<P> Two commonly cited sources of income inequality data are the CBO and economist Emmanuel Saez, which differ somewhat in their sources and methods . According to Saez, for 2011 the share of "market income less transfers" received by the top 1% was about 19.5% . Saez used IRS data in this measure . The CBO uses both IRS data and Census data in its computations and reported a lower "pre-tax" figure for the top 1% of 14.6% . The two data series were approximately 5 percentage points apart in recent years . </P> <P> Pioneers in the use of IRS income data to analyze income distribution are Emmanuel Saez and Thomas Piketty at the Paris School of Economics showed that the share of income held by the top 1 percent was as large in 2005 as in 1928 . Other sources that have noted the increased inequality included economist Janet Yellen who stated, "the growth (in real income) was heavily concentrated at the very tip of the top, that is, the top 1 percent ." Follow - up research, published in 2014, by Emmanuel Saez and Gabriel Zucman revealed that more than half of those in the top 1 percent had not experienced relative gains in wealth between 1960 and 2012 . In fact, those between the top 1 percent and top . 5 percent had actually lost relative wealth . Only those in the top . 1 percent and above had made relative wealth gains during that time . </P> <P> The comparative use of Census Bureau data, as well as most sources of demographic income data, has been questioned by statisticians for being unable to account for' mobility of incomes' . At any given time, the Census Bureau ranks all households by household income and then divides this distribution of households into quintiles . The highest - ranked household in each quintile provides the upper income limit for each quintile . Comparing changes in these upper income limits for different quintiles is how changes are measured between one moment in time and the next . The problem with inferring income inequality on this basis is that the census statistics provide only a snapshot of income distribution in the U.S., at individual points in time . The statistics do not reflect the reality that income for many households changes over time--i.e., incomes are mobile . For most people, income increases over time as they move from their first, low - paying job in high school to a better - paying job later in their lives . Also, some people lose income over time because of business - cycle contractions, demotions, career changes, retirement, etc . The implication of changing individual incomes is that individual households do not remain in the same income quintiles over time . Thus, comparing different income quintiles over time is like comparing apples to oranges, because it means comparing incomes of different people at different stages in their earnings profile . </P> <P> Gary Burtless of the Brookings Institution notes that many economists and analysts who use U.S. census data fail to recognize recent and significant lower - and middle - income gains, primarily because census data does not capture key information: "A commonly used indicator of middle class income is the Census Bureau's estimate of median household money income . The main problem with this income measure is that it only reflects households' before - tax cash incomes . It fails to account for changing tax burdens and the impact of income sources that do not take the form of cash . This means, for example, that tax cuts in 2001 - 2003 and 2008 - 2012 are missed in the census statistics . Furthermore, the Census Bureau measure ignores income received as in - kind benefits and health insurance coverage from employers and the government . By ignoring such benefits as well as sizeable tax cuts in the recession, the Census Bureau's money income measure seriously overstated the income losses that middle - income families suffered in the recession . </P>

In which form does most of the average american family’s wealth exist