Citizen Portal
Sign In

Get Full Government Meeting Transcripts, Videos, & Alerts Forever!

Get email alerts on the Revenue Forecast Sales Tax topic

No spam. Unsubscribe anytime.

City's revenue forecasting team outlines neural-network sales-tax model and warns one-time funds won't close 2026 gap

Denver City Council Finance & Business Committee · October 14, 2025
AI-Generated Content: All content on this page was generated by AI to highlight key points from the meeting. For complete details and context, we recommend watching the full video. so we can fix them.

Summary

Denver budget office presented its sales-tax forecasting approach using a custom neural-network model, reported 2024 collections were 98.8% of budgeted estimates, and said one-time revenues that helped stabilize 2025 (about $36M) are largely absent from the 2026 outlook, leaving the city to rely on expenditure reductions and reserve strategies.

Denver's Budget Management Office briefed the Finance & Business Committee on Oct. 14 about the city's revenue outlook for 2026, explaining modeling methods, one-time revenue uses in 2025 and the five-year forecast and reserve strategy.

Chief economist Lisa Martinez Templeton described the city's consumption-based sales-tax model, a custom program written in Python that uses monthly tax collections at six-digit industry detail, local foot-traffic indicators, tourism and hotel occupancy data, macroeconomic variables and special-event adjustments. Templeton said the model is paired with judgmental adjustments and scenario analysis to produce a baseline and optimistic/pessimistic scenarios.

Budget Management Office Director Justin Sykes said the city maximized one-time revenues in 2025 (just over $36 million), including $22 million transferred from the border crisis response fund and $6 million from the Stapleton Reserve Special Revenue Fund, but projects only about $5 million in one-time sources for 2026. He warned the city cannot rely on one-time transfers to the same degree next year and described proposed structural expenditure reductions to bring the operating budget closer to balance.

Sykes also walked the committee through the five-year operating forecast and the city's reserve policy. He said the city anticipates roughly $1.664 billion in 2026 general fund revenues, with more than two-thirds coming from sales and use tax and property tax. The presentation explained the reserve policy bands: amounts above 15% of expenditures can be used for one-time items, 10—15% allows stabilization during below-average growth, and below 10% is reserved for severe emergencies. Staff currently project 2025 year-end reserves around 10.3% and a proposed 2026 target of about 11%.

Council members questioned modeling accuracy and vendor provenance. Templeton said she developed and runs the Python-based neural network in-house and noted that the city collected 98.8% of budgeted core sales-and-use-tax estimates for 2024, meaning model errors were around 1.2% in that comparison.

Council members also raised aging fees (for example, the right-of-way vacation fee set at $1,000 in prior years) and requested a holistic review of fee schedules to determine where adjustments could be made, with sensitivity to equity impacts. Finance staff agreed to pursue an off-season review of top revenue sources and fees and to provide a breakdown of what funded the border crisis response fund and where those dollars were reallocated.

The committee closed with follow-up requests for additional detail on one-time transfers, fee-update proposals and clarifying assumptions in the five-year forecast ahead of the budget vote next month.