Clive Granger (1934–2009) was a down-to-earth Welsh econometrician who won the
2003 Nobel Prize in Economics — and who cheerfully insisted he was no great mathematician, just
someone with a good nose for spotting the right problem. That modesty undersells him wildly: his
ideas,
Granger's first big idea sidestepped a bottomless philosophical swamp. Rather than ask whether X truly causes Y — a question philosophers have chewed on for centuries — he asked something a statistician can actually test: does knowing the past of X help you predict Y better than Y's own past alone? If yes, we say X "Granger-causes" Y. It isn't causation in the deep metaphysical sense, and Granger was scrupulously honest about that, but it is precise, testable, and astonishingly useful. The name stuck to it forever.
The story of cointegration is a lovely accident. A colleague — the statistician David Hendry — argued that you could combine two individually wandering, non-stationary series and get a stable combination out. Granger was sure that was nonsense and set out to prove him wrong, expecting the maths to collapse. Instead the maths politely informed him that Hendry was right. Trying to demolish the idea, Granger ended up formalising it — and cointegration became the very work the Nobel committee cited. Sometimes the best way to find a truth is to attack it and lose.
Cointegration is famous for its favourite bar-room analogy: picture a tipsy person weaving home with their dog. Each one wanders unpredictably — neither path is going anywhere in particular. But they never stray far from each other, because every so often one calls or tugs and the gap closes. Two random walks, tethered by an invisible leash. That is cointegration: variables that each drift without bound yet stay bound together, and Granger's insight was to show how to find that hidden leash in real economic data.
The fuller story is on Wikipedia: Clive Granger — Wikipedia.