The ebb and flow of school enrolments

29th November 2019 by Timo Hannay [link]

Much is (rightly) made of the need for an adequate supply of good school places. But how well does this supply meet demand, does it vary by school type or location, and how does it change over time as different pupil cohorts work their way through the system? This blog post will look at all of these questions in an analysis that we have conducted in collaboration with TES.

Our main conclusons are:

• Year-group cohorts can vary greatly in size, sometimes by 20% or more, and often rise and fall in waves. This seems to have measurable effects on class sizes and school occupancy (pupil headcount as a percentage of school capacity).
• Overall primary school occupancy is currently falling because the cohorts that have recently entered Reception and Key Stage 1 are smaller than their recent predecessors. We expect overall primary school enrolments to fall further as these smaller Key Stage 1 cohorts enter Key Stage 2 and replace the larger cohorts currently occupying those year groups.
• Furthermore, recent declining birth rates suggest that future Reception cohorts will be even smaller, so the overall decline in primary school enrolment is likely to continue for at least the next few years.
• The resulting reductions in primary school occupancy are not uniform, but tend to affect certain types of schools more than others. In particular, smaller primary schools and those with lower Ofsted ratings not only had lower occupancies in 2015, before the current declines began, they have also fallen further than other schools since then. These are the kinds of schools that are likely to face the biggest challenges with any ongoing decline in primary pupil numbers.
• In contrast, overall secondary school occupancy has risen since 2015. Nevertheless, many of the same trends are evident: low Ofsted ratings, small school sizes and non-selective status all correlate with relatively low and decreasing school occupancy.
• Depending on your point of view, these trends are increasing systemic inequality or are simply signs of an effective education market in action. In either case, they pose questions for those tasked with running schools and planning future provision.

Users of SchoolDash Insights can see further analysis in the 'Pupils' section using the new cohorts and capacity data that we have released to coincide with this study.

Wave of the future

The numbers of children within each year group are not consistent but vary from year to year, often in cycles, as shown in Figure 1. (These statistics come from the DfE's annual pupil census, which is conducted each January – ie, roughly in the middle of the academic year¹.)

In 2012, there were peaks in Reception and Year 10, with a trough in Year 5. These fluctuations aren't trivial either – the Reception Year cohort was nearly 14% bigger than the Year 5 cohort.

Obviously this pattern tends to shift to the right over time as children age², so by 2014 the children in the demographic trough had reached secondary school (Year 7) and a new even bigger peak had entered Year 1. Another two years later, in 2016, children in the 'trough' cohort were at the end of Key Stage 3 (Year 9) and the numbers of pupils in early primary years (Reception to Year 3), while still high, had reached something of a plateau. By 2018, as the trough cohort reached the end of secondary school (Year 11), the number of children entering primary school began to drop. This trend has continued into 2019 and the decline has become more marked, with the Reception Year cohort about 5% smaller than the peak Year 2 cohort.

Note also that the net effect between 2012 and 2019 has been for the numbers of primary school pupils (Reception to Year 6) to increase greatly, those in the the early years of secondary school (Years 7-9) to increase a bit and those in the latter stages of secondary school (Year 10 and 11) to drop slightly.

Figure 2 shows these temporal changes more clearly. Pupils in Reception and Years 1-2 (red line) rose from 2012 to 2017 but have since declined. Those in Years 3-6 (yellow line, corresponding to Key Stage 2) have grown strongly, increasing by nearly 17% between 2012 and 2018. However, in 2019 they have flattened off and will surely start declining over the next few years as the current smaller cohorts of younger pupils graduate from Key Stage 1.

Among secondary schools, Years 7-9 (green line, Key Stage 3) fell gradually between 2012 and 2014 but have since grown. Finally, Years 10-11 (blue line, Key Stage 4) fell slowly from 2012 to 2017 and have since been virtually flat.

Class room

So much for demand, what of supply? Or to put it another way, what has happened to school capacity as these demographic changes have worked through the system. Figure 3 shows the changes in overall primary and secondary school capacity and pupil numbers between 2015 and 2019³.

The first thing to notice is that primary schools as a whole are very close to capacity. In 2015, the overall occupancy (pupil numbers as a percentage of school capacity) was over 98%. Capacity has since increased by around 217,000 places (5%), but so has demand, which means that even in 2019 overall occupancy remains relatively high at about 96%. In general, however, capacity and enrolment have tracked one another quite closely, suggesting that the system is relatively good at adjusting to demand, at least over the time scale shown.

Overall secondary school occupancy is much lower – around 86% – and has remained relatively stable over this period.

Even if total school capacity tends to track total demand, it is important to understand where is it added or taken away since this may affect individual choice and could even shift the overall balance of the education system. Between 2015 and 2019, primary school capacity grew by about 5% while secondary schools shrank by about 1%. But beyond this there are some big differences within each phase, as shown in Figure 4.

Most strikingly, schools with high Ofsted ratings grew more than those with lower ratings. Free schools also grew exceptionally strongly (as you would expect from newly founded institutions) as did non-Christian faith schools (something we have written about before). Among secondary schools, single-sex and grammar schools outpaced other groups. Use the menu in Figure 4 to explore other categories of school.

There are also substantial regional differences, as shown in Figure 5. Capacity tended to grow most in London, the South East and the North West of England. The lowest growth rates among primary schools and greatest reductions among secondary schools were in the north and east of the country.

Size up

What effect does all this have on schools? One interesting aspect to study is class size, data for which are shown in Figure 6. (Note that in order to make the relatively small changes easier to see, the y-axis does not begin at zero.)

Of course, class sizes aren't affected by pupil numbers alone. Among other things, they can also be influenced by school budgets, availability of teachers and general labour market conditions. Nevertheless, the patterns here seem to be broadly consistent with the general trends in pupils demographics seen above. Specifically, average class sizes at Key Stage 1 (Years 1-2, red line) crept up slightly between 2012 and 2017 but have since declined. In contrast, Key Stage 2 class sizes (yellow) have risen throughout.

Secondary schools (blue), as well as having smaller average class sizes, also show a different pattern. From 2012 to 2014/2015 – the same period during which the demographic trough was working its way through Key Stages 3 and 4 – class sizes dipped. Then they rose again from 2015 to 2019 as relatively large cohorts entered Key Stage 3.

Occupying forces

But perhaps the most pressing question, especially from the point of view of those running schools, is what effect all this has on school occupancy. To provide some context, we will start by looking at average occupancies over the whole period from 2015 to 2019. As already mentioned, across all years and school types, occupancy among primary schools (96%) has been much higher than among secondary schools (86%). Beyond this, however, it can vary greatly by school type within each phase, as shown in Figure 7.

For example, Ofsted 'Outstanding' schools as a whole are virtually fully populated, while 'Inadequate' schools are heavily under-populated, especially among secondary schools. Similarly, schools with high proportions of poor pupils have lower occupancies, and this effect is much bigger among secondary schools than primary schools. Large and urban schools also tend to be fuller than their small and/or rural counterparts. Of the different faith categories, Catholic schools have the highest occupancies.

Converter academies (typically high-performing schools) have higher occupancies than sponsor-led academies (usually previously underperforming schools). Note that overall occupancy among free schools is very low. This is not necessarily a cause for concern because they are often new and still in the process of building up their pupil numbers.

Among secondary schools, those with small proportions of previously low-attaining pupils are more fully subscribed, as are single-sex schools and grammar schools. Use the menu in Figure 7 to explore these and other school types.

Though there are also differences by region, these are not as marked, as shown in Figure 8.

Finally, we will look at how school occupancy has changed since 2015. Figure 9 combines occupancy in 2015 (horizontal axis) with changes in occupancy from 2015 to 2019 (vertical axis) for different types of primary school. The large grey dot in the centre represents all primary schools, which, as we've already seen, were close to 100% occupancy in 2015, though this proportion has since fallen by 2.6 percentage points.

Looking by Ofsted rating, 'Outstanding' schools not only had the highest occupancy in 2015, they have also fallen by the smallest amount since then. In contrast, 'Inadequate' schools started well below full capacity in 2015 and have fallen by more than 4 percentage points since that time. (Hover over the dots in Figure 9 to see accompanying details.) A similar pattern is apparent for different sizes of school, with smaller schools showing both lower initial occupancies and larger declines. Thus we see some schools that were already struggling come under more even pressure – an example of the so-called 'Matthew Effect'. Depending on your point of view, this either increases systemic inequality or demonstrates the education market in action (or perhaps both).

This self-reinforcing trend isn't universal, however: some school types show the opposite effect in which initial low occupancies tend to correlate with higher increases (or at least lower reductions) in occupancy over time. This tends to reduce inter-group differences. We can see this, for example, when looking by region, local deprivation level, faith group and academy type. (For that last grouping, notice the rescaling of the axes to accomodate free schools, which are represented by the tiny dot in the far top-left.) To explore further, click on the legend in Figure 9 to turn different groups on or off; double-click to show one group on its own.

Figure 10 shows the same analysis for secondary schools. As already described above, these have lower occupancies (87% overall in 2015) and this has grown slightly since (up 1.9 percentage points). When looking by school type, we once again see a 'Matthew Effect' for Ofsted ratings and school size, with lower-rated and smaller schools tending to decline in occupancy while others grow. We can also add grammar schools, which rose from around 101% occupancy in 2015 to a remarkable 105% in 2019.

In contrast, we see the opposite effect when looking by school deprivation, local deprivation and proportions of pupils with previous low attainment. (Note that these three factors often correlate, so it's not surprising that they show similar patterns.) The same is also true when looking by religious affiliation, sixth-form status and academy type (though that last trend is thanks mainly to free schools, represented by the small dot in the top-left corner).

Years to come

What can we expect in the next few years – will the number of children entering primary school continue to fall, or will they stabilise or perhaps even start to rise again? One obvious way to try and answer this question is to look at recent birth rates (as reported by the ONS). These are shown in Figure 11. The red columns show the numbers of live births in England and Wales during each calendar year; the blue columns show 2019 cohort sizes by year group, exactly as we saw in Figure 1. The latter have been lined up with their putative year of birth: Year 11 corresponds to 2004, Year 10 to 2005 and so on. (Note that this correspondence isn't precise because births are tallied by calendar year while pupils are assigned to cohorts based on academic years. For example, some current Year 11 pupils were born in 2003, not 2004. But the alignment is close enough to observe long-term trends, which is what concerns us here.)

The first thing to note is that the number of births each year exceed the corresponding school cohort sizes (ie, the red columns are taller than the blue columns). This is for two main reasons: (i) the numbers of births are for England and Wales, while the school cohort sizes are for England only; and (ii) the school cohort sizes only capture pupils in the state system, not those in private education. There may also be some effects from immigration and emigration.

But these differences in totals aside, the overall patterns of rises and falls are strikingly similar. This isn't surprising: we would expect birth rates to be a (perhaps the) major driver of school cohort sizes. In particular, the peak birth rate in 2012 corresponds very closely to the peak cohort sizes currently in Years 2 and 3. Since then, we have seen a decline in birth rates and a corresponding fall in school cohort sizes. Importantly, the birth rate has continued to decline markedly over the last three years, indicating that primary school enrolments are very likely to drop further.

Given this information, it's even possible to estimate the approximate size of the changes to come. This requires some assumptions (described in Footnote 4), but based on these it seems plausible that the total state primary school population in England could fall by about 2% over the next three years. That may not sound like a lot, but it corresponds to around 95,000 children, or roughly 400 medium-sized primary schools. If these changes fall unevenly, which appears likely given our analysis above, then some schools may be profoundly affected.

It is important to add that DfE makes it own more elaborate forecasts of state school numbers. These currently foresee a slower decline in which total primary school enrolment is flat for the next two years, followed by a reduction of about 100,000 pupils, mostly during 2022-2024. There are many potential reasons for this discrepancy. One is the fact that the DfE's forecast includes nursery provision, where falling birth rates are currently being offset by rising participation rates. Nevertheless, both their analysis and ours suggest a decline of roughly 100,000 primary pupils over the next 3-5 years.

Demography is not destiny, but it certainly offers some guidance to the future for schools. This analysis suggests that primary school headcounts are falling and likely to continue to do so. In time, of course, these changes will feed into secondary schools too. Furthermore, they are affecting some schools much more than others. So if you're running a primary school, especially one that is small or has a low Ofsted rating, then brace yourself: forewarned is forearmed.

Footnotes:

1. Note that the pupil numbers shown for year-group cohorts in Figure 1 and Figure 2 are slightly lower than total enrolment numbers shown in Figure 3. This is because the year-group breakdowns are not available for all schools. But this effect is small and broadly consistent from year to year, so does not significantly affect any of the analyses presented here.
2. Even assuming flawless data collection, the number of (eg) Year 6 pupils at any given time is unlikely to be the same as the number of Year 5 pupils 12 months earlier, principally because children move in and out of the state system and the country.
3. Comparable data for 2011-2014 are not readily available so have not been analysed.
4. Our assumptions are as follows: Over the next three years, the cohorts currently in Reception to Year 3 will move through the state primary system without changing in size; during that time the current Year 4, 5 and 6 cohorts will leave primary school. These latter groups will be replaced by new Reception, Year 1 and Year 2 cohorts composed of children born between 2016 and 2018. The sizes of these new school cohorts will be in the same proportion to their corresponding birth-year cohorts as the current Reception cohort is to the group of children born in 2015 (ie, 88.5%, which is a bit lower than immediately preceding years but close to the mean for all years going back to 2004).

Hire education: What's happening in secondary school recruitment?

1st November 2019 by Timo Hannay [link]

With the turn of the academic year and the unmistakable whiff of electioneering in the air, the politically fraught topic of school resourcing is rarely far from the news. The talk is not just of money but also teachers – and sometimes even money for teachers. In particular, there are persistent concerns that it is becoming harder to hire classroom staff, especially for in-demand subjects and in certain parts of the country with tight labour markets.

This post attempts to add some objective analysis using data derived from vacancies posted on school websites. It updates our previous interim analysis using full data for the 2018-19 academic year, including year-on-year comparisons with 2017-18. We also cast our net wider, looking at trends across all major subjects areas, not just science. As before, we are delighted to be carrying out this study in collaboration with The Gatsby Foundation.

The main findings are:

• We detected 1,279 more teacher vacancy advertisements in the 2018/19 academic year than in 2017/18, representing an increase of 4.6%.
• There was considerable variation by subject area. Increases were seen in Expressive Arts (+15%), Technology (+12%), English (+9%), Maths (+7%) and Science (+3%). In contrast, there were reductions in the numbers of advertisements seen for Humanities and Social Sciences (-2%) and Languages (-5%). (For details of which classroom subjects fall into each of these areas, see Footnote 1).
• Recruiting activity also varied by region. After controlling for underlying teacher populations, the highest levels were seen in the South East and London, while the lowest were in the North East and North West.
• Advertisements for maternity, temporary or part-time positions all showed increases in the 2018-19 academic year compared to the previous year. They were most common in the Humanities, English the Expressive Arts, possibly reflecting higher proportions of female teachers in those fields. They also tended to be more prevalent in relatively affluent areas and at schools with high Ofsted ratings.
• Technician positions showed a similar overall year-on-year increase (4.8%) to teachers. These were concentrated in Science (+21%) and Technology (+10%), with Expressive Arts (-3%) showing a year-on-year reduction.
• Consistent with our previous analysis of data to March 2019, these full-year results confirm that the proportions of science teacher positions that require a specialist in biology, chemistry or physics vary greatly by school location and type. This suggests big disparities in the provision of science education across the country. We also continue to find no evidence that specialist science positions are any harder to fill than general science positions.

Further real-time information is available in our Jobs section. Users of SchoolDash Insights can also see related analyses in the 'Recruitment' and 'Staff' sections.

Ads up

In order to better understand trends in school recruiting, we have indexed the website (where one exists) of every English secondary school, sixth-form college and college of further education (FE), capturing information about any teacher and technician vacancies advertised there².

Figure 1 shows the numbers of teacher vacancy advertisements found in each subject area during the 2017-18 and 2018-19 academic years. Categorised in this way, the highest number of advertisements was in Science (about 5,500 a year) and the lowest was in Languages (about 2,000 a year). Most subject areas showed year-on-year increases in numbers of advertisements: Expressive Arts (+15%), Technology (+12%), English (+9%), Maths (+7%) and Science (+3%). However, Humanities and Social Sciences (-2%) and Languages (-5%) posted declines. The overall change was a rise from 27,877 advertisements in 2017-18 to 29,156 in 2018-19, an increase of 4.6%.

(Hover over the columns in Figure 1 and subsequent figures to see corresponding values.)

Such changes in online recruiting activity can be interpreted in a number of ways. For example, higher levels may be a sign of underlying growth or greater web usage by school administrators. Alternatively, they may indicate high staff turnover or difficulty in recruiting. Of course, these effects are not mutually exclusive and, as we shall see below, there are signs that they sometimes combine. Nevertheless, these results are consistent with the commonly held view that teachers are becoming harder to recruit and retain, as well as with certain shifts in the relative popularity of different subject areas (eg, the decline in foreign languages).

Overall increases in recruiting activity are not limited to classroom teaching positions. Figure 2 shows a similar analysis for technician vacancies. The range of subject areas covered here is narrower, being concentrated in science and technology. Both of these subject areas showed year-on-year increases (Science +21%, Technology +10%), while Expressive Arts showed a small decline (-3%). Overall, technician advertisements rose 4.8% between 2017-18 and 2018-19, very similar to the increase seen above in teacher advertisements.

Hunting season

Teacher recruiting is highly seasonal. As shown in Figure 3, year-on-year increases in online advertising activity were seen throughout the year, though there were slightly smaller proportional increases during peak season in January-June than in the off-season months. In the 2017-18 academic year, 76% of activity took place in January-June, while in 2017-18 this proportion declined slightly to 75%. With such a small change, it is hard to know whether this reflects a genuine reduction in seasonality (ie, bringing forward of some recruitment to earlier in the academic year) or simply random year-to-year variation.

Technician positions show very different seasonality (and lower numbers), as shown in Figure 4. There are slight peaks on either side of the summer break but broadly similar levels of activity at other times of the year. Here, too, increases between 2017-18 and 2018-19 were spread throughout the year.

Group work

What can we say about recruiting activity among different types of school? Figure 5 shows relative levels of online teacher recruiting activity for various school groups. Here we have corrected for the number of teachers employed at each type of school by looking not at raw numbers of advertisements but at the number per 1,000 teachers (otherwise any differences would be determined mainly by school size).

Those with the highest relative levels of activity tend to be free schools (which are typically new and growing), small schools (due to their low baseline teacher headcounts) and schools with high levels of deprivation (which presumably have more trouble and attracting and retaining staff). Conversely, grammar schools tend to show slightly lower levels of recruitment activity. Interestingly, local deprivation levels (as opposed to those in the school itself) don't seem to make much difference. All of these trends are consistent between the 2017-18 and 2018-19 academic years, suggesting that they are robust effects. Use the menu in Figure 5 to explore these and other school groupings.

Note also the presence in Figure 5 of columns on the right of each chart labelled "Multi-school groups". These represent schools that recruit as part of a consortium (either a multi-academy trust or some other inter-school alliance). Since such positions are typically not assignable to an individual school, they have been put in their own category. These schools showed a marked increase in number, albeit from a low base, doubling from 5% of all advertisements in 2017-18 to 10% in 2018-19. This suggests a considerable increase in pooled teacher recruitment activity.

Figure 6 shows a similar analysis by region. (It excludes advertisements issued by multi-school groups, which is why there appears to be no obvious overall year-on-year increase between 2017-18 and 2018-19.) There were considerably higher levels of recruitment activity in London and the South East and much lower levels in the North East and North West. The ordering is more or less consistent between academic years and presumably reflects the relative tightness and/or churn rates of labour markets in different parts of the country.

Map 1 shows the same data as in Figure 6. Use the menu to switch between 2017-18 and 2018-19 academic years. Hover over regions to see accompanying data. (The year-on-year changes are small so you may need to look carefully to see them.)

Cover to cover

Some advertisements specify maternity cover, temporary or part-time status³. Figure 7 shows the proportions in each of these categories, which ranged between roughly 4.5% and 7%. Since the categories are not mutually exclusive (eg, a given position might be both temporary and part-time), these percentages are not additive. Each group showed increases between the 2017-18 and 2018-19 academic years.

Figure 8 shows the proportions of advertisements for maternity cover, part-time and temporary positions by subject area across both academic years (2017-18 and 2018-19). Humanities and English show the highest overall levels, especially for maternity and temporary positions, while Expressive Arts showed the highest proportion of part-time positions. The variation in proportions of these positions – particularly maternity cover positions – presumably reflects at least in part the differing proportions of female teachers in each subject area.

There are also clear patterns by school type, as shown in Figure 9. Lower proportions of advertisements for maternity, part-time or temporary position were seen in schools with high percentages of disadvantaged pupils, those located in poor or urban areas, those with large proportions of pupils showing low prior attainment, small schools and those with low Ofsted ratings. Girls' schools tend to show higher proportions of these positions, consistent with the idea that teachers in those schools are more likely to be female (see our previous analysis of the School Workforce Census).

Figure 10 shows the trend by region. Overall, the South West, North West and North East show the highest proportions of maternity, temporary and part-time positions while London and the East of England show the lowest proportions.

Map 2 shows the same data as in Figure 10. Use the menu to switch between maternity, part-time and temporary positions. Hover over regions to see accompanying data.

Recruitment science

Finally, this section provides an update of our previous interim analysis of science teacher recruiting, this time using data for the full academic year.

One of the most striking findings in the earlier study was the fact that there were huge variations in the proportions of science positions that specified a specialism in biology, chemistry or physics as opposed to generalist positions covering all three disciplines. Overall, the proportions of specialist positions rose slightly from 29% in 2017-18 to 30% in 2018-19, but as shown in Figure 11, there continue to be large disparities by region (as well as by school type, though these are not shown here).

Map 3 shows the same data as in Figure 11. Once again, use the menu to switch between 2017-18 and 2018-19 academic years, and hover over regions to see accompanying data.

Another important finding from the previous analysis was the fact that repeat advertising rates were no higher for specialist science positions than for generalist science positions, which suggests that they are no harder to fill. Figure 12 shows full aggregate results for the 2017-18 and 2018-19 academic years. (If anything, repeat advertising rates seem lower for specialist positions, though this outcome needs to be interpreted with care – see Footnote 5). Note also that repeat advertising rates for both specialist and generalist science positions vary greatly by region, being relatively high in London and the South East but much lower in the North East, North West and South West.

Annual appraisal

We see again in these full-year results that the market for secondary school teachers (and technicians) in England is far from monolithic. There are very distinct patterns by subject, school type and location, not to mention time of year. Evidence of disparities and challenges abound, but in the age of the web these vacancies are visible not just to potential applicants but also those of us trying to understand the system as a whole. We hope that these findings prove useful to those in a position to help make it work better.

Footnotes:

1. 'Expressive Arts' include Art, Drama and Music. 'Technology' includes Design & Technology and Computing. 'Humanities and Social Sciences' include Geography, History, Economics, Sociology and Religious Studies. 'Other' subjects include Physical Education, Business Studies and Media Studies.
2. This analysis is based on data gathered by indexing school websites every weekday to search for any new vacancies advertised. This process does not capture all vacant positions because: (a) not all positions are advertised on school websites, (b) even when they are, they are not necessarily presented in a way that can be automatically indexed, and (c) websites are sometimes unresponsive or otherwise unavailable. For this reason, the data presented should be thought of as being based not on a comprehensive list of all vacancies but on a subset. However, positions were detected for well over 90% of schools and the representation of different subgroups was generally consistent with the overall population of schools, so there are no obvious reasons to suspect systematic biases. This means that any variations, such as differences between subjects or changes over time, are likely to represent genuine underlying trends. All data presented in this analysis are from mainstream state secondary schools, sixth-form colleges and FE colleges in England. Primary schools, special schools, independent schools and establishments outside England have been omitted.
3. These were defined as positions for which the part-time, temporary or maternity status was explicitly stated within or close to the job title. It does not include positions where this was specified elsewhere or not included at all in the online information.
4. Consistent with our previous analysis of science teacher recruiting, repeat advertisements are defined as new advertisements that overlap with the subject area of an earlier advertisement for the same school which appeared at least 21 days but no more than 90 days previously. This method almost certainly creates both false positives and false negatives, but manual inspection suggests that the results are in line with classifications made by human experts.
5. It is important to qualify this finding. Because general science positions are more numerous than biology, chemistry and physics positions, it is more likely that we will overestimate the repeat rate for the former. For a longer explanation, see the relevant footnote in our earlier blog post.