It’s remarkable to see a conservative and a democratic socialist unite in condemning major American companies. Both Senator Bernie Sanders and Fox News host Tucker Carlson have recently slammed Amazon, Walmart, Uber, and other large companies for paying workers and contractors too little. In a withering monologue last week, Carlson claimed that the companies are all effectively subsidized by the taxpayer because many of their employees’ incomes are supplemented by various federal welfare benefits, such as food stamps. Sanders agrees. Yesterday, he introduced legislation (the so-called Stop BEZOS Act) to tax large corporations one dollar for every dollar their workers receive in government food stamps or health-care benefits.

If nothing else, it is amusing that neither Sanders nor Carlson fully acknowledges the logical implications of their position. If Sanders is right that programs such as food stamps modestly subsidize employers who pay low wages, then his hugely expensive Medicare-for-all and free-college-tuition proposals would constitute a massive subsidy to low-wage employers. If Carlson truly believes that large firms have the power to suppress wages below competitive rates, then he should support raising the minimum wage to combat that power — something that he has, in the past, sensibly advocated against.

Snark aside, the pair are simply wrong on the economics of the matter, and shortsighted to boot. An employer’s responsibility is to pay employees for the work they do, not to ensure that they have some societally agreed-upon level of livable household income. Indeed, it is a peculiar worldview that suggests that, when setting wages, a company employing low-skilled workers should ignore the value of the tasks the employee actually undertakes for them.

In competitive labor markets, we usually assume that firms pay workers according to their productivity, the marginal revenue product of their labor. Market wages are determined by where this demand interacts with the supply of workers. Firms can’t underpay workers without losing the best to rivals. Nor can they routinely pay employees for more than they add to company revenue without losing capital to rivals at home and abroad and risking going out of business. There is no evidence that Amazon, Walmart, or Uber have high-enough degrees of labor-market power that they are the single hirer of workers in any one geographical area. For Carlson to imply that their pay rates are evidence purely of corporate greed is the worst form of populism.

There is a basic conundrum hanging over this debate: In a world with no minimum-wage laws, no out-of-work benefits, and no in-work benefits, some workers with low productivity levels would obtain work but find it difficult to live comfortable lives on market income. The real questions then are: Who should help, and if it is the government, will that end up subsidizing firms?

One form of help comes in the form of means-tested programs that apply regardless of work status, such as basic food stamps. These explicitly do not subsidize employers as Sanders and Carlson allege. Actually, we’d imagine that transfers of this kind would have the opposite effect, because they replace income obtained from work: The more you earn, the less in transfers you receive. These programs therefore reduce the supply of workers, by raising the wage people would have to be offered to return to work, which in turn raises market wages if the supply of workers is upward-sloping. Far from a “subsidy,” then, means-tested federal welfare benefits are more like a “tax on employers.”

Indeed, the only forms of welfare that can theoretically subsidize employers through lower wages are transfers that supplement income from work and so increase labor supply, such as the earned-income tax credit (EITC). As a wage subsidy, the EITC encourages more potential workers to seek low-paying employment, because the earnings from that employment plus the subsidy are higher than the means-tested benefits they forfeit by going back to work. The EITC thus increases the labor supply by design, which is great for EITC recipients but can hurt ineligible groups, such as those without children, who see their wage rates fall as a result.

The benefits of such supplemental subsidies are indeed captured by a combination of the employer and all employees. Academic experts Auston Nichols and Jesse Rothstein, using reasonable assumptions, estimate that for every $1 put into the EITC program, employees receive $0.64 of the outlay and employers capture $0.36 owing to reduced market wages. One therefore could, if he were so inclined, call the EITC an “employer subsidy,” though strangely neither Sanders nor Carlson has bothered to mention it at all.

Given that Sanders’s and Carlson’s critique focuses instead on means-tested welfare programs, it makes no sense. Cajoling companies to pay more by imposing high minimum-wage rates or taxing those whose employees receive government assistance would simply make it more difficult for lower-productivity workers to find jobs, putting taxpayers on the hook for more safety-net spending. Only supplementary welfare programs such as the EITC have the effect that Sanders and Carlson describe, and cutting these programs would certainly hurt the workers who rely on them as much as, if not more than, employers.