Are Labour Markets Segmented in Developing Economies? A Clustering Approach for Colombian Workers

Labour markets in developing economies are usually thought to be segmented. Differences in productivity, red tape, and high taxes create a divide between a modern and an excluded traditional sector. More recently, some scholars have challenged this view. In this article, we propose to test the segmented markets hypothesis using a clustering method applied to Colombian workers. Following Anderson et al. (1987) we hypothesize that if the first view prevails, the labour market has well-defined worker clusters that our empirical strategy could uncover. Using the FAMD-K-means algorithm we find three clusters: one comprises half the workforce, has workers with secondary education or vocational training, without labour contracts, and median earnings slightly above the minimum wage. The second group comprises 37% of the workforce, older workers with even lower earnings and educational achievement, with more precarious jobs. The last cluster comprises good quality jobs, mostly with indefinite labour contracts, with workers with university degrees and median earnings close to four times the minimum wage. We statistically tested the differences between the informality definition and our method and found that the traditional measures have an important correlation with the clusters resulting from our model.