Skip to content

Introduction of Month 2 Bias Adjustment for Business Investment

Summary

As announced in the last Business Investment bulletin, a bias adjustment will be introduced in the next Business Investment release.  This note provides an overview of the month two bias issue and the bias adjustment. 

Background

Month two business investment bias

Business Investment estimates are released provisionally in month two and then finalised in month three.

They are compiled from the quarterly Capital Expenditure Survey data.

At month two only provisional quarterly survey data are available; month two estimates are therefore produced using a combination of the latest returned quarterly survey data and estimation using historical data.

Increased responses to the quarterly survey in month three results enables the estimated data to be replaced by actual results.

Five year average response rates at month 2 are 70.7 per cent against  the month three average response of 83.2 per cent.

As part of the continual improvements to outputs, and in response to user inquiries, Gross Capital Formation branch performed a quality assurance check on the Business Investment release for the period 2007 Quarter 1 to 2010 Quarter 3.

The quality assurance indicated a negative bias in the provisional (month 2) outputs from the Capital Expenditure survey leading to continual positive revisions in the revised (month 3) data; and regression analysis on the time series involved showed a statistically significant bias in the estimates.

The Business Investment Summary Quality Report (128.9 Kb Pdf) provides further details on the survey response.

Bias adjustment

To reduce the effects of this bias, the Office for National Statistics has derived a multiplicative factor to uplift the provisional results thereby reducing the scale of the revisions at month 3.

Various techniques were tested by the Methodology Directorate (ONS) with the conclusion that applying a simple multiplicative adjustment factor to the month 2 data would be the most appropriate based on simplicity of use and the reduction in the average differences between provisional and revised results.

Due to movements in the survey response rates the multiplicative factor will be reviewed periodically and no later than annually to maintain suitability.

Implementation

As announced in the Business Investment Q4 2011 release dated 28 March 2011, the month 2 bias adjustment will be included for the first time, in the Business Investment Q1 2012 provisional results release on 24 May 2012.

Content from the Office for National Statistics.
© Crown Copyright applies unless otherwise stated.