By Carl Tannenbaum

I worked in Europe for two weeks earlier this month.  Among the regional treats I enjoy most are pretzels, which local bakeries take to a high level.  There is a divide, however, between those who enjoy soft pretzels and those who prefer hard pretzels.

A divide has recently developed between soft and hard economic data.  At a time when conditions are changing rapidly, understanding the difference between the two is terribly important.

In the vernacular of economists, “soft” data is generated by surveys.  It gauges impressions about current and future conditions.  Examples include measures of consumer and business confidence.  The construction of soft data is not uniform: some series are indexed relative to a baseline date, while others are diffusion indices.  (A common diffusion index takes the difference between positive and negative responses to a survey.)

“Hard” data results from more rigorous accounting for things like spending and output.  Sampling and assumptions are commonly used to assemble these statistics, but they are considered to be more objective than softer metrics.

Sentiment surveys ask about prospects going forward, in an attempt to be forward-looking.  Hard data is retrospective.  Economists can use histories of hard data to model what might happen in the future: this is the bedrock of forecasting.  But during times of rapid change, history may not be a reliable guide to what lies ahead.

The current environment may fall into that latter category.  Tariffs are on their way to levels that we have not seen in a century, depressing growth and placing upward pressure on the price level.  It is too early to see direct impacts on quantities like gross domestic product (GDP).  And so soft data is getting additional attention.

Some of the survey data provides cause for concern.  The University of Michigan consumer confidence index has fallen sharply in the last several months, and stands close to a record low.  Surveys of business confidence are off their peaks, likely the result of very high levels of trade policy uncertainty.

Survey measures aren’t called soft solely because of their construction.  Their ability to presage trends in hard data is not altogether strong.  People often say one thing in response to surveys and then do very different things with their wallets: the statistical link between confidence levels and future changes in spending is not very strong.  This may be due, in part, to the role that non-economic factors play in household sentiment.

Business sentiment is slightly more telling.  A lack of confidence inhibits the willingness of firms to commit to forward-looking investments and increases in capacity.  Companies that are apprehensive about the future are typically more conservative in their hiring plans.  This, in turn, limits wages and consumption.

The general concern, though, is that we can talk ourselves into a recession.  Households worried about what lies ahead will curtail spending; worried business won’t hire and expand.  In this way, negative sentiment can become self-fulfilling.

We should all hope that is not the case this time around.  Surveys of U.S. households reflect pessimism over the labor market and expectations for higher inflation.  Partly on the basis of this soft data, U.S. forecasters have been raising odds that a recession will occur during the next twelve months.

For now, there is scant evidence of a downturn in any of the hard data.  While U.S. consumer spending had a sluggish start to the year, seasonal factors have played a heavy hand in driving the results.  Layoffs within the Federal government have cast a pall over the nation’s capital, but the balance of employment data offers few hints of trouble.

When it comes to pretzels, I prefer the soft variety.  The same goes for economic outcomes; I’ll take a soft landing over a hard landing any day.