Graphic content – November; US Data Seasonality
29th November 2018
US data, very unusually for this time of year, is coming in below expectations
Back in August 2017, I mentioned on this blog that US economic data had exhibited a strong and consistent seasonal bias since 2010. Relative to consensus forecasts, high frequency US data had a habit of starting the year above them, weakening sharply from April through to July, and then strengthening through the summer and particularly into year end. Since 2010, US economic data has not once come in above expectations in June, but has beaten expectations in December every year bar 2015. 2017 proved to be a textbook year, with US data exceptionally strong through Q4 relative to prior expectations.
It isn’t straightforward why seasonally adjusted economic data still appears to demonstrate seasonal bias (so called ‘residual seasonality’). There is a fairly well-known bias that US Q1 GDP has historically tended to be weaker than other quarters, and yet Citi’s Surprise Index shows that historically US data has typically been stronger than expected in Q1 (perhaps economists are overestimating the Q1 GDP seasonal errors!).
The habit of Q1 and particularly Q4 to be stronger than expected, and Q3 to be much weaker, is possibly due to extreme economic weakness in Q4 2008 and Q1 2009. The severe recession post-GFC would distort some of the seasonal calculations for many of the data releases that form the Citi Surprise Indices, as this weak data would then bring down the average for the time of year, thus increasing the likelihood of subsequent data coming in above expectations.
The reason why this is relevant is that economists do not appear to be taking account of the residual seasonality problem, otherwise their forecasts would be higher in Q1 and Q4. And more importantly, neither is the market.
As illustrated in the chart below, the US Treasury market has tended to rally from March, just after data has started to turn for the worse, and then tends to sell off in Q4 as data beats expectations.
US economic data closely followed the usual pattern in H1 2018, but it has started to derail. US data is now coming in slightly below expectations, which is very unusual for Q4, with the year 2015 being the only occurrence since 2010. Indeed there are some similarities to 2015 – back then, China was having a material economic slowdown, both oil and non-oil commodities fell sharply, corporate bonds started a major wobble (some US high yield mutual funds were suspended as they couldn’t meet client redemptions), and the Fed was building up to a December rate hike after a lengthy pre-commitment. The omens therefore do not look particularly good for the US economy, and investors have already tentatively started to price in a ‘dovish hike’, just as in 2015 when the Fed initiated its hiking cycle, indicated 4 hikes in 2016, but only eventually hiked again one year later.