CentrePiece - Centre for Economic Performancecep.lse.ac.uk/pubs/download/CentrePiece_12_3.pdf · CentrePiece Winter 2007/08 1 page 2 School segregation and its consequences Stephen

The Magazine of The Centre for Economic Performance Volume 12 Issue 3 Winter 2007/08

ISSN 1362-3761

School segregationIntergenerational mobilityForeign relations and tradeComputers and crime-fighting

ImmigrationThe death of distanceInequality and productivitySmoking during pregnancy

THE PSYCHOLOGY OF SAVINGS

C e n t r e Pi e c e

Our schools are becoming more sociallysegregated. The life chances of childrenfrom poor families are lower nowadays.Immigrants are threatening the jobs of British-born workers. Blunt assertionsof this kind are commonplace in the language of politicians andcommentators from across the politicalspectrum. But how true are any ofthese claims? Which ones reflect realityand which are myths?

The answers can only come fromtaking a proper look at the evidence.Much of the work of researchers at theCentre for Economic Performance (CEP)aims to do just that, gathering dataacross a wide range of social phenomenaand rigorously applying the tools ofeconomic analysis to see what we canlearn about such pressing policy concernsas education, inequality and immigration.

An underlying theme in thisCentrePiece is the value of researchevidence for informing public debate.For example, on whether high- and low-achieving children are being educated

in separate schools, Stephen Gibbonsand Shqiponja Telhaj find that there isstrong segregation by ability but it doesnot appear to be increasing over time.And Marco Manacorda and colleaguesfind that while immigration does havesome labour market effects, they seemto be on the wages of earlier migrants.

On intergenerational mobility, aseries of CEP studies has revealed that ithas indeed fallen between the cohortof British children who grew up in the1960s and early 1970s and those whogrew up in the 1970s and 1980s. But westill know little about changes sincethen or what has happened in othercountries. Here, Maia Güell andcolleagues explain a new technique for measuring the importance of familybackground for people’s outcomes inlater life.

One area in which it is particularlycommon to hear bold statementsunencumbered by evidence is theimpact of computers. Informationtechnology has led to the ‘death of

distance’, various pundits have asserted.Computers may be in every workplacebut they’ve failed to make us moreproductive, argue others.

Three studies reported here touchon these issues. One uses patentcitations to see if innovations areflowing more quickly betweencountries. Another examines whetherthe rise in inequality is linked to firms’ use of new technology. And thethird investigates whether computershave helped US police departments tofight crime.

As always, your feedback onCentrePiece is welcome. And look outfor other CEP efforts to summarise thelatest research evidence in the policyanalyses available on our website:http://cep.lse.ac.uk/_new/publications/series.asp?prog=CEPPA

Romesh [email protected]

CentrePiece is the magazine of the

Centre for Economic Performance at the

London School of Economics. Articles in this

issue reflect the opinions of the authors, not

of the Centre. Requests for permission to

reproduce the articles should be sent to the

Editor at the address below.

Editorial and Subscriptions Office

Centre for Economic Performance

London School of Economics

Houghton Street

London WC2A 2AE

Annual subscriptions for one year (3 issues):

Individuals £13.00

Students £8.00

Organisations (UK and Europe) £30.00

Rest of world £39.00

Visa and Mastercard accepted

Cheques payable to London School of

Economics

CEP director, John Van Reenen

CEP research director, Stephen Machin

Editor, Romesh Vaitilingam

Design, DesignRaphael Ltd

Print, Ghyllprint Ltd

© Centre for Economic Performance 2008

Volume 12 Issue 3

(ISSN 1362-3761) All rights reserved.

Centre Piece

Editorial

CentrePiece Winter 2007/08

1

page 2 School segregation and its consequencesStephen Gibbons and Shqiponja Telhaj explore the extent to which high- and low-achieving pupils are separated into different schools

page 6 Is distance dying at last?John Van Reenen and colleagues use patent citations to show that innovations now flow more quickly between countries

page 14 The psychology of savings and investmentDavid Laibson talks about his work in the new field of behavioural economics

page 18 Inequality of individual wages and the dispersion of firm productivityJohn Van Reenen and colleagues find a major source of rising inequality in the greaterdispersion of firms’ productivity

page 21 The labour market effects of immigrationMarco Manacorda, Alan Manning and Jonathan Wadsworth investigate the effectsof increased immigration on wages and employment

page 24 What’s in a name? Information on intergenerational mobilityMaia Güell and colleagues use surnames to measure the importance of familybackground for people’s outcomes in later life

Contents in brief...

p21The labour market effects of immigration

p12Freedom Fries:

when internationalrelations damage

international trade

page 11 Smoking during pregnancyEmma Tominey argues that women whosmoke in early pregnancy should not bewritten off as 'too late' to help

page 12 Freedom fries: wheninternational relationsdamage international tradeGuy Michaels and Xiaojia Zhi ask whathappens to trade when public attitudestowards a foreign country turn negative

page 27 Do computers help policefight crime?Paul Heaton and Luis Garicano studythe impact of information technology onthe productivity of US police departments


2

School segregationand its consequences


3

Expansion of choice hasbecome a central themeof recent educationalpolicy debates inEngland andinternationally. Some

argue that freedom of choice ensures that pupils and schools are efficientlymatched, and that ‘quasi-market’discipline induced by open competition forpupils encourages schools to adopt moreefficient teaching methods.

Others point to the possibly adverseconsequences of a more ‘segregated’school system, in which pupils are lesslikely to mix with children unlikethemselves in terms of background andability. According to these arguments,more choice is bad – either becausesegregation seems to imply inequality andis inherently socially undesirable; orbecause it is claimed that a segregatedschool system is educationally inefficient.

Setting aside broad and subjectivearguments over the desirability of ethnicand social mixing, the most compellingcause for concern is that the separation ofpupils into academically advantaged anddisadvantaged groups may work toexacerbate inequalities in educationaloutcomes without producing any overall benefits.

This can occur if children areinfluenced by the achievements andbehaviour of their schoolmates – so-called‘peer group’ effects – or simply becausedisadvantaged pupils place greaterpressure on teaching resources and soharm the chances of others in theirclassroom or school.

These issues have become highlyrelevant in the light of policies that seek toexpand parental choice, and our research

has sought some answers to the keyquestions: what is the extent ofeducational segregation? Are the patternschanging? What are the implications forindividual pupils in terms of theirachievements? And what contributiondoes this make to educational inequality?

Are secondary school pupilseducationally segregated?Most of the debate about schoolsegregation (and hence much previousresearch) has been concerned withdemographic and socio-economiccharacteristics. Relatively little attentionhas been paid to the important issue ofsegregation that is explicitly along lines ofeducational advantage and disadvantage –presumably the key concern to thoseworried about inequality in education. Forparents too, the real consideration seemsto be the stratification along lines of pupil‘ability’ or capacity to achieve.

In the first part of our research, welook explicitly at the extent to which high-achieving and low-achieving pupils areseparated into and educated in differentsecondary schools – and how this changedbetween 1996 and 2002. We do this byexamining differences in the compositionof secondary state schools in terms of theacademic achievements of pupils at thetime they start school.

The analysis is based on pupils’ resultsin their maths, science and English keystage 2 tests at the age of 11, the end ofprimary school, coupled with informationon which secondary schools they move on to. These data are collected for nearlyall state school pupils, so we can studywhat is happening for almost all schoolsand almost the entire population of pupilsin England.

What is the best way to analysedifferences in the composition of schools?Previous research has typically used varioussegregation or inequality indices – singlenumbers that summarise the level ofinequality at a particular point in time –but we adopt a more graphical approach.

Imagine ranking all pupils enteringsecondary schools in increasing order oftheir test scores at age 11. Pupils in thetop 1% receive a score of 100, those in thebottom 1% receive a score of 1 and thosein the middle a score of 50, and so on. Foreach secondary school, we can calculatethe position in that ranking of the averagepupil, a score that summarises the averageintake ability in each school.

Then, with these average school scores,we can rank schools in the same way. The 1% of schools receiving the lowest-achieving average pupil get a score of 1 and the 1% with the highest-achievingaverage pupil gets a score of 100, and so on.

Plotting each school’s average pupilranking against its own ranking in thedistribution of schools provides a fairlycomplete visual description of the way inwhich pupils with different priorachievements are segregated into differentsecondary schools. If there were very littlesegregation, then there would be very little

To what extent are high-achieving and low-achieving pupils in England separated into and educated in different secondary schools?Research by Stephen Gibbons and Shqiponja Telhaj looks at patterns ofsegregation by ability and their impact oneducational outcomes.

The average ability ofchildren going into the

best comprehensiveschools is way abovethe average ability in

the worst

difference between the achievement of theaverage pupil as we move from the worstto the best schools, and the plot wouldtend towards the horizontal. As pupilsbecome segregated along lines of ability,the plot steepens towards a 45-degreeslope with the top 1% of pupils in the top1% of schools and the bottom 1% ofpupils in the bottom 1% of schools.

Figure 1 shows the picture for Londonfor three years: 1996, 1999 and 2002. Thetop panel shows comprehensive schoolsonly, while the lower panel shows allschools, including religious schools andgrammar schools. For comparison, we alsodraw the 45-degree line that represents ahypothetical school system in which pupilsare completely segregated according totheir prior achievement.

It is clear that even for comprehensiveschools, there is quite strong segregation byability. The gap between the average pupilin the highest-ranking school and theaverage pupil in the lowest-ranking schoolspans almost one-third of the distributionof pupil achievement. In other words, theaverage pupil in the bottom school is in thebottom third of pupils ranked by test scoresat age 11, while the average pupil in thetop school is in the top third. In the bottompanel, which adds in other types of schools,the differences are even more pronounced.

What drives differencesbetween school intakes?Our analysis does not answer this questiondirectly. But since comprehensive schoolshave no scope to choose pupils, these

patterns are most likely to arise becausethese schools serve local neighbourhoodsand different neighbourhoods containpupils from very different socio-economicbackgrounds. All the children in a richneighbourhood go to school together, andall the children in a poor neighbourhood goto school together. Such differences arisenot because of the freedom for parents andpupils to choose schools, but because of alack of choice given where a pupil lives.

In contrast, some segregation couldalso arise because the highest-abilitychildren are free to choose the sameschools as other higher-ability childrenwherever they live – the story of self-selection that underpins the critique thatschool choice generates inequalities. Thisconsequence of school choice is illustratedby the bottom panel: educationalsegregation is even greater when weinclude schools that can ‘cream skim’ pupilsby picking according to ability or have otherattributes – such as religious ethos – thatmake them likely to attract or choose pupilsof different types and abilities.

Whatever the causes of educationalsegregation, a striking feature of Figure 1is that there has been almost no change inrecent years. Comparing the lines for 1996to 2002 shows that almost nothing haschanged over these years in terms of theway that pupils of different abilities aresorted into different schools. The picturein other regions of England is invariablythe same: there are wide inequalities inintake between schools, but there hasbeen no systematic change over theperiod of our analysis.

Other approaches to the questionproduce similar answers: for example,there are no regions in which theproportion of schools accommodatingpupils with abilities among the top 5% isless in 2002 than in 1996, or where theproportion of schools accommodatingpupils from the bottom 5% hasdecreased. In fact, our findings suggestthat the general trend is towards a more


4

Figure 1:

Segregation by achievement in secondary school intakes in London

-20

-10

0

10

20

30

40

80 1006040200

■ 1996■ 1999■ 2002

-20

-10

0

10

20

30

40

80 1006040200

■ 1996■ 1999■ 2002

Comprehensive schools

All schools

Pri

or a

chie

vem

ent

of t

he

aver

age

pu

pil

in

a s

choo

lP

rior

ach

ieve

men

t of

th

e av

erag

e p

up

il i

n a

sch

ool

School rank from worst to best intake

School rank from worst to best intake

Educationalsegregation is even

greater when theanalysis includesschools that can

‘cream skim’pupils by ability

or religion


5

research, we find that any contribution thatpeer groups make to a child’s academicprogress is quite small: a move from theworst to the best comprehensive school (interms of the intake ranking in Figure 1)would make only a slight difference to howwell a child progressed in the first few yearsof secondary school. It seems unlikely thatthe balance of educational success orfailure will be tipped according to whethera child attends a school alongside high- orlow-ability children.

This claim might seem puzzling giventhat parents seem to make great efforts tofind schools with good peer groups. Butbetter peer groups may provide otherbenefits – physical safety, emotionalsecurity, familiarity, lifetime friendshipnetworks or simply exclusivity – whichmakes schools with good peer groups verydesirable, even if they offer only slightacademic advantages. Perhaps it is herethat individuals really win or lose outthrough both socio-economic andachievement-based school segregation.

even distribution of these groups across schools.

So overall, we do not find any dramaticor systematic changes in schoolcomposition in terms of the abilities ofpupil intakes. This is an important result asit runs counter to the tales of increasedstratification and segregation that arecommonplace in academic, media andpolitical circles. The idea that pupils of high-and low-ability have become increasinglysegregated seems to be something of amyth, at least in recent years.

But there are large and stabledifferences in intake between schools ofthe same general type even if they havelittle autonomy in control of pupiladmissions: the average ability of pupilsgoing into the ‘best’ comprehensive schoolsis way above the average ability in theworst. It is surely this fundamental contrast– driven for the most part by geographicaldisparities in pupil background arising fromresidential segregation – that drivesperceptions of inequity in school provisionand of failings in the system.

Does segregation matter?Whether the patterns of segregation thatwe find are cause for concern depends inpart on whether such segregation isconsidered socially desirable, but also onwhether peer group ability has any realimpact on individual pupils. Anecdotally,schools often seem to be judged andchosen by parents on the basis of the kindof children they enrol, rather than thequality of their teaching or other facilities.

So do parents’ apparent preferencesmean that peer groups have a big impacton how well a child does at school and inlater life? Certainly, pupils in schools withlow-achieving peers are more likely to dobadly later on, and pupils in schoolssurrounded by high-achievers are morelikely to do well. But is this just becausehigh-ability pupils tend to go to schoolswith other high-ability pupils and low-abilitypupils go to schools with other low-abilitypupils? Or does our peer group reallymatter for our own success?

Answers to these questions areimportant for a number of reasons. First, ifpeer groups matter, then a segregatedschool system means that children could bedisadvantaged by the school they attend,even if teaching standards are as high thereas anywhere else. This could lead topersistence or growth in inequality as the

lowest-achievers get the worst deal andhigh-achievers the best deal in terms ofpeer group quality. Whether, on balance,this inequality generates benefits or costsfor society depends on whether thewinners gain more than the losers – aquestion underlying the familiar debateabout the benefits of streaming versusmixing in schools.

Aside from these issues, it is worthknowing if individual pupils respond to theirpeers, because it can mean thateducational interventions that appearbeneficial to one pupil in isolation may beeven more effective when rolled out to thewhole population. This ‘social multiplier’effect arises because an intervention, suchas a new teaching method, benefits a pupildirectly but also indirectly via its impact onhis or her schoolmates.

Our evidence on these questions isbased on what happens to educationaltrajectories when pupils move from primaryto secondary school and meet newschoolmates. We use this change in peergroup quality to examine whether thedifferences between school intakeillustrated in Figure 1 have any influence ona pupil’s subsequent progress in tests up tothe age of 14.

We find that pupils do make betterprogress in maths and English in the earlystages of secondary school if their newschoolmates have a good record of priorachievement. And it really is priorachievement that seems to matter: otherfactors, such as ethnic mix, agecomposition and low-income schoolmates,have no direct effects on a child’s progress.

This is encouraging because pupils’prior attainments are surely moreamenable to early interventions than socio-economic and demographic characteristics.But it also means that the patterns ofsegregation in secondary schools couldhave real consequences in terms ofeducational inequality.

Even so, in line with most previous

This article summarises ‘Peer Effects and

Pupil Attainment: Evidence from Secondary

School Transition’ by Stephen Gibbons and

Shqiponja Telhaj, Centre for the Economics of

Education (CEE) Discussion Paper No. 63

(http://cee.lse.ac.uk/cee%20dps/ceedp63.pdf)

and ‘Are Schools Drifting Apart? Intake

Stratification in English Secondary

Schools’ by Stephen Gibbons and Shqiponja

Telhaj, CEE Discussion Paper No. 64

(http://cee.lse.ac.uk/cee%20dps/ceedp64.pdf).

Stephen Gibbons is a senior lecturer in

economic geography at LSE. Shqiponja

Telhaj is a lecturer in economics at the

University of Sussex. Both are research

associates in CEP’s education and skills

programme.

Patterns of segregationin secondary schoolscould have realconsequences in termsof educational inequality

Commentators have for yearsbeen claiming thateconomic activity no longerrespects internationalfrontiers. But while cheap

communication, falling shipping costs andthe internet make things easier, face-to-face interaction remains as important asever, even in high-tech sectors likesoftware and biotechnology – just think ofSilicon Valley or Bangalore. And if onelooks at trade patterns between countries,international boundaries still matter a lot.

Our new research finds the firstevidence that distance really is dying – atleast in the world of ideas. Having lookedat over two million patent citations over aquarter of a century and broken themdown by the country of the inventors(covering just about every nation in theworld), we find that national barriers arecrumbling when it comes to the flow ofinnovations, as measured by the relativespeed of patent citations across countries.

We find that there was a great deal of‘home bias’ between 1975 and 1990: forexample, German inventors cited otherGermans 14% faster than Americaninventors cited German patents. But the

really interesting thing is how thesecitations have changed over time: since1990, the Americans have been only 5%slower at citing Germans than theGermans themselves; and the French only1% slower. So even though information isspreading faster within countries, it hasalso started to flow really quickly betweencountries.

The importance ofgeographyEconomists dating back to Alfred Marshallin the late nineteenth century haveemphasised the importance ofgeographical proximity for the flow ofideas and the diffusion of technologies.Inventors find it hard to captureknowledge: like water, it ‘spills over’ toother people. Geographical proximityfacilitates this process if face-to-faceinteraction is important, for example,

where knowledge is ‘tacit’ so that it is notcodified in writing or in standard industrypractices. Workers and managers meetingat formal and informal meetings andmoving between firms will help totransmit new ideas betweenorganisations.

The idea that closeness matters forthe diffusion of knowledge also liesbehind a number of government policies.Subsidies for research and development(R&D) generally target activities that aregeographically located within the homecountry. These typically seek to attractshiny high-tech firms to locate in specificareas, based on the premise that theiractivities will have the highest ‘positiveexternalities’ – benefits not only for theinvestors but also for firms and workersnearby. Other policies aim to encourageclusters of high-tech firms, based on thebelief that the sum of benefits from


7

Is distance dying at last?

When it comes to theflow of new ideas,

distance is now dying

Economists have long been sceptical of claims aboutthe ‘death of distance’ – the idea that new technologyhas diminished the significance of geography foreconomic outcomes. Research by Rachel Griffith,Sokbae Lee and John Van Reenen on patentcitations over time finds the first hard evidence thatdistance is indeed becoming less important.


8

innovations such as the internet. We usepatent citations to address whether therumours of the death of distance havebeen somewhat exaggerated. In particular,we look at the speed of knowledge flowsbetween countries.

A simple look at the raw data suggestssubstantial evidence of home bias in theway that knowledge is transmitted: beinggeographically close does makeknowledge transfer easier. But we also seeevidence that distance has become lessimportant over time for the internationaltransmission of ideas.

Figure 1 plots the relative speed ofpatent citations across countries and over

clusters of firms is greater than theconstituent parts.

In contrast to the idea thatgeographical proximity is important fortechnology transfer, the notion of the‘death of distance’ has recently flourishedin the popular imagination (for example,Friedman, 2005; Cairncross, 1997; Coyle,1997). Messiahs of globalisation claimthat information travels around the globeat rapid speed from California to Calcuttathrough the internet, conferences,telephone and other communicationdevices. Under this view, geography playslittle role.

There are plenty of anecdotes, butwhat is the hard evidence that distancematters? And to what extent have theychanged over time?

Existing research:distance mattersAnswering these questions turns out tobe difficult for a number of reasons. First,how do we trace knowledge flows? Andsecond, how can we distinguish theimportance of geographical proximity forknowledge flows from other factors thatmay be associated with geography?

Patent citations have become animportant source of information on theway that knowledge flows between firmsand countries. When an inventor takesout a patent, they have to providecitations to the prior technology from which their ideas are drawn. This is a pretty direct measure ofknowledge flows.

One prominent study that uses patentcitations in this way is by Adam Jaffe andcolleagues (1993). They show thatinventors are far more likely to cite otherinventors living close by than inventors inmore distant locations.

Several studies have followed thisapproach, and the consensus that hasemerged is that knowledge really issubject to a significant degree of homebias. Our earlier study (Griffith et al,2006) also shows that British firms wholocate R&D labs in the United States arebetter able to tap into Americanknowledge than those that do not.

New research:distance starting to dieBut much of this research uses older data.Our latest study aims to establish whetherthings have changed recently with

-5

0

5

10

15

20

Rest of worldUSJapanEUUKFranceGermany

Figure 1:

Distance matters less for the flow of ideas

Innovations areflowing morequickly betweencountries, asmeasured by therelative speed ofpatent citationsover time

Speed of citation: Example of inventions from Germany

German cited patents

� Early (1975-1989)

� Late (1990-1999)

Notes: This graph shows the relative percentage time (in mean number of days) from the date

that a German inventor was granted a patent until the first citation of that patent (by the

location of the inventor that made the citation). For example, the first bar (blue) for France in the

early period indicates that the French inventor took 4% longer to cite a German invention than a

German inventor took to cite another German inventor. This fell to 1% in the 1990s (red bar).


9

-10

-5

0

5

10

15

20

25


-10-505

101520

253035


-10-8-6-4-202468

1012


0

10

20

30

40

50


-9

-6

-3

0

3

6

9

12

15


Figure 2:

Distance matters less for the flow of ideasSpeed of citation: All other countries

French cited patents

EU cited patents

UK cited patents

Japanese cited patents

US cited patents

� Early (1975-1989)

� Late (1990-1999)

Notes: As in Figure 1. Height of bar

indicates how much slower a country is in

citing another country.

time for the example of inventionsdiscovered in Germany. These are allsuccessful applications to the US PatentOffice for inventors living in Germany in an ‘early’ period (1975-89) on the leftand then in a ‘later’ period (1990-99) onthe right.

Looking first at the early period, theheight of each bar indicates how muchslower foreign inventors were in being firstto cite Germans relative to other Germaninventors. In the 1970s and 1980s,American inventors were about 14%slower in citing Germans than theGermans themselves.

Figure 2 looks at inventions cited inthe other OECD countries. The fact thatthe bars in both Figures 1 and 2 arealmost all positive suggests that thephenomenon of home bias in ideas is aliveand well – Germans are quicker at citingother Germans, British quicker at citingother British, and so on. What is moreinteresting is how home bias has changedover time – on average the bars in thelater period are lower than the bars in theearlier period. This suggests that homebias in ideas has fallen. After 1990 inFigure 1, the French are only about 1%slower in citing Germans, and theAmericans only about 5% slower in citingGermans inventors, than the Germansthemselves (only the British seem to beslower off the block in citing Germans inthe 1990s!).

This looks promising and suggestsdistance is dying. But there are manyreasons why the simple patterns in theraw data might be misleading. Inparticular, knowledge may spread more orless quickly due to many patentcharacteristics that may be poorlycaptured by observable characteristics, andmay be associated with geography. Forexample, if high quality patents are citedmore quickly than lower quality patents,and if high quality patents are


10

geographically clustered for some otherreason, then this could give the impressionof home bias whereas in fact it is to dowith the higher average quality ofinventors in one location.

The traditional way to deal with this sortof problem in econometric research is tocontrol for unobserved and correlated ‘fixedeffects’ (things which we don’t measure butare pretty much fixed over time). Lee (2007)develops an econometric estimator thatallows us to do this by taking ‘quasi-differences’ over time between multiplecitations on the same patent.

Our research suggests that controllingfor these unobserved characteristics makesan important difference to estimates ofthe importance of home bias in innovativeactivity:� First, the evidence for home bias is

much weaker once we control for fixedeffects. The non-fixed effects models(which are standard in the researchliterature) suggest home bias in amajority of cases, whereas our preferredmodels indicate home bias in only aminority of cases.

� Second, home bias is much stronger inthe ‘traditional’ sectors of the economy(such as chemicals and mechanicalengineering) than in more ‘modern’sectors (such as computing). This isconsistent with the idea thatinformation diffuses fasterinternationally in the modern sectors.

� Finally, and most interestingly, we findevidence that home bias has declinedover time, being much stronger in thepre-1990 period than the post-1990period. We interpret this as suggestingthat information flows more easilyacross national boundaries as the costof international communication andtravel has fallen.

Don’t bury economicgeography just yetSo the bottom line is that when it comesto the flow of new ideas, we find hardevidence that distance has become less

important over time. But for many sectors,especially in the more traditional andmature technologies, it is not yet dead.

One policy implication of this researchis that it makes less sense to subsidisecorporate R&D if the ideas generatedbenefit other countries very quickly.Similarly, firms may worry more that theirinnovations may earn them profits for lesstime than in the past, as foreign firmslearn to imitate and leapfrog them.

If new ideas benefit other countries more quickly nowadays, it may make less sense to subsidisecorporate R&D

For traditionalsectors withmaturetechnologies,distance is not yet dead

This article summarises ‘Is Distance Dying at

Last? Falling Home Bias in Fixed Effects

Models of Patent Citations’ by Rachel Griffith,

Sokbae Lee and John Van Reenen, CEP

Discussion Paper No. 818 (http://cep.lse.ac.uk/

pubs/download/dp0818.pdf).

Rachel Griffith and Sokbae Lee are

at University College London.

John Van Reenen is director of CEP.

Further reading

Frances Cairncross (1997), The Death of

Distance: How the Communications

Revolution Will Change Our Lives, Harvard

Business School Press.

Diane Coyle (1997), The Weightless World:

Strategies for Managing the Digital

Economy, Capstone/MIT Press.

Thomas Friedman (2005), The World is Flat:

A Brief History of the Twenty-first Century,

Farrar, Strauss and Giroux.

Rachel Griffith, Rupert Harrison and John

Van Reenen (2006), ‘How Special is the

Special Relationship: Using the Impact of US

R&D Spillovers on British Firms as a Test of

Technology Sourcing’, American Economic

Review 96(5): 1859-75.

Adam Jaffe, Manuel Trajtenberg and Rebecca

Henderson (1993), ‘Geographic Localization

of Knowledge Spillovers as Evidenced by

Patent Citations’, Quarterly Journal of

Economics 108(3): 577-98.

Sokbae Lee (2007), ‘Estimating Panel Data

Duration Models with Censored Data’,

Econometric Theory, forthcoming.

Alfred Marshall (1890), Principles of

Economics, Macmillan.


11

Maternal smoking during pregnancy remains prevalent inmany countries despite decades of research testifying tothe harm it imposes on the unborn child. Following the1998 ‘Smoking Kills’ White Paper, which identified endingmaternal smoking as a target of future UK governmentpolicy, a national telephone helpline was launched to helppregnant mothers devise strategies for quitting smoking.

But since one in five mothers in the UK still smoke whilepregnant, current policy does not seem to be having a bigimpact. To gain a better understanding of how policy canget its message across more strongly and target pregnantmothers more effectively, research by Emma Tomineyexplores how smoking during pregnancy lowers childhealth at birth.

The study confirms that mothers who smoke duringpregnancy will have smaller babies – typically 5.4%(6.5oz) lighter than other babies. But around half of thisdamage is because of ‘unobservable traits’ of the mother– including other health risks she might take, such asdrinking alcohol, and her nutrition and knowledge ofhealthy behaviour.

The research also finds that the lasting harm to babies ofsmoking during pregnancy is greatest if the mothers havea lower level of education. Children born to mothers wholeft school at the age of 16 suffer double the harm foreach cigarette smoked. This suggests that the governmentmust target its anti-smoking policy directly at poorlyeducated families.

And it is important to note that women who do smoke inthe early stages of pregnancy should not be written off asbeing too late to help. Surprisingly, the research showsthat the harm to the baby is essentially reduced to zero ifthe mother quits by month five of the pregnancy. This ismuch longer than conventional wisdom and previousresearch have suggested.

Possible causes for this finding could either be that theharm from smoking accumulates during the final monthsof pregnancy, or because the mothers who quit smokingsimultaneously reduce their health risks in other ways.Either way, this tells us that there is more time than wethought to help the mothers change their behaviourduring pregnancy.

Overall, the research suggests that while stopping motherssmoking during pregnancy is important, it is only half thebattle. Other studies have shown that the effects of beingborn underweight stay with a child throughout its life,affecting its health, education and earnings potential (seeBehrman and Rosenzweig, 2004, and Black et al, 2007).This study shows that in order to lower the incidence ofunderweight births, stopping a mother from smokingmust be combined with helping her to be healthier inother areas of her life.

Previous studies have identified a link between smokingand low birth weight, but none have looked in such depthat whether the education of the mother can alter this andhow the harm accumulates during pregnancy. EmmaTominey’s study analyses data on the lives of 6,500children and their mothers, and goes into exceptionaldetail about the mothers’ lifestyle, tracking them fromtheir child's birth until the age of 42. This proved to bevery important, as not only does the harm vary accordingto the mother’s education, but traits of the mother that are often unobservable account for around half ofthe harm.

The study calls on the government to alter radically itspolicy on helping pregnant women quit smoking, inparticular targeting the children of poorly educatedmothers. A much more holistic approach to improvingchild health in pregnancy is needed to help thousands ofchildren break out of the poverty trap.

This article summarises ‘Maternal Smoking During Pregnancy and

Early Child Outcomes’ by Emma Tominey, CEP Discussion Paper

No. 828 (http://cep.lse.ac.uk/pubs/download/dp0828.pdf).

Emma Tominey is an occasional research assistant at CEP.

Further reading

Jere Behrman and Mark Rosenzweig (2004), ‘Returns to Birth Weight’,

Review of Economics and Statistics 86(2): 586-601.

Sandra Black, Paul Devereux and Kjell Salvanes (2007), ‘From the

Cradle to the Labour Market? The Effect of Birth Weight on Adult

Outcomes’, Quarterly Journal of Economics 122(1): 409-39.

in brief...Smoking during pregnancy


Freedom fries:when international relationsdamage international trade

in brief...

12

When public attitudes towards a foreign country turn negative,what are the effects on trade, and especially trade in firm inputs? Guy Michaels and Xiaojia Zhi explore this question using thedeterioration in Americans’ opinions of France following the twocountries’ dispute over the invasion of Iraq – a deterioration so drastic that French fries were renamed ‘freedom fries’.

International trade flows maybe sensitive to large changesin relations and attitudes

When the French government opposed US efforts toobtain a United Nations mandate to use military forceagainst Iraq in 2002, the resulting standoff led to amassive deterioration in Americans’ opinion of France.Our research uses this episode to track the impact of suchworsening international relations on the businessdecisions of firms involved in trade between the twocountries. We find significant effects: worsening attitudesdecreased trade.

According to US polls of public opinion, in February 2002,83% of Americans viewed France favourably; by March2003, this proportion had declined to 35%. Even three

years later, in February 2006, the fraction of Americanswith a positive view of the French had recovered only asfar as 57%.

Very negative attitudes towards France became commoneven among affluent, college-educated Americans, so theywere likely to be prevalent among managers. By contrast,attitudes towards Germany worsened much less andrecovered quite quickly, and attitudes towards Britain, Italyand Spain changed very little.

The US government also singled out France in particular:Condoleezza Rice, then the president’s national security


13

adviser, was quoted by Reuters in March 2003 tellingassociates to ‘punish France, ignore Germany and forgiveRussia’. There were also calls in the media to ban Frenchproducts, and the House of Representatives cafeteriabegan to serve ‘freedom fries’ instead of French fries.

This shock to relations between France and the USprovides an ideal opportunity to examine the effect ofattitudes on trade. In other circumstances, internationaltrade may in itself affect attitudes. For example, perceivedgains from trade may motivate the US and China toimprove relations. But at the same time, increased trademay cause Americans to fear that China threatens theirjobs, worsening attitudes.

But in the case of the United States and France, there areclear reasons for the worsening attitudes and thosereasons are unrelated to trade. The US-France setting alsoavoids the problem that worsening attitudes may becorrelated with increased personal risk; this may be aproblem in other circumstances, where worseningattitudes may cause risk of war or terrorism. And usingother OECD and eurozone countries as plausiblecounterfactuals for France makes it possible to net out theshort-run changes in trade that may have taken place inthe absence of worsening relations.

Our study finds that US imports from France fell by about15% and US exports to France fell by about 8%compared with other OECD or eurozone countries. Thisdecline was due in large part to a fall in France’s share ofthe quantity of inputs traded with the United States.There was a similarly large decline in both US businesstrips and tourist visits to France, suggesting thatworsening relations did indeed affect transactionsbetween firms.

One interesting aspect of this natural experiment is that itis difficult to rationalise the decline in trade of inputsusing standard arguments of firms choosing inputs thatminimise their costs of production. The attributes of inputsproduced in France and by competing input producers,and the characteristics of firms’ production processes inthe United States are not likely to have changed in theshort run. This suggests that tastes – and not only simplecost-minimising calculations – may affect firms’ choice ofinputs.

These results also suggest that international trade flowsmay be sensitive to large changes in relations andattitudes. We conclude that the effect of attitudes may beparticularly strong where there are strong incentives topunish a foreign country – as in the case of commoditiesused by governments – or where the availability of closesubstitutes lowers the cost of changing a firm’s inputsuppliers. This result may be especially important forunderstanding the robustness of trade flows betweenWestern countries and other important trade partners.

US imports from France declinedsharply following the two

countries’ standoff over Iraq

This articles summarises ‘Freedom Fries’ by Guy

Michaels and Xiaojia Zhi, CEP Discussion Paper No.

815 (http:/cep.lse.ac.uk/pubs/download/dp0815.pdf).

Guy Michaels is a lecturer in economics at LSE

and a research associate in CEP’s labour

markets programme. Xiaojia Zhi is a research

economist at CEP.


14

Romesh Vaitilingam: David, you’re aneconomist but your lectures on savingsand investment are apparently aboutpsychology. What is it that you’re thinkingabout here?

David Laibson: Well, I am an economistbut I spend a lot of time thinking aboutthe psychological factors that influencepeople’s choices and I’ve titled theselectures ‘the psychology of savings andinvestment’ because I want to contrast thepsychological approach with thetraditional classical economic approach.

When I went to graduate school, wewere taught that everyone was rationaland that they optimised. More recently,the view has developed that while most ofour economic behaviour is pretty rationaland pretty optimal, we do occasionallydepart from that, and such departures canbe studied, measured and modelled. Thislecture series is about those efforts toenrich our understanding of economicbehaviour by adding the psychologicalcomponents that, in essence, completethe picture.

RV: So really this is a fairly recentinnovation, bringing these psychological

understandings into economics.

DL: The ‘big bang’ for this field came in1979, when a very important paper byDanny Kahneman and Amos Tversky waspublished. The paper was aboutsomething called prospect theory – or howpeople think about risky outcomes. HadAmos not died, they both would havebeen awarded a Nobel Prize; as it turnsout, Danny received the Nobel Prize in2002.

But even though the field started in1979, it had very few followers until the1990s. There was a period in thewilderness in the 1980s when there werejust a handful of people actively doingbehavioural economics. Then in the 1990s,life really got started in terms of more andmore people signing on to this endeavour.And now it’s an area that has a great dealof interest, enormous graduate studentinterest and a lot of research taking place.You could almost say that it’s become afad.

RV: So tell me a little more about how thisactually developed. For years and yearsnon-economists have said you guys havethis very peculiar view of human nature.

What was it that suddenly made itpossible to introduce psychology? Was itsomething to do with the tools andtechniques that economics haddeveloped?

DL: I think there were two sets ofdevelopments that happened during thelast 25 years. The first thing was that webegan to find empirical evidence thatcontradicted the rational actor model.Now just to be clear: no economist todayor 50 years ago believes that people aretruly perfectly rational.

What economists have always said isthat the rational actor model is a verygood approximation of how peoplebehave. It gets things mostly right just likea map gets things mostly right. A map ofLondon may miss out the hills and valleysbut it basically tells you how London is laidout. So just as a map can be useful even ifit’s not perfectly right, so the rational actormodel was always felt to be good enough.

But in the last few decades, more andmore evidence – both experimentalevidence from the laboratory and evidencefrom real markets – like financial markets– has contradicted many of the predictionsof the rational actor model.

We need a combination of psychology andeconomics to understand people’s savings andinvestment decisions, according to DavidLaibson, who recently delivered the 2007 Lionel Robbins Memorial Lectures at LSE.CentrePiece editor Romesh Vaitilingaminterviewed him about his work in the new field of behavioural economics.

The psychology of savings and investment


15

The other set of developments is thatwe have developed theoretical frameworksthat enable us to formalise thesedepartures. We have alternative modelsfor mathematically representing humanbehaviour. These alternatives don’t leavethe rational actor model aside but add toit and improve it. Now we have thecombination of new data andsupplemental models that are jointlymoving economics forward.

RV: Tell me about your own researchprogramme, and particularly how it relatesto one of the great policy challenges,getting people to save enough to pay fortheir old age. What kind of findings areyou coming up with and what lessons canwe learn from them?

DL: There are essentially two kinds ofthinking in my research. The first aims tounderstand the foundations of humanpreferences and human decision-making.And the second is to think about how wecan build institutions that help people do

what they want to do.The underlying psychology is the

psychology of instant gratification. It is thepsychology of a decision-maker in ahousehold who puts enormous weight onthe present and then drastically discountsevents that might only be a week away intime. It’s the person who says, I know Ishould exercise and I’ll do that next weekbut right now these chips look very good.Or the person who says, I know I shouldsave and I’m going to start saving veryseriously next month but tonight, howabout a bottle of champagne?

This psychology can be modelled andmeasured, and I’ve spent a lot of myacademic career trying to do that. Butthen the question arises, if people do putenormous weight on the present, it mightbe awfully hard for them to save. We seethat people say that they care a lot aboutthe future. They say, look, I’m not goingto save today but I will save tomorrow. I’mnot going to exercise now but I willexercise later. So it’s not as if we think thatour futures are irrelevant, just that we

By understandingthe psychologicalfoundations of humandecision-making,we can buildinstitutions thathelp people do what theywant to do


16

aren’t willing to invest in them right now.We prefer to make those investments nextmonth.

So we need to build institutions thathelp get us over that hump. And a lot ofthe work that I’m doing is trying to figureout how to create pension plans orsavings systems that help people save forretirement. These are people who mightnot be able to do it on their own but arethrilled when you make it easier for themby providing, say, a defined contributionpension plan. This would be a system, forexample, where you’re a new employee ata firm and you’re automatically enrolled inthe pension scheme, with some fraction ofyour salary deposited every month directlyinto a retirement account.

These automatic enrolment plans arehighly effective. People like them and theydon’t opt out of them. They accumulatelarge bodies of wealth for retirement. Yet,if you were to change the system and setit up so that instead of people beingautomatically enrolled they are only giventhe option to enrol, you find that very fewpeople join these plans. It takes years forthe typical person to enrol in a planwithout automatic enrolment.

So we have a situation where ourtendency to procrastinate undermines ourwillingness to save. One natural wayaround that is to make savings automaticand when you do that, people are verysatisfied with the net result. A lot of whatI’ll be talking about in these lectures isdescribing the underlying psychology ofresisting savings and the institutions thathelp people save, which is what they tellus they want to do anyway.

RV: So your work is informing policy thatwill, in effect, save people fromthemselves.

DL: That’s right. But we don’t want to betoo paternalistic. I’m an advocate of givingpeople good defaults and always givingthem the option to opt out if they don’twant to save. I recognise that governmentsometimes goes too far and that we don’twant to tell people how to live their lives.But I also recognise that left to their owndevices completely, people sometimesmake bad choices.

So we’ve defined the middle ground, ahybrid system that nudges people in theright direction, encouraging them to dothings that they view as desirable without

compelling them or coercing them in waysthat are too paternalistic or heavy-handed.Basically what we want to do is to makesavings easy. If you don’t want to save,that’s fine, you can opt out of the system.But if you want to save, which mostpeople think they should do, then it’smade simple for you, not difficult andtime-consuming.

RV: Is there also a role for education inthis? Few people seem to have a goodgrasp of basic numbers, but can we teachthem to manage their finances moreeffectively?

DL: Well, it’s true that people have verylow levels of financial literacy, and some ofthe work that I’ve been involved in hasbeen studying these financialvulnerabilities in terms of low levels ofeducation or knowledge about finance.But if we think that education is animportant piece of the puzzle, we have tonot only document that education ismissing but also show that educatingpeople leads them to behave in betterways.

Surprisingly, there’s very little evidencethat supports that kind of policy. It’s notclear that education programmes can

increase people’s financial literacy andimprove their behaviour in ways that arecost-effective. While it’s true that if wegave everyone a PhD in finance, we’dprobably see the world saving much morecorrectly and investing more rationally,that’s an awfully expensive intervention.

Then the question becomes, can westrip it down and generate much simpler,cheaper, quicker interventions that get thejob done? And the answer so far is,maybe not. It doesn’t look like there areeasy educational interventions that trainpeople to make optimal or nearly optimalfinancial decisions.

For example, it’s a good idea to have ahigh school curriculum that emphasisesmore economic and financial issues. Buteven if we did, it might well be the casethat by the time that 18-year-old reaches48, the world will have changed. Andwhat she learned in high school may nolonger be that relevant for what she willhave to do as an adult.

If I had been taught how to save forretirement when I was in high school, Iwould have been taught, find a companywith a good defined benefit pension planand work there for the rest of your life.And that turns out to be terrible advice.First, all of those plans are being


17

terminated. And second, in the job marketof today, you don’t stay at one companyyour whole life: you move around and ifyou move around, as you should toadvance your career, you’re going to losethose pension benefits.

So if we think education is important,we have to first prove that we actually cancost-effectively educate people and helpthem make better choices. I think thatautomaticity is a more cost-effective toolin helping people make good choices.Rather than spending a lot of time andenergy educating someone, which may ormay not get them to make a betterchoice, I know that if I automatically enrolthem in a defined contribution pensionplan with a 5, 6 or 7% savings rate and Iautomatically allocate their assets to a lifecycle fund, they’re going to do very well.And they’re going to basically stick withthis good default.

So the question is, do I want to spendthousands of dollars on education for thatperson or do I want to automatically enrolthem in the pension plan with the defaultsavings rate of 7% and the default assetallocation to a life cycle fund withrelatively low fees. I know the latterworks. It still remains to be shown that theeducation works.

RV: What about the psychology ofinvestment? How does that fit into youranalysis?

DL: Many of the investment choices thatthe typical investor makes are self-defeating, whether it’s return-chasing,avoiding risks because of loss aversion orfailing to recognise the importance of feesin reducing returns. Again and again,economists are documenting that peoplemake financial mistakes. We have toidentify those mistakes, understand theirsources and build institutions that helppeople avoid them.

So that’s where the investment sidecomes in. It’s easy to make bad choices infinancial markets: choosing the wrongmutual funds, choosing the wrong assetclasses, failing to diversify, failing to hold aworld portfolio and instead just holding adomestic portfolio. The typical investorneeds help, perhaps through educationand perhaps through defaults, to improvetheir balance sheet.

Another issue is employer stock. In theUnited States, investors in defined

contribution pension plans often hold analarmingly high fraction of their wealth inthe stock of their employer, which is, ofcourse, putting all their eggs in the samebasket. You don’t want your job to be onthe line and your retirement account tocollapse at the same time, which is whathappened at Enron and many other firmsin the early 2000s.

We’ve got a lot to do in measuringand then understanding the psychologicalunderpinnings of this kind of investmentmistake, which will lead us to build betterinstitutions that help people avoid thesemistakes.

RV: So in terms of these institutions you’relooking to design, what kind of impact areyour research findings and those of yourcolleagues in this field starting to have onpolicy-makers? Are these ideas starting tofeed into practice?

DL: The work that my collaborators and Iare doing and the work of many othereconomists in this field, have had an effecton what’s going on in financial marketsand on the regulators both in the UnitedStates and around the world.

In 2006, the United States passed thePension Protection Act (PPA), whichbasically made defaults the official policyof the US government. The PPA usesdefaults to help people save in theirretirement accounts. The Department ofLabor recently produced a set ofregulations that establish what is andwhat is not an appropriate investment forthese default funds. The UK is about toinitiate its own major policy programmeusing defaults. And we’re seeing defaults show up in many countriesaround the world.

So policy-makers are embracing theidea that a) investors sometimes makemistakes and b) institutions that aren’tthat complex or that controversial canavoid a lot of these problems. And I’mhappy to say that I think the pensionsworld of 2010 is going to look a lotdifferent and a lot more practical anduseful to unsophisticated investors thanthe pensions world of 2000.

David Laibson’s Lionel Robbins lectures were

delivered on 19, 20 and 21 November 2007 in

the Old Theatre at LSE. Recordings of the

lectures and slides from the lectures are

available here: http://cep.lse.ac.uk/interviews/

David Laibson is professor of economics

at Harvard University. More on

his research can be found here:

http://www.economics.harvard.edu/

faculty/laibson

The pensionsworld of 2010 isgoing to look a lot more useful tounsophisticatedinvestors than the pensionsworld of 2000


18

Wage inequality hasincreased substantially inboth the United Statesand the UK in the last

30 years. This fact, which is now firmlyestablished, has fuelled debate aboutappropriate policy responses, includingmore progressive taxation, education andtraining, tougher corporate governance tocontrol the remuneration of topmanagement, strengthening unions andminimum wages, curtailing trade andbeefing up social security. Some critics ofpro-market reforms claim that risinginequality is evidence that liberalisationhelps the rich at the expense of theaverage worker. For others, risinginequality is used to argue forprogrammes to upgrade skills.

But a proper policy response requiresan understanding of what has causedwage inequality. Much research shows(see, for example, the summary in Machin

and Van Reenen, 2007) that an importantpart of the rise in inequality comes fromincreases in inequality among similarworkers – what economists call ‘within-group’ inequality. This means that workerswith a similar age, gender, skill andindustry group tend to be paid increasinglydissimilar wages.

Within-group inequality may be drivenby the same factors explaining the broaderdistribution of wages between workers.For example, if technology increases thepayoff to a university education it is alsomore likely to increase the wages of moreable graduates more than those of lessable graduates. But the increase in within-group inequality may be driven by otherfactors – such as declining unionmembership and the minimum wage(Goldin and Katz, 2007).

Technology-based theories of within-group wage inequality share a commonprediction: since workers’ pay is linked to

their productivity, increases in wageinequality should be accompanied byincreases in ‘productivity dispersion’. Inother words, the productivity distributionfor firms should have widened.

Our new study looks first atindividuals’ earnings and firms’ averagewages and shows that the increase inwage dispersion between people is mainlydue to an increase in wage inequalitybetween firms. In other words, theinternal structure of wages within firmshas not widened as much as thedifference in average wages between firms(in the same industries). This is animportant finding when looking for thecauses of wage inequality. It suggests thatlittle of the growth in inequality is theresult of changes in the way firms treattheir own workers.

Hundreds of papers have documenteda close correlation between firms’ averagewages and firms’ average productivity

Many blame globalisation for growing wageinequality in the UK. But according to research by Giulia Faggio, Kjell Salvanes andJohn Van Reenen, the rise in inequality isbetter explained by increasing dispersion in the productivity of firms related to their use ofnew technology.

Inequality ofindividual wagesand the dispersion of firm productivity

beginning with the pioneering work of ex-Monetary Policy Committeemembers Sushil Wadhwani and SteveNickell in the 1980s at the CEP (see, forexample, the summary in Layard et al,2005). We also find that this relationshippersists and has grown stronger,suggesting that firms’ characteristics areimportant in understanding the evolutionof wage inequality.

To test the link between productivityand inequality, our research looks atchanges in firm productivity, mainlymeasured by value added per worker, overtime. We focus on the differencesbetween high- and low-productivity firmsamong representative samples ofhundreds of thousands of British firms.

Previous research has studied themanufacturing sector (Dunne et al, 2004;Haskel and Martin, 2002), but we analyseboth the manufacturing and servicesectors of the UK economy since the twosectors have experienced very differenttrends. In particular, the manufacturingsector has been in rapid decline, whichmeans that the least productive firms are disappearing from view, compressingthe observed distribution. Consistent with this fact, productivity inequality hasrisen much faster in services than in the manufacturing sector since the early 1980s.

It is well-known and entirelyunderstandable that different firms have


19

Figure 1:

Productivity dispersion in the UK economy, 1984-2001

Much of the rise in wageinequality is driven by

increasing differences inwages among firms in the

same industry

0.8

1

1.2

1.4

1.6

1.8

1990 1992 1994 1996 1998 2000198819861984

Bottom tenth

Top tenth

Median

Pro

du

ctiv

ity

(198

4=1)

Year

Note: Productivity is defined as the log of value added per worker. Values are indexed to be

1 in 1984, so the lines show the growth in productivity at different parts of the distribution.

The upper line is at the 90th percentile, the middle line is the median and the bottom line is at

the 10th percentile.


20

comparisons are useful because when‘general-purpose technologies’ such as ICTbecome ubiquitous, the impact on firms isfelt worldwide. Cross-country comparisonsmake it possible to distinguish generaleffects from changes specific to the UK.

In terms of policy, our findings suggestthat the causes of rising inequality areprimarily structural and related to newtechnology rather than to trade orinstitutions. Thus, contrary to what somehave argued, greater trade protectionismor the re-energising of unions may dorelatively little to reverse the increase in inequality.

A better strategy would be toconcentrate on raising the skill andeducation levels of the workforce,particularly of those at the bottom of the ability distribution. This means not just improving the quantity ofschooling for disadvantaged groups, but increasing the quality of school andpre-school education.

different productivity levels. Some firmsare better managed and use bettertechnology than other firms. But anoriginal finding of our work is that thisproductivity gap between firms haswidened. We look at two groups of firmsfrom opposite ends of the productivitydistribution: one at the top (90thpercentile) and one at the bottom (10th percentile).

So productivity inequality rosesubstantially over the period (see Figure 1).Between 1984 and 2001, productivity inour sample rose by about 49% (40 logpoints) for the median firm. But for firmsnear the bottom (the 10th percentile)there was a rise in productivity of only22% (20 log points) whereas for firmsnear the top (90th percentile) productivityhas risen by about 82% (60 log points).

Our research shows that changes inlabour productivity are mainly driven bychanges in ‘total factor productivity’,rather than changes in the amount ofbuildings or old capital equipment used.These results suggest that technologicaldifferences among firms may be theunderlying cause of increasing dispersion.Using data on the use of information andcommunication technologies (ICT), weshow that productivity dispersion rosemost strongly in sectors where ICT usegrew most rapidly and intensively. The rise in productivity dispersion did not seem closely related to imports or totrade unions.

Finally, we also document a rise inproductivity dispersion in France andNorway in the 1990s, although the risewas smaller than in the UK. Such

This article summarises ‘The Evolution

of Inequality in Productivity and Wages:

Panel Data Evidence’ by Giulia Faggio,

Kjell Salvanes and John Van Reenen,

CEP Discussion Paper No. 821

(http://cep.lse.ac.uk/pubs/download/

dp0821.pdf).

Giulia Faggio is an economist at Citigroup

Inc. in London and a CEP research associate.

Kjell Salvanes is at the Norwegian School of

Economics and Business Administration.

John Van Reenen is director of CEP.

Further reading

Timothy Dunne, Lucia Foster, John

Haltiwanger and Kenneth Troske (2004),

‘Wage and Productivity Dispersion in US

Manufacturing: The Role of Computer

Investments’, Journal of Labor Economics

22(2): 397-429.

Claudia Goldin and Lawrence Katz (2007),

‘Long-run Changes in the US Wage Structure:

Narrowing, Widening, Polarizing’, National

Bureau of Economic Research Working

Paper No. 13568.

Jonathan Haskel and Ralf Martin (2002),

‘The UK Manufacturing Productivity Spread’,

Ceriba Working Paper

(http://www.ceriba.org.uk).

Richard Layard, Stephen Nickell and Richard

Jackman (2005), Unemployment:

Macroeconomic Performance and the Labour

Market, Oxford University Press.

Stephen Machin and John Van Reenen (2007),

‘Changes in Wage Inequality’, CEP Special

Paper No. 18, forthcoming in The New

Palgrave Dictionary of Economics

(http://cep.lse.ac.uk/pubs/download/special/

cepsp18.pdf).

The increase inthe productivitydispersion of firmsis mainly inservices, especiallyin sectors whereICT use grew most rapidly and intensively

A major source ofrising inequality is

greater dispersion infirms’ productivity

related to their useof new technology


21

During the past ten years therehas been a significant increasein the number of individualscoming to live and work in

the UK. By 2007, 12.5% of the workingage population had been born overseas, up from 8.5% at the end ofthe last recession in 1993 and 7% in themid-1970s.

The addition to the UK labour forceover this period caused by the rise in thenumber of working age immigrants from2.3 to 4.2 million is about the same asthat stemming from the increase in thenative-born working age populationcaused by the baby boom generationreaching adulthood. The relative rise in immigrant numbers in recent years has been greater among those with higher levels of educational attainment(See Figure 1).

These trends have stimulated a heateddebate about the effects of immigration, adebate that has taken on renewed vigourfollowing the wave of immigration fromthe countries that have recently joined theEuropean Union and following evidence ofa loosening of the UK labour market.

Yet, despite immigration being at theforefront of the political debate, research-based evidence on its effects on thelabour market is far from conclusive. Thelargest body of evidence comes from the

United States, where researchers havereached different conclusions. In a numberof papers, David Card at Berkeley findslittle discernible impact of immigration onthe wages of native-born workers, whileGeorge Borjas at Harvard argues thatimmigration has a pronounced effect onthe native wage structure.

Evidence for the UK is scant and weshould not automatically assume that theimpact of immigration will be similar to

that found in the United States. The mostconvincing study – by Christian Dustmannand colleagues – finds that immigrationhas no large discernible effect on the levelof native wages.

At first glance, this finding is puzzlingsince simple economic reasoning andperhaps popular belief suggest that alarge increase in labour supply – such asthe one brought about by immigration –should lower the wages and/or

What are the effects of increased immigrationon the wages and employment of people whoare already here, whether they are native-born or earlier immigrants? Research byMarco Manacorda, Alan Manning andJonathan Wadsworth investigates.

The labour market effects

of immigration

Figure 1:

Immigrant shares in male population of working age

0%

5%

10%

15%

20%

25%

200520032001199919971995199319911989198719851983198119791977

Per

cen

tage

of

imm

igra

nts

in

th

e m

ale

wor

kin

g ag

e p

opu

lati

on

Year

� Graduates

� All men

� School leavers


22

The main labour market impact ofincreased immigration to the UK is feltby immigrants who are already here

Table 1:

Male immigrants and native-born workers in Britain(five year averages)

1975 1985 1995 2005

Percentage share of university graduates among:

Native-born 6.0% 12.2% 15.4% 21.6%

Immigrants 9.9% 22.7% 32.9% 40.4%

Percentage wage gap between university graduates and school leavers among:

Native-born 12.3% 12.2% 16.4% 24.4%

Immigrants 25.7% 24.3% 29.6% 29.9%

employment of workers in the recipientcountry. But this conclusion is based onthe often unspoken assumption thatnatives and migrants are highlysubstitutable in production – that anemployer can easily replace a native-bornworker with an equally skilled migrantworker. This need not to be the case andit is therefore a matter for empiricalverification.

A simple test of the degree ofsubstitutability between these differentproduction inputs is the responsiveness ofthe wage differential between native andmigrant workers with any given level ofskill to labour force changes broughtabout by migration. The intuition behindthis test is that if natives and migrants areperfectly substitutable in production, wewould expect their wages to respondsimilarly to changes in labour supply, be it

from natives or migrants, leaving thedifferential unaffected,

Conversely, if we find that the native-migrant wage differential is sensitive tothe share of migrants in the population,and in particular if this differentialincreases as migration rises, this will beevidence of imperfect substitutability,potentially explaining why natives appearto suffer few losses from migration.

To do this, our study focuses on men’slabour market outcomes using micro-dataspanning the period from the late 1970sto the mid-2000s. Because it is wellknown that workers with different labourmarket experience and skills are imperfectsubstitutes in production, we decomposethe labour force into different groupsdefined by age and education in additionto migration status. We estimate thetrends in immigration along these

dimensions as well as the correlationbetween these trends and the wages andemployment of natives and previousmigrants.

Measuring migrants’ skills is not aneasy task with UK data since the definitionof qualifications varies between nativesand migrants, with large numbers ofmigrants being classified in official surveysas holding unspecified ‘otherqualifications’. To cut through thisproblem, we measure completededucation in terms of the age at whichthe individual left full-time education.According to this metric, over the threedecades of analysis, migrants have beenon average more skilled than natives andincreasingly so.

This is shown in Table 1. The relativesupply of skilled to less skilled labour hasgrown more among immigrants. At the


23

same time, the returns to education haverisen faster for natives. A way torationalise this is that the returns toeducation among each of these groupsrespond to their own relative supply,implying imperfect substitutability.

Our estimates indicate that a 10% risein the share of immigrants in the UKpopulation is associated with an increasein the native-migrant wage differential ofaround 1.9%. We take this as evidencethat migrants and natives are imperfectsubstitutes in production.

This finding is highly robust andsquares well with our other finding thatthe extent of occupational segregationbetween migrants and natives in the UKlabour market is sizeable. We show thatbetween 20% and 34% of immigrantswould have to change jobs to match theoccupational distribution of employmentamong natives.

Interestingly, we find that migrationparticularly affects wage differentialsbetween natives and recent migrants aswell as those who migrated in adulthood.It appears that newly arrived migrants bear

disproportionately the cost of newmigration inflows. This cost manifests itselfin terms of lower wages rather thanemployment losses.

We use these estimates to determinethe effect of increased immigration on thelevel of native wages. Our calculationsshow that this effect is negligible. Thisresult depends on the imperfectsubstitutability between migrants andnatives coupled with the fact thatmigrants still account for a relatively lowshare of the workforce.

Our conclusion does not imply thatimmigration to the UK has no costs, butrather that the effects on groups otherthan immigrants themselves has been, onaverage, small.

As a caveat to our conclusion, recallthat most of the changes in the labour

force brought about by immigration overthe three decades of our analysis are dueto the inflow of skilled migrants. Becauseof this, our study is unable to shed muchlight on what appears to have been arecent increase in unskilled migration. Thisinflow might have rather different effectson the wage and employment structure ofnatives than the one estimated in ourresearch.

Indeed our analysis shows thatunskilled natives are relatively moresubstitutable with respect to migrantsthan skilled natives, although differencesbetween the two groups are notstatistically significant. More data andlonger time series are needed to estimatethis effect precisely.

Further reading

Brian Bell (1997), ‘The Performance of

Immigrants in the United Kingdom: Evidence

from the GHS’, Economic Journal 107: 333-44.

George Borjas (2003), ‘The Labor Demand

Curve is Downward Sloping: Re-examining

the Impact of Immigration on the Labor

Market’, Quarterly Journal of Economics

118: 1335-74.

George Borjas, Richard Freeman and

Lawrence Katz (1996), ‘Searching for the

Effect of Immigration on the Labor Market’,

American Economic Review 86(2): 246-51.

David Card (1990), ‘The Impact of the

Mariel Boatlift on the Miami Labor

Market’, Industrial and Labor Relations

Review 43: 245-57.

David Card (2001), ‘Immigrant Inflows, Native

Outflows, and the Local Labor Market

Impacts of Higher Immigration’, Journal

of Labor Economics 19(1): 22-63.

David Card (2005), ‘Is the New Immigration

Really So Bad?’, Economic Journal 115:

F00-323.

David Card and Thomas Lemieux (2001), ‘Can

Falling Supply Explain the Rising Return to

College for Younger Men? A Cohort-based

Analysis’, Quarterly Journal of Economics

116(2): 705-46.

Christian Dustmann, Francesca Fabbri and

Ian Preston (2005), ‘The Impact of

Immigration on the UK Labour Market’,

Economic Journal 115: F324-41.

Gianmarco Ottaviano and Giovanni Peri

(2006), ‘Rethinking the Effects of Immigration

on Wages’, National Bureau of Economic

Research Working Paper No. 12497.

This article summarises ‘The Impact of

Immigration on the Structure of Male Wages:

Theory and Evidence from Britain’ by Marco

Manacorda, Alan Manning and Jonathan

Wadsworth, CEP Discussion Paper No. 754

(http://cep.lse.ac.uk/pubs/download/

dp0754.pdf).

Marco Manacorda is reader in economics at

Queen Mary, University of London and a

research associate in CEP’s labour markets

programme. Alan Manning is professor of

economics at LSE and director of CEP’s

labour markets programme. Jonathan

Wadsworth is at Royal Holloway, University

of London and a senior research fellow in

CEP’s labour markets programme.

Newly arrived migrantsbear disproportionately

the cost of increasedmigration in terms of

lower wages

generation, you need the lifetime incomeof parents and children, a panel of at least40 years. To measure mobility over twogenerations, you need lifetime incomedata for children, parents andgrandparents – a minimum of a 70-yearpanel.

Even when the data are available, it isalmost impossible to make comparisonsacross countries or over time. So we knowvery little about intergenerational mobility– whether it has risen or fallen over time;whether it is larger in the United States,the UK or continental Europe; or whetherit is more prevalent in growing or stagnantsocieties, in richer or poorer societies or insocieties where inequality is high or low.

Our research offers a new way ofmeasuring intergenerational mobility that

escapes from the slavery of panel data.Instead, our method measures theinformational content of surnames: themore information a surname reveals aboutthe economic welfare of its bearer, themore that inheritance determines people’seconomic outcomes. In other words, themore a surname tells us about thewellbeing of an individual, the lessintergenerational mobility there is in that society.

The data requirements of this methodare infinitely less demanding than thetraditional method for measuring mobility,as we use census data (a cross-section ofsurnames as well as measures of incomeand education), which are collected inmost countries. And by studying theinformational content of surnames, it is


24

Almost all children get theirsurname from theirparents. While it isunlikely that a specificsurname has much effect

on its bearer’s wellbeing, surnames areinherited together with other things thatactually do affect a child’s future life –such as genes, wealth, beauty andeducation. Surnames can thereforeprovide us with information about thewellbeing of individuals, not because theymatter in themselves but because theytravel across generations with things thatdo matter.

In this way, surnames offer a potentialsource of data for tracking the importanceof family background for outcomes in laterlife and the degree to which people’seconomic and social status changesbetween generations. Surname data canshow us how this ‘intergenerationalmobility’ compares across countries andhow it evolves over time.

It is notoriously difficult to measurethe probability that the child of poorparents will become rich, and vice versa.This is because the traditional procedure(comparing the lifetime income of parentswith that of their children) demands verylong panels of data, which are hard toobtain. To measure mobility within one

It is very difficult to measure changes ineconomic and social mobility across generationsand to make comparisons across societies.Research by Maia Güell, José V. RodríguezMora and Chris Telmer finds that surnamescan provide valuable insights into theimportance of family background for people’soutcomes in later life.

What’s in a name?Information on intergenerational mobility

SáGil de B

mating is that surnames are inherited onlyfrom the father, but background dependson both parents. An increase in assortativemating – so that people are more likely tomarry someone from a similar familybackground to themselves – results in adecrease in mobility and an increase in theinformational content of surnames.

Surnames are not only informativeabout the family to which an individualbelongs, but also about his or herethnicity. Not controlling for ethnicitywould bias the results, as theinformational content of surnames mightreflect ethnicity in addition to specificfamily background. Fortunately, we canuse the surnames themselves to controlfor ethnicity.

The second stage of our analysis is totest the methodology against some data.We use data from Catalonia, where theSpanish naming convention comes inhandy: individuals have two surnames (thefirst from the father and the second fromthe mother); they pass only the first ofthese to their children (the standardwestern tradition of inheriting the paternal

easier to measure changes in mobility overtime since every census contains all therelevant information for many generations.We can assess how mobility evolves overtime by comparing the informationalcontent of surnames among older andyounger cohorts.

In the first stage of our analysis, wedevelop a methodology for measuring thedegree of intergenerational mobility sinceit is not immediately obvious whysurnames should be informative or thatthe amount of information they containreveals the degree of intergenerationalmobility. Imagine that there are very fewsurnames, each shared by manyindividuals who are not necessarily related.In this case, surnames would not beinformative, since they would not indicatefamily linkages. For example, twoindividuals called Smith are very unlikely tobe related, which makes any similaritybetween their incomes a product ofchance and unrelated to the degree ofinheritance.

Fortunately, the distribution ofsurnames is extremely skewed. In otherwords, there are some very commonsurnames (their bearers unlikely to berelated), but the huge majority ofsurnames are quite infrequent, accountingfor a very large fraction of the population.Two bearers of an uncommon surnameare likely to be related. Uncommonsurnames are central to our method,because income similarities between

individuals linked by an infrequentsurname indicate the significance ofbackground for economic outcomes.

The reason for this skewness is thatthe process of generation and inheritanceof surnames is akin to the genetic processthat determines the distribution of DNA.Surnames (lineages) die when the lastmale bearer of a surname dies without amale descendant (as surnames areinherited via the male line in most westernsocieties). Lineages are born whenever aperson changes his surname or animmigrant arrives carrying a distinctsurname.

Our research develops a ‘genetic’model of the joint distribution ofsurnames and income. It shows that bylooking at the informational content ofsurnames, we can infer the importance ofbackground. Extensions of the modelallow for the possibility of ‘assortativemating’ (in which ‘like marry like’) and theintroduction of ethnic differences inincome (due to discrimination or any otherreasons).

The rationale for including assortative


25

Vázquez MontalbánGaudí Cornet

The more a surname tells us aboutthe wellbeing of an individual, theless intergenerational mobilitythere is in that society

ánchezBiedma AlbaDalí Domènech


surname); and women never change theirsurname when they get married.

Thus it is possible to identify familybackground through the first surname andethnicity through the second surname.Our measure of the informational contentof surnames captures how much firstsurnames explain the variation inindividuals’ outcomes beyond ethnicityand individual observable characteristics(gender, place of birth, etc.). We find thatthis is sizeable and, consistent with themodel, larger when we focus oninfrequent surnames, as these provide acloser approximation to family.

The combination of the two surnamesalso allows us to identify siblings (as twoindividuals who share two infrequentsurnames in the same order are almostcertainly siblings) and to determine thedegree of assortative mating among theparents (how much the surname of thefather helps to explain the surname of themother).

Our analysis of the data shows that inCatalonia surnames are informative in away that is perfectly coherent with thepredictions of the model. Surnamescontain information about both ethnicity(individuals with Catalan surnames dobetter) and family background.Furthermore, the amount of informationthat surnames contain has increasedsteadily over time, indicating a decrease in mobility.

One way to validate these results (andthe methodology) is to identify siblings byusing the two surnames. Doing this revealsthat the correlation between siblings hasincreased over time, which also indicatesdecreased mobility. Moreover, this decrease

in mobility is explained by an increase inassortative mating that predates theincrease in the informational content ofsurnames by one generation.

There are two readings of our results.The first is literal: despite dramaticincreases in the provision of publiceducation in modern Spain, the degree ofintergenerational mobility has decreased.This fall in mobility is a consequence ofassortative mating.

Second, surnames enable us to look atthe relevance of family background. Thisview is strongly reinforced by the fact thatthe results are identical whether using onlyone surname (for countries other thanSpain) or using two and concentrating onsiblings. Thus, the methodology is stronglysupported by the results. We can learn alot by looking at how much surnames say.

This article summarises ‘Intergenerational

Mobility and the Informative Content of

Surnames’ by Maia Güell, José V. Rodríguez

Mora and Chris Telmer, CEP Discussion

Paper No. 810 (http://cep.lse.ac.uk/pubs/

download/dp0810.pdf).

Maia Güell is professor of economics at the

University of Edinburgh and a research

associate in CEP’s labour markets

programme. José V. Rodríguez Mora is

professor of economics at the University of

Edinburgh. Chris Telmer is an associate

professor at Carnegie-Mellon University.

In Spain, theamount ofinformation that surnamescontain hasincreasedsteadily overtime, indicating a decrease in mobility

Montalbán26

Sánchez VicarioMiró Ferrà

Gasol Sáez


27

Do computers help police fight crime?

The adoption of information technology by policedepartments in the United States is a relatively recentphenomenon. Before 1987, less than 2% of the 2,200 USpolice departments with fewer than 100 employees usedcomputers. And as recently as 2003, only 40% of policedepartments had mobile computer terminals.

So did police departments that adopted informationtechnology earliest see changes in organisational practicesand drops in crime relative to those that adopted later?Our research addresses this question by comparing thecrime-fighting effectiveness and organisational practices ofpolice departments that were early adopters to those thatadopted the technology less rapidly (factoring outdemographic and other changes, as well as time trendsand department-specific effects).

We use two data sources: first, the triennial LawEnforcement Management and Administrative Statistics(LEMAS) surveys, which document US police departments’use of technology between 1987 and 2003; and second,the FBI’s uniform crime reports, which collect crime andarrest statistics for the entire country. We also usedemographic data from the US censuses to control forpopulation characteristics.

We find that the departments that adopted informationtechnology grew in size by around 8%, increasing thenumber of specialised jobs and share of support staff

needed as well as the use of written (as opposed to oral)instructions within the department. These departmentsalso increased the educational standards for new officersby raising both the college requirements and their hoursof training by around 200 hours. The increase in traininghours appears to be mostly related to the use of mobileterminals.

Thus, the greater use of computers by police departmentsimplies a larger, increasingly complex and more highlyeducated police force. This outcome is not too surprising.Where our results are puzzling is when we turn to theproductivity gains related to this technological revolution.

We measure productivity in two ways: first, crimeclearance rates (crimes solved relative to crimescommitted); and second, offence rates (crimes per head ofpopulation). Surprisingly, we find that the use ofinformation technology neither increased crime clearancerates nor decreased offence rates. In fact, if anything, theuse of computers is associated with an increase inproperty crimes. This is a puzzling set of results,particularly given our finding that the use of technologyraises the quality of police staff.

So is there some causal link between these results? Ourempirical strategy suggests that the answer is yes, giventhat the results control for geographical and department-specific characteristics as well as time trends. Of course, as

Does the increasing computerisation of police work result in more effective crime-fighting? Paul Heaton and LuisGaricano study the impact of information technology on theorganisation and productivity of US police departments.

Computersincrease theeffectiveness ofpolice work – but only if it isreorganised to take advantage oftheir presence

in brief...

n


28

always in social science, it is hard to prove causalityunequivocally. To increase our confidence in the findings,we study several alternative hypotheses:

� First, it is not simply the case that the more troubledpolice departments are the ones that adoptedinformation technology faster (which would yield aspurious link between high investment in informationtechnology and small changes in clearance rates).

� Second, it is not the case that future levels ofinformation technology use are connected to changesin organisation and productivity, as would be expectedif information technology were the result, rather thanthe cause, of the changes studied.

� Third, the results for productivity are unrelated to citysize or population growth or decline: they hold acrossthe board for growing and shrinking cities and policedepartments.

� Fourth, controlling for the size of a department’sbudget does not alter the result, suggesting that it isnot the outcome of growing or shrinking budgets.

� Finally, we use the adoption of information technologyby state as a plausible ‘exogenous’ source of variationfor the rate of adoption by police departments, andshow that the results are still unchanged.

All of this suggests that the conclusion is causal and raisesan important question about whether technology reallydoes help police departments and, if so, how.

One key advantage of computers is improved record-keeping. Our research shows that introducing computersfor record-keeping did increase the amount of recorded‘minor crimes’ such as larceny, accounting for some, but not all, of the measured ineffectiveness ofcomputers. As would be expected, improved record-keeping did not affect the number of homicides andmajor crimes recorded. This supplies an explanation forpart of the puzzle.

We develop the hypothesis that just as in the privatesector, it is not only the increased use of computers that isrequired for better outcomes, but also changes in theorganisation of police work. In the context of policing, theorganisation of work around computers might change bythe use of geographical patterns to allocate resources andto measure the crime-fighting progress of individualofficers.

The best-known example of this use of informationtechnology is the New York City Police Department’sCompStat system, associated with Bill Bratton’s tenure aspolice chief during the 1990s under then-Mayor RudyGiuliani. CompStat is an integrated information andmanagement system for policing, which includes several

key elements that the LEMAS data capture: data-drivenproblem identification, internal accountability,geographical organisation of deployment, and measurablegoals.

Our research finds that police departments that adoptedcomputers together with CompStat-style organisationalpractices did experience reduced property and violentcrime and significantly increased crime clearance rates (byaround 8%). Thus, we conclude that computers doincrease the effectiveness of police work, but only if policework is substantially reorganised to take advantage oftheir presence.

Computers plus an integratedinformation and management

system for policing increasecrime clearance rates

This article summarises ‘Information Technology, Organisation

and Productivity in the Public Sector: Evidence from Police

Departments’ by Luis Garicano and Paul Heaton, CEP

Discussion Paper No. 826 (http://cep.lse.ac.uk/pubs/

download/dp0826.pdf).

Luis Garicano is director of research in LSE’s department of

management, professor of economics and strategy at the

University of Chicago’s graduate school of business, and a

research associate in CEP’s research programme on

productivity and innovation. Paul Heaton is an associate

economist at the RAND Corporation.

New from the CEP

For further information on CEP publications and events

please contact:The Publications Unit


Houghton Street, London WC2A 2AE

Telephone +44 (0) 207 955 7673

Fax +44 (0) 207 955 7595

Email [email protected]

We accept Visa, Access and Mastercard

for payment by post, fax and phone.

Overseas rates on request.

http://cep.lse.ac.uk/pubs/books.asp

CEP Discussion Papers are now available as electronic copiesfree to download from the Centre’s website:

http://cep.lse.ac.uk/pubs/dp.asp?prog=CEPDP

WHAT ARE THE LONG-TERM EFFECTS OFUI? EVIDENCE FROM THE UK JSA REFORMBarbara PetrongoloCEP Discussion Paper No. 841December 2007

ROBUSTLY OPTIMAL MONETARY POLICYKevin SheedyCEP Discussion Paper No. 840November 2007

THE UNEMPLOYMENT VOLATILITY PUZZLE:IS WAGE STICKINESS THE ANSWER?Christopher PissaridesCEP Discussion Paper No. 839November 2007

INFLATION PERSISTENCE WHEN PRICESTICKINESS DIFFERS BETWEEN INDUSTRIESKevin SheedyCEP Discussion Paper No. 838November 2007

INTRINSIC INFLATION PERSISTENCEKevin SheedyCEP Discussion Paper No. 837November 2007

LAGS AND LEADS IN LIFE SATISFACTION:A TEST OF THE BASELINE HYPOTHESISAndrew Clark, Ed Diener, YannisGeorgellis and Richard LucasCEP Discussion Paper No. 836November 2007

CREATIVE DESTRUCTION WITH ON-THE-JOB SEARCHJean-Baptiste MichauCEP Discussion Paper No. 835November 2007

SCYLLA AND CHARYBDIS: THE EUROPEAN ECONOMY ANDPOLAND'S ADHERENCE TO GOLD, 1928-1936Nikolaus WolfCEP Discussion Paper No. 834November 2007

INNOVATION IN BUSINESS GROUPSSharon Belenzon and TomerBerkovitzCEP Discussion Paper No. 833November 2007

INCENTIVES IN COMPETITIVE SEARCH EQUILIBRIUMEspen Moen and Åsa RosénCEP Discussion Paper No. 832November 2007

U.S. LABOR MARKET DYNAMICSREVISITEDEran YashivCEP Discussion Paper No. 831November 2007

GROWTH AND THE QUALITY OF FOREIGNDIRECT INVESTMENT: IS ALL FDI EQUAL?Laura Alfaro and Andrew CharltonCEP Discussion Paper No. 830November 2007

COST-BENEFIT ANALYSIS OFPSYCHOLOGICAL THERAPYRichard Layard, David Clark, MartinKnapp and Guy MayrazCEP Discussion Paper No. 829October 2007

MATERNAL SMOKING DURINGPREGNANCY AND EARLY CHILDOUTCOMESEmma TomineyCEP Discussion Paper No. 828October 2007

TESTING THE “WATERBED” EFFECT INMOBILE TELEPHONYChristos Genakos and Tommaso VallettiCEP Discussion Paper No. 827October 2007

INFORMATION TECHNOLOGY,ORGANIZATION, AND PRODUCTIVITY INTHE PUBLIC SECTOR: EVIDENCE FROMPOLICE DEPARTMENTSLuis Garicano and Paul HeatonCEP Discussion Paper No. 826October 2007

INTRA-INDUSTRY FOREIGN DIRECT INVESTMENTLaura Alfaro and Andrew CharltonCEP Discussion Paper No. 825October 2007

TECHNOLOGICAL DIVERSIFICATIONMiklos Koren and Silvana TenreyroCEP Discussion Paper No. 824October 2007

CentrePiece is the magazine of the Centre for

Economic Performance at the London School of

Economics. Articles in this issue reflect the

opinions of the authors, not of the Centre.

Requests for permission to reproduce the

articles should be sent to the Editor at the

address below.

Editorial and Subscriptions Office


London School of Economics

Houghton Street

London WC2A 2AE

© Centre for Economic Performance 2008

Volume 12 Issue 3

(ISSN 1362-3761) All rights reserved.

Centre Piece

The back issues

Documents

CentrePiece - Centre for Economic Performancecep.lse.ac.uk/pubs/download/CentrePiece_12_3.pdf · CentrePiece Winter 2007/08 1 page 2 School segregation and its consequences Stephen