Growing up in Scotland: change in early childhood and the impact of significant events

CHAPTER 5: JOB LOSS

Previous research using GUS data (Barnes et al., 2010) has identified low family level work intensity as a factor with particular bearing on child poverty. Analysis of British Household Panel Survey ( BHPS) data has linked job loss with psychological distress in adults (Thomas et al., 2007) and poorer long-term outcomes for children, with the experience of parental worklessness in early childhood (age 0-5) having the strongest influences on later educational attainment and economic inactivity (Ermisch, 2004). In this chapter we investigate what happens to families when a significant reduction in parents' combined working hours occurs.

5.1 Key findings

• A sustained job loss or substantial reduction in working hours was much more likely among lone parents than couple families.
• Lone parents more likely to experience job loss included younger mothers and those who had older children in addition to the study child and those with poorer physical health.
• Couple families more likely to experience job loss included those living in social rented housing and those living on low income. Among couple families the likelihood of experiencing job loss was in fact lower for those who had older children in addition to the study child.
• Both lone parent and couple families who experienced job loss were more likely to subsequently have a high level of home chaos, low income and high conflict in the parent-child relationship.

5.2 How many families experience job loss?

As explained in Chapter 2, to investigate changes in employment at the family (household) level, we create a measure of Work Intensity Ratio ( WIR). This is based on the average use of household workforce. To assess changes in the level of employment in the household, we calculate wave-on-wave differences in WIR by subtracting WIR in a given wave from WIR in the next wave. Negative values of this measure indicate decrease in the level of household employment and positive values indicate increase. We focus on those families who experienced a wave-on-wave decrease in WIR of at least 0.5, which was not followed by a subsequent recovery (see Chapter 2 for a more detailed definition of this event).

The timing of the event may also be important - a job loss experienced early in a child's life may have different consequences for its outcomes than a job loss experienced later. Table 5.1 presents the distribution of years in which we observed the drop in employment.

Table 5.1 Timing of the job loss event

Age of the child when job loss occurred	%
1-2 years in 2006/07 2-3 years in 2007/08 3-4 years in 2008/09 4-5 years in 2009/10	25 15 12 48
Bases
Weighted	171
Unweighted	167

Base: All families taking part in all five years and experiencing a job loss after the first interview.

Note: Column per cent.

The proportion of families who experienced a job loss when the child was aged 1-2 years was higher than over the two following years. This could partly be because of mothers taking a career break and not returning to employment at the end of maternity leave ¹¹ . However, job loss peaked when the child was aged 4-5. This is at least to some extent due to the way in which we defined the event - a job loss followed by a lack of recovery - as clearly it is easier to observe a recovery for events that occur earlier. Note that we cannot be sure how many of the families who experienced job loss between the last two sweeps of the survey would increase their employment levels over the next year or so. This problem is known in the statistical literature as 'censoring', and we include the timing of the job loss in our statistical modelling to at least partially control for this problem.

The job loss event is equivalent to a single parent losing a part-time job or, in the case of couple families, one parent losing their full-time job. The change is hence substantial and is likely to significantly affect the circumstances of the whole family, including children, particularly because it is sustained over time. The main comparison we carry out is between families who continued to have a high level of employment ( WIR of at least 0.5) in all five sweeps of the survey, and those families who initially had a high level of employment, but then experienced a substantial drop in their employment level, and never recovered after the drop. In the subsequent analyses we will refer to these groups of families as having 'stable employment' and 'job loss' histories respectively.

To put these categories of interest in context, Table 5.2 presents their frequencies relative to other possible employment history profiles: a 'stable lack of employment' profile, that is families with WIR below 0.5 at all five years, and all the remaining families with mixed or unstable employment histories - the profile labelled as 'other' in the table.

Table 5.2 Employment history profiles by family type

	Lone parent families	Couple families	Total
	%	%	%
Stable employment Job loss Stable lack of employment Other	21 13 42 24	82 5 3 9	75 6 8 11
Bases Weighted Unweighted	343 242	2501 2766	2844 3008

Base: All families taking part in all five years.

Note: Column per cent.

Patterns of employment vary considerably according to the family type: 82% of couple families are in the 'stable employment' category, compared with only 21% of lone parent families. Conversely, only three per cent of couple families are in the 'stable lack of employment' category, compared with 42% among lone parents. The event of job loss, as defined in this report, is also more likely to happen in lone parent families (13%) than in couple families (5%). This pattern suggests that the differences between the family types should be taken into account in further analyses. Hence all subsequent descriptive analysis in this chapter presents separate estimates for lone parent and couple families.

5.3 Which families are most likely to experience job loss?

Table 5.3 presents the association between job loss and family background characteristics measured in 2005/06 (when the child was aged 0-1, see section 2.1.3 for full list of characteristics considered). The main findings are that for both lone parents and couple families there is an increased likelihood of job loss among families:

• with younger mothers;
• with mothers with poorer physical health;
• with a main earner in lower occupational class (social class);
• living on a low income; or
• living in more deprived areas.

A number of background characteristics had different associations with job loss for lone parents and couple parents:

• The likelihood of job loss among lone parent families increased with the number of children; such a pattern is not evident among couple families.
• Among couple families, the likelihood of job loss is lowest for owner-occupiers and highest among families renting in the social sector. However, among stable lone parents the likelihood of job loss is highest among those in the privately rented and 'other' accommodation.
• Among couple families, the likelihood of job loss does not vary by the urban-rural classification of the local area. However, among lone parents job loss is more likely in urban areas than in small towns and rural locations.

Highlighted in the tables are those associations which remained significant for either lone parents or couple parents after taking other factors into account (see below).

Table 5.3 Job loss by background characteristics of child and parents

Base: All originally job-rich families taking part in all five years.

Note: Row per cent.

Note: ¹Three least deprived quintiles; ²Two most deprived quintiles.

Note: Shaded cells show characteristics with statistically significant relationships with job loss (separately for lone parent and couple families), after controlling for other factors in multivariate regression analysis.

Note: [ ] Percentages are based on fewer than 50 cases and should be interpreted with caution. * Percentages are based on fewer than 20 cases and are not robust, therefore results are not presented.

Table 5.4 Job loss by background characteristics of household

Base: All originally job-rich families taking part in all five years.

Note: Row per cent.

Note: ¹Three least deprived quintiles; ²Two most deprived quintiles.

Note: Shaded cells show characteristics with statistically significant relationships with job loss (separately for lone parent and couple families), after controlling for other factors in multivariate regression analysis.

Note: [ ] Percentages are based on fewer than 50 cases and should be interpreted with caution. * Percentages are based on fewer than 20 cases and are not robust, therefore results are not presented.

Multivariate analysis was used to identify the factors associated with job loss when controlling for other variables. In this analysis, and for the remainder of the chapter, we focus on those who experienced a job loss and compare these with those in stable employment. Lone parents face different circumstances when combining working with family responsibilities and as noted above job loss is more commonly experienced by lone parents than among couple families, hence we undertook separate analyses for the two family types ¹² . However, any interpretation of the results of the multivariate analysis for lone parents should bear in mind the small sample size - only 92 originally work-rich stable lone parents were included in the analysis. Conducting multivariate analysis on a sample this small affects the power of the tests, meaning certain associations could be missed (see Table C.3 in the technical appendix for full results).

For lone parents the likelihood of job loss was:

• lower as mothers got older;
• lower for families in small towns compared with large urban cities;
• lower for lone mothers with higher physical health scores on the SF-12 scale; but
• substantially higher for families with other children older than the GUS child.

For couple families, the likelihood of job loss was:

• higher for families living in social rented housing, compared with owner occupiers; and
• higher for families living on low income.

In contradiction to the finding for lone parents, the likelihood of job loss for couple families was actually lower for families with other children older than the GUS child.

5.4 What happens to children whose parents experience job loss?

Table 5.5 presents the drivers of child outcomes, measured in 2009/10, for families who experienced a job loss and for those with stable employment histories.

• Home chaos is higher in families that experienced job loss in both lone parent and couple families.
• Income poverty is higher among lone parent families than couple families overall, but in both family types the job loss families have a higher risk of income poverty than the stable employment groups.
• Lone parents that have experienced job loss have a higher likelihood of poor mental health. The difference is not as large among couple families. This may indicate that the support of a partner may have a protective influence following job loss, or it may indicate a more voluntary (or unconstrained) reduction in work intensity among couple families.
• There seem to be some differences between those who experienced a job loss and those with stable employment in terms of the mother-child relationship on both the warmth and the conflict dimension, particularly among lone parent families.

Table 5.5 Drivers of child outcomes by employment history

		Lone parent	Couple	Total
		%	%	%
Home chaos (% high level chaos)	Stable employment	20	29	20
	Job loss	43	40	40
	All	43	32	35
Income poverty (% poor)	Stable employment	41	9	10
	Job loss	81	47	56
	All	56	11	14
Maternal mental health (% poor mental health)	Stable employment	13	10	10
	Job loss	32	14	19
	All	21	10	11
Pianta warmth (% lower or least warmth)	Stable employment	16	19	19
	Job loss	23	26	25
	All	19	19	19
Pianta conflict (% higher or most conflict)	Stable employment	18	14	14
	Job loss	27	20	22
	All	21	14	15
Bases
Weighted		373	2451	3404
Unweighted		256	2699	3415

Base: All originally job-rich families taking part in all five years.

Of course the table above does not show at what level the driver was prior to the event, this is included in the multivariate analysis along with other control factors. Initial modelling revealed the effects of job loss were not different for the two family types and the job loss event for any of the drivers so for each driver both family types were included in the same regression model. Table 5.6 summarises the results from the regression models (see Tables D.11 to D.15 in the technical appendix for full results).

For both lone parents and couple families, compared to stable employment, job loss was significantly associated with a higher likelihood of:

• high level of home chaos;
• income poverty; and
• high conflict in the parent-child relationship.

Table 5.6 Relationship between job loss and drivers of child outcomes controlling for other variables

	Drivers of child outcomes
	High level home chaos	Income poverty	Poor maternal mental health	Parent-child - Low warmth	Parent-child - High conflict
Job loss	? 1	?			?
Driver present at year 1	n/a	?	?	?	?
Interaction: Driver present (year 1) & job loss	n/a

Note: All factors other than the Event are measured at the sweep 1 interview (2005/06).

Note: Arrows indicate whether an event or year 1 driver category is associated with significantly higher (?) odds of the driver of negative child outcomes occurring, compared with the reference category.

Note: All factors with arrows (?) are significant at 5% level, unless otherwise indicated. Blank cells indicate no significant relationship.

Note ¹: Significant at 10% level

The timing of job loss was significantly associated with income poverty, with more recent job loss associated with a higher likelihood of income poverty (models not shown). However, the small number of job losses recorded in some of the years should be borne in mind when interpreting this finding.