I’ve been thinking a lot about forecasting lately, mostly because I’m writing a dissertation about cost and ridership forecasts that are prepared for public transit projects.
I made a little list of things we do in transportation planning and engineering that rely on forecasts:
- Figuring out whether an investment in infrastructure is likely to be cost effective. This is at the top of my list because it’s what my dissertation is about. Particularly for investments in transit, we want to know how many people will be riding the system in the future and how much it will cost to serve them.
- Figuring out if we need more transportation capacity. This is closely related to the first item. If trains or highways or bike lanes or parking lots are pretty close to being full, then we might want to increase capacity somehow (note: this is one approach to reducing congestion, but it is not the only one, nor is it usually the best one). But that’s only if demand is going to stay as high as it is now. And it might not.
- Timing traffic signals. Many traffic signals can adjust their timing based on sensors or other real-time information (#smartcities). But others have fixed timing plans based on typical traffic patterns. This is a forecast. The assumption is that what is typical today will continue to be typical until we get a chance to reprogram the signal.
- Charging traffic impact fees. In many cities, when a new development goes in, the developers pay a fee based on how much traffic their development will generate. Or rather, the fee is based on a forecast of how much traffic the development is expected to generate. You don’t know for sure until it’s open for business.
- Charging transportation utility fees. I’ve written about these before. Again, the fee is based on how much traffic you expect a property to generate. Which is a forecast.
- Setting rates for other user fees. Tolls, transit fares, and even fuel tax rates are all based, at least in part, on how many people we expect to pay them, and how that number will change when the rate increases or decreases. This is a forecast.
Some of these forecasts involve more uncertainty than others. Some aren’t much better than a wild guess. Others have so little uncertainty that they hardly seem like forecasts at all (either because the are predicting something that is very well-understood or because they don’t look very far into the future). But they all involve some uncertainty and risk of wasting something – public money, private money, or the traveling public’s time (and time is, after all, money).
There are ways to reduce risk (for example, by getting better at forecasting), but we can’t eliminate risk. So a major part of transportation planning (and all planning) has to be about contingency planning and risk management.