Economic policy has an array of ethical concerns. Social choice overrides these concerns—voters may validly vote to make themselves less-well-off because they are not infinitely-selfish and have a desire to see others be better off—and otherwise we must reason on why a policy is ethical.
Pareto and Kaldor-Hicks
An economic policy change is a Pareto improvement when nobody is worse off, and some people are better off. Economists call this welfare, such that those who are worse off have lower economic welfare, and those better off have greater economic welfare.
If, for example, a new tax policy cost the wealthiest 1% more in taxes but reduced their out-of-pocket costs by 1%, they would be no better off; if at the same time the poor and middle-class paid 1% more in taxes but received services worth more than 1% of their income, they would be better off. This is a Pareto improvement.
Pareto improvement is nearly-impossible and makes almost all economic policy unethical. A number of economists eventually formed a solution called a Kaldor-Hicks increase: if a Pareto improvement is possible through further wealth redistribution, the policy is ethical. For example: if the middle-class are 2% better off and the rich 0.1% better off, and 1% can be taken from the middle-class to make the rich no worse-off and no better off, then the policy is a Kaldor-Hicks increase.
The Income Effect
A Kaldor-Hicks increase does not require taking such action as to produce a Pareto improvement. Doing so has its own ethical concerns, notably due to the income effect: a marginal dollar has more value to someone with less income than to someone with more, and so taking from one income class to give to another does not represent an equal transfer.
For example: taking from the middle-class to offset an earlier tax increase on high-income earners, particularly, is a loss: $1,000 is as valuable to a suitably-rich recipient as $100 to a middle-class earner, and so taking $1,000 from the middle-class earner to offset $1,000 of additional taxes to high-income earners represents a loss of what the middle-class earner would value $900.
Tax Rate Models
Taxes only climb to 100%, and at times reasonable adjustments are imperative. The income effect above creates difficult ethical issues with adjusting high-income tax rates downward.
By setting a tax rate target model and adjusting to redistribute productivity gains, it is possible to sidestep the above ethical issue. If the middle-class become 2% better off through productivity gains and the top tax rate is above the target model, then the middle-class can be left 1% better off after the adjustment.
It is possible to avoid a raise of middle-income taxes entirely. Such a strategy has a stronger ethical basis than the productivity transfer approach.
Productivity gains imply more real GDP and so more tax revenue relative to government expenses at the same tax rates. If the tax model shows some are overtaxed, then the reduction is taken from this additional relative revenue. Because of the income effect, taxes would be reduced more on overtaxed lower-income earners than on higher-income earners.
This approach allows all taxpayers to keep their productivity gains, and returns government productivity gains to the overtaxed. If some are underserved, then government productivity gains should instead increase available government programs. The better served the population is, the more loosely we can handle economic ethics: each option becomes indistinguishable, and so one cannot say with certainty which is ethical, and we leave it up to the taste of policymakers and their read of their constituencies.