Bets on the future
The future bit is the hardest
Are you any good at predicting the future?
Probably, you’re not. Most people aren’t.
The most celebrated study was by Philip Tetlock, who asked 284 experts to make a total of 28,000 predictions - all in areas of their own expertise. Tetlock then did something very unusual. He waited for time to pass and then checked the accuracy of their forecasts once sufficient time had passed for it to be possible to know whether they were right. He found that, once they were being asked to forecast a range of about 3-5 years forward, their forecasts were generally no more accurate than a chimp throwing a dart.
(As he points out in his book Superforecasting, he actually found that “the average expert did about as well as random guessing. Except that’s not the punch line because ‘random guessing’ isn’t funny. The punch line is about a dart-throwing chimpanzee. Because chimpanzees are funny.”)
Now, that’s a bit of an issue, given that most investment decisions are predicated on at least a 3-5 year time horizon. If even the most expert of experts can’t predict that far out, how on earth can investment decisions be made?
Some people do this for a living
Like many startups, Snap has been primarily funded by Venture Capital funds.
Venture Capital funds (VCs) draw money from wealthy individuals and financial institutions and invest the cash into early-stage startups.
What is interesting is the degree to which VCs embrace uncertainty.
They are placing bets on the future, and they know it.
There are potential lessons for the corporate world.
Most of your guesses will be wrong
Every year a typical VC firm will receive many thousands of business plans to review. Of these, several hundred founders will be invited to pitch. Of these, around 40 will receive investment. Of these, typically, 20 will fail entirely. Of the remainder, one will generate returns that exceed all the others combined.
Obviously, it would be much better only to invest in that one and skip the other 39.
But VCs don’t do that, and the reason is that it’s absolutely impossible to predict which one is the one.
The ratios described above are true not just for a London-based VC like Kindred Capital (the first VC to place a bet on Snap) but the most successful and valuable VCs in the world. Even firms like Benchmark in Silicon Valley (who backed eBay, Uber, Instagram, Snapchat and Twitter) back more firms that fail than firms that go on to win big.
Investment is a prediction game. If the best of the very best can’t predict the future accurately, why do so many corporates persist in trying?
Business Cases Don’t Work
The reality is that some projects will crash and burn, and others will make it big. And no-one can possibly predict which.
So instead of focusing on a detailed financial forecast (which is actually just one scenario, but consumes most of the time spent preparing an internal pitch - time that could be spent not on honing the spreadsheet but on honing the actual proposal), maybe corporates should instead focus more on potential.
Let me give you an example.
When TfL first came up with the idea of contactless payments back in the 2000s, the business case was predicated on time savings through the gates. As predicted peak demand increased, the pace of throughput would require central London stations to expand to take wider gatelines. Real estate and construction in London is expensive, so it was worth a tech investment to avoid this cost.
What happened in reality was that contactless throughput turned out to be slower than Oyster, not faster. But then Covid came along and - in all probability - permanently suppressed high peak demand.
So the problem that Contactless was created to solve no longer exists and, even if it did, the solution doesn’t work to solve it.
So Contactless failed as a project?
Hell no, it’s been a triumph. It came into its own during the pandemic for obvious reasons, but it was wildly popular before that. For a while, TfL was the largest single contactless merchant in Europe. No-one will be able to prove it, but I suspect it probably delayed the decline in bus ridership by two years.
Contactless was a project with a huge range of potential benefits and upsides. Some of them came true, some of them didn’t. Some were included in the final business case; some weren’t. But it was a decent bet on the future.
ITSO, the Government’s smartcard standard, was another decent bet on the future. Back in the 1990s, smartcards were the future and setting a national standard was a sensible approach. Unfortunately, ITSO was a bad bet. The standard was too inflexible, the security requirements too onerous and the capabilities overtaken by alternative technologies.
When it becomes clear that a VC has made a bad bet, they stop investing. Immediately. And they don’t feel embarrassed doing so, nor do they feel a failure, because they knew that a proportion of their bets would be wrong.
But tied to a business case that is interpreted as making promises not predictions, a corporate will keep investing long past the moment when they should have drawn stumps. DfT has spent over £1 billion on ITSO, and most of it was after the moment when a VC would have given up.
The business case process
I’m sure lots of corporates do it differently, but everywhere I’ve worked, the capital investment process looks much the same:
1) A capital budget is allocated
2) Business cases are invited from managers
3) Each business case includes a financial forecast which predicts a return on investment over multiple years.
4) Those with the highest return on investment are allocated cash
Corporate business cases therefore become an exercise in trying to engineer a high ROI based on a single forecast of the future. Often the version of the future is constrained by the corporate assumptions as to what the future will look like. If you’ve got a great project for a scenario that is perfectly plausible but different to the corporate central case scenario, you’re not going to get funded. If you can find a way to create an excel spreadsheet that creates high ROI based on the template assumptions, you’ll win the jackpot for your project.
The forecasts you’ve made will now be baked into a budget and you’ll be held accountable to deliver precisely what you forecast. Managers that sponsor projects which fail will be lambasted while those that luck out will be heroes. Those lacking the self-confidence to make a five-year prediction with absolute certainty will stay quiet all the way through, and their ideas will never be heard.
Now, a VC doesn’t look at things like that.
What a VC will ask is: what is the maximum possible upside of this project? If everything goes best in all possible worlds, how good could this be? And they will look at the transformative potential of a project. They will review the numbers but they will exercise judgement.
Once the’ve made their bets, they won’t then hold each individual investment responsible for delivering precisely what they forecast in their central case. (Though they will support, encourage and motivate them to achieve they most they can). Internally, they will not make budget returns against their individual investments until it becomes clear how they perform in the real world. Instead, they will invest their fund and seek an aggregate return. If they undercut their aggregate forecasts, it means their judgement is wrong - and they’ll focus on how they can improve their judgement so as to make better bets in future.
I remember asking Leila Zegna (a partner at Kindred Capital and the best VC I had the privilege of working with) why they backed FiveAI (an autonomous car startup seeking to create a fleet of robotaxis to compete with Uber on the streets of London) given the extraordinary amounts of capital that FiveAI would need to burn before it became apparent whether or not they were going to make it. Her answer was that the upside was massive so it was worth a bet and even if their precise vision didn’t come off, they would be creating something that was valuable somehow.
Well, the vision didn’t come off. FiveAI have not and are not going to be running robotaxis around London. But, equally, they have built something of value: as they are using their technology as an assurance product for other autonomous vehicle products. Who knows if FiveAI is going to be The One for Kindred Capital, but that’s fine. FiveAI are to Kindred Capital what Contactless turned out to be for TfL: a thoroughly good thing, just not for the reasons they thought it would be.
So if a corporate finance department wanted to adopt the approach of accepting that the future is unpredictable, here’s what they might do:
1) Allocate a capital budget. That bit doesn’t change.
2) Budget a return that will be generated, in aggregate.
3) Invite pitches from managers. Ask managers to present their best case possible scenario and the ways in which their proposal will improve the business. They should obviously include numerical metrics (my new ticket machine is 15% cheaper than our current one) but do not need to be held accountable for things that are impossible to predict and largely outside their control (78 new ticket machines will be deployed to stations by 2024)
4) The finance team use their judgement to back bets. Some will be high impact, low probability. Others will be lower impact but more likely to come off
5) The managers are set to work on their projects. They should follow the agile principles I’ve described here
6) Finance should monitor closely, not because they’re holding managers to account for delivery against a notional forecast but because they’re watching the metrics to see which projects are generating good results so they can provide them with further support and capital
Finance may not like this approach, because it makes the finance team accountable for the quality of their decisions, as opposed to making managers accountable for the delivery of their business cases. But it is the choices of which projects to back that is crucial; so that feels like where accountability should sit.
In this world, no manager should feel compelled to plough on spending money on a project that, in their heart, they know isn’t going to deliver anything because the alternative would to admit failure.
Instead, as soon as a project looks like it’s not going to work, there should be no shame in saying “we tried this one and it was a bad bet.” If the company doesn’t achieve its forecast returns, that means finance need to improve their decision-making on which bets they back.
And it acknowledges the truth that none of us know what’s going to happen in the future.
It’s weird how hard humans find it to accept the future is uncertain. But, if a manager in a transport corporate submitting a business case really was so talented at forecasting that (unlike Benchmark capital or Tetlock’s 284 world experts) they could predict financial results five years hence - well, they wouldn’t be a manager in a transport corporate, would they?
