Economists sometimes cite tax breaks to be equivalently inefficient as subsidies, as they incentivize behavior in a similar way. This is often used as an argument against the tax breaks on employer based health insurance.
Unfortunately the logic here is incorrect. The effect of the tax break on behavior patterns may be identical to the subsidy, but the source of inefficiency is different. A partial equilibrium analysis can be used to see this.
Say we have goods A and B, with the average person’s preferences being A ≻ B ≻ [nothing] at market prices. This means people strictly prefer A to B to not buying anything. With no taxes (and therefore no tax breaks) or subsidies, people will buy A over B. If we apply a subsidy on B, then people will buy good B even though they would normally prefer good A. This is the same outcome as if we had applied a tax on just A, and therefore a tax break on B. However, in the latter case the inefficiency is coming from the tax on A, not the tax break on B.
A tax on A and on B will cause people to substitute to nothing. Removing the tax on B will cause them to substitute back to B, which is a more efficient outcome. While the shift away from A is still inefficient, it is of course less inefficient than a non-discriminatory tax pushing everyone to buying nothing, since moving from nothing to B increases utility.
If A and B are untaxed and a subsidy is provided for B, then people will shift their consumption to B, which is just as inefficient as when A was taxed. However, the change here is different, we went from full efficiency (no taxes) to inefficiency (B is subsidized).
Just as removing the subsidy will increase efficiency, removing the tax increases efficiency also. This is because tax breaks themselves aren’t inefficient, taxes are. So while saying that a tax break on B will shift behavior towards B, which is an inefficient outcome, it is misleading because the inefficiency isn’t coming from the tax break on B, rather it is coming from the tax on A. If there were tax breaks on everything, then there would be no taxes, and that would provide maximum efficiency in all markets.
Sources: consultation from professor
P.S. This is a partial equilibrium argument. There are arguments which use general equilibrium which state that the inefficiency from the tax is offset by solving some other inefficiency elsewhere, such as provision of some non-excludable good. So the pro-tax economic argument is that the inefficiency of the tax is lower than the inefficiency of leaving some other good underprovided. Most of the time when people are talking about taxes, tax breaks, and subsidies, they are talking in terms of partial equilibrium, so the argument against tax breaks is broken in that context. The general equilibrium argument is different and actually holds logical weight.
Updated 2020-06-05 (cleaned things up a little)