Measuring development

Last week: debated on what development is

• Modernisation — stages, catch-up possible → measure the level and its change

• Marxism — accumulation, exploitation → measure surplus distribution

• Dependency — core and periphery → measure the position

There is no “right indicator”

• Where theory enters the measurement

• What each indicator family shows — and hides

• How to choose the one for your task – and defend it

• How to get the data yourself

“First observe, then interpret” does not work

• The tidy view on empirics: theory enters only after the data

• The promise: disagreement is about interpretation

• The universal remedy: go back to the data

All observation is theory-laden

• Claim original due to Norwood Russell Hanson (1958)

• To say “Country A is more developed than B” you need a number

• To get a number you need an indicator

• To choose an indicator you must already have decided what development is

This is not the claim that data are bad

Theory enters at four points

The HDI makes all four choices — and many users hide three

• Conception: capabilities — but only three dimensions

• The three: a long and healthy life · knowledge · a decent standard of living — proxied by life expectancy, schooling and GNI per capita

• Why three? Data exist for nearly every country — unlike freedom or environment

• Aggregation: geometric mean since 2010 — income cannot buy off dying young

Same twelve countries, four conceptions

• Does the choice actually change anything? This can be checked!

The ranking depends on the conception

Output and sustainability point in opposite directions

• Output ≈ health — but Costa Rica outlives the US on a third of the income

• Output and equity are essentially unrelated

• Output vs. sustainability strongly negative — Ethiopia last on income, lowest emissions; Qatar the reverse

The numbers are theory-laden too

• India’s World Bank Gini is about 25 — lower than Norway’s

• India: consumption survey. Norway: income data

• The rich save \(\rightarrow\) consumption always looks more equal

Five families, five different questions

GDP and GNI come apart exactly where it matters

• Ireland, Luxembourg — profits booked in, then repatriated (Ireland publishes a corrected GNI*)

• Qatar — income flows out to migrant workers

• Philippines — remittances flow in; GNI > GDP

An average cannot see a distribution

GDP per capita is a plutocratic average

• Plutocratic mean: total income / heads — the usual GDP per capita

• Democratic mean: one person, one vote — tracks the typical household

• Germany 1991–2021: +0.7 % p.a. plutocratic, +0.25 % democratic, -0.2 % for the bottom quarter

The data are free, and you can get them today

Comparing development levels yourself

How to actually do it

library(WDI)
WDI(country = c("BR","VN","ET"),
    indicator = c(
      GDPpc = "NY.GDP.PCAP.PP.CD"),
    start = 2020, end = 2023)

What to take away

• No neutral measurement — concept, conception, indicator, aggregation

• The choice is a claim — put it in the argument accordingly

• Conceptions rank differently — output vs. sustainability almost inversely

• Check robustness — if it flips, that is a finding

Before next week

• Write up open questions you have had with the exercise and post them into Moodle

• Write one page — countries, indicators, rankings, and which choice mattered

• Catch up with the mandatory reading — the indicator families on the Measuring Development track (Gräbner-Radkowitsch 2026c; Gräbner-Radkowitsch 2026d; Gräbner-Radkowitsch 2026b)

• Read Amartya Sen’s Development as Capability Expansion (Sen 1990)