IO-Snap

When a paid license expires, IO-Snap will continue to launch and operate but will be excluded from automatic software and data updates.

The full version has virtually the same classification as the annual BEA input-output accounts. BEA refers to this as the Summary level.

We plan to make available U.S. benchmark accounts with full detail in IO-Snap format before the end of 2023. We also are working to develop methods for adding or disaggregating a Summary-level sector based on a combination of user input and national benchmark accounts.

The data for the US and for all states (including Washington, D.C.) for the years 1998-2020 are included with IO-Snap.

Data that come with the Demo/Trial version are limited to a smaller number of historical years at a higher level of aggregation.

We use the Make and Use tables Before Redefinitions and Reallocations. The most important reason for this decision is that there are no data sources for employment that correspond to the After Redefinitions and Reallocations accounts.

There is a very nice paper that we recommend, called "From Make-Use to Symmetric I-O Tables: An Assessment of Alternative Technology Assumptions."

Yes! But, you’ll have to use your own regional data in the format of the state data we provide. National table regionalization is driven by regional personal income and value-added data by industry. You can enter employment or income data (or both) for a sub-state region if you have them, or even if you simply want to make them up for a hypothetical region of some idealized type. Once you've entered the regional driver data, you can generate the regional accounts.

You also can contact us to discuss special order IO-Snap data for custom regions and levels of sectoral detail.

Yes, we have a long list of potential features, from shift-share modeling thourgh industrial clustering analyses to hypothetical extractions and more. We are also designing graphical displays for select data.

Register for our Newsletter to be kept apprised!

We do our best to impute data using inter-variable relationships from national data and from data for Census Division. So, for example, if we have employment but no employment compensation for a given industry in a region, we use the employment / employment compensation ratio from the relevant Census Division to estimate the missing value. This is sometimes referred to in the economics literature as the Kendrick-Jaycox method.

More importantly, we mark all imputed input data with a light yellow background so that users have the option to apply their own imputation methods if and as preferred. We trust in the judgment and ability of our users, who oftentimes will have greater local or regional expertise to apply to the imputation process, and we support data editing to take advantage of that expertise.

Yes! This step is often neglected by casual users and can result in some substantial impacts estimate errors. You will note that the conversion option is only available in “commodity-driven” analyses, and this is because industry output is almost always composed of more than one commodity, so the adjustment needs to occur in commodity space, not industry space.

Great question! We don't see IO-Snap as competing directly with IMPLAN, which can be a terrific solution for a wide variety of analytical purposes. We've used it often ourselves and highly recommend it. The folks at IMPLAN go to great lengths to generate and provide highly detailed, disaggregated IO accounts for not only states, but also counties and even Zip codes, whereas we only provide state level summary data by default. And let's not forget that IMPLAN also has the ability to generate estimated interregional IO accounts. IO-Snap will be a useful tool for those who wish to have data for one or more states at what BEA refers to as the Summary level of reporting.

Many applications don't require highly disaggregated IO accounts, and the fact that we provide all data for all states over a series of years -- and in a friendly format -- facilitates interregional and inter-temporal comparisons. We also think there is a group of users who would prefer to effectively construct their own IO accounts. IO-Snap provides a good foundation for these analysts, with a high level of control over what goes into the process.

We also acknowledge that working at greater levels of sectoral and geographic detail comes at the cost of relying more heavily on data imputation. That is, each increase in levels of detail results in a higher proportion of missing data, and in some cases, also requires additional assumptions. This suggests that there exists a trade-off between, e.g., sectoral detail and accuracy. Given the need for a significant amount of imputation even at the Summary level, we have chosen to sacrifice some detail in exchange for greater parametric certainty (accuracy).

We also suspect that we put somewhat less effort into regionalizing final demand than does IMPLAN (though we don’t know all of the IMPLAN procedures). We make some basic assumptions, but leave any necessary refinement of regional final demand estimates to the user. Prior to generating direct requirements and multipliers, final demands, industry production functions and output distributions can be directly edited by the user.

IO-Snap will be useful for both professional application and in the classroom. We hope that the simple and intuitive IO-Snap interface will make input-output data and analysis more accessible to students.

Finally, we hope that the affordability of IO-Snap relative to Implan will make input-output analysis more accessible and to a wider audience.

So, the bottom line is that those who need high levels of detail and those who have become accustomed to IMPLAN's functionality will probably continue to find IMPLAN very well suited to their purposes. But we also think there is room for a product that fills a different set of needs, and IO-Snap is that product!