https://www.rand.org/content/dam/rand/pubs/technical_reports...
> First, depending on how congestion pricing is implemented, it can be either regressive or progressive. This depends in large part on how toll revenues are used. For instance, if regions spend revenues in ways that benefit low-income individuals, congestion pricing is more likely to be progressive. However, if regions use revenues in a way that benefits all individuals equally, congestion pricing may be, overall, regressive. This is the strongest finding in the economic literature.
> Second, even when low-income and other transportation-disadvantaged groups benefit as a whole from congestion pricing, it is very likely that some individuals will still be worse off.
> Third, for all forms of congestion pricing (but more for some than for others), the distri- bution of residents and job opportunities (not to mention shopping, schools, places of worship, and other important destinations) has a large impact on the equity implications.
> Fourth, high-occupancy toll (HOT) lanes, the most common form of congestion pricing in the United States, tend to raise fewer equity concerns among motorists, since they provide drivers with an additional choice of using a set of priced lanes while allowing them to continue using parallel, free lanes if they prefer.
> Fifth, while congestion pricing has been shown to reduce emissions in general, there is scant evidence showing that congestion pricing can specifically reduce negative environmental consequences for neighborhoods disproportionately affected by emissions.
> Given the risk of negative impacts to low-income and other groups under congestion pric- ing, we looked at suggested ways to diminish these impacts. Two mechanisms are in common use: (1) revenue redistribution and (2) discounts and exemptions.