Revisions in the annual national accounts

In this report, we look at the revisions that arise in the annual national accounts, from the first publication of preliminary estimates to the final figures. We do this by comparing several national accounts aggregates, including GDP, as they were published in three preliminary versions against the final figures for the years 1972 to 2016. The purpose is to show whether preliminary national accounts figures have systematic biases relative to the final figures, as well as to quantify the uncertainty in the preliminary estimates.

For the entire period, the results show that all figures in the national accounts have tended to be estimated too low in the first two preliminary versions relative to the final figures. This is caused by the vintages from the 70’s to the mid 80’s. Since then, the preliminary estimates have on average been fairly close to the final figures, and what appears to remain of systematic biases can be argued to be due to single years with major revisions. Gross fixed capital formation stands out as an aggregate that shows a significant bias, and this bias has not diminished over time. It appears that this aggregate tends to be estimated about 2 percentage points too low in the preliminary figures right up to the publication of the final annual estimate. GDP Mainland Norway has typically been revised 0.3 to 0.5 percentage points.