Testing Gender and Race Non-linear Unemployment Invariance Hypothesis in South Africa: Evidence from Threshold Regression

Abstract

This study employed South African data from 2008Q1 to 2021Q1 and a combination of OLS, FMOLS and threshold regression to test the validity of the existence or absence of a nonlinear Unemployment Invariance Hypothesis across race and gender, with the goal of determining whether the relationship between the unemployment rate and labour force participation rate is dependent on the unemployment regimes. The threshold regression results revealed that the relationship between unemployment and labour force participation varies by regimes. In other words, the impact of unemployment on labour force participation varies by gender and race and depends on the state of unemployment. In most cases, the discouraged worker hypothesis dominates, particularly at high unemployment rates (beyond predicted thresholds). This indicates that the relationship between the labour force participation rate and unemployment across different categories is dependent on the characteristics of the South African labour market, which is marked by structural and cyclical unemployment with structural inequities across races, sectors and groups. As a result, a persistently high unemployment rate in the South African labour market may lead to lower labour force participation, exacerbating the existing unemployment problem and eventually leading to permanent unemployment. Given our findings, the South African government should pursue substantial labour reforms that include educating and retraining existing unemployed workers, as well as creating an enabling environment conducive to job creation.