One of the basic principles of public finance is that it makes no difference whether a tax is legally borne by the buyer or seller. The burden of the tax will depend on the relative elasticity of supply and demand, and the economic incidence of a tax doesn't depend at all on the legal incidence. Stephen Gordon did a post discussing this issue, and provided a nice set of graphs for comparison:
In this case most of the burden of the tax falls on buyers, in the form of higher prices. But his post also shows the opposite case, which is more likely to apply to labor markets (where sellers of labor, i.e. workers, are believed to absorb most of the burden of a payroll tax, whatever its legal incidence (which is 50-50 in the US.))
The same is of course true of subsidies, which are merely negative taxes. If there is no minimum wage then it makes no difference whether you subsidize workers $4/hour by directly paying them that sum of money for each hour worked, or paying the company that hires them that sum of money for each hour worked. This is a point on which all well-informed economists agree, whether on the left or the right. (And of course the whole point of the proposal is to offer an alternative to the minimum wage.)
Perhaps this is all obvious, but I mention these facts because several commenters have told me that many "on the left" are opposed to subsidies on low wage employment, like the one proposed here. They believe it subsidizes firms who pay low wages. I've favored this policy since grad school, and still strongly believe it is the best way of reducing economic inequality. It is similar to the Earned Income Tax Credit, with two possible advantages---ease of preventing fraud, and better incentives in terms of the decision on how many hours to work. With an hourly wage subsidy, low wage workers would always be rewarded for working harder, something not true of the EITC.