Every method in the Impact Evaluation course, every indicator in the Development course, every fiscal number in the Public Finance area — all rest on data, and if the data are wrong, the cleanest method gives the wrong answer. This course is about that foundation: how policy numbers are actually made, and why they are often far shakier than they look — especially in Africa. We begin with why measurement matters and how bad it can be.
Garbage in, garbage out
Methods can't rescue bad data
The sophistication of a method cannot compensate for the quality of the data it's applied to — 'garbage in, garbage out'. A perfectly-designed RCT measuring a badly-mismeasured outcome gives a precise estimate of the wrong thing. A careful poverty analysis built on a flawed consumption survey produces an authoritative-looking but wrong poverty rate. An elegant inequality decomposition on data that misses the rich understates inequality no matter how good the math. Measurement is logically PRIOR to method: you must measure the right thing well before any analysis is meaningful. Yet measurement is chronically underappreciated and underfunded relative to method — economists train extensively in econometrics and barely at all in how the data were collected. This course corrects that imbalance, on the premise that the data are where most of the real uncertainty in policy numbers lives, and that understanding measurement is as important as understanding method.
The African data gap
Africa's statistical tragedy
African economic data are often sparse, infrequent, outdated, and of poor quality — a problem serious enough that Shanta Devarajan (then the World Bank's Africa chief economist) called it 'Africa's statistical tragedy' and Morten Jerven documented it in 'Poor Numbers' (2013). The symptoms: many countries go years or decades between censuses and household surveys; vital registration (births and deaths) is incomplete, so even population and mortality are uncertain; national accounts (GDP) rest on outdated base years and weak underlying data; and the large informal and subsistence sectors are poorly captured. The consequences are serious: policy is made partly blind (you can't target poverty you can't measure, or judge growth you can't track accurately), international comparisons and rankings are built on shaky numbers, and the data weakness compounds the state-capacity problems of the Governance course (a state that can't see its economy struggles to govern it). The data gap is itself a development problem — and improving statistical capacity is an underappreciated priority.
What the GDP rebasings revealed
Nothing illustrates the data gap more vividly than the GDP rebasing episodes. A country's GDP is estimated using a 'base year' of detailed economic structure; over time the economy changes (new sectors — mobile telecoms, services — grow) but if the base year isn't updated, the GDP estimate misses them. When countries finally 'rebase' (update the base year), GDP can JUMP overnight: Ghana's 2010 rebasing raised measured GDP by ~60% in a stroke; Nigeria's 2014 rebasing nearly DOUBLED its GDP (making it Africa's largest economy overnight), revealing a large services and entertainment economy (Nollywood, telecoms) the old base year had missed. Nothing real changed on rebasing day — the economy was always that size; the MEASUREMENT was wrong before. The rebasings revealed how far official figures can drift from reality when statistical systems are underfunded, and they are a sobering reminder that headline numbers like GDP — used in every debt ratio, growth comparison, and per-capita figure — can be off by tens of percent. If GDP itself is this uncertain, every ratio built on it (debt-to-GDP, tax-to-GDP — the Public Finance area) inherits the uncertainty.
Measurement as a public good
Why is measurement so underfunded if it's so important? Because good data is a PUBLIC GOOD (the Public Finance course): the benefits of a reliable census or survey are diffuse (everyone who uses the data benefits) and long-term, while the costs are concentrated and immediate, so it is systematically underprovided — the same political-economy logic that underfunds other public goods, applied to statistics. National statistical offices are often under-resourced, politically vulnerable (governments sometimes prefer not to measure inconvenient things, or to manipulate the numbers — the de jure/de facto and transparency themes of the Governance course), and low-prestige. The case for investing in statistical capacity — better surveys, censuses, vital registration, and (increasingly) the new data sources of module 7 — is strong precisely because measurement is the foundation everything rests on, and the returns to better data (better-targeted, better-evaluated policy across every sector) are large though diffuse. Recognising measurement as essential public infrastructure, not a technical afterthought, is the framing this course argues for — and the improving data picture (more surveys, satellite and mobile data) offers real hope of closing the gap.
Exercise
A country's official statistics show GDP growing 5% a year and a poverty rate of 35%, and a minister cites these as proof the country is developing well. The country last rebased its GDP 15 years ago and last ran a household consumption survey 9 years ago. (1) Explain why these headline numbers should be treated with caution. (2) Explain what a long-overdue GDP rebasing might reveal and why. (3) Explain how the 9-year-old survey undermines the poverty figure. (4) Explain why this measurement weakness is a development problem, not just a technical one.