Transcripts

NEET seminar - video transcript

(Eyal Apatov - a Senior Analyst within the Oranga Tamariki Evidence Centre)

Thanks everybody for coming.

So this is work I've done while I still at MBIE, as you can recognise the background.

As Kelly said, I used data from the IDI and the results have been tested to pass all the confidentiality tests and another disclaimer, this was work that was mainly done for economists and people who work in the economics sector so my guess is that if you've been involved in the social sector and programmes that influence need(?), that look at youth outcomes will probably more confirm things you already knew as opposed to telling you something new, but, yeah it would be good to get that sort of feedback as well.  So it is mostly for a wide audience and looking at a range of factors as opposed to focusing on specific things like education or health, etc.

So shortly, the studies about that gap.  Over here we see the rate of youth, 15 to 24 that are not in employment, education or training.  This is a break down by ethnicity and you can see that there is a persistent gap between the NEET rate of Māori and Pacifica or pacific peoples compared to other ethnic groups.  Roughly about twice.  And that gap, even though it kind of closes down it is still constant and persistent.

So that is one fact.  We know that the NEET population is diverse but there are some characteristics that are more common, for example, early school leavers, early parenting, residing in certain areas, having family characteristics.  We know that these factors that generally predict NEET, NEET status, are also more common amongst Māori and Pacific youth.

So the next question is what's going to happen to that NEET rate gap if Māori and Pacifica had on average the same distribution across these characteristics if those gaps between risk factors were eliminated would we expect that NEET rate gap to be eliminated as well.

Another thing I looked at is are certain risk factors more important to explain that gap for some sub groups than for others.  To go into a bit more detail about what I was talking about, what happens if we eliminate the gap in risk factors and how that effects the NEET rate gap, we have here an example with two groups: group A and group B.  The horizontal axis shows the average hypothetical education level for both groups.  So the average for group A is level 6, the average for group B is level 2.  On the vertical axis we see the corresponding weekly pay.  For both groups the lines show the relationship between education level and learning.  So as I say, group A has on average level 6 qualifications and earn about $700, group A has level 2 qualification earns an average $250.  So there is a gap in earning and there is a gap in qualification level.

If we had a programme that pushed the qualification level of group B, because the relationship is more or less the same, the wage gap will be eliminated, or largely eliminated. 

But there could be a situation where it's not just the average qualification level is different, but their return to qualification.  How much extra do you earn when you increase your qualification level?  So in this sort of story when we increase the qualification level of group B from two to six, we do close that earning gap, but not all the way.  So it explains some of the gap, but not all of it. 

So we in MBIE wanted to know how much of the gap -- how much of the need gap can be closed if those differences and risk factors are closed.  What story do we have?  Is this just a story about differences in characteristics or are there other things?  So for example, the chart on the right is more of a story of the gender wage gap.  You take males, you take females, you equalise education and other background factors you explain about a fifth of the gap. 

If you like equations more than pictures this is the Greek letter version.  Basically what is says, it repeats the pictures.  Your likelihood of being a NEET depends on having certain risk factors, which is those Xs and other stuff, which is that (inaudible).  You look at each ethnicity in the study and you calculate that relationship and then if you take any two groups you can break down the gap in NEET rate into three factors, broad factors, highlighted in three different colours.  The first part is the that the two groups -- the gap is explained by the two groups having differences in risk factors.  So there was that horizontal axis.  Group A had level 6 education.  Group B had level 2 education and that is why we see a gap in earning. 

The second part is differences in returns to risk factors.  So there were the differences in slopes. 

The third part is a bit more tricky but the intuition is that -- in this study I'm not just going to look at education, I'm going to look at a lot of factors and what this term(?) asks is are the risk factors that have the biggest gap also have the biggest difference in slopes.  So that could tell, for example, some groups are more likely to reside in poor areas and if they reside in more poor areas they're also more likely to be NEET.  So the effects of things were we see the biggest gap is also more damaging, for example.

So I'm keeping with the official definition of NEET and looking at youth age 15 to 24, focusing on 2016 and I look at long term spells of NEET and that's because past studies found that this is a certain -- like the risk of experiencing adverse outcomes in future years is greater if you experience long terms spells of NEET as opposed to one week of being NEET.  So I'm comparing outcomes and NEET rates between ethnicities and I got four broad ethnic groups.  One is youth that identifies solely as Māori and it's termed Māori only.  Youth that identify as Māori and other ethnicities, termed as Māori and Pacific people, their label will be also Pacifica, I didn't have the time to change that.  

The control group, the group I'm comparing is everybody in my sample that didn't identify as Māori or Pacifica, some (inaudible) Māori Pacific people and (inaudible) call other.  And rather than looking at the entire ethnic group I'm breaking it down by sub age group and gender.  So this is based on administrative record from the IDI.  So I'm following the approach that McLauren Toman took from the Treasury to define it.  It is basically assigning activities to an individual in each month.  So I'm asking for each individual, have you been overseas in a month for 15 days or more, if not have you been in custody for 15 days or more, if not have you been enrolled to an educational course for a day or more in that month, if not, did you earn at least $10 from wages, salaries or self-employment, if the answer to all of these is no I'm saying you are a NEET in that month. 

To be long term NEET you need to have six of those in a row, consecutive months.  So my sample has almost 600,000 observations.  This is about 90% of the overall estimated population.  The biggest difference is I exclude temporary migrants. 

I collect information about the usual suspects, educational achievements, other personal details, family characteristics and area factors that have all in past studies been found to predict youth outcomes. 

Looking at the data, this chart will kind of look familiar to the first chart.  This is the long term NEET rate by ethnic group and just like the official rate it has been falling since 2012 and at a slightly faster rate for Māori and Pacifica but still there's a quite large gap between that and that of other, non-Māori and Pacifica. 

If I look at the break down by the 15 to 19 year old, just like in the official stats, there's not much difference across gender at this age group and as the overall group the highest rate is for Māori only followed by Māori Pacifica and other.  When I look at all the age groups you can see the NEET rate increase and the gap in terms of percentage point difference also increase, and that is similar to the official stats.  What is also similar to the official stats is the higher rate for females at this age group. 

The official stats allow you to look at types of NEETS and a lot of the difference between -- most of the difference between male and female is to do with caregiving duties.  Now I don't have access to caregiving duties, but what I've done is broken down the NEET rates by whether you have one or more children.  So on the left, this is the NEET rate for 20 - 24 year olds without children and on the right what is the NEET rate if you have at least one child. 

The rate for male increases, doubles, but the rate for females is over half.  And at the bottom, in percentages, that's the portion of all NEETs that are from that sub-group.  So 5% of all long term NEET are males with one or more child compared to almost one out of four.  So one out of four long term NEET is mothers.  So that is something that changed my perspective about long term NEET.  I kind of imagined street kids being all, you know, street kids, but actually it's mothers that is a big part of that story.

Other high level findings, so as in past studies I found that Māori and Pacifica have greater prevalences of the factors that predict they're going to be NEET.  So low decile -- attending low decile schools, parents with benefit dependency. 

On the other hand when I compare the slopes, how do these factors -- what is the likelihood of if you have this factor to become NEET.  The relationships weren't that different.  So if you think about the example, the slopes weren't that far apart.  There were some differences but generally if you're an early school leaver you're probably going to be NEET regardless of your ethnicity or ethnicity doesn't matter as much at this point. 

Just a quick point about (inaudible) basically I could see from the returns that having a driving licence of a bachelor degree reduces your likelihood of being a NEET.  If you're a mother as well there's a compound effect.  So more than the sum of the parts. 

So looking at the results of the composition now.  So I am going to go slowly because there is quite a lot in these tables.  So these tables look at the composition results between Māori, Māori only Pacifica and other.  My control group for those ages 15 to 19.  And there's a break down by male and female labelled M and F. 

The first row shows the difference in long term NEET rates.  So if you remember the bar charts, that is the gap between any two groups.  So the NEET rate for Māori only was around 13% males.  The one for other was 9.5 percentage points lower, so that 0.095. 

Then the total gap, the total NEET rate gap is broken down in to three factors.  Risk factors, observable risk factors, that's that horizontal axis, how much of it relates to that.  Returns, how much of it has to do with slopes, differences in slopes and the interaction affect. 

In brackets I'm showing how much of the gap is attributed to each factor.  So for Māori only male 15 to 19 93% of the gap is explained by differences in risk factors. 

If you just follow that first row you see it varies from 77 to 126%.  What it means when more of the 100% of the gap is explained, is that if, in this example, Pacifica had on average the same characteristics of non-Māori and Pacifica their NEET rate is going to be relatively lower. 

Returns too is important, but slightly less, between 23% and 32%.  And if we look at the 24 year old group this pattern holds. 

Risk, differences in risk factors, explain between 80% and 112% of the gap. 

Now I wouldn't necessarily that it's 80 and not 79 and not 82 but I would believe the relative importance, and as you can see the returns, differences in the slopes are even less significant for the older age group and even more so for interactions. 

If I look at specific factors I want to know which factors are important for each group.  Here I'm focusing on Māori only, on the NEET rate for Māori only males aged 20 to 24. 

The vertical line is the actual gap, almost 13 percentage points.  And now I'm asking if everything was the same but Māori only had the same share of youth that didn't receive any suspension and warning by age 16, what is going to be my new gap? 

So this chart shows that the gap would fall by about 1 percentage point.  There would be a slightly larger effect if they are the same share of parents that receive benefit.  Slightly bigger effect for holding a current driving licence and a bigger effect if the distribution across highest qualification was the same between Māori only and other. 

If you add all the factors you can see the gap almost eliminated, just over one percentage point.

If we look at Pacifica females, we look down at the 1, 2, 3 -- the first four factors are the same as from the previous slide with relatively similar effects.  But the effect of having children, differences in the proportion of Pacifica females with children explains about half the gap.  Much bigger than qualification or driving licence provision.  If we add all the factors we see that we are just left the zero meaning that the gap will reverse because Pacifica are expected to have a lower long term NEET rate compared to non-Māori and Pacifica.

So what do we do with all this?  This is good news, I mean, I the rate is falling and what explains the gap is things that we can see and if we can see them we can target them.  It is a better story than only 10% of the gap is explained by things we can see. 

Even though there is repetition across risk factors there is some variation across sub-groups.  So the biggest difference was parenting, much larger effect on females compared to males.  Other effects were area level effects.  Area deprivation was much stronger for 20 to 24 year olds, was stronger for Māori.  Qualifications were important for the older age group and driving licence seems to have a stronger -- explaining a stronger portion of the gap for males.

But overall it says it's supporting interventions that promotes school performance and provision of driving licence have the potential to substantially reduce the NEET rate gap. 

What is less clear, there is some area level and parental level -- family level effects that were as strong as qualification or driving licence.  What does that mean in terms of policy intervention?  I can't think of any direct response, but it does highlight risk groups.  Groups we can look at further and better understand.  So, for example, if we think about the -- well, there's a lot of Māori residing in highly deprived areas, but Māori that reside in these areas are also more likely to be long term NEET.  Why?  Is the experience of being a NEET in a highly deprived area the same as a non-deprived area?  What is the difference between being NEET or non-NEET in a highly deprived area.  Why having parents that are currently on benefit explain your likelihood of being NEET as much as your highest qualification?  

And just a final thought is that a lot of the stories about motherhood and my thought was the IDI is usually based on individual level unit of analysis and I was thinking, "Is this the right level?"  So if we had a policy and it would cut the rate for mothers by half, is this a success story necessarily?  What would be the outcomes for dependant own children, females doing caregiving duties not for their children, for other dependants?  So for me it was like a wake up call to think about beyond the individual level, to also consider family level analysis. 

And I'll just leave you with that. 

Thank you very much.