Unexpected Developments at the Bureau of Labor Statistics

The July employment report showed 73,000 new jobs added, heavily driven by the healthcare and social assistance sectors. But then things took a surprising turn: President Trump dismissed the Director of Labor Statistics, leading to an uproar in the press. Economists from various cities quickly dropped their planned discussions about issues like authoritarianism and data independence. Yet, as usual, the focus missed the bigger picture.

The revisions for May and June were not just significant; they were exceptional by most measures, particularly considering the context of the post-pandemic environment. These weren’t mere daily updates; they felt more like the ground shifting beneath us.

To clarify, the Bureau of Labor Statistics initially stated that June had seen an addition of 147,000 jobs, but later adjusted that number to 133,000. A closer look revealed a mere 14,000 jobs added when everything was accounted for. While it sounded promising at first, the two rounds of revisions ended June with only 19,000 jobs—totaling a downward adjustment of 258,000 for both May and June.

This marks the most significant revision outside of pandemic-related changes in recent history. We don’t use “most” lightly here; it reflects a statistical reality that’s hard to ignore.

Understanding the Revisions

When we say the revisions for May and June were extreme, we don’t mean they were just noteworthy or somewhat surprising. We’re talking about statistically abnormal events—far beyond what would be considered typical fluctuations. These revisions underscore the chaotic nature of current economic conditions.

Looking at June, the downward shift of 133,000 jobs represents a 3.6 sigma anomaly when measured against historical data, indicating that such a deviation is a rarity—perhaps something one might see only once in a generation. Utilizing Median Absolute Deviation, that June revision stands out prominently, raising eyebrows in ways few instances before have.

And perhaps even more striking is May’s second revision of -125,000, with more than half of that revision reflecting zero job additions. The statistical implications for May are unprecedented; since 2003, we haven’t seen anything like this.

When combined, the adjustments for May and June represent a staggering 6.8 sigma event—placing it in the near-impossibly rare percentile of occurrences. This defies standard explanations, reinforcing the notion that “these things don’t just happen.” And when they do occur, they serve as important signals.

The Misleading Explanations

This brings us to the common justifications presented by Trump detractors. They suggest that declining survey response rates are leading to larger revisions, implying there’s little to be concerned about. While it’s true that response rates have dipped into the low 40% range, making initial figures more volatile, the remarkable nature of the May and June revisions goes well beyond mere instability. They stand as clear outliers, even in this new normal.

So, what’s truly happening? Four key factors emerge.

• A considerable federal layoff: The government has been reducing its workforce at levels rarely seen, and while this often doesn’t show up in monthly statistics, it can distort the revision process significantly.
• Monetary policy changes: Following Trump’s election, the Federal Reserve quickly shifted from a supportive stance to a more restrictive one, even as core inflation was slowing. This rapid policy change might suggest that the labor market is weakening at a pace faster than historical norms.
• Immigration enforcement: The ongoing crackdown on illegal immigration is likely affecting job availability and pay in ways that aren’t directly visible in unemployment claims, adding a structural change that muddles the data.
• Customs duties: This factor is perhaps the weakest. Uncertainty regarding tariffs can delay hiring decisions, particularly among small and medium enterprises that depend on international resources.

Panic in the Political Sphere

The political landscape has erupted in response to Trump’s firing of the BLS Commissioner. Claims have surfaced asserting that our statistical methods may soon be as unreliable as those from other countries—a rush to judgment. Critics assert that revisions are just typical when, in reality, these adjustments are among the largest and most statistically significant on record. Ignoring this reality just makes Trump appear worse.

Moreover, Trump didn’t appoint a crony; he promoted a long-time BLS veteran as a member of the committee, which actually suggests a desire for continuity.

The real issue isn’t Trump’s loss of faith in the previous commissioner but rather the broader distrust within the commentary community regarding data interpretation—especially among those whose narratives don’t align with the numbers. They seemed untroubled by the anomalies until a convenient, anti-Trump story was available.

In truth, the only craziness here is pretending otherwise.