Measurement Frame Impossibility: why consumption welfare can rise while system welfare falls

MFIT says consumption-based welfare misses losses outside prices, quantities, and household reports

The paper defines a bilateral measurement space B made from household transactions, surveys, self-reports, capability indicators, and distributional accounts. It then defines true welfare as W* = W(B) + ΣΦ (s_i), where class-S channels sit outside B.

• B includes functionals of (p, q, h): prices, quantities, and household characteristics
• S includes non-priced, non-excludable channels co-generated with bilateral transaction volume
• When Φ(S) is negative and scales with volume, measured welfare can rise while true welfare falls

Deaton’s welfare tools share one bilateral measurement architecture

The paper maps more than 20 Deaton publications across roughly 50 years and argues that their instruments observe household-market transactions, individual reports, or household characteristics, then compute welfare inside that frame.

• AIDS maps prices and expenditures to budget shares and cost-of-living welfare
• Household-survey methods turn expenditures, assets, and reports into poverty and inequality measures
• Subjective well-being scales map individual reports into welfare scores
• Deaths of Despair names system channels that the consumption frame did not register

From 1999 to 2019, consumption rose while mortality rose for non-college whites

The paper’s empirical fingerprint is the Case-Deaton mortality reversal. US consumption per capita rose over the same period in which white non-Hispanic Americans without a four-year college degree accumulated about 600,000 excess deaths.

• Age-adjusted midlife mortality for non-college whites rose from about 381 to 415 per 100,000
• Reported poverty held steady and evaluative well-being trended upward in the paper’s account
• The two data streams point in opposite welfare directions because they observe different objects

The missing channels are opioid capture, rent extraction, labor-market exit, and community collapse

Class-S channels are not omitted survey variables. In the paper, they are system-welfare losses that do not generate a clean price-quantity-household observation, even when they are caused by activity that does enter B.

• Non-priced: institutional erosion and community-infrastructure decay do not appear as household budget items
• Non-excludable: community collapse affects residents whether or not they made the relevant transactions
• Co-generated: opioid prescriptions and pharmaceutical revenues enter B while institutional damage enters S

The theorem needs Bilateral Substrate, Class-S Existence, and Volume Complementarity

The result is built from three assumptions. Axiom 1 puts welfare instruments inside B. Axiom 2 establishes at least one class-S channel. Axiom 3 links the class-S welfare term to bilateral transaction volume.

• Bilateral Substrate: each instrument is a functional of (p, q, h)
• Class-S Existence: at least one welfare channel is non-priced, non-excludable, and co-generated with volume
• Volume Complementarity: Φ(S) is monotone in bilateral transaction volume
• The paper states that each pair of axioms gives only a proper subset of the full result

If class S has zero mass, the standard instruments can be welfare-complete

The paper does not claim that AIDS, poverty measures, SWB scales, MPI, or DINA are bad instruments. It says they are strong inside B and complete only when system-welfare channels carry no residual mass.

• When Φ(S)=0, W* = W(B)
• AIDS can recover consumer welfare through the expenditure function inside the bilateral frame
• Poverty lines, SWB scales, and MPI can classify welfare states inside their observed domains
• The theorem concerns divergence when Φ(S) becomes negative and remains outside B

For any welfare function on B, there are paths where W(B) rises and W* falls without bound

Under the three axioms, the paper proves that measured welfare can increase monotonically while true welfare decreases monotonically. The divergence comes from Φ(S), which is outside the domain of any W defined only on B.

• Axiom 1 places W(B) on the bilateral substrate
• Axiom 2 puts Φ(S) outside that substrate
• Axiom 3 lets Φ(S) scale with transaction volume
• The divergence lemma states that for any ε>0, sufficiently high bilateral volume makes W(B)-W* exceed ε

AIDS, LSMS-style surveys, SWB, MPI, and DINA inherit the same blind spot

The nesting result recovers major welfare instruments as projections of B. Adding more bilateral dimensions improves measurement within B but does not make institutional integrity, community infrastructure, or political accountability observable as class-S channels.

• AIDS sees drug prices and quantities, not physician-network integrity
• Household surveys see expenditures and assets, not community-level institutional collapse
• SWB sees individual reports, not whether meaning-generating institutions are eroding
• DINA sees distribution of national income, not the system that produced the distribution

Beyond-GDP, capability, SWB, deaths-of-despair, DINA, and MPI do not prove a frame limit

The paper positions MFIT against six literatures. Each broadens welfare measurement or documents a failure, but none proves that class-S channels are structurally outside any bilateral dashboard regardless of dimensionality.

• Stiglitz-Sen-Fitoussi dashboards add dimensions inside B
• Sen and Nussbaum move welfare beyond consumption but assess functionings at the individual level
• Case and Deaton document the mortality reversal without turning it into a general measurement theorem
• Alkire-Support MPI adds household dimensions: health, education, and living standards

The simulation checks the theorem against opioid mortality, education gradients, rents, and labor-force exit

The paper reports Monte Carlo calibration rather than treating the theorem as only abstract. The stated calibration uses 100,000 draws, three distribution families, and seed=42 within the paper’s assumptions and empirical scope.

• Benchmarks include opioid excess mortality and life-expectancy reversal by education
• Other benchmark channels include pharmaceutical rent extraction and labor-force-exit trajectories
• The calibration supports the divergence pattern: positive bilateral measures can coexist with negative system-welfare mass

Closing the gap requires direct measures of S, not more variables inside B

MFIT’s constructive answer is an external welfare ledger. The proposed instruments measure system-welfare channels directly instead of trying to infer them from household transactions, individual reports, or distributional surfaces.

• Institutional-integrity indexes measure institutional quality outside household budgets
• Domain-specific β_W estimates attach welfare weights to system channels
• Decision Accounting records price losses that bilateral welfare instruments cannot observe
• Consumption data remain useful, but W* requires W(B) plus separately measured ΣΦ (s_i)

MFIT is the measurement-side analog of the Missing System Theorem

The paper links MFIT to the SAPM research program. The Missing System Theorem says bilateral pricing cannot price system welfare; MFIT says bilateral measurement cannot see system welfare.

• Pricing floor: bilateral prices miss system welfare
• Measurement floor: bilateral instruments miss class-S channels
• Aggregate bound and governance resolution complete the stated foundational architecture
• Policy and managerial evaluation need institutional metrics alongside consumption and distribution data