State Export Data: Strengths and Limitations
WISER-Census statistics on state exports to Mexico are fairly reliable, although like most data series the numbers have important limitations. Users need to keep in mind several things when utilizing these data, especially for policy purposes.
First statistics on merchandise exports of specific industries within states are, as a rule, somewhat less precise than state export totals. We believe that industry-level export figures are generally close to the mark, but accuracy does vary across states and industries. Consequently, common sense needs to be applied when using industry numbers. To take an extreme example, most would agree, contrary to what the statistics show, that the District of Columbia probably did not both produce and export $5 million in forestry products to Mexico in 1990. This is almost certainly a misleading statistic given the highly urbanized nature of the DC area.
Second, state and industry export data for manufacturing industries are considerably more reliable than figures for non-manufacturing industries. Data on state agricultural exports, for example, tend to be weak for reasons outlined in the following section.
Third, the WISER-Census data are likely biased in favor of border states. Exports of border states--particularly Texas--are probably overstated, while exports of non-border states are likely understated. Texas' exports to Mexico are inflated because the state is a primary staging area for supplying Mexican "maquiladora" enterprises. A significant portion of Texas' exports to Mexico is undoubtedly produced in other states. Even allowing for this distortion, however, exports to Mexico that are produced in Texas are unquestionably very large.
Unfortunately, we are unable to
provide quantitative estimates of data reliability for specific states
or specific industries within states. This is partly due to resource constraints
and partly due to the nature of the data-collection process, which is reviewed
below.
Origin of State Export Statistics.
World Institute for Strategic Economic Research (WISER) State Data Center data on state exports are drawn from United States Census Bureau data tape number EQ912 that is issued quarterly. This tape aggregates export origin information from Shippers' Export Declarations (SEDs), that are filed for virtually all merchandise exported from the United States.
On every SED, shippers are asked, among other things, to provide the "point (state" of origin" of the export. This question was originally inserted on the SED to identify transportation patterns and requirements. The goal was to pinpoint the origin of movement (the state where an item began its export journey), not the origin of manufacture (the state where an item was produced).
To identify the origin of movement, instructions for filling out the SED specify that shippers should choose from among the following definitions for "point of origin":
2. The state of the commodity having the greatest dollar value in a multi-product shipment.
3. The "state of consolidation," that is the state in which various parts of a multi-product export order are readied for shipment.
4. The Foreign Trade Zone for exports leaving an FTZ (an FTZ is similar in principle to a bonded warehouse). Using the FTZ "state of origin" definition, an export produced in Ohio, but shipped from an FTZ located in Florida, would show up in state export statistics as a Florida export.
Nevertheless, our judgment is that the state of origin coincides with the state of manufacture much of the time. We believe that the export journey often begins at the factory gate--or close to it. This conclusion stems from numerous discussions with exporters, statistical specialists at the Census Bureau, and other relevant sources.
We recognize, however, that an indeterminate percentage of shippers utilize point-of-origin definitions that have little bearing on the state of manufacture. Sometimes, for example, shippers will specify states where exports are temporarily warehoused.
The problem is most acute with respect to non-manufactures. Exporters of agricultural products, for instance, almost always specify the location of loading and storage terminals, not the location of producing farms, as the state of origin.
Another problem is that some shippers simply leave the "point of origin" block blank on the SED. On the Census tape, about 25 percent of the value of US exports is typically unallocated by state. WISER tries to deal with this shortcoming by applying a formula to "unallocated" data, breaking down the category and reassigning export values to individual states. Although resulting numbers give a much neater picture of state exports, the degree of precision of the figures is unknown because WISER's reallocation procedure cannot be validated.
Finally, WISER-Census state export data for the years 1987-90 are not fully comparable across the entire time series. First, revisions in the Standard Industrial Classification system were implemented in 1988. Thus 1988-90 state exports for individual industries are classified somewhat differently than in 1987. Further complicating the situation is that, beginning with 1989 data, reporting of US export statistics shifted from the long-standing "Schedule B" nomenclature to the new Harmonized System (HS). This move introduced a number of classification problems that the Census Bureau is currently working to resolve.
Although the WISER-Census data have major weaknesses, the fact remains that these data are the most up-to-date source of information on state exports. The only alternative source for state export figures is a Census Bureau series which is typically issued with a three-year lag.
To sum up, care must be taken when drawing conclusions from the WISER-Census series. One should not rely solely on WISER-Census statistics when making statements about the contribution of exports to a given state's employment and overall economic health. Such statements should be supported, not only by WISER-Census data, but also by other evidence furnished by state agencies, trade associations, or other sources.
