The algorithm is winning. We live in a time where every click we make and every step we take generates metadata; sometimes just uttering a thought out loud can trigger an ad on a social media feed. This is also an age where disinformation and political leanings have been weaponized to great effect on social media. So how does the algorithm affect the leading sentiment indicators studied by investors today? Based on a comparison of the current economic cycle and the Great Financial Crisis in 2008, quite a lot.
Late last year, the market had already decided the Federal Reserve Bank had reached the end of its interest rate tightening cycle. Investors were mildly optimistic the Fed would engineer a “soft landing” with respect to a recession. Inflation, which peaked in June 2022 at more than 9%, had drifted back down to 3.2% by October. This downward trend in inflation was happening not because demand had cratered. In fact, the U.S. gross domestic product (GDP) was still up nearly 5% in the third quarter of 2023.
The U.S. consumer was still spending. And they were still working. The U.S. unemployment rate had drifted up slightly but was still under 4% — or what the Fed considers “neutral” or full employment. To top it all off, the S&P 500 was up more than 20%. Most economic indicators were positive. Since it’s the direction of the economy that typically drives expectations, it would have been reasonable to see consumer sentiment ticking up in kind. That was not happening.
Comparing sentiment during big events
One of the longest-running and most referenced measures of consumer sentiment, the Michigan Consumer Sentiment Index (MCSI) was launched in the 1940s and aims to gather opinions from Americans on their expectations for spending and inflation and their overall feeling about the business climate and employment. At the end of 2023, the MCSI came in at 69.7. That is not exactly bubbling with enthusiasm. For context, the MSCI’s all-time low was 50 in June 2020, during the COVID-19 pandemic, and its all-time high was 112 in January 2000, during the tech bubble. The average reading is 84.9.
By comparison, on September 30, 2008, the MCSI came in at a very similar 70.3. However, the backdrop was quite different. Lehman Brothers had just collapsed earlier that month. The S&P 500 had already slid 20% anticipating the subprime mortgage crisis. In stark contrast to the fourth quarter in 2023, U.S. GDP had plummeted to 2.3% from 5.0% a year earlier. On top of the data, the average person was hearing daily about doomsday scenarios and a global financial meltdown. And yet, the MCSI was actually higher than in December 2023. What could explain that?
Perhaps inflation was the culprit? It was much higher in 2023 than 2008 and the highest it had been since the early 1980s. But expectations for inflation (and expectations in this case really do matter) had improved dramatically. Inflation in September 2008 was at 2.5% vs 3.8% at the end of 2023. But that 3.8% had been rapidly descending throughout the year (as predicted). Since the long-term average for inflation is 3.7%, it is unlikely the rate alone would lead to Great Financial Crisis-levels of pessimism.
Fear of a Fed misstep?
Another possibility is that the fear of a Fed misstep had the average person pessimistic. After all, as inflation ticked up, so too did the retrospectives about the Fed’s missteps in the 1970s that ultimately led to 4 million jobs being lost before inflation was whipped. However, the market certainly didn’t seem to fear that result en masse in 2023. The S&P 500’s 11% gain in the fourth quarter, although AI-driven to a large degree, was fueled in part by a belief the Fed would not misfire.
It is also possible that those creating economic news are affected by these trends, too. The Brookings Institution studied the San Francisco Fed’s Daily News Sentiment Index and compared it to economic data. The Daily News Sentiment Index creates a daily time series of sentiment from analyzing economic news stories from 24 daily newspapers. Brookings found that economic news itself had become more pessimistic since 2016 and had diverged from economic data in the same manner as consumer sentiment. Brookings found that from 1988 to 2016, news “sentiment” changed consistently with the actual economic data. Since 2016, the economic data and news sentiment decoupled.
Social media’s rise
There is one explanation for the 2023 year-end consumer (and news) cynicism that seems far more likely: social media — and its evolving influence in society since 2008.
In 2008, Facebook and Twitter were nascent endeavors still years away from going public and wielding massive influence. TikTok was nearly a decade away from relevance. It was the first year that the internet overtook print newspapers as a primary news source. TV news in 2008 had already morphed from the days when CBS Evening News anchor Walter Cronkite was voted the Most Trusted Man in America in 1972. But in 2008, television was still the primary news source and had yet to compete with mass clicks, tweets and disinformation for viewership. This changed dramatically over the past 15 years.
How information is disseminated and used can be transformative to society. Starting in the 15th century, Gutenberg’s printing press distributed religious doctrine as well as scientific discovery to the masses. Franklin Delano Roosevelt leveraged the new radio technology to deliver Fireside Chats that could bypass the press and talk directly to Americans about the New Deal. The FCC’s 1987 repeal of the Fairness Doctrine empowered the rise of politically aligned media and allowed conservative talk shows, in particular, to wield considerable influence.
Echo chambers of like-minded views
Social media transformed society, too. It hastened the speed of news dissemination, something critical during natural disasters or global pandemics. But it enabled users to self-select their experience, building an echo chamber of like-minded views. The algorithm amplified the echo, serving up ads (informative or disinformative) and content in service to other actors. The result(s)? Studies have connected social media use to political polarization and decreased mood and consumer sentiment. Recently, a study out of the U.K. correlated MCSI data to social media influencer posts, finding “consumer sentiment is affected by social media influencers and is not just a reflection of economic variables.”
The influence of social media on society affects more than just the consumer sentiment indicators used by investors. But with artificial intelligence just kicking in, the algorithm is going to wield even more influence going forward. Investors will need to develop some sensitivity analysis when evaluating consumer sentiment — and seek to better understand the intersectionality of consumer sentiment, political polarity, macroeconomic data on the markets and consumers’ own personal finances.
The information in this publication is based primarily on data available as of March 31, 2024, that has been obtained from sources believed to be reliable, but its accuracy, completeness and interpretation are not guaranteed. In addition, this publication contains the opinions of the authors as of that date and such opinions are subject to change without notice. We do not think this publication should necessarily be relied upon as a sole source of information and opinion. The use of Generative Artificial Intelligence may present risks such as undetected errors, defects, and security vulnerabilities. This publication is not a recommendation of, or an offer to sell or solicitation of an offer to buy any particular security or investment product. Bailard cannot provide investment advice in any jurisdiction where it is prohibited from doing so.
