When I explain the difference between correlation and causation I like to tell the story of the South Pacific cargo cults that sprung up at the end of WWII.
A Cargo Cult
American forces had been active in beating back Japanese forces across the South Pacific and had hopped, island by island, towards Japan as the war was coming to an end. Supplying the soldiers, sailors, and airmen in the Pacific proved an exercise in logistics but one that the United States embraced and excelled at. The Americans set up supply depots in friendly countries, arranged regular R&R for combat troops in allied nations, and built first dozens and eventually hundreds of air strips in some of the remotest parts of the world.
Some of these air strips were located on barely habited islands. The local people watched as the Americans cleared and then flattened the land, installed lights and rudimentary air traffic control systems, and then welcomed planeload after planeload of cargo to the previously isolated islands. Millions of tons of cargo were airlifted in this way and the supply chains that the Americans built across the South Pacific played a significant role in the eventual victory in the region.
However, after the war the airfields were of little value. An island that might have been of prime strategic importance in supplying soldiers in 1945 returned, in the post-war era, to its status as an isolated, barely inhabited outpost in the world’s largest ocean.
But while the Americans departed, the idea of the airfield and the cargo that arriving US planes would deliver stuck. Indeed, so-called ‘cargo cults’ emerged that would construct elaborate airfields complete with imitation lights, huts where a local with a radio made of wood and coconuts would ‘control’ the arrival of planes, and a team of ground workers ready to service each and every cargo-packed plane that arrived.
But of course, not a plane did.
The error of the islanders, of course, is that it was not the presence of the airfield and the associated huts, ground crew, radio and lights that caused the airplanes to arrive packed with food and supplies. While the airfield was correlated with the arrival of the plane – indeed, it was a truly necessary condition for the arrival of a cargo plane on the island – it was not the cause of the plane’s arrival.
The cargo cults of the South pacific, however, were not the only ones to fall for attraction of correlation.
A Housing Crisis
Consider, for example, the housing crisis of 2008. While the reasons for the crisis, the collapse of major financial institutions, and the global economic slowdown that followed are many and varied, a major part has to do with an assumption that certain things would not change.
In particular, it was assumed that the housing market would weather any storm because, in the past, the housing market had weathered almost every storm. Sure, there was the odd global depression that saw the housing market (and most every other market) collapse temporarily, but there was a reason that a good investment was colloquially deemed “safe as houses” and not “safe as stocks”.
In the past an investment in property was a smart move. You borrowed money, bought a house, paid off the house, and sold it for a profit. It was a long term strategy for wealth acquisition, with the bonus that you could live in the house throughout the time it grew in value. In other words, a homeowner could extract value from owning the house and then additional value from selling the house – the perfect investment.
Because the housing market had traditionally been stable and because buying a house – even multiple houses – was an established path to future wealth, more people bought more houses. There was a strong correlation between homeownership and future financial stability, even future financial success. Hence, if you wanted stability and success, you should buy a house.
And, of course, this turned out to be a misplaced assumption.
Problems with Betting on Correlations
The correlation between people who owned homes in 2007 and financial stability or success in 2009 is vastly different to the same correlation in 1997 and 1999 or 1987 and 1989. Just as with the South Pacific cargo cults, something had changed.
World War II, the root cause of the cargo planes arriving on the island, had been concluded and so there were no more flights.
The home loan market and the financial services industry had changed so as to little resemble the stable markets of the past.
But anyone relying on the accumulation of data and the extraction of correlations from that data would have missed it.
The majority of people that were surprised by the 2008 housing collapse are little different to the islanders who expected that the correlations that had held to that point in time would continue to hold in the future. They bet big on the idea that the broader financial system was capable of producing wealth in the same manner as it always had.
But what should we base our decisions on if not experience, track record, and history? What should we seek in the masses of data that inform our choices in everything from what meal to enjoy at a restaurant through to what country should host a new production and distribution complex?
The answer: we should look for causes, not correlations.
The Smarter Bet? Causes.
In the case of the South Pacific cargo cults it is easy to identify how a misidentification of a correlation (airfields to cargo planes) ignored the cause (a war enabled planes to visit the island) that was more important in determining whether a plane would arrive on an airstrip.
In the case of the 2008 housing crisis, the correlation (investing in houses leads to wealth) also ignored the cause (a stable, regulated housing market enabled wealth creation) that was also more important in determining whether an investment would succeed.
In addition, in spite of the fact that both the airfield and cargo planes, and the housing investor and housing lender, too, were part of a broader system (one geopolitical, one financial), ignorance of that system and of the attributes of that system meant that correlations – even when strong – would eventually fail.
This is the reason that searching for correlations and basing major decisions on those correlations that emerge is a sure path to failure. Whether you are left wondering why the cargo plane has not arrived or why your housing investment didn’t work out the way your parents or grandparents investments did, the end result is the same.
Identifying causes, on the other hand, means understanding the real reasons behind why things happen, and betting on causes is the path to real success.