For some services, it is impossible to estimate PPPs solely from price data collected through within-country surveys. In most countries, there are no markets or weak and imperfect markets for education, medical services, general government services, and services of non-profit institutions. PPPs must be estimated from a combination of national data on expenditures on these services, some locally collected prices, and national and cross-country datasets on the quantity and quality of inputs and/or outputs of services.
There is general agreement that to the extent possible parities in the areas of health, education and general government should employ prices, expenditures deflated by output, or productivity adjusted wage rates for well specified types of employees. This involves price comparisons in the health sector, and somewhat different treatment of private and publicly provided services. There are major quality issues in price (bed-day rates), or input cost (doctors and nurses salaries) for hospitals across countries. For education regions have used student measures of output, numbers of teachers as input/output, and teacher salary by level and education as parities, sometimes adjusted for accompanying school capital. Finally, general government has most often used salary comparisons by detailed occupational category as parities for government, often modified by estimates of productivity differences. Two papers were commissioned to review past practices and provide proposals for the future. Earlier works were also revisited to help shed light on comparison resistant services.