Fed comments at Jackson Hole indicate that they are in wait and see mode about both domestic data and financial market/China stresses. But for now they’re apparently sticking to their 2015 liftoff plan and not ruling out a September rate hike. Specifically, Stanley Fischer argued “Given the apparent stability of inflation expectations, there is good reason to believe that inflation will move higher as the forces holding down inflation dissipate further” and concluded “I will not, and indeed cannot, tell you what decision the Fed will reach by September 17”. US rates have sold off a little as a result, but nevertheless are only pricing around a 40% chance of a September rate hike, rising to around 60% for a December liftoff. So the market would apparently be surprised by the Fed following through on those words. Friday’s payrolls release, particularly the earnings component, is the near term market focus.
So it’s important to confront what is likely to happen to USD with an (unexpected) Fed rate hike. And I illustrate that on average USD has fallen in the months following previous four Fed rate hikes periods, representing an important prima facie challenge to current USD bulls. Moreover, ‘buy the rumour, sell the fact’ behaviour is an inherent aspect of exchange rates’ forward-looking nature, which strong positioning behaviour (as per recent stretched USD long positions) can exacerbate.
But several more detailed findings indicate that USD is not necessarily doomed to again depreciate after a 2015 Fed rate hike: (i) USD did not usually appreciate prior to Fed hikes (not much buying the rumour!); (ii) initial USD falls after Fed hikes were usually relatively small, indeed USD sometimes appreciated immediately afterwards; (iii) US yields generally continued rising after Fed hikes (not much selling the fact!); (iv) despite that USD was not supported because non-US rates rose as much or more, resulting in some OK correlations with USD moves, but this seems less likely to happen this time around; (iv) USD’s moves after the June 2004 Fed hike, which are initially the most puzzling, also seem driven by factors unlikely to be repeated – higher oil prices driving a record US current account deficit, US inflation and inflation breakevens rising, US investors buying substantial foreign securities and USD acting as a funding currency.
USD nevertheless faces challenges with a prospective 2015 Fed rate hike – the unwind of stretched USD long positons (which also seems to have contributed somewhat to USD’s 2004 fall) and likely support to EUR, JPY and CHF if risk appetite is damaged seem the most prominent in G10. So USD longs against commodity-exposed G10 currencies and fragile EM currencies seem most attractive. But there would also seem to be diversification benefits in selected EUR/JPY/CHF positions of against commodity-FX and fragile EMs ahead of a Fed hike, especially give Fed messages from Jackson Hole. Obviously everything is extremely data-dependent at present, so we could be in for a very interesting few weeks. My argument that an October Fed hike shouldn’t be ruled out received some support from St Louis Fed Governor Bullard at Jackson Hole.
USD has tended to fall after previous Fed rate hikes but there are important nuances
The charts below illustrates that on average USD has fallen in the four previous Fed rate hike periods (April 1987, February 1994, June 1999 and 2004). The big picture is that the Fed USD majors TWI was on average 4% below its pre-Fed move after six months, with the fall peaking at just over 6% nine months after the Fed started hiking. Results are similar for DXY.
But there are some important nuances/details within that big picture.
First, the USD falls were initially relatively small: USD TWI was on average only 0.1% down one month after the Fed hiked and 1.3% lower three months after. And obviously other US and non-US factors are more likely to affect TWI the further away we go from the initial Fed move.
Second, USD TWI actually initially rose in the two weeks immediately following the 1999 Fed move, before then heading back down. And while USD initially fell in the two weeks after the June 2004 Fed rate hike it subsequently bounced back and was generally range-bound before starting to depreciate more significantly from mid-October (i.e. a bit of a lag from the initial Fed hike, although Fed rate hikes were ongoing).
Third, USD’s behaviour around Fed hikes isn’t really ‘buy the rumour sell the fact’:.
• That pattern only really happened around the February 1994 Fed rate hike – when USD TWI had appreciated 1.5% (2.5%) in the three (six) months prior to the Fed action only to deprreciate 2.5% (4.8%) in the following three (six) months.
• While USD was also mildly appreciating prior to the June 1999 rate hike (up 1.7% in the previous three months) that rise initially continued for a couple of weeks after the Fed started hiking.
• In contrast, USD was already depreciating prior to the April 1987 and June 2004 Fed rate hikes. Indeed, after initially falling marginally after the April 1987 Fed hike USD TWI subsequently appreciated reasonably sharply: it was 4% above the pre-hike level four months after the Fed moved. So the 1987 Fed hike can potentially be interpreted as having temporarily interrupted USD’s previous trend depreciation.
• The only way you can really conclude that USD was “buying the rumour” prior to Fed hikes is to exclude the 1987 Fed hike (which looks a bit arbitrary, although it’s obviously the most dated case study) and focus on a pretty long window like six months in order to conveniently overlook USD’s depreciation in the two months before the June 2004 Fed hike.
While those basic facts are thought provoking, they only take us so far in determining whether the past is likely to repeat itself around any forthcoming Fed rate hike. To do that we really need to first dig deeper and try and infer why those developments occurred. And we then need to think whether the same forces are likely to apply this time around (or whether other will be more important).
Interest rate differentials reveal further interesting facts
And analysing movements in US and foreign yield curves produces several important findings.
First, USD’s post-Fed hike falls generally don’t reflect ‘buy the rumour sell the fact’ in US interest rates:
• Both US 10 year treasury yields and US 2 year swap rates were on average rising prior to the previous four Fed hikes and continued rising after them, although the rate of increase slowed a little for 10y rates. Specifically, 10y Treasuries (2y swaps) were 25bp (45bp) higher three months after the first Fed hike and 85bp (90bp) higher after six months.
• The sharpest rises occurred after the Feb 1994 Fed rate hike: US 10y yields rose 140bp and 2y rate were up 180bp after six months.
• The exception was the June 2004 Fed rate hike, when 10y yields (2y swaps) were 60bp (27bp) lower three months after the start Fed’s first hike. Indeed, this pattern eventually became known as the ‘bond yield conundrum’, where strong Chinese purchases of US Treasuries were important (see here). I analyse the 2004 experience in more detail below.
Second, US-foreign interest rates differentials neverthless didn’t support USD. Indeed they were consistent with USD depreciations some of the time. This illustrates the importance of considering non-US developments (and non-US weakness has been at least as important as US strength in driving USD’s 2014/15 appreciation).
• US-foreign rate differentials were generally flat in the two to three months after the Fed rate hikes, following general rises pre Fed action. Obviously this reflected non-US rates actually rising in parallel to US rates in the couple of months after the Fed acted.
• But this only seems to potentially reflect growing non-US rate hike expectations in 1999: BOE followed the Fed in raising interest rates in September 1999 while ECB and BOJ both raised rates in November 1999. By contrast, Buba and BoE cut interest rates after the 1987 Fed hike, Buba and BoJ loosened in 1994 and the Fed largely acted alone in 2004 (ECB started raising rates right at the end of the Fed hike period).
• The rate convergence could well reflect US macro variables like industrial production and retail sales often being broadly in the G10 pack around Fed hikes, although you can’t definitively rule out the impact of bond market technical factors (which could be repeated).
Third, the link between USD and US-foreign interest rate differentials movements tended to reassert itself a couple of months after the Fed hikes.
• They broadly matched the more pronounced USD TWI fall from June 1994, as non-US interest rates rose even more than the already substantial rise in US interest rates.
• USD’s September-October 1999 depreciation and subsequent bounceback also appears well accounted for by US-foreign rate movements.
• So history provides some evidence that USD deviations from rate differentials tend to be temporary (giving hope to current USD bulls betting on growing US-foreign rate differentials).
• That said, US-foreign interest rate differentials don’t match any of USD’s fall (slow initial depreciation which subsequently accelerated) after the June 2004 Fed rate hike, even though US yields were actually falling.
US Economic Strains in 2004
But the ‘puzzling’ USD depreciation after the June 2004 Fed hike seems to reflect the combination of oil price rises, deteriorating US current account, rising US inflation and inflation breakevens and US investors buying foreign assets.
The charts below illustrate the little-discussed stylised fact that oil prices have tended to rise after Fed rate hikes. Indeed, oil price rises generally tend to precede the USD TWI depreciations after Fed rate hikes, downplaying the argument that oil prices only rise because the USD was falling. Oil price rises matter for the Fed and the USD because they put upward pressure on inflation and adversely affect the US current account (when the US was more oil dependent before shale came on stream). Such dynamics seem to have mattered most for USD around the June 2004 Fed rate hike. Specifically, USD’s further leg down after October 2004 was preceded by the oil price rising sharply. And this was associated with:
• US CPI inflation rising quite sharply after the Fed hike. This suggests that USD was undermined by the Fed getting behind the curve due to worries about the after effects of the dot com bubble bursting. Cleveland Fed Governor Mester recently warned of the dangers of repeating this “Greenspan put” mistake.
• Consistent with that, US 10y inflation breakevens also rose quite sharply three months after the Fed hiked. The corollary was that the eventual rise in nominal rates didn’t translate into the higher real rates which would have been more likely to have supported USD.
• The US oil trade deficit and current account deficits deteriorated sharply, in contrast to the situation around previous rate hike cycles. Indeed, the deterioration of the US current account around this period, to the worst percent of GDP on record, sticks out like a sore thumb in G10 space.
So this is a nice example of the general lesson that the FX impacts of rate rises depends on the economic backdrop: 2004 looks like the Fed hiking from a position of growing economic strains rather than one of a sustainably improving economy. Linked to this, USD’s 2004 depreciation may also (unusually) have been affected by capital flows. The record US current account deficit meant, of course, that there were strong net foreign capital inflows into the US. But it’s notable that underneath the surface US investors were actually purchasing substantial foreign assets. And those gross capital outflows correlated quite closely with USD’s fall around this period.
Some evidence of positioning contributing to 2004 USD depreciaion
There is, however, also some evidence from CFTC positioning data that long USD positions built up before the 2004 Fed rate hike were unwound around Fed action. And that could have contributed to the eventual USD depreciation – although the timing raises some questions. Specifically, the USD longs, which had built quite quickly (unlike recently) were sharply unwound five weeks before the Fed hiked and were temporarly rebuilt a little when the Fed acted. But USD net positioning did turn negative alongside the more significant USD depreciation from October 2004, although that could be as much to do with the oil price/current account/rising inflation issue as ongoing Fed hikes. This apparent use of USD as a funding currency during this period could also be linked to US rates remaining relatively low despite the Fed’s hike, with only gradual increases signalled.
Applying the lessons: what’s similar or different now? How should we trade the Fed?
Of course, there are definite limits to what you can infer so much from past behaviour. Each previous period can have its own distinct characteristics and four events is also hardly a robust sample size. But most importantly the current (global) situation seems very different from a couple of decades ago (i) non-US growth is generally weak; (ii) several major central banks are conducting unprecedented monetary policy and may do more which is affecting FX dynamics in probably unintended ways; (iii) China is significantly more important in the global economy but is going through significant transition issues; (iv) EMs in general are also more important but many are suffering from the after effects of enjoying very loose US monetary policy and have significant China exposures.
In particular, the previous tendency for USD to be dragged down by non-US yields rising by as much as/more than US yields (rather than US yields falling) seems less likely to occur with a 2015 Fed hike given the general weakness of non-US activity. Rather the risks seem skewed to further monetary easing by a number of non-US central banks (both G10 and AxJ): the well-known policy divergence theme seems more likely to have legs than in the past given the global macro situation.
And it’s notable that falling non-US yields seem to have been at least as important as rising US yields in driving the broad 2014/15 USD strength (before the recent stall). Thursday’s ECB press conference may well produce forecast (skew) downgrades and more dovish language given the significant tightening of financial conditions, although ECB may take a glass half full view after last week’s slightly better lending data and Monday’s steady (core) HICP read. The BoJ’s expectation of a sharp end-2015 CPI uptick looks very optimistic, although the political economy of BoJ encouraging further JPY falls shifted a couple of months ago. The risks also seem skewed to further easing by commodity-country central banks (both G10 and EM) given weaker than anticipated commodity prices and growing evidence of adverse second round impacts on investment. BoC could yet surprise again on 9 September, given weak business investment in Tuesday’s above-consensus GDP report and that oil prices lie significantly below the level assumed by BoC. And a range of AxJ countries seem to be suffering from deflationary (producer price) pressures (see here) and are adversely affected by weak global trade and China exposures (see here).
The 2004 experience – when the Fed moved alone from low rates (as per now) and US yields displayed some buy the rumour sell the fact tendencies – nevertheless cautions that interest rate differentials could nevertheless still work against USD. But importantly the underlying causes of USD’s 2004 weakness aren’t currently active: (i) Oil prices seem unlikely to spike sharply, given strong global oil supply and weaker global demand; (ii) the Fed seems less likely to get behind the curve than in 2004 given that they seem to be taking seriously the lesson that delaying hiking too long stores up inflation and asset price problems. Rather the concern is that they are moving too soon given moribund inflation (see here on weak inflation dynamics, exemplified by last week’s 1.2% yoy core PCE deflator figure and declining breakevens); (iii) The US current account is not a major concern; (iv) US investors seem less likely to make major purchases of foreign securities – rather a continuation of recent asset liquidations seems a more likely base case; (v) USD isn’t likely to become a funding currency.
Stretched USD positioning is, however, a major challenge to USD strengthening with a 2015 Fed hike. Indeed, the finding that USD longs were unwound five weeks before the Fed’s 2004 rate hike, with further unwinds two months later (on macro concerns) potentially contributing to USD’s later fall, may send a shiver down USD bulls’ spines in light of the apparent USD longs unwind in recent weeks. Moreover, USD longs currently look even more stretched than in 2004 even after their recent unwind, although their buildup has been more gradual. That said, CFTC data are well known to only cover a small portion of the FX universe.
Moreover, a Fed hike will likely support EUR, JPY and CHF if it (significantly) damages risk appetite – given that they all are acting like funding currencies/safe havens (due to negative ECB and SNB interest rates on top of JPY’s and CHF’s usual safe haven characteristics). Moreover, it has previously tended to be difficult for USD to rise against other G10 currencies when EURUSD is falling. But the ECB may well be getting more nervous about the effective monetary tightening engendered by EUR’s rise, equity price falls, higher real rates and a steeper yield curve (see here) and so may well act to cap EURUSD’s rise in a severe risk off scenario (hence why Thursday’s ECB meeting is interesting). And the wildcard of potential China-induced risk aversion will also be hovering in the background: in itself that reduces the chances of a September Fed hike (see here) although Fed Jackson Hole comments indicate that they are trying to look through recent volatility.
Importantly, however, history is not that auspicious about the Fed’s ability to not overly damage risk appetite with hikes: the Chart below illustrates that SPX tended to be flat to falling after initial Fed hikes, especially if the 1987 experience is stripped out. The picture is, however, more mixed for the VIX, which spiked sharply with the February 1994 Fed hike, fell after the June 1999 one and was little changed after the June 2004 hike. Should the Fed decide to hike in 2015 they will likely try to calm market concerns by again adopting forward guidance that rate hikes are likely to be slow, gradual and data dependent. But that is likely to be a more difficult trick for the Fed to pull off should they surprise the market by hiking in September (which itself militates against early Fed action, alongside weak apparently weak US inflationary pressures).
Given all this, going into potential 2015 Fed hikes USD longs appear most attractive against:
• Commodity-exposed G10 and EM FX. Commodities are unlikely to react positively to a Fed hike (exacerbating fears of weaker global demand generated by China’s slowdown) and there’s growing evidence of adverse second round impacts on investment (e.g. in recent Australian investment plans data although RBA have stepped back from jawboning AUD hoping that Fed hikes will do the work for them).
• AxJ countries with PPI deflationary pressures, significant exposures to China (see here) and weak world trade growth as well as reportedly large USD-denominated debt. They risk being squeezed both by a Fed hike and/or a China slowdown and face difficult policy choices.
• EM currencies with funding vulnerabilities given large current account deficits, especially those with political issues (e.g. BRL, TRY).
But given that EUR, JPY and CHF may well be supported near-term by a Fed hike (via the risk aversion channel, perhaps especially with a surprise September move) there would also seem to be diversification benefits in selected long positions of them against commodity-FX, AxJ and other vulnerable EMs (subject to liquidity issues). Precise currency choice would, of course, require more detailed analysis of individual country situations.