| Popis: |
The Estate Multiplier Technique is an important example of augmentation of a probability sample with a non-probability sample, which is used in practice. The Statistics of Income Division (SOI) of the IRS has for many years used this technique to produce estimates of wealth for the living population from a sample of estate tax return data (ETD) collected primarily for program administration purposes.(1, 2) A Federal estate tax return must be filed within 9 months of death for every U.S. decedent whose gross estate equals or exceeds the applicable filing threshold of $675,000 in gross assets for 2001.(3) All of a decedent’s assets, whether owned solely or jointly, and debts are reported on the return. These estimates, although limited by the estate tax filing threshold, provide coverage of the wealthiest 1 to 2 percent of the population that controls between 25 and 30 percent of total U.S. personal wealth. Estimates of the wealth holdings of the living population are derived by applying a multiplier to the ETD (Mallet 1908). The multiplier m is equivalent to a sampling weight where the probability of selection includes both the probability r of being a decedent and p being included in the SOI sample. Multipliers are calculated separately based on relevant characteristics such as age and gender, denoted by the index a = 1, ..., A, and for each SOI sample strata, denoted by the index i = 1, ..., I: mai = 1/(ra * pi). Estimated total wealth W = ∑ waimai (Atkinson and Harrison 1978, p. 23). In practice, the probability that a person will die in a given year is not random; it is conditional on factors such as age, sex, and family history. For the wealthy, additional factors, such as access to better health care, better nutrition, and less hazardous occupations, also play a role (Attanasio and Emmerson |
