Online Program

Investigating estimation methods under a reduced sample size – Application to the survey of capital expenditure
*Matthew David Greenaway, Office for National Statistics, UK 
Salah Merad, Office for National Statistics, UK 
Ria Sanderson, Office for National Statistics, UK 


Keywords: Cut-off sampling, skewed data, generalised calibration, model-based direct estimation, response burden

In an effort to reduce response burden and survey costs in the ONS quarterly survey of capital expenditure (Capex), we have investigated two strategies that make use of administrative data such as VAT turnover and expenditure. The first strategy involves using cut-off sampling and estimation via generalised calibration, as in Haziza et al (2010). The second strategy involves sampling from the whole population but with lower fractions than at present, and using model-based direct estimation, as in Chandra and Chambers (2011). The responses from the Capex survey are very lumpy – there are many zeros and some very large values, which presents a challenge for estimation. The performance of the methods considered will be evaluated through a simulation study using real survey data.