NFP +115K Hides a Breakeven Labour Market

Headline: 115,000. Look at the trailing three months. February: -156,000 (revised). March: +185,000. April: +115,000. Average: 48,000.

That 48,000 number is what economists call the breakeven rate — the level of payroll growth needed to absorb new workers entering the labour force from population growth and immigration. Below breakeven, the unemployment rate rises mechanically. At breakeven, it stays flat. Above breakeven, it falls.

Three months at breakeven, with the trajectory deteriorating from the +200K average that prevailed through 2025, is exactly the pattern that preceded the December 2007 onset of the Great Recession-era cuts. The Fed cut from 5.25% in September 2007 because the three-month payroll average had fallen to 80K — and the recession had already started two months earlier.

The consensus reads +115K and writes 'beat'. The Pantheon Macroeconomics forecast — U-rate to 4.7% by year-end, first cut in December — reads the same +115K as the second-derivative collapse it actually is.

Average hourly earnings: +0.2% MoM, +3.6% YoY. Both miss consensus. Both confirm what survey data has been saying for two months.

3.6% YoY wage growth, with productivity growth around 1.5%, implies unit labour cost growth of roughly 2.1% — *consistent with the Fed's 2% inflation target*. The wage channel is no longer pushing services inflation higher. It is no longer the obstacle to rate cuts.

This is precisely the data the FOMC's dovish faction has been waiting for. Miran's 25bp cut dissent on April 30 looked aggressive when wages were running at 4%+. With the April print in hand, it now looks prescient. The hawks (Bowman, Mester, Schmid) wanted to *hike* on the oil shock. Wage data this soft makes the hike case nearly impossible to defend at the next meeting.

Both tails of the dissent map move toward the centre. The next FOMC reads more dovish, not more hawkish, on this data.

Of the 115,000 net new jobs, healthcare added 37,000 — about 32% of the total. Healthcare is the most lagging sector in the US labour-cycle taxonomy. It is structurally driven by demographic ageing and largely insensitive to the business cycle until the cycle has fully turned.

When healthcare is carrying a third of payroll growth, the cyclical sectors are *not* contributing. Manufacturing, construction, professional and business services, retail (despite this month's positive contribution) — these are the categories that move first into a downturn. The mix in the April report tilts heavily toward the non-cyclical.

This is the same composition pattern the BLS reported in late 2007: healthcare carrying the headline, cyclical sectors flat, then negative. The 2019 mini-cycle had a similar fingerprint before the Fed cut three times in 2019 H2.

The unemployment rate held at 4.3%. The composition of *how* it held there is the contrarian read.

The labour force participation rate contracted in April. Fewer working-age people are seeking work. That mechanically holds the unemployment rate down even when actual hiring slows — the denominator shrinks faster than the numerator can fall.

If participation had been flat instead of contracting, the April unemployment rate would likely have printed 4.4% or 4.5%. The 'unchanged at 4.3%' headline obscures a labour market where workers are being discouraged out of the count. That is a recession-tilted read of the same data the hawks treat as evidence of resilience.

This matters for the Fed. The dot plot pays attention to unemployment levels. It does *not* pay similar attention to participation dynamics. So the Fed's reaction function will lag the underlying signal — exactly the policy-error setup that produces a too-late easing cycle followed by aggressive cuts.