Number |
CREPEDP-115 |
Publication Date |
January 2022 |
Title |
Are Japanese Wage Statistics Representative?
|
Author |
Daiji Kawaguchi and Takahiro Toriyabe
|
Abstract |
This paper assesses the representativeness of two major government wage surveys of Japan: the Basic Survey on Wage Structure (BSWS) and the Statistical Survey of Actual Status for Salary in the Private Sector (SSPS). We examine whether the two-step procedure that involves random sampling of establishments and random sampling of workers within selected establishments ensures the representativeness of sampled workers in the population. We find lower response rates for establishments operating in the service industry, with fewer employees, and located in urban prefectures, but no relationship between the response rate and past wages. The mean wages calculated from the sampled individual payroll records coincide with establishment-level aggregate records, indicating that workers are randomly selected within establishments. Overall, we only find evidence of non-random sample selection based on observed characteristics of establishments, thus the selection is considered to be ignorable and the unbiased mean wage can be estimated with an appropriate weighting. We also discuss coverage issues due to the sampling design.
|
Keywords |
Wages, Random Sampling, Comparison, Coverage, Non-response, Bias
|
Other information |
Paper in English (52 pages) |