Executive Summary
The major events of 2021 directly involved flawed forecasting by Government and related institutions. These include the Pandemic response, the energy crisis, the cost of living crisis and COP26.
These flawed forecasts have negative real-world effects; and ultimately undermine trust with voters.
There are structural reasons government bodies cannot reliably forecast for themselves. These relate to Groupthink, flawed incentives and explicit agendas. The forecasts are not intended to be correct, but to support a predetermined narrative.
Government needs to separate its forecasts from its agencies.
Forecasts from outside government applying market structures and market-based incentives for accuracy would deliver better predictions. In addition they would be cheaper, more adaptive and able to self-improve.
The government should aim to get all of its forecasts from external providers via market-based systems that incentivise and reward forecasting accuracy.
It has been another lamentable year for government forecasters. Pandemic predictions have been wildly off the mark again and again, and are now coming under increasing scrutiny. Central Banks told us inflation would be transitory, only to be proved wrong within months. Indeed, The Bank of England’s latest inflation forecasts were blown away within a fortnight. Even the call for Climate Change action is being undermined by forecasts that are already turning out to be exaggerated or entirely false. All of which is undermining public confidence in science and economics.
There is nothing new about terrible government predictions. It’s one of the many reasons centrally planned economies prove such disastrous failures. There are three ago-old reasons why governments cannot forecast:
1. Groupthink. Too often government forecasts are made by the same sort of people. For example, the Bank of England’s economists live and work in the same place, read the same things, studied at the same colleges, earn similar salaries and do much the same job as each other. As a result, they lack cognitive diversity. Instead you get an echo-chamber, with each contributor amplifying the other’s biases.
2. Flawed Incentives. Bad economists and bad epidemiologists don’t get fired or lose their bonuses. Indeed, the London School of Hygiene and Tropical Medicine has the worst forecasting record of the pandemic; and yet it is called upon to produce the forecasts that dictate pandemic policy!
3. Conflicting Agendas. In hierarchical bureaucracies, the most beneficial action for the predictor will rarely be to produce the most accurate forecasts. Most forecasts are there to justify an outcome that has already been decided on, or to tell the boss what they want to hear. We saw these flaws emerge in a fascinating exchange this week between Fraser Nelson and The Chairman of Sage, Profesor Medley. Professor Medley admitted Covid forecasts were ‘not predictions.’ Indeed, they were not even supposed to be accurate, or take account of probability. Instead, the ‘forecasts’ exist to force ‘decision makers to get decisions made.’ For many people this has proved an alarming discovery, but it is by no means unique. When central banks do not want to tighten monetary policy, they simply suppress their inflation forecasts to justify the agenda. When Chancellors want to spend more money and still look prudent, they just hike up their forecasts for GDP growth. And how can the UN produce credible climate forecasts while pushing for countries to take greater climate action?
The dangers of biased forecasts like this cannot be overstated. Central banks frequently lose credibility; and with it control of their currency and inflation. Turkey, Venezuela and Lebanon being the latest culprits. Similarly, by producing hyperbolic and inaccurate Covid forecasts, the government is playing straight into the hands of anti-vaxx groups and conspiracy theorists. Those with longer memories will recall that The New Labour Project was effectively killed off by the calamitous invasion of Iraq: an action justified with ‘sexed up’ documents masquerading as objective evidence. When you lose confidence like that, you can never win it back.
If the government cannot be trusted to produce honest and reliable forecasts, then who can?
The answer (as with so many things) is to get rid of the government bureaucrats, and let the markets have a go.
This is quite simple. In some cases we already have the data available. Predicted inflation rates can already be implied from the market price of index-linked bonds versus nominal bonds. Weather derivatives are already a commonplace financial instrument. But we could go further and predict anything with markets. Millions of punters are already familiar with the world of spread-betting, via contracts for difference. The more accurate your prediction is the more you win. The further away you are and the more you lose. Government could create crowd-sourced prediction markets for anything where it needs a forecast, from pandemic deaths to GDP growth to climate forecasts. Registered organisations (such as university faculties and financial analysts) could bring to bear their expertise for financial gain or loss. And at last we would see digitised government in action.
And there are many reasons to believe market-sourced forecasts would be more accurate than centrally produced forecasts.
For example it has been consistently found that betting markets are the most reliable way to predict election results. With a financial bet, the raw incentive is to only bet if you are sure you are right. No-one is gaming the system or expressing their ideological biases. As a perfect example of the power of financial incentives, look how at how the Omicron forecasting accuracy of investment bank, JP Morgan’s have trounced the expertise of SAGE and the London School of Hygiene and Tropical medicine
Markets also benefit from ‘The Wisdom of Crowds.’ The Wisdom of Crowds was coined by Victorian mathematician Frances Galton. Galton noticed that average estimates sourced from a varied group of non-experts were invariably more accurate than the estimates of individual experts. Why? Because the group was more diverse; with each member bringing unique insights to the decisions but without amplifying each other’s biases.
Markets also benefit from Darwinian Meritocracy. As time goes on, successful forecasters will win more money, dominating a larger share of the market. Meanwhile, poor forecasters will quickly lose their money and exit. Thus, the market concentrates itself on the best players. Had Covid predictions been done this way, Professors Ferguson and Medley would have been out of the game a long time ago, and better methodologies would have emerged.
Crucially, markets adapts in real time to changing information. If a new blockbuster Covid treatment comes out, forecasts would instantly change, just as stock prices do. If oil prices go up, implied inflation rates instantly move. There is no need to wait days or weeks for committee meetings and lengthy reports. This is modern world forecasting: reacting instantly to today’s rapid information flows.
This is not to say market-based forecasts will always be correct. For example, the TMT bubble of 2000 and the 2008 sub-prime bust were in part caused by flawed financial models. Yet even in these cases it wasn’t just markets that erred: government fanned these bubbles with deregulation and moral hazard while public funds were ploughed in. Even when the market gets it wrong, the government rarely does better. What matters is that market predictions are significantly better most of the time AND the incentives are there to improve them when they are wrong.
Britain’s leaders have relied on bad forecasters for millennia. While chicken gizzards and bubbling cauldrons have been replaced by sophisticated models, today’s crop of experts are sadly no better. Market-based predictions would save money, produce more accurate predictions and provide the incentives to improve. 2022 should be the year we give the markets a go.
