The Effect of Raising the Minimum Wage on Employment : Causal Inference Bootcamp

The Effect of Raising the Minimum Wage on Employment : Causal Inference Bootcamp


[MUSIC] Next we are going to look at one of
the most famous examples of difference-in-differences. What is the effect of the minimum wage on
employment? If you pass the minimum wage, are more people get jobs or do people get
fired and employment goes down? Well that’s exactly what Card and Krueger looked at in
their 1994 study. Here’s the data, and here’s the main table
– let’s see what they found. Well, in New Jersey, in 1992, the minimum wage was increased
from $4.25 to $5.05. So what they did is they surveyed fast food restaurants in New Jersey
before the minimum wage law was passed, and after it was enacted. So here’s New Jersey. “Full-time employment
before, all available observations.” So 20.44 people on average worked in a fast food
restaurant in New Jersey. So here they say, “each part-time worker counts as half of
a full-time worker.” So that’s what a full-time equivalent means: we’re counting part-time
people as half of a full-time worker. So about 20.5 people worked there before the minimum
wage was increased. Then after the minimum wage was increased, employment went up to
21 people. So it increased by about half a worker, so that’s what the before and after
comparison suggests: that the minimum wage law increased employment by half a worker.
But as we’ve discussed before, maybe there’s some other trends, some other variables that
are changing over time that would explain this this difference of 0.59 workers. So what
are we going to do? We need to find some control group we can get the common trend from, and
extract out the common trend here, so that we are left with just the effect of treatment,
the effect of a minimum wage law. So what Card and Krueger did is they also
surveyed stores in neighboring Pennsylvania. They surveyed fast food restaurants there
before and after the minimum wage law changed in New Jersey. So in Pennsylvania, the minimum
wage was also 4.25 an hour, but it was not increased. So there, the stores did not get
treated in either time period. So before the New Jersey law was passed, in Pennsylvania
the average full-time equivalent employment was 23 people per store. And then after the
law was passed, it went down to 21.17 people per store, for a difference of 2.16 people. So that suggests that the overall trend in
employment in fast food restaurants was going down. Maybe they’re introducing new technologies
or whatever that are causing stores to need less workers in general. So overall trend
was going down, but in New Jersey, employment actually increased. So that suggests that
if we look at the difference between the two differences of New Jersey’s change before
and after, and then subtract off the common trend from Pennsylvania, we get this number
here: our difference-in-difference estimate of the causal effect of the minimum wage on
employment: 2.76. So that suggests that the minimum wage actually
increased employment in fast food restaurants by almost three people, which is quite a big
effect. And if you look at the standard deviation here, shown in parenthesis, and you use this
and our rule-of-thumb to get a confidence interval, you find that this effect is also
actually statistically significant. So, not only do we have a large, positive
effect on employment of the minimum wage, but it is statistically significant. That
right there, this number is the main finding of the Card and Krueger 1994 paper. Now this
paper is extremely famous because this isn’t what people usually expected the minimum wage
to do. Most people would have thought that the minimum wage would lower employment, would
mean that it’d be harder for people to get jobs, and this finding is exactly the opposite.
So, as you might expect there’s still a huge debate about this policy, and what’s the best
thing or what’s the actual effects of these policies, and this paper is one of the things
that kickstarted an enormous literature on this topic. So if you’re interested in this policy and
this topic, definitely I encourage you to read more. That’s the basic finding they use
just in application of difference-in-differences. [MUSIC].