Key points are not available for this paper at this time.
The shape of the earnings-age function and the way education affects it can explain all the empirically observed relations of migration measures to distance moved, age, and education. No other assumptions (which are often made)--such as attributing lower risk aversion and higher efficiency of job search to more-educated persons, or higher psychic cost of moving to older persons--are needed. The individual's search for jobs and the firm's search for employees are affected by the shape of the earnings function--and, thus, by age and education in a predictable way. In turn, the observed systematic behavior of the migration measures with respect to age and education is responsive to the search behavior.
Aba Schwartz (Sun,) studied this question.