US Treasuries: buyer beware

The significant decline of Treasury yields from their peak at the end of March is puzzling given the growth forecasts and the recent inflation data. This suggests that investors side with the Fed in thinking that inflation will decline. It also reflects the weakening of data in recent weeks, which implies that markets focus more on the change in the growth rate than on its level. The sensitivity of bond yields to economic data moves in cycles. One should expect that, as seen in the past, a less accommodative US monetary policy would increase this sensitivity because these data will shape expectations of more tightening or not. Before reaching that stage, we should already expect an increased sensitivity in the course of 2022, because it is quite likely that inflation will remain above the FOMC’s target.

Since its peak of 1.74% at the end of March, the 10-year US Treasury yield has dropped approximately 40 basis points. At first glance, such a movement is counterintuitive considering the growth outlook – the Bloomberg consensus forecast is anticipating above potential real GDP growth for the entire projection horizon, which runs until the end of 2022 – and the high inflation numbers in recent months.

The latest consumer price inflation data have surprised to the upside, triggering an increase in Treasury yields. The reaction was more outspoken for short and intermediate than for longer maturities (chart1), reflecting a market expectation that the Federal Reserve might start to normalise its policy earlier than previously expected. However, the following day, yields went into reverse in reaction to dovish comments of Jerome Powell in his semi-annual report to Congress.

It seems bond investors share his view that the increase in inflation will be transient. Market participants may find comfort in the fact that Covid-19-sensitive sectors (apparel, cars, travel, recreation services, hotels) explain the bulk of the latest monthly inflation number. When the Fed chairman commented that there will not be a major increase in inflation because the central bank will not tolerate it, the market was unfazed. It seems to think that it’s very unlikely that the Fed would be required to act forcefully. Such a relaxed view of the inflation outlook could explain why yields have not moved higher this year, but to rationalise the decline in long-term rates and break-even inflation – a measure of inflation expectations that has declined about 20 basis points since its peak in spring – one needs to look elsewhere.

One possibility is that investors expect supply pressures – as captured by the assessment of input prices and delivery lags in the ISM and PMI surveys – to decline quite soon, considering the elevated levels that have been reached. Historically, there has been a close correlation between these pressures and break-even inflation.

Another potential explanation is rising concern about new infections, which could become a headwind to growth. Finally, lower bond yields could reflect that the weakening of growth has already started. Assessing the growth dynamics during a given quarter is challenging because, given the multitude of data that are published on a regular basis, it is difficult to see the forest for the trees. The Federal Reserve of New York’s weekly nowcast of quarterly real GDP growth provides an attractive solution to this problem¹.

Chart 2 shows that in recent months the nowcast has been declining – pointing to slower growth – and this development has been accompanied by lower bond yields. The relationship between 10 year Treasury yields and the nowcast of GDP growth fluctuates over time.

As shown in chart3, based on a five-year rolling regression between bond yields and the nowcast, since 2007 there have been three cycles in the regression beta with peaks in 2009, 2014 and early 2020, each time followed by significant declines. This means that the extent to which a given change in the nowcast is reflected in a change in bond yields depends on the economic environment².

The stance of monetary policy is an obvious candidate as a factor conditioning the correlation between Treasury yields and the GDP growth nowcast. Focusing on the size of the Federal Reserve’s balance sheet, the chart suggests that the stabilisation of its size (stopping net purchases in the context of a QE programme) has been accompanied by a closer relationship between economic data and bond yields.

The same applies when, in a later stage, the balance sheet started to shrink (‘quantitative tightening’) as well as during the tightening cycle which started at the end of 2015 and which led to a peak in the federal funds rate in December 2018. These results are particularly relevant at the current juncture: one should expect that a less accommodative monetary policy would increase the sensitivity of bond yields to economic data because these data will shape expectations of more tightening or not. Before reaching that stage, we should already expect an increased sensitivity in the course of 2022, because it is quite likely that inflation will remain above the FOMC’s target.

¹ A nowcast is a real-time estimate of real GDP growth based on a broad range of economic data. The inputs in the nowcast are weighted based on their timeliness and quality. For a description of the methodology, see: Liberty Street Economics, Just Released: Introducing the New York Fed Staff Nowcast, 12 April 2016.

² Interestingly, the chart suggests that in recent months the beta has been very low. This result depends on the window chosen for the regression. A long window –which is the case here- implies a high degree of smoothing. This may cause an underestimation of the intensity of the correlation over shorter intervals (as illustrated in chart 2).