The Business Case Against Business Cases

From railways to mobile phones, the numbers always miss the behaviours that matter and why relying on the quantifiable past alone is a terrible guide to the future.

A few days ago I got kinda riled up by a LinkedIn post from someone who has a lot of influence on Toronto, and indeed Canadian urbanism. Basically it was suggesting that the business case for building high speed rail connecting Toronto and Montreal makes no sense because they’ve got 40 years worth of passenger data between the two cities.

There’s a flaw in this thinking. It’s basically scientific looking bullshit, which we call ‘business casing’. Ostensibly it exists to justify doing something new, or at least ensure a deflection of blame if it’s wrong. Unfortunately, it is often the single most effective tool for ensuring nothing happens at all.

Business cases worship at the altar of the past, they are basically historical artefacts masquerading as prophecy. They pore over what happened yesterday and then assume tomorrow will look suspiciously similar. It is forecasting by rear view mirror.

But the whole point of innovation, whether it’s a new railway, a cycling network, telecommunication infrastructure, new energy grids etc. is precisely that tomorrow will not look the same. It is meant to change behaviour, not perpetuate it.

When London Outgrew the Spreadsheet

Let’s look at London’s Elizabeth Line. The official projections predicted steady uptake, when what Londoners delivered was a stampede. In just three years it has already clocked over 500 million journeys, and on some days 800,000 trips are taken. Forecasts had it coasting at perhaps 130 million trips a year by now, however it’s now comfortably approaching 200 million … and that’s without counting the fact that about a third of those journeys are new. These are trips people simply wouldn’t have made at all if the line didn’t exist.

This is a universal blind spot in that many models assume new infrastructure is primarily rearranging the existing demand, plus a little bit more. They act as if the only people who ride trains are those who already do so today, plus the odd existing car journey or flight. What they miss is that better options create new behaviour. “Meet a client in Toronto and still be home for dinner” doesn’t appear in a Canadian business case for high-speed rail, because the option doesn’t exist yet.

If I want to go on a short notice trip to Toronto, I’m looking at $1000 in airfare, or 12 hours of slow train travel. A hotel, should I not want to sleep at a friend’s place can be $500 a night. The business case doesn’t take this into account, that I will simply choose to not go sometimes unless I can stick it on expenses. Likewise, lots of other people simply don’t want to spend the amount of time and money that is required to go between the cities. But if a return train fare was say $250 and they could do it in a day, they probably would!

The Excel model can only imagine faster versions of today’s tedium. It cannot imagine day trips that don’t currently happen without some completely made up assumptions.

Bogotá’s Buses and Seville’s Bicycles

It’s the same in Bogotá. When the city built its bus rapid transit system, forecasts suggested perhaps 670,000 daily passengers. Within two years it was carrying 900,000. Today it serves 2.4 million, more than many European metro systems. The ‘unrealistic’ demand was simply invisible until the buses existed.

Or bikes Seville. In 2006, cycling accounted for half a percent of trips, basically a rounding error. Traditional analysis would have written it off, but when the city laid 80 km of segregated cycle lanes in one go, daily trips jumped from 6,000 to 70,000. Cycling now commands close to 10% of all journeys. The data showed was that nobody cycled, however the infrastructure revealed plenty of people wanted to, they just didn’t fancy doing it on the roads as they were.

Technology Is No Better, Just Ask McKinsey

If you think this afflicts only infrastructure, technology is not all that different either. In 1980 AT&T hired McKinsey to forecast the U.S. mobile phone market. The consultants predicted 900,000 users by the year 2000. The actual number was 109 million. AT&T, acting rationally on the best available data, pulled back on the emerging industry, and spent billions buying its way back in once reality intervened.

OPEC’s forecasts fell into a similar trap for electric vehicles, for eight years in a row they underestimated adoption. Their 2015 forecast for 2022 was wrong by 93%. If you had believed them, you’d have assumed charging networks were a frivolous waste (which is probably what they hoped too). Instead, governments who bet on unrealistic EV growth now look prescient.

There are also the casualties of over-caution. In the 1960s Britain, armed with meticulously gathered data showing declining rural ridership, took a hacksaw to its rail network, and over 2,300 stations closed under the Beeching Axe. Decades later, with road congestion endemic and rail booming again, the government is now spending hundreds of millions to reverse some of those closures. What was presented as rationalisation now looks like vandalism.

The Real Error & The Remedy

The underlying mistake is simple. We limit our inputs to forecasting activity to being ‘data driven’. The issue being all data comes from the past. So our forecasts lack foresight. Our forecasts are extrapolations of existing behaviour, when foresight requires imagination about how behaviour changes once people have a new option. As the urbanist Lewis Mumford put it in 1955: “Adding lanes to address traffic is like loosening your belt to cure obesity.” The planners still haven’t caught on.

What we need instead is to ask what would have to be true for behaviour to change? What previously impossible things would having a new railway unlock? Further, what features of such a railway would influence that behaviour? That might be the timetable, the pricing, the location of the stations, the colour of the trains etc.

Scenario planning multiple possible realities, rather than single-point prediction is far more helpful … going beyond the obsession that economists have of things being ceteris paribus. Especially in the context of the likes of transportation, urbanism and infrastructure, demand is elastic, not necessarily linear and that people are not robots. Above all, recognising that the purpose of infrastructure is not to satisfy existing habits but to unlock new, previously impossible ones!

The safest business case in the world will always be to do nothing. But if you follow that logic, you eventually end up with nothing worth defending. The real business case isn’t in the spreadsheet, but the behaviours that don’t yet exist. And if your models can’t imagine them, that isn’t a problem with the project, it’s a problem with your imagination!

(disclaimer: this isn’t to say don’t do some thinking, about whether there might be some demand, just don’t rely only on what can be quantified today because it looks more rigourous!)

To being Challengers!