Measuring Human Capital When the Baseline Is Missing: Designing MEAL systems for systems change, not just output counts

Open almost any programme report in health, education, or social protection and you will find the same shape of evidence. People trained. Clinics built. Teachers deployed. Workshops held, materials distributed, beneficiaries reached. The numbers are large, the trend lines point up, and the report reads as success.

Then ask the question the report is quietly avoiding. Did anyone get healthier, learn more, or escape poverty because of it? And the evidence goes quiet, because the report was never designed to answer that. It counted what was easy to count and called it results.

This is not laziness, or not usually. It is a rational response to a genuinely hard measurement problem, and it deserves to be understood before it is criticised. But it produces a monitoring practice that can run for years, report faithfully the whole time, and never establish whether the money changed a single life. In human capital work, where the entire point is a change in people that is slow, diffuse, and hard to attribute, that is a serious failure disguised as diligence.

Why everyone counts outputs

Outputs get counted because outcomes are hard, and it is worth being specific about why.

Outputs are immediate and controllable. A programme decides to train five hundred health workers and, budget permitting, it does. The number is clean, it arrives on time, and the programme fully controls it. That is exactly what makes it comfortable and exactly what makes it a weak measure of anything that matters.

Outcomes are the opposite on every dimension. They are slow: better health or learning shows up over years, well after the reporting cycle closes. They are diffuse: a child’s education is shaped by the school, the household, nutrition, income, and a dozen forces no single programme controls. And they are hard to attribute: even where the outcome improves, proving your intervention caused it, rather than the harvest, the peace, or the parallel programme next door, is a real methodological challenge.

Faced with something immediate and controllable and something slow and diffuse, most systems measure the first and hope it stands in for the second. In stable, data-rich settings that hope is sometimes reasonable. In fragile ones, with weak baselines and volatile conditions, the gap between output and outcome can be total. You can train every worker you promised and improve nothing, and a pure output count will never tell you.

MEAL built for systems change

The response is not to demand impossible precision. It is to build the monitoring, evaluation, accountability, and learning system around the kind of change human capital work actually produces, which is systems change: a shift in how a health system, a school system, or a social system performs over time. That calls for a different toolkit than output counting, and it starts from different assumptions.

It begins with a serious theory of change. Not the decorative diagram in the annex, but a real, testable account of how this activity is supposed to lead to that outcome, with the assumptions between each link made explicit. The value of a theory of change is that it can be wrong, and finding out where it is wrong is the most useful thing an evaluation can do. A theory nobody could disprove is decoration.

It measures contribution rather than chasing attribution. In an open system you will rarely prove your programme alone caused an outcome, and pretending otherwise invites false credit and false blame in equal measure. The honest and more useful question is whether the intervention plausibly contributed, alongside everything else, and what the evidence for that contribution actually is. Contribution analysis and theory-based evaluation exist precisely for the messy causal environments where clean experimental attribution is unavailable, which describes most human capital work in the region.

It treats the missing baseline as a design problem to solve, not an excuse. Where no baseline exists, and often none does, you reconstruct one: recall methods, comparison groups, phased rollouts that let earlier cohorts serve as a reference for later ones, proxy measures for outcomes that resist direct measurement. None is perfect. All beat the alternative, which is measuring change from an origin you never established and quietly assuming it was zero.

And it is built for adaptation, not just judgment. The point of measuring is to steer while the programme is still running, not only to grade it once it is over. A MEAL system that produces a verdict at the end and nothing usable in between has failed at its most valuable job. The good ones feed a loop: measure, learn, adjust, measure again. That loop is worth more than any single evaluation, because it improves the thing while it can still be improved.

Telling the truth about uncertainty, to people who want certainty

There is a communication problem sitting underneath all of this, and it is at least as hard as the measurement.

Stakeholders want clean results. A funder wants to know the programme worked. A minister wants a number for the podium. The honest evidence from systems-change work is rarely that clean. It is a contribution, not a proof. A plausible causal story, not a settled one. A range with caveats, not a headline with a decimal point.

Communicating that honestly, without either overclaiming or drowning a decision-maker in methodological hedging, is a genuine skill, and it is where measurement meets strategic communication. Overclaim, and you build the next round of decisions on a result that will not survive scrutiny, and you spend down your credibility doing it. Bury the finding in qualifications, and you have told a busy decision-maker nothing they can use. The craft is to convey what is known, how well it is known, and what it means for the choice in front of them, in language that respects both the evidence and the reader.

That is not a footnote to the measurement work. It is part of it. Evidence that cannot be communicated in a way stakeholders can act on has not finished doing its job.

Measurement as a learning discipline

The instinct to reduce human capital programmes to output counts comes from a real difficulty. But it produces monitoring that reassures without informing, and in fragile settings that is close to worthless.

The better path treats measurement as a discipline for learning rather than an exercise in accounting. It asks harder questions, tolerates the honest uncertainty in the answers, and turns what it finds into better decisions while the programme is still live. That is more demanding than counting workshops. It is also the only version of measurement that tells you whether any of the money actually changed the lives it was meant to change, which was the entire point of spending it.