Modelling income data with exogenous measurement factors

With the increasing use of administrative data for analysis it is necessary to understand possible measurement artefacts that can arise from the way the administrative data is collected and recorded. One possibility is when the period covered by the data does not conform to a standard time period. Data collected on a weekly or fortnightly basis but reported monthly is a common pattern, the supply of employee earnings data to the New Zealand tax department being an example. In this case the month to month changes in the reported time series do not reflect true changes in monthly earnings. It is possible to extend time series models to identify the pay period used by a business and potentially adjust the data to take account of this measurement effect.