The problem with earnings forecasts

Craig Botham

Craig Botham

Senior Emerging Markets Economist

See all articles

We recently published our forecast returns for a range of asset classes over a seven-year horizon, which can be found here.

We made some major changes this year to the equity forecasting methodology, which is explained in the main document.

As part of the revision process, we took a look at analysts’ earnings-per-share (EPS) forecasts, in an attempt to include a wider range of information into at least the immediate forecast for equities.

However, we did not blindly take the latest consensus estimate. As economists, we are all too aware of the tendency for systematic biases to creep into forecasts: witness US GDP in recent years.

With this in mind, we looked at the historic over- or underestimate, by comparing 12 month ahead EPS forecasts with the actual realised EPS. The average bias is shown below for a range of markets.

Source: Bloomberg, Schroders Economics Group. July 2016. Used max history available, usually 10-15 years.

The bias for some markets is quite low; in the case of the US and Asia Pacific ex Japan, for example, there is almost no systematic bias.

Note that this is not the same as saying analysts get these markets right; only that their errors average out to roughly zero, so there is no bias to adjust for.

By contrast, analysts seem to be consistently optimistic on the earnings outlook for emerging markets, the UK and Europe ex UK. Something to bear in mind when receiving the next batch of equity analysts’ reports.