Risk events and learning from error: when are assessments of the risk of unemployment revised?

Abstract

Data from the British Household Panel Study are used to investigate the effect of changes in personal circumstances on employed panel members' assessment of the risk of unemployment. Adverse personal changes (‘risk events’) are shown to increase the likelihood of people, who felt safe from unemployment, to re-assess their risk and to become risk ‘pessimists’ one year later. Similarly, panel members who initially feared they might lose their jobs re-assessed their risk and became risk ‘optimists,’ if they experienced a number of (positive) risk events. After controlling for risk events, ‘error of judgment,’ which measures the accuracy of the initial risk assessment, was only independently significant for panel members who had judged themselves to be safe from unemployment, but had nevertheless lost their jobs. Risk re-assessments among panel members who had initially felt their jobs were insecure were significantly influenced by the time passed since re-employment, often exceeding the survey period. Overall, panel members who changed their initial assessment of their risk of unemployment had typically experienced a greater number of risk events than those who did not change their assessment.

Notes

¹Thus accounting for inflationary and other incremental adjustments.

Percentages may not add to 100 due to rounding.

¹More accurately the odds describe the increased likelihood of respondents experiencing a specified risk event or displaying a given characteristic revising rather than not revising their risk assessment compared to the reference group revising rather than not revising their risk assessment. This complex wording is avoided here.

Significance levels (2-tailed test) *10%, **5%, ***1%, ****0.1.

Note: This table only shows the statistically significant variables, after removing non-significant variables from the analysis.

²The constructed risk event variable ‘change in health conditions,’ by contrast, did not significantly affect risk re-assessments.

Significance levels (2-tailed test) *10%, **5%, ***1%, ****0.1.

[ ] indicates absolute case numbers reported due to small total number of cases in sample.

³The exclusion of these variables, while changing the frequency counts, did not change the general findings of this analysis.

⁴In the case of incorrect pessimists, the risk events involving changes in employment contracts were retaining or gaining a permanent contract, which had a positive or neutral effect on the odds of an optimistic risk re-assessment.

**Difference in means significant at 1% level; * difference in means significant at 5% level. Statistical significance tested using Analysis of Variance (ANOVA) test function of Means comparison in SPSS. Figures in round bracket refer to maximum numbers applicable to the respective sub-group of optimists and pessimists and the mean percentages based on these maximum numbers.