This variable was measured by one item indicating the extent of the delegation of HRM practices to line ministries in the central government. Data were collected through the 2010 OECD Survey on Strategic Human Resources Management (OECD 2011). Respondents were senior officials in the central government’s HRM departments, and the information referred to HRM practices in the central government. The index is composed of the following variables: the existence of a central HRM body and the role of line ministries in determining the number and types of posts within organizations; the allocation of the budget between payroll and other expenses; staff compensation levels; position classification, recruitment, and dismissals; and conditions of employment. The values of this variable are on a percentage scale.
This variable was measured using the production costs in the general government as a percentage of GDP in 2009. We suggest that this measure is a good indication of the resources that were available to the government at that time. Furthermore, it incorporates the GDP, which is a macroeconomic variable that may influence outputs and outcomes. The concept and methodology of production costs build on the existing classification of public expenditures in the System of National Accounts (SNA) (OECD 2011). Specifically, government production costs include (1) the compensation costs for general government employees, including cash and in-kind remuneration plus all mandatory employer (and imputed) contributions to social insurance and voluntary contributions paid on behalf of employees, (2) the costs of goods and services produced by nongovernment entities paid for by the government (these include goods and services provided to both government and citizens), and (3) the consumption of fixed capital (depreciation of capital). The data include government employment and intermediate consumption for output produced by the government for its own use, such as roads and other capital investment projects built by government employees. The values of this variable are on a percentage scale.