The Effects of Education on Labor Productivity: Differences between tradable and non-tradable industries

Transcription

1 Preiminry version eptember 206 Te ffects of duction on Lbor Productivity: Differences between trdbe nd non-trdbe industries ise. Brezis* nd Gid Brnd** Abstrct Tis reserc focuses on te differences in bor productivity between trdbe nd non-trdbe industries, nd te contribution of eduction nd umn cpit to te differences between tese two sectors. e find tt wie productivity rose substntiy in te trdbe sector, in te non-trdbe sector, output per worker s remined te sme; toug, te rise in umn cpit in bot sectors does not differ significnty. Tis pper empsizes tt eterogeneous biity of individus s we s doube duity bot in te bor mrket nd te iger-eduction migt expin tese fcts. Te doube duity in te economy permits seprtion of individus by teir biity, wic cn expin te differences in bor productivity between sectors. Keywords: productivity, trded goods, services, duity, iger eduction, umn cpit, wge premium. JL cssifiction: F2, F6, I26, J24, O4. * ed, Aron Meir Center (AMCB), Br-In University eise.brezis@biu.c.i ** P.D student, Br-In University, nd Tub Center gidmbrnd@gmi.com

2 2 I. Introduction Tis reserc focuses on te contribution of eduction nd umn cpit to te differences between te trdbe nd non-trdbe sectors. A priori, we coud ssume tt te non-trdbe sector, wic mosty incudes services, woud ve ow content of umn cpit, nd tt tis fct woud be te min difference between tese two sectors, i.e., te trdbe sector s ig content of umn cpit, nd te non-trdbe s ow content of umn cpit. Indeed, Figure sows te retive content of trdbe vs. nontrdbe, nd tere seems to be corretion between te trdbiity of n industry nd te verge yers of scooing of its workers. o tis difference in umn cpit coud expin te gp in bor productivity. owever, over te st two decdes, tere ws so n increse in te content of umn cpit in te non-trdbe sector. In Figure 2, te rise in umn cpit in te non-trdbe sector does not differ significnty from te rise in umn cpit in te trdbe sector. Despite tis increse in umn cpit in te non-trdbe sector, we find tt productivity rose substntiy on te trdbe side, but on te non-trdbe side, output per worker s remined te sme. Tese fcts cn indicte tt eduction migt be ssocited wit rise in productivity ony in te non-trdbe sector. Te purpose of tis pper is to deveop frmework wic cn ssembe te vrious pieces of te puzze togeter. Tis pper empsizes tt doube duity in te economy migt expin te fcts presented bove. Indeed te economy dispys duity in bot te output sector nd te eduction sector: Te bor mrket is divided into two very different economies: Te trdbe sector wic comprises miny ig tec but so trdition mnufcturing industries, nd te non-trdbe sector wic comprises miny trde nd service industries. Te bor productivity in tese two ggregted sectors is very different (Tbe ). But tis is not te ony duity in te economy: Te economy dispys duity in iger eduction. Tere is not one cnne of iger eduction, but two: grduting from prestigious university, or grduting from oc coege. 2 Te iger eduction sector in mny OCD countries dispys two seprte cnnes of eduction, due to seprte entry exms in te vrious types of universities nd fcuties. e present dt for te Isrei economy, but we find simir empiric regurities for oter OCD countries. Moreover, te mode is gener. 2 Te st decde, tere ws uge increse in te intke of tertiry eduction, but tis ws miny due to rise in enroment rtes in oc coeges (see Figure 3).

3 3 ie one needs score in te top 5 t percentie on cognitive test 3 (equivent to n AT exm) to enter computer science nd score in te top 35 t percentie in conomics in eite universities, one needs ony to pss te bottom 30 t percentie in te fcuty of umnities, nd no exm t to enter coeges (wic usuy do not offer science courses or computer science). Tis pper wi sow tt te duity in te sector of iger eduction permits to differentite individus wit ig nd ow biity, nd tis is prt of te expntion of te strong division between te trdbe nd non-trdbe sectors. Te reted expntion is te compementrity between cpit nd umn cpit in te trdbe sector, weres te non-trdbe sector is function ony of bor nd umn cpit. A tese eements togeter permit us to ssembe te vrious pieces of te puzze dispyed by some OCD economies. Indeed our min proposition stresses tt te doube duity in te economy permits seprtion of individus by teir biity. In oter words, individus wit ig biity wi work in te trdbe sector, wie individus wit ow biity wi work in te non-trdbe sector. Tis seprtion of biities coud expin te difference in bor productivity between sectors. 4 Te pper is divided into 6 sections. In te next section, we present te iterture. In section III, te empiric regurities re put in evidence. Te mode is presented in section IV, nd in section V te empiric nysis is discussed. ection VI is te concusion. II. Reted Literture 2. eterogeneity of workers Bss (964) nd mueson (964) ssumed tt rise in productivity on te trdbe side woud resut in trnsfer of workers from te non-trdbe to te trdbe side of te economy. Terefter te non-trdbe firms woud be subjected to wge pressure resuting in n increse in te over wge eve nd rise in te price of non-trdbe commodities. Yet Lvy nd Friedmn (2007) found te Bss nd mueson effect to be bsent in te Isrei economy nd ttribute tis to segmented bor mrket. Tey find tt workers crcteristics in te trdbe industries differ grety from tose on te non-trdbe side resuting in ow mobiity between te sectors. Brnd nd Regev (205) find tt te trde iberiztion reform resuted in quick nd substnti movement of ow skied workers 3 Acceptnce typicy is bsed on combintion of mtricution grdes nd cognitive tests. 4 Tis version of te pper is very preiminry, since tere re no dynmics, wic wi be introduced in te next version.

4 4 from trdition mnufcturing industries to te ower wge tier of te non-trdbe industries. Among te ig skied, movement in te opposite direction ws observed. 2.2 Duity in iger eduction nd te effects of eduction on productivity Te duity in iger eduction s been miny empsized in retion to soci mobiity nd inequity, nd not so muc in retion to productivity. Indeed, Brezis nd eier (206) sow tt du iger-eduction system crcterised by te concomitnce of bot stndrd nd eite universities genertes permnent soci strtifiction, ig soci immobiity nd sef-reproduction of te eite. Moreover, Kerckoff (995) suggests tt te effect of fmiy bckgrounds coud be mgnified wen te eduction system is igy strtified nd seective. Tis rgument s been confirmed by sever empiric works (nusek nd oessmnn, 2006; Pfeffer, 2008; Dronkers et., 20). Indeed, in te recent economic iterture on eduction nd umn cpit, tere s been growing interest into te nysis of strtifiction of eduction systems (see Brezis nd Temin, 999, 2008). Te impct of eduction on economic growt s been mjor topic in economic reserc. From te ery studies by Becker (962), Mincer (974) nd cutz (97) to te iterture wic foowed Romer s (990) endogenous growt teory, umn cpit ccumution s been regrded s key determinnt of ong-term economic growt. Yet tere is n emergence of n empiric iterture wic cst doubt on te positive effect of n increse in umn cpit on economic growt. Te resuts of Pritcett (200), Krueger nd Lind (2002) nd Benbib nd piege (994) sow tt wen umn cpit is one omogenous vribe, usuy depicted by te verge yers of eduction, ten umn cpit does not ffect economic growt. o, umn cpit in te form of omogenous eduction does not seem to ffect economic growt in te dt. it te use of scores on interntiony comprbe exmintions, nusek nd oessmnn (2008, 202) nd Brro (203) ve stressed te importnce of scoo quity nd cognitive skis rter tn scoo quntity. imiry, Atinok nd Aydemir (206) sow tt te effect of scoo quity on growt differs cross regions nd by te economic eve of countries. Brezis nd Crouzet (2006) sow tt differences of quity nd recruitment mong universities ed to dopt different types of new tecnoogies, wic ffect te eve of economic growt. 2.3 Productivity eres ggregte productivity is expected to converge bsed on country s unique crcteristics, Rodrick (20) finds tt GDP per worker in te mnufcturing industries

5 5 tends to converge unconditiony. Tis coud expin te different trends of productivity growt in te trdbe nd non-trdbe industries. pence nd tswy (202) find tt te increment productivity growt in te U in te period stems from te trdbe sector, wie virtuy empoyment is generted in te non-trdbe industries. De Miceis et. (203) nd Junnkr (203) find negtive retionsip between bor force growt nd productivity growt. Terefore te different trends in productivity growt my be tied to teir opposite empoyment trends. Trdbe industries re more prone to knowedge diffusion vi gob connections, FDI nd mobiity of skied bor. Te nture of te ctivity of tese industries is more tecnoogicy dynmic nd more ikey to be revoutionized by innovtion. As consequence, productivity tends to rise fster in te trdbe industries, penomenon weknown for mny yers. Furtermore, since iger productivity resuts in fow of skied bor nd cpit investments, te productivity of te trdbe retive to te non-trdbes tends to widen in ccordnce wit ggregte economic growt. Rodrick (206) finds tt rpidy growing non-trdbe sector, nd in prticur swift decine in empoyment in mnufcturing industries, cn inder te productivity convergence in midde income economies. Micro dt on firm eve furter strengtens tis ssertion, nd evidence sowing tt exporting firms tend to outperform te non-exporters s been we documented in te iterture. Greenwy nd Kneer, (2008) describe te empiric findings of erning by exporting effect, empsizing tt exposure to foreign competitors, new cients nd fierce competition in interntion mrkets wi resut in incresing efficiency nd greter investments in new tecnoogies. III. mpiric Regurities Tis prt presents dt specific to te Isrei economy, but most of te fcts re simir to te regurities in some oter OCD countries. e strt by focusing on eterogeneity of workers, nd differences in productivity, nd ten differences in returns of fctors of production. Te division between te trdbe nd non-trdbe industries nd te source of te dt re expined in section eterogeneity of workers between sectors Te distribution of te workers cognitive scores re dispyed in Figure 5 nd summry sttistics re sown in Tbe 2. As is sown, wie workers biity in te non-trdbe sector is normy distributed, biity mong workers in te trdbe sector is strongy

6 6 skewed to te rigt, suggesting tt substnti differences exist in workers biity in ec industry group. 5 Accordingy, we so find significnt differences in te quity of obtined eduction. eres 68% of workers in te trdbe industries (in te ge group 27-32) ve obtined teir eduction from university, tis number drops to ony 5% mong te nontrdbe industries. Moreover, BOI (206) finds simir resuts. Combining te resuts of te recenty pubised OCD urvey of Adut kis (PIAAC), te BOI (206) finds, on n industry eve nysis, tt weres te sre of workers wit iger eduction retive to te OCD is not positivey correted wit te productivity gps; PIAAC test resuts re positivey correted. Tis finding suggests tt te quity of eduction my vry significnty cross industries. 3.2 Different Productivity in te trdbe nd non-trdbe industries TFP growt rtes in te trde nd services vs. mnufcturing industries re different (Figure 6), representing te trdbe nd non-trdbe side of te economy, ccordingy. Tis comprison, wic spns from 975 up unti 205, sows tt unti te ery 990s, tere ws n most perfect corretion between productivity growt in te two industry groups, weres fter tt period, growt in mnufcturing industries begn to cceerte, ong wit minor decrese in productivity in te non-trdbe industries. A more precise division between te two industry groups reves simir resuts (Figure 7). Ony minor differences in productivity eves in te two industry groups were found in 995, yet te rpid growt on te trdbe side nd ck of growt in te non-trdbe industries s resuted in widening of te gps. Tis cn so be sown by ccuting te cumutive distribution of productivity in 995 nd 200 (Figure 8). eres on te trdbe side te tter s first order stocstic dominnce over te former, tis is cery not te cse in te non-trdbe industries. 3.3 ges of skied workers in trdbe nd non-trdbe industries. Te trdbe industries tend to py iger wges tn non-trdbe industries. Figure 9 sows strong corretion between te extent in wic n industry is open to interntion trde nd te nnu verge wge eve. iger wges re found so wen controing for workers' observbe crcteristics. As sown in Figure 0 wge gp between te two sectors exists nd te gp s furter incresed over te exmined time period. Accordingy, return on eduction ws found to be iger in te trdbe industries. Te resuts iustrted in Figure sow tt te coege wge premium s remined retivey 5 Tis concusion is supported by simpe Komogorov mirnov test (p<0.0).

7 7 stbe in te non-trdbe industries wie stedy increse ws found in te premium on te trdbe side. Moreover, te premium ws found to be on te sme eve in bot industry groups in 995, yet by 200 te premium in te trdbe side ws notby iger. To concude, wie eduction rtes ve incresed cross segments of te bor mrket, productivity s tken muc ess uniform course. On te trdbe side, productivity s risen srpy, but on te non-trdbe side productivity s reced ner stndsti. Te mode presented in te next section wi try to give n expntion to tese fcts. IV. Te mode 4.. Introduction Our frmework ssumes tt tere is eterogeneity (i) mong sectors, (ii) in te iger eduction mrket (iii) mong individus. In oter words, not ony individus differ in teir biities; but sectors ve different production functions nd terefore different needs for umn cpit, nd so te iger eduction system is not omogeneous. In tis mode, some of te workers cquire iger eduction. o te workers re eiter workers witout skis, L or workers wit skis due to te fct tt tey ve invested in iger eduction,. owever eduction is not omogenous. Individus cn eiter ern in top university, or equiventy in sciences ( for exct science, or eite universities) or ern in stndrd university (or in fcuty of umnities ( N for non eite, or umnities). Indeed, te duity in iger eduction cn tke two quite equivent forms. iter one wnts to focus on te duity in te type of fcuties (science versus umnities), or one prefers to focus on te duity between eite universities vs. stndrd universities. For bot duities, te resoning is te sme. 6 Te type of eduction te individu cquires is common knowedge, since it is cknowedged on is dipom Abiity e ssume tt individus re born wit different biities, eiter ig denoted, or ow denoted. e ssume proportion gmm of individus wit ig biity. Moreover, for ske of simpicity we ssume tt were. Tis difference in biity ffects te economy troug two cnnes. First, smrter peope ern more rpidy, nd terefore for getting te sme grde, tey ve to invest 6 In tis version of te pper, since we ve gtered dt on te duity between eite nd stndrd universities, wic is ced coeges in Isre, we wi focus on te second type of duity, i.e., eite vs stndrd universities.

8 8 muc ess effort tn te person wit ow biity. e cn cite exmpes tt in te sme css, some students cn just, for once, ook t te mt book, nd wi sove most exercises. ome oters wi ve to redo 0 times te sme exercise to know to sove it. Te second cnne is troug te mrket bor. Abiity ffects te productivity of individus, nd individus wit ig biity wi ve iger productivity for doing te sme job. e strt te presenttion of te mode wit te utiity function, ten we move to te production section, nd ten to te eduction mrket Te utiity Tere re two sorts of goods in te economy, trded, T nd non-trded, NT. Consumers wnt tem bot, nd we ssume n esticity of substitution of between tese goods, so te utiity function wi tke Cobb-Dougs form suc s: U ( T, NT ) T NT. were 0 () In te next section, we present te production functions. e strt wit te trdbe sector. 4.4 Te trdbe sector. Te trdbe sector uses 3 fctors of production, L, nd K. e ssume C function between nd L, tt is umn cpit nd bor re substitute fctors of production, nd we ssume tt bor, L nd cpit K ve constnt rte of substitution of. is not omogenous, we ve in fct two different types of umn cpit, nd N. Te two types of umn cpit re perfect substitute. Te productivity of ec is function of te verge biity of te skied workers ving cquired tis type of eduction (, 2 ). Moreover, umn cpit cquired troug eduction in eite university/science dds productivity of. o te trdbe sector s te foowing production function Y T K [( N 2 ) ( u L) ]. (2) ere, re bot between 0 nd. Te respective costs of te fctor of productions of L, N, nd K re: u,,,nd r.

9 9 4.5 Te non-trdbe sector. Te non-trdbe sector uses 2 fctors of production, nd L. Te two types of re perfect substitute. o te NT sector s te foowing production: Y NT N 2 ) ( u ) ] [( L. (3) 4.6. Te iger duction sector. Tere re exms for entry to te different fcuties/ universities. Tere re fcuty of sciences / eite universities, in wic wen grduting, te student cquire umn cpit of type ; nd tere re fcuties of umnities/coeges, in wic te student cquires umn cpit of type N. Te grdes on te entry exm (AT) to gin ccess to te fcuty of sciences/eite universities, re muc iger tn te grdes to enter umnities/coeges. 7 In consequence, we get te foowing prtition. tudents wit ig grdes on is AT wi get ccess to science/eite universities nd cquire umn cpit of type. tudents wit ower grdes (but wit bccuret) wi register to umnities/coeges, nd wi cquire umn cpit of type N. Finy, individus witout bccuret wi sty unskied, nd dispy fctor of production, L. Te individus wit bccuret cn register to csses wic re eping tem to tese AT exms. Te costs for tking tese exms is te price per our of tese csses, P, mutipied by te number of ours necessry for prepring for tese exms. Individus wose biity is ow need penty of time for te cquisition of te knowedge (i.e., e needs to invest ig effort, e ), weres individus wose biity is ig need ow investment ( e ). For mtters of simpicity, we ssume tt efforts re inverse to te biity eve, so tt e / nd e /. o te costs for ec individu for entering sciences/eite universities re: C P P. e for individus wit ig biity (4) 7 In different countries, it s different nme. In te U, it is AT, in Isre, te psycometric exms, in Frnce te prep exms.

10 0 C P P. e for individus wit ig biity (5) And we get tt C C e ssume tt te costs for entering umnities/coeges re 0 for ig-biity individus wie te costs for ow-biity is ow but not zero, nd we ssume it is: c P / wit 2 nd /( ) Te quiibrium. Let us find out, weter tere is seprtion between types of biity. In order to do so, we wi present tree emms. Let us first define condition I on wges, nd ten present emm. Condition I: P P ) ( Condition Ib: c u Lemm. Under condition I, individus wit ow biity wi cquire iger eduction of type N, wie individus wit ig biity, wi get ccess to eite universities nd cquire umn cpit of type. Individus wo do not ve bccuret, nd did not finis ig scoo, wi not ve ccess to university or coege. Proof Let us ssume tt indeed individus of ig biity cquire, nd individus wit ow biity go to ern in umnities/coeges. e sow tt tis is n equiibrium, i.e., no individu wnts to diverge from tis equiibrium. ). For ig biity person, from te rigt nd side of Condition I, it is esy to sow tt we get te foowing inequity : C Te inequity mens tt ig biity individus get iger income from investing in eduction in science/eite university tn from getting degree in umnities/coege (remember tt costs for ig biity to ern in umnities is 0). In consequence we ve sown tt indeed ig biity individus prefer to ern t eite universities/ science.

11 b). For ow biity person, under condition Ib nd te rigt nd side of condition I, we get te foowing inequities (remember tt for ig biity, cost of erning in coege/umnities is c). o: c C u In consequence ow biity individus enter coege/ umnities. o we get tt tere is duity in iger eduction: individus wit ig biity cquire, nd individus wit ow quity cquire N. e now ceck weter tere is so duity in te bor mrket. Let us define Condition II: Condition II: 2 e ten get te foowing Lemm. Lemm 2 Individus wit umn cpit of type individus wit umn cpit of type N wi work in te trdbe sector, nd te wi work in te non-trdbe sector. Proof (i) Let us nyze te trdbe sector. From te production function dispyed in eqution (2), umn cpit of types nd N re perfect substitute. In consequence te producer wi empoy te type wic is te cepest for im for producing te sme mount of output. One worker of type is producing 2 t costs, wie te worker of type N is producing t cost. It is ess expensive to ire workers ving grduted from science/eite universities if: wic is equivent to te eft nd side of condition II. 2 (ii) About te non-trdbe sector, from eqution (3), one worker of type 2 t costs, wie te worker of type N is producing t cost. It is ess expensive to ire workers ving grduted from coeges/umnities if: is producing

12 2 wic is equivent to te rigt nd side of condition II. 2 e now turn to te Proposition. QD. Proposition Under Conditions I nd II, individus wit ig biity wi work in te trdbe sector, nd individus wit ow biity wi work in te non-trdbe sector. Proof From Lemm 2, workers in te trdbe sectors re wit eduction of type. From Lemm, tose wit eduction type re of ig biity. In consequence, individus wit ig biity work in te trdbe sector. Foowing te sme resoning, individus wit ow biity wi work in te non-trdbe sector. ince te ony skied workers in te trdbe sector re of ig biity, we ten get tt 2, nd te production function tkes te foowing form: Y T K [( ) ( u L) ]. (6) Foowing te sme resoning, te eqution of te non-trdbe sector is: Y NT N ) ( u ) ] [( L. (7) 4.8. ges nd ge premium. Let us now find out te wges, Te mrgin products of u,,. nd L re equ to teir wges, so: nd: YT u K u [( ) ( u ) ] L L L. (8)

13 3 u T L K Y ] ) ( ) [(. (9) o tt te wge premium of eduction of type is: ) ( ) ( ) ( L u u. (0) From te non-trdbe function of production, te mrgin products of N nd L re equ to teir wges, so: ] ) ( ) [( u N u NT u L L Y. () ] ) ( ) [( N u N NT L Y. (2) And te wge premium of eduction of type N is: 2 ) ( ) ( N u u L (3) From (0) nd (3), we get tt te wge premium of eduction of type vs. type N is: ) ( 3 ) ( ) ( ) ( ) ( L L u N u (4) Lemm 3 it production functions s presented in equtions (6) nd (7), Conditions I nd II ods. Proof (in te next version) Lemm 3 permits us to simpify te equtions presented in gener mnner (equtions 2 nd 3) to equtions (6) nd (7). ince conditions I nd II ods, tis mens tt te mode is

14 4 stbe nd tt indeed te seprtion equiibrium is stbe nd no individu s reson to devite from is position. Terefore, ow biity workers indeed grdute from coeges nd wi go to work in te non-trdbe sector. For ig biity workers, tey wi grdute from eite universities, nd work in te trdbe sector. Tis seprtion permits us to ccute te wge premium, wen workers wit different biities work in different tsks. ince teir biity is different, nd te eite universities offer better eduction, te ig biity workers ve doube reson for iger productivity ( nd ). Tis expins te gp in productivity Concusion of te mode Te mode s permitted to expin te foowing fcts. (i) From Proposition we get tt peope wit ig biity (i.e., ig grde on AT) re te ones working in te trdbe sector nd te opposite. Tis is wt s been empsized in section 3.. (ii) Productivity of workers in te trdbe sector is iger tn in te NT sector (since one is function of nd te oter one.); Tis is wt s been empsized section 3.2. (iii), from eqution 4 we see tt te retive wge is ower tn, so wges of skied workers in te trdbe sector is iger tn in te non-trdbe sector. Tis is wt s been empsized in section 3.3. In concusion, te doube duity in te bor nd iger eduction sectors is essenti for expining te empiric regurities presented in section III. V. mpiric Anysis In tis prt of te pper, we nyse te retionsip between output productivity, openness nd eduction. e do so by running simpe regression in wic te nnu verge growt rte of productivity determined by te initi og eve of productivity nd te initi og scoo yers. e so contro for te degree in wic te industry is exposed to interntion trde nd n interction term between te initi og eve of eduction nd te openness rtio. More specificy te regressions we wi perform is: Δgdp per our α α 6 openness rtio i i α 2 n(gdp per our ) α n(scoo ) ε i i i 3 n(gdp per our ) i 2 α 4 openness rtio i α 5 n(scoo i ) ere i is n industry index; Δgdp per our is te nnu verge growt rte of productivity; gdp per our is te eve of productivity in 995; scoo is te verge yers

15 5 of scooing in 995; from 0-. openness rtio is mesure of te openness of ec industry rnging 5. Dt 5.. Te smpe Using istoric time series dt nd bor surveys conducted by te Centr Bureu of ttistics of Isre (CB), te study buids upon dt set rnging from 995 up to 200, owing for empoyment, output nd eduction trends to be exmined. e focus ony on te non-mining/non-re-estte business sector nd furter excude te gricutur nd construction industries wic rey eviy on foreign bor (i.e non-isrei workers from brod empoyed on somewt temporry bsis). In consequence, we ve 42 sectors wic ccount for bout two-tirds of te bor mrket. 8 Te dt is vibe up unti Te trde openness index In order to mesure te openness of ec industry we empoy te Input-Output tbes, vibe for te yer 2006, nd devise trde openness index. Te index mesures te sre of exports nd competitive imports witin te tot suppy of te industry. Tis mesurement yieds n index rnging from 0-, indicting te extent in wic n industry is exposed to foreign competition. ubsequenty we divide te industries into two distinct trdbiity groups: trdbe nd non-trdbe sectors. Using te trde openness index we set te cutoff point t 0.4, mening industries wit iger openness rtio re incuded in te trdbe sector nd te remining in te non-trdbe industries. Using tis division eves us wit 9 trdbe industries wic consist mosty of ig-tec nd te buk of te mnufcturing industries. Te 23 remining non-trdbe industries consist of most of te trde nd services industries. Finnces nd bnking ctivities were cssified s non-trdbe since tese industries in Isre re engged mosty in te oc mrket (see Tbe ). e find tt tis division represents te trdbe nd non-trdbe side of te economy we nd te resuts re quite robust nd not sensitive to te cutoff point. In ddition, we ve gtered dt on te workers in te trdbe nd non-sectors by te type of university tey ttended, teir cognitive exm score, nd industry in wic tey re empoyed. Te dt is vibe to us for te yer 20 for workers in te ge group (n=68,664). Te dt ws restricted to 8 Tis dtbse consists of surveys of Trde, ervices, Trnsport, Communictions nd mnufcturing conducted by te Isrei Centr Bureu of ttistics. Micro dt on wges nd empoyment crcteristics were coected using income nd bor surveys.

16 6 workers wo competed t est n undergrdute degree nd were empoyed t te time of te survey on fu time bsis Te Resuts. Resuts re presented in Tbe 3. Our min interest is te interction term found to be positive nd significnt (p<%) indicting tt scoo x openness rtio is infuenti for productivity growt. Omitting te interction term from te regression reduces te expntory power of te regression, nd te coefficient of n(scoo ) remins insignificnt. Tese resuts indicte tt trde is key determinnt of te ink between eduction nd interntion growt. Putting our resuts togeter, we cn concude tt s n economy becomes more open to trde, eduction ccumution cn gin stronger effect on productivity growt in te trdbe industries. 5.3 Furter nysis e perform some more exercises wic re more in pse wit te stndrd modes in wic it is ssumed tt individus re omogeneous in biity. Indeed, if individus were ssumed omogeneous, ten we woud find difference in te mtcing between occuption nd eduction. Indeed, wen we compre te growt rte of workers wit iger eduction to te rise in occuptions wic require suc eduction, we find tt on te trdbe side tese two vribes tend to move togeter, wie in non-trdbe industries, te tter is growing t pce tree times fster tn te former (Figure 2). As consequence, wen we ssume omogeneity of workers, tis is strong indiction tt suppy of educted workers s risen fster tn demnd. owever, ccording to our mode, wic ssumes eterogeneity of workers, tis exercise cn be interpreted s if te type of occuption woud differentite between skied wit ig biity, nd skied wit ow biity. Te sme dicotomy of interprettion exists on te wge premium. Our pper sows tt indeed wge premium s incresed in different mnner in bot sectors, due to te fct tt individus ve different biities (Figure ). If we woud ppy te Ktz nd Murpy (992) frmework, wic ssumes omogeneity of workers, wen suppy grows fster tn te demnd for skis, ten te coege wge premium wi f. en suppy growt fs sort of tis rte, te ski premium wi increse. en exmining te different segments of te bor mrket, we find tt te wge premium s incresed in te trdbe sector nd reced ner stndsti on te non-trdbe one. Tis finding refutes te ssumption tt suppy of ski s exceeded te demnd.

17 7 Our interprettion is different. ince te type of iger eduction cn sign te biity of te worker, te workers in ig-tec in prticur nd te trdbe sector in gener ve doube vue-dded retive to te workers in te non-trdbe sector. On te one nd, tey possess better biity ( vs. ) nd on te oter nd, tey obtined n eduction wic gives tem ed in productivity ( ). Terefore, te doube duity, nd te fct tt indeed we ve found tt workers re of different biities in te two sectors, expin te fcts reted to te wge premium, witout entering te debte reted to suppy nd demnd. VI. Concusion Tis pper nyzes te difference in te type of eduction nd umn cpit between te trdbe nd non-trdbe sectors. Indeed, te dt sow tt umn cpit vries between te trdbe nd non-trdbe industries, moreover, te verge biity is iger in te trdbe sector. Terefore iger-eduction nd umn cpit nd its cus effect on productivity woud be better nysed s eterogeneous fctor of production, s done in our mode. It is interesting to note tt te roe of iger eduction nd umn cpit s evoved over te decdes. In te pst, umn cpit ws interpreted s te fctor of production wic represented te skis of workers. It ws omogenous. Tody, umn cpit tkes different forms nd umn cpit is not nymore omogenous fctor of production. In consequence, in tis pper, we differentite between types of eduction nd biities of individus. On te one nd, individus cn be wit ow or ig biity. On te oter nd, individus cn grdute eiter from prestigious universities or from stndrd coeges. Our mode s sown tt tere is retionsip between biities nd types of eduction: Individus wit ow biity re more prone to grdute from stndrd universities, wie ig biity students grdute from eite scoos nd from prestigious fcuties. It is tis disprity wic expins te vrious differences between te service sector nd te mnufcturing ones. On te empiric side, we ve nyzed te Isrei economy from 995 to 20. More specificy, we buid n openness rtio index in order to indicte te sensitivity of te economic sectors to foreign competition. ubsequenty, we divided te bor mrket into trdbe nd non-trdbe components enbing bot sectors to be exmined. e ve found tt productivity is different between bot sectors, tt te wge premium is iger in te trdbe sector, nd tt te biity of workers is ower in te non-trdbe sector.

18 8 References Atinok, N. nd A. Aydemir "Te impct of cognitive skis on economic growt." orking Pper Bnk of Isre Bnk of Isre Annu Report 204. Bnk of Isre "Bsic skis of workers in Isre nd industry productivity." urvey nd coection of reserc issues. Bss, B., 964. "Te purcsing-power prity doctrine: A reppris." Te Journ of Poitic conomy 72, Brro, R. J duction nd economic growt. Anns of conomics nd Finnce, 4(2), Becker, G.., 962. "Investment in umn cpit: A teoretic nysis." Te Journ of Poitic conomy 70(5), Benbib, J. nd M. piege "Te Roe of umn Cpit in conomic Deveopment. vidence from ggregte cross-country dt". Journ of Monetry conomics. 34: Bouo,., de erres, A., Monr, M., Te contribution of economic geogrpy to GDP per cpit. OCD Journ: conomic tudies 2008, -37. Brnd, G. nd. Regev Te Du Lbor Mrket: Trends in Productivity, ges nd umn Cpit in te conomy, in tte of te Ntion Report: ociety, conomy nd Poicy in Isre 205, Tub Center for oci Poicy tudies in Isre pp Brezis,.. nd J. eier oci Mobiity t te top nd Duity of te iger- duction ystem" mimeo Brezis,.. nd P. Krugmn Immigrtion, Investments nd Re ges, Journ of Popution conomics, 9: Brezis,.. nd Crouzet, F Te roe of iger eduction institutions: recruitment of eites nd economic growt. In Institutions nd conomic Growt, ed. T. icer nd C. Grci-Penos. Cmbridge, MA: MIT Press. Brezis,.. nd P. Temin ites nd conomic Outcomes, in: Duruf nd Bum ds., te New Pgrve ncycopedi. De Miceis, M. A., stevo M. M. M., ison, M. B. A., 203. "Productivity or empoyment: Is it coice?." IMF orking Pper P/3/97. Dronkers J., vn der Veden R. nd Dunne A. 20. Te effects of eduction systems, scoo-composition, trck-eve, prent bckground nd immigrnts origins on te cievement of 5-yers od ntive nd immigrnt students. A re-nysis of PIA 2006 Gev, A "y Is Isre s Productivity Lower Retive to Oter Deveoped Countries?" conomics nd Reserc Deprtment, Ministry of Finnce (ebrew).

19 9 Greenwy, D., Kneer R., "Firm eterogeneity, exporting nd foreign direct investment." Te conomic Journ 7(57), F34-F6 Junnkr, P. N., 203. "Is tere trde-off between empoyment nd productivity?" IZA Discussion Pper, no Friedmn, A. nd Y. Lvi "Isre's Re xcnge Rte nd Foreign Trde." Bnk of Isre Review 79 (ebrew), nusek,.a. nd oessmnn, L Does duction Trcking Affect Performnce nd Inequity? Differences-in-Differences vidence cross Countries, conomic Journ, 6, C63. nusek,. A., oessmnn L., Te roe of cognitive skis in economic deveopment. Journ of economic iterture, 46(3), nusek,. A., oessmnn L., 202. "Do better scoos ed to more growt?" Cognitive skis, economic outcomes, nd custion. Journ of conomic Growt 7(4), Ktz, L. F., Murpy, K. M., 992. Cnges in retive wges : uppy nd demnd fctors. Ntion Bureu of conomic Reserc, no. w3927. Kerckoff, A.C Institution Arrngements nd trtifiction Processes in Industri ocieties. Annu Review of ocioogy, 5, Krueger AB nd Lind "duction for Growt: y nd for wom?". Journ of conomic Literture 39:0-36. Mincer, J. A., 974. "cooing nd ernings, in cooing, xperience nd rnings." NBR. Pfeffer, P.T Persistent Inequity in duction Attinment nd its Institution Context, uropen ocioogic Review, 25(5), Pritcett, L "ere s te duction Gone?" ord Bnk conomic Review 5: Rodrik, D., 20. "Te future of economic convergence." Ntion Bureu of conomic Reserc, no. w7400. Rodrik, D Premture Deindustriiztion, Journ of conomic Growt, 2 (), 33. Romer, Pu M., 990. ndogenous Tecnoogic Cnge. Journ of Poitic conomy 98(5) prt II, Regev,. nd G. Brnd, 205. Cuses of te idening Productivity Gps Between Isre nd te OCD: A Mutiyer Industry-Leve Comprison in tte of te Ntion Report: ociety, conomy nd Poicy in Isre 205, Tub Center for oci Poicy tudies in Isre pp mueson, P. A., 964. "Teoretic notes on trde probems." Te Review of conomics nd ttistics 46(2),

20 20 cutz, T.., 97. "Investment in umn Cpit. Te Roe of duction nd of Reserc." New York: Te Free Press. pence, M., nd. tswyo, 202. "Te evoving structure of te Americn economy nd te empoyment cenge." Comprtive conomic tudies 54(4),

21 Figures nd Tbes Figure : re of duts (25-64 yer-ods) wit iger eduction (on te eft) nd verge yers of scooing (on te rigt) in te trdbe nd non-trdbe industries, 20. ource: Te Isrei Centr Bureu of ttistics Figure 2: re of duts (25-64 yer-ods) wit iger eduction (on te eft) nd verge yers of scooing (on te rigt) in te trdbe nd non-trdbe industries, ource: Te Isrei Centr Bureu of ttistics.

22 22 Figure 3: Number of students enroed in iger eduction for bceor s degree by type of cdemic institution * Not incuding students t te Open University. For consistency in comprison, Arie University (formery Arie Coege) is considered n cdemic coege for yers. ource: Te Isrei Centr Bureu of ttistics nd te Tub Center. Figure 4: GDP per our worked in Isre nd G7 countries. UD, constnt prices, 200 PPPs,

23 23 ource: OCD stt nd te bnk of Isre. Figure 5: Te distribution of te workers cognitive scores (equivent to n AT exm) in te trdbe (on te eft) nd non-trdbe industries (on te rigt) Te dt represents te yer 20 for workers in te ge group (n=68,664). Te dt ws restricted to workers wo competed t est n undergrdute degree nd were empoyed t te time of te survey on fu time bsis. ource: Te Isrei Centr Bureu of ttistics. Figure 6: Tot Fctor Productivity in te mnufcturing industries nd te trde nd services industries, Index: 975=00, ource: Te bnk of Isre nd te Tub center for soci poicy.

24 24 Figure 7: GDP per our worked re growt in te trdbe nd non-trdbe industries , NI, constnt prices, 200. *Trdbe industries were defined s industries in wic teir openness rtio exceeds>40% ** Te index mesures te sre of exports nd competitive imports witin te tot suppy of te industry t te yer 2006 ource: Te Isrei Centr Bureu of ttistics. Figure 8: Cumutive distribution of productivity in 995 nd 200 trdbe (on te eft) nd non-trdbe (on te rigt) industries..8.8 Cumutive Probbiity.6.4 Cumutive Probbiity gdp_per_our_worked gdp_per_our_worked ource: Te Isrei Centr Bureu of ttistics.

25 25 Figure 9: Annu verge wge eve in 200 nd openness to interntion trde. ource: Te Isrei Centr Bureu of ttistics. Figure 0: Te wge premium for empoyment in te trdbe industries* * 'Te crt exibits te wge differenti between te trdbe nd non-trdbe industries controing for observbe crcteristics. Te premium for empoyment in te trdbe industries is estimted wit te og oury wge for fu time workers (25-64 yer-ods) regressed in ec yer on te foowing expntory vribes: potenti experience nd potenti experience squred, yers of scooing nd yers of scooing sqered, dummies for sex, immigrnts nd non-jewis workers. A dummy for te trdbe industries is dded nd two yer moving verge of its coefficient sown in te figure. ource: Te Isrei Centr Bureu of ttistics.

26 26 Figure : Te coege wge premium in te trdbe nd non-trdbe industries * * 'Te coege wge premium is estimted wit te og oury wge for fu time workers (25-64 yer-ods) regressed in ec yer on te foowing expntory vribes: potenti experience nd potenti experience squred, yers of scooing, dummies for sex, immigrnts nd non-jewis workers. A dummy for te trdbe industries is dded, n interction between te dummy nd scoo yers reves te coege wge premium in ec industry group. ource: Te Isrei Centr Bureu of ttistics.

27 27 Figure 2: re of workers (25-64 yer-ods) wit iger eduction nd te sre of workers in occuptions wic require iger eduction. Trdbe (on te eft) nd non-trdbe (on te rigt) industries * * According to te Centr Bureu of ttistics cssifiction (CB 994) of occuptions nd te pre interntion cssifiction, occuptions wic require iger eduction re cdemic professions wic incude for exmpe engineers, scientists, economists, nd ccountnts (mjor group 0). Aso incuded re ssocite professions nd tecnicins (mjor group ) wo incude occuptions tt require tertiry eduction nd not necessriy cdemic eduction. ource: Te Isrei Centr Bureu of ttistics.

28 28 Tbe : Industry crcteristics in 200 Brnc GDP per our worked Averge yers of scoo mpoyees (In tousnds) Openness rtio Trdbiity torge, prking ots nd freigt termins % Non-Trdbe irdressing nd beuty prours % Non-Trdbe Resturnts nd dining services % Non-Trdbe Post nd courier ctivities % Non-Trdbe Lbour recruitment nd provision of personne % Non-Trdbe ecurity nd cening ctivities % Non-Trdbe et services % Non-Trdbe Motor veices, motorcyces nd bicyces - se, mintennce nd repir nd reti trde of fue % Non-Trdbe duction % Non-Trdbe Recretion, cutur nd sporting ctivities % Non-Trdbe Lnd trnsport % Non-Trdbe Pubising nd printing % Non-Trdbe Teecommunictions % Non-Trdbe oese trde (exc. motor veices nd motorcyces) % Non-Trdbe Reti se (exc. ses, mintennce nd repir of motor veices) % Non-Trdbe Insurnce nd soci insurnce funds % Non-Trdbe Bnking nd oter finnci institutions % Non-Trdbe otes nd ccommodtion services % Non-Trdbe Mnufcture of food products % Non-Trdbe Business ctivities n.e.c % Non-Trdbe Mnufcture of soft nd cooic beverges nd tobcco products % Non-Trdbe Mnufcture of wood nd wood products % Non-Trdbe Mnufcture of pper nd pper products % Non-Trdbe Mnufcture of furniture % Trdbe Mnufcture of non-metic miner products % Trdbe Mnufcture of met products (exc. mcinery nd equipment) % Trdbe Mnufcture of bsic met % Trdbe Computer nd reted services % Trdbe Auxiiry trnsport ctivities % Trdbe Mnufcture of pstic nd rubber products % Trdbe Mnufcture of wering ppre (except knitted) % Trdbe Mnufcture of refined petroeum nd its products % Trdbe Mnufcturing n.e.c % Trdbe Mnufcture of industri equipment, medic nd scientific equipment % Trdbe Mnufcture of eectric motors nd eectric distribution pprtus % Trdbe Air trnsport % Trdbe Mnufcture of eectronic components % Trdbe Mnufcture of texties % Trdbe Mnufcture of mcinery nd equipment % Trdbe Mnufcture of eectronic communiction equipment % Trdbe Mnufcture of trnsport equipment % Trdbe Mnufcture of footwer nd eter products % Trdbe

29 29 Tbe 2: cognitive test scores for workers in te trdbe nd non-trdbe industries* ector Obs Men td. Dev Min Mx Trdbe 49, Non-Trdbe 9, A 68, *Te dt represents te yer 20 for workers in te ge group (n=68,664). Te dt ws restricted to workers wo competed t est n undergrdute degree nd were empoyed t te time of te survey on fu time bsis. ource: Te Isrei Centr Bureu of ttistics. Tbe 3: stimtion resuts Dependent vribe nnu growt rte in GDP per cpit per our worked, () (2) (3) (4) (5) Log (GDP per our worked in 995) (0.053)*** (0.055)* (0.52)** (0.088) Log (GDP per our worked in 995) (0.006)** (0.006) (0.006) (0.00) Log (scoo yers in 995) (0.0344) (0.034) (0.040) (0.052) Openness rtio (0.0699)** (0.007)*** (0.007)*** (0.099) Openness rtio X scoo yers (0.005)*** (0.007) Number of observtions R-squred * p<0. ** p<0.05 ***<0.0 Averge yers of scoo re used for yer-ods. Te Openness rtio mesures te sre of exports nd competitive imports witin te tot suppy of te industry in Tis mesurement yieds n index rnging from 0-. Te nysis is conducted on 42 non-mining/non-re-estte/non-gricutur/non-construction business sector industries. Te dependent vribe is te re nnu growt rte in GDP per cpit per our worked during te yers

30 30 Tbe 4: ummry sttistics Vribe Obs Men td. Dev. Min Mx GDP per our worked in Openness rtio scoo yers in Openness rtio X scoo yers