Tim Toohey's Latest Insights - US Inflation Outlook - Ellerston Global Macro Fund
September 2017 | BY Tim Toohey
Basket case: A different perspective on the US inflation outlook.
“A nickel ain’t worth a dime anymore.” Yogi Berra
Yogi Berra was perhaps the greatest accidental philosopher in recent history. He was a gifted baseball player for the Yankees, but his jumbled phrases became more famous in the sporting world, as much for their unintended wisdom as their humour. After the laughter had ceased, his unusual perspective on everyday events often revealed a truism that escaped those accustomed to thinking in a standard way. Looking at common problems from a different perspective can sometimes illuminate a path not visible from the common vantage point.
The US Bureau of Labor Statistics (BLS) has been collating monthly inflation data since 1967. Each month the financial world eagerly anticipates its arrival and transacts financial instruments based on often cursory glances at a few summary statistics. What was the monthly increase? Did the annual rate rise? What happened to ‘core’ inflation (ex food and energy)? Were there any revisions? What happened to rents…they are 1/3rd of the CPI index you realise? As soon as it arrives it is gone again, and the market moves on to the next data point, rumour or, more recently, tweet.
If you marvel at the array of analysts willing to go on TV and dictate exactly what the CPI data print means for markets within seconds of the data release then you are not alone. Sometimes the very first reaction to a data release may not be the correct response. Sometimes a modicum of analysis, let alone reflection, is appropriate to distil the signal from the noise. And occasionally even the Fed, despite its stock of PhD economists (over 300 at the Washington based Board of Governors at last count, let alone the number at the 12 regional Federal Banks) errs in its interpretation of the direction of inflation pressures. So what can someone based in Australia hope to tell the US about what is happening to its inflation data? A different perspective, a different analytical framework and perhaps a different conclusion to the current narrative.
Dispersion within the basket: Broad based and persistent disinflation
“You can observe a lot by watching.” Yogi Berra
One of the more straight forward techniques typically employed by Australian inflation analysts is to assess the number of items rising in the consumer price index basket versus the number of items falling in price. Such dispersion indices can be created using a number of different approaches and are useful for gauging how generalised inflation pressures are and the skew in the distribution of price pressures. For instance, researchers have often found that the more generalised prices are, and how far the distribution has skewed relative to its history, as useful for forecasting near term inflation trends.
Exhibit 1 shows one of our calculations of dispersion within the US inflation basket. The FOMC in January 2012 agreed that an annual rate of inflation (measured by the PCE deflator) of 2% is consistent in the long run with meeting the Fed’s statutory mandate of maximum employment and price stability. We calculate the percentage of items in the basket greater than 1% below and above the Fed’s target of 2%, and the percentage of items that fall within the 1% either side of the 2% target.
Exhibit 1: Historic highs in the number of items increasing by <1%
Three points that stand out from the exhibit;
- The number of items with prices rising at a pace consistent with the Fed’s inflation target has been rising in recent months and is close to the historic average.
- The number of items rising by more than 3% on an annual basis was recently challenging for a historic low.
- The number of items rising by less than 1% on an annual basis has eased from 3Q16 peak of over 60% of the basket, but remains elevated and near the highs recorded at the depth of the financial crisis.
Such broad based disinflation pressures skewed greater than 1% below the target was not only unusual, it was suggestive of ongoing low inflation readings. We noted the distribution had kicked lower in May 2017 and, as a result, became cautious on near term future inflation prints.
US inflation analysts, instead took their lead from the Fed and concluded that the decline in inflation in March was merely due to some ‘one-off’ items. Exhibit 1 was suggesting quite the opposite. Inflationary pressures were again becoming less broad-based and slipping back into the disinflationary pattern exhibited through 2012 to 2016.
One of the key questions is whether the dispersion is showing signs of turning back up into the target band? At the margin, yes. Exhibit 2 takes the same data and looks at the difference between the number of items rising above 3% and the number of items rising by less than 1%. On this measure, the worst appears to be over, but a compelling break higher is clearly yet to occur.
Exhibit 2: The number of items >3%yoy less the number <1%yoy
Exhibit 3 shows the more common headline and core measures of US inflation and its alarming path over the prior 6 months. Encouragingly a small recovery is now evident. A 0.4%mom rise for headline inflation in August and a 0.248%mom rise for core inflation were the highest inflation readings since January 2017 and an encouraging relief from the persistent undershooting of recent months. But how can we be sure that this is more than just an aberration?
Exhibit 3: Does August mark the break in a worrying disinflationary trend?
A deeper look at the individual components contributing to and subtracting from the inflationary pressures reveals rent (both primary and owners’ equivalent), motor vehicle registration and education are almost always the dominant sources of inflationary pressure. Each item has been in the top 7 each month for the past 2 years and in the top 7 85% of the past 5 years. Similarly, declines in the price of wireless telephone plans and used cars – 2 of the items called out by the Fed as being one-off price declines – are not particularly unusual. Indeed, since 2012 these are the 2 most persistent sources of disinflation in the US. The bottom line is the items that have in recent years been driving the inflationary and disinflationary impulse are still doing so. So what has been driving the disinflation process since early 2017?
Technically speaking, technology is not the source of recent disinflation.
“I’m not going to buy my kids an encyclopaedia. Let them walk to school like I did.” Yogi Berra
When a forecast is not going to plan it’s human nature to seek explanations in new and unforeseeable events, rather than admitting forecasting error. The price destructive power of new technology, the dislocation of labour and profit capture by the corporate is not a new concept. For those who also endured Thomas Hardy’s 1891 Tess of the d’Urbervilles at secondary school, the realisation that technological innovation results in price falls and labour market distress will not have been lost upon you.
Yet the string of low inflation prints from March through to July 2017 had the Fed initially pointing to “one-off” factors, but as disinflation became more persistent the Fed’s collective finger increasingly pointed towards technology as the culprit. US Fed Governors, who previously were steadfast on the need to continue to normalise interest rates as the US steadily drained its spare capacity, suddenly had a change of heart. William Dudley at the New York Fed on the 8th of September noted;
“The persistence [of low inflation prints] suggests that more fundamental structural changes may be playing a role. These include the increased ability of prospective buyers to compare prices across different sellers quickly and easily, the shift in retail sales to online channels of distribution from traditional brick-and-mortar stores, and the consequences of these changes on brand loyalty and business pricing power.”
Robert Kaplan (Dallas Fed) and Loretta Mester (Cleveland Fed) expressed similar concerns in recent weeks, while Neel Kaskhari of the Minneapolis Fed was even more effusive in embracing the technological disinflation wave;
“Technological disruption is a new and powerful structural factor that is influencing inflation”.
The irony is that only a little more than 12 months ago ‘secular stagnation’ theorists held sway. Advocates such as Larry Summers and John Taylor advocated that the reason for why demand was so lacklustre was the world couldn’t possibly expect to innovate at the same rate as prior decades. Their arguments were not just influential; by mid-2016 secular stagnation was essentially priced by financial markets as the base case.
So in sum, in mid-2016 the consensus was there was not enough innovation and not enough growth; by 2H17 the world had apparently tipped on its axis and embraced the idea of there being too much innovation and the globe is running short of spare resources. But what is the evidence that a wave of technological innovation has hit the Western world and is now driving down inflation at a faster rate than could have been detected 6 months earlier?
To answer the question, we went through the US inflation components and assigned each item to whether they were predominantly domestically produced (denoted non-tradables) or if they are imported or compete with imports (tradables). Curiously, this approach is not standard practice in the analysis of inflation in the US.
The tradables basket excluding the impact of oil prices for US inflation is shown through time in Exhibit 6. We then excluded the items in the CPI basket that were readily identifiable as high technology items; Personal Computers and Peripheral Equipment, Computer Software and Accessories, Telephone Hardware, Calculators, and Other Consumer Information Items, Televisions, Other Video Equipment, Audio Equipment, and Audio Discs Tapes and Other Media. This basket of goods, labelled “Tradables ex-oil and ex-tech”, is also shown in Exhibit 4.
Exhibit 4: The exchange rate is more important than technology
There are 2 important observations that arise from Exhibit 4;
- The tradable goods sector ex-oil exerted significant variation over the past 2 years. This basket of goods represent 24% of the US CPI basket and accelerated from an annual pace of -1%yoy in October 2015 to 3%yoy by March 2016 before slowing significantly. By late 2016 this basket declined at a 4% annual pace – at exactly the same time when inflation was cycling higher at the headline level into the end of 2016.
- The exclusion of technology related items makes little difference to measured prices. The combined weight of the technology related items listed above in the CPI basket of just 0.8%. Consumers may love to buy technologically enhanced goods, but they don’t buy them very often, so their impact on the CPI is relatively limited.
In other words, the reason for the surge in inflation in the US late-2016 was not due to rising prices of the tradable goods basket, tradables were pushing in the opposite direction. Moreover, technologically advanced manufactured goods are simply too small in the US CPI basket to explain the gyrations in inflation over the past 12 months.
Of course, this ignores items that are technologically driven services, which resides in our non-tradables basket. However, items such as wireless telephone services and internet services are also relatively small in the composition of US CPI. Both items currently have a 1.5% weighting and since March 2017 have collectively subtracted just 0.07% from inflation. Not immaterial, but not enough for the Fed to wet the bed over fears of a technologically driven disinflation wave. It’s hardly enough for Tess to raise an eyebrow.
Trouble with the curve: Why the Phillips Curve may not be the best way to approach inflation analysis.
“If you don’t know where you are going, you might wind up someplace else.” Yogi Berra
Flat Phillips curves discombobulate macro forecasters. Without their main touchstone of falling unemployment rates generating rising inflation expectations, economists are at something of a loss to describe the inflationary process and policy makers tend to become less forward looking. In essence, market participants begin gravitating to the completely unsatisfying view that the best forecast of inflation is where inflation was last month.
The US Fed took a noticeable step in this direction at the July meeting with the minutes revealing some members in open defiance of the importance of the Phillips curve;
“…a few participants cited evidence suggesting that this framework was not particularly useful in forecasting inflation”, while “most participants thought that the framework remained valid“.
The Philadelphia Fed went further and released a paper suggesting that Phillips curves are only useful for forecasting inflation when the economy is slowing and the unemployment rate gap to NAIRU is increasing – not in an environment when the unemployment rate is steadily falling. That conclusion was likely to cause discomfort for economists inside and outside the central banking universe. But for a Fed Chair who is at heart is labour market economist, it is an anathema.
While we don’t agree with the Philadelphia Fed’s paper, it is clear that the market will only believe inflation is an issue when it is clearly visible in the data and recent inflation data has been sufficient to give some FOMC members pause.
As alluded to in the introduction, inflation data in the US is usually presented in the US using the ‘exclusion’ approach. That is, inflation is typically analysed excluding something from the headline measure; ex-oil, ex-food, ex-housing….etc and looking at the trend in what is left. It’s a common approach and it is replicated globally, but of course once you have excluded several items you are still left wondering what is driving the inflation process in over a hundred separate other items in the basket.
Another approach is to systematically remove the volatile items at either end of the weighted distribution of price movements in the basket. This trimmed mean approach has been used extensively by the RBA, is calculated in New Zealand (although the RBNZ tend to prefer their principal components model to derive an underlying inflation path), and is also calculated in the US by the Cleveland Fed for the CPI and the Dallas Fed for the PCE deflator. This approach is very useful, but at its heart it’s a technique to look through the noise to determine an underlying trend, rather than a forecasting aid to estimate where inflation is likely to be in the future. If Yogi Berra had assessed these approaches he would likely conclude that they are great for telling us where inflation currently is, but not particularly good at telling us where inflation is going.
The main advantage of approaching forecasting by creating different slices of the CPI basket is that it breaks the problem into digestible chunks. Indeed, we believe our bespoke baskets for US inflation provide some very useful insights.
A fork in the road, but the path is clear.
“When you come to a fork in the road, take it.” Yogi Berra
Janet Yellen in the past week admitted that in assessing why inflation had slowed from above 2%yoy at the start of the year to 1.6%yoy by mid-year that “I will not say that the committee clearly understands what the causes are of that” and that while one off factors like cell phone plans were part of the explanation, recent weak inflation data were a bit of a “mystery”.
Approaching the problem from the standard exclusion-method does make it difficult to see the underlying causes of the weak inflation. Approaching the problem by studying which items have been most significant in driving the inflation process and their historical frequency using our baskets approach to the problem sheds a different light.
Recent US inflation misses were not being driven by a narrow number of one-off price falls and were not being driven by a new technological wave hitting the US economy. The disinflationary pressure was historically broad based and persistent. Items that can reasonably be attributed to technology are neither falling rapidly in price nor of sufficient size in the US consumer basket to be a valid explanation of the recent trend.
Our basket based analysis suggests the answer to Yellen’s mystery is multifaceted;
- The peak impact of the disinflationary impulse from the US$ strength was during 1H17 – subtracting 30ppts from annual inflation. But this is now changing. Our tradables basket appears set to be contributing 30ppts to US inflation by mid-2018.
- The halving of the oil price in 2H14 and the persistent disinflationary influence through 2015 flipped in 2016 as oil prices recovered from a trough in February 2016. In the 12 months to February 2017 gasoline prices contributed 119ppts to annual inflation. The relative stability in oil prices saw this impetus fade to zero by mid-17, taking headline inflation down with it. A hurricane induced spike in retail gasoline prices in September threatens to contribute a sizeable 30ppts to inflation in the month of September. Although this impact will surely fade in coming months it will by itself be sufficient to alter the 6-month annualised rate of inflation from near zero to close to a 1.5% pace in the very near term.
- Rents have been blamed as a significant disinflationary force in 1H17, however, we showed that rents are always weak in the 1H of most years. 2017 is not particularly unusual for rents, and this seasonality suggest that rents will comfortably contribute 90ppts to inflation in the final 5 months of 2017 compared to the 40ppts contributed in the first 5 months of the year.
- Public administered prices clearly eased in 1H17, in contrast to fears of escalating health care and tobacco prices. The combination of lags and changing policy suggests the public sector will be a significant source of inflation in the coming 12 months. We expect this basket to contribute 50ppts to annual inflation over the next 6 months.
- Finally part of the mystery as to why inflation eased was that the strength of underlying inflation into the start of 2017 was inflated by base effects and the dwindling downdraft from a glut in unprocessed foods. As these impacts normalised, so too did headline inflation. Yet there is reason to believe that unprocessed food may soon start to be a modest inflationary impact in coming quarters.
When it comes to the inflation outlook the US has hit a fork in the road. Most investors remain uncertain as which path to take; the path pointing to subdued future inflation due to unknown mysterious causes, or the path pointing to a quick snap back in inflationary pressures.
From our analysis the evidence clearly points to a relatively sharp acceleration in inflationary pressures in coming months. Our expected path for inflation is shown in Exhibit 5. Base effects will still see the annual rate of inflation oscillate around 2%yoy in the coming months, ebbing again in early 2018, but the Fed will be looking at the 6-month annualised rate.
Exhibit 5: We expect a sharp re-acceleration in in inflationary pressure
Such a path for inflation will not only embolden the Fed to deliver on its ‘Dots’ for the Fed funds rate, it will cause markets to seriously question whether the Fed is behind the curve.
As Yogi made clear, forecasting is tough. But a different frame of reference and a different perspective can bring insight. Our baskets approach to forecasting suggests US consumer inflation is set to accelerate and accelerate relatively rapidly.
For financial markets, the combination of a sharp reacceleration in inflation coinciding with the commencement of quantitative tapering (QT) and robust economic growth momentum leaves developed market bonds, the ‘yield trade’, risk parity strategies and minimum volatility strategies highly vulnerable. QE has underwritten the most extended period of positive correlations between equities and bonds in modern history. Traditional asset classes are set to endure a more challenging period during the recalibration. However, this is precisely the investment climate we have been waiting for. In the lexicon of baseball; “Batter up!”
“It’s tough to make predictions, especially about the future.” Yogi Berra
This newsletter has been prepared by Ellerston Capital Limited ABN 34 110 397 674 AFSL 283 000, responsible entity of the Ellerston Global Macro Fund without taking account the objectives, financial situation or needs of individuals. For further information, contact email@example.com. This material has been prepared based on information believed to be accurate at the time of publication. Assumptions and estimates may have been made which may prove not to be accurate. Ellerston Capital undertakes no responsibility to correct any such inaccuracy. Subsequent changes in circumstances may occur at any time and may impact the accuracy of the information. To the full extent permitted by law, none of Ellerston Capital Limited, or any member of the Ellerston Capital Limited Group of companies makes any warranty as to the accuracy or completeness of the information in this newsletter and disclaims all liability that may arise due to any information contained in this newsletter being inaccurate, unreliable or incomplete.