A Novel Sampling Strategy for Surveying High Net‐Worth Individuals—A Pretest Application Using the Socio‐Economic Panel

• Author: Rainer Siegers,Martin Kroh,Charlotte Bartels,Carsten Schröder,Johannes König,Markus M. Grabka
• DOI: http://doi.org/10.1111/roiw.12452
• Published date: 01 December 2020
• Date: 01 December 2020

© 2019 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of

International Association for Research in Income and Wealth

825

A NOVEL SAMPLING STRATEGY FOR SURVEYING HIGH

NET-WORTH INDIVIDUALS—A PRETEST APPLICATION USING THE

SOCIO-ECONOMIC PANEL*

Carsten sChröder

SOEP at DIW Berlin and Free University Berlin

Charlotte Bartels

SOEP at DIW Berlin

Markus M. GraBka

SOEP at DIW Berlin

Johannes köniG

SOEP at DIW Berlin

Martin kroh

SOEP at DIW Berlin and University of Bielefeld

AND

rainer sieGers

SOEP at DIW Berlin

High-wealth individuals are typically underrepresented or completely missing in population surveys.

The lack of comprehensive national registers on high-wealth individuals in many countries challenged

previous attempts to remedy this under-representation. In a novel research design, we draw on public

data on the shareholding structures of companies as a sampling frame. Our design builds on the empiri-

cal regularity that high-wealth individuals are likely to hold at least part of their assets in the form of

shareholdings. Based on data from over 270 million companies worldwide, we select all individuals

who are both German residents and registered shareholders of companies. In a pretest, we interviewed

124 households from a gross sample of 2,000 anchor persons. Our analysis shows that values of share-

holdings from register data highly correlate with individual ranks in the wealth distribution, that the

quality of personal information, particularly the residential address, is sufficiently high for subsequent

interviewing, and that the approach can fill a major data and research gap in the study of high-wealth

individuals.

JEL Codes: C83, D31

Keywords: sampling method, wealth, top wealth

Note: We would like to thank the German Federal Ministry of Labour and Social Affairs for finan-

cial support; Bureau van Dijk for technical support; Paul Brockmann, Konstantin Göbler, Christoph

Halbmeier, Fabian Nemeczek, Christopher Prömel, Axel Ramstein, and Thomas Rieger for research

assistance; Deborah Anne Bowen and Adam Lederer for editing.

*Correspondence to: Carsten Schröder, SOEP at DIW Berlin, Mohrenstr. 58, 10117 Berlin,

Germany (cschroeder@diw.de).

Review of Income and Wealth

Series 66, Number 4, December 2020

DOI: 10.1111/roiw.12452

This is an open access article under the terms of the Creat ive Commo ns Attri butio n-NonCo mmerc ial-

NoDerivs License, which permits use and distribution in any medium, provided the original work is

properly cited, the use is non-commercial and no modifications or adaptations are made.

bs_bs_banner

Review of Income and Wealth, Series 66, Number 4, December 2020

826

© 2019 The Authors. Review of Income and Wealth published by John Wiley & Sons Ltd on behalf of

International Association for Research in Income and Wealth

1. introduCtion

Wealth concentration is on the rise in many developed and developing coun-

tries (Alvaredo etal., 2017). In light of the crucial importance of wealth in modern

economies and its high concentration, the Commission on the Measurement of

Economic Performance and Social Progress (Stiglitz-Sen-Fitoussi report) and the

G20 Data Gaps Initiative emphasize the need to improve information on the dis-

tributions of wealth, income, and consumption.

Indeed, there are major deficits in the available data on high-wealth individ-

uals and households, here defined as the top 1 percent, in many countries. Many

countries do not impose a common general tax on wealth holdings and, thus, lack

population registers with detailed wealth information. There is also evidence that a

great deal of the wealth of the wealthiest is hidden in tax havens and, as such, unob-

served in administrative data (Alstadsaeter etal., 2018). The resulting limitations

of register data not only impedes drawing substantive conclusions in empirical

studies on the entire wealth distribution, but also consistently estimating inequal-

ity or concentration indices like the Gini coefficient or quantile ratios, which are

sensitive to the inclusion of the richest households in the sample. The introduction

of dual taxation of labor and capital income in many European countries since the

1990s exacerbated the data problem.

Household surveys are another potential data source. Such surveys allow

researchers to link wealth holdings with the broad set of surveyed background

variables in domains like employment, family composition, psychology, health,

education, etc. Yet, it is unlikely, by definition, that high-wealth individuals end

up in a probabilistic sample of a few thousand individuals. Furthermore, the indi-

vidual willingness to participate in surveys declines systematically with increasing

assets (see D’Alessio and Faiella, 2002; Sànchez Muñoz, 2011; Westermeier and

Grabka, 2015), reducing the share of high-wealth individuals in surveys to levels

even below their actual share in the population. While the low sampling probabil-

ity of the wealthiest reduces the precision of statistical estimates, the non-random

nonresponse introduces bias.

Administrative register data can help in oversampling wealthy households

in scientific surveys. Successful examples that rely on tax data include the Survey

of Consumer Finances (conducted by the US Federal Reserve Board), Encuesta

Financiera de las Familias (Bank of Spain), and Enquête Patrimoine (Bank

of France/INSEE) (Vermeulen, 2018). Following these positive examples, the

Eurosystem’s Household Finance and Consumption Network (HFCN) made

attempts to oversample relatively wealthy households. Yet, administrative data

for an appropriate sampling frame was unavailable for all Euro-area countries

except France and Spain (see Finances and Network, 2013). As an alternative,

oversampling individuals by regions with above-average tax returns, as exempli-

fied in the PHF Survey of the German Bundesbank, was implemented, but did

not substantially improve the representation of high-wealth individuals in the

survey.

In sum, there is a major data gap regarding the representation of the high-

wealth group in many countries. Figure 1 depicts this data gap, taking the German

Socio-Economic Panel (SOEP) as an example.The SOEP is an ongoing panel