That average is a false equivalency since the cost subsidy will be localized to the regional power provider typically the state-run monopoly or some other large provider. It's not an even distribution and the core thesis of the article, that the lower-income folks are the ones subsidizing it, likely still holds because land is historically cheaper and is generally less developed by municipal and state govts when its residents are lower-income and/or a minority population, and local distribution (and costs) impact those with less (usually no) disposable income by which they can improve their household energy efficiency.