Here, we're comparing the initial expectation with the actual outcome. The values shown according to start year (31 December) are “initial expectation” minus “actual outcome” (simple), with positive values implying an over-estimate. This matters hugely because just 1% pa is worth around 25% of a fully RPI-linked liability. We think that is a significant corporate financial distortion.

The expectations have been based upon periods of either 10 years or 15 years, the approaches being based upon “Bank of England spot”, “Bank of England forward” and “long gilts”. For each period, we've shown each approach on the same chart.

Over periods of 15 years until the end of 2019, all three measures ended up very slightly negative, all quite close to one another. On average, forward (0.26%) was far closer to an accurate prediction than either spot (0.99%) or gilts (1.36%). However, forward (1.01%) was only slightly less variable than either spot (1.18%) or gilts (1.47%), making it best of class but arguably still not good enough unadjusted. We still believe there is a bias.