US labour market statistics for July suggest that the recession has ended, for two reasons. First, the (smaller) fall in payroll employment last month was offset by a rise in the length of the average working week. Aggregate hours worked in the private sector were therefore unchanged – the first month not to register a fall since last August. Since labour productivity has continued to rise during the recession, stability in hours worked implies an increase in output.

Secondly, the unemployment rate – derived from a survey of households rather than employer payrolls – fell slightly in July. Historically, a monthly decline following a sustained steep increase has marked the end of a recession – see chart. This is true even when the rate subsequently rises to a new peak, as it did in the aftermath of the 1990-91 and 2001 recessions.

Some commentators have dismissed the significance of the fall in the unemployment rate on the basis that it reflected a contraction of the labour force rather than a rise in the number of households in employment. This may be unwise. Historically, the first monthly decline after a significant peak has often been associated with a fall in the labour force, e.g. in 1975, 1980, 1982 and 2003. Moreover, a measure of payroll jobs derived from the household survey increased, by 70,000, last month.

The improvements in both the employer and household surveys are consistent with other labour market evidence, including a recent large decline in corporate layoffs (see here), falling initial unemployment claims, a stabilisation of help-wanted advertising and less negative employment readings in business surveys.