How investors are using alternative data to follow the inflation debate
With the inflation debate dominating financial markets as the global economy reopens after the COVID-19 pandemic, investors are turning to a number of so-called alternative data sources in a bid to get a more timely reading of the evolution of prices around the world.
While readings of the U.S. Consumer Price Index, the Producer Price Index, and the Federal Reserve’s preferred measure – the personal consumption expenditure deflator – all take place once a month, they offer a retrospective picture of the evolution of prices. three to four weeks earlier.
“This is where real-time alternative data comes in,” Yin Luo, senior analyst at Wolfe Research, said in a phone interview, offering “a more timely measure of price movements than official data, which trends to be delayed “.
Register: Barron’s Live at noon ET on August 4: Alternative data – What investors need to know
The ability to get a faster read of trends in market-sensitive data has led to the widespread adoption of alternative data for years, but the COVID-19 pandemic may have excited interest as investors and economists have struggled to make sense of almost all unprecedented. shutdown of economies and subsequent reopening, thanks to vaccination, which has been plagued by supply chain issues and complicated by the spread of coronavirus variants.
Interest in macro-focused alternative data has likely quadrupled since the pandemic, Luo said.
Meanwhile, so-called reflation trading – a bet that assets more sensitive to the business cycle or otherwise prepared to benefit from spike in inflation would outperform their peers – has risen and fallen in 2021. Its path has depended whether investors see inflation as in danger of soaring and perhaps spiraling or more likely to turn out, to use the Fed’s preferred term, transient.
One of the best-known resources comes from PriceStats, a company born out of a Massachusetts Institute of Technology initiative known as the Billion Prices Project in 2008. The project grew out of research conducted by economists Alberto Cavallo, now at Harvard, and Roberto Rigobon.
PriceStats, which partnered with State Street in 2011 to bring its data to the financial services industry, scours the web, tracking daily price changes on a huge basket of goods in 23 economies. PriceStats focuses only on “multi-channel” retailers that offer products both online and offline, eliminating large online-only retailers, including the e-commerce juggernaut Amazon.com Inc. AMZN,
PriceStats retrieves product descriptions and prices, sorting them by classifications. It calculates the price changes each day and uses the official weights used in a particular country’s inflation gauge to create an index.
The result is a real-time overview of a large basket of goods for each country. Of course, not everything is online. And PriceStats, for example, doesn’t track used car prices, which have skewed US inflation data in recent months. It also doesn’t track house prices, for example, but offers proxies, such as building materials, which have historically tracked those prices closely.
The data offers a snapshot of inflationary pressures with a lag of around two days, as opposed to the multi-week lag offered by official data, said Marvin Loh, senior global markets strategist at State Street, in an interview. Equally important is having a set of prices for 21 different countries, he said, because inflation is a global topic, which makes it possible to see how global price pressures play out in each. country and region.
Loh said he has seen prices online adjust faster than prices on the shelves. In 2020, when the pandemic hit, prices tracked by PriceStats fell much faster than official data reflected. On the rebound, the basket showed stabilization in prices, while official data reflected non-basket items such as used cars and airline tickets.
The bottom line is that the data showed upward pressure from what a normal June or July would look like, “but not accelerating like it did from the start. from spring to summer “.
Investors can also try to discern expectations about inflation by following the news. Ravenpack, an alternative data provider specializing in using natural language processing (NLP) to analyze news, said the approach helped shed light on a shift in investors’ expectations for inflation in May.
Ravenpack sets up systems for customers who follow particular topics, with each news item being assigned a sentiment score, said Inna Grinis, the company’s head of macro strategy.
An article about rising inflation would receive a negative sentiment score, for example. The score also takes into account all the sizes mentioned in the story, she explained. For example, if an article reported that inflation increased by 4%, the score would be higher than if prices increased by 1%.
The data provided information to puzzled clients about a decline in US Treasury yields after a spike earlier this year that was attributed in part to expectations of a lasting resurgence in inflation. The graph below tracks the weekly number of news articles containing the words “reflation” versus “transient” and “inflation” in headlines.
Grinis said the approach allows users to track the importance of “reflation” and “transitory inflation” narratives in real-time market commentary.
The chart shows that the reflation narrative started to gain momentum in the summer of 2020, peaking in late February and early March, ahead of the yield on 10-year T-bills TMUBMUSD10Y,
peaked at the end of March at just under 1.80%. On June 10, when the consumer staples index posted a larger-than-expected monthly rise of 0.7%, bond yields continued to decline, in line with the tale of transient inflation, Grinis noted.
Since mid-June, narratives of reflation and transient inflation appear to have oscillated in importance, Grinis said. The stimulus narrative has risen around the release of June CPI data on July 13, which is again much warmer than expected. The 10-year yield first rose and then fell back, in line with lower reflation fears, she said.
The 10-year rate fell below 1.15% on Monday to reach its lowest level in nearly six months. The US stock market, meanwhile, continued to climb through 2021, but with different sectors and styles coming in and out of favor as expectations about inflation shifted.
The Dow Jones Industrial Average DJIA,
is up more than 14% year-to-date, while the S&P 500 SPX,
grew by more than 17% and the high-tech Nasdaq Composite COMP,
is up more than 14%.
NLP analysis can also be used by businesses to monitor conference calls and corporate 10-Q and 10-K filings, which can provide a holistic view of rising input prices, pass-through plans. price increases and other markers of inflation, said Luo of Wolfe Research.
While not as timely as spending or price data, such readings can offer insight into labor costs and labor shortages, such as increasing salaries at most major banks for junior bankers, Luo said. It is also useful to scan central bank policy announcements, speeches, and other documents to identify trends.
Of course, many alternative data sets are well outside the price range of individual investors. Additionally, no single data set should be viewed as a silver bullet that will give an investor the key to instant outperformance.
Using alternative data in addition to traditional data provides a more diverse set of inputs. As Luo described earlier, this echoes the investment industry adage that describes holding a diverse basket of assets as the only “free lunch” in finance. The same principle applies to data.
But investors don’t necessarily have to break the bank to integrate real-time insight into data, said Nicholas Colas, co-founder of DataTrek Research.
“Our approach to data is to use freely available resources to assess market narratives,” he said in a July 23 memo. “Due to the economic upheavals of the past 18 months, headlines – and the stories they seem to tell – are often misleading or plain wrong. “
How the American consumer reacts to inflation is a key question for investors in the second half of this year, Colas said. DataTrek’s approach is to look at the volume of US-based Google searches for terms like “cheap” and “discount”.
“A rise would signal growing concerns about inflation which could lead to lower consumer demand,” he said. An updated graph (see below) showing search volumes since January 2019 shows that this has not started to happen. With stable volumes for both terms, stagflation doesn’t look “even close to being a problem for the US economy.”
The example, said Colas, shows that with “the wealth of free information available today, it is possible to find the real story. You just have to know where to look.