While the unemployment rate is the single most widely used statistic for providing a snapshot of labour market conditions, there is considerable controversy over its measurement. The International Labour Organization (ILO) proposes three tests: to be counted as unemployed an individual must be without work, currently available for work and seeking work.

The first and second criteria are not controversial; the third, however, is subject to debate. To satisfy the third criterion, the individual must take specific actions that qualify as ‘search’ in a given reference period: the longer the period for satisfying the search criterion, the higher the unemployment rate.

Leaving this aside, the ILO recognises that the condition of ‘seeking work’ may not be meaningful in countries where the labour market is largely unorganised, where job search can operate through non-standard channels, and where the self-employed constitute a large segment of the labour force.

Our research addresses the limitations of the standard unemployment definition and the three-state labour market classification on which it rests: employment, unemployment and non-participation (Özkan-Değirmenci and Tunalı, 2014). We analyse data from the Household Labor Force Survey collected by the Turkish Statistics Institute for the period 2000-02.

One way of assessing the adequacy of the unemployment definition is to compare transition rates from non-participation and unemployment to employment. If we find that a considerable number of non-participants flow into employment, we might conclude that the three-state classification is not very useful.

The next logical step is to identify sub-groups among non-participants that are responsible for these flows, determine in what ways they are different from the unemployed and, if necessary, augment the unemployment definition and the three-state classification.

In the data for Turkey, employment is typically at its peak in the third quarter (Q3), lowest in the first quarter (Q1) and somewhere in between in Q2 and Q4. This pattern is attributable to seasonality in agricultural activity and tourism services. While nearly 90% of those employed in the first quarter are also employed in the second, about 84% of the non-participants and only 37% of the unemployed remain in their previous state.

Conditional on not being in the labour force in 2002 Q1, 14% of individuals become employed one quarter later. Conditional on being unemployed in 2002 Q1, 38% of them become employed one quarter later. Thus, an individual who is classified as unemployed is 2.7 times more likely to get employed than someone who is classified as non-participant.

But in Q1, the number of non-participants is 11 times higher than the number of unemployed individuals. This indicates that for every unemployed individual who successfully gets a job, there are four non-participants who also successfully make this transition.

Put another way, for every five previously non-employed and newly employed individuals, only one of them was unemployed three months earlier, while four were non-participants. This is an astounding statistic that underscores the hazards of focusing on the unemployed alone, and the inadequacy of the three-state classification of the labour market.

We conclude that there are significant flows from non-participation to employment. In fact, there is abundant evidence that some individuals who did not seek work during the reference period and were therefore excluded from the labour force display some level of attachment to the labour market.

This might arise because during economic downturns, some non-employed individuals give up on search when their efforts fail to produce results. Seasonal economic fluctuations push others outside the labour force, even though they are ready to re-enter when work is available. This leads to the idea of the ‘marginally attached’ – a group of individuals among the non-employed who are the best candidates to be counted as unemployed, above and beyond the official ones.

Identifying these marginally attached individuals is important in the MENA region where the market economy and labour market institutions are still developing. In the absence of a nationwide network of employment agencies, clearly identified channels might not exist for the exchange of labour market information. The presence of large rural sectors exacerbates the difficulties. As a result, the distinction between the unemployed and the non-employed becomes blurred.

To render the marginal attachment notion operational, we first examine sub-groups of non-participants who display elevated mobility into employment. Students emerge as a group worthy of distinction between Q2 and Q3.

Next we examine data on search behaviour. We define the marginally attached as individuals who did not search during the reference period, but did so earlier – behaviour that suggests willingness to work.

We then divide non-participants into three groups: students, marginally attached and others. Finally, we examine transition rates between five different labour market states: the three sub-groups of non-participants we identify plus employment and unemployment.

The empirical question is whether the marginally attached can be combined with the unemployed or with non-participants. Our results convincingly separate the marginally attached from the other non-participants. The line of demarcation between the unemployed and the marginally attached is blurred. The upshot is that the marginally attached individuals deserve as much attention as the unemployed.

What our study shows is that the official unemployment rate conceals identifiable segments of the population who are keen to respond to the employment generation capabilities of the labour market.

Our critical assessment should not be mistaken for a proposal to drop the standard definition of unemployment: that definition is a measure that allows useful international comparisons and should be maintained. Rather, our point is that the standard measure may not be adequate for assessing the degree of labour market attachment that is present in a given economy at a given point of time.

By using additional measures of unemployment that relax one or more of the tests used in classifying non-employed individuals, it is possible to obtain alternate measures. Tracking the different measures over time provides a more accurate sense of labour market developments. For a more elaborate example of this approach, we recommend The Employment Situation news release of the Bureau of Labor Statistics in the United States.

Further reading

Özkan-Değirmenci, Hayriye Özgül, and İnsan Tunalı (2014) ‘Labor Market Mobility and Marginal Attachment in Turkey: Evidence from HLFS, 2000-2002’.