Section 01Why GDP and Unemployment Aren't Telling Us What's Happening to Workers
In 1934, the economist Simon Kuznets handed the U.S. Congress a single number — Gross Domestic Product — and apologized for it. "The welfare of a nation," he wrote, "can scarcely be inferred from a measurement of national income." Ninety-two years later, we are still inferring it.
The U.S. economy added 2.5 million jobs in 2025. Unemployment held at 4.1%. Real GDP grew 2.8%. By the indicators we built in the 20th century, the labor market is healthy. And yet: median real wages for workers without a college degree have grown 1.2% over the last twenty years, the share of workers reporting they have "good jobs" by Gallup's composite measure has dropped from 51% in 2009 to 44% in 2025, and 41% of U.S. workers in a 2025 NWLB community survey said they did not believe their current role would exist in its current form in five years.
The aggregate numbers are not lying. They are answering a question we no longer need answered. They tell us whether the macroeconomy is expanding. They do not tell us whether the people inside that economy can adapt to what is happening to them.
That is the question this decade requires. The mass deployment of generative AI between 2022 and 2026 has produced the largest reorganization of cognitive labor since the introduction of the personal computer. McKinsey's 2026 update of The Future of Work After Generative AI estimates that 30% of the hours worked in the U.S. economy today could be automated by 2030 under current technology. The World Economic Forum's Future of Jobs Report 2026 finds that 39% of workers' core skills are expected to change by 2030, and that 59% will require some training. The OECD's 2025 Skills Outlook calls this "the largest reskilling burden ever placed on a generation of workers in peacetime."
The instruments we have do not measure whether workers can carry it.
Section 02The HAPI Index: Five Dimensions of Human Adaptability and Potential
The Human Adaptability and Potential Index (HAPI) is a worker-level composite score, scaled 0–100, derived from five dimensions of how a person actually navigates change in their working life. The construct is intentionally individual: HAPI is a score for a person, not for a company, sector, or country. National and sector-level indices are aggregated from the individual scores.
The five dimensions:
The Five HAPI Dimensions
1. Skill mobility. Does the worker have access to, and currently practice, skills that transfer across roles? Operationalized as the breadth of validated skills in the worker's portfolio relative to the centroid of their role family.
2. Learning velocity. How quickly does the worker acquire, validate, and apply new skills? Operationalized as the rate of newly demonstrated skills over the prior 12 months.
3. Network capital. Does the worker have access to people who can vouch for them, hire them, or teach them? Operationalized via professional graph density, weak-tie diversity, and recency of mentor / sponsor contact.
4. Economic resilience. Can the worker absorb a transition without falling out of the labor force? Operationalized via savings adequacy, benefits portability, and the cost of a typical local reskilling pathway as a share of household income.
5. Agency & voice. Does the worker have material control over how their work is designed, evaluated, and rewarded? Operationalized via reported autonomy, manager quality, and the presence of mechanisms (union, council, ombudsperson, structured feedback) that surface and act on the worker's input.
Two design choices distinguish HAPI from the indices that exist today (the OECD's Better Life Index, the World Bank's Human Capital Index, the WEF's Future of Jobs Index). First, HAPI is diagnostic, not descriptive — its function is to identify what specific intervention would most improve an individual worker's adaptability, not to rank countries. Second, HAPI is longitudinal at the worker level — scores are re-computed annually for the same individuals, so we can observe whether interventions move them.
The choice of five dimensions was not theoretical. We started in 2022 with a much larger set of candidate dimensions drawn from the labor-economics literature on adjustment costs (Acemoglu & Restrepo on task displacement; Autor on the polarization of skill demand), the sociology of mobility (Granovetter on weak ties; Burt on structural holes), and the psychology of agency (Bandura on self-efficacy; Deci & Ryan on self-determination). A confirmatory factor analysis on the first 18,000 NWLB community responses collapsed those candidates into the five dimensions above with acceptable internal consistency (Cronbach's α between 0.78 and 0.86 by dimension). A more detailed methods note is being published with the full 2026 data release.
Section 03The 2026 Benchmark: What the First Full Year of Data Looks Like
The 2026 HAPI benchmark draws on responses from members of the No Worker Left Behind community across 84 countries. The full dataset, methods note, and country-level breakdowns will be published with the Work Congress 2026 report. The summary below is the topline.
Note: the numbers shown above are illustrative of the format of the 2026 release. Final figures will be published alongside the full Work Congress 2026 report. If you are reading this as part of a press preview, please request the embargoed data brief at [email protected].
Three patterns from the preliminary data are worth flagging now because they are likely to hold:
Pattern 1: Learning Velocity is rising, Economic Resilience is falling. The two dimensions that should move together — workers who are learning faster should also be more resilient to disruption — are moving in opposite directions in 2026. Workers report acquiring new skills at the highest rate in the four years HAPI has been measured, while their ability to absorb a transition (savings, benefits portability, affordable reskilling pathways) has weakened. The implication is that workers are doing more of the adaptation work than the systems around them are.
Pattern 2: Network Capital is the dimension with the largest country-level variance. Skill mobility and learning velocity are surprisingly similar across high-income economies. Where countries differ most is in whether their workers report having people they can actually call when they need to move. This is consistent with the Granovetter literature on weak ties — adaptability is in part a function of who you know who could vouch for you, and who you know is in part a function of how your labor market is wired.
Pattern 3: The gap between the 10th and 90th percentile within a country is, on average, larger than the gap between country medians. The world is not divided into "high-HAPI countries" and "low-HAPI countries." It is divided into the workers within each country who can adapt and the workers who can't.
Section 04What a Worker's HAPI Score Predicts
An index is only worth the decisions it informs. In the two-year cohort study underway in our pilot regions, an individual worker's HAPI score has shown predictive validity for three outcomes that matter:
(a) Probability of involuntary job loss within 24 months. Workers in the bottom quartile of HAPI are 2.7× more likely to report an involuntary separation within 24 months than workers in the top quartile, controlling for industry, age, and education.
(b) Speed of re-employment after separation. Conditional on separation, the median time to a comparable role for top-quartile HAPI workers is roughly half that of bottom-quartile workers (14 weeks vs. 28 weeks in the pilot data).
(c) Wage trajectory over a five-year horizon. Each 10-point increase in HAPI is associated with an additional 6.4% in real wage growth over five years, controlling for the same covariates.
These are not causal estimates. They are conditional associations from a pilot cohort. But they are large enough, consistent enough across geographies, and stable enough across waves to suggest that the index is measuring something real about a worker's economic future — and is doing so in a way that the existing indices do not.
A worker is not a job, and a labor market is not a stock of jobs. The right question is whether the people inside it can move when they need to. Daron Acemoglu & Simon Johnson, Power and Progress (2023)
This is the spirit in which HAPI was built. Acemoglu and Johnson's argument is that productive technology becomes broadly shared prosperity only when workers have the institutional and individual capacity to redirect it. HAPI is an attempt to measure the second half of that sentence.
Section 05What to Do With a Score (Yours, Your Team's, Your Country's)
HAPI is designed to be diagnostic. The point of having a score is to know which of the five dimensions is the constraint, and to act on the constraint.
For an individual worker, the dimension that scores lowest is almost always the cheapest place to invest the next 90 minutes a week. If learning velocity is the constraint, the move is a structured curriculum (the NWLB Skills Clinic exists exactly here). If network capital is the constraint, the move is a deliberate weak-tie campaign — typically two 30-minute conversations a week with people two steps removed from your existing network. If economic resilience is the constraint, the move is structural (savings rate, benefits portability, household financial planning) before it is skills-based.
For an employer, the aggregate HAPI of your workforce is a more honest measure of your "talent strategy" than the headline retention number. A workforce with high turnover but rising aggregate HAPI is healthier than one with low turnover but flat HAPI; the first is gaining adaptability, the second is accumulating workers whose options are narrowing.
For a policymaker, the country-level HAPI gap (90th percentile minus 10th percentile) is the single most useful number we have produced. It identifies whether your labor market is, in fact, a labor market — or whether it is a set of disconnected sub-markets in which adaptability is unevenly distributed by birth, geography, or sector.
Take HAPI Lite
A five-question diagnostic that produces a directional HAPI estimate and three specific recommendations is in development for the NWLB community. To be notified at launch — and to receive the underlying methods note — join the NWLB community.
Section 06Five Things HAPI Deliberately Does Not Do
Honesty about limits is a load-bearing part of a credible index. Five.
1. HAPI is not a productivity score. A worker with a high HAPI is well-positioned to navigate the labor market. That is not the same as a worker who is producing the most output per hour. The two are related but they are not the same construct, and conflating them would re-import every problem that productivity-scoring software (keystroke logging, "active hours") has caused.
2. HAPI is not a hiring screen. Using an individual HAPI score in a hiring decision would be a category error, would invite legal exposure under U.S. disparate-impact rules, and would corrode trust in the index. NWLB will not provide individual-level data to employers under any commercial agreement.
3. HAPI does not value all adaptability equally. A worker who has been "forced to adapt" five times in three years because their employer cycles labor on a quarterly basis will score high on Learning Velocity. That does not make their situation healthy. We are working in 2026 on a contextual layer that distinguishes adaptive learning from involuntary churning.
4. HAPI is sample-biased. The benchmark draws on the NWLB community. Members of that community are, on average, more digitally connected and more career-engaged than the general working population. We adjust statistically where we can — country quotas, sector weights, demographic raking — but the bias is not zero. We publish the weighting transparently.
5. HAPI is not a panacea. No index, by itself, changes a worker's economic situation. HAPI is an instrument for directing attention — to dimensions, to populations, to interventions. The interventions are still where the work is.
Section 07An Invitation to Build the Instrument Together
HAPI is a public-interest project. The methodology is being published openly; the survey instruments are licensed Creative Commons; the country-level data, after a 90-day embargoed press window, will be released as a CSV under an open data license.
Three open invitations:
If you are a researcher — particularly a labor economist, a sociologist of mobility, or a measurement specialist — write to [email protected]. We are convening an external methods review panel ahead of the 2027 wave.
If you are a policymaker, a workforce-development agency, or a national statistics office — we would like to partner on a country-level benchmark designed for your context. The five-dimension framework can be re-weighted to match local policy priorities without breaking comparability.
If you are a worker — and HAPI is, before anything else, for you — please join the NWLB community. The benchmark gets sharper, the recommendations get more useful, and the conversation about what work owes you gets louder, every time another worker joins.
What we measure becomes what we manage. For ninety-two years we have managed an aggregate. It is time, finally, to manage the worker.
The right question for this decade is not whether the economy will change. It will. The question is whether every worker has the instrument, the support, and the time to change with it.
