No Hard Feelings: Soft vs. Hard Divide Persists

May 27, 2025 • Liz Ann Sonders • Kevin Gordon
There is still a wide divergence between hard and soft data, and a recovery in the latter is likely to be weak absent a meaningful reduction in policy uncertainty.

In economic analysis, the distinction between "soft" and "hard" data is crucial for interpreting the health and trajectory of the economy. Soft data refers to sentiment-based measures such as surveys, expectations, and confidence indicators, while hard data encompasses quantifiable economic outputs like employment numbers, retail sales, and industrial production. Recent divergences between weaker soft economic data and more resilient hard data have muddied the outlook. The weaker soft data undoubtedly has been weighed down by the ongoing trade war, along with myriad additional policy uncertainties.

Shown below, Bloomberg's Economic Surprise Indexes track how economic data releases compare to consensus expectations. In recent months, a growing divergence has emerged: the Soft Data Surprise Index has turned negative, while the Hard Data Surprise Index has remained positive, reflecting better-than-expected data—in particular, across labor market readings.

Surprise!

In recent months, a growing divergence has emerged: the Bloomberg Soft Data Surprise Index has turned negative, while the Hard Data Surprise Index has remained positive, reflecting better-than-expected data—in particular, across labor market readings.

Source: Charles Schwab, Bloomberg, as of 4/30/2025.

Bloomberg "Hard Data" and "Soft Data" Surprise Indexes measure the difference between actual data and analysts' forecasts. Indexes are unmanaged, do not incur management fees, costs and expenses and cannot be invested in directly. Past performance is no guarantee of future results.

Consumers' woes

The expectations component of The Conference Board's Consumer Confidence is a component of the Index of Leading Economic Indicators (LEI). Shown below via the black vertical dotted lines and where they fall on the orange correlation coefficient line, are all monthly Consumer Confidence Expectations readings since last November and through May. The 40-point plunge in expectations from the post-election peak implied weaker real gross domestic product (GDP) readings and indeed, first-quarter real GDP was reported in slight negative territory. However, here is an example of soft data playing some catch-up to hard data: notice that May's reading (green vertical dotted line) was a significant improvement from April. That's in keeping with stronger expected second-quarter GDP estimates to date.

Ascending expectations?

Both 1-year and 5-10-year inflation expectations have spiker higher while core CPI has continued to trend lower in year/year terms.

Source: Charles Schwab, Bloomberg, The Conference Board, The Leuthold Group. 1967-5/27/2025.

Blue dots represent year-end readings. Vertical dotted lines represent Consumer Expectations Index reading for corresponding month. Correlation is a statistical measure of how two investments have historically moved in relation to each other, and ranges from -1 to +1. A correlation of 1 indicates a perfect positive correlation, while a correlation of -1 indicates a perfect negative correlation. A correlation of zero means the assets are not correlated. Indexes are unmanaged, do not incur management fees, costs and expenses and cannot be invested in directly. Past performance is no guarantee of future results.

Inflation woes

Components of Consumer Sentiment data from the University of Michigan (UMich) include inflation expectations. This is another example of the soft vs. hard dichotomy. As shown below, both shorter- (blue line) and longer-term (orange line) inflation expectations have spiked higher; while the actual core Consumer Price Index (CPI, green line) has continued to trend lower in year-over-year terms.

Inflation expectations > actual inflation

Both 1-year and 5-10-year inflation expectations have spiker higher while core CPI has continued to trend lower in year/year terms.

Source: Charles Schwab, Bloomberg, Bureau of Labor Statistics.

UMich (University of Michigan) expectations as of 5/16/25. CPI as of 4/30/2025. Forecasts contained herein are for illustrative purposes only, may be based upon proprietary research and are developed through analysis of historical public data.

More "seasoned" readers might remember the Misery Index, created by economist Arthur Okun in the 1960s. It is calculated by adding the seasonally adjusted unemployment rate to the annual inflation rate. Shown below—and highlighting yet another wide soft vs. hard gap—the expected Misery Index (combining UMich one-year inflation expectations with UMich higher unemployment expectations, orange line) is up on a dramatic spike. That's in contrast to the actual Misery Index (blue line), which continues to trend lower.

Expected misery > Misery Index

The "Expected" Misery Index is up on a dramatic spike while the actual Misery Index continues to trend lower.

Source: Charles Schwab, Bloomberg, The Leuthold Group, University of Michigan (UMich).

Misery Index as of 4/30/25. "Expected" Misery Index as of 5/16/25. Forecasts contained herein are for illustrative purposes only, may be based upon proprietary research and are developed through analysis of historical public data.

Small businesses' woes

Small businesses are likely in the least favorable position in terms of the global trade war—not least because of their inability to easily move production, trim costs, and/or reduce headcount without seeing harm to operations. There was a ton of outlook optimism in the small business community in the immediate aftermath of the election, but about half of that spike has reversed this year, as shown below.

Small business optimism fading

The share of small businesses expecting a better economic backdrop is fading quickly, after a post-election surge.

Source: Charles Schwab, Bloomberg, National Federation of Independent Business (NFIB), as of 4/30/2025.

We have mentioned this before, but the small business optimism data from the National Federation of Independent Businesses (NFIB) tends to be politically skewed: optimism tends to surge after a Republican takes the White House and vice-versa when a Democrat comes in. In the past decade, swings in confidence have gotten more aggressive.

Most of those swings have been driven by the soft components. As shown below, NFIB breaks down the subcomponents of its overall optimism into hard and soft categories. In keeping with the "vibecession" (when the soft data fell to recession-like levels but the hard data never did) we experienced a couple years ago, the soft components (orange line) plunged to all-time lows; in fact, and stunningly, they declined in 2022 to levels worse than what occurred during the Global Financial Crisis.

Conversely, the hard components (blue line)—though having deteriorated—did not see nearly as strong of a decline. They fell to levels consistent with prior slowdowns but did not come close to matching their soft counterparts.

Soft and hard stories not similar

In keeping with the "vibecession" we experienced a couple years ago, the soft components in the NFIB survey plunged to all-time lows a couple years ago; in fact they fell to levels worse than what we saw during the Global Financial Crisis.

Source: Charles Schwab, Bloomberg, NFIB (National Federation of Independent Business), as of 4/30/2025.

Hard and soft represents the sum of their respective components.

The same goes for the upswing after the election. Soft components soared back into net positive territory—presumably because of the anticipation of a more favorable policy backdrop. The hard components saw a bump higher but again, nowhere to a degree that matched what the soft data saw.

The May update for the NFIB survey will incorporate some more tariff war easing, given the series of pauses on tariffs the White House has implemented over the past month. We're not sure that confidence will come roaring back, though—not least because there is still uncertainty as to what the goal is when it comes to tariffs, what the ultimate tariff rate will be for each country, and what sectoral tariffs will stick, among other things.

There are early signs that the recent tariff pauses are not causing a significant rebound in soft data across the board. Shown below, that is evident in the Philadelphia Fed's Manufacturing (blue line) and Services Indexes (orange line). Through late May, the former had a rather sharp bounce while the latter's move up was minuscule. Both remained in contraction for the month, indicating that overall activity is still depressed.

Services struggling in Philly

Through late-May, the Philly Fed Manufacturing Index saw a rather sharp bounce while the Services Index's move up was minuscule. Both remained in contraction for the month, indicating that overall activity was still depressed.

Source: Charles Schwab, Bloomberg, Philadelphia Federal Reserve, as of 5/20/2025.

The Philadelphia Fed Manufacturing Index measures manufacturing activity in the Third Federal Reserve District, covering Pennsylvania, New Jersey, and Delaware.  The Services Index monitors nonmanufacturing activity and gathers "soft" data in the form of responses from business owners, executives, and managers. Indexes are unmanaged, do not incur management fees, costs and expenses and cannot be invested in directly. Past performance is no guarantee of future results.

The rebound in manufacturing isn't much of a surprise when considering that manufacturing is at the epicenter of the global tariff war. What might be increasingly worrisome is if the services sector deteriorates more, or takes longer to recover. This is a good time to emphasize that the United States has a sizable services surplus, and given services is a larger chunk of the economy, a sizable hit to services demand will have larger economic implications.

Each month, the Philadelphia Federal Reserve (and other regional Fed banks) adds special questions to its survey, and one from this month stuck out regarding inflation. A shown below, when businesses were asked if their customers have become less price sensitive, 0% said "yes." Conversely, half of respondents said they've become more sensitive.

Consumers increasingly price sensitive

When asked by the Philadelphia Fed if their customers were becoming less price sensitive, 0% of businesses said "yes" in May.

Source: Charles Schwab, Philadelphia Federal Reserve, as of 5/20/2025.

The Philadelphia Nonmanufacturing Business Outlook Survey is a monthly survey of nonmanufacturers in the Third Federal Reserve District, covering Pennsylvania, New Jersey, and Delaware.

It's a stark, but necessary reminder that when it comes to the everyday consumer, inflation is often about psychology more than precise mathematics. Wall Street folks often obsess over month-over-month, year-over-year, or three-month annualized changes in the consumer or producer price indexes. Yet, the reality is that the masses will focus on how much they're paying for goods and services—and whether the price is higher than their chosen reference point. Given the increase in price sensitivity driven by tariffs, we think anxiety over price volatility will likely keep some downward pressure on soft data. That doesn't mean confidence can't improve, but the strength of the rebound might be dampened by an unknowable path forward (for now) when it comes to trade policy.

Analysts' woes

Given corporate leaders' concerns about policy—notably re: tariffs—a rising share of S&P 500 companies have either withdrawn forward-looking earnings guidance or have lowered it. This has resulted in a meaningful decline in the earnings revision ratio (orange line); while actual earnings growth (blue line) has been trending higher, at least through this year's first quarter. In other words, the trade war did not have a material impact on trailing earnings through the first quarter; but analysts have decidedly not extrapolated that strength into estimates for the remainder of this year.

No extrapolating of strong 1Q25 EPS

There has been a meaningful decline in the earnings revision ratio; while actual earnings growth has been trending higher, at least through this year's first quarter.

Source: Charles Schwab, Bloomberg, LSEG (London Stock Exchange Group) Datastream, as of 5/23/2025.

Earnings revision ratio shows the number of forward 12-month estimates that have been revised upwards vs. downwards, divided by the total number of estimates. Indexes are unmanaged, do not incur management fees, costs and expenses and cannot be invested in directly. Past performance is no guarantee of future results.

Conclusion

We have no clearer crystal ball as it relates to ongoing policy-related uncertainty. The mercurial nature of the Trump administration's trade policy is likely to keep expectations a bit shaky. Businesses have proven time and again they can be nimble and make adjustments to policy changes. The rub in this era is that there is little sense of the playing field—or even the rules of the game—on which companies are operating. Some clarification (or finalization) on tariffs could lead to some catch-up by the soft data. Absent that clarification, it is more likely that the hard data catches down to the soft data. Soft data should continue to be viewed as an early-warning system and both hard and soft data must be interpreted in that context—especially as economic cycles have become more complex and nonlinear.

Diversified? Tax efficient? Cost effective?

Explore more topics