The 2005 TSI Annual Revision

The Transportation Services Index (TSI) is systematically revised once a year, in June, to incorporate changes in methodology, revisions to source data, and changes in estimated seasonal factors. The TSI revision policy was adopted in November 2004 to provide the most accurate and up-to-date measure while maintaining month-to-month stability by limiting each data point to one revision in the fifth month after its initial release. The annual revision allows the latest methodological research to be incorporated systematically and ensures that seasonally adjusted historical index numbers reflect recent revisions to estimated seasonal factors and the source data.

Revisions in source data and their impacts

Data revisions received during the calendar year for each TSI component change the individual index numbers as well as the overall index. First, these revisions change the index numbers of the individual months for which source data were revised. Second, the source data revisions change other index numbers through their effects on the estimation of the seasonal factors. Because the index is adjusted by the seasonal factors, all numbers in the index are affected.

Under the TSI revision policy, the most recent four months (the previous three plus the new release), will change one month at a time with every new release and will, at all times, be preliminary, while the fifth and earlier months’ values will remain as “revised” and will not be changed until the year-end comprehensive revision. Without this TSI revision policy, all index numbers would fluctuate each month as source data were updated.

Table 1 shows the impact of data source revisions before seasonality adjustments. Revisions in the source data generally have a greater impact on the TSI for recent years. The largest upward revision in the TSI due to revisions in source data was 0.6 percent for November 2002, while the largest downward revision was 1.0 percent for December 2004.

For annual average TSI, 2003 had the largest upward revision at 0.3 percent and 2004 had the largest downward revision at 0.2 percent.

For monthly freight TSI, the largest upward revision was 0.4 percent in August 2003 and the largest downward revision was 1.6 percent in December 2004.

For the annual freight average TSI, 2003 had the largest upward revision at 0.3 percent and 2004 had the largest downward revision at 0.4 percent.

For monthly passenger TSI, the largest upward revision was 1.7 percent in November 2002, while the largest downward revision was 1.6 percent in December 2001. For annual average passenger TSI, 2003 had the largest upward revision at 0.6 percent and 2001 had the largest downward revision at 0.3 percent.

Table 1. Percentage Changes in the TSI due to Data Revisions, 1990-2004

Excel | CSV

NOTE: Minimum means the largest downward revision or, in years where there was no downward revision, the smallest upward revision. Maximum means the largest upward revision, or, in years where there was no upward revision, the smallest downward revision.

SOURCE: Bureau of Transportation Statistics

Revisions to weights and their impacts

The weights for the TSI are used to combine the output indexes of the nine component transportation services into aggregated indexes, namely Total TSI, Freight TSI, and Passenger TSI. Weights give the relative importance of individual components in the combined total. The weights are calculated as the shares of the component transportation services in total transportation GDP. The primary data sources for the TSI weights are the estimates of GDP by industry produced by the Bureau of Economic Analysis (BEA). Weights are also estimated for splits of modal GDP among different services for aviation, rail, and pipelines (e.g., freight service v. passenger service, railroad carload service v. intermodal service, and petroleum products v. natural gas). Revisions in weights are due to updates and revisions to GDP data and modal source data. For given changes in the output indexes of the component transportation services, the TSI will tend to increase when the estimated weight of a faster growing service increases, and vice versa.

Revisions in the weights had the greatest affect on index numbers of the most recent year, 2004 (Table 2). The largest upward revision in the monthly TSI due to revisions in weights was 0.3 percent for October 2004 and the average revision for 2004 was an increase of 0.2 percent. The downward revisions to the TSI due to revisions to weights were all less than 0.05 percent and all in the months of the pre-2000 years. The largest upward revision in the Freight TSI was 0.2 percent for October 2004, and the largest downward revision in the Freight TSI was 0.06 percent in 1991. The largest upward revision in the Passenger TSI was 0.3 percent for April 2004, while the largest downward revision was 0.2 percent in September 2001. Revisions in weights had no affect on Passenger TSI in pre-2001 years because there was no change in passenger weights for those years.

Table 2. Percentage Changes in the TSI due to Changes in Weights, 2001-2004

Excel | CSV

NOTE: Minimum means the largest downward revision or, in years where there was no downward revision, the smallest upward revision. Maximum means the largest upward revision, or, in years where there was no upward revision, the smallest downward revision.

SOURCE: Bureau of Transportation Statistics

The changes in seasonal factors and the impacts

Changes in approaches to handling outliers (extreme observations), the type of filters that are used, how far back to use the data for seasonal factor estimation, and choices to treat the seasonality as additive or multiplicative cause changes in estimates of the seasonal factors. In addition, changes in data, either revisions to previous data or adding new data points to the data series, will also result in changes in the estimates of seasonal factors. However, because the index numbers are held unchanged after their preliminary period during the calendar year per TSI revision policy, changes in the estimates of seasonal factors will not be incorporated into the TSI until the annual revision.

The Change of TSI Base Year from 1996 to 2000 and Its Impact

The TSI base year was changed from 1996 to 2000 in the July 2005 TSI release. The base year was updated to provide a more meaningful reference to the most recent data because the most recent data is of greatest interest to TSI users. The new base year is now equal to 100 and the other index numbers are indexed to the new base year instead of the old base year.

Indexing to the new base year is performed by multiplying the old index numbers with the ratio of 100 over the old index number of the new base year. This ratio is referred to as the conversion factor. For the change of TSI base year from 1996 to 2000, the conversion factor was 0.8668 for the Total index, 0.8736 for the Freight index, and 0.8472 for the Passenger index. Changing the base amounts to a quasiparallel shift (or a proportional increase or decrease) of the level of the index and does not affect the month-to-month growth rates in the index. The graphs (Figure 1) below illustrate the changes in the Total TSI, Freight TSI, and Passenger TSI index numbers due to changing the base year from 1996 to 2000.

Figure 1. The impacts of changing base year from 1996 to 2000 on the TSI

Total TSI, 2000 based versus 1996 based

Freight TSI, 2000 based versus 1996 based

Passenger TSI, 2000 based versus 1996 based