The measurement of transport activity and fuel use over time is of critical importance to policy makers and analysts seeking to understand the sector. Since 1970, the Australian Bureau of Statistics (ABS) has conducted its Survey of Motor Vehicle Use (SMVU) every 3 to 5 years (and recently annually). The survey provides crucial estimates of activity and fuel use by all classes of motor vehicles, from motorcycles to heavy trucks. However, changes in survey procedures over the years have made long-term comparisons difficult to make using the 'official' figures coming out of each individual survey. To be useful for time-series analysis, the raw estimates from each survey have to be adjusted, for both logical and empirical reasons. The logical reasons for adjustment arise when survey classifications are changed, sample populations are varied, survey questions are changed, and so forth. Each of these changes introduces a possibly important data discontinuity into one or several data series. These 'logical' discontinuities have to be allowed for by an analyst, before he or she can make use of the data for econometric or even trend analysis. 'Empirical' adjustments are sometimes necessary to account for the sampling variability. All things being equal, a smaller sample size increases the sample variability. Recent reductions in the sample size of the SMVU have increased the range of variation in point estimates between SMVUs, such that the analyst's judgement is necessary to adjust the 'official' figures in order to derive consistent estimates. This paper summarises the results of recent BTE analysis of the SMVU, with particular emphasis on the 1998 and 1999 SMVU results. The paper provides a description of the methods used to adjust the SMVU results and presents consistent time-series estimates of total passenger vehicle travel and road freight transport between 1970 and 1999. (Author/publisher) For the covering entry of this conference, please see ITRD abstract no. E205861.
Abstract