Forecasting Transportation Revenue Sources: Survey of State Practices

Transcription

1 NCHRP Project Synthesis Topic Forecasting Transportation Revenue Sources: Survey of State Practices Martin Wachs Department of Urban Planning, UCLA Benton Heimsath HR&A Advisors, Inc.

2 2 Revenue Forecasting Financial forecasts provide baseline information needed for transportation planning and system management.

3 3 Purpose of the Synthesis To help practitioners improve methods of forecasting and to assist DOTs as they explore ways to enhance future revenue, including the development of new methods of revenue generation Questions: How do states forecast transportation revenues? Which states have the most sophisticated forecasting tools? Is that sophistication helpful or important? What are states doing to account for innovative (non-traditional) transportation revenue sources?

4 4 Transportation Funding Context State transportation revenues are stagnant and declining. Raising motor fuel tax is politically challenging. The federal gasoline tax is unchanged since Gasoline consumption has declined each year since Vehicle Miles Traveled Gas Consumption with Increased Efficiency VMT Growth Revenue Loss

5 5 Background Fuel shocks in 1970s led to an interest in accurate forecasting of future revenues. Main variable examined was demand for gasoline (VMT, fleet fuel efficiency) Regression models using socioeconomic factors such as income, employment, population.

6 6 Surface Transportation Funding By Revenue Source 21% 36% Federal State Local 43% (Source: AASHTO Center for Excellence in Project Finance)

7 7 Elements of the Synthesis Information on current standard methods and new developments in revenue forecasting by state DOTs How states project revenue from both traditional and new and non-traditional sources What issues they have faced implementing new revenue generating mechanisms and approaches, and how those issues were addressed

8 8 Research Methods Used Review of the literature recent academic research reports, articles in professional journals, agency newsletters National survey of state DOTs 47 of 52 DOTs responded (87% response rate) Telephone interviews of state officials Reviews of documents describing state practices.

9 9 Example Survey Questions Does your agency produce long-term revenue forecasts, or are they produced by another entity? Which revenues are typically projected? Has your agency changed its methods recently? What are the major challenges your agency faces?

10 10 Responses: 44 states and DC 87% response rate Respondents were economists, planners, financial analysts, transportation commissioners, etc. Yes 13 No 31 Unsure 1

11 11 Types of Revenues Forecast Revenue Type # of States % State gasoline per gallon taxes 43 96% State diesel fuel per gallon taxes 42 93% State vehicle registration and/or license fees 39 87% Federal funds allocated to the state 39 87% State sales taxes designated for transportation 23 51% Allocations of state general revenues 15 33% Other 14 31% Tolls 9 20% Local (county or regional) sales and property taxes 6 13%

12 12 Approaches to Conducting Forecasts Consensus of Experts Simple Extrapolation Econometric Regression

13 13 Interesting Observations Not Anticipated State DOTs reported that they see themselves as operating agencies, which refrain from making forecasts for the purpose of influencing policies or suggesting new options to legislators or governors. If the state wants to evaluate the consequences of a new revenue source, they should not ask us, but rather should rely on legislative staff, governor s staff, or consultants DOT revenue forecasts of revenue are more associated with revenue management and operational planning than with strategic innovation

14 14 Other Agencies Involved CO: General Assembly Legislative Council Staff and Office of State Planning and Budgeting CA: Department of Finance DC: Office of Tax and Revenue + Bureau of Transportation Statistics to estimate fuel efficiency FL: Twice-yearly Revenue Estimating Conference for 10 yr forecast, members include Legislature, Dept of Revenue, Dept of Highway Safety, DMV GA: Cambridge Systematics does long term forecast that is used in the development of Statewide Transportation Plan LA: Revenue Estimating Conference for short-term, simplified growth assumptions for long-term forecast

15 15 Econometric Regression Majority of state respondents used some form of econometric analysis. Simple linear regression vs. time series analysis Often used to estimate quantities, such as gallons of gasoline consumed, which are then multiplied by taxes/fees to determine revenue forecasts.

16 16 Washington The Washington gasoline forecast is derived from a regression model that estimates per capita gasoline consumption using 10 independent variables. Quarterly Model Equation The equation for gasoline consumption in Washington is defined as ln (Gas) = α + φln(wa_emp) + φ(wa_gasp*eff) + δ(dummy) + ε Where 1. Gas= Quarterly gross gas consumption from Treasurer Reports (log), 2. WA_Emp= Quarterly Washington non-farm employment(log), 3. WA_GasP*Eff = Quarterly Washington gas prices* US average fuel efficiency(log)lagged 2 quarters, 4. Dummy = Quarterly dummy variable for periods of severe oil shortages. 5. ε = Stochastic disturbance on gasoline consumption. Independent variables are able to explain most of the variations in gasoline consumption.

17 17 Sources of Data for Washington DOT s Econometric Models Variables WA personal income Population Inflation (2 measures: CPI and IPDC) Employment Oil price index Fuel efficiency U.S. sales of light vehicles U.S. fuel prices (retail gas and diesel and index of petroleum products) Sources Based on the Washington Economic and Revenue Forecast Council in short-term (through 2017) based on forecasts from Blue Chip average US GDP growth rates and NYMEX fuel prices; and long-term Global Insight forecast Preliminary Office of Financial Management population projections Washington Economic and Revenue Forecast Council for short-term and Global Insight forecast for long-term WA non-ag. employment; WA TTU, WA retail trade and national unemployment rates 2014 Global Insight forecast 2014 short and long-term Global Insight forecast 2014 Global Insight forecast and November 2013 long-term Global Insight forecast3 EIA for short-term and Global Insight for long-term

18 18 Washington - Motor Vehicle Fuel Consumption Forecast, June 2013 Source: Transportation Revenue Forecast Council June 2013 Transportation Economic and Revenue Forecasts

19 19 Washington - Motor Vehicle Fuel Tax Revenue Forecast, June 2013 Source: Transportation Revenue Forecast Council June 2013 Transportation Economic and Revenue Forecasts

20 20 Missouri In 2006, MODOT commissioned a private consulting firm to address decreasing accuracy in revenue forecasts. New proposed model used six independent variables (up from four), a more sophisticated risk analysis process and time-series regression models. The framework is no longer used by MODOT staff. Takeaway: More complexity does not necessarily result in more useful or accurate models.

21 21 Accuracy Measurement Several states reported that they commonly performed retrospective analysis to determine forecasting accuracy. Vermont Emergency Board Consensus Update to Fund Forecasts Staff Recommended Consensus Forecast Update Difference from July 2013 Forecast Dollars Percent Dollars Percent Dollars Percent General Fund $ % -$ % -$ % Transportation Fund $ % $ % -$ % Education Fund $ % -$ % -$ % Total $ % $ % -$ % Source: Revised Revenue Outlook (dollars in millions)

22 22 Interesting Excerpts from Responses AK: All our forecasting techniques are based on assumptions. CT: Purposely very conservative. DE: Forecasts have been very accurate, except in recessionary periods. IN: 5 year forecast each year within 98% match. LA: The economists that provide forecasts are more interested in the state general fund forecast and sometimes do not fully analyze the factors affecting the forecast of gasoline tax.

23 23 Many States Use Takedowns from Federal Agency Forecasts Federal Revenue Forecasting Model (FRFM) maintained by FHWA - a fairly simple multiple spreadsheet models manually updated using some data series which may be out of date. Twenty classes of vehicles, 36 weight groups many revenue types; limited public documentation available National Energy Modeling System (NEMS) by Energy Information Agency (EIA) of the Department of Energy (DOE). More complex and better documented; includes a specific transportation module.

24 24 Minnesota: A Blended Approach Minnesota s DOT s Fuel Consumption Inputs, Source: Minnesota s DOT 2014 Transportation Funds Forecast (2014).

25 25 Conclusions There is little evidence of attempts to forecast innovative, non-traditional revenue sources More research is needed to share knowledge and develop state-of-the-art forecasting econometric tools for state DOT needs State forecasting needs vary widely in terms of sophistication and breadth States should proactively develop forecasts to inform policymakers about the revenue implications of new revenue sources

26 26 Potential Initiatives DOT staff responsible for revenue forecasting in some states regularly consult with those having similar responsibilities in other states, but many other state forecasters do not. There is insufficient knowledge to lead us to suggest that states adopt standard methods. Information sharing among the states could be improved.

27 27 Potential Initiatives A regularly updated website on transportation revenue forecasting, perhaps maintained by AASHTO s Center of Excellence in Project Finance A blog to facilitate exchanges between relevant staff and different state agencies Regular sessions on transportation revenue forecasting at AASHTO and TRB Annual Meetings

28 28 For More Details Please See Our Report NCHRP Synthesis Report 479 Forecasting Transportation Revenue Sources: survey of State Practices Thanks for your attention It s time for Questions, Comments, and.

29 29 Contact Us Martin Wachs Benton Heimsath