zombification, & lop-sided global external flows), and how they have been brought to extremes rendering the regime unsustainable. However, it remains to be seen if markets will fully transit from an unhinging of the old regime to the entrenchment of a new one. We see two main outcomes (with differing impact on rates trajectory) over the medium term (2-3 years): 1) transition into a newly-entrenched higher growth/inflation regime via a significant shift in major policy frameworks, or 2) oscillation among various plausible regimes and where key elements of the old regime remain, accompanied by circumstantial ebb and flow in major policy frameworks.
We will attempt to lay out in detail the conditions we think necessary for a permanent phase shift to a higher growth/inflation regime in future posts. These necessarily entail structural changes to the macroeconomic ‘superstructures’ in order for the drivers to be self-reinforcing and long-lasting. For the domestic economy, we suspect it will inherently be conditions which allow the rate of change in “wealth stock” & its inherent “velocity” to permanently increase (policy mix that entail some degree of redistribution, a material shift in power dynamics, more persistent and dominant demand-side fiscal policies, explicit/implicit debt forgiveness). On the external front, necessary conditions may include a proactive/growth-conducive rebalancing towards domestic demand from economies generating persistent external surpluses. A much higher degree of global coordination will also be needed to address structural issues with the global financial architecture.
Less Relevant Consideration
A. Size of Fiscal Deficits without Context
We have previously mentioned why one should not relate government debt accumulation/higher fiscal deficits (as % GDP) to higher interest rates (high government debt levels and low interest rates may at times correspond as they reflect similar drivers). Supply dynamics (a function of fiscal policy and the treasury’s issuance plans) have also historically had no direct relation to UST yields. Ultimately, the outcome of any amount of fiscal deficit/government debt is entirely contextual (the need to be analyzed in aggregate via the flow-of-funds and sectoral balance approaches) – we explained the key considerations and emphasized the importance of macroeconomic ‘superstructures’ when determining such outcomes.
B. Identities of Direct Buyers/Sellers
Often, we encounter examples of economists/strategists using identities of direct buyers/sellers of US treasuries as ‘drivers’ of market movements (e.g. rate declines are described as being driven by specific buyers; when rates rise, the focus turns to the identities of net sellers). However, we find gauging underlying UST demand via the factual identities of direct buyers & sellers to be uninformative at best and misleading at worst (notwithstanding the fact that for every market transaction, there must be both a buyer and a seller – yet few explore the identities of specific sellers when rates decline for instance). This is because the nature of flows into the global risk-free asset is unique vis-à-vis other asset classes, given its role in absorbing a significant portion of global ‘excess’ liquidity and currency transaction flows. Direct buyers of USTs partly reflects where excess liquidity sits, depending on how global financial flows are recycled, rather than necessarily a unilateral innate desire/motivation by specific buyers to accumulate US treasuries.
As a simple illustration, consider the contrast between a) a foreign country choosing to recycle its BoP net inflows via FX reserves (the foreign central bank will be recorded as the final direct buyer of USTs) vs. b) a foreign country recycling BoP net inflows via purchases of USD-denominated corporate debt through domestic agents (these inflows may eventually get recycled back into US treasuries via US corporates or the banking system). Both scenarios entail the recycling of external surplus (e.g. perhaps to prevent FX appreciation of the foreign country) as a key driver for US treasuries accumulation, but with contrasting identities of the direct UST buyer. Another straightforward example would be the implementation of QE – which constitutes an asset swap for the banking system via a switch from USTs to excess reserves (in this case, it would also be inaccurate to attribute any rise in term premia – as we had “counter-intuitively” witnessed during past periods of QE – to supposed ‘forced selling’ by the banking system).
Instead, the right question to ponder is how the recycling of flows impact global growth/inflation (e.g. can QE stimulate the economy? Or does easing beget more easing? Are external surpluses sustainable/recycled to productive purposes?), rather than precisely identifying the direct buyers/sellers of USTs. The answers are therefore entirely contextual/circumstantial; we argued – back in 2015 – why global capital flows in aggregate are unlikely to be absorbed in a growth-conducive manner, considering the current global financial architecture (the underlying reasons remain highly relevant even today).
C. Spot Level Comparisons vs. Pre-Covid-19
We have also encountered several strategists/commentators arguing for a near-term rebound of longer-end rates (10Y/30Y) to pre-Covid 19 levels or above, given improved cyclical growth/inflation dynamics. Such forecasts will need to specify if the spot curve will reflect i) a much higher forward rates curve with the short-ends near the ZLB (this implies a much higher neutral vs pre-pandemic; assumptions going beyond a strong cyclical recovery), or ii) an accompanying increase in short-end (this implies a near-term weakening of the AIT framework), with only moderate moves in medium-term forwards. Else, such forecasts violate bond math, with the embedded assumptions in long-end spot levels and the implied forward curve inconsistent.
D. Like-for-Like Taper Tantrum Comparisons without Context
With regards to tapering, context matters. Like-for-like comparisons to the 2013 taper tantrum should consider the following as well:
– The communicated sequencing (completion of taper to precede rate hikes) and short-end market pricing suggest tapering should no longer be a surprise to market participants.
– EM current accounts are less stretched vs. 2013 (market turbulence from “fragile-5”) in general; especially in Asia where dependence on speculative external capital flows is comparatively less.
– The 2013 Evans Rule was no longer a constraint for future rate hikes (thus taper announcement pulled in and pushed up rate hike expectations), this is in contrast to the AIT framework, which by definition, ensures a lower neutral and/or a slower path for the same level of inflation/UE rate vs. 2013 (again, unless one assumes AIT loses its credibility). The gap between the FOMC’s longer run projection and market pricing were also much larger in 2013.
– Fed communication is now much more frequent (press conf at every meeting) and incorporates feedback provided by a wider subset of market participants (via regular surveys) – this reduces the element of surprise which was a major factor back in 2013.
– The 2013 taper tantrum narrative was amplified by the potential rumors surrounding the appointment of Larry Summers as next Fed chair (who was perceived as much more hawkish relative to Bernanke/Yellen).
Conclusion
We find rate forecasts based upon well-substantiated assumptions on the relevant considerations more reliable. On the contrary, forecasts largely premised upon what we deem as irrelevant drivers are less so. Positioning should also reflect one’s degree of conviction on probabilistic outcomes of the relevant drivers.
Note: All spot/forward rates cited above reflect 02 July 2021 EOD levels.
