When a U.S. control number or “CONNUM” is the same as the comparison market CONNUM the USDOC determines the products to be identical. However, when the identical comparison market CONNUM was not sold, how does the USDOC’s SAS computer programming go about identifying the “most similar” comparison market match? In an original investigation the USDOC defines the physical characteristics of the merchandise under investigation that best define the product, and assigns a hierarchy to those characteristics from most important to least. In addition, The USDOC assigns to each variation within a given characteristic a numerical code. Those codes are concatenated into a CONNUM from left to right in order of importance. These codes also define how similar the variations within a given physical characteristic are to each other. The USDOC weights the codes such that the closer the codes are numerically, the more similar the variations. As a result, the closer the overall CONNUMs are numerically, the more similar the US and comparison market models are to one another.
Let’s demonstrate this concept using a fictional case called “Frozen Penguins from the South Pole.” The physical characteristics chosen as best defining the frozen penguins are 1) species (most important), 2) size/weight and 3) color (least important). The variations within each characteristic and their assigned numerical codes are as follows:[table id=27 /]
In this example, the USDOC has determined the “white tail” and “black tail” species are “more similar” to each other as the absolute difference between their assigned codes is only 1 – while the difference between a “white tail” and a “Maximus” is 2 and a “black tail” and “Maximus” is 3. In this way, the weighting inherent to the numerical codes assigned mathematically defines similarity.
To determine the overall similarity between CONNUMs, the USDOC compares the codes for each U.S. physical characteristic to those for each comparison market penguin (that passes the cost test) and calculates the absolute difference for each. The USDOC’s SAS programming then reads the absolute differences between each physical characteristic, in order of importance, from left to right, as a whole number; sorts the data such that the lowest whole number rises to the top, and selects that as the “most similar” comparison market match. For example, assume we are trying to find the most similar match to a U.S. frozen penguin that is a 9 lb., Maximus with 90% black solid coloring. The CONNUM for this penguin is “1401.” Consider the chart below. U.S. CONNUM 1401 is compared to three different comparison market penguins with CONNUMs “3809”, “4220” and “1103.” When compared to the first model, the US CONNUM has an absolute difference of 2 between species codes, a difference of 4 between weight codes and a difference of 8 between color codes. These differences, read as a whole number from left to right is 248. When compared to the second comparison market penguin, the differences, read as a whole number from left to right is 3219, and when compared to the third comparison market penguin the number is only 32.[table id=29 /]
The “most similar” model is the one that is closest to the U.S. model numerically, as a whole number. In this case it is the third comparison market penguin, the 2lb., Maximus that has 50% solid black coloring. As is clear, while the physical characteristics identified in an investigation are important, perhaps even more critical is the order of importance of those characteristics, and the manner in which the USDOC ultimately weights the numerical codes assigned to the variations within each physical characteristic.
This blog post is part of a larger “Topics in Dumping” series, please go to the introduction blog post if you would like to see the list of other posts in this series.