understanding significant impact changes 2012-2013
Hi,
I just purchased the 2013 California dataset and am seeing significant differences in the impacts for the construction industry and I'd like to understand why. While I expect there to be year to year changes, particularly in light of the BEA's revisions, I don't understand why new construction impacts are so radically different.
For example, here are outputs for a $100 million direct spend on housing construction. The 2012 sector is [b][i]37:Construction of new single & multi-family structures[/i][/b] and the 2013 sector is [b][i]60:Construction of new MF structures[/i][/b]
2012 2013 Change
[table]
[tr]
[td]EMPLOYMENT Direct 593 481 -19%[/td]
[/tr]
[tr]
[td]EMPLOYMENT Indirect 275 478 74%[/td]
[/tr]
[tr]
[td]EMPLOYMENT Induced 1,158 345 -70%[/td]
[/tr]
[tr]
[td][b]EMPLOYMENT TOTAL 2,026 1305 -36%[/td]
[/tr]
[tr]
[td][/b][/td]
[/tr]
[tr]
[td]OUTPUT Direct 99,999,999 100,000,004 0%[/td]
[/tr]
[tr]
[td]OUTPUT Indirect 47,257,950 70,678,896 50%[/td]
[/tr]
[tr]
[td]OUTPUT Induced 183,514,611 53,387,048 -71%[/td]
[/tr]
[tr]
[td][b]OUTPUT TOTAL 330,772,559 224,065,948 -32%[/td]
[/tr]
[tr]
[td][/b][/td]
[/tr]
[/table]
I see similar changes with other construction sectors. Could someone please explain what happened?
Thanks
Was this post helpful?
-
I should also note that the changes also appear in HH income impacts. For example, here’s the output from a $100M HH income spend in CA (using the 50-75K income category) 2012 [table] [tr] [td]Impact Type Employment Labor Income Value Added Output[/td] [/tr] [tr] [td]Direct Effect 0.0 $0 $0 $0[/td] [/tr] [tr] [td]Indirect Effect 0.0 $0 $0 $0[/td] [/tr] [tr] [td]Induced Effect 1,656.6 $99,234,157 $163,908,476 $264,582,207[/td] [/tr] [tr] [td]Total Effect 1,656.6 $99,234,157 $163,908,476 $264,582,207[/td] [/tr] [tr] [td]2013[/td] [/tr] [tr] [td]Impact Type Employment Labor Income Value Added Output[/td] [/tr] [tr] [td]Direct Effect 0.0 $0 $0 $0[/td] [/tr] [tr] [td]Indirect Effect 0.0 $0 $0 $0[/td] [/tr] [tr] [td]Induced Effect 850.8 $44,995,498 $79,411,089 $133,927,846[/td] [/tr] [tr] [td]Total Effect 850.8 $44,995,498 $79,411,089 $133,927,846[/td] [/tr] [/table]0
-
IMPLAN SupportHi Alex, Thank you for your post! we were able to duplicate your 2013 results assuming the Event Year is 2015 however we were unable to duplicate 2012 results. In order to make sure we are diagnosing the correct issue or explaining the correct difference we need to be able to verify your 2012 results. Could you please send in your 2012 model to: implangroup@implan.com If you are not comfortable sending in your model, could you send to us a screen shot of your model overview or excel export, the help > about window from the IMPLAN Software, and what Even Year you are using for your Activity. Thanks!0
-
i will put the model into a dropbox and send you a link since its too big to send over email.0
-
IMPLAN SupportHi! Thank you for sending the Model. I have received it and our team will review.0
-
IMPLAN SupportThank you for sending us the Model. Sector 37 for IMPLAN Model years 2008-2012 is divided into two Sectors in the 2013 IMPLAN Model: sectors 59 and 60. The majority of Sector 37, at the national average, becomes sector 59. Sector 60 is a relatively small share (about 10%). Sector 60’s profile in terms of how much they spend on intermediate goods (which drives indirect impacts) and Employment Compensation (which drives Induced impacts) is less comparable to 2008-2012 sector 37 versus 2013 Sector 59. For example, in California, Sector 60 spends 60% of Output value on intermediate purchases, whereas Sector 59 spends about 48% on intermediate purchases (which is close to the 2011 value for sector 37 of 49%). This would account for much of the increase in Indirect impacts, in both Employment and Output. The decline in Direct Employment is the result of overall lower Employment per Output. This is generally consistent with increasing productivity over time. In 2011, Sector 37 has Output per Worker at $144k. In 2013, sector 59 has Output per Worker of $168K, with Sector 60 at $198k. So, this appears to be the effect both of increasing productivity and of one of the constituent Sectors (Sector 60) have higher productivity than the other Sector (Sector 59). Unless increases in compensation per worker fully make up for the relative decrease in Employment per Output1 (which often is not the case), the lower Direct and Induced Employment impacts will lead to lower Indirect Effects, overall, all else equal. Recall that Induced Effects measure the effect of spending of compensation. So, if there is less total money paid to workers, Induced effects will be lower. Please let us know if you have any more questions.0
-
i'll have to review this in depth (i understand SF & MF homes have different production functions), but this still doesn't explain why the induced HH income spends are so different. as you can see from the tables i provided the 100M spend is halved in the induced impacts.0
-
i'll have to review this in depth (i understand SF & MF homes have different production functions), but this still doesn't explain why the induced HH income spends are so different. as you can see from the tables i provided the 100M spend is halved in the induced impacts.0
-
IMPLAN SupportHi Alex, When we do a comparison of a Labor Income Change in CA 2011 (your "2012" Model is using 2011 data) vs. CA 2013 , the total effect is about 10% lower in 2013, which is a reasonable level of change given changing tax rates, household counts by income group, BEA Benchmark, and RPCs. There was a national payroll tax break that was in effect in 2011 that is not in effect in 2013, meaning that households had less disposable income in 2013 all else equal. Please see the attached release notes for other changes incorporated into the 2013 data sets. Also, we noticed that all of the Events in your 2011 model have an Event Year of 2007, which means that IMPLAN will inflate your values (to 2011) before applying to the multipliers, which would increase impacts. Was this intentional? You will also need your Dollar Year for View to be the same between both years' results when comparing between the two years. We're continuing to investigate and will let you know if we find any other major reasons for the differences. [attachment=634]h7aa025a.docx[/attachment]0
-
IMPLAN SupportHi Alex, In the native-default build of the Model the variance is the 10%, but this is clearly not equivalent to the results that you are getting, so we went back into the Model you provided us and discovered that your Multiplier values were significantly higher than the default build that we were comparing to, and that this was clearly the source of the differences between your 2011 CA Model (based on Model Overview) named 2012 CA Statewide. Doing a little additional digging we discovered that the source was that the analyst who built the 2011 Model endogenized all Institutions, thus significantly increasing the amount of locally captured spending included in the Multipliers. We have included a link that describes how this impacts the Multipliers and the rare circumstances where you may want to include additional effects of Institutions beyond Households and Labor Income into the Induced Effects. The source of the differences that you are describing lie in the analysts choice to modify the Model from the default Multiplier build to include non-household and income related Institutions. This is typically not recommended as it will drastically increase the effects. http://implan.com/index.php?option=com_content&view=article&id=364 Thanks!0
Please sign in to leave a comment.
Comments
9 comments