GDP measures all activities taking place on the market place, and thus neglects and production in the home that never hits the market. With the increase of female labor participation, and consequently the increase of production of goods previously produced in the household, GDP growth exhibits so bias if it is supposed to measure total production in an economy: part of the growth stems from an accounting change. Affected are goods like meals, daycare, laundry services, cleaning services.
Christopher House, John Laitner and Dmitriy Stolyarov show that this bias is negligible and accounts only for a cumulated 2.5% of GDP, or 25% of women's measured earnings: 2.5% of today's GDP would have been produced at home fifty years ago.
There is no easy ways to figure these numbers out, as home production is not measured. The trick used here is to build a micro-founded model with home production, looks at the impact of changes in home production on other variables, measure the latter and reverse engineer the model to find what home production ought to be. There are clearly lots of margins for errors (model specification, calibration, measurement of observables), but the exercise is very carefully done and the results are strong.