Education, Occupational Class, and Unemployment in the Regions of the United Kingdom. Vani K. Borooah * University of Ulster.

Transcription

1 Educaton, Occupatonal Class, and Unemployment n the Regons of the Unted Kngdom Van K. Borooah * Unversty of Ulster January 2007 Abstract Estmatng the returns to educaton s an mportant aspect of emprcal economcs. Usually ths s acheved by estmatng the addtonal earnngs provded by an extra year of schoolng. However, gven the dffculty of obtanng relable earnngs data, ths approach s not always easy to mplement. Ths paper proposes an alternatve measure of returns to educaton based on the probablty of labour market success assocated wth dfferent levels of qualfcaton. Returns to educaton, so conceved, are estmated on data from the 2001 UK Census for the dfferent regons of the UK. Two measures of success are used: frst, the lkelhood of persons n employment beng n good jobs; second, the lkelhood of persons n the labour force beng n employment. The results show that, n every regon of the UK, better qualfcatons are sgnfcantly and strongly assocated wth hgher probabltes of labour market success. * School of Economcs and Poltcs, Unversty of Ulster, Newtownabbey, Northern Ireland BT37 0QB, Unted Kngdom. E-mal: VK.Borooah@ulster.ac.uk. I am grateful to the Department for Employment and Learnng, Northern Ireland for supportng ths project and to the Cathy Marsh Centre for Census and Survey Research at the Unversty of Manchester for the use of data from the Sample of Anonymsed Records of the 2001 Census for the Unted Kngdom of Great Brtan and Northern Ireland. However, needless to say, I am entrely responsble for the paper and, n partcular, for ts shortcomngs.

2 1. Introducton Conventonal wsdom has t that educaton s the handmaden of prosperty. Indeed, one of the most popular textbooks n economcs (Mankw, 1998) asserts that Educaton s at least as mportant as nvestment n physcal captal for a country s long-term success.. one way n whch government polcy can enhance the standard of lvng s to provde good schools and encourage the populaton to take advantage of them (p. 528). Ths theoretcal underpnnng now forms the rason d etre of government sklls polcy; at the natonal level the Letch Revew s consderng what the UK s long-term ambton should be for developng sklls n order to maxmse economc prosperty and productvty. In Northern Ireland the Department for Learnng and Employment n ts Sklls Strategy seeks to provde an overarchng framework for the development of the local sklls base. Gven the mportance of educaton n determnng a person s materal well-beng, t s natural to enqure about the returns to educaton. In partcular, s t possble to quantfy the extent to whch a person s made better off through more educaton? Economsts, begnnng wth Mncer (1958), have sought to answer ths queston by estmatng the net effect of years of schoolng on the earnngs of ndvduals. The recognton that not all the addtonal earnngs of better educated workers can be ascrbed to ther superor educaton has led to controls for the noneducatonal characterstcs of ndvduals beng mposed before estmatng ther returns to educaton. These controls may relate to age, gender, place of resdence, ndvdual ablty or motvaton. Based on a survey of 97 studes that estmated returns to educaton, Ashenfelter et. al. (1999) concluded that the return to an addtonal years schoolng was between 6 and 9 percent. 1

3 However, such studes, relatng earnngs to years of schoolng, suffer from two problems. 1 Frst, there s a dearth of relable data on earnngs. Second, t s qualfcatons, rather than years of schoolng, whch are the passport to hgher earnngs; acqurng qualfcatons requres not just spendng tme n an educatonal nsttuton but also havng somethng to show for the tme spent. In the face of these problems, we develop an alternatve concept of returns to educaton. In essence, ths nvolves estmatng the amount by whch the probablty of labour market success ncreases wth hgher educatonal qualfcatons. We use two concepts of labour market success : for persons who are employed, success means dong a hgh, rather than a low, occupatonal status job; for persons n the labour force, success means beng employed rather than unemployed. Data from the 2001 Census for the UK were used to put emprcal flesh on ths conceptual skeleton. The Sample of Anonymsed Records (SARS) of the 2001 Census for the Unted Kngdom reported the hghest qualfcaton for 1,031,840 persons between the ages of years. In addton, the SARs nter ala also provded detals of a person s non-educatonal characterstcs: gender, regon of resdence, age, employment status, occupaton, ethncty, place of brth. Usng ths nformaton we estmated the probabltes of: () employed ndvduals beng n jobs of dfferent occupatonal types; () persons n the labour force beng unemployed. Ths allowed us to estmate, after mposng the approprate non-educatonal controls, the lnk between dong well n the labour market and educatonal qualfcatons. 2. Educaton Qualfcatons and the 2001 Census for the UK 1 See Dutta et.al. (1999) and Harmon et.al. (2003) for surveys of the lterature on returns to educaton. 2

4 As mentoned earler, the SARS of the 2001 Census for the Unted Kngdom reported the hghest qualfcaton for 1,031,840 persons between the ages of years. The qualfcatons were reported as: () no qualfcatons; () Level 1 qualfcatons (roughly, at least one GCSE); () Level 2 qualfcatons (roughly, 5+ GCSEs at A-C grades); () Level 3 qualfcatons (roughly, 2+ A levels); (v) Level 4 qualfcatons (frst degree or hgher, HNC, HND). These qualfcaton levels are detaled n the notes to Table 1. 2 Table 1 shows the percentage of respondents wth these qualfcatons n the dfferent UK regons. So, for example, of the 30,482 Northern Ireland respondents aged 16-74, 38 percent had no qualfcatons whle 18 percent had Level 4 qualfcatons. The most strkng feature of Table 1 s the marked dsparty between the north and the south of the Unted Kngdom n the proporton of respondents wthout any qualfcatons. Approxmately, 30 percent of the respondents n the North East, the North West, the East and West Mdlands, Wales, and Scotland had no qualfcatons compared to around 22 percent for the East, the South East, the South West, and nner and outer London. Of all the regons of the Unted Kngdom, Northern Ireland had the hghest proporton of respondents (38 percent) wth no qualfcatons. At the other end of the qualfcatons spectrum, Table 1 shows that 44 percent of respondents n nner London, 31 percent n outer London, and 27 percent n the South East had Level 4 qualfcatons. At the other extreme, only 18 percent of respondents n Northern Ireland and n the North East had level 4 qualfcatons. In 2 It should be noted that these were the levels reported at the tme that the 2001 Census was undertaken. As from September 2004 there has been a movement towards the Natonal Qualfcatons Framework (NQF) whch comprses nne levels (Entry level to level 8). Whlst entry level and levels 1 to 3 dd not change, levels 4 to 5 have been dvded nto the more precse levels 4 to 8. Whle a qualfcaton ttle mght have changed to reflect the more precse levels, the qualfcaton content and demand on the learner has not changed. 3

5 the other regons the proporton of respondents wth Level 4 qualfcatons was farly equal at approxmately 21 percent. Tables 2 and 3 show that when the upper age lmt of respondents was reduced to, respectvely, 45 and 30 years the proporton of respondents wth no qualfcatons fell dramatcally: the proporton of respondents n Northern Ireland wthout any qualfcatons fell from 38 percent (Table 1: aged 16-74) to 24 percent (Table 2: aged years), to 12.9 percent (Table 3: aged years). At the other extreme, reducng the upper age lmt also ncreased the proporton of respondents wth Level 4 qualfcatons though, compared to proportons wth no qualfcatons, the results were much more muted: for example, the proporton of respondents n Northern Ireland wth Level 4 qualfcatons rose from 18 percent (Table 1: aged 16-74) to 20.4 percent (Table 2: aged years), to 21.3 percent (Table 3: aged years). Table 4 shows for, Northern Ireland, dfferences between the proportons of Catholcs and Protestants wth dfferent qualfcatons: the assocated t values provde a test of whether these nter-communty dfferences were sgnfcantly dfferent from zero. Whle there was no sgnfcant dfference between the proportons of Catholcs and Protestant respondents, taken n ther entrety (.e ), wthout any qualfcatons, the proporton of Catholc respondents wthout any qualfcatons was sgnfcantly hgher than that for Protestants for the truncated age groups, and years. However, at the other end of the qualfcatons spectrum, the proporton of Catholc respondents wth Level 4 qualfcatons was sgnfcantly hgher than that for Protestants for all the age groups: 16-74, 16-45, and years. Compared to Protestants, Catholcs were more lkely to be wthout qualfcatons but also more lkely to have the hghest level of qualfcatons. 4

6 It s mportant that the data n Table 4 s set wthn the context of Hgher Educaton mgraton n Northern Ireland. The emprcal evdence tends to suggest that a hgher proporton of Protestants mgrate to Great Brtan (GB) for HE purposes wth a sgnfcant proporton not returnng. Ths seems a strong explanatory factor n explanng the fndng that Catholcs have the hghest levels of qualfcaton. Table 7 shows the relatonshp between qualfcatons and occupatonal class. For example, n Northern Ireland, only 11.9 percent of respondents wthout any qualfcatons, compared to 73 percent of respondents wth Level 4 qualfcatons, were n manageral, professonal or techncal (PMT) occupatons; on the other hand, 25 percent of those wthout any qualfcatons, compared to only 2.5 percent of those wth Level 4 qualfcatons, were n elementary occupatons. Table 7 also shows that the proporton of persons wth Level 4 qualfcatons n PMT occupatons n all the other regons of the Unted Kngdom (approxmately, 77 percent) was (statstcally) sgnfcantly hgher than n Northern Ireland (73 percent): compared to other UK regons, persons wth Level 4 qualfcatons n Northern Ireland were underrepresented n PMT occupatons and over-represented n sklled manual/non-manual occupatons. Table 8 shows the relatonshp between qualfcatons and two states of economc status: unemployment and beng permanently sck or dsabled (PSD). In Northern Ireland, 49.5 percent of unemployed persons (seekng work and avalable to start wthn 2 weeks) and 75.1 percent of the PSD had no qualfcatons; conversely, only 9.6 percent of those unemployed and only 4.5 percent of the PSD, n Northern Ireland, had Level 4 qualfcatons. In ths respect, Northern Ireland s experence contrasts wth that of the other UK regons where persons wth Level 4 qualfcatons were proportonately more strongly represented among the unemployed and the PSD: 5

7 for example, n nner London, the proporton of the unemployed who had no qualfcatons was, at 26 percent, the same as the proporton of the unemployed who had Level 4 qualfcatons. Table 9 shows unemployment rates (the proporton of respondents n the labour force seekng work and avalable to start wthn 2 weeks) and the sckness/dsablty rates (proporton of the non-student, non-retred populaton whch s permanently sck or dsabled) by hghest level of qualfcaton. Ths Table shows that, n every UK regon, better qualfcatons were assocated wth lower unemployment and dsablty rates. The unemployment rate for unqualfed persons was sgnfcantly above the regonal average and the unemployment rate for those wth Level 4 qualfcatons was sgnfcantly below. For example, n the context of an overall unemployment rate of 6.7 percent n Northern Ireland, the unemployment rate n that regon, for respondents wthout any qualfcatons, was 11.5 percent compared to a rate of 2.9 percent for persons wth Level 4 qualfcatons. Comparng LFS and Census data There s a rather large dscrepancy between the Labour Force Survey (LFS) and the Census n relaton to sklls reportng. Undoubtedly the Census of Populaton offers the most accurate recent baselne for numbers of people wth each skll. However, the fact remans that Census fgures show a sgnfcantly dfferent dstrbuton of sklls from the LFS. For example, n the Census a lower proporton of people have level 4/5 sklls (15.8% compared wth 18.5% n the LFS n 2001). A much hgher proporton have level 1 or o sklls (58% compared wth 36.5% n the LFS). The largest dfference of all s n Level 3 sklls, held by only 9% of workngage people n the census compared wth 23% n the LFS for These dfferences 6

8 are large and are not due to samplng fluctuatons. The Northern Ireland Statstcal Research agency (NISRA) s currently nvestgatng the reasons behnd the dfferences: they have suggested that t may be due to the fact that respondents to the LFS are ntervewed by an experenced ntervewer, qualfcatons are more correctly assgned to the relevant sklls level. On the other hand respondents to the Census are completng the forms by themselves and therefore they may ncorrectly categorse ther qualfcatons. It should be emphassed that ths Report s entrely Census based and does not, therefore, seek to make any such comparsons. 3. Econometrc Methodology: Ordered Logt Suppose there are N persons (ndexed =1 N). Let the values taken by the varable Y represent the occupatonal status of these persons such that: Y = 1 f the person was employed n a PMT occupaton; Y = 2 f the person was employed n a sklled manual or non-manual occupaton; and Y = 3 f the person was an elementary worker. Snce these outcomes are nherently ordered n the sense that the outcome assocated wth a hgher value of Y s of lower occupatonal status than that assocated wth a lower value the approprate method of estmaton s that of ordered logt. The dea behnd ths model (Borooah, 2001) s that the occupatonal status of a person may be represented by the value of the latent varable, H, wth hgher values of H representng lower status. One may consder ths latent varable to be a lnear functon of K status-determnng factors whose values for ndvdual are: X, k = 1... K. Consequently, k 7

9 K H = X β + ε = Z + ε (1) k k k = 1 where: β s the coeffcent assocated wth the k th k varable and Z Xkβk k =. An ncrease n the value of the β < 0 and to deterorate f β > 0. k Snce the values of th k factor wll cause the status of a person to mprove f k H are, n prncple and n practce, unobservable, equaton (1) represents a latent regresson whch, as t stands, cannot be estmated. However, what s observable s a person s occupatonal status and the categorsaton of persons n the sample n terms of occupatonal status s mplctly based on the values of the latent varable < ) such that: ( δ 1 δ 2 H n conjuncton wth threshold values, δ 1 and δ 2 Y Y Y = 1, f H δ = 2, f δ < H δ = 3, f H > δ 2 (2) The δ1, δ 2 of equaton (2) are unknown parameters to be estmated along wth the β k of equaton (1). A person s classfcaton n terms of hs/her occupatonal status depends upon whether the value of a partcular occupatonal status are: H crosses a threshold and the probabltes of a person beng n Pr( Y = 1) = Pr( ε δ Z ) 1 Pr( Y = 2) = Pr( δ Z ε < δ Z ) 1 2 Pr( Y = 3) = Pr( ε δ Z ) 2 (3) If t s assumed that the error term ε, n equaton (1), follows a logstc dstrbuton then equatons (1) and (2) collectvely consttute an ordered logt model 3 3 The assumpton that the ε are normally dstrbuted results n an ordered probt model. 8

10 and the estmates from ths model permt, through equaton (3), the varous probabltes to be computed for every person n the sample, condtonal upon the values of the status-determnng factors for that person. The ordered logt estmates for estmatng equatons (1) and (2), wth Y as the dependent varable (defned above), are shown n Table 10 for each of seven broad areas of the UK: the North, North West and Yorkshre; the East and West Mdlands; the East, South East and South West; nner and outer London; Wales; Scotland; and Northern Ireland. The equaton was estmated on data for all persons who were n employment, ether as employees or as self-employed. The estmated parameters ˆk β, ˆ δ 1 and ˆ δ 2 maxmse the lkelhood of observng the values of the dependent varables, condtonal upon the values assumed by the determnng varables. The z-ratos, shown parenthetcally n Table 10, are the ratos of the estmated coeffcents to ther estmated standard errors: the z-ratos are (asymptotcally) dstrbuted as N(0,1) under the null hypothess that the assocated coeffcents are zero. 4. Estmaton Results: Margnal Probabltes of Occupatonal Status A natural queston to ask from the ordered logt model s how the probabltes of beng n the dfferent categores of occupatonal status changed n response to a change n the value of one of the status-affectng factors, the values of the other factors remanng unchanged. Unfortunately, the coeffcent estmates themselves do not provde a clear answer to ths queston. If β k < 0, then, n response to an ncrease n the value of the k th determnng factor, Pr(Y =1) wll rse and Pr(Y =3) wll fall. However, snce the change n probabltes across all three outcomes must sum to zero, t s not clear what would happen to the mddle probablty, Pr(Y =2): t may rse or fall. Gven a change n the value of a determnng varable, t s mpossble, 9

11 therefore, to nfer, from the sgn of ts coeffcent estmate, the drecton of change n all the probabltes. For ths reason Greene (2000) cautons that we must be very careful n nterpretng the coeffcents n ths model...snce t s the least obvous of the [dscrete choce] models" (p. 878). For ths reason, the estmaton results are dscussed n terms of margnal probabltes. The margnal probablty, assocated wth a determnng varable, of beng employed n a partcular occupaton, s the change n the probablty of employment n that occupaton consequent upon an unt change n the determnng varable, the values of the other varables remanng unchanged. For dscrete varables, the margnal probabltes refer to changes consequent upon a move from the resdual category for that varable to the category n queston. Table 11 shows the margnal probablty of beng employed n a PMT job assocated wth a unt change n each of the determnng varables. For example, Table 11 shows that n Northern Ireland a change n qualfcatons from no qualfcatons to Level 1 qualfcatons ncreases the probablty of employment n a PMT job by 18.3 ponts; a change n qualfcatons from no qualfcatons to Level 4 qualfcatons ncreases the probablty of employment n a PMT job by 68.2 ponts; and, by corollary, a change n qualfcatons from Level 3 to Level 4 qualfcatons ncreases the probablty of employment n a PMT job by 30.3 ( ) ponts. Table 11 shows that, for employed persons, the probablty of PMT employment rses wth hgher qualfcatons: across the UK regons, the margnal probabltes of beng n PMT employment, for Level 4 qualfcatons, were between approxmately 66 (lowest)-70 (hghest); the correspondng ranges for the PMT margnal probabltes for Level 3, Level 2, and Level 1 qualfcatons were, respectvely, 36-45, 30-36, and ponts. 10

12 An mportant feature of Table 11 s that n every UK regon, the margnal probablty of women beng n PMT employment was negatve: ceters parbus women, compared to male, employees were less lkely to be n PMT employment. Table 11 also shows that the margnal probablty of beng n PMT employment rose wth age, reached a maxmum at years, and then fell. Asan and Black employees n the Mdlands, the South, and n London were less lkely to be n PMT employment ceters parbus than Whtes; however, there dd not appear to be an ethnc effect (except a negatve effect for Blacks n the North, North West, and Yorkshre) assocated wth PMT employment n the other UK regons. It should be noted that ssues of ethncty were not relevant for Northern Ireland where the sample was, almost exclusvely, Whte. 5. Estmaton Results: Margnal Probabltes of Unemployment The precedng secton addressed the queston of the occupatonal status of persons who were employed. A related queston of nterest s how the probablty of a person, who was n the labour force, beng unemployed was nfluenced by hs/her personal characterstcs and crcumstances. In order to answer ths queston we estmated a logt model n whch the dependent varable Y =1 f person was unemployed and Y =0 f person was employed (employee or self-employed). 4 The results of estmatng ths equaton for the dfferent regons of the UK are shown n Table 12. A postve (negatve) coeffcent estmate ndcates that the probablty of beng unemployed rses (falls) wth an ncrease n the value of the varable assocated wth the coeffcent. However, the coeffcent estmates do not provde a gude to the 4 K Pr( Y = 1) The logt equaton s = exp{ X k β} = exp{ z} for K coeffcents, β and for 1 Pr( Y = 1) k = 1 z z observatons on K varables where Pr( Y = 1) = e /(1 + e ) 11

13 amount by whch the probablty of unemployment ncreases or decreases n consequence of a change n the varable value. These margnal probabltes are shown n Table In the present context, the margnal probablty, assocated wth a determnng varable, of beng unemployed s the change n the probablty of unemployment consequent upon an unt change n the determnng varable, the values of the other varables remanng unchanged. For dscrete varables, the margnal probabltes refer to changes consequent upon a move from the resdual category for that varable to the category n queston. Table 13 shows that, for persons n the labour force, the margnal probablty of unemployment falls wth hgher qualfcatons: across the UK regons, the margnal probabltes of beng unemployed, for Level 4 qualfcatons, were between -2.6 (lowest) to -6.7 (hghest). The equaton predcted that, after controllng for noneducatonal factors, the probablty of an unqualfed person n Northern Ireland beng unemployed was 10.5 percent. So, for example, compared to an unqualfed person n the Northern Ireland labour force beng unemployed, the probabltes of persons wth Level 4, Level 3, Level 2, and Level 1 qualfcatons beng unemployed were, respectvely, 6.2, 4.6, 4.1, and 3.3 percentage ponts lower than 10.5 percent. The probablty of beng unemployed also vared by age. Compared to the youngest age group, years, the margnal probablty of unemployment for the other age groups was always negatve, ndcatng that they were less lkely to be unemployed - beng most negatve for persons n the labour force between years. Table 13 also shows that, except n Wales, there was a sgnfcant ethnc dmenson to unemployment: compared to Whte persons, the probablty of beng 5 The margnal probablty s defned as Pr( Y = 1) X k 12

14 unemployed was hgher for Asans, Chnese, and Blacks. In Northern Ireland, the probablty of a Catholc beng unemployed was 3.3 ponts hgher than for Protestants. Lastly, n every regon of the UK, the probablty of women n the labour force beng unemployed was always lower than that for men. 6. Conclusons Estmatng the returns to educaton s an mportant aspect of emprcal economcs. Usually ths s acheved by estmatng the addtonal earnngs provded by an extra year of schoolng. However, as noted earler, gven the dffculty of obtanng relable earnngs data, ths approach s not always easy to mplement. Ths paper proposed an alternatve measure of returns to educaton based on the probablty of labour market success assocated wth dfferent levels of qualfcaton. Returns to educaton, so conceved, were estmated on data from the 2001 UK Census for the dfferent regons of the UK. Two measures of success were used: frst, the lkelhood of persons n employment beng n good jobs; second, the lkelhood of persons n the labour force beng n employment. The results showed that, n every regon of the UK, better qualfcatons were sgnfcantly, and strongly, assocated wth hgher probabltes of labour market success. Aganst ths general concluson, we may dstngush dfferent patterns of nterregonal detals. 1. There s the marked dsparty between the north and the south of the Unted Kngdom n the proporton of respondents wthout any qualfcatons. Approxmately, 30 percent of the respondents n the North East, the North West, the East and West Mdlands, Wales, and Scotland had no qualfcatons compared to around 22 percent for the East, the 13

15 South East, the South West, and nner and outer London. Of all the regons of the Unted Kngdom, Northern Ireland had the hghest proporton of respondents (38 percent) wth no qualfcatons. At the other end of the qualfcatons spectrum, Table 1 shows that 44 percent of respondents n nner London, 31 percent n outer London, and 27 percent n the South East had Level 4 qualfcatons. At the other extreme, only 18 percent of respondents n Northern Ireland and n the North East had level 4 qualfcatons. In the other regons the proporton of respondents wth Level 4 qualfcatons was farly equal at approxmately 21 percent. 2. The proporton of persons wth Level 4 qualfcatons n PMT occupatons n all the other regons of the Unted Kngdom (approxmately, 77 percent) was (statstcally) sgnfcantly hgher than n Northern Ireland (73 percent): compared to other UK regons, persons wth Level 4 qualfcatons n Northern Ireland were under-represented n PMT occupatons and over-represented n sklled manual/non-manual occupatons. 3. In Northern Ireland, 49.5 percent of unemployed persons (seekng work and avalable to start wthn 2 weeks) and 75.1 percent of the PSD had no qualfcatons; conversely, only 9.6 percent of those unemployed and only 4.5 percent of the PSD, n Northern Ireland, had Level 4 qualfcatons. In ths respect, Northern Ireland s experence contrasts wth that of the other UK regons where persons wth Level 4 qualfcatons were proportonately more strongly represented among the unemployed and the PSD: for example, n nner London, the proporton of the unemployed who 14

16 had no qualfcatons was, at 26 percent, the same as the proporton of the unemployed who had Level 4 qualfcatons. 4. In every UK regon, better qualfcatons were assocated wth lower unemployment rates and low dsablty rates. The unemployment rate and the dsablty rate for unqualfed persons was sgnfcantly above the regonal average and the unemployment rate for those wth Level 4 qualfcatons was sgnfcantly below. For example, n the context of an overall unemployment rate of 6.7 percent n Northern Ireland, the unemployment rate n that regon, for respondents wthout any qualfcatons, was 11.5 percent compared to a rate of 2.9 percent for persons wth Level 4 qualfcatons. 5. Ceters parbus the probablty of PMT employment rose wth hgher qualfcatons: across the UK regons, the margnal probabltes of beng n PMT employment, for Level 4 qualfcatons, were between approxmately 66 (lowest)-70 (hghest); the correspondng ranges for the PMT margnal probabltes for Level 3, Level 2, and Level 1 qualfcatons were, respectvely, 36-45, 30-36, and ponts. 6. The margnal probablty of women beng n PMT employment was negatve: ceters parbus women, compared to male, employees were less lkely to be n PMT employment. The margnal probablty of beng n PMT employment rose wth age, reached a maxmum at years, and then fell. 7. For persons n the labour force, the margnal probablty of unemployment falls wth hgher qualfcatons: across the UK regons, the margnal probabltes of beng unemployed, for Level 4 qualfcatons, were 15

17 between -2.6 (lowest) to -6.7 (hghest). The equaton predcted that, after controllng for non-educatonal factors, the probablty of an unqualfed person n Northern Ireland beng unemployed was 10.5 percent. So, for example, compared to an unqualfed person n the Northern Ireland labour force beng unemployed, the probabltes of persons wth Level 4, Level 3, Level 2, and Level 1 qualfcatons beng unemployed were, respectvely, 6.2, 4.6, 4.1, and 3.3 percentage ponts lower than 10.5 percent. 8. The probablty of beng unemployed also vared by age. Compared to the youngest age group, years, the margnal probablty of unemployment for the other age groups was always negatve, ndcatng that they were less lkely to be unemployed - beng most negatve for persons n the labour force between years. 16

18 References Ashenfelter, O., Harmon, C., and Oosterbeek, H. (1999), A Revew of Estmates of the Schoolng/Earnngs Relatonshp, wth Tests for Publcaton Bas. Labour Economcs, Vol. 6, pp Borooah, V.K. (2001), Logt and Probt: Ordered and Multnomal Models, Quanttatve Studes n the Socal Scences, Thousand Oaks, CA: Sage Publcatons Inc. Dutta, J., Sefton, J., and Weale, M. (1999), Educaton and Publc Polcy, Fscal Studes, vol. 20, pp Harmon, C., Oosterbeek, H. and Walker, I. (2003) The Returns to Educaton A Revew of Evdence, Issues and Defcences n the Lterature. (wth). Journal of Economc Surveys, vol. 17, pp Hornbeck, D.W., and Salamon, L.M. (1991), Human Captal and Amerca s Future, Batlmore: The Johns Hopkns Press. Greene, W.H., (2000), Econometrc Analyss, Englewood Clffs: Prentce Hall. Mankw, N.G. (1998), Prncples of Economcs, New York: Worth Publshers Inc. Mncer, J. (1958), Investment n Human Captal and Personal Income Dstrbuton, Journal of Poltcal Economy, vol. 66, pp Pras, S.J. (1995), Productvty, Educaton and Tranng, London: Natonal Insttute of Economc and Socal Research. 17

19 Table 1 Level of Hghest Qualfcatons by UK Regon, 2001, aged Percentage of Respondents, aged 16-74, wth relevant qualfcaton No Level 1 Level 2 Level 3 Level 4 Qualfcatons North East (43,202) North West (114,879) Yorkshre (84,138) East Mdland (71,878) West Mdland (89,649) East (92,521) South East (138,810) South West (83,081) Inner London (54,751) Outer London (79,533) Wales (49,118) Scotland (99,798) Northern Ireland (30,482) England, Wales, and Northern Ireland (932,042) Source: 2001 UK Census, Sample of Anonymsed Records 18

20 Notes to Table 1 Fgures n parentheses are numbers n sampled n regon Educaton levels are defned as follows: England and Wales: Level 1: 1+ O level passes, 1+ CSE/GCSE any grades, NVQ level 1, Foundaton GNVQ Level 2: 5+ O level passes, 5+ CSE (grade 1), 5+ GCSE (A-C grades), School Certfcate, 1+ A levels/as levels, NVQ level 2, Intermedate GNVQ Level 3: 2+ A levels, 4+ AS levels, Hgher School Certfcate, NVQ level3, Advanced GVNQ Level 4: Frst degree, Hgher degree, NVQ levels 4 and 5, HNC, HND, Qualfed Teacher status, Qualfed Medcal Doctor/Dentst/Nurse/Mdwfe/Health Vstor Scotland: Level 1: O grade, Standard Grade, Intermedate 1, Intermedate 2, Cty and Gulds Craft, SVQ level 1 or 2 or equvalent Level 2: Hgher Grade, CSYS, ONC, OND, Cty and Gulds Advanced Craft, RSA, Advanced Dploma, SVQ level 3 or equvalent Level 3: HND, HNC, RSA Hgher Dploma, SVQ level 4 or 5 or equvalent Level 4: Frst degree, Hgher degree, Professonal qualfcaton Northern Ireland: Level 1: GSCE (grades D-G), CSE (grades 2-5), 1-4 CSEs (grade 1), 1-4 GCSES (grades A-C), 1-4 O level passes, NVQ level 1, Foundaton GNVQ or equvalents Level 2: 5+ O level passes, 5+ CSE (grade 1), 5+ GCSE (A-C grades), Senor Certfcate, 1 A levels, 1-3 AS levels, Advanced Senor Certfcate, NVQ level 2, Intermedate GNVQ or equvalents Level 3: 2+ A levels, 4+ AS levels, NVQ level 3, GNVQ Advanced or equvalents Level 4: Frst degree, Hgher degree, NVQ levels 4 and 5, HNC, HND In England and Wales, the hghest level of qualfcaton was derved from responses to both the qualfcatons qualfcaton and the professonal qualfcatons queston; n Scotland and Northern Ireland, t was based only on the qualfcatons queston. Source: 2001 UK Census, Sample of Anonymsed Records 19

21 Table 2 Level of Hghest Qualfcatons by UK Regon, 2001, aged Percentage of Respondents, aged 16-45, wth relevant qualfcaton No Level 1 Level 2 Level 3 Level 4 Qualfcatons North East (26,752) North West (70,769) Yorkshre (52,406) East Mdland (44,051) West Mdland (55,004) East (56,697) South East (85,647) South West (49,395) Inner London (40,882) Outer London (52,955) Wales (29,073) Scotland (59,761) Northern Ireland (19,269) England, Wales, and Northern Ireland (582,900) See Notes to Table 1 Source: 2001 UK Census, Sample of Anonymsed Records 20

22 Table 3 Level of Hghest Qualfcatons by UK Regon, 2001, aged Percentage of Respondents, aged 16-30, wth relevant qualfcaton No Level 1 Level 2 Level 3 Level 4 Qualfcatons North East (9,943) North West (26,950) Yorkshre (20,315) East Mdland (16,477) West Mdland (21,122) East (21,028) South East (31,650) South West (18,271) Inner London (18,355) Outer London (20,819) Wales (11,237) Scotland (22,424) Northern Ireland (7,395) England, Wales, and Northern Ireland (223,562) See Notes to Table 1 Source: 2001 UK Census, Sample of Anonymsed Records 21

23 Table 4 Level of Hghest Qualfcatons n Northern Ireland by Relgon *, 2001 Percentage of Respondents, aged 16-74, wth relevant qualfcaton No Level 1 Level 2 Level 3 Level 4 Qualfcatons Catholc (12,015) Protestant (13,845) t value Percentage of Respondents, aged 16-45, wth relevant qualfcaton Catholc (7,986) Protestant (8,077) t value Percentage of Respondents, aged 16-30, wth relevant qualfcaton Catholc (3,205) Protestant (2,920) t value * Only Catholcs and Protestants (Presbyteran, Church of Ireland, Methodst, Other Chrstan). Persons of other relgons, no relgon, and not stated relgon were excluded. See Notes to Table 1. The t values result from testng the null hypothess of there beng no dfference between the Catholc and Protestant mean proportons at the relevant qualfcaton. Source: 2001 UK Census, Sample of Anonymsed Records 22

24 Table 5 Level of Hghest Qualfcatons n England and Wales by Ethncty, 2001 Percentage of Respondents, aged 16-74, wth relevant qualfcaton No Level 1 Level 2 Level 3 Level 4 Qualfcatons Whte (822,750) Mxed (7,859) Asan (40,720) Black (20,917) Chnese (9,314) Percentage of Respondents, aged 16-45, wth relevant qualfcaton Whte (504,189) Mxed (6,849) Asan (29,726) Black (16,085) Chnese (6,782) Percentage of Respondents, aged 16-30, wth relevant qualfcaton Whte (189,639) Mxed (3,435) Asan (14,554) Black (5,367) Chnese (3,172) See Notes to Table 1 Source: 2001 UK Census, Sample of Anonymsed Records 23

25 Whte (97,849) Indan (351) Pakstan (1,108) Chnese (153) Whte (58,268) Indan (262) Pakstan (859) Chnese (117) Table 6 Level of Hghest Qualfcatons n Scotland by Ethncty, 2001 Percentage of Respondents, aged 16-74, wth relevant qualfcaton No Qualfcatons Level 1 Level 2 Level 3 Level Percentage of Respondents, aged 16-45, wth relevant qualfcaton Percentage of Respondents, aged 16-30, wth relevant qualfcaton Whte (21,712) Indan (129) Pakstan (407) Chnese (56) See Notes to Table 1 Source: 2001 UK Census, Sample of Anonymsed Records

26 North & Yorkshre Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers East and West Mdlands Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers East, South- East, South- West Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers London, Inner & Outer Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers Table 7 Occupaton and Level of Hghest Qualfcaton n UK Regons Percentage of persons wth relevant qualfcaton n occupatonal class No Qualfcatons Level 1 Level 2 Level 3 Level

27 Scotland Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers Wales Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers Northern Ireland Manageral, Professonal, Techncal Sklled manual, nonmanual Elementary workers England, Wales, and Northern Ireland Manageral, Professonal, Techncal Sklled manual, non-manual No Qualfcatons Level 1 Level 2 Level 3 Level Elementary workers Source: 2001 UK Census, Sample of Anonymsed Records 26

28 Table 8 Unemployment and Permanently Sck or Dsabled and Level of Hghest Qualfcaton n UK Regons, Non-students Aged Percentage of persons n economc status wth relevant qualfcaton No Qualfcatons Level 1 Level 2 Level 3 Level 4 North & Yorkshre Unemployed Permanently sck or dsabled East and West Mdlands Unemployed Permanently sck or dsabled East, South- East, South- West Unemployed Permanently sck or dsabled London, Inner & Outer Unemployed Permanently sck or dsabled Scotland Unemployed Permanently sck or dsabled Wales Unemployed Permanently sck or dsabled Northern Ireland Unemployed Permanently sck or dsabled England, Wales, and Northern Ireland Unemployed Permanently sck or dsabled Source: 2001 UK Census, Sample of Anonymsed Records 27

29 Table 9 Unemployment Rates and Dsablty Rates by Level of Hghest Qualfcaton n UK Regons Percentage of persons wth relevant qualfcaton n the labour force who are unemployed or permanently sck/dsabled North & Yorkshre Unemployment Rate Permanently sck/dsabled rate East and West Mdlands Unemployment Rate Permanently sck/dsabled rate East, South- East, South- West Unemployment Rate Permanently sck/dsabled rate London, Inner & Outer Unemployment Rate Permanently sck/dsabled rate Scotland Unemployment Rate Permanently sck/dsabled rate Wales Unemployment Rate Permanently sck/dsabled rate No Qualfcatons Level 1 Level 2 Level 3 Level 4 All respondents

30 Northern Ireland Unemployment Rate Permanently sck/dsabled rate England, Wales and Northern Ireland Unemployment Rate Permanently sck/dsabled rate Unemployment rate: proporton of the labour force that s unemployed and avalable to start work wthn 2 weeks. Permanently sck/dsabled rate: proporton of the non-student, non-retred populaton whch s permanently sck or dsabled. Source: 2001 UK Census, Sample of Anonymsed Records 29

31 Table 10 Ordered Logt Estmates of Employment n Occupatonal Class, by UK Regon North, North West and Yorkshre Mdlands East, South East, and South West London Wales Scotland Northern Ireland Female 0.202*** 0.291*** 0.360*** 0.255*** 0.178*** 0.129*** 0.071** (19.62) (23.60) (41.61) (17.96) (7.63) (7.97) (2.14) Level *** *** *** *** *** *** *** (60.02) (49.42) (69.92) (34.03) (24.51) (32.85) (16.08) Level *** *** *** *** *** *** *** (88.49) (75.62) (108.94) (54.91) (38.04) (51.14) (26.53) Level *** *** *** *** *** *** *** (88.73) (73.64) (112.00) (62.37) (37.58) (60.72) (26.21) Level *** *** *** *** *** *** *** (190.54) (157.88) (214.66) (133.29) (85.47) (122.54) (60.02) yrs yrs yrs Irsh born European born Non- European born *** *** *** *** *** *** *** (35.73) (31.76) (47.28) (25.49) (18.22) (25.43) (14.55) *** *** *** *** *** *** *** (45.69) (37.77) (58.52) (21.13) (22.99) (28.16) (18.32) *** *** *** *** *** *** *** (34.48) (28.83) (43.85) (12.62) (17.48) (20.61) (14.28) ** *** ** *** (0.98) (2.38) (0.92) (4.07) (0.74) (2.36) (2.62) * *** * (0.57) (1.73) (1.40) (10.43) (0.03) (1.76) (0.08) *** *** 0.185*** * *** (3.94) (0.05) (2.94) (8.31) (1.71) (4.77) (1.58) Mxed 0.136* *** (1.87) (1.55) (0.05) (5.27) (0.12) Asan *** 0.219*** 0.413*** (0.46) (4.65) (5.69) (15.14) (1.46) Black 0.311*** 0.205*** 0.370*** 0.723*** (4.28) (3.77) (7.08) (27.68) (0.91) Chnese *** (0.24) (0.03) (1.11) (5.47) (0.99) Catholc (0.17) Indan (1.09) Pakstan *** (4.66) Other 0.387*** (2.64) N

32 Notes to Table 10: The equaton was estmated over all those persons n employment (employees or selfemployed). Dependent varable: =1, f person s employed n a professonal, manageral, or techncal occupaton; =2, f employed n a sklled manual/non-manual occupaton; =3 f employed n an elementary occupaton. Levels of Qualfcaton are defned under Table 1: resdual s no qualfcatons. Resdual age category s years. Resdual brthplace s UK born. Resdual ethncty n England and Wales, and n Scotland s Whte. Resdual relgon (Northern Ireland only) s Protestant. Absolute value of z statstcs n parentheses; * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1% 31

33 Table 11 Margnal Probabltes of Beng Employed n Professonal, Manageral, Techncal Occupatons, by UK Regon North, North West and Yorkshre Mdlands East, South East, and South West London Wales Scotland Northern Ireland Female *** *** *** *** *** *** ** (19.6) (23.6) (41.6) (18.0) (7.6) (8.0) (2.1) Level *** 0.223*** 0.240*** 0.203*** 0.213*** 0.186*** 0.183*** (59.6) (49.3) (72.6) (37.2) (24.2) (32.4) (15.4) Level *** 0.345*** 0.363*** 0.304*** 0.318*** 0.329*** 0.317*** (92.54) (80.1) (121.7) (64.4) (39.2) (53.3) (26.4) Level *** 0.425*** 0.432*** 0.361*** 0.416*** 0.450*** 0.379*** (105.2) (88.9) (149.9) (86.0) (43.9) (74.1) (27.4) Level *** 0.689*** 0.664*** 0.661*** 0.696*** 0.692*** 0.682*** (327.1) (272.9) (362.4) (200.6) (143.6) (201.8) (93.8) yrs yrs yrs Irsh born European born Non- European born 0.108*** 0.119*** 0.131*** 0.109*** 0.127*** 0.121*** 0.127*** (31.7) (47.5) (25.8) (18.2) (25.3) (14.4) *** 0.189*** 0.112*** 0.190*** 0.161*** 0.212*** (44.7) (37.1) (59.4) (21.7) (22.6) (27.5) (17.5) 0.166*** 0.161*** 0.173*** 0.086*** 0.184*** 0.151*** 0.212** (33.5) (28.2) (44.6) (10.5) (17.0) (20.0) (13.6) ** *** ** 0.067** (0.99) (2.5) (0.92) (4.1) (0.76) (2.3) (2.5) *** (0.58) (1.8) (1.41) (10.5) (0.03) (1.7) (0.08) 0.038** ** *** *** (3.9) (0.1) (2.9) (8.3) (1.7) (4.6) (1.5) Mxed ** *** (2.0) (1.5) (0.05) (5.3) (0.12) Asan *** *** (0.46) (4.8) (5.8) (15.3) (1.4) Black ** *** *** (4.55) (3.9) (7.4) (29.0) (0.96) Chnese *** (0.24) (0.03) (1.12) (5.5) (0.96) Catholc (0.17) Indan (1.1) Pakstan 0.112*** (4.5) Other ** (2.9) N See notes to Table 10 32

34 Table 12 Logt Estmates of Unemployment Employment Equatons by UK Regon North, North West and Yorkshre Mdlands East, South East, and South West London Wales Scotland Northern Ireland Female *** *** *** *** *** *** *** (22.23) (11.87) (8.93) (12.16) (6.92) (14.06) (5.20) Level *** *** *** *** *** *** *** (26.43) (17.82) (16.37) (15.36) (7.67) (13.14) (9.79) Level *** *** *** *** *** *** *** (30.09) (21.38) (20.58) (18.64) (10.90) (19.75) (11.56) Level *** *** *** *** *** *** *** (26.63) (19.75) (17.26) (16.56) (9.29) (16.42) (11.53) Level *** *** *** *** *** *** *** (41.26) (30.60) (29.30) (36.22) (14.79) (26.56) (16.18) yrs yrs yrs Irsh born European born Non- European born *** *** *** *** *** *** *** (30.88) (22.66) (25.20) (14.97) (14.54) (18.13) (9.31) *** *** *** *** *** *** *** (35.80) (25.37) (27.73) (17.29) (16.52) (21.55) (10.26) *** *** *** *** *** *** *** (29.83) (23.00) (21.85) (15.48) (13.14) (19.10) (9.77) *** 0.434*** 0.243** (1.39) (3.75) (3.75) (2.54) (1.16) (0.64) (1.23) 0.274*** 0.348*** 0.349*** 0.500*** * (2.90) (3.04) (5.20) (8.83) (1.46) (1.77) (1.22) *** *** (0.38) (1.58) (0.59) (5.87) (0.10) (0.44) (3.25) Mxed 0.811*** 1.099*** 0.649*** 0.867*** 0.747*** (7.97) (10.70) (6.33) (11.73) (2.99) Asan 0.671*** 0.830*** 0.642*** 0.515*** (10.68) (14.08) (8.72) (11.31) (0.82) Black 1.057*** 1.276*** 1.017*** 1.157*** (10.61) (17.27) (11.41) (29.90) (1.39) Chnese 0.666*** 0.865*** 0.384*** 0.529*** (5.08) (5.40) (2.93) (6.75) (0.95) Catholc 0.647*** (10.40) Indan (0.96) Pakstan 0.555*** (3.90) Other 0.863*** (4.11) Constant *** *** *** *** *** *** *** (20.36) (26.60) (45.38) (26.63) (12.79) (13.93) (10.23) N Notes to Table 12: The equaton was estmated over all persons n the labour force.e. employed (employees or selfemployed) or unemployed Dependent varable: =1, f person s unemployed; =0, f employed. Levels of Qualfcaton are defned under Table 1: resdual s no qualfcatons. Resdual age category s years. Resdual brthplace s UK born. Resdual ethncty n England and Wales, and n Scotland s Whte. Resdual relgon (Northern Ireland only) s Protestant. Absolute value of z statstcs n parentheses; * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1% 33

35 North, North West and Yorkshre Table 13 Margnal Probabltes of Beng Unemployed, by UK Regon Mdlands East, South East, and South West London Wales Scotland Northern Ireland Female *** *** *** *** *** *** *** (22.6) (11.9) (9.0) (12.2) (7.0) (14.4) (5.2) Level *** *** *** *** *** *** *** (31.4) (20.8) (18.8) (18.7) (8.8) (14.6) (11.9) Level *** *** *** *** *** *** *** (36.5) (25.8) (24.1) (22.7) (12.8) (26.1) (15.2) Level *** *** *** *** *** *** *** (39.2) (29.1) (22.7) (21.8) (12.8) (24.3) (18.1) Level *** *** *** *** *** *** *** (55.6) (41.0) (36.2) (38.1) (18.8) (35.4) (23.3) yrs yrs yrs Irsh born European born Non- European born *** *** *** *** *** *** *** (31.8) (23.5) (26.3) (15.3) (15.2) (18.8) (9.7) *** *** *** *** *** *** *** (46.1) (31.9) (34.4) (21.3) (21.3) (26.9) (12.9) *** *** *** *** *** *** *** (45.0) (33.6) (29.6) (21.3) (19.7) (28.4) (14.5) ** 0.017** 0.014** (1.27) (3.0) (3.1) (2.3) (0.99) (0.7) (1.1) 0.014** 0.017** 0.013*** 0.032*** (2.6) (2.6) (4.5) (7.4) (1.3) (1.6) (1.0) *** ** (0.38) (1.7) (0.58) (5.6) (0.10) (0.45) (2.3) Mxed 0.054*** 0.078*** 0.027*** 0.066*** 0.050** (5.8) (7.1) (4.8) (8.6) (2.2) Asan 0.042*** 0.050*** 0.027*** 0.032*** (8.2) (10.4) (6.7) (9.6) (0.7) Black 0.079*** 0.097*** 0.051*** 0.092*** (7.2) (11.1) (7.6) (21.5) (1.1) Chnese 0.042*** 0.055** 0.014** 0.034*** (3.9) (3.9) (2.5) (5.5) (1.2) Catholc 0.033*** (10.1) Indan (0.86) Pakstan 0.034** (3.1) Other (3.0) N See notes to Table 12 34