Triggering Movements Into and Out of Child Poverty: A Comparative Study of New Zealand, Britain and West Germany

Suzie Ballantyne ¹
Simon Chapple
Ministry of Social Development

Dave Maré
Jason Timmins
Motu Research

Abstract

We know quite a lot about cross-sectional child poverty rates. But we want to move closer to answering the dynamic question of why children move into and out of poverty.

Using a longitudinal data set developed out of the Income Supplement to the Household Labour Force Survey, this research examines trigger events (like losing a job or losing an adult from the household) and responses to these triggers by families, as a means of considering child poverty dynamics in New Zealand. It compares New Zealand's dynamic experiences with Britain and West Germany.

The comparative approach provides information on whether it is differences in frequency of trigger events or in responses to trigger events across countries that drives cross-national differences in chances of children moving into or out of poverty.

A study of the trigger events and responses associated with transitions gets us one step closer to understanding causes of child poverty, an important part of making policy to reduce poverty.

Introduction

What events are related to children moving into and out of poverty? This paper considers the impact of trigger events and responses to these events for chances of moving into and out of child poverty in New Zealand. It compares New Zealand's experience with Britain and (West) Germany. This comparative approach provides information on whether it is differences in events or in responses to events across countries that drives cross-national differences in chances of children moving into or out of poverty. A study of the trigger events associated with poverty transitions also gets us one step closer to understanding causes of poverty and of the different factors contributing to inflows and outflows from child poverty, an important part of making policy to reduce poverty.

The base New Zealand data are considered briefly, followed by a description of child poverty rates and entry and exit rates from poverty in the three countries. The next sections - the core of the paper - deal separately with trigger events for two types of child households - children in lone-parent households and children in couple households - moving into and out of poverty. The conclusion summarises the key elements of the study.

This paper is short and relatively accessible. It draws on a longer, more detailed, and more technical paper (Ballantyne et al. 2003). Readers interested in the full results and technical detail should refer to the longer version.

The Linked Income Supplement Data

New Zealand's quarterly Household Labour Force Survey (HLFS) is a rotating panel, designed to facilitate the inter-temporal reliability of cross-sectional estimates of labour force status. The basic unit of sample selection is the geographic address. Each geographic address is in the panel for up to eight consecutive quarters. Thus the entire panel turns over in a two-year period. While designed for cross-sectional purposes, the rotating panel of the HLFS provides potentially useful longitudinal information, albeit over a relatively short period.

The Income Supplement (IS) of the HLFS occurs annually every June quarter. The IS has been running since 1997. It is designed to collect information on current hourly and weekly earnings and income from self-employment and from government benefits over the reference week of the HLFS.

Because of the rotating nature of the panel, in theory, up to half the dwellings in one IS will also be in the IS the following year. Due to sample attrition, exclusion of imputed records and accelerated sample rotation during the study period, the achieved linkage rate is about half of this theoretical figure (Ballantyne et al. 2003). The longitudinal data and the household demographic data provide information on short-term household income dynamics, which is employed here to examine child poverty transitions.

The unit of analysis is the child, defined as less than 18 years of age. The unit of income accounting is the household. Data on gross weekly total income of those who live in a household with at least one child, observed in both ISs, are used. Households with imputed incomes are excluded from the analysis. Current pay-as-you-earn (PAYE) tax rates are applied to gross income to obtain after-tax personal weekly income. A series of adjustments for child support payments is made, assuming that weekly-income questions are not picking up annual IRD family tax credits. Net income is then summed across the household and equivalised using the square root of the number of people in the household. This net weekly household equivalised income is attached to each individual child in the household. Poverty is defined as an equivalised disposable household income below 60% of the median income for all persons (adults and children).

Descriptive Information on Child Poverty Transitions

This section describes annual exit and entry rates from child poverty and undertakes a cross-national comparison with Britain and (West) Germany. British and German data are taken from Jenkins and Schluter (2003).

Basic cross-national comparisons are shown in Table 1. New Zealand has a child poverty rate (23.2%) between that of Germany (19.4%) and Britain (30.1%). However, the poverty rate for New Zealand children in lone-parent households (47.0%) is similar to that in Germany (49.1%), which in turn is lower than that experienced in Britain (68.1%). New Zealand's couple-household child poverty rate (17.7%) is similar to both Britain's (22.4%) and Germany's (16.3%).

Table 1 Annual Poverty Rates and Poverty Exit and Entry Rates for Children in New Zealand, Britain and Germany

Source for the United Kingdom and Germany: Jenkins and Schluter (2003).

Table 1 also contains information about poverty dynamics, in the form of exit and entry rates. New Zealand has the highest child poverty exit rate (38.6%). This high exit ranking is due primarily to high exit rates for children in couple households (46.3%), since New Zealand's exit rate for children in lone-parent households (25.2%) is actually between Britain's (20.4%) and Germany's (33.4%). New Zealand's child poverty entry rates for both types of household are similar to Britain's but higher than Germany's.

In New Zealand, the difference in exit rates between children in lone-parent households and those in couple households is particularly stark. The chance of exit for a child in a lone-parent household in New Zealand is about half the exit chances of a child in a couple household (25.2% compared to 46.3%), whereas for Britain the ratio is higher - more like two-thirds (20.4% compared to 27.3%). Germany does even better - the chances of a child in a lone-parent household exiting poverty are over 90% of the chances of a child in a couple household (33.4% compared to 36.2%). However, the pattern of relative disadvantage for New Zealand lone parents is not as apparent for child poverty entry rates - in all three countries, children in lone-parent households are around two-and-a-half times more likely to enter poverty.

Trigger Events for Poverty Exit for Children In Lone-Parent Households

In this section, we consider events that may trigger exit from poverty for lone-parent households. Several trigger events are detailed in the first column of Table 2. These events are not mutually exclusive. For instance, it is possible that a child experiences a fall in household size while exiting a lone-parent household, gaining a full-time worker and getting an increase of 20% or more in labour earnings. Additionally, it is worth pointing out that the rise in labour earnings considered here is not a pure price effect, since it does not control for hours worked for a given level of household employment. Thus for many children's households experiencing such an event, parental hours of work may increase.

Table 2 Poverty Exits by Children in Lone-Parent Households for New Zealand, Britain and Germany

Note: Pr(event) represents the chance (probability) of the event occurring. Pr(not poor at t/event) represents the chances of a move out of poverty if the child's family experiences the particular trigger event.

The first line of Table 2 presents the baseline exit rates for all children of lone-parent households in poverty. As shown in Table 1, chances of exit in New Zealand are about one-quarter (25.2%), which sits between the high of one-third in Germany (33.4%) and the low of one-fifth in Britain (20.4%).

There are cross-national differences in the chances of trigger events occurring. In New Zealand, 12.4% of lone-parent households fell in size. The chances of the event in the two other countries were negligible (1.3% for Britain and 0.0% for Germany). It is likely that this finding is an artefact of different data collection methods by the three countries.

Compared to Britain (17.0%) and Germany (12.6%), New Zealand's lone parents have a lower chance of repartnering (10.3%). On the other hand, in terms of generating full-time jobs for poor lone-parent households, New Zealand (10.5%) sits between Britain (8.9%) and Germany (14.6%). However, the joint probabilities of repartnering and gaining a full-time worker (either through changes in the initial lone parent's labour status or because the new partner has or gets a job) are very similar across the three countries (New Zealand 5.7%; Britain 6.0%; German 5.3%).

There are very large differences between New Zealand and the other two countries in terms of the chances of having a 20% or more increase in labour earnings with the number employed held constant. In New Zealand the chances of this happening to lone parents are a modest 6.3%, while it is experienced by two-thirds (65.2%) of lone parents in Britain, and by half (50.0%) of lone parents in Germany. The extreme difference is puzzling. It may be that, in Germany and Britain, many lone-parent households experience increases in hours worked from a low base. It may also be the case that variations are generated by data-set differences.

Now consider the comparative chances of exiting poverty conditional on experiencing a trigger event. The point of comparison is the first row - the baseline chances of exiting poverty - which is 25.2% for New Zealand, 20.4% for Britain and 33.4% for Germany.

Leaving a lone-parent household raises the probability of a New Zealand child exiting poverty only very marginally above the baseline (28.6% compared to 25.2%). However, the effects are much stronger in Britain (46.0% compared to 20.4%) and Germany (48.8% compared to 33.4%). Thus changes in household structure work best for removing children in lone-parent households from poverty in Britain.

Gaining a full-time worker raises child poverty exit chances in all three countries. The chances are lowest in New Zealand (61.6%), with British chances a solid three-quarters (74.1%) and German chances even higher (82.2%). Thus labour markets work better for getting children in lone-parent households out of poverty in Britain and Germany than in New Zealand.

Experiencing the combined event of exit from lone parenting and gaining one or more full-time workers reveals a similar ranking. While the event nearly guarantees exit in Britain (83.5%) and even more so in Germany (91.7%), the chances of leaving poverty for New Zealand children of lone parents are only half (51.8%).

Gaining an increase of 20% or more in labour earnings does not raise exit chances above baseline in New Zealand (26.3%). The German result, too, is similar to baseline (32.3%). In Britain, the chances of exit on this event are actually lower than the baseline exit rate (11.1%). Thus most of those who experience such an event must be coming from a long way below the poverty line for so few to make it over the line.

Trigger Events for Poverty Exit for Children in Couple Households

This section presents cross-national comparisons of poverty exits for children in couple households in the same way as has been done for children in lone-parent households. The relevant information is contained in Table 3.

Table 3 Poverty Exits by Children in Couple Households for New Zealand, Britain and Germany

Note: Pr(event) represents the chance (probability) of the event occurring. Pr(not poor at t/event) represents the chances of a move out of poverty if the child's family experiences the particular trigger event.

Overall, children in couple households are more likely to exit poverty in New Zealand (47.0%) than in Britain (27.3%) and Germany (36.2%).

Consider the different probabilities of trigger events across the three countries. Falls in household size are much more common in New Zealand (8.1%) than in Britain (1.7%) and Germany (2.2%). Some of this difference may be due to differences in survey design - a temporary absence of a household member will appear as a reduced household size in New Zealand, but not in the other countries. The chances of gaining a worker are a little under a third in New Zealand (29.0%), which is higher than the one-in-five chance in Britain (20.1%) and the 15.1% chance in Germany. The chances of gaining a full-time worker follow the same ordering, with a one-in-five chance of the event in New Zealand (20.7%) compared to 15.4% in Britain and 10.9% in Germany.

For children in couple households, a 20% rise in earnings is least likely to lead to exit in New Zealand (17.4%). The chances of such an event are more than twice as high in Britain (40.8%) and almost twice as high in Germany (32.0%). While these country differences for poor couple households are substantial, they are not as disproportionate as the cross-national differences for children in poor lone-parent households (6.3% compared to 65.2% and 50%, shown in Table 2).

In examining the probabilities of leaving poverty conditional on a trigger event, the crucial row for comparison is the baseline data on exit rates in the first row of Table 3.

New Zealand children in couple households who experience a decline in household size have an 81.8% chance of leaving poverty (a three-quarters increase over baseline), compared to Britain and Germany, where the rare event barely raises the chances of leaving poverty above the baseline at all.

Gaining a worker raises chances of a child poverty exit in New Zealand to 65.3% while gaining a full-time worker is even better - an 80.7% chance of exit (with the baseline being 47.0%). Relative to both the lone-parent situation in New Zealand and to conditions in Britain and Germany, gaining a worker in New Zealand has a much stronger impact on child poverty exit in a couple household. Chances for exit conditional on the same event in Britain (50.0%) and Germany (56.5%) are similar to each other and lower than in New Zealand.

Gaining an economically significant rise in labour earnings gives a chance of poverty exit of 70.8% in New Zealand, comparable to Germany at 62.9%, but over double that in Britain, where the chances of exit (28.3%) are no more than the baseline rate (27.3%). Thus children in couple households must be closer on average to the poverty line than children in lone-parent households for this event to shift much larger numbers of couple households out of poverty.

Overall, New Zealand is much more successful than Germany and Britain in generating jobs for poor couple households with children and, once they get a job, in moving them out of poverty.

Trigger Events for Poverty Entry for Children in Couple Households

The analysis now shifts to consider movements into child poverty. Poverty entry can only be considered for children in couple households since the sample numbers for non-poor children in lone-parent households are too small to sustain further analysis.

Again, trigger events are considered first (see Table 4). Household size rises for about one in 10 children in New Zealand couple households not in poverty (9.5%). The odds of this happening are lower in Britain (6.0%) and Germany (4.8%).

Table 4 Poverty Entry by Children in Couple Households for New Zealand, Britain and Germany

Note: Pr(event) represents the chance (probability) of the event occurring. Pr(not poor at t/event) represents the chances of a move into poverty if the child's family experiences the particular trigger event.

Risks of non-poor children entering a sole-parent household are small and very similar in New Zealand (1.8%) and Germany (1.6%), and higher but still small in Britain (3.2%). Given the geographic basis of the New Zealand data and the likelihood that, in many cases, becoming a sole parent will be associated with a movement in geographic location, couple separations are likely to be underestimated for New Zealand.

Experiencing the loss of a worker is most likely in Britain (18.0%) and least likely in Germany (8.7%). New Zealand lies between the two comparison countries (13.9%). Similar relative patterns are found for the event of losing a full-time worker.

The probability of a child in a non-poor household experiencing the twin events of entering a lone-parent household and having their household lose a worker is low for all countries, lying between 1% and 2%.

Country rankings of the chances of experiencing a significant fall in labour earnings for a non-poor couple household are quite different from country rankings for experiencing a significant rise in labour earnings for lone-parent and couple households in poverty. Children in New Zealand couple households who are not in poverty have a one-in-five chance of experiencing this fall in earnings (18.7%), which is much higher than chances in either Britain (8.4%) or Germany (8.0%).

Finally, the chances of children in non-poor families experiencing the event of a newborn child are nearly twice as high in New Zealand (7.5%) as in Britain (4.3%), where the chances are in turn nearly four times higher than in Germany (1.0%). The reason for the large German-British difference is due to the different nature of the two surveys (Jenkins and Schluter 2003:461 footnote 15). The geographic basis of the New Zealand data set and the fact that families about to have children are less likely to be geographically mobile is likely to be driving some of the difference between New Zealand and Britain. Additionally, New Zealand's fertility rate is higher than both Britain and Germany.

Now consider conditional probabilities of children from couple households entering poverty. Baseline entry rates are 8.2% for New Zealand, 9.9% for Britain and 6.4% for Germany.

Increases in household size do not push the chances of children in couple households falling into poverty above baseline in New Zealand, but almost double the baseline chances in Britain (18.0%) and increase them by almost one-third in Germany (9.0%).

While joining a lone-parent household is a comparatively rare event, it leads to a big increase in all three countries in the probability of falling into poverty compared to the baseline chances. However, the chances of a split leading to poverty are lowest in New Zealand (43.7%) compared to 61.8% in Britain and 58.9% in Germany.

The experience of losing one or more workers - or one or more full-time workers - generates very similar cross-national chances of children in couple households falling into poverty in all three countries. The chances are one in five for New Zealand (20.5%), Britain (22.0%) and Germany (21.5%)

While significant falls in labour earnings are most likely in New Zealand, the chances of this moving children in couple households into poverty is close to baseline (8.8%). It is much more likely to move both British and German children into poverty, with a chance of over one in four for Britain (27.7%) and about one in five for Germany (19.3%).

Finally, having a newborn in Britain or Germany substantially increases the chances of children in couple households falling into poverty - in both cases, the odds are over one-quarter (Britain 27.2%; Germany 25.9%). The event has less of an impact in New Zealand, with the odds only a little above baseline (9.5%).

The overall impression is that there is at least as much variation in event probabilities across countries as in the chances of leaving or entering poverty conditional on experiencing the trigger event. This conclusion was arrived at semi-formally. There were 16 cases across the three tables where coefficients of variation (standard deviation divided by the mean) could be calculated across the three countries for the trigger events as well as the transition probabilities and the coefficients of variation compared for trigger events versus transition probabilities. In 12 out of the 16 cases (75%), country variation in experience of trigger events exceeded country variation in transition probabilities.

How Much Do The Trigger Events Contribute To Poverty Dynamics?

Another insightful way of cutting up the data is to consider the percentage of those exiting poverty who had a particular trigger event associated with that exit. This provides an indication of the relative importance of trigger events in association with exits or entries. The events should not be expected to sum to 100%. Because there are exits not associated with trigger events, the events could sum to less than 100%. In addition, the events are not mutually exclusive, so the same exit may be associated with more than one trigger event, and thus the sum may be over 100%. This method provides some information on comparing one country with a low event prevalence but a high exit probability with another with a high event prevalence and a low exit probability. The information is provided in Tables 5-7.

Table 5 Share of Poverty Exits (%) of Children in Lone-Parent Households who Experienced the Trigger Event; New Zealand, Britain and Germany

Table 5 considers children in lone-parent households. Over a quarter of children exiting poverty from a lone-parent family gain a full-time worker in New Zealand (25.7%), compared to around one-third in Britain (32.3%) and Germany (35.9%). Only 11.8% of children in lone-parent New Zealand families left poverty with another adult moving in, whereas 38.3% of British exits are accompanied by repartnering. The biggest single event in Britain associated with poverty exit for children of lone parents is a household-structure event, not a labour-market event as it is in New Zealand. Germany is more like New Zealand, with the household-structure event being relatively less important and the labour-market events being more common.

Table 6 shows that nearly one-half of child poverty exits from couple households are associated with the gain of a worker in New Zealand (41.0%), much higher than the one-third in Britain (30.2%) and the one-fifth in Germany (20.9%). On the other hand, over half of children in poor couple households in Germany who leave poverty are in households experiencing a significant rise in labour earnings (55.6%) compared to one-quarter in New Zealand (26.6%).

Table 6 Share of Poverty Exits (%) of Children in Couple Households who Experienced the Trigger Event; New Zealand, Britain and Germany

Finally, consider the relative importance of trigger events in association with entry into poverty, shown in Table 7. Fully one in five British children in couple households who fall into poverty are associated with parental separation (20.0%), compared to one in seven in Germany (14.7%) and less than 10% in New Zealand (9.6%). Overall, more children take adverse labour market events with them when they fall into poverty than take adverse marriage market events (or, to be more accurate, probably mostly marriage market). Of children in couple households in New Zealand who move into poverty, 36.3% have a transition associated with someone in the household losing a job, compared to 41.8% in Britain and 27.2% in Germany.

Table 7 Share of Poverty Entries (%) of Children in Couple Households who Experienced the Trigger Event; New Zealand, Britain and Germany

Conclusion

While having a high overall exit rate from child poverty, New Zealand generates a much lower probability of exit from poverty for children in lone-parent households relative to children in couple households compared with Britain and Germany.

For lone-parent households, New Zealand is much less successful than both Britain and Germany in translating positive events, like changes in household structure and the household gaining a full-time worker, into child poverty exits.

On the other hand, for poor children in couple households, New Zealand appears to be far more efficient than either Britain or Germany in translating labour-market events, like gaining a full-time worker and experiencing an economically significant rise in labour earnings, into poverty exits.

A common pattern for poor children in both lone-parent and couple households in New Zealand is that New Zealand again appears more efficient than either Britain or Germany in translating a significant earnings rise into a poverty exit.

Additionally, the analysis re-emphasises the importance of labour-market events and responses for generating child exits from poverty compared to demographic events. Demographic events and marriage-market events are less frequent and, in New Zealand, are less likely to generate the positive events of exit and the negative events of entry into child poverty.

References

Ballantyne, Suzie, Simon Chapple, Dave Maré and Jason Timmins (2003) "Movements into and out of child poverty in New Zealand: Results from the Linked Income Supplement" Ministry of Social Development Working Paper.

Bradbury, Bruce, Stephen Jenkins and John Micklewright (eds.) (2001a) Falling In, Climbing Out: The Dynamics of Child Poverty in Industrialised Countries, Cambridge University Press, Cambridge.

Bradbury, Bruce, Stephen Jenkins and John Micklewright (2001b) "Conceptual and measurement issues" in Bruce Bradbury, Stephen Jenkins and John Micklewright (eds.) Falling In, Climbing Out: The Dynamics of Child Poverty in Industrialised Countries, Cambridge University Press, Cambridge.

Bradbury, Bruce, Stephen Jenkins and John Micklewright (20001c) "Child Poverty Dynamics in Seven Nations" in Bruce Bradbury, Stephen Jenkins and John Micklewright (eds.) Falling In, Climbing Out: The Dynamics of Child Poverty in Industrialised Countries, Cambridge University Press, Cambridge.

Jenkins, Stephen and Christian Schluter (2003) "Why are child poverty rates higher in Britain than in Germany? A longitudinal perspective" Journal of Human Resources, 38(2):441-65.

Picot, G., M. Zyblock and W. Pyper (1999) "Why do children move into and out of low income: Changing labour market conditions or marriage or divorce?" Statistics Canada, April, No. 132.

¹ Acknowledgements
The opinions expressed here and errors made are the responsibility of the authors and not the Ministry of Social Development. We thank Martha Hill, Stephen Jenkins, Tim Maloney, Rebecca Martin and Bryan Perry for very helpful comments. We also thank Sandra MacDonald and the team from Statistics New Zealand for their hard work on the data set. Errors and omissions remain our own. Access to the data used in this study was provided by Statistics New Zealand, under conditions designed to give effect to the confidentiality provisions of the Statistics Act 1975. The authors are listed alphabetically.
Correspondence
Simon Chapple, Strategic Social Policy, Ministry of Social Development, PO Box 12 136, Wellington, telephone: +64 4 916 3847, fax: +64 4 916 3775, email: simon.chapple006@msd.govt.nz