Just to add to that, our surveys are cross-sectional. We ask people at the same time whether they are employed or not and we assess their mental health. On slide 9, you can see that a lower proportion of people who declare a disorder are employed, but the statistics don't allow us to do the kind of thing you are suggesting: to show whether you return to better mental health once you have a job. It would require longitudinal data where we know a person's state before employment and measure their state after employment, but that we don't have.
On January 31st, 2017. See this statement in context.