The URL can be used to link to this page

Home My WebLink About9/8/2026 - Revenue Projection Process and Forecasting Model. , Revenue Projection Process and Forecasting Model Board of Supervisors Work Session September 8, 2026 IZOA~,'.'.,~'. ~~'.',~,~I Y Roanoke County Revenue Projections • Analysis by Finance and Management Services Staff • Judgmental forecasting based on trends • Dashboards • Different models for different revenue sources • Focus analysis efforts on harder to predict revenues • Approximately 135 different revenue sources • Virginia Finance Network • Revenue team approach Revenue Forecasting Process • Analyze prior year revenue trends and monitor monthly revenues • Review economic indicators • Consult Finance Network resources • Revenue Team meetings (December -February) • Analyze possible state code changes • Real Estate and Personal Property Assessments • Present preliminary revenue outlook to the Board of Supervisors • Make final adjustments before Proposed Budget FY 2027 General Government Revenue Collection Summary Category Date(s)/Period of Collection Real Estate 1st Half in December (CY 2026), 2 nd Half in June (CY 2027)* Personal Property Receive in May-June, billed annually Sales Tax Receive monthly, 2 months delay Business License Receive in February-March annually Current Public Service Receive in November -December annually Corp Meals Receive monthly, 1 month delay Hotel/Motel Tax Receive quarterly, some received monthly * 1st Half of Real Estate based on CY 2026 Assessment, 2 nd Half based on CY 2027 Assessment finalized in January 2027 % of Total Budget 50.5% 17.2% 5.9% 3.4% 2.9% 2.4% 0.9% Current Forecasting Methods • Use of Excel statistical tools and judgmental forecasting methods • 2 month and 12 month moving averages for revenues collected at regular monthly intervals • Perform regular monthly analysis and update budget projections throughout the year informing revenue forecasting for upcoming year • Use of Al tools for revenue forecasting including ChatGPT, Perplexity, and Claude Virginia Finance Network • Began during COVID to share information related to CARES Act and ARPA • Information shared from various organizations including: • ICMA • Virginia Association of Counties • Virginia Municipal League • GFOA and VGFOA • Networking between local governments in Virginia on common issues Revenue Team Departments Represented Revenue Team Strengths • Obtain input from revenue generating departments • Subject matter experts on various aspects of the local economy and items that affect revenue collection • Broad perspectives on economic trends and legislative impacts • Consensus approach to revenue budget • Considerations of Board of Supervisors preferences Roanoke County Results • Comparing original budgeted revenues to actual revenues, Roanoke County has done well compared to other localities in Virginia • Roanoke County took a very conservative approach during COVID and the following high inflation years • Revenue growth is slowing and returning to normal • While revenue growth is slowing, expenditure growth continues at a higher level Caleb Eng Finance Intern -Budget Division I grew up in Roanoke, VA attending Faith Christian School. I now study at Northwestern University where I major in Economics and Mathematical Methods in the Social Sciences. I also minor in Math and am working towards a certificate in Financial Economics through the Kellogg School of Business. I am interested in pursuing my PhD in Economics. I also serve as a research assistant to Robert J. Gordon, a well-respected macroeconomist. Content Overview • As an intern, I have been focusing on projecting the revenues from year to year • This presentation details some of the challenges involved with doing so as well as some of the methods for forecasting • These methods and challenges were all considered in my revenue forecasting model, which I have created over the course of this internship Challenges with Revenue Forecasting • This graph displays one method of projecting the 315 revenue. • The main point is that projections are inherently uncertain, we can do the best we can to try to establish a baseline estimate, but any forecast carries uncertainty.· 295 275 255 235 215 195 Historical and Projected Revenue ($M) 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 95% Prediction Interval 80% Prediction Interval 50% Prediction Interval • To illustrate how wide an accurate forecast can be: • For 2027, • Forecast Value: 287.0M • 50% chance: 282.7 -291.3M • 80% chance: 279.0 -295.3M • 95% chance: 274.8 -299.8M • Note: this is just one projection method, economic indicators suggest we will likely be below the forecast value. Revenue Growth Rates (%) Real and Nominal 2026 2027 95% Prediction Interval 80% Prediction Interval 50% Prediction Interval Inflation Also Presents a Challenge • Historically, our budget growth has been driven very heavily by inflation, however in FY 2022 & FY 2023 we had a period -of unusually high inflation followed by FY 2024 & FY 2025 which saw high growth in real dollars. • The unusually high real growth rates over the past few years could be reversed. 2026 is evidence of this. • Some real growth is simply delayed inflation growth (partially attributable to real estate assessment timing) Real Growth Rate of Budget (%) 8 (growth rate without inflation; adjusted for budget code changes; log growth) 7 6 5 1 0 2017 2019 2020 2021 2023 2024 2025 2026 -1 -2 ■Real 8 7 6 5 ~ 4 QI .. Ill ~ 3 i 0 ~ 2 1 0 -1 -2 2017 Growth Rate of Budget {%) (adjusted for budget code changes; log growth} 2019 2020 2021 2023 ■ Real ■ Nominal 2024 2025 2026 8 7 6 5 1 0 -1 -2 I 2017 Growth Rate of Budget (%) and Inflation (%, PCE) (adjusted for budget code changes; log growth) 2019 2020 2021 2023 2024 ■ Real ■ Nominal D Inflation 2025 2026 18 ROA~'~;~~~ ~'.1,~,~TY General Methods for Forecasting Revenue ROA~,'.'.,~,; ffli ~:",\',~ I Y Method Overview Over the course of my internship with Roanoke County, I have explored a few primary methods for projecting the revenues. For funds that exhibit semi-stable growth I consider the following: • Average Historical Growth Rate (seen earlier) • Moving Average Growth Rate • Autoregression • Exponential Smoothing Historical Average Growth Rate • Takes the whole average growth rate of the budget, or budget line-items and applies it forward. • Pros: • Good baseline • Easy to calculate na·ive confidence intervals • Cons: • Does not factor more current information Moving Average Growth Rate • Takes some of the most recent revenues and computes their growth rates and applies it forward • Pros: • captures trends • Cons: • trends reverse, 2026 is evidence Autoregression • Looks at the most recent growth rate and moves it closer to the average at a rate determined by historical growth • Pros: • Considers recent and past information • Cons: • Takes more data than we have to be highly accurate Exponential Smoothing • Estimates a growth trend and level by weighting more recent information more heavily • Pros: • captures trends while still accounting for the whole history • Cons: • trends reverse Model Accuracy Using an appropriate, predetermined, weighted average of methods for FY 2025 and FY 2026 and correcting with only information that would be available at the time we obtain the following: FY Actual ($M) Forecast ($M) Error ($M) Error(%) 2025 266.0 268.6 2.6 0.97 2026 275.2 274.5 -0.7 -0.26 This is not a definitive range of plausible errors but should provide context for how well the math can and cannot do. Forecasting Real Estate • It is very rare, but sometimes we can use one index to predict a budget line-item • Given assessment timing, it 25 20 15 is possible to use this fact to 10 project assessment values and therefore revenue from 5 rea I estate tax • This approach is still error prone (average +/-2.01% on assessments) -5 -10 Case-Schiller Housing Price Index (Blue) vs Total Assessment Value (Orange),% Growth Rates 2010 2012 2014 2016 2018 2020 2022 2024 2026 Questions & Comments