As part of a broader goal to better inform policy makers and researchers of the key trends in labour markets that ultimately determine the status and well being of the world's workers, the Trends Team produces estimates of labour market indicators such as unemployment, employment by sector and status, labour productivity and working poverty, among others, with many estimates disaggregated by age and sex. These estimates are utilized in ILO publications, including the Key Indicators of the Labour Market (KILM), the Global Employment Trends and related reports, and a number of regional trends reports. The data are also utilized in reports from other agencies, including many related to the Millennium Development Goals (MDGs).
Trends Econometric Models
The biggest challenge in the production of aggregate estimates is that of missing data. In an ideal world, producing world and regional estimates of labour market indicators, such as employment, for example, would simply require summing up the total number of employed persons across all countries in the world or within a given region. However, because not all countries collect and/or report data in every year and, indeed, some countries do not report data for any years at all, it is not possible to derive aggregate estimates of labour market indicators by merely summing across countries.
To address the problem of missing data, the ILO Trends Unit has designed and actively maintains three econometric models which are used to produce estimates of labour market indicators in the countries and years for which no real data exist. The Trends Labour Force Model is designed to produce estimates of labour force participation rates and the economically active population. The Global Employment Trends Model is used to produce estimates – wherever possible disaggregated by age and sex – of unemployment, employment-to-population ratios, status in employment, employment by sector, labour productivity and employment elasticities. Finally, the estimates of the working poor around the world come from the Trends Working Poverty Model. The models are collectively referred to as the “Trends Econometric Models”.
Each of these models uses multivariate regression techniques to impute missing values at the country level. The first step in each model is to assemble known pieces of real information (i.e. every real data point) for each indicator in question. Only data that are national in coverage and comparable across countries and over time are used as inputs. This is an important selection criterion when the models are run, because they are designed to use the relationship between the various labour market indicators and their macroeconomic correlates (such as per capita GDP, GDP growth rates, demographic trends and country and time dummy variables) in order to produce estimates of the labour market indicators where no data exist. Thus, the comparability of the labour market data that are used as inputs in the imputation models is essential to ensure that the models accurately capture the relationship between the labour market indicators and the macroeconomic variables.
The last step of the estimation procedure occurs once the datasets containing both the real and imputed labour market data have been assembled. In this step, the data are aggregated across countries to produce the final world and regional estimates.
For a detailed description of the methodology used in the Global Employment Trends Model, please consult the following paper:
- Trends Econometric Models: A Review of Methodology (ILO, Geneva, 2010)
For further information on the Trends Econometric Models, readers can consult the following technical background papers:
- S. Kapsos, J. Pasteels and V. Bourmpoula: ILO Estimates of the Economically Active Population 1990-2010 (Sixth Edition) (ILO, Geneva, 2007)
- S. Kapsos: World and regional trends in labour force participation : methodologies and key results (ILO, Geneva, 2007)
- S. Kapsos: Estimating growth requirements for reducing working poverty : can the world halve working poverty by 2015? (ILO, Geneva, 2004)
- G. Crespi: Imputation, estimation and prediction using the Key Indicators of the Labour Market (KILM) data set (ILO, Geneva, 2004)
- W. Schaible and R. Mahadevan-Vijaya: World and regional estimates for selected key indicators of the labour market (ILO, Geneva, 2002)
The world and regional indicators produced from the Trends Econometric Models serve as input for numerous ILO products, chief among which are:
ILO regional reports, for example:
- Labour and social trends in ASEAN 2008: Driving competitiveness and prosperity with decent work (see also previous editions)
- African Employment Trends, April 2007
- Realizing decent work in Asia: report of the Director-General, Fourteenth Asian Regional Meeting, Busan, Republic of Korea, August-September 2006
The ILO also collaborates with numerous other organizations to produce the world and regional estimates of labour market indicators that meet their specific analytical needs. Reports by other international organizations that make use of ILO estimates are: