V2-6.05| THE COUNTER
01| Everything so far has been about the machine. This chapter is about the encounter — one person, one office, one form — because that is where the state is actually experienced by anyone in the barrio, and because it is the one part of government performance that can be measured cleanly. The measurement is called non-take-up: the share of people who are entitled to something and do not receive it. What makes it so useful is that eligibility is not in dispute, so whatever the gap turns out to be, the programme built it.
02| The gaps are large and they are not confined to poor or badly run countries. In the United States, SNAP non-take-up ran at 16–17 percent in 2019, Medicaid at 35 percent for adults and 35 percent for children between 2014 and 2017, and TANF at 72 percent by 2012, up from 18 percent in 1995 after the 1996 reforms. In Europe the figures are worse: Germany averaged 56 percent between 2005 and 2014, France's RSA 34 percent in 2018, the Netherlands 35 percent in 2018, Belgium between 57 and 76 percent in 2005, and Sweden between 70 and 80 percent in the mid-1980s and again in the late 1990s. In middle-income countries they are worse still — China's rural Dibao leaves roughly 90 percent of eligible people unreached, and 59 percent of the Brazilian poor were not receiving Bolsa Família [Ko & Moffitt, Take-up of Social Benefits, IZA DP 15351 — page numbers for individual figures to be recovered].
03| Read that list again with one thing in mind: these are the successes. Every programme named there is funded, operating, staffed and considered a going concern, and several are internationally admired. Majority non-take-up is entirely compatible with a programme being described, honestly and without irony, as working. This is the most important thing to carry out of this chapter, because it means the difference between a government's stated coverage and its actual coverage is not an accusation — it is a routine property of functioning systems.
04| The causes are known and decompose into five, of which four belong to the administration itself. Benefits are sometimes too small to justify the cost of claiming them, and people decline when "benefits do not exceed costs" [Ko & Moffitt, p. 20]. Administrative barriers do a great deal of the work, with claimants reporting that "application would take too much time" and that "offices are too far away" [p. 23]. Information failure is the single most commonly reported reason of all, recorded simply as "lack of knowledge" [p. 23]. Stigma operates both externally and internally — "it would feel like begging" [p. 23] — and finally the offices themselves make mistakes, with US audits finding that "error rates in incorrectly denying eligibility ranged across the states from 0.3 percent to 4.7 percent" [p. 25].
05| Only stigma originates partly outside the administration, and even stigma is shaped by design decisions — whether a benefit arrives through a welfare counter or through the tax code changes how it feels to receive it. The other four are produced by choices about forms, offices, opening hours, publicity and staff training, all of which are inside somebody's control. That is what makes non-take-up a performance measure rather than a fact about poor people. It describes the programme, not its claimants.
06| The costs of claiming are also borne asymmetrically, in a way that is easy to miss from inside an office. Filling in a form is a minor irritation to a person with a printer, a fixed address, a bank account and a free afternoon. The same form is a day of lost wages, two bus fares, a photocopy shop, a childcare problem and a real chance of being turned away for a missing document to someone without those things. The burden is nominally identical and materially regressive, and it falls hardest exactly where the need is deepest — which means the programme filters most severely against the people it was designed for.
07| There is a second asymmetry inside the office, and it explains more official behaviour than any amount of attitude does. The two errors an official can make are not equal in consequence for the official: admitting someone ineligible surfaces in an audit with a name attached to it, while refusing someone eligible produces no record at all, because the person leaves and the file is never opened. Under those incentives caution is not a personal failing but exactly what the incentive structure has purchased. The exclusion error is the invisible one, and invisible errors grow.
08| The strongest exception is also the strongest positive finding in this whole part. Where entitlement is inferred from data the state already holds and the benefit is paid without an application, take-up approaches full coverage — automatic or default enrolment is the only intervention known to close the gap rather than narrow it. What it does is change the question from "did this person apply" to "does the state's data reach this person", which is a question about the state rather than about the claimant. That reformulation is the whole trick, and it reappears from the procedural side two chapters later.
09| Two weaker exceptions are worth recording. A universal benefit has almost no take-up problem because there is nothing to prove, at the cost of paying people who do not need it — the coverage-versus-depth trade arriving here from a different direction. Outreach experiments, in-person and digital, move take-up measurably but only partially, because they treat the symptom while leaving the burden exactly where it was. Both are real improvements and neither closes the gap.
10| The counter-case has to be stated: some non-take-up is chosen rather than imposed. People decline benefits they judge not worth the trouble, refuse them out of pride, or expect to be off the programme within a few months and see no point in starting. Treating every non-claimant as a victim of administrative failure overstates the case and insults the judgement of people who made a reasonable decision. The honest formulation is that most non-take-up is administrative and some of it is preference, and separating the two requires survey data most programmes have never collected.
11| What this leaves for anyone entering a barrio is a measurable and unflattering fact about the terrain. There is very likely a substantial population inside it that is entitled to money or services it is not receiving, the reasons are mostly documented and mostly fixable, and nobody currently has an incentive to count them — since the exclusion error appears in no audit and on no dashboard. That is simultaneously the clearest unmet need available and the clearest demonstration of what a system that actually reaches people would be worth. Whether closing that gap is something residents can do from below, or requires the state's own data and therefore the state's cooperation, is the question this chapter hands forward.
