It is common practice in the literature to compute labor flows from data on stocks.
To use these flows in standard search models, it is assumed that the economically relevant
movements occur between employment and unemployment. If there are significant flows
between labor force participation and inactivity, ignoring the participation decision can lead
to biased results. This paper shows that while with three states it is impossible to identify
all the flows from publicly available data on stocks, partial identification is possible with
the help of data on unemployment duration and job tenure. A new method is described,
which allows the computation of the transition probabilities that are most relevant from a
macroeconomic perspective. The method is easy to use, and the paper describes the detailed
steps for its implementation to potential users.