Today’s U.S. Consumer Price Index (CPI) numbers were notably firm, confirming that weakness in the fourth quarter of 2016 was the result of residual seasonality in core goods prices rather than a more worrisome decline in the underlying trend pace of inflation.
With today’s print, the annualized core rate of inflation in the three months through January ticked up to 2.3% after having held at around 2.0% for much of the second half of 2016. As we mentioned in a recent blog post, we were careful not to read too much into the weakness in the latter part of 2016, as seasonal adjustments around the holiday shopping season likely contributed to the softening.
Similarly, we are taking care not to extrapolate too much from the strength of today’s numbers: When looking past the noise, we believe the data have continued to confirm our forecast for 2.2% core inflation in 2017.
Core goods surprise
Digging into the details of today’s report, the standout surprise was the month-over-month jump in core goods inflation. The 0.4% gain was the largest one-month change since 2009 and masked any softness related to the post-election rise in the U.S. dollar. Household furnishings (+0.4%), apparel (+1.4%), and video and audio (+0.6%) were particularly firm.
However, it’s important to keep in mind that the strength in these categories came on the heels of soft prints in November and December, which likely reflected an inability of the seasonal adjustments to completely capture price fluctuations related to the holiday shopping season. Retail industry holiday sales surveys reported greater-than-average price discounting by brick-and-mortar stores to compete with online shopping outlets. Meanwhile, reported price normalization after the holidays likely contributed to the larger-than-seasonally-average advance in inflation in January.
More broadly, core CPI inflation (which excludes food and energy prices) has typically been higher in the first half of the year before moderating in the second, a tendency that has garnered the attention of Federal Reserve Board staff and persists despite the index being seasonally adjusted. Among the CPI components, durable goods prices tend to follow a pattern similar to that of headline residual seasonality, and shifts in consumption preferences and retail price-setting around the holiday shopping season are also likely partly to blame.
Meanwhile, core services inflation of 0.3% was in line with the recent trend, but the 0.2% rise in the heavily weighted shelter categories was slightly softer than the recent trend (+0.3%). The weakness in shelter could reflect the lagged effect of the deceleration in housing prices as it finally feeds through to rents; however, we think it’s more likely an artifact of the way the Bureau of Labor Statistics (BLS) calculates rents. To estimate the change in implicit rents based on the rental survey, the BLS removes any landlord-provided utilities and furniture. In practice, this tends to result in below-trend rent and owner-equivalent rent during months in which utilities prices surge, as they did in January (+1.5%). If this is the case, we should see an uptick next month.
When all is said and done, we see January’s inflation numbers as supportive for Treasury Inflation-Protected Securities (TIPS). Here’s the math: Core CPI is running at 2.3% and the inflation accrual in March will be 0.58%, which translates into 12 basis points (bps) of carry (annualized inflation accrual) for five-year TIPS and 6 bps for 10-year TIPS.
More fundamentally, we believe TIPS should trade with a positive inflation risk premium; to put it a different way, we think market-implied breakeven inflation (BEI) levels should be above the inflation run rate, not below. In our view, the 10-year BEI, which is trading closer to 2%, does not account for the risk that the Trump administration’s trade and fiscal policies could push inflation even higher.
Tiffany Wilding is a PIMCO economist focusing on the U.S. Jeremie Banet is a portfolio manager on the real return team. Both are regular contributors to the PIMCO Blog.