This post considers whether sterling’s post-April rally represents an overshoot and dissects the May Inflation Report’s treatment of Brexit-affected asset prices. The main points are:
- Sterling has, as I expected, continued to closely track betting market Brexit odds, recently down to the low-20%’s from over 40% in April, but has also been supported by the continued risk-friendly market environment (despite the Fed’s more recent hawkishness) and an apparent sterling short position squeeze.
- EU referendum opinion polls have, however, moved significantly less and the apparent recent shift in support for “remain” reflects recent polls being predominantly telephone based (they tend to be more pro-remain and report less “undecided” voters). Moreover, markets may well demand more decisive polling evidence – e.g. the “undecided” share falling or greater evidence of younger voter engagement – to further support sterling as 23 June approaches.
- But already record-high sterling near-term sterling implied volatilities and risk reversals, make fresh GBP short positions expensive to initiate and market participants continue to presume the undecided voters will eventually vote for the status quo (as in the Scottish referendum). Market participants may consequently instead search for cheaper non-sterling based Brexit hedges or focus on trades expiring before the referendum.
- So while my bias is for some of GBP’s recent bounce to unwind if the “remain” lead doesn’t become more firmly established, an “uneasy truce” around sterling could easily develop in the run-up to 23 June, with consequent range-bound GBP trading (but likely continued rises in near-term implied vols).
- Evidence of slowing UK growth and weakening pay pressures continues to accumulate, causing MPC members to ponder whether it’s all Brexit-related. While UK data will likely continue playing second-fiddle Brexit odds in driving sterling over the next few weeks they will likely return to focus post-referendum.
- The May Inflation Report decision to strip out around half of sterling’s post-November fall but not similarly adjust the market rates profile raises consistency issues, even if it’s a larger step. Contrary to the Bank analysis, I find fairly-compelling simple evidence that Brexit uncertainties have also driven OIS rates. It seems no coincidence that MPC rate cut expectations have been priced out as Brexit odds have fallen.
- I agree with MPC that Brexit concerns have significantly impacted sterling (and anticipated them making an adjustment) but view the impact as intermediated via market interest rates. Notably, UK 2-year OIS rate moves can account for 70% of the sterling’s moves between November and the May IR (with foreign rates adding only a little explanatory power).
- The May IR inflation projections would have been commensurately lower if MPC had stripped out the apparent Brexit impacts on OIS rates. That would have provided more ammunition to MPC doves like Vlieghe, but he’s already signalled his bias to cut rates if UK data don’t relatively-quickly bounce post-referendum.
- That, alongside poor UK data per se, the record UK current account deficit and remain being increasingly priced, will likely mute sterling’s bounce on a remain vote (sterling would likely fall sharply, probably 10-20%, on a surprise “leave” vote).
How fragile is sterling’s recent bounce?
Sterling rallies on lower betting market Brexit odds and positive market risk environment
In less than four weeks time we’ll know the result of the recently all-consuming UK EU referendum. But from sterling’s recent relatively-sharp appreciation you might surmise that the result has become pretty certain. Specifically, £ERI has unwound around half of it’s 11%-plus post-November slide since early April, rising against most G10 currencies along the way (see Chart below). It’s been one of the few currencies to recently rise against USD while the relatively-small GBPSEK bounce backs my previous view that GBP short positions were less attractive against USD and EUR.
I’ve been highlighting for several months how sterling’s prospects would remain dominated by developments in betting market Brexit odds. And the second Chart above illustrates that this relatively-close relationship has indeed continued in recent weeks. Betting market Brexit odds currently lie just above 20%, down from a peak of over 40% in mid-March and 30-35% at the start of May (Number Cruncher Politics report recent sub-20% probabilities). Moreover, there’s a bit of evidence of betting markets becoming more skewed to low-probability odds: there seems to have been fewer high-probability intra-day odds reported (see Chart below). But overall the spot market message is that previous GBP shorts have been unwound on the better Brexit news.
By contrast, option markets have continued to price substantial Brexit risks. Indeed, as I expected, 1-month GBP implied volatilities have already spiked to record high as soon as the referendum entered their window and 1-month GBP risk reversals are similarly stretched towards sterling weakness (see Chart above). But longer-tenor vols and, especially, risk reversals have hardly fallen: option market participants continue to not bet on any major sterling post-referendum recovery.
The corollary, however, of these record-extended 1-month GBP implied vols and risk reversals is that initiating fresh short GBP positions is pretty expensive, thereby potentially reinforcing the support for sterling from lower Brexit odds. In other words, given how much option markets are already pricing it may be difficult for short GBP positions (re)extend unless the Brexit campaign receives a substantial boost e.g. via immigration concerns starting to outweigh the apparently recently dominant economic concerns of Brexit. So market participants may well instead search for cheaper non-sterling based Brexit hedges or focus on trades expiring before the referendum.
Sterling’s rally has, however, also been helped by market risk appetite remaining apparently reasonably healthy. G10 FX implied volatility has fallen sharply, and not just (directly) due to lower Brexit risks, as equity volatility has remained subdued (see Chart below). That said, global equities have continued to struggle to make gains and the GBP-equity price link has been fairly weak. And the Fed’s recent more hawkish stance, importantly including by Yellen on Friday (with market-implied odds of a June Fed hike up to 34% from around 4% prior to the Fed minutes) plus China issues could yet disturb this relative calm, undermining sterling.
Lower betting market Brexit odds could prove fragile
The previously-discussed fall in betting market Brexit odds is notable because it hasn’t really reflected a substantial shift in underlying EU referendum opinion polling. Specifically, the recent rise in “remain” poll of polls probabilities (see Chart below), which seems to have provided some fresh downward impetus to Brexit betting market odds, was both short-lived and also seemed to have been driven by most recent EU referendum opinion polls being telephone-based. As is well-know, telephone polls tend to provide more support for remain than internet polls and have lower “undecided” voters (perhaps because the telephone interviewers are a little more probing than an internet site). And the second chart below illustrates that internet-based EU referendum polls have remained relatively stable, confirming the impact of the recent polls being predominantly phone-based.
Moreover, the deeper question of whether telephone or internet polls provide the more accurate guide to the referendum outcome remains hotly debated. Number Cruncher Politics, recently argued that phone polls’ known demographic and eductational biases cause them to overstate the “remain” case by only 3pp. That’s substantially less than previously argued argued by YouGov.
But recent polls continue to feature relatively-significant proportions of undecided voters – leaving the referendum potentially wide-open (if immigration starts trumping economic concerns), rather than Brexit being a 20% probability event. But market participants seem to think that at-present undecided voters will eventually plump for the status quo, following the Scottish referendum pattern (although near-term GBP implied volatility also then spiked sharply). But there’s probably more that could disturb the recent “remain” lead than before the Scottish referendum, with no last minute bribes being offered. A bad few weeks of migration headlines could turn undecided voters who’ve become weary of economic statistics being traded. Or the “remain” camp could suffer from low voter turnout if voters extrapolate recent developments as signalling that the win is in the bag.
And the meta-point is that markets remain highly-sceptical of opinion polls, in light of the Scottish referendum and (especially) 2015 UK General election experiences. But, as I’ve previously discussed, despite their better recent track record betting market odds aren’t perfect. So opinion polls could yet come into greater focus, especially if they remain close with large numbers of undecided voters.
Referendum result may just represent the start of UK political uncertainties
Moreover, markets may be underweighting the political fall-out from the referendum, irrespective of the result. The Conservative party continues to take chunks out of each other and over the weekend William Hill cut the odds of David Cameron standing down before end-2016 to only 2/1 (even if Cameron’s critics don’t yet seem to have the 50 MPs necessary to call a no-confidence vote). Such post-referendum political uncertainties will tend to constrain the likely post-referendum GBP bounce on a “remain” vote, which is in any case already increasingly priced in.
UK fundamentals have continued underperforming, pay growth weakening
I’ve previously highlighted how the UK economy shows definite signs of slowing (see here and here), with Brexit uncertainties playing a significant role (higher option value of waiting for firms considering investing or hiring or consumer considering major purchases) plus some potential more fundamental slowdown. And overall UK data have continued to disappoint in recent weeks (see Chart below), contrary to recent better-than-expected Euro Area macro data (although the ECB still faces several dark clouds, see here). Last week’s backward-looking GDP release confirmed UK growth slowing to 0.4% in Q1, mainly due to weaker stockbuilding but with a 0.5% q/q fall in business investment hinting at some Brexit uncertainty effects (although it also fell 2.0% in Q4 2015). Moreover, the April UK PMI surveys pointed to only 0.1% Q2 growth: so this week’s PMI releases will be closely-watched (the consensus is for both manufacturing and services to recover a little while further construction weakness is expected).
On the household side, the latest GfK consumer confidence survey failed to pick-up from it’s reportedly Brexit-driven recent decline, with confidence about general economic prospects especially weak (see Chart above). But last week’s XpertHR report that median pay settlements dipped to only 1.7% in the three months to April, with almost half the awards lower than in 2015 and only a fifth higher. This first sub-2% figure for two years, and the weakest since 2012, is especially worrying given that around 40% of UK pay deals are signed in April. Such continued pay weakness will strengthen the hand of MPC doves, although market MPC rate cut expectations have diminished on the recent lower Brexit odds (see blow).
But MPC do seem to be getting more concerned about the slowdown. Gertjan Vlieghe recently signalled that he’ll consider a rate cut vote if there isn’t a relatively-rapid UK data improvement after a remain vote. Andy Haldane probably also isn’t too far away from that position. And Kristen Forbes is concerned that something more fundamental (non-referendum) is driving the recent UK slowdown.
It would also calm jitters if the UK’s record 7% of GDP current account deficit showed some signs of falling. That said, valuation effects have prevented the UK’s net international investment position from commensurately deteriorating and an improving Euro Area would both support trade and hopefully raise the low return on foreign investment which has been an important driver of the record deficit (see here).
Should MPC have also adjusted OIS rates in May?
The MPC decided, in the May Inflation Report, to strip out half of sterling’s post-November depreciation: their analysis suggested that portion had been driven by Brexit uncertainties, which I anticipated. MPC and Bank staff will likely now feel vindicated in making this unusual adjustment, given sterling’s recent bounce.
But MPC didn’t make an equivalent adjustment to the (OIS) market path of interest rates feeding into their forecast: my preview explained that the 20-30bp fall in OIS forward rates between the February and May IRs (see Chart below) helpfully supported inflation. The May IR just briefly stated that “It is harder to identify a significant impact on UK short-term market interest rates related to the referendum….interest rates appear to have moved little in response to referendum-related news.” I disagree. While I’ve obviously got significantly less resources than the Bank’s economists, collectively several simple findings provide reasonably-compelling simple evidence that Brexit uncertainties have also driven OIS rates.
Specifically, the two charts above illustrate that betting market Brexit odds have been important drivers of OIS rates. The first chart shows that the whole OIS forward curve has shifted up in recent weeks, unwinding the downward shift in the run-up to the May IR, at the same time as Brexit odds have fallen. Most provocatively, market expectations of further MPC rate cuts have been priced out in recent weeks: it’s fairly implausible that this is unrelated to lower recent Brexit odds.
Of course, at the IR press conference Carney downplayed the likelihood of definite MPC rate cuts in the event of Brexit (reiterated at the TSC). But this fails the timing test as an explanation of higher OIS rates: the second chart illustrates that OIS rates were rising several days before Carney’s comments and OIS rates were unaffected by the comments themselves (markets are currently generally sceptical about central bankers e.g. recent Fed hawkishness and ongoing BoJ dovishness). Rather, the second Chart above nicely illustrates is the continued strong correlation between Brexit odds and OIS rates: the R-squared since start-2016 is nearly 30% for 2-year OIS rates and 25% for 1-year OIS rates.
Betting market odds are, however, an imprecise measure of the evolving underlying market Brexit views. So it’s pretty revealing that the charts below indicate the continuing close correlation between GBP’s moves and UK OIS rates. The first, an update from my May IR preview, illustrates that this multi-month link has doesn’t seem to have been impeded by Brexit concerns dominating UK financial markets. Indeed, the second chart shows that: (i) notably, UK 2-year OIS rate moves on their own account for 70% of the sterling’s moves between November and the May IR ; and (ii) GBP has became more sensitive to OIS rates (higher slope coefficient) in the six months after the November IR than earlier in 2015. My view is that this extremely strong relationship has been driven by both being affected by the dominant common factor of evolving market Brexit views (currently doubted, but with that view being potentially fragile as discussed above).
GBP’s increased sensitivity to interest rate moves since Brexit concerns emerged is even more apparent when UK-foreign relative 2-year swap rates are considered (see Chart below). And it confirms that UK OIS rates have been the main drivers of GBP: the R-squared only increases marginally (to 75%) when UK-foreign relative rates are considered.
Based upon the collective, albeit simple, evidence above to me it seems fairly clear that both GBP and UK OIS rates have been impacted significantly by the common factor of evolving market Brexit views. I completely back MPC’s decision to downplay GBP’s Brexit-driven depreciation, indeed I anticipated them making an adjustment and my simple evidence suggests they might have stripped out even more of sterling’s post-November fall (although as discussed above I’m also doubtful that there will be a huge GBP bounce on a remain vote). But I differ from them by concluding that the Brexit impact has importantly continued to be intermediated via market interest rates. Moreover, theory backs that conclusion given that there’s no real signs of UK assets generally being regarded as substantially riskier: UK sovereign CDS premia have continued retracing while UK equities haven’t significantly underperformed.
So the empirical evidence indicates that MPC’s decision to not adjust their market rate profile in a similar way to the GBP conditioning path is questionable. The BoE staff’s surprising failure to detect a referendum effect on OIS rates could reflect the complicated approach employed or them not moving beyond imperfectly-measured Brexit proxies (e.g. they examine the frequency of Bloomberg Brexit references) to examine cross-asset price links as I did above. But theoretically it also seems a little strange to make a strong judgement about one asset price (sterling) but steer away from doing the equivalent to another one (OIS rates). This could reflect a reluctance to face the complications caused by OIS rates being expectations of the MPC’s own actions and the significant impacts it would have on the rest of their model (GBP’s impacts are narrower). Of course, this could also open up a pandora’s box, making a large number of adjustments to various model variables (business investment, consumer confidence, employment….). But such judgements are the essence of forecasting.
The bottom line is that if MPC had overwritten some of the fall in OIS rates in the run-up to the May IR, the all-important inflation projection would likely have been lower (less inflation stimulus from lower OIS rates if they are likely to unwind after the referendum alongside less sterling weakness). And that would have provided more ammunition to MPC doves like Vlieghe, although as discussed above he’s already signalled his bias to cut rates if UK data don’t relatively-quickly bounce post-referendum.