But in surveys of employers taken just before and after changes
in the minimum wage, David Card and Alan Krueger (1995) showed that
this just isn’t true.
The law passed in 1996 (and was probably known about some time before that), but didn't actually raise the minimum wage until during 1997. If the study is measuring employment just before and after the raise, it's going to miss any changes that happened in anticipation of that raise. Studies of employment differences over longer periods, of course, run into the problem of other possible factors affecting results. Economics is a pain that way: we can't really run controlled experiments.
Using the example of the fast food restaurant, a minimum wage worker receiving a 10% increase will sell about 100$ of merchandise with a cost increase of less than 1$. The cost will be passed on to the customer. The store has many more customers than employees, and the customers are quite likely in the range of people whose wages are affected by the MW increase. (That effect is not limited to minimum wage earners, but it does vanish as you approach the average industrial wage.) So it's hard to argue that the price increase will reduce the sales, and that means that staff reductions are unlikely since fast food outlets aren't prone to overstaffing.
So we have a theoretical model that predicts that employment will not decrease (that's my theory and what it is too) and empirical evidence that it does not. Therefore: raise the minimum wage QED.
Not so fast. First off, fast food restaurants have fairly inelastic labor demands, but that doesn't mean that the economy as a whole does. And second, quite frankly, there's another consideration which gets consistently overlooked by the left: minimum wage laws (and rent control) make racism easier.
In a free market, racism hurts not only the person being discriminated against, it hurts the person doing the discriminating too: if you pay white employees more than black employees, you're wasting money on payroll. You'd save money by either paying the white employees less, or by hiring black employees instead.
But what if you have to pay everyone more than their labor is really worth, but you can't afford not to hire them? Well, in that case, discrimination is free. You'll get more job applicants than you would at lower free-market wages, and you can hire white employees and not hire black employees, and there are no consequences to you as an employer for making that discriminatory choice.
Same with rent control: if you are prevented by law from charging more than a certain amount for an appartment, then you'll have more people wanting the appartment than you would in a free-market environment. That means you can discriminate for free. But if the price isn't set by law, then if you don't rent to the highest bidder (black OR white), then the choice to discriminate costs you.
So we should be considering more than just employment and wage levels here. Tampering with the market also encourages discrimination by removing the economic costs associated with discrimination in a free market.