From 2009 onwards, the slowdown in French real wages was less acute than that in labour productivity, which pulled French firms'margin rate down, falling to its lowest level since 1985. How can this resilience of wages be explained? Econometric models of wages show that it is usual for a slowdown in labour productivity not to be passed on in full to wages, but also that this mechanism is not sufficient in itself to explain the recent disconnection between real wages and productivity. Two potential explanatory hypothesis are sometimes proposed for this disconnection: a change in the labour force structure driven by greater job losses among lower-paid workers, and downward nominal wage rigidities for those employees who have not changed companies. At the end of 2008 and in 2009, the fall in employment was accompanied by a change in the labour force structure. With a decline in the proportion of less qualified employees, job separation among those earning the lowest wages thus contributed to buoying up the average wage. Excluding employed population structure changes during the crisis, average wage would have grown by 0.7 points less each year between 2009 and 2012. However, the rise in the qualification level of the working population was already making a positive contribution towages of 0.4 points per year prior to the recession. The effect of the change in composition due to the recession therefore seems quite weak. In parallel, almost no downwardwage rigidities were observed. For those persons remaining in the same firm, a significant amount of wage dropped, while wage freezes are rare. In addition to this, those enterprises that suffered a specific fall in their activity in 2009 adjusted wages downwards and these reductions were greater in scale, on average, than the price rises granted by companies whose activity improved. Nominal rigidities therefore cannot explain the inertia in wage dynamics, as shown by a macroeconomic analysis of wages.