A policy changes the choices inside the model
An equation can trace a direct effect cleanly: raise a fee, and—if everything else stays fixed—the cost of an activity rises. But people do not stay fixed. They compare alternatives, respond to one another, and sometimes change the very conditions the policy was meant to influence.
Consider a new charge for driving into a busy area. Some commuters may pay it; others may shift their travel time, take transit, share a ride, or choose a different route. Employers might adjust schedules, and shops may see different patterns of customers. The final result depends not just on the charge, but on which options people have and how they value them.
Small changes can trigger different responses
People facing the same policy may make different choices because their budgets, schedules, habits, and access to alternatives differ. Those choices can also interact. If many drivers move to a side street, it may stop being a useful shortcut. If enough people change routes, the payoff for everyone else changes too.
This makes prediction a problem of responses and feedback, not arithmetic alone. A model must represent how people choose—and how those choices alter one another. Often, the response is not known precisely in advance, so analysts compare plausible assumptions rather than treating one forecast as certain.
The forecast is part of the policy question
That uncertainty does not make equations useless. Models can clarify which outcomes depend on which assumptions and identify what to watch after a policy begins. The practical test is whether the predicted pattern appears in real choices—and whether the model needs updating as people adapt.
A policy’s effect is therefore more than what it directly costs or offers. It also includes the new routes, routines, and decisions people build around it.
