IMPLAN Pro: Analysis-by-Parts Working with Government Institution Spending Patterns

Introduction:

The Institution Spending Patterns, found under the Activity Options>Import> Institutional Spending Pattern menu in the IMPLAN Pro software, represent a general spending distribution for measuring broad Institutional activity in your region. These include general spending for governments, capital, enterprises and households. This article is designed for the consideration of the government spending patterns.  Unlike the government 'payroll only' Sectors, these spending patterns provide the full picture of the Institution on the region, but include a few important caveats.

Detailed Information:

Whenever we want to examine the impacts of U.S. Institutions in IMPLAN, we need to do this through an Analysis-by-Parts like methodology, because the this allows us to examine the impacts of Institutional programs even though they have no underlying Multipliers in the system. Because the industries that these Institutions purchase from are included in the underlying Multiplier matrix, using an Analysis-by-Parts method allows us to capture Institutional impacts without Multipliers. By examining government purchases and their labor payments, we can estimate the impacts of these government activities based on their local purchasing and the resultant Indirect Effects and Induced Effects, as well as the impacts of their employees' spending . These spending patterns, unlike the Industry Spending Patterns, will report Direct, Indirect, Induced and Total results. Direct Effects are reported for Institution Spending Patterns because the first-round of Institutional spending is defined as a Direct Effect.

The government spending patterns, available through the Activity Options>Import> Institutional Spending Pattern menu, provide a high level view of government activity in two categories: State & Local Government or Federal government activities. The following three spending patterns are available for each government type:

• State & Local Government
  • Education,
  • Non-education,
  • Investment.
• Federal Government
  • Defense
  • Non-defense,
  • and Investment.

(Please note also that the behavior of the spending patterns found using the Activity Options>Import> Institutional Spending Patterns are different from those spending patterns found in the Activity Options> Import> From Another Model, Operations_Patterns_for_509Industries_MIG_Lib440.impdb (for data sets 2007-2012) or Operations_Patterns_for_509Industries_MIG_Lib536.impdb (for data sets 2013- current). These spending patterns accessed through the Utilities folder behave as the normalized spending patterns that exclude payroll, as described in the Usage section below. While older data sources are used for these spending patterns, the purchases in the spending patterns describe more specific aspects of State & Local Government behaviors, allowing you to split the State & Local Government Non-Education Spending Pattern into more discrete elements such as fire, police or parks. These additional spending patterns also include some investment and construction functions.)

These general spending patterns are unique in that they describe both the Intermediate Expenditure component and the Value Added component within the same spending pattern. This results in these spending patterns producing 'mixed results'; where the reported Direct Effects describe both what we would generally consider Direct Effects: the income, Employment and Value Added of the government Institution type, and the first-round Indirect Effects that arise from the government spending its budget. This mixing of the results requires some additional work on the part of the analyst to parse the results, or requires the analyst to divide the spending pattern into the traditional spending pattern prototype that includes only Intermediate Expenditures, and then create a separate Activity to account for the labor/payroll component. This article describes both methodologies and the advantages and disadvantages of both methods.

Usage:

Using the State & Local Government Education spending pattern to estimate the effects of our local public university.

The spending pattern we will import includes the spending for all types of State & Local Government Education, not just university education, thus it is a mix of K-12 and university spending. While there is no harm in using this spending pattern, it is important to be aware that for this reason you may want to edit the spending with IPEDS or other organizational level data.

To access the spending pattern, go to Activity Options>Import> Institutional Spending Pattern,

 

and select the State & Local Government Education Spending Pattern.

Once we selected our State & Local Government Education Spending Pattern, we can select Import to bring it into the Model.

If this is not the first Activity in the Model, we will need to click on the spending pattern description in order to view the associated Events.

Note that the Events are on a commodity basis and that the LPP is based on the regional availability of each commodity's supply to meet demand. This allows the university to purchase the necessary local commodities based on all available sources, and on the basis of the region's ability for supply to meet demand. Note also that the Sum of Event Values is 1.00. This is important because it means that every dollar entered in as the Activity Level will be spent across the spending pattern.

It is also important to ensure that the Event Year of the spending pattern matches the value we will enter as Activity Level. For this example, we will be looking at a public university whose total operational value was $455,000,000 for the 2015-2016 academic year. Since the deflators between the two years will not show a significant variance, we can choose one year and use that to represent the entire year's budget. In this example, we will use the deflators for 2015.

 

At the bottom of the Events list, note the two Sectors designed with an * and the wording 'Employment and payroll'. These Sectors are special Sectors that have no associated spending pattern and are used to calculate the Employment value and Value Added associated to government activity.

At this point in the analysis, we have two general options for proceeding.

Option 1: Leaving the Payroll Sectors in the Institution Spending Pattern

If we leave the spending pattern as is, we can use the full value of our university annual operations as the entered Activity Level. To enter the appropriate Activity Level, we will click the Edit Activity button and then enter in the operational value of our public university. For this example, the university operational value is $455,000,000 including tuition, room, and board.

Once we have saved our edits to the Edit Existing Activity screen, our Activity will be prepared for analysis.

Click Next, or Analyze Scenarios, to move to the Scenarios screen where we can create a new Scenario for our public university. Because the full annual value of operations is included in our Activity Level, we won't alter the Scenario Level.

Once our Activity is selected, we are ready to Analyze Single-Region.

It is important to remember that the reported Direct Effects are a mixture of the effects of the university itself and the first-round of Intermediate Expenditures in the region. This also explains why the reported Indirect Effects appear to be so low, and why the Employment, Labor Income, and Output reported for the Direct Effects do not match our our university numbers.

Since the first-round of Indirect Effects are captured as Direct Effects for Institution Spending Patterns, the LPP Imports screen will describe the amount of the leakage that occurred from the first-round of university purchases. In this instance, $53,552,990 dollars of university expenditures left the defined Study Area to purchase goods and services that weren't available to them within the region.

The Direct Institution Change  describes the production of the first-round of Indirect Effects that was provided by Institutional production and was thus did not generate any Indirect or Induced Effects.

 

We can get additional detail and can extract the effects from our university from those of their purchases by going into the Detail Results> View By:Output, Value Added, Labor Income and Employment and sorting by Direct Effects. The two Sectors showing the greatest Direct impacts are those of our university payroll.

Unlike the other listed Direct Effects in the Scenario, the values in for these Sectors represent the Employment, Employee Compensation and Value Added for the University we are modeling, so in order to get a clearer impression of what these results mean, we will want to separate out the university effects from the remaining impacts.

Exporting our Summary Results to Excel will allow us to remove the effects of the university from the remaining impacts.

The first step involves removing the Employment, Labor Income, Value Added and Output values from the reported Direct Effects that are derived from Sectors 532 and 534, as these are the 'true' Direct Effects. So in addition to our Summary Results, we will need to copy the following cells out of each Detail Results view in Excel.

Output:

Value Added:

 

Labor Income:

 

Employment:

 

Once we have these two Sectors captured in Excel for each of the views, we can sum the respective values for each element into our Direct Effect.

Note that when we do this, Value Added = Output. This occurs because these special 'payroll only' Sectors do not include any Intermediate Expenditures, thus the difference between the reported $387,547,839.86 and the $455,000,000 we entered is the value that went to purchase the first-round university purchases. We can now adjust the reported Direct Effects.

Once the values from Sectors 532 and 534 are removed, the remaining Direct Effects are shown above. Depending on how we want to talk about these values, it may be useful to leave the table in this condition.

The advantage of leaving the table as above, is that it tells the story of the impacts of the university itself (4,071 jobs, $339M of Labor Income, $387.5M of Value Added, $455,000,000 of Output) and provides a more discreet view into the effects of the university's purchases: 218 jobs are supported along with approximately $9M of additional Labor Income and $22.5M of production. These businesses supported directly by the universities operations, in turn, support another 45 local jobs and $2M of Labor Income. The Labor Income resulting from the jobs supported by all these activities then supports an additional 1,530 jobs and $59M of Labor Income.

Alternatively, formatting the results to match the 'expected' results can be done quickly and easily at this point. Here are the steps involved:

1. Remove the effects of the payroll Sectors from the reported Direct Effect, as above.
2. Sum the remainder in the Direct Effect row with the values reported as Indirect Effects and leave these values in the Indirect row.
3. Take the values for the Direct Effects (university impacts) for Employment, Labor Income and Value Added and copy these into the Direct Effects row along with the starting Output value used as Activity Level.
4. Check and re-sum the totals.

After following steps 1-4 the final results appear as above.

Option 2: Removing the Payroll Sectors from the Institution Spending Pattern

To remove the two payroll Sectors, click on Sectors 3532 and 3534 to select them and then select the Delete Event button to remove them.

Note that after the payroll Sectors have been removed the spending pattern Sum of Event Values is only 0.16. This indicates that by default the Intermediate Expenditures associated to State & Local Government Education activity make up only 16% of their total production value.

One advantage of removing the payroll Sectors is that we can easily, and predictably, change the assumption of the amount of spending that goes towards the purchase of goods and services for university operations to our known value. If we only have total operational value, and don't have a budget and payroll value, we can certainly use this percentage to approximate the split between Intermediate Expenditures and Value Added and use the $455,000,000 as our Activity Level. However, if we know the university goods and services budget, we can normalize the spending pattern so that we can use the budgetary value as our Activity Level.

Note that after we apply the Events Options> Change All> Normalize Events, the Sum of Event Values is 1.00 again, thus indicating that every dollar of Activity Level will be spent on the spending pattern. We can now set the Activity Level to the cost of goods and services for the university.

Now that the university goods and services have been accounted for, we need to see the impacts of their payroll effects. To do this we can create an Industry Change Activity Type.

For the College Payroll Activity, we will choose the state payroll Sector, 532. It's important also that the Event Years between the spending pattern and the Industry Change for payroll match, and both reflect the year in which our university budget and payroll are reported.

 It is important with this Event to be sure that you are entering the correct value into the correct field.

• If you are entering a university payroll value, your value should be entered in the Employee Compensation field and then IMPLAN can estimate the Employment and Industry Sales.
  • After entering compensation, you can edit Employment to a known value.
• If you are using the Intermediate Expenditures:Value Added ratios to estimate the Event values, the Value Added dollars should be entered into Industry Sales.
  • Once this is done Employment can be adjusted, if known.
• The Industry Sales value for this Event type is equal to Value Added, not payroll, and should be somewhat larger than the Employee Compensation value.
  • When summed with the Activity Level used for a normalized budget the Industry Sales value listed here plus the budget value should equal the operational value of the university, in this case $455,000,000.

Once the payroll Event is setup, we are ready to create our two Scenarios.

One Scenario for the spending pattern:

and one Scenario for our payroll.

Two Scenarios are used because this avoids having the 'mixed' set of results we saw for the original spending pattern analysis. From these two Scenarios we now get two sets of results.

The spending pattern results which describe the effects of our budgetary expenditures on the local economy. Again these results do not include Employment, Labor Income, Value Added or Output of the university, but reflect the Employment, Labor Income, Value Added and Output associated to the first-round of Intermediate Expenditures of the university in the local economy.

The payroll results which provide the university Employment and show the impact of university staff spending their income in the local economy.

 If we want a single set of combined results, we can export these to Excel.

1. Sum the Direct Effect and Indirect Effect rows in our spending pattern table together to create our total Indirect Effect.
2. Sum the Induced Effects rows together from both sets of analyses into the Induced Effect row.
3. Use the Direct Effects reported in the Payroll Activity for Employment, Labor Income and Value Added.
4. Overwrite the Direct Output value with our known University Operational value of $455,000,000.

Additional Questions & Considerations

What if I don't want to have both local and state Employment, because I know that my university is staffed by state employees?

• You can resolve this with either method. If you want to use the method that leaves the payroll Sectors in the spending pattern you can sum the two coefficients for payroll Sectors together and use the new coefficient with the relevant Sector.
• Alternatively you can use the second method as described. Removing the payroll Sectors, and  creating your new Industry Change Activity with the Sector that is appropriate for your known Employee Compensation or the ratioed Value Added. If you do have your own ratios, this method allows you more control over the analysis, as described below.

What if I know all the elements of my university operations and don't want to let the Model estimate the Direct Effects?

• If you have this information then the second method is advised. By removing the payroll Sectors and normalizing the spending pattern you have complete control over the ratios and how your values are applied.

Why is Value Added greater than Employee Compensation for a public university?

• While the majority of Value Added for a public Institution will Employee Compensation, there is a small additional value that these Sectors account for that describes the Institution's capital consumption allowance.

Why use the payroll Sectors instead of a Labor Income Change?

• The advantage is that it captures or estimates the additional Value Added element of the capital consumption. A Labor Income Change could also be used with the payroll value and 5001 Employee Compensation Event, but you would need to solve for Value Added separately.

Why don't the two Scenarios produce the same results?

• Ideally they won't because in the second Scenario you'll be using more known data to drive the results. However, even if you try to match them perfectly, some variance will arise between the two methods as a result of coefficient rounding, since the spending pattern Sum of Event Values shows only two decimal places.

Do I enter the income for my university in the payroll only Sector as Employee Compensation or Industry Sales?

• Always enter the payroll value in the Employee Compensation field.

 

 Related Topics

Case Study on Analysis-by-Parts