Abstract:
The US Survey of Consumer Finances (SCF) and the Eurosystem’s Household Finance and Consumption Survey (HFCS) provide evidence that wealth is heavily concentrated at the upper tail of the wealth distribution. A commonly cited number for the US is that 1 percent of the households hold 30 percent of total household wealth. I investigate the reliability of upper tail wealth estimates from household wealth surveys in the presence of survey differential non-response, i.e. the fact that richer households have lower response rates than poorer households. Differential non-response can often not be remedied by adjustment of survey weights, as wealth of the non-responding households remains unobserved. Differential non-response biases tail wealth estimates downwards. Monte Carlo evidence shows that such a bias can be quite substantial. I provide a method that greatly reduces the bias. The method combines survey data with data from rich lists and uses them jointly to estimate a Pareto (power-law) distribution for tail wealth. Using this method, the paper combines the SCF and HFCS data with Forbes World’s billionaires data to provide new estimates of tail wealth. For surveys with low or no oversampling of the wealthy, these estimates tend to indicate a higher concentration of wealth at the top than those calculated from the wealth surveys alone.