Unemployment: Blame the Victim?

This paper investigates a problem of 'state dependency'. People who have been unemployed in the past are much more likely than others to become unemployed in the future. But is it unemployment itself that causes future unemployment or is there something else, some measured or unmeasured form of heterogeneity, in individual characteristics or environmental circumstances, that causes both the past and present unemployment? There is an extensive economic literature on this issue which comes to no absolutely firm conclusions as to the existence of state dependency in this case. We present a simple sociological model, of the 'recursive determination' of the employment state. We estimate that time invariant personal characteristics not included in the model (or left altogether unmeasured) can only play a small part in the determination of unemployment. Which implies in turn that the association between successive periods of unemployment must be mostly the result of acquired or other time-varying characteristics (though we cannot tell whether these relate to the individual or to circumstances in his or her social or geographical environment). Non-technical summary People with recent histories of unemployment are more than averagely prone to becoming unemployed. So does unemployment cause future unemployment? James Heckman first raised this question in 1979, yet today it remains unresolved and the focus of many papers. The unresolved problem concerns the issue of 'state dependency'. Is it some aspect of the experience of unemployment itself that causes future unemployment? Or are there some measured or unmeasured individual characteristics (or environmental circumstances)-what economists describe as 'unmeasured heterogeneity'-that influence the probability of experiencing unemployment. The economic literature on this subject has developed various econometric models addressing the general question of the balance of causation across these numerous factors. The focus of this paper is on a more limited question. Is it possible that some long-established personal characteristics are responsible for an individuals proneness to unemployment making it legitimate statistically speaking to 'blame the victim'? We aim to provide a reasonably conclusive answer to this question using panel and retrospective lifetime employment data collected by the British Household Panel Study (BHPS) in successive years since 1991. iv The BHPS evidence allows us to construct accounts of individuals' employment circumstances and recent history for each month since entering the labour force. However, because of the problem of under-reporting of unemployment occurrences in the more distant past, we just focus on unemployment spells experienced since September …