August 15, 2025

The world has gone topsy turvy since the BLS revised downwards significantly the employment data for May and June.  Trump claimed that the downward revisions were politically motivated and designed to make him and his Administration look bad.  Nonsense.  Here’s a news flash.  All data are flawed and all data get revised as additional information is received.  Get used to it.  Politics do not influence the outcome.

No economic indicator is the absolute truth.  They are all flawed because they are nothing more than an estimate based on the available data.   The first glimpse at an indicator for any given month is particularly at risk.  It will get revised in subsequent months as additional data are received.  And even after several months of revisions the data are still inaccurate.  The data collecting agencies use sample data to estimate the behavior for a much larger group of people or businesses.  Despite sophisticated data collection procedures the end result is simply an estimate of what is  happening.  That is why economists, investors, and policy makers need to look at a variety of indicators to get a sense for what is actually happening to GDP growth, the labor market, or inflation.  While all of these indicators  may be inaccurate, they are not politically biased.

Suppose for a moment that the BLS wanted to skew the employment data to fit some political agenda.  The payroll data are derived from the establishment survey of businesses.  The BLS simultaneously conducts a survey of households to determine the unemployment rate.  The first problem would be to get both surveys to tilt in the desired direction.

Then, recognize that  those are not the only two sources of information about the labor market for any given month.  The ADP employment report, for example, is published two days prior to the BLS report.  The two surveys can differ significantly from month to month.  And both initial estimates will be revised in each of the next two months.  Initial differences may disappear as more data are collected.

In addition, there are data on initial unemployment claims, the number of people receiving unemployment benefits, job openings, hiring, and separations.  They all tell part of the story.  If the BLS decided to skew the outcome, a wide number of public and private organizations would have to in on the game.  If such a widespread conspiracy were to occur lots of people would know and, almost certainly, somebody would want their 15 minutes of fame and become a whistleblower.

We are convinced that BLS employees are hard working individuals trying to do the best job they can with insufficient  money and not enough bodies to produce a reasonably accurate portrayal of what is happening in the labor market.  They deserve praise, not ridicule.

As has been widely reported, the  primary reason why revisions to the employment data have gotten bigger is because the participation rates have fallen markedly.  Why is that?  We can think of a couple of reasons.  First, the number of surveys to which a firm is asked to respond has risen dramatically.  Each form requires time and costs money to complete.  The internet has made it possible for anybody to conduct a survey and ask businesses or consumers to participate.  Think about the number of surveys each of us as consumers are asked to complete these days.  Any time somebody comes to our house to do anything, we go to a doctor’s office, or visit a government agency they send  out a survey asking how they did.  And they sometimes proudly assert that it will only take 5-10 minutes of your time to complete.  We suffer from survey overload.

Second, every time a firm or consumer responds to one of these surveys it is exposing its computer system to potential bad guys.  It is risky.  Everybody runs the risk of being hacked.

Third, responding to one of these surveys can reveal important corporate strategy which could find its way into the hands of a competitor.

While all surveys are flawed, consumer surveys seem particularly vulnerable.  The University of Michigan does a monthly survey of consumer sentiment.  The Conference Board does a similar survey of consumer confidence.  They ask consumers how they feel about current conditions in the economy, their expectations about the economy  six months from now, and what they expect the future inflation rate to be.  People’s views are almost certainly swayed by whatever they read on the internet.  As we all know, anybody can put up any senseless item they want on the internet.  Social media influencers are motivated by how many followers they have so they tend to write sensationalist pieces that are designed to attract readers.  They are not interested in the accuracy of whatever they are saying.  Then, AI searches incorporate that nonsense into its responses.  It becomes an exercise in groupthink.

The  point of all this is that every economic indicator is based on a relatively small sample of firms or consumers designed to estimate what is happening in a much larger universe.  As a result, they are all inaccurate and potentially misleading.  Economists, consumers, and politicians should view all of them with a healthy dose of skepticism.  They are based on small samples.  Adjustment factors designed to account for normal seasonal movements can be inaccurate.  They can skew a monthly economic indicator in one direction one month and in the opposite direction the following month. For this reason always look at the entire spectrum of data on employment or GDP or inflation before reaching a conclusion about what is happening.  Also  be aware that every indicator is going to get revised as additional data are received.  Most of the time the revisions will be fairly small but sometimes they are not.  Question the data, but do not shoot the messenger.

Stephen Slifer

NumberNomics

Charleston, S.C.