# Induced Output and Employment Effects

Hi Implan, I have a few questions about multipliers. 1) How does Implan calculate type SAM multipliers by default? I think that type I multipliers are calculated as inv(I-A) where A is the square matrix of technical coefficients for the 440 Implan industries. Is this correct? Which industries, institutions, and factors are included in the matrix A that is used to calculate type SAM multipliers? 2) How does Implan get employment from output? From what I've read, it sounds like Implan uses the study area data to calculate (employment/output) for a given industry, and multiplies that ratio by direct output to get direct employment. However, when I check this, it appears not to be the case. Thanks in advance for your help. Stephen
0 out of 0 found this helpful

• Hello Stephen, The default Type SAM multiplier is all industries, labor income (employee compensation and proprietor income) and Households (all nine) internalized in the Leontief inverse. [url=http://implan.com/v4/index.php?option=com_docman&task=cat_view&gid=150&Itemid=60]This paper[/url] describes and shows the mechanics for the various forms of Type SAM multipliers, including the factors you mention below. We get our Employment data for the background Study Area Data from a variety of different sources delineated [url=http://implan.com/v4/index.php?option=com_multicategories&view=categories&cid=258:employmentdata&Itemid=71]here[/url]. That ratios that you are describing are derived from the Output values and Employment values pulled from the various federal sources we use, and these ratios are used to estimate Employment from Output or Output from Employment in a Event when the user is only able to supply the information for one of those two Event fields. Please let usknow if this addresses your concern or if you have any additional questions.
• Thanks, that's very helpful! I'm still a little confused about how direct employment is derived from output when only output is supplied, though. For example, using the 2009 California data file from Implan, I get that a \$1,000,000 industry change in Oilseed Farming (code 1) directly results in 8.8 jobs. When I look at the study area data, I see that, in 2009 in California, output for oilseed farming was \$56,243,936 and employment was 449.3. Based on my understanding, which is probably very incorrect, of how employment is derived from output, I would expect a direct employment change to correspond to (employment per \$ output) * (direct output change). However, (449.3 / 56243936) * 1000000 = 7.99, not 8.8. What am I missing here? Thanks again for your help.
• Hi Stephen, You calculation below is correct and the value of the Employment Multipliers for 2009 CA Sector 1 is 7.99. However, you will only get that exact Employment Multiplier if your Event Year and Dollar Year for View are also both 2009. We are guessing that you are probably using and viewing in current year dollars, so what you are seeing is the effect of the deflators being applied. Please let us know if you have any additional questions.
• You were right, thanks! I do have another question, though, that is related to tax impacts. I understand how IMPLAN calculates local, state, and federal tax impacts for employee compensation, proprietor income, and indirect business tax. However, I'm a little unsure about how IMPLAN calculates tax impacts for Households and Corporations. My first guess is that IMPLAN calculates household tax impacts as follows: (household tax impact) = (value added change) * (% of value added that goes to households) * (% of household income that goes to government) Could you please tell me whether this is incorrect? Thanks again for your help!
• Hi Stephen, Sorry that it took us a little while to get back to you. Your guess is correct. The only denotation we would add is: (household tax impact) = (value added change) * (% of value added that goes to [b]LOCAL[/b] households) * (% of household income that goes to government) but we presume this was likely implied by your context, we just wanted to add for clarification. Please let us know if you have any additional questions. IMPLAN Support Team
• Great, thank you! Sorry for all of the questions, but I have another one. I'm still struggling with how IMPLAN gets the household tax impact numbers. I've been trying to replicate IMPLAN federal tax impact results for households by doing the following: 1) (household impact from employee compensation) = (employee compensation impact) * (% of employee compensation that goes to each household bucket) * (% of household income that goes from each household bucket to government) where: a) "% of employee compensation that goes to each household bucket" is a vector of length nine that gives the percentage of employee compensation that goes to each of the household income buckets (element 1 is the % of employee compensation that goes to code 10001, element 2 is the % of employee compensation that goes to code 10002; etc.) b) "% of household income that goes from each household bucket to government" is a vector of length nine that gives the percentage of household income from each bucket that goes to government (element 1 is the % of code 10001 spending that goes to government code 11001, element 2 is the % of code 10002 that goes to government code 11001; etc.) 2) (household impact from proprietor income) = (proprietor income impact) * (% of proprietor income that goes to each household bucket) * (% of household income that goes from each household bucket to government) where: a) "% of proprietor income that goes to each household bucket" is a vector of length nine that gives the percentage of proprietor income that goes to each of the household income buckets (element 1 is the % of proprietor income that goes to code 10001, element 2 is the % of employee compensation that goes to code 10002; etc.) b) "% of household income that goes from each household bucket to government" is a vector of length nine that gives the percentage of household income from each bucket that goes to government (element 1 is the % of code 10001 spending that goes to government code 11001, element 2 is the % of code 10002 that goes to government code 11001; etc.) 3) Taking the sum of all of the elements in the vector created in (1) and adding it to the sum of all of the elements created in (2). Could you please tell me where I am going wrong? In your previous message, you mentioned local households. If I have all LPP values set to 1.00, will this make an impact? Let me know if this question is unclear or doesn't make sense, too. Thanks again for all of your help.
• Hello Stephen, What you are doing looks correct. However, if your number is low, then you may wish to replicate with Other Property Income payments to households as well. Corporate profits are treated as a leakage, but rental income, interest, etc. accrued by households may be re-spent by households. If you still have questions after looking at this adjustment, please let us know.
• Thank you. I am actually getting numbers that are too high. As a simple example, I'm looking at a \$1,000,000 industry sales increase in the oilseed industry (industry code 1) using 2009 California data. I calculate the change in employee compensation as \$210,017. (1) I find that the percentage of employee compensation (code 5001) spending that goes into households is as follows for each household bucket: Value.10001 0.003519218 Value.10002 0.004153971 Value.10003 0.022576744 Value.10004 0.040438686 Value.10005 0.084314186 Value.10006 0.174809680 Value.10007 0.146259688 Value.10008 0.185549822 Value.10009 0.221726872 These numbers are calculated by dividing payments from employee compensation to a given household bucket by total payments from employee compensation to any sector. (2) I multiply these numbers by \$210,017 to find how much money flows from employee compensation to each of these household buckets. I get the following (each row shows the percentage of employee compensation spending that goes to the given household bucket): Value.10001 739.0966 Value.10002 872.4057 Value.10003 4741.5065 Value.10004 8492.8230 Value.10005 17707.4362 Value.10006 36713.0542 Value.10007 30717.0624 Value.10008 38968.6696 Value.10009 46566.4754 (3) Then, I find the percentage of household spending that goes to the federal government. Each row here represents the percentage of household spending from the given bucket that goes to the federal government (code 11001). Value.10001 -0.004216124 Value.10002 -0.024781378 Value.10003 -0.006793951 Value.10004 0.016023332 Value.10005 0.038724749 Value.10006 0.063692183 Value.10007 0.077846984 Value.10008 0.095219457 Value.10009 0.106116626 These numbers are calculated by dividing spending from a given household bucket on federal government (code 11001) by total spending for that household bucket. (4) Next, I multiply the numbers calculated in (2) by the numbers calculated in (3). I get the following: Value.10001 -3.116123 Value.10002 -21.619414 Value.10003 -32.213563 Value.10004 136.083324 Value.10005 685.716020 Value.10006 2338.334583 Value.10007 2391.230673 Value.10008 3710.575562 Value.10009 4941.477237 (5) Next, I sum these numbers to get federal tax spending from employee compensation via households. I get \$14,146.47. (6) I follow the same process for proprietor income (proprietor income impact of \$126,317) and get \$11,417.15. (7) I add the numbers from (5) and (6) together to get \$25,563.61. IMPLAN gets \$22,315. Thank you for bearing with me. I'd really like to understand why the results that I am getting are inconsistent with the results from IMPLAN.
• Hi Stephen, If your numbers are too high, and based on your listed methodology, are you removing payroll and income taxes as well as savings from your compensation and income numbers before doing your calculations? If not that may be the problem. The Labor Income Change Activity removes all these factors before running the remaining income values against the spending patterns. Please let us know if this helps, or if you are taking this into account and still having issues with your calculations.
• Hi, Do you mean that the \$210,017 should be lower, the shares of employee compensation spending that go to the household buckets should be smaller, the shares of proprietor income spending that go to the household buckets should be smaller, or the shares of household spending from each of the buckets to the federal government should be smaller? The \$210,017 employee compensation impact is consistent with the result that IMPLAN gets for employee compensation value-added. Is this the incorrect starting point (I am able to replicate taxes from employee compensation using this number)? I see that employee compensation flows to households via wages and salary (w/out Social Security), other labor income, and transfers. Currently, I am aggregating these three items into one number, which represents any expenditure from employee compensation to households. Should I not be aggregating these up? Similarly, the \$126,317 proprietor income impact is consistent with the result that IMPLAN gets for proprietor income value-added. Is this the incorrect starting point (I am able to replicate taxes from proprietor income using this number)? As far as I can tell, proprietor income is flowing to households only through proprietor's income (w/out Social Security and CCA), so I can't think of anything that I could change there. I also see that households send money to the federal government via income tax, estate and gift tax, and fines and fees. I am aggregating these three numbers into one number, which represents any expenditure from the given household income bucket to government. Should I not be aggregating these up? Thanks again for all of your help - I really appreciate it.
• Hi Stephen, If you are starting with Employment Compensation or Proprietor Income, you'll need to first remove payroll taxes and commuters based on the EC or PI column's payments to government and domestic trade. Then you'll need to remove income taxes and savings based on the Household columns' payments to government and capital. [url=http://implan.com/v4/index.php?option=com_kunena&func=view&catid=83&id=12965&Itemid=35#12967]Here's a great forum post [/url]that explains it all nicely. Hopefully this will help to resolve the issues you are seeing.
• Great, thanks! That was very helpful. I'm getting much closer now. Currently, I am taking the change in employee compensation and proprietor income from the industry sales change that I specified earlier (\$210,017 and \$126,317 respectively), and plugging them into IMPLAN as labor income changes. Then I am taking the employee compensation and proprietor income impacts that result from this labor income change (\$97,015 and \$18,108 respectively) and using them as the starting point for the calculation that I outlined in my post two days ago. I have also incorporated Other property income as you suggested by using the other property income impact that I get from the industry sales change (this number is \$536,124). I multiply this number by (% of other property income that goes to each household bucket * % of each household bucket that goes to government) and take the sum to calculate taxes from households from other property income. For federal taxes, I get within \$10 of IMPLAN, but for state taxes, my number is nearly twice as high as the number that IMPLAN gets! Can you think of why this methodology would work for federal taxes but not state taxes? Thanks again for your help and sorry for all of the questions.
• Hi Stephen, The methodology you outlined in that post appears to be correct, so there may be a deflators thing going on again. The Dollar Year for View of the tax impacts need to be in the same year as the Employment Compensation and Proprietor Income figures you are using in the calculations. Please let us know if this does not resolve the issue.
• Thanks. I just double checked and the dollar year for view is correct. I am getting the same numbers as IMPLAN for employee compensation, proprietor income, tax on production and imports, and corporate state and federal tax impacts, but I'm still seeing a difference in household state and federal tax impacts. The difference between what I am getting for household federal tax impacts and what IMPLAN gets is very small, but I am getting a number that is about twice as big as what IMPLAN gets for household state tax impacts.
• Hi Stephen, To provide additional assistance, we think it may be helpful to look at your model. If you could please email your model, and any other information that you think may help us to help you resolve this to implangroup@implan.com, we would greatly appreciate it. Thanks!
• Thanks again for your help! I've emailed some additional details to implangroup@implan.com.
• Hi Stephen, Thanks for sending the spreadsheet. A couple of quick questions: [ul] [li]What region are you working with?[/li] [li]What year of data are you working with?[/li] [/ul] Thanks!
• Thanks. I am using 2009 California state-level data.
• Hi, I just wanted to check in to see whether you had a chance to look at the model. Thanks again!