Inheritance Reporting in the Health and Retirement Study Data : Evidence of Forward Telescoping

Intergenerational transfers are often under-reported in household surveys (Laitner and Sonnega, 2010). This paper demonstrates that over-reporting can be an issue as well. Using data from the 1992-2008 Health and Retirement Study, a biannual longitudinal survey, we find that the first autocorrelations of inheritance reports are substantially higher than autocorrelations at longer lags. This evidence is consistent with over-reporting due to forward telescoping. We estimate that, of all inheritances reported in a survey wave, 5 to 10% had been already reported in the prior wave. This results in a similar magnitude overestimation of total two-year amounts of inheritance transfers. Additionally, this reporting error affects regression estimates, when inheritance measures are used as independent variables. In an example of regression analysis, we find that parameter estimates of inheritance variables are biased toward zero by about 2% to 12%. To mitigate forward telescoping in inheritance reports, it is advisable to employ bounded/dependent interviewing in longitudinal surveys