Guide

Multifactor productivity explained

Harbor Manufacturing's capital-intensity desk ran a familiar playbook in 2024: labor productivity in durable goods stalled near +0.3% year-over-year while unit labor costs climbed. The firm priced in persistent margin squeeze across three factory sleeves. Then the annual Bureau of Labor Statistics multifactor productivity (MFP) release landed: private nonfarm business MFP rose +1.4% while capital services inputs grew faster than labor hours. The “efficiency” problem was not workers moving slower — it was capital deepening. Robots, software, and extended equipment life raised output per hour only after you accounted for the machines doing more of the work. Harbor's model had treated every labor-productivity miss as a wage-inflation signal and overestimated ULC pass-through by 52 basis points on two industrial lines.

Multifactor productivity — also called total factor productivity (TFP) or the Solow residual — measures how much output grows after subtracting the contribution of measured inputs: labor hours, capital services, and (in detailed KLEMS tables) energy, materials, and purchased services. It answers a question quarterly labor productivity cannot: is growth coming from working smarter, or from adding more machines and materials? Fed researchers, macro strategists, and industrial equity analysts use MFP to judge long-run potential GDP, the sustainability of profit margins, and whether wage gains reflect scarcity or genuine efficiency. This guide covers MFP definition and growth accounting, BLS release anatomy and sector detail, capital services measurement, how MFP differs from output per hour and industrial production, the Harbor Manufacturing refactor, an indicator decision table, pitfalls, and an investor checklist.

What multifactor productivity measures

MFP is the portion of output growth not explained by growth in measured inputs. In the standard two-factor decomposition BLS publishes for major sectors:

Output growth ≈ MFP growth + weighted labor-input growth + weighted capital-input growth

Labor input is measured in hours worked (sometimes adjusted for labor composition — education and experience). Capital input is measured in capital services — the flow of productive capacity from equipment, structures, intellectual property, and inventories, not simply the dollar stock of capital on a balance sheet. When output rises 3% and combined input growth accounts for 2%, MFP contributes roughly 1% — the efficiency gain economists associate with technology, process improvement, and better organization.

MFP is not a direct measure of innovation. It is a residual. Mis-measured inputs, utilization swings, and aggregation error all land in the MFP line. Treat single-year moves as suggestive; decade averages are more informative for structural trend.

Growth accounting and the KLEMS framework

BLS productivity accounts build on KLEMS — capital (K), labor (L), energy (E), materials (M), and purchased services (S). The headline two-factor MFP series collapses energy, materials, and services into intermediate inputs or omits them depending on the sector table; detailed industry releases expose all five when data allow.

Input weights and value shares

Each input's contribution to output growth equals its growth rate times an income share weight — roughly how much of output revenue pays for that input in competitive equilibrium. Labor's share has drifted from ~65% in the 1970s toward ~60% in recent decades; capital's share rose. When capital's weight is higher, faster investment boosts measured output growth even if MFP is flat — a pattern visible in post-pandemic manufacturing capex cycles tied to industrial production recoveries.

Capital services vs capital stock

A \$10 million CNC machine contributes more to current-year capital services in its first productive years than a 20-year-old asset of the same book value. BLS constructs capital services using rental-price weights that reflect depreciation, interest, and expected returns by asset type (equipment, structures, IP products). Software and R&D capital grew as a share of services after BEA expanded IP investment in national accounts — lifting measured capital input growth and sometimes depressing MFP when output lags the new intangible stock.

Labor composition

Raw hours understate labor quality when the workforce gains education. BLS publishes labor composition–adjusted hours in detailed tables: a hour from a more experienced worker counts as more than one standard hour. Composition adjustments can explain part of productivity gains that look like pure MFP in unadjusted data.

BLS multifactor productivity release anatomy

Unlike quarterly Productivity and Costs, the major MFP release is annual, typically publishing in spring for the prior calendar year (e.g., March 2026 for 2025). Supplemental industry detail and revised vintages follow in summer and fall.

Series / table	What it shows	Typical use
Private business MFP	Headline TFP for private nonfarm economy	Long-run potential GDP, Fed research
Private nonfarm business	Excludes farms; matches much of GDP	Macro strategy, equity sector allocation
Manufacturing (total, durables, nondurables)	Factory MFP with capital/labor splits	Industrial margin and capex models
20+ 3-digit NAICS industries	Industry MFP when sample size permits	Stock picking, supply-chain analysis
Contribution tables	Percent of output growth from each input + MFP	Decompose whether growth is input-heavy

Revision and benchmark cycles

MFP revises when BEA updates GDP, investment, and IP capital stocks. Major benchmark revisions (every five years) can shift decade-average MFP by several tenths. The 2024 vintage incorporated expanded software and R&D capital, revising post-2010 manufacturing MFP slightly lower and capital contribution higher — a reminder that “weak productivity” headlines sometimes mean “we finally counted the robots.”

Link to GDP and GDI

Output in MFP accounts aligns with value-added measures in GDP and the income side in GDI. When statistical discrepancy widens between expenditure and income GDP, productivity residuals can look unstable until BEA reconciles. Cross-check MFP narratives against corporate profit growth and capex deflators, not labor data alone.

MFP vs labor productivity vs unit labor costs

Three BLS productivity concepts sit on a ladder of increasing input coverage:

Measure	Inputs accounted for	Frequency	Best question it answers
Labor productivity (output per hour)	Labor hours only	Quarterly	Timely wage-inflation bridge via ULC
Unit labor costs	Compensation vs output per hour	Quarterly	Near-term price pressure
Multifactor productivity	Labor + capital (+ materials in detail)	Annual	Long-run efficiency and potential growth

Labor productivity can rise because firms add capital per worker (capital deepening) even when MFP is flat — each worker tends more machines. Conversely, MFP can rise during layoffs if low-productivity hours drop faster than output (a cyclical distortion). Harbor's error was equating a labor-productivity stall with inefficiency; MFP showed capital input growth explained most of the gap, leaving unit nonlabor costs — not ULC — as the margin lever.

Harbor Manufacturing capital-intensity desk refactor

Harbor's industrial sleeve model used quarterly output per hour as a proxy for structural efficiency. When durables labor productivity printed +0.3% YoY in 2024 while wages ran +4.1%, the desk projected ULC-driven PPI pass-through of 80–90 bps on two lines. After integrating annual MFP:

Capital services growth +3.8% on the matched NAICS slice — automation and extended asset lives.
MFP contribution +1.4% — positive, not the stagnation labor-only data implied.
Labor hours −0.6% — headcount cuts raised output per hour mechanically without MFP gain.

Refactored margin forecasts cut assumed PPI pass-through from 85 bps to 33 bps on the two sleeves (52 bps total error removed). The desk now gates wage-inflation alerts: labor productivity below +1% triggers a ULC watch only when MFP is also below its 10-year sector median and capital services growth is below capex plan. Capital-heavy quarters route to unit nonlabor payment analysis instead.

Technique decision table: when to use MFP vs labor productivity

Your goal	Use labor productivity / ULC	Use multifactor productivity
Fed meeting / CPI quarter-ahead	Yes — timely, wage-linked	No — too lagged
10-year potential GDP estimate	Partial — misses capital	Yes — primary input
Factory automation capex thesis	Risky — confuses deepening with TFP	Yes — split capital vs MFP
Single-stock margin after layoffs	Misleading — hours denominator shock	Yes — check MFP vs composition
Monthly payroll-week trading	Yes — AHE and hours	No
AI / software productivity narrative	Insufficient	Yes — IP capital + MFP residual

Historical context and reading the trend

U.S. private business MFP averaged roughly +1.0% per year from 1995–2004 (productivity boom), faded toward +0.3% pre-pandemic, and spiked volatile around 2020–2022 as output and hours desynchronized. Post-2022 normalization saw MFP recover toward +1% while labor productivity swung negative — classic capital-deepening signature as firms invested in equipment and software faster than they rehired.

Decade averages matter more than single prints. Compare MFP to hours worked trends and to corporate profits: sustained MFP plus flat profit share suggests gains flow to workers; MFP stagnation with rising profit share often points to markups or market power rather than measured efficiency.

Pitfalls

Treating MFP as quarterly — annual release lags; do not trade CPI off March MFP alone.
Confusing capital deepening with TFP — more machines per worker raises labor productivity without innovation.
Ignoring utilization — running existing plants harder boosts output without MFP; recessions invert the bias.
Single-industry MFP with thin samples — BLS suppresses noisy NAICS cells; check footnotes.
Equating MFP to “technology” — residual absorbs measurement error, markups, and reallocation.
Using book-value capex as capital input — BLS capital services use rental prices and age-efficiency profiles.
Overlooking IP capital revisions — software/R&D reclassifications shift history materially.

Investor and analyst checklist

Bookmark BLS Multifactor Productivity release calendar (annual, ~March).
Download contribution tables: labor, capital, MFP shares of output growth.
Compare manufacturing durables vs nondurables MFP before industrial earnings.
Cross-check capital services growth against BEA fixed investment and capex plans.
Pair annual MFP with quarterly ULC for near-term vs structural inflation read.
Adjust labor productivity reads when hours swing from layoffs or rehiring.
Track 10-year MFP averages; ignore single-year pandemic-era outliers.
For sector stocks, use 3-digit NAICS MFP when published; else use parent sector.
Reconcile MFP narrative with corporate profit share and PPI margin data.
Re-benchmark models after BEA five-year national accounts revisions.

Key takeaways

Multifactor productivity measures output growth after labor and capital inputs — the efficiency residual, not just output per hour.
Capital services, not capex dollars, are the correct capital input; IP and software increasingly matter.
Labor productivity and MFP diverge when firms automate: rising output per hour can mask flat TFP.
Harbor Manufacturing cut PPI pass-through error 52 bps by gating ULC alerts on MFP and capital services, not labor productivity alone.
Use quarterly labor productivity for timely wage inflation; use annual MFP for long-run potential growth and capex theses.