Memorandum
City of Lawrence
City Manager’s Office
TO: David L. Corliss, City Manager
Diane Stoddard, Assistant City Manager
CC:
FROM: Roger Zalneraitis, Economic Development Coordinator/Planner
DATE: October 14th, 2008
RE: Recommended Updates to the Benefit-Cost Model
City Staff has recently completed a first draft of the benefit-cost model. Upon completion, staff conducted three meetings with external parties to review the model and receive feedback. The first meeting was conducted on August 29th with representatives from the City Manager’s Office, Douglas County and the Chamber of Commerce. The second meeting was on September 25th with Commissioner Chestnut and David Corliss. The third meeting was on September 26th with three professors from the University of Kansas and two citizens of Lawrence who are actively involved in economic development issues.
From these meetings, several suggestions were made. After review, the following appeared to be key items for which to seek direction from senior management and the City Commission:
Conceptual issues:
1) Should a “but-for” estimate be included in the benefit-cost model?
2) Should the discount rate be adjusted to include firms that fail to fully comply with performance agreements?
3) How should we treat city assets that are built, acquired, or sold?
Operational issues:
1) Which cost and revenue items from the city and county budgets should be included when calculating the impact of new residents and employees on government expenses?
2) Similarly, how should “persons” within the city be counted- in particular are residents who work in the city being double-counted?
3) Should we use average costs or marginal costs for capital improvements?
4) Should spending by construction activity be incorporated into the model?
Presentation issues:
1) What results would the City Commission like to see presented when reviewing incentive requests?
The following memo provides a discussion these items. Options and recommendations will be made where appropriate, but should not be considered final.
Conceptual Issues
“But-For”
The “but-for” issue has been raised both in the past and by some of the attendees at the meetings. There is a strong argument to be made that incentives should only be extended if they will be the determinative factor in a firm locating in the City. In other words, “but for” the incentives, the firm would not come to Lawrence.
The “but-for” provision affects not only whether to extend a subsidy, but how to account for it. As noted by Commissioner Chestnut in his November 2007 memo, if a firm would have located in the City but received an abatement anyway, this should be considered a cost to the City. However, if the firm would not have located here “but for” the abatement, the abatement should not be considered a cost.
A simple example may help clarify. Suppose you had a roommate who wanted to have a party. He was going to charge each guest one dollar, there would be 20 guests, and afterwards it would cost about five dollars to clean up the room. However, he needs five dollars from you because he would not otherwise be able to buy the supplies for the party. Since he has agreed to let you keep the profits, he’s asking to borrow the money but not return it- in effect he’s receiving a subsidy.
The revenue in the above example case would be $20 (20 guests times one dollar each). The costs would be $5 to clean. With no subsidy, the profits would be $15. The benefit-cost ratio would be the revenues (20) divided by the costs (5), or 4.
If you give your roommate a subsidy, the resulting benefit-cost ratio would depend on whether the subsidy was needed or not. If the party would not have occurred, then the appropriate way to account for the subsidy is to reduce the numerator- or revenues received- by the subsidy amount. Now, instead of $20 of revenue, the calculation would show $15 of revenue. Dividing by the five dollars in costs, the benefit-cost ratio is 3.
If it turns out, however, that the party would have occurred without the subsidy, then the subsidy needs to be counted as a cost. In other words, it gets added to the denominator rather than subtracted from the numerator. Now there is $20 of revenues, but $10 of costs, for a resulting benefit-cost ratio of 2. Therefore, how the subsidy is accounted for can be extremely important for determining the net benefits.
In the Kansas University tax abatement model, there was an implicit assumption that the subsidy was necessary to attract the firm. In the new model, the assumption is the opposite: the subsidy is counted as a cost. Neither model attempts to answer whether a firm would locate without a subsidy.
Since the initial reviews, two other outputs are now available in the new benefit-cost model. The first result shows the benefit-cost ratio before any subsidy is offered at all (the equivalent of the “4” in the above example). The second output was the original one provided, the benefit-cost ratio if the subsidy is counted as a cost (the “2” in the above example). A third output was added in response to this discussion- the benefit-cost ratio if the subsidy was necessary (the “3” in the above example).
Conducting a preliminary analysis of whether a firm would locate “but for” the subsidy is not advisable for the following reasons. First, the data necessary to do a proper analysis is virtually impossible to obtain (for example, one would need to know the incentive packages offered by other cities). Second, there are no agreed upon methods for a “but for” analysis, other than to ask a firm if they need the subsidy. There is, however, fairly broad agreement that a firm when applying for a subsidy will answer that question in the affirmative. Finally, the City may have public objectives articulated in its economic development policies that can be advanced through offering incentives to targeted firms or industries, which might be restricted by a “but for” analysis of this type.
Recommendation: consider and approve one of the following alternatives:
1) provide one benefit-cost ratio for each analysis, but note which one of the three is being provided and note that depending on one’s assumptions about the necessity of the subsidy, the benefit-cost ratio may actually be different; or
2) provide all three benefit-cost ratios with a brief explanation for each (eg- “with no subsidies,” “assuming a subsidy is necessary and is offered,” and “assuming a subsidy is not necessary and is offered”). This would allow the City Commission to determine with their own judgment whether they believe the subsidy is necessary or even desired, and vote accordingly.
The Discount Rate
In order to properly account for future costs and revenues, these monetary streams need to be valued in present day dollars. This is done through a “discount” of those dollars. Since the public subsidy is treated in this benefit-cost model as an investment, we believe that the proper discount rate is to identify a “risk-free” rate of return and add a “risk” component to that rate. Currently, the “risk” component is based on past abatement performance. Prior abatements were analyzed over a 15 year period. The current discount rate is adjusted for the number of firms that received abatements but ceased operating in the community within 15 years.
There are two competing arguments regarding the discount rate. The first argument is that the investment is actually less risky than mentioned above, particularly when the investment is being used to create an employment center. The second argument is that the discount rate is underestimating risk because it is not including firms that failed to fully comply with their performance agreement.
The first argument is valid in particular for infrastructure projects. A key consideration, however, is what happens if the first firm at an employment center leaves. A higher discount rate helps account for negative future possibilities, such as the employment center remaining vacant for a period of time, or a new entrant seeking and receiving an incentive. A higher discount rate, in other words, is a conservative assumption for unknown future adverse outcomes.
The second argument is also important. However, in the new economic development policy being crafted, a section is being considered that would establish compliance requirements and offsetting incentive reductions if compliance is not met. If that section is included in future economic development policy, a discount rate that also accounts for failure of previous firms to comply would be a double-count. As of this time, we believe that compliance will be accounted for in the new economic development policy. If it is not, then we may wish to revisit the discount rate analysis.
Recommendation: continue with the current discount rate and methodology.
An additional consideration is that a separate discount rate may have to be developed for TIF and TDD applications, should those applications require a benefit-cost analysis. The reason for this is because of the different time period involved.
Treatment of Asset Sales and Acquisitions
In some cases, the City may be selling assets, such as land, to a firm. In almost every case, the City will be accounting for capital projects to accommodate additional economic activity (part of this related to average and marginal costs, which will be discussed later). For example, a new office building might require a turn lane be constructed.
The model currently accounts for any asset sales as a revenue, and any asset construction as a cost. An issue was raised concerning whether asset sales should be counted as a revenue. An asset already has a book value, and the sale of the asset merely exchanges the book value for cash. Thus, there would be no change in net assets for the City, and hence no revenue.
In order to provide consistent treatment of assets throughout the model, a similar approach would need to be applied to new public capital assets as well. Currently the model treats the creation of additional assets (fire stations, roads, sewer and water systems) as a pure cost. However, many of these items have a market value and are treated as a capital asset for financial purposes.
In reviewing other city models and model building guidelines, no other models treated new assets built as anything except a cost. For the purpose of consistency in asset treatment, it is important to count assets built and assets disposed of the same way.
Recommendation: continue treating asset divestitures as revenue and asset construction as a cost. This will result generally in a more conservative outcome, as capital costs apply in almost every circumstance, and asset sales in a more limited number. This is consistent with other cities’ approaches, and avoids treating assets sold and assets acquired differently.
Operational Issues
Cost and Revenue Items from City and County Budgets
In the model, costs and revenues are generally estimated by additional persons in the City and County. These costs and revenues are derived by taking the general operating budget for both jurisdictions and dividing by the “population” (the question of who the population is will be discussed shortly) to arrive at a per capita output.
Some meeting attendees suggested that this may not accurately reflect the impact of population on the budget. Several budget items are not driven by population. Thus, by including those non-population-sensitive items, the model is overstating the impact of additional people in the community.
Internal reviews were conducted with both the City and County. After these reviews, we determined that the following City budget items could be counted as non-sensitive to changes in the population (an asterisk * denotes accounts with less than $250,000 in them):
City Revenue receipts:
City Expenses:
For the County, the revenues that are not population sensitive are:
County Expenses that are not population sensitive are as follows:
Generally, City and County believe that expenses associated with internal operations are more sensitive to technology improvement than population changes. County officials noted the Appraisers office, for example, is now able to use fewer appraisers to survey more properties every year due to technological advances. County officials believe that their primary expenses driven by population growth were related to public safety. Policies limit the impact of population growth on planning and parks for the County.
Finally, it is important to note that one of the key goals of the model is to keep the inputs as simple as possible. Any review of line items should be mindful of this goal.
Based on this review, and with consideration to keeping the model as simple as possible, one of the four following options is recommended:
1) Keep the model “as-is”- all costs and revenues from the General Operating Fund are included;
2) Exclude transfers only- this line item is largely part of the budget for reconciliation purposes, and has nothing to do with population, technology, or any other driver of City and County budgets;
3) Exclude only items above a certain cost threshold- many expenses and revenues, as noted with an asterisk, above, are small and have little impact on overall costs and revenues. In order to simplify, we may wish to exclude only items over a certain dollar threshold (in this case $250,000) that are on this list;
4) Exclude all items identified by City and County staff as being non-sensitive to population.
Counting “Persons” in Lawrence and Douglas County
Many benefit-cost models, including this one, look at the impact of a new business through business activity and population growth. The population includes both new residents as well as new employees. Both residents and employees have a certain impact on revenues gained and expenses occurred in taxing jurisdictions.
The model currently counts all new “persons” in the City and County. “Persons” are defined as all new residents, plus all new employees. Some participants in the review sessions expressed concern about double-counting. Basically, if a person lives in Lawrence and works here, the model counts that person twice in order to figure out costs and revenues. There was also some question as to whether a person who only works in the City or County but does not live here, has the same impact as a person who is here all day.
In order to reduce the potential of double-count, a new estimate was taken of “persons” in Lawrence and Douglas County. The count of persons compared to the old method is described in the following table:
Type of “Person” Original Method Proposed Method
Resident, Unemployed* 1 person 1 person
Resident, Works in City/County 2 persons 1 person
Resident, Works Elsewhere 1 person .5 persons
Non-Resident, Works in City/County 1 person .5 persons
*An “unemployed” resident is defined as any person who lives in the City or County who is not in the labor force. This includes children, the retired, and those who either choose not to work or cannot find a job.
In the above example, the original method would have yielded 5 new “persons”, whereas the proposed method would show only 3 new persons. The original method implies that a person who both lives and works within the taxing jurisdiction has twice the impact on costs and revenues as any other person in the locale. It also implies that a commuter to a job in the taxing jurisdiction has the same impact as an “unemployed” person who lives in the City or County.
Recommendation: adopt the proposed method for determining how many new “persons” there are in both the City and County. This will make the calculation slightly more complicated, but will also eliminate a double-count on residents who work in the jurisdiction.
Average Versus M arginal Costs for Capital Improvements
New population and new firms have an impact on the need for capital infrastructure in a location. Accounting for these capital costs incrementally is an “average” cost method, whereby each new entrant is charged a certain amount of additional infrastructure, even if the infrastructure is not added today. In contrast, a “marginal” cost method only charges a new entrant infrastructure costs when the infrastructure is built.
The model uses an average cost for calculating capital improvements. There are two reasons for this. First, using a marginal cost would allocate all of the costs to one entrant, even when multiple new entrants are responsible. For example, suppose that a business park did not require a stoplight on a highway until the last firm entered the park. Under a marginal cost approach, only the last firm entering the park would be charged for the stoplight, when in reality every firm entering the business park contributed to the cost.
Second, recommended practice is to use an average cost when a city is growing moderately or there is capacity in the current infrastructure to handle additional entrants. Some cities in the Kansas City metro area do use marginal cost approaches, but the cities that do tend to be landlocked- there are near-term limits to their growth potential. We believe that there is generally spare capacity in the City and County, growth has been slowing in recent years, and there is room through annexation for further City expansion.
There is an option to turn off the calculations for capital improvements though. This allows the user to manually enter any capital costs for the new project, but only for the firm. Capital costs for new residents are always calculated. In terms of the firm, this allows the user to enter all or partial infrastructure costs associated with a new firm.
Recommendation: under most scenarios, an average cost approach is acceptable. If, however, a particularly large firm moves into the City or an under-served area, staff can use a marginal cost method through manually entering infrastructure costs. The output should note whether capital costs were calculated or manually entered so the Commission knows how costs were assessed against the firm.
Spending on Construction
Initial construction of a new facility is counted as a cost in the model. However, construction activity is also an economic benefit to a community. Some respondents suggested that the model should account for the economic benefit that comes from construction activity.
In the Kansas City metro area, some cities use economic benefit from construction, and some don’t. The benefit that the model would pick up for City and County governments is chiefly sales taxes from goods purchased for building a new facility. These sales taxes tend to be fairly small compared to the net benefit of the project overall, however, and thus have little impact on the final outcome of the model.
One of the chief impediments to incorporating construction is that the jobs are transient. There are few if any lasting spin-offs from new construction activity. Further, there are also costs associated with construction activity that would have to be taken into account, such as increased heavy vehicle traffic leading to deteriorated road conditions. The net result is that the benefits and costs from construction activity may very well be zero, even if included.
Recommendation: do not include construction activity in the model. There are few if any benefits and many controversies surrounding how to incorporate construction benefits and costs. It is also a conservative assumption to withhold some benefits, which is generally a preferable approach to measuring benefits.
Output
The new model is entering the final stages of development. At this point, it is reasonable to seek feedback on what management and the City Commission would like to see in terms of output. At the very least, the following is recommended:
Guidance is sought on whether these outputs are appropriate for the Commission to make recommendations, and if any additional outputs would be desired. Further, a discussion of which variables to conduct a sensitivity analysis upon is required.