The Treasury’s record of economic forecasting is so poor, it should have given up trying years ago

The Treasury’s record of economic forecasting is so poor, it should have given up trying years ago

Lies, damned lies and Whitehall models: this is the take-home message from a new report, Alternative Brexit Economic Analysis, issued today by Roger Bootle, Gerard Lyons, Julian Jessop and Patrick Minford.

It turns out that the mandarins have been up to their old Project Fear tricks again and this time they have been well and truly outed.

By way of a quick rehash: in the Brexit referendum campaign, the Treasury used a rigged version of its preferred gravity model in an all-out effort to scare the electorate into voting Remain. By “rigged”, I mean just that: the model ruled out any possible benefit from Brexit by assumption, and was set up so that it could not possibly give any forecast other than that Brexit would be bad. That model and its forecasts came in for withering criticism and its pessimistic Brexit forecasts turned out to be way off the mark. The subsequent performance of the UK economy has instead vindicated the forecasts of Patrick Minford, Tim Congdon and others who had forecast business as usual. The Treasury had destroyed its own credibility.

Fast forward two years and the good news is that Whitehall have got themselves a better model. The bad news is that the new model is proving to be of no benefit because they are abusing it in much the same way as they abused their earlier model. And they are doing their best to keep details secret, lest anything leak out that might trigger a fresh round of criticism.

In the past couple of weeks, a Whitehall-influenced cross-departmental Brexit impact assessment, EU Exit Analysis – Cross Whitehall Briefing, has been carefully ‘leaked’ to Buzzfeed. Its message is essentially Project Fear 2: Brexit will be bad for the economy, same old.

This has led to a flurry of commentary, with the BBC giving its usual uncritical assessment. The timing, not just of the leak and of the report, is significant as it coincided with the Brexit Cabinet sub-committee’s meetings to decide the position of the Government in the trade negotiations. It looks like the Treasury is meddling in politics again…

According to numerous leaks, these cross-departmental officials have redone the Treasury’s Brexit long-term forecasts with a superior new approach. This new model is a standard Computable General Equilibrium (CGE) model similar to the World Trade Model built by the Cardiff University macroeconomics research group. Because the Government Economic Service does not have the in-house capability to develop such sophisticated models, they sensibly have decided to use the Global Trade Analysis Project (GTAP) model, a workhorse built at Purdue University in Indiana by multiple universities, government and international agencies. So far, so good: Whitehall officials have dumped the Treasury’s useless gravity approach and adopted a full world causal CGE trade model, something they should have done years ago.

Despite the model upgrade, however, Whitehall are still making the same basic forecasting errors that were being made before – and they are egregious ones too: blatant bias, circular reasoning, silly assumptions and the hubris of false certainty. They start with an otherwise reasonable model, but input the most anti-Brexit assumptions they think that they can get away with. The model then produces results showing that Brexit would be a disaster, and the mandarins claim that Brexit will be a disaster because that is what their state-of-the-art “scientific” forecasting model tells them. Who do they think they are kidding? The model’s outputs simply reflect the inputs fed into it: garbage in, garbage out.

The mandarins’ forecasts are also undermined by obviously implausible assumptions. These assume scenarios that are not consistent with announced government policy and policy assumptions that no sensible government would implement. They absurdly assume that lowering enormous trade barriers against the rest of the world would have virtually no effect, while maintaining existing barrier-free trade with the EU under a Canada+ trade agreement would create huge costs in lost trade with the EU. For example, they claim that our EU trade would diminish by 5% of GDP, which is an astonishing claim to make considering that our current trade with the EU is only 12% of GDP. The Whitehall analysis also takes no account of the gains to the UK from improving regulation once the UK escapes the stifling confines of the Single Market regulatory regime or even of the gains from no longer paying our annual EU budget payment.

What appears to matter here is whether (a) the correct Brexit policy assumptions are fed into the model and (b) the model is ‘tuned’ to fit UK trade facts faithfully. If the right Brexit policies are fed in, it seems that all of the models – including the GTAP, Cardiff and gravity models – all produce directionally the same results – all clustered around a positive 2 per cent to 4 per cent of GDP range. The positive gains also get larger as the model gets closer to the facts of UK trade.

The mandarins then compound their errors by treating their forecasts as if they were precise and unchallengeable, e.g., by claiming that they could predict regional growth 15 years out to the first decimal point. Actually, you can use a model to forecast any output to any number of decimal points you want, but that doesn’t mean that the forecasts are any good. Forecast precision is one thing, but – as the Treasury’s earlier Brexit forecast fiasco showed – accuracy after the event is quite another. They have learned nothing from their own past mistakes.

Indeed, their judgments have been wrong on every major issue over the last 87 years. They were wrong on each of Brexit, the Global Financial Crisis, joining the Euro, the Exchange Rate Mechanism, the 1981 budget, early 1970s inflation, the Philips curve and the abandonment of the Gold Standard in 1931. That is some track record.

There is much the Treasury can to improve its forecasting performance. Ideally, they should exit the business on the grounds that almost a century of experience shows that they are no good at it. But at the very least, they might listen to the views of outsiders who are better at it.