<img src="//bat.bing.com/action/0?ti=5564067&amp;Ver=2" height="0" width="0" style="display:none; visibility: hidden;">

IMPLAN Blog

Validating Updates to IMPLAN's Gravity Model

May 25, 2018 by Ross Conroy

Recently I sat down with IMPLAN economist, James Squibb, to discuss his recent paper acceptance at the upcoming International Input-Output Association (IIOA) conference in Brazil this June. This paper, an extension of Incorporating Port-Level Foreign Trade Data into IMPLAN’s Gravity Model to Estimate Region-Specific Foreign Trade Rates published in 2017 by James and Jennifer Thorvaldson, attempts to validate the reliability of the new gravity model updates. Here is what I learned.

Since 2005, IMPLAN has used a double-constrained gravity model to estimate intra- and inter-county flows of goods and services. By having these trade data between all the counties, you can build what's called MRIOs, or Multiregional Input Output models, from which you can analyze the effect that a change in a U.S. region can have on another region.

The issue with the current gravity model approach is that MRIO analysis can only examine U.S. county-to-county, state-to-county, and state-to-state trades. To estimate the impact of foreign trade, we’ve historically used a simpler assumption based on available national-level data. In this approach, we assumed that every commodity-producing county has the same export rate as the United States as a whole. This assumption works in many cases, but is definitely too simple in others.

To overcome this deficiency, we updated IMPLAN’s gravity model in 2017 to incorporate port-level foreign trade data to better estimate region-specific foreign trade rates. The first step was to estimate foreign trade with some local specificity. But to do that with the gravity model, we needed certain data points. This was achieved using Census Bureau data to determine to which ports in the United States certain commodities are shipped and the value of those commodities. We can now treat each port as though it were a county with regionally specific foreign trade rates for each commodity.

The extension does two things. First, it attempts to determine how reliable the regionally specific foreign trade rates in the gravity model are by aggregating and comparing them against tabulated, state-level Origin of Movement (OM) and State of Destination (SD) data.

Second, the extension attempts to decompose all these data by country-level trade partner. The Origin of Movement and State of Destination data, port level data, and national rates all tell you (for any commodity) to what country it’s being exported or from what country it’s being imported. So with a little bit of data bridging, we come up with the trade partners.

What this means is we now have another option to tell how important imports and exports are and how important certain countries’ markets are. That can be valuable when examining the effects of changes in commodity exports and imports to specific regions. For example, producing regions closer to a port would experience a greater impact than those further away.

The intent of this change in the gravity model is to provide good data to researchers that really want to dig into doing international trade research and modeling. An added benefit is that once we know all the country-level trading partners identified, we are closer to building an international MRIO model to connect the United States as a whole to other countries as well as individual states and counties to other countries.

Topics: Data, Economics, Methodology

Why IMPLAN?

Put simply, IMPLAN is built for everyone.

With IMPLAN there are no black boxes. All of the background data used to create your personalized input-output models can be viewed and customized to reflect your Study Area. So now, what used to take economists weeks can be done in minutes. By anyone!

Reasons why IMPLAN is the best choice:

  • Easy to learn
  • Easy to use
  • Outstanding customer support
  • Instills confidence in your analyses

Subscribe to Email Updates

Recent Posts