How Tim Harford left academia to build a career writing about economics

Executive overview

Tim Harford trained in game theory at Oxford but pivoted away from academia after his advisor told him he was doing better as a writer than he ever would as a theorist. A stint in Shell's scenario planning team, followed by writing The Undercover Economist in his spare time, launched a career built on applying economic thinking to everyday life.

Harford's model is diversification by design: an FT column, BBC radio, books, and podcast work spread risk and create creative cross-pollination. His core thesis on data consumption mirrors his career philosophy — slow down, prioritise context over volume, and ask what a number actually measures before reacting to it.

The right question is never "what does the number say?" but "what does this number actually mean, and am I the right person to be reading it daily?"

Leaving academia and finding a voice

• Harford's Oxford advisor Paul Klemperer told him directly: you're flourishing as a writer — why would you come back to be a mediocre academic?
• Between his master's and the FT, he worked at Shell's scenario planning team — interdisciplinary, long-horizon thinking that shaped how he frames problems
• Wrote The Undercover Economist with no agent, no publisher, no guarantee anyone would read it; carved out the time alongside a World Bank job
• Got an FT internship, moved onto the editorial board, spent five years writing leaders — a senior-sounding role that let a junior journalist fix mistakes before they went public
• Undercover Economist launched six months after Freakonomics — not planned, but well-timed to catch readers hungry for more accessible economics

The business model: diversification as risk mitigation

• Harford describes his setup not as a grand plan but as grabbing opportunities as they arrived and building gradually
• Key income streams: FT column (solo, fast), BBC Radio 4's More or Less (team of ~5, weekly), books (~every two years), public speaking, Pushkin podcast Cautionary Tales
• When speaking fees collapsed during COVID, BBC commissioned more episodes; the portfolio absorbed the shock
• Each stream has a different production rhythm — solo column vs. team radio vs. intensive scripting for Cautionary Tales — which keeps the work varied
• Radio series and books compound: he twice turned BBC radio series directly into books, doubling output from one research effort
• Works part-time on any single project; books are the most flexible — deadlines are "squishy" and can give in the short term without disappearing entirely
• Takes his agent's advice to do single-book deals rather than multi-book contracts; does not know the reasoning, just trusts it

What made the undercover economist style work

• Curiosity came first: he noticed odd pricing in coffee shops before he had any intention of writing about it
• The style is economic thinking applied to things you walk past every day — non-trivial questions that turn out to be tractable
• The Undercover Economist was, in his words, a naive book — he did not know how to sell it; the charm may have been part of the appeal
• It never hit the top 10 but sat in the top 30 non-fiction paperbacks for over a year — driven by word of mouth, not a single publicity spike
• The FT column and the book were the same voice, which reinforced both

Building a media presence over time

• More or Less began as a small niche programme and grew to a prime-time Radio 4 slot over nearly 15 years
• The prime-time shift happened to coincide exactly with the pandemic — the BBC immediately accelerated the timeline when COVID hit
• His OBE citation was for "services to improving economic understanding" — the through-line across all formats
• Pushkin's Cautionary Tales is structured differently: Harford writes every word of the script, does a table read, revises, then hands it to a production team who add actors, music, and mixing
• The Gladwell/Pushkin model lets him retain creative control at the front end while professional production adds polish at the back

On data, numeracy, and slow consumption

• Most numeracy mistakes are not maths failures — they are failures of context, definition, and comparison
• Example: a UK health secretary claimed losing 5 lbs each would save the NHS £100m over five years; per person per year that is 30 pence — the maths required is division, not statistics
• On vaccine efficacy: "95% effective" is a ratio of illness rates between vaccinated and placebo arms — not intuitive, but not technically complex once unpacked
• Background rates matter: 1,000 people over 80 die in the UK every day; once that group is vaccinated, deaths will occur near vaccination dates purely by coincidence
• Slower news consumption is more informative: weekly sources give context that daily rolling counts strip away; monthly is fine for most purposes
• Emotional self-monitoring is the underrated skill — noticing whether a claim makes you feel vindicated or angry is more useful than checking the source first
• The most important data habit: "Teach people to distrust what they think, not just what they read"

On finding what the best economists share

• The theorists Harford admired most — including Nobel winners Milgrom and Wilson — moved constantly between abstract theory and real-world problems
• His own thesis modelled sequential auctions under budget constraints, a theoretical problem that came directly from a practical question asked by a Shell colleague
• The best questions are non-trivial but tractable — the same quality that makes a good maths problem makes a good popular economics question

Practical habits and setup

• Lives and works in central Oxford; uses university libraries as a deliberate change of cognitive context — not for research alone but for the environment
• On location: changing physical space satisfies the need for variety more productively than checking email or phone
• Book writing schedule is intentionally unstructured; the flexibility lets him take short-notice opportunities without derailing the work
• Explains ideas out loud or to another person as a diagnostic — if the first follow-up question exposes a gap, that gap was always there