Getting smaller results on new software
Hi. I'm trying to update some work I did a few years back using the IMPLAN desktop software now that I'm using the IMPLAN web based software. Both analyses use the USA file (i.e. they're national). The original work used 2010 tables, for the new one I'm using 2014 tables.
I wanted to check my old results before updating things, so I took the exact inputs I ran back then in the 440 sectoring scheme (downloading the old inputs into excel), converted it to the 536 sectoring scheme using your crosswalk, and then uploaded into the new system. In both cases, both all the event years and the output display year are 2013. In both cases, it's an industry change. I checked to make sure the overall sum of the inputs was the same, and it is.
I'm finding though that the impacts of the new model are a full 13% lower than the impacts of the old model (judged pretty much any way you'd like, but I'm referring specifically to value added for the 13% number). I can understand that the results wouldn't be identical, but this is a huge difference.
Do you know what could be causing this? Thanks.
-
Hi Dan! Thank you for your Forum Post. So that we can compare the 2010 Model to the 2014, would you mind sending to us the 2010 Model? You can email it to info@implan.com As for the 2014 Model, could you send in your UserName and Password for us to view the Model. Thanks -
Hi Dan! Thank you for sending in your 2010 Data Model and the access information for your 2014 Model in IMPLAN Online. Even though both Models Activities are in the 2013 Event Year, since they are different base year data's (2010 vs 2014) they will yield different results. The year of the data set reflects how the economy behaved that particular year. To compare the IMPLAN Pro version to the IMPLAN Online version, I recommend building a 2010 Model in your IMPLAN Online account with the same Inputs. This should result in similar results. I hope that this helps! -
The online 2010 model does indeed yield identical results, which rules out a software issue, but I guess my question was really why such a large difference. Of course a different base year won't produce identical results, but is there intuition for why it's so much lower? I found part of the answer myself: the process I was using to crosswalk the old to the new sectoring (in Excel) wasn't working right, so even though the totals were right, the sectoring was still off. I fixed that, and now the new (i.e. 2014 base year as opposed to 2010) results are *only* 4% lower, not 13%. I guess that's close enough to be pretty reasonable; still, I wish I understood better if there's a simple reason why we'd expect these impacts to be lower. I guess I could mention that the spend scenario was built off of personal consumption expenditure tables, by way of context. Anyway, thanks for your help. -
Hi Dan. In comparing the 2010 Pro Model to the 2010 Online Model; it looks as if your Event Inputs are not identical and the Sum of Event Value is different. There should not be a variance of Results when the two studies are a identical match. There were some decent-sized changes between 2010 and 2014: new BEA Benchmark, new Census of Agriculture, and new Decennial Census during that period, plus 4 years of incremental methodology improvements. If helpful, here is a link to the Release Notes: http://implan.com/index.php?option=com_content&view=article&id=399 Thanks!
Please sign in to leave a comment.
Comments
4 comments