Properly measuring income inequality is difficult, and not just because figuring out income from all sources is hard. The problem is that market income is not the relevant measure. So-called home production matters as well. When a spouse is not not getting a wage on the market but is providing services at home that have a market value, these services should be counted as income (and consumption). Of course, this is difficult to measure, so a short-hand trick could be to find how home production is related to market income.
One could expect home production to be negatively related to market income, because of the substitution of market time for home time. Harley Frazis and Jay Stewart find, however, a very weak correlation. They argue that one should rather treat home production like a constant. They come to this conclusion by using the American Time Use Survey, from which the value of home production can be estimated. Luckily, it also has income data. Frazis and Stewart use a flexible form (a Fourier series expansion) to allow any non-linearities to be captured.
This means that to accurately measure income inequality, one needs to add a constant, $8000 to $16000, depending on the type of household and whether secondary child care is included or not. This implies that trends in income inequality are not affected by home production, and we can continue to use market income for inequality analysis, as long as the composition of household types does not change.