In this collection of essays. Edmond Malinvaud aims at explaining what he learned as a government statistician, particularly with respect to the unemployment problems of the last two decades. The government expert must forecast for diagnosing spontaneous trends or assessing the likely impact of public decisions. Such forecasts rely on a more or less intensive analysis. To understand the main distinction between frictional and disequilibrium unemployment requires a more rigorous conceptual apparatus than is often acknowledged; this leads to a properly defined Beveridge curve playing the major role. The most vexing issue concerns the effect of real wages on the medium term trend of labour demand; it cannot be well grasped without a good understanding of investment, for which the author presents his reference model.