Memorandum
City of Lawrence
City Manager’s Office
TO: David L. Corliss, City Manager
CC: Cynthia Boecker, Assistant City Manager
Diane Stoddard, Assistant City Manager
FROM: Roger Zalneraitis, Economic Development Coordinator/Planner
DATE: March 24, 2009
RE: Final Report on the Benefit-Cost Model
Background
The state of Kansas requires that a benefit-cost analysis be conducted for any property tax exemption offered by a local government (KS 79-251(a)(1)). To fulfill this requirement, the City of Lawrence contracted with the Institute for Policy and Social Research at the University of Kansas. The Institute developed a model and provided analysis for all tax abatement applicants.
Earlier this year, KU notified Lawrence that the model needed to be updated. At the same time, there was a desire on the part of the City to have an in-house model in order to be able to simplify the analytical process, both in terms of the number of variables used and in terms of the ability to run multiple analyses on an application. For these reasons, the City chose to develop its own model.
Research
Staff conducted extensive research in developing the benefit-cost model. In order to create the model, staff did the following:
· Reviewed best practices as noted by existing literature;
· Met with KU staff and reviewed the KU benefit-cost model to understand how the analysis has been conducted to-date;
· Reviewed the State model that is available for municipalities as well as the questionnaire they recommend for applicants for tax abatements;
· Identified key issues and conducted several internal meetings to develop preliminary approaches to modeling these issues;
· Visited Lee’s Summit, MO; Kansas City, MO; Lenexa, KS and Manhattan, KS to review their models and how they handled some of the more difficult issues in modeling (these issues included multipliers, discount rates and costing infrastructure);
· Spoke with consultants to discuss where to obtain certain variables such as multipliers; and
· Met with City, County, and USD 497 officials to apprise them of progress and better understand their budgets and costs.
Output- First Draft
A first draft of the model was ready by the end of August. The draft version of the model measures costs and revenues for the City, Douglas Coun ty, USD 497, and the State. Revenues and costs are measured both for the firm as well as new residents that move to the community. Revenues include sales tax, property tax, any sale or lease of property owned by the City or County, franchise fees, state transfers to the school district, and income and corporate income taxes for the State. Costs include any new infrastructure built for the project, ongoing operating costs for the taxing jurisdictions, interest paid by taxing jurisdictions for bonds issued, and for the State any new transfers to the School District.
All data is derived either from the applicant’s questionnaire or from easily obtainable public sources. Much of the data is entered on a single page and the source of the data is clearly identified for users of the model. Some data- such as Census information and City and County budgets- is included as additional worksheets.
Some key features of the model are more abstract. In particular, this includes the multiplier, the number of new residents, and the discount rate. The multiplier is taken from the Bureau of Economic Analysis (BEA) RIMS II database. The multiplier measures the relationship of industries to one another in a local economy, and thus helps estimate the number of additional jobs and the salaries of those jobs when a firm relocates, expands, or contracts in a local economy. These additional jobs are known as indirect jobs. The multiplier will need to be updated every few years.
The number of new residents uses a procedure that derives its estimates from the U.S. Census’ Local Employment Dynamics (LED) database. The LED database measures job location and commuting patterns of every employee covered under unemployment insurance (it thus excludes federal employees as well as the self-employed and individual contractors). The model uses the data from LED to estimate where new firm employees will live, as well as where new indirect jobholders will live as well.
Finally, the discount rate attempts to value the stream of future revenues and costs in today’s dollars, under the key assumption that a dollar tomorrow is not worth as much as a dollar today. The discount rate in the model values “tomorrow’s” dollar at a rate equal to a risk-free rate of return plus a risk-adjustment for the likelihood that the total projected return will not be made.
Presentation
Upon completion of the first draft of the benefit-cost model, staff held a series of meetings to introduce the model, explain it, and receive feedback. A series of meetings was conducted with City, County, Chamber, and community representatives. These meetings helped identify several issues that needed further research and resolution. Several items were brought up that required further review. A memo was prepared with recommendations as to whether these items should be incorporated into the model. After feedback was received on the memo, an updated version of the model was prepared. The updated model incorporates the following changes:
1) Two errors were found in the original model and corrected. One was an overcalculation of sales taxes in the first year, the second was an undercalculation of property taxes for new residents who have indirect jobs;
2) All census and community information was updated, as needs to be done on an annual basis;
3) The calculation of the benefit-cost ratio was changed. The ratio had been calculated as if the firm did not need the incentive. It is now calculated as if the firm does need the incentive (further discussion on this issue can be found in the follow-up memo, along with Vice-Mayor Chestnut’s response);
4) In November, Lawrence residents approved three new sales taxes for the next ten years that will add .55% to the local tax rate. These new sales taxes were added as “sunset” taxes- that is, they expire after several years (in the 87 acre test case, we assume the firm will not be operational until 2010, so we only count these sales taxes for 8 total years);
5) Interest rates were updated due to large changes. This resulted in lower mortgage rates for home purchases and a lower discount rate for future revenue and cost streams;
6) Sales tax revenues generated from construction of the facility was added to the model;
7) Population numbers were adjusted. There was a double-count in the way that new “persons” in the community were accounted for in the first version of the model. Specifically, a person who both lived in Lawrence and worked in Lawrence would count as two “people.” This meant that the person would generate twice as much revenue and cost as someone who, for example, lived here but did not have a job. Two alternative population counts were developed and tested;
8) Revenues and expenses in the General Fund budget that are not related to population growth were removed;
9) Additional Funds that are related to population growth were incorporated into the revenues and expenses; and finally
10) A slight adjustment was made to income calculations to match supplemental income (for example, earnings from dividends and interest, self-employment) more closely to wages from primary jobs.
Results
Of these ten changes, two of them (items 2 and 10) had minimal impact on the results. The adjustment to the benefit-cost calculation (item 3) had no effect on the revenues and costs. The change in interest rates (item 5) resulted in a greater increase in revenues than costs. Part of the reason for this has to do with the specific example tested, and might vary under other scenarios. The addition of city and county revenues and expenses outside the General Fund (item 9) raised costs more than revenues. Three of the remaining five items (items 1, 4, and 6) resulted in more revenues than costs, although the overall impact of each on the benefit-cost ratio was modest.
The population calculation (item 7) determines the number of unique jobs and residents, and thus how much additional costs and revenues will be generated by these people. As mentioned, there was a double-count in the original model. Removing the double-count reduces the number of new “persons” in the community substantially. When combined with the elimination of non-population sensitive items in the budget (Item 8), this results in a significant impact on the model.
Staff has since conducted additional meetings with City and County employees in order to ensure the model accurately portrays how new employees, commuters and residents impact the City and County budgets. The results of these discussions resulted in slight changes to the benefit-cost ratio, as well as an increase to both revenues and costs to each budget.
Action Item
Staff believes that the benefit-cost model is now ready to be used for evaluation of incentives, and recommends that the City Commission adopt the model for future use.