Graph #1: The number of jobs in social science and humanities research has been contracting for a decade, according to Bureau of Labor Statistics data. I don't know this specific labor market well. Does anyone have an explanation?
Graph #2: Looking at General Social Survey data, I found that the answers to one question -- "How long have you worked in your present job for your current employer?"-- create a useful economic dataset on employee turnover in American labor markets. What we see from a sample of 4,645 is pretty amazing: a constant 11% annual rate of attrition which holds strong from roughly two years out until retirement. (Before two years, the annual rate of attrition is slightly higher.) If you were to re-plot the same graph in a semilog scale, what you see is a very tight "linear" fit. You can see, also, the periodic discontinuities at multiples of five: that's likely because people are rounding. (Note: I removed the data point at 39 years, which had zero respondents and broke the exponential fit in Excel. Beyond 40 years, the sample size is very small, so I excluded that from the fit.)