centrepiece - welcome to lse research online - lse...

The Magazine of The Centre for Economic Performance Volume 13 Issue 2 Autumn 2008

ISSN 1362-3761

C e n t r e Pi e c eSUPERMARKETS AND THE BRITISH HIGH STREET:THE UNINTENDED CONSEQUENCESOF PLANNING REGULATIONS

Life chancesRising prices

University fees

Inequality and attitudes to redistributionSwing states and US trade policyCentral bank communications

Spend it like Beckham

All three of the main political parties inBritain are now putting the idea offairness at the heart of their policyagenda. The Conservatives propose ‘ablueprint for fairness’; the LiberalDemocrats want ‘to restore fairness toBritish families’; and the prime ministeraspires to create ‘a fair Britain for thenew age’.

What these words mean concretelyremains unclear, but there appears tobe a common concern with equality –of outcome and/or opportunity – anarea in which research by the Centrefor Economic Performance (CEP) hasmade many pioneering contributions.

Over the past two decades, work byresearchers at the Centre hasdocumented rising labour marketinequalities in Britain – and uncoveredtheir causes in large-scale technologicaland institutional change. And a seriesof CEP studies has revealed that socialmobility has fallen between the cohortof British children who grew up in the1960s and early 1970s and those who

grew up in the 1970s and 1980s – asocial science ‘fact’ that is nowembedded in public debate.

In this CentrePiece, CEP’s researchdirector Stephen Machin, who wasrecently appointed to the government’snew National Equality Panel, describesthe ‘big ideas’ that have emerged fromthe Centre’s longstanding programmeon wage inequality. Future issues willreview CEP research on and evaluationof key policy responses such aseducation and the minimum wage.

In a separate article with JoBlanden, Machin outlines the latestevidence on what has happened to thelife chances of children born into poorfamilies since 1970. This shows thatcontrary to the oft-repeated claims ofmany politicians and commentators,the downward trend has not continuedbut appears to have flattened out.

Two further articles touch on thetheme of inequality: one on howinflation hurts the poor more than therich; the other charting British people’s

diminishing enthusiasm forredistribution.

We also feature two more storiesthat illustrate the complex interactionsbetween politics and economics. As theUS presidential race nears the finish,one article shows that industries in theso-called ‘swing states’ are more likelyto enjoy trade protection.

And our cover story examines theimpact of planning regulationsdesigned to restrict the development of‘big-box’ supermarkets. This is an issueon which politicians clearly do takedifferent positions – and it turns outthat those most strongly opposed tobig boxes have set in motion theopposite effect from what theyintended, actually accelerating thedecline of smaller independent storeson the British High Street.

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

Cover image,

www.flickr.com/photos/pavlovapumpkin/

© Centre for Economic Performance 2008

Volume 13 Issue 2

(ISSN 1362-3761) All rights reserved.

Centre Piece

Editorial

CentrePiece Autumn 2008

1

page 2 Does planning regulation protect independent retailers?Raffaella Sadun reveals that planning regulations supposed to ‘save the traditionalBritish High Street’ have in fact done the opposite.

page 8 Big ideas: rising wage inequalityStephen Machin surveys CEP’s pathbreaking research findings on wage inequality.

page 14 A continuing downward trend in intergenerational mobility?Jo Blanden and Stephen Machin look at evidence on the life chances of children borninto poor families since 1970.

page 18 How to measure the rising cost of livingNicholas Oulton explains the problems in creating an accurate measure of inflation –and shows how inflation hurts the poor more than the rich.

page 21 Do swing states influence US trade policy?As the US election campaign nears its climax, Mirabelle Muûls and Dimitra Petropoulou find that industries in battleground states are more likely to be protected.

page 24 Words that work: comparing the effectiveness of central bank communicationsCarlo Rosa finds that Fed announcements move market interest rates significantly morethan those from the European Central Bank.

Contents in brief...page 6 Are the top universitiesworth paying for?Iftikhar Hussain, Sandra McNally andShqiponja Telhaj investigate whetherdegrees from the UK’s ‘better’ universitiesare worth more than others.

page 12 Spend it like BeckhamAndreas Georgiadis and Alan Manningwonder why voters have not forcedsuccessive governments to increase taxeson the rich.

page 27 The end of trade unionismas we know it? Alex Bryson and David Blanchflowerexplore whether the decline of unions in Britain really matters for employees and firms.

Conference reportspage 11 The emergence of China and India in the global economypage 17Intergenerational mobility

page 18How to measure the rising cost of living page 12


high street, the opposite effect of the one intended.

The new planning regulations –introduced in 1996 and reinforced in1999 – stipulated that retailers wanting toopen a store of more than 2,500 squaremetres had to pass a ‘sequential test’ anda 'test of need'. These tests demandedproof that large out-of-towndevelopments could not be created inalternative in-town or edge-of-town


2

Given that we claim tobe so attached to thetraditional BritishHigh Street, it seemsodd that until themid-1990s, fewer

and fewer of us spent much of our timeshopping there. Instead we weremigrating to the out-of-town behemoths:gigantic mega-stores run by, amongothers, Tesco, Sainsbury’s and Asda (now

owned by Wal-Mart, the world leader insuch superstores).

But over the course of the 1990s – inpart as a result of pressure from groupsconcerned about the plight of smallindependent retailers – the planningregulations changed, with the effect thatlarge retail chains significantly slowed theexpansion of ‘big boxes’. My researchshows that this actually may have harmedsmall independent retailers in the British

Does planning regulation protect independent retailers?

In 1996, new regulations made it much harder forUK supermarkets and other retailers to developnew out-of-town outlets – so-called ‘big boxes’.In part, these regulations were supposed to ‘savethe traditional British High Street’ by protectingsmall retailers. Research by Raffaella Sadunshows that they might have actually acceleratedthe decline of independent stores.

locations, and that these new retaildevelopments were ‘needed’ in the area.

The reforms also increased the role oflocal authorities in the implementationand interpretation of these planningguidelines. Significantly, this meant thatlocal politicians could select which largestores could open in their area.

Under the new planning system, thenumber of successful planningapplications in an area depended both onlocal demand conditions (whether firmswould want to open a new store there)and on local politics (whether politicianswould let them). Overall, these changesadded significant monetary and time coststo the application process. Unsurprisingly,the development of new big boxesdeclined sharply, as Figure 1 shows.

But the fall in the opening of bigboxes did not coincide with a reduction inthe total number of new stores, ratherwith a change in their size and location.In the years following the introduction ofthe reforms, the major UK retail chainsstarted to open more small stores on highstreets and in city centres. Griffith andHarmgart (2005) show that since the late1990s, the top four UK retail chainssubstantially increased the number ofsmall convenience stores opened in towncentres relative to investments in largestores in out-of-town locations.

Figure 2 shows how the median sizeof stores operated by the big supermarketchains fell between 1997/8 and 2002/3.Over the relatively short period of four


3

Figure 1:

Planning grants for large retail stores declined sharplyafter regulations introduced in 1996

Note: The figure reports the number of

major retail applications granted across

304 English local authorities between 1993

and 2003.

2.8

2.6

2.4

2.2

2

1995 200520001990Year

Figure 2:

After 1996, supermarket chains began opening smaller stores

2

1.5

1

0.5

0

0 100 200 300

2

1.5

1

0.5

0

0 100 200 300

Store size distribution (based on employment)

Store size distribution (based on employment)

2002/3

1997/8

Fra

ctio

nF

ract

ion

Ave

rage

nu

mb

er o

f ap

pli

cati

ons

gran

ted

by

loca

l au

thor

itie

s

Note: The figures show the changing average size of ‘non-specialised stores’ (mostly

supermarkets) operating nationally (that is, in all 11 UK regions) in 1997/8 and 2002/3, as

measured by employment in each store. The vertical lines mark the 10th, 50th and 90th

percentiles of the distribution.

Big boxes maybe good news for ‘mom-and-pop’ stores

years, the median size of a store belongingto a large supermarket chain fell from 75to 56 employees. This trend contrasts withthe retail chains’ development in othercountries. For example, over a comparabletime period, the average store size ofnational retail chains in the United Statesincreased from 142 to 152 employees(Haskel et al, 2008).

In my research, I investigate what arethe direct and indirect effects of the newplanning regulations on employmentamong independent retailers. In particular,I estimate the effect of the number ofplanning grants won by big boxes on theemployment growth of independentretailers in a local authority.

This is far from straightforward, as thesame factors can influence both thenumber of grants and the growth ofindependent retailers. For example, if anarea experiences rising incomes, this willlead to increasing demand for retailproducts, which will be spread across


4

Ranges

� Over 3.3

� 2-3.3

� 1.2-2

� 0-1.2

Figure 3:

Average number of planning grants for large retail storesacross English local authorities, 1993-2003

Conservativecouncils are much less likely to grant planningpermission fornew big boxesthan othercouncils

Note: The figure shows the

substantial variation in

planning grants across the UK.

Each local authority has a

different colour, corresponding

to the average number of

grants they gave over the

period 1993-2003, with darker

colours indicating that more

grants were given.


5

both big boxes (thus generating moreplanning applications) and smallindependent retailers.

To identify whether the number of bigbox grants affects the employmentgrowth of independent retailers, I exploitthe fact that there is a political, as well asan economic, dimension to the planningprocess. The two key assumptions at thebasis of my methodology are thatpoliticians’ preferences about planningregulations depend on their partyaffiliation, and that the electoral successof the different parties is independent offactors that might affect employmentamong independent retailers. Under thesecircumstances, it becomes possible to usethe political composition of the localauthorities to identify the effect of grantsof big boxes on the employment growthof independent retailers.

More than any other party in the UK,the Conservatives have traditionally beenassociated with Nimby-ism (‘not in mybackyard’) – a strong opposition towardsnew retail developments. Typically justifiedon environmental grounds, this alsoprobably reflects the political weight ofmiddle-class homeowners, who worryabout the effect on the value of theirproperties, and small retailers, who oftenfear competition from big boxes.

The influence of Conservativepoliticians on local planning activitiesaffects the number of applicationsgranted across the UK, leading toconsiderable variation in the number ofgrants across local authorities (Figure 3).This is the variation I use to identify theeffect of big boxes on the employmentgrowth of independent retailers.

I find that the number of successfulplanning grants for out-of-townsupermarkets is positively related to theemployment growth of small independentretailers. This implies that regulation thatprevents the development of more bigboxes may actually harm independentretailers and the people that they employ.

A possible interpretation of this resultis that when planning regulations preventthe entry of large supermarkets, retailchains move to a business model basedmore on smaller in-town stores. Thesesmaller format stores compete moredirectly with independent stores, andaccelerate their decline.

To provide more evidence of this, Iproceed in two steps. First, I show that

the growth of smaller in-town storesbelonging to retail chains is negativelyassociated with the number of big boxeswinning a planning grant. This suggeststhat the increasing movement of themajor UK chains towards smallconvenience stores can be directly linkedto the increasing planning hurdles facedby big boxes.

Second, I look separately at the effectof big boxes on the entry and exit rates ofretailers into and out of the market, andon the growth or contraction of retailersalready in the market. Interestingly, thereappears to be almost no effect from moregrants for big boxes on changes in thesize of existing retailers, nor on thenumber of new independent retailers thatenter the market.

The entire effect from more big boxeswinning planning permission appears tocome from fewer independent retailersleaving the market. This suggests thathaving more big boxes has the effect ofreducing the total amount of competitionthat independent retailers face.

My estimates suggest that the sharpdecline in new big boxes can account forabout 15% of the decline in employmentamong independent retailers between1998 and 2004. It should be stressed thatit is not clear whether all independentretailers benefit equally from more bigboxes, and it is too early to drawconclusions about any long-run effects.But so far, at least, it seems that bigboxes may actually be good news for‘mom-and-pop’ stores – at leastcompared to the likely alternative of moreTesco Metro stores.

This article summarises ‘Does Planning

Regulation Protect Independent Retailers?’

by Raffaella Sadun, CEP Discussion Paper

No. 888 (http://cep.lse.ac.uk/pubs/download/

dp0888.pdf).

Raffaella Sadun is a research officer in

CEP’s productivity and innovation

programme.

Further reading

Rachel Griffith and Heike Harmgart (2005)

‘Retail Productivity’, The International

Review of Retail, Distribution and Consumer

Research 15(3): 281-90.

Jonathan Haskel, Ron Jarmin, Kazu

Motohashi and Raffaella Sadun (2008)

‘A Three Country Comparison of the

Retail Sector Using Micro Data’,

CEP/CeRiBA mimeo.

The fewer bigboxes allowed by councils, themore small chainstores wereintroduced – and the moreindependentretailers suffered

More big boxesbeing givenplanningpermission isassociated withfewerindependentretailers beingforced out ofbusiness


6

Are the top universitiesworth paying for?

From 2006, new UK undergraduates – ‘freshers’ – wereasked to pay up to £3,000 in fees for each year of theirdegree. The tripling of tuition fees was politicallycontentious at the time. But for many backbenchmembers of parliament, the key sticking point was notthat universities were allowed to charge much more thanbefore, but that the £3,000 a year limit represented a capnot a fixed rate. This established the principle thatdifferent universities could charge different fees.

In the event, only a handful of universities charged feesbelow the £3,000 cap. But if the limit were to be raised,there could be more variation in the fees charged, withmore universities charging a lower fee than the maximumpermitted. Indeed, this might make sense. After all,universities’ costs vary: the more prestigious ones typically(but not always) pay higher salaries to their staff and maywant to recoup more of their costs through raising thefees to the higher levels permitted.

But would this variation in fees be paralleled by a variationin the benefits that students get from differentuniversities? For example, does going to a moreprestigious and costly institution pay off in terms of itsgraduates receiving a higher salary on entering the labourmarket?

We know that the average starting salary of students fromthe Russell Group of 20 elite UK universities is higher thanthat for students from all other UK universities. But thiscould be because the Russell Group universities takehigher calibre, more highly motivated students in the firstplace.

Can evidence from the United States help us here? ManyUS studies show that the relationship between universityquality and graduate wages is positive, even aftercontrolling for the high school grades (and thereforepossibly the ability and drive) of each student on entry.Black and Smith (2006) show that the positive impactmight even be understated, as the studies only use asingle measure of university quality – students’ averagehigh school test scores.

But we should be careful about assuming that this appliesin the UK. For one thing, similar studies find no effect

from the quality of US high schools on wages (as noted inHanushek, 2003). This might be because US high schoolsoperate in a less competitive market than US universities.Since UK universities operate in a more centralised, lesscompetitive system than US universities, there might belittle or no difference to a graduate’s earnings fromattending a prestigious university.

The principal difficulty with establishing whether universityquality affects graduate salaries is that high-qualityuniversities select (and are selected by) the best students.But if universities select students entirely on the basis ofthings that we can observe – such as their grades at GCSEand A-level – then we might be able to remove any bias inthe results of our research by controlling for themappropriately.

To make sure that we fully capture those traits that wedon’t observe – such as ambition and effort – we alsocontrol for background family factors such as parents’education.

Our study is based on four cohorts of students: those thatgraduated in 1985, 1990, 1995 and 1999. The 1995 and1999 cohorts were surveyed three and four years aftergraduation, respectively. These two cohorts are the mostcomparable since the surveys were deliberately designedto be similar.

The 1985 and 1990 cohorts were surveyed 11 and sixyears after leaving university, respectively. For both ofthese cohorts, we study their wages six years aftergraduation (the earlier cohort were asked to recall theirwages six years after they graduated). The differencesbetween the cohorts mean that we have to be carefulabout saying that the ‘returns’ to studying at differentuniversities have changed over time.

One difficulty for the research is measuring the quality ofan institution. A single measure is likely to be inaccurate,as quality can be demonstrated in different ways.

With the introduction of tuition fees for undergraduates, theUK government accepted the principle that since graduatesreceive much of the benefit of their degrees, they should beasked to bear some of the cost. But are degrees from ‘better’universities worth more than others? Iftikhar Hussain,Sandra McNally and Shqiponja Telhaj investigate.

in brief...

Going to an ‘elite university’ (in thetop quartile of quality) rather than a

university in the bottom quartileincreases graduate pay by 10-16%.


7

Consequently, we use a single composite measure thatcombines research quality, the faculty-student ratio,expenditure per student, the A-level scores of incomingstudents and the dropout rate.

The results of our analysis suggest that there is asignificant premium to attending a high-quality universityover an average university in terms of the wages thatgraduates can command in the labour market. This impliesthat even if two graduates have the same A-level gradesand family background and studied the same degreesubject, they will earn different wages if they went todifferent universities. The graduate from the moreprestigious university will, on average, earn more.

Specifically, an increase of one standard deviation inuniversity quality leads to a roughly 6% increase ingraduate wages. To put this another way, a student whogoes to a university that is in the second highest quartile(in terms of quality) can expect to earn 5-7% more than astudent with the same parental background and A-levelswho goes to a university in the bottom quartile of quality.

There is also evidence that students who go to the verybest universities can benefit even more: our resultssuggest that there may be increasing returns to quality. Astudent who attends a university in the top quartile ofquality may earn 10-16% more than a student whoattends a university in the bottom quartile, depending onthe measure of quality used.

Our results indicate that students who go to a topuniversity do benefit more from their degrees thanstudents who go to universities of lesser quality. As itturns out, our estimates for the ‘returns to quality’ aresimilar to those found by Black and Smith (2006) for theUnited States, as well as those found by Conlon andChevalier (2003) for the premium for attending a RussellGroup university in the UK.

As far as we can tell, the returns to going to a betterquality university have increased over time. This mayreflect the large increase in the number of universities in1992. But because our samples differ so much betweenthe first two cohorts and the second two, we cannot becertain about this.

Over a lifetime in the labour market, going to a universityof higher quality can make a big difference to totalincome. The average annual earnings of graduates fromthe 1999 cohort were £22,828 in 2003 (four years aftergraduation). If we assume that the returns to quality are6% of this amount and that this stays constant inabsolute terms over the total time in the labour market,this is worth in total £35,207 (assuming 25 years in thelabour market and using a discount rate of 3.5%). Since itis likely that the premium to going to a higher qualityinstitution increases over time, this may be anunderestimate.

Such evidence suggests that there is some justice inrequiring graduates to contribute to the cost of theiruniversity education, and in allowing different universitiesto charge different fees. That said, we should be clear thatalthough there are increased average returns to graduatesof higher quality institutions, these are still small incomparison with the overall value of higher education.

Blundell et al (2005) find that the average returns tohigher education are 48% (of earnings) in comparisonwith leaving school at age 16 with no qualifications. If wetranslate this into lifetime earnings in the same (veryrough) way, this amounts to a £281,594 differencebetween a graduate and a school-leaver. So we can seethat encouraging more people to go into higher educationshould still be the major policy priority.

Further reading

Dan Black and Jeffrey Smith (2006) ‘Estimating Returns to

College Quality with Multiple Proxies for Quality’, Journal of

Labor Economics 24(3): 701-28.

Richard Blundell, Lorraine Dearden and Barbara Sianesi

(2005) ‘Measuring the Returns to Education’, in What’s the

Good of Education? The Economics of Education in the UK

edited by Stephen Machin and Anna Vignoles, Princeton

University Press.

Gavan Conlon and Arnaud Chevalier (2003) ‘Does it Pay to

Attend a Prestigious University?’, Centre for the Economics of

Education Discussion Paper No. 33

(http://cee.lse.ac.uk/cee%20dps/ceedp33.pdf).

Eric Hanushek (2003) ‘The Failure of Input-based Schooling

Policies’, Economic Journal 113: F64-98.

This article summarises ‘University Quality and Graduate

Wages in the UK’ by Iftikhar Hussain, Sandra McNally and

Shqiponja Telhaj, Centre for the Economics of Education

Discussion Paper No. 99 (forthcoming). The study was funded

by the Economic and Social Research Council through its

Teaching and Learning research programme.

Iftikhar Hussain is a research economist in CEP’s education

and skills programme. Sandra McNally is director of CEP’s

education and skills programme. Shqiponja Telhaj is a

lecturer in economics at the University of Sussex and a

research economist in CEP’s education and skills programme.

A degree from a top universityis worth more, but a degree

from any university still paysoff in terms of higher earnings


8

Accurately documenting –and understanding thecauses of – rising labourmarket inequalities havebeen a major

preoccupation of CEP researchers overmany years. While the study of wage andemployment structures dates back a longway in economics (at least as far as AdamSmith), the large body of more recentacademic literature on rising inequalitybegan in the early to mid-1990s. It sprangfrom a recognition that wage gapsbetween higher and lower paid workerswere rapidly widening in a number ofcountries, notably the UK and the UnitedStates.

There are at least two main aspects ofthis research to which CEP economistshave made significant contributions andshaped the debate. The first is careful useof large-scale microeconomic data onindividuals’ wages and employment todocument what happened and to measureaccurately the extent to which the wagedistribution widened.

For many years, economists hadpointed to the stability of the wagestructure. But there was a sense in the late1980s and early 1990s that things werechanging. In the early to mid-1990s, CEPresearchers Stephen Machin and John

Schmitt (then a doctoral student at CEPand now a senior economist at the Centerfor Economic and Policy Research inWashington, DC), using data sources suchas the General Household Survey and theNew Earnings Survey, were the first todocument the substantial increase in wageinequality that had occurred since the late1970s (Schmitt, 1995; Machin, 1996).

Simultaneously, CEP research fellowRichard Freeman of Harvard University wasdocumenting a similar increase in wageinequality in the United States – whichwas worse than in the UK as those at thebottom were experiencing large real fallsin their earnings.

The second significant contributionwas to develop a better understanding ofthe proximate causes of rising wageinequality. Influential cross-country workfrom the late 1980s by Stephen Machin,John Van Reenen and various co-authorspinpointed ‘skill-biased technologicalchange’ as a force operating in the labourmarkets of a number of industrialisedcountries as generating greater labourmarket inequality than in the past (Bermanet al, 1998; Machin and Van Reenen,1998; Nickell and Van Reenen, 2001).

This change consists of technologicaladvances (such as computers) that havebeen observed to benefit more skilled or

educated workers and, at the same time,to be detrimental to the wages andemployment prospects of less skilled oreducated workers. As these technologiesdiffused into modern workplaces, therelative wages of skilled versus unskilledworkers rose or the employment rates ofthe skilled versus the less skilled rose – orin some countries, notably the UK and theUnited States, both happened.

CEP researchers have continued tomake significant contributions to thisdebate and we now know a lot more thanwe did when the initial research wasundertaken. It has become evident thatthe sharpest burst of rising wageinequality in the UK occurred in the1980s. In that decade, the whole wagedistribution widened as successively higherpercentiles of the wage distributionexperienced higher relative wage growth(Machin, 1999).

This can be illustrated by thinking of‘upper tail wage inequality’ as the wage atthe ninetieth percentile (the wage receivedby someone 10% below the top of thewage distribution) relative to the fiftiethpercentile (the person right in the middle),and ‘lower tail wage inequality’analogously as the tenth percentile wagerelative to the fiftieth.

Figure 1 shows decade-by-decade

In the second of CEP’s ‘big ideas’ series,Stephen Machin surveys significant researchfindings on wage inequality that have emergedfrom the Centre over the past three decades.

Big ideasRising wage inequality


9

changes in overall, upper tail and lower tailwage inequality (in annualised percentagepoints). In the 1980s, both lower and uppertail inequality rose significantly as the 90-50wage differential increased and the 10-50wage differential fell. A key aspect to thiswas rising wage gaps between more andless educated workers – in other words, thewage ‘returns’ to education rose.

The next two decades proved

somewhat different, however. In the 1990s,wage inequality continued to rise but at amore muted pace. In the 2000s, lower tailinequality narrowed as the 10-50differential increased but upper tailinequality continued to expand.

How does recent research reconcilethese patterns of change? First of all, the‘naïve’ story of skill-biased technologicalchange has evolved into a more nuanced

one. It may be more plausible to think oftechnological advances as damaging thewage and employment prospects forworkers in jobs that can be replaced bymachines and computers – those jobs thatrequire workers to perform routine tasks.At the same time, jobs requiring workers todo more complex, non-routine tasks cannotbe replaced or downgraded in terms of thewages they pay in the same way.

Figure 1:

Overall, upper tail and lower tail wage inequality rosesignificantly in the 1980s and at a slower pace in the1990s; in the 2000s, lower tail inequality fell while uppertail inequality continued to rise

2

1

0

-1

Non-entrepreneurs

Entrepreneurs

2000s1990s1980s

■ 90-10 wage differential■ 90-50 wage differential■ 10-50 wage differential

The rises in UK and USwage inequality are due to a combination oftechnological change andinstitutional changes, suchas union decline, affectingthe labour market

An

nu

alis

ed p

erce

nta

ge p

oin

t ch

ange

s


10

So it was not only unskilled workerson the production line who saw their jobsdisappear and wages fall, but also thoserelatively skilled workers like bank clerkswho performed routine clerical tasks.Their jobs were increasingly replaced bynew information technology such as ATM machines.

CEP work by Maarten Goos and Alan Manning showed that this processeffectively caused a polarisation of thelabour market (Goos and Manning,2007). At the top end of the occupationalstructure, job growth since the late 1970shas been rapid – and wages have risensignificantly. But the middle has hollowedout, as the jobs that could be replaced byautomation have been lost. At the bottomend, there remain jobs that cannot bereplaced by machines or computers – likecleaning or caring jobs – and the demandfor some very low wage jobs has risen.

While we can say that the significantgrowth in upper tail inequality in the UKsince the late 1970s was down in largepart to skill-biased technological change,we had to find other mechanisms toexplain the big rise in lower tail inequalityin the 1980s and the subsequent bounceback of the 2000s. Here CEP researchpointed to changes in the role of labourmarket institutions, in which there havebeen two important episodes.

First, the rapid decline of tradeunions, which had traditionally proppedup the wages and employment of low-skilled workers, played a role in risinglower tail inequality. Second, asdocumented in research by RichardDickens and Alan Manning, theintroduction of the National MinimumWage in April 1999 has been important insecuring sizeable relative wage gains forlow paid workers in the 2000s and thuscontributed to the more recent narrowingof lower tail inequality (Dickens andManning, 2004).

Where does this leave us today? CEPresearch on this important question hasbeen vital in documenting and explainingwhy labour market inequalities have risenover time in the UK, and in placing thisrise into its appropriate internationalcontext. But many important issues arestill outstanding.

For example, our work found that therole of developing countries and foreigncompetition in fostering lower wages forthe less skilled in the UK was, perhaps

surprisingly, minimal. Although in theory,trading with poorer countries could leadto the wages of unskilled British workers‘being set in Beijing’, this did not seem tobe happening.

But this literature mainly pre-datedthe rise of China and India in the mid-1990s, and ignored some of the ways inwhich trade itself could stimulate greatertechnical change. Both these issues arebeing pursued in our future work usingnew sources of data.

In conclusion, our work has shownthat the recent rises in UK wageinequality have mainly been due to acombination of skill-biased technologicalchange and institutional changes affecting

the labour market. But what kind ofpolicies could be implemented to reversethe trend in inequality?

The long-term policy has to be thebuilding up of human capital. But what isthe best way to do this? In the shorterterm, reform of labour market institutionssuch as the minimum wage and tradeunions could also be important. We willturn to policies with the potential totackle distributional problems in futureissues of CentrePiece.

Further reading

Eli Berman, John Bound and Stephen Machin

(1998) ‘Implications of Skill-biased

Technological Change: International

Evidence’, Quarterly Journal of Economics

113: 1245-80.

Richard Dickens and Alan Manning (2004)

‘The National Minimum Wage and Wage

Inequality’, Journal of the Royal Statistical

Society Series A 167.

Maarten Goos and Alan Manning (2007)

‘Lousy and Lovely Jobs: The Rising

Polarization of Work in Britain’, Review of

Economics and Statistics 89: 118-33.

Stephen Machin (1996) ‘Wage Inequality in

the UK’, Oxford Review of Economic Policy 7:

49-62.

Stephen Machin (1999) ‘Wage Inequality in

the 1970s, 1980s and 1990s’, in The State of

Working Britain edited by Paul Gregg and

Jonathan Wadsworth, Manchester University

Press.

Stephen Machin and John Van Reenen (1998)

‘Technology and Changes in Skill Structure:

Evidence from Seven OECD Countries’,

Quarterly Journal of Economics 113: 1215-44.

Stephen Nickell and John Van Reenen (2001)

‘Technological Innovation and Economic

Performance in the United Kingdom’, in

Technological Innovation and Economic

Performance edited by Benn Steil, David

Victor and Richard Nelson, Princeton

University Press.

John Schmitt (1995) ‘The Changing Structure

of Male Earnings in Britain, 1974-88’, in

Differences and Changes in Wage Structures

edited by Richard Freeman and Lawrence

Katz, University of Chicago Press.

Stephen Machin is research director of CEP

and professor of economics at University

College London.

The long-term policy response torising inequality has to be thebuilding up of human capital


11

This three-day conference debated all aspects of theimpact of the emerging economic powers on Europe andthe United States and the changes within India and China themselves. A highlight of the conference was apolicy panel featuring Athar Hussain of LSE’s AsiaObservatory, Alan Winters, recently appointed chiefeconomist at the UK’s Department for InternationalDevelopment, Martin Wolf of the Financial Times and Yale professor Peter Schott.

CEP’s director John Van Reenen, chaired a lively discussion,which he kicked off by asking three questions aboutlessons for growth, the effect of globalisation oninequality and threats to the political coalition forglobalisation.

The panellists focused on the fact that the lessons fromChina and India’s development may be limited for othercountries. A key factor has certainly been an emphasis on exports, a degree of opening up of markets to imports and some liberalisation of the economy. But Chinahas embarked on a more classical development paththrough manufacturing exports, whereas India’s growth

has been startlingly service-driven, a very new mode of development.

There was some dispute over the role of trade in drivinginequality in the rich countries. Martin Wolf argued fortechnology having the major role whereas the otherpanellists saw a greater role for trade. John Van Reenen’sresearch shows that the two are inter-related: trade withChina appears to have stimulated faster technologicalprogress in the West.

Although the panel agreed that the prospects for theDoha round were grim, Martin Wolf pointed to the factthat globalisation had created footloose multinationalswith much less vested interest in protectionism in any onecountry. This by itself undermined one of the key lobbiesagainst trade – vested domestic business interests.

The conference on the emergence of China and India in the

global economy was co-organised by Stephen Redding and

John Van Reenen of CEP. Full details on the conference are

available here: http://cep.lse.ac.uk/_new/events/event.asp?id=50

The emergence of China and India in the global economyJuly 2008 conference

CHINAINDIACHINDIACHINDIACHINACHINACHINACHINAINDIAINDIACCCC


12

The gap in income between rich people and the averageBriton has widened very markedly over the last threedecades. Andreas Georgiadis and Alan Manning wonderwhy voters have not forced successive governments toincrease taxes on the rich.

‘I warn you that there are going to be howls ofanguish from those rich enough to pay over 75%on their last slice of earnings.’Denis Healey, Labour Party Shadow Chancellor, 1973.

‘The justice for me is concentrated on liftingincomes of those that don't have a decent income.It's not my burning ambition to make sure thatDavid Beckham earns less money.’Tony Blair, Labour Party Leader, 2001 election campaign.


The gap in income between rich people and the averageBriton has widened very markedly over the last threedecades. In the early 1970s, the top 10% received about one quarter of total income but their proportion of total income now is more like one third. We mightexpect that the political response to this – at least from aLabour government – would take the form of moreredistribution, including higher marginal tax rates for therich. But this has not happened.

The simplest way to see it is that the top rate of incometax fell from 83% in the late 1970s to 40% in 1988 –since when it has not changed. There is now no majorpolitical party proposing substantial rises in the topmarginal rate of income tax. Looking across the tax andbenefit system, we can see that redistribution has notincreased even as inequality has risen.

In a democracy, it is the voters in the middle (the medianvoter) who have the most influence since they are the‘swing’ voters who determine the outcomes of elections.And as the median voter has a below average income(due to a handful of super-rich people), we might expectthem to put pressure on the government to redistributeresources from the rich to themselves. According to thisview, a rise in income inequality will lead to moreredistribution as the median voter tries to get a slice ofthe extra pie going to the rich.

Of course, there is substantial redistribution: the share ofthe top 10% in final income is about one quarter, whichis considerably lower than their one third share in original

income. But the prediction that more inequality will leadto more redistribution is not supported by the evidence.Not only has redistribution in the UK failed to keep pacewith rising inequality, but there is also little evidence thatit applies across other countries. Perhaps the starkestcomparison is between the United States and Europe: theformer has higher levels of inequality than the latter andless redistribution.

So why has the rise in inequality in the UK been met withso little demand for higher taxes on the rich? One viewmight be that our democracy is closer to ‘one pound, onevote’ instead of ‘one person, one vote’. If this were so,then a rise in the share of income going to the richwould also lead to a rise in their share of political power,hence potentially explaining the lack of a redistributiveresponse.

Our research has been considering other possibleexplanations. We analyse the British Social AttitudesSurvey (BSAS) to see whether attitudes towardsredistributive taxation have been changing.

The answer is that they have. Figure 1 shows the averageresponse of individuals to the statement ‘governmentshould redistribute income from the better-off to thosethat are less well-off’ from 1 (‘strongly disagree’) throughto 5 (‘strongly agree’) – so that higher average valuesrepresent a greater demand for redistribution.

Figure 1 indicates that the demand for redistribution fellin the early 1980s but then rose from the mid-1980s to

in brief...


13

the mid-1990s. This was the period in which inequalitygrew fastest and it seems likely that the lack ofredistribution by the Thatcher and Major governmentswas not particularly popular.

Tony Blair inherited a large demand from voters for moreredistribution in 1997. Since the election of New Labour,support for redistribution plummeted to a new low. Oneexplanation might be that the Labour governmentsucceeded in redistributing income so that voters’demand were met. But the data suggest that little hasactually happened to inequality or redistribution since1997.

There has been some redistribution towards children inpoverty and poor pensioners. But these are relativelysmall deals in the big scheme – the underlying trends ininequality in the economy as a whole. We are left withthe job of explaining the puzzle that inequality now ismuch higher than in the 1970s but the demand forredistribution is much lower.

What determines attitudes towards redistribution? Ouranalysis of the BSAS indicates a number of importantfactors, all in line with the theory of the median voter:

� First, your personal circumstances – the rich aremarkedly less in favour of redistribution than the poor.

� Second, if you care a lot about the poor or are veryenvious of the rich, it is quite likely that you support alot of redistribution.

� Third, if you believe that high taxes discourage work,

then you are less likely to support redistribution.� Finally, if you think the government cannot be trusted,

then you are not likely to support as muchredistribution.

The views of the British population have changed overtime and our analysis attempts to link the fall in thedemand for redistribution to changing attitudes. We findthat we can do a pretty good job: changes in attitudestowards redistribution can explain almost two thirds ofthe decline in the demand for redistribution in the UK.

The single most important change is that many fewerpeople now seem to believe in the ‘class war’ – that thereis one law for the poor and one for the rich, or that bigbusiness benefits owners at the expense of workers.

So people’s beliefs are as important as their economiccircumstances in explaining attitudes to political issueslike redistribution. And these beliefs can change fast.That politics is a battle for ‘hearts and minds’ is notsurprising, but why have the beliefs of British voterschanged in this way?

It seems that British views are becoming more like theviews of Americans: those in the middle are no longerenvious of the rich – instead they aspire to be the rich.But as we enter a recession in which the average Briton isquite likely to feel the pinch, it may once again becomean attractive political policy to seek to increase the shareof taxes paid by the rich.

This article summarises ‘Spend It

Like Beckham? Inequality and

Redistribution in the UK, 1983-

2004’ by Andreas Georgiadis and

Alan Manning, CEP Discussion

Paper No. 816 (http://cep.lse.ac.uk/

pubs/download/dp0816.pdf).

Andreas Georgiadis is a

research economist in CEP’s

labour markets programme.

Alan Manning is director of CEP’s

labour markets programme.

Figure 1:

Agreement with the statement that ‘government shouldredistribute income from the better off to those that areless well-off’ (1=strongly disagree; 5=strongly agree)

3.5

3.4

3.3

3.2

3.1

3.0

2.9

2004200019951990198619791974

Year

Dem

and

for

red

istr

ibu

tion

’

Voters seem to value redistribution less – and this is the main reason why it hasn’t increased


14

Since 1997, the governmenthas launched a variety ofinitiatives designed toimprove the life chances ofpoor children. The logic was

clear: every child should have an equalchance to succeed in life regardless of howwell-off their parents are. Whether parentalbackground is becoming more or lessimportant in determining child outcomes isof interest in itself, and it may help us toevaluate the policy initiatives designed toincrease social mobility from onegeneration to the next.

We have shown in work with PaulGregg (Blanden et al, 2005, 2007) that, onaverage, the life chances of a child borninto a poor household in 1970 were worsethan those of a child born into a similarhousehold in 1958. In particular, weshowed that the earnings of individualsborn in 1970 were more strongly related tothe income of their parents than those ofthe earlier cohort. But until now, we havenot been able to assess whetherintergenerational mobility has declinedfurther since then.

To do so is especially important sincechildren even from the second cohortwould have been in their late twenties bythe time New Labour came to power in1997 and would not have been affected

by policies such as Sure Start (whichprovided services for pre-school agechildren) and the Excellence in Citiesprogramme (which directed more resourcesto inner city schools).

Our new study, funded by the SuttonTrust, uses more recent data to attempt todetermine what has happened to socialmobility since 1970. It uses information onthe children of both the 1958 and 1970cohorts: on average, the children of those born in 1958 were born in 1985,and the children of the 1970 cohort wereborn in 1999.

In addition, we use information onyoung adults from the British HouseholdPanel Survey (BHPS), a nationallyrepresentative sample of 5,500households, as well as children born in2000 and 2001 from the MillenniumCohort Study.

One problem with looking at themobility of these later cohorts is thatchildren born recently have not entered thelabour market yet. So instead, weinvestigate the link between parents’income and the intermediate outcomes oftheir offspring. These outcomes includeacquiring a degree by the age of 23,cognitive test scores during the early years,and parents’ reports of behaviour duringchildhood.

We then analyse whether the linkbetween parental income and theseoutcomes has been strengthening overtime based on the principle that if therelationship has strengthened, then this islikely to lead to a decrease inintergenerational mobility later in life. Wecan check if this principle is true by usingthe 1958 and 1970 cohorts, for which we

It was CEP research that first established thatthe life chances of children born into poorhouseholds in 1970 were worse than those for children born twelve years earlier.Now Jo Blanden and Stephen Machin look at more recent evidence to find out whathappened to the life chances of children borninto poor families since then.

Recent trends in intergenerational mobility:

Will the downward trend continue?

Illus

trat

ion:

Jen

ny M

umfo

rd


Figure 2 repeats this exercise forreading test scores (at age 11 in the 1958cohort and age 10 in the 1970 cohort) andshows the average test score percentileachieved by those from differentbackgrounds. Inequality also grows acrossthe cohorts by this measure.

We then compare the changingrelationship between these outcomes and family income between 1958 and1970 with recent trends. If intermediateoutcomes are ever more strongly related to parental income, this suggests that the trend of declining social mobility has continued.

A key assumption is that educationaloutcomes for children are a good (and reasonably constant) predictor ofwhat they will earn as adults. This has been borne out by other studies: the more educated you are, the more you earn.

have information on intermediateoutcomes and for whom we know thatintergenerational income mobility fell.

Figures 1 and 2 show how theintermediate outcomes that we use arerelated to parental income. Figure 1 shows that although children born in 1970were more likely to have a degree by theage of 23 than children born in 1958, the growth in degree-level education was strongest among those from therichest backgrounds.

Although 7% of children born in 1970 from the poorest fifth of householdshad a degree (up from 5% of the children born 12 years earlier), degreeattainment from those from richbackgrounds grew much more quickly:from 20% to 37%. So inequality in degreeattainment widened from 15 percentagepoints to 30 percentage points betweenthe two cohorts.

Figure 3 compares the degreeachievement of the 1970 cohort with thatof more recent ‘pseudo-cohorts’ drawnfrom the BHPS. Unlike the earlier period,there is little to suggest that the gapbetween the rich and the poor (in terms ofdegree attainment) widened. Indeed, thegap in degree attainment between childrenfrom the poorest households and therichest remains static between the 1970cohort and the first BHPS sample (who areon average six years younger).

The gap appears to widen slightlybetween the first and second BHPS groups,but the change is too small for us to ruleout the possibility that it is an artefact ofthe data. So the recent evidence should beinterpreted as ‘no change’: it appears(from this measure at least) as though thedecline in intergenerational mobility hasbeen arrested.

Degree attainment is only one of the

Inequality in degreeattainment and testscores widenedbetween the 1958 and1970 cohorts

Figure 1:

Inequality in degree attainment – 1958 and 1970 cohorts

40%

35%

30%

25%

20%

15%

10%

5%

0%

Non-entrepreneurs

Entrepreneurs

1970 cohort age 231958 cohort age 23

■ Poorest 20%■ Richest 20%■ Gap in degree achievement (richest-poorest)

Figure 2:

Inequality in test score percentiles – 1958 and 1970 cohorts

70%

60%

50%

40%

30%

20%

10%

0%

Non-entrepreneurs

Entrepreneurs

1970 cohort age 101958 cohort age 11

■ Poorest 20% – test score■ Richest 20% – test score■ Gap in test score (richest-poorest)

intermediate outcomes in which we areinterested. But as it turns out, the storyfrom the other measures is similar: the gapbetween rich and poor children has notwidened in recent years.

Figure 4 shows this for test scoreperformance at around the age of 5. Thebars look almost identical across the threedatasets, indicating that there has been nochange in the relationship between testscores and family income in the cohortsborn from the mid-1980s to the turn ofthe millennium.

Evidence from behavioural testsconfirms and consolidates our findings.The relationship between family incomeand externalising or ‘acting out’ behaviourincreases in the earlier cohorts but shows

no change for those born since the mid-1980s.

As far as we can tell, therefore, itseems that our previous finding of a fall inintergenerational mobility between the1958 and 1970 cohorts will not continuefor children born more recently. Looking atthe connection between interveningfactors - educational attainment, testscores and behavioural measures - andfamily income for more recent cohorts wefind little evidence of change.

It appears that the decline in socialmobility may well have flattened out. Thismay be either good or bad news for policy:while it may have stopped the previousdecline, it has failed to lead to an overallimprovement in mobility.


16

This article summarises ‘Up and Down the

Generational Income Ladder in Britain: Past

Changes and Future Prospects’ by Jo Blanden

and Stephen Machin, National Institute

Economic Review 205: 101-17. This work

originally appeared in December 2007 as the

Sutton Trust report ‘Recent Evidence on

Intergenerational Mobility’. The authors

gratefully acknowledge the financial support

of the Trust.

Jo Blanden is a lecturer in economics at the

University of Surrey and a research associate

in CEP’s education and skills programme.

Stephen Machin is research director of CEP

and professor of economics at University

College London.

Further reading

Jo Blanden, Paul Gregg and Stephen Machin

(2005) ‘Educational Inequality and

Intergenerational Mobility’, in What’s the

Good of Education? The Economics of

Education in the UK edited by Stephen

Machin and Anna Vignoles, Princeton

University Press.

Jo Blanden, Paul Gregg and Lindsey

Macmillan (2007) ‘Accounting for

Intergenerational Persistence: Non-cognitive

Skills, Ability and Education’, Economic

Journal conference volume: C43-60.

Figure 3:

Inequality in degree achievement - 1970 onwards

50%

45%

40%

35%

30%

25%

20%

15%

10%

5%

0%

Non-entrepreneurs

Entrepreneurs

BHPS age 23 1999 BHPS age 23 20021970 cohort age 23 1993

■ Poorest 20%■ Richest 20%■ Gap in degree achievement (richest-poorest)

Figure 4:

Inequality in test score percentiles - more recent cohorts

60%

50%

40%

30%

20%

10%

0%

Non-entrepreneurs

Entrepreneurs

Children of 1970 cohortage 4-6 – 2004

MCS age 5 – 2005/6

Children of 1958 cohortage 5-7 – 1991

■ Poorest 20% – test score■ Richest 20% – test score■ Gap in test score (richest-poorest)

The fall inintergenerationalmobility betweenthe 1958 and1970 cohorts hasnot continued forchildren bornmore recently

On 23 June 2008, thePrime Minister gave aflagship speech to schoolleaders in which he saidthat ‘raising social

mobility in our country is a nationalcrusade in which everyone can join andplay their part’ and pledged that a WhitePaper on social mobility would beproduced by the end of the year.

On the same day, CEP – which hasbeen at the forefront of recent researchinto intergenerational mobility – hosted a one-day conference on the subject, co-organised with the Centre forLongitudinal Studies at the Institute forEducation. The day brought togethermany of the leading academics in the field– what Polly Toynbee described in TheGuardian as ‘a roll call of top sociologistsand economists’.

The first session contrasted theapproaches and results from thesociological and economic approaches tomeasuring intergenerational mobility. ColinMills presented joint work with JohnGoldthorpe from the sociological sidewhile Jo Blanden provided the economists’perspective, presenting a paper she haswritten with Paul Gregg and LindseyMacmillan. Discussion from Robin Naylorand John Hills brought out the similaritiesand differences in findings and approach.

Paul Gregg and Lindsey Macmillanthen brought the evidence onintergenerational mobility right up to date,looking at how the relationship between

family background and educationalachievements has changed under NewLabour and the implications of this forfuture mobility. Miles Corak gave the lastpaper of the morning, showcasingCanadian data which indicates that 40% of men follow their father into thesame firm.

In the afternoon, Marco Francesconipresented joint work with John Ermisch,using data from the Millennium CohortStudy to examine the impact of familybackground on birth weight, one of thevery earliest ways in which familyinvestments can influence children’soutcomes. Maia Güell followed with apaper (joint with Sevi Mora and ChrisTelmer) showing how the ability ofsurnames to explain income in thepopulation is related to, and can estimate,intergenerational mobility from onegeneration to the next.

Steve Nickell responded by questioningthe worth of estimating intergenerationalmobility, arguing that one number isinsufficient to capture all the complexsocial and economic processes involved. In the discussion that followed,intergenerational elasticity was comparedwith the growth rate and the Ginicoefficient measure of inequality as one ofthe essential statistics that describe aneconomy.

The final academic paper of the daywas by Anders Björklund (joint with LenaLindahl and Matthew Lindquist), usingdata from Sweden to explore the value of

sibling correlations in income as analternative measure of mobility.

The audience for the day came from abroad background: representatives ofcharities, thinktanks and governmentdepartments were present, alongsideacademics from across the UK. Inrecognition of this, the conference aimedto combine academic papers with abroader discussion of the policyimplications of social mobility research.

To this end, the day closed with aroundtable discussion chaired by StephenAldridge of the Cabinet Office. AcademicsStephen Machin and John Goldthorpe setthe scene by building a bridge betweenthe academic and policy viewpoints.Contributions then followed from LiberalDemocrat MP Lynne Featherstone,Conservative MP David Willetts, Lee ElliotMajor of the Sutton Trust and PollyToynbee.

Social mobility remains at the top ofthe policy agenda and the discussions atCEP have contributed to bringing theacademic and policy communities a littlecloser together in their thoughts.

The conference on intergenerational mobility

was co-organised by Jo Blanden of CEP and

Kirstine Hansen of the Centre for

Longitudinal Studies at the Institute for

Education. Further information is available

here: http://cep.lse.ac.uk/_new/events/

event.asp?id=48


17

Intergenerational mobilityJune 2008 conference


18

The British public is increasingly scepticalthat official inflation numbers reflect the truerise in the cost of living. Nicholas Oultonexplains the problems in creating an accuratemeasure of inflation, and outlines a newmethod of measuring rising prices.

How to measure the rising cost of living

When I tell my friendsthat I have beenstudying theproblem of path-dependence bias in

chain index numbers of the cost of living,their eyes tend to glaze over. But asidefrom its purely intellectual interest, thestudy of cost-of-living (and other) indexnumbers has many important practicalimplications. After all, whenever you readthat the consumer prices index (CPI) roseat an annual rate of 2.5% last month orthat GDP in the first quarter rose by 1.5%,you are being told something about indexnumbers.

To what extent do price indices likethe retail prices index (RPI), used to updatepensions and social security benefits inBritain, or the CPI, used by the Bank ofEngland as its inflation target, measure thetrue cost of living? An index is a group (or‘basket’) of goods, and by measuring theprice change of each good you canestimate the price change of all. Butbecause some goods take up a largerproportion of our disposable income thanothers, we need to adjust (or ‘weight’) theindex to reflect this.

In the past, such price indices quicklybecame out-of-date as they were basedon a fixed basket of goods: the weight ofdifferent goods in the baskets didn’tchange over time. And as people tend tosubstitute goods that have quickly risingprices for ones that have slowly risingprices, the share of our spending thateach good takes up varies, and becomesless and less similar to that of the originalindex.

So in Britain the basket of goods is notfixed but is changed every year in line withthe latest spending patterns – in thejargon, the CPI or the RPI is a chain index.But conventional price indices like the CPIand the RPI, considered as measures of thetrue cost of living, do suffer from what Icall ‘path-dependence bias’. This second

type of bias arises because the poor spendtheir money in different ways from therich, even when both face the sameprices.

One of the earliest and most robustempirical findings in economics is Engel’sLaw, named for an 1857 study by ErnstEngel of household budgets in Saxony.Engel found that richer households spentmore on food but that the richer thehousehold, the lower the share of food intotal expenditure.

Engel’s study is not just of historicalinterest. The share of the householdbudget devoted to food in Britain hasfallen by around half in the last 30 yearsand it’s likely that this is mostly due to therising average level of prosperity over thisperiod.

Engel’s Law illustrates a generalproblem with measuring the cost of living.Consider a household with a very lowstandard of living, spending say 60% of itsbudget on food, just like the majority ofthe households Engel studied in 1857 orindeed hundreds of millions of people inpoor countries today. Suppose the price offood rises by 20%, with other pricesremaining unchanged. Then moneyincome will probably have to rise by closeto 12% (0.60 x 20%), to leave thestandard of living unchanged.

Notice that we can’t say it must rise byexactly 12% since it might be possible forthe household to substitute some other

goods for food. For example, the need forfood could be reduced a little by burningmore fuel (this is the substitution biaspoint). But at such a low standard ofliving, the substitution possibilities areclearly limited.

Compare this household with amodern day British one, spending only15% of its budget on food. Now themaximum rise in income required tomaintain the standard of living intact isonly 3% (0.15 x 20%). In fact, it may be agood bit less as substitution opportunitiesare greater: if the price of food rises,households can buy cheap food instead ofexpensive food (more bread and less meat)or substitute other products for food,though poorer households will obviouslyfind this harder than richer ones.

The general point is that though pricestend to rise over time, they don’t all rise atthe same rate. The increase in the cost ofliving depends on the pattern ofexpenditure; this pattern differs betweenricher and poorer households and alsochanges as the average standard of livingrises over time. So a single number thattries to measure the change in the cost ofliving will be necessarily inaccurate, aspoor families spend their money ondifferent things than rich families do.

So when we find that a conventionalprice index like the RPI has risen at anaverage rate of about 6% per year overthe period 1974-2004, the key question is:whose standard of living is being taken asthe base for measuring the cost of living?The answer is, roughly, that of someonewith the average standard of living in themiddle of this period – around 1989.

But what’s so special about 1989?Why not take the viewpoint of peopletoday, who are of course considerablyricher on average than their counterpartsin 1989? Or the viewpoint of people in1974, who were considerably poorer?

Because people’s incomes change overtime, so do their spending habits. But if


19

From 1974 totoday, inflation

in the UK has hurt the

poor more thanthe rich

we want a true picture of how the cost ofliving is changing, this change in incomesconfuses things. We can’t simply use theoriginal shares of each item in thehousehold budget (in the base year, here1974) as substitution bias means that theprice index will overstate the cost of living:people would naturally buy fewer of thegoods that have seen quickly rising prices.But we shouldn’t really use the actualbudget shares in 2004 either, as this willlead to path-dependence bias – sincepeople now are richer than they were in1974, they want to buy different things.

What we would ideally like to know are the hypothetical budget shares:what the shares would have been if prices were at their actual level in 2004but the standard of living had been at its 1974 level (or any other year takenas the point of comparison). Of coursethese hypothetical shares cannot bedirectly observed.

Fortunately, there is a way ofestimating the hypothetical sharesindirectly. This can be done using onlyavailable data, namely prices and (actual)budget shares. When the hypotheticalshares have been estimated, we canconstruct an RPI for any level of thestandard of living we like.

For example, taking 1974 as the baseyear for the standard of living, we couldask: by how much would the income of ahousehold with the typical 1974 standardof living have to rise in order for it toenjoy the same standard of living at theprices of 2004? The same could apply toany other year.

Figure 1 shows the results ofestimating an RPI appropriate to thestandard of living of each year between

1974 to 2004. It reveals a very interestingpattern. The growth rate of the cost ofliving tends to be higher the further back in time the base year is. As incomeshave grown steadily over time, this impliesthat the rise in the cost of living wasgreatest if we take the viewpoint of thepoorest (the British in 1974) than if wetake the viewpoint of the richest (theBritish in 2004).

In other words, over this period,inflation hurt the poor more than the rich.The difference between the highest andthe lowest inflation rate is around 0.5%per year: this is the maximum size of thepath-dependence bias.

This method can also be appliedbetween countries. Policy towards povertyand development should be informed by


20

Figure 1:

Average growth rate of true cost-of-living index, 1974-2004by base year for standard of living (per cent per year)

the size of the gap between the rich andthe poor – and so how the cost of living ischanging for different groups is important,not least due to the rapidly rising foodprices that we are seeing at the moment.

But exactly the same problem of path-dependence bias arises here too. And the bias is potentially much larger:across the globe, living standards vary bymuch more than they did over the period1974-2004 in Britain, during which theyapproximately doubled.

The World Bank has just completedthe latest and fullest round of theInternational Comparison Program, whichhas delivered detailed price and spendingcomparisons for virtually every country of any size in the world, including (for the first time) China. Applying thenew method to these data will allow us to develop better measures of livingstandards across the globe.

6.4%

6.3%

6.2%

6.1%

6.0%

5.9%

2004200019951990198519801974

Base year for standard of living

Accuratemeasurement of the rising cost of living inthe developingworld is neededto informdevelopmentpolicy

This article summarises ‘Chain Indices of the

Cost of Living and the Path-dependence

Problem: An Empirical Solution’ by Nicholas

Oulton, CEP Discussion Paper No. 797

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

dp0797.pdf) and Journal of Econometrics

144(1): 306-24 (May 2008).

Nicholas Oulton is a senior visiting research

fellow at CEP.


21

John McCain and Barack Obamaare spending the lion’s share oftheir time on the campaign trailin just a few states – because it is only these so-called ‘swing states’

that could conceivably vote for eithercandidate. Neither candidate is likely todevote many resources to New York(safely Democrat) or South Carolina (solid Republican).

But do swing states just receive morevisits from presidential hopefuls comeelection time, or does their electoralimportance translate into somethingmore tangible? In particular, is it possiblethat swing states with concentrations ofcertain industries, notably those that are

fast becoming uncompetitive in the faceof cheaper imports from China andelsewhere, are able to push forprotectionist trade policies?

In 2002, President Bush introducedtariffs of up to 30% on imported steel. Itseemed to many that the president wasless worried about supposed dumping(the ostensible reason for the tariffs)than the potential loss of jobs in ‘rustbelt’ swing states like Michigan, Ohio,Pennsylvania and West Virginia.

The World Trade Organisation (WTO)ruled that the tariffs were not justified,and allowed the European Union (EU) toretaliate. The EU leaders were certainlyunder the impression that President Bush

was following political incentives, astheir proposed retaliatory tariffs weretargeted at goods exported from otherkey swing states, such as oranges(Florida) and cars (Michigan). Thepresident duly repealed the steel tariffs.

The attention paid to swing states,especially large swing states (which havemore votes in the ‘electoral college’ thatultimately chooses the president), shouldnot be understated. Florida was the mostintensively contested state in the 2004election, receiving a fifth of all candidatevisits and more than a quarter (27%) ofall money spent on television advertising.According to data from ‘Who Picks thePresident?’, a report by FairVote

As the US election campaign approaches itsclimax, the presidential candidates jet from onebattleground state to the next trying to drum upsupport. Research by Mirabelle Muûls andDimitra Petropoulou finds that the USgovernment is more likely to try to protectindustries concentrated in these states.

Do swing statesinfluence trade policy?

Industries that employ manypeople in politically decisive statesare more likely to be protected

Illus

trat

ion:

Jen

ny M

umfo

rd

protection by incumbent governments soas to attract votes. If we are right, thenindustries that receive protection arethose which employ many people.

We first show, using a model of theelectoral college, that an incumbentpresident has a strong incentive tomanipulate trade policy so as to build areputation of protectionism with theelectorate, especially in the large swingstates. By signalling his preferences ontrade policy, an incumbent president (or

party) can influence swing voters and thechances of re-election. This theoreticalwork lends support to the hypothesis thatthe 2002 US steel tariffs were introducedfor reasons of political expediency.

We test our theory by looking at the1984 presidential election. This election,between Ronald Reagan (the Republicanincumbent) and Walter Mondale(Democrat), provides as good a set ofdata as any to test the hypothesis thatindustries concentrated in states expected

(www.fairvote.org/presidential), this wasmore money than was spent onadvertising in 45 other states and theDistrict of Columbia combined.

The map shows the swing states,such as Florida with 27 electoral votes, in yellow, while blue and red statesdenote the Democratic and Republican‘safe states’, respectively, such asCalifornia (55 electoral votes) and Texas(34 electoral votes).

Previous research has tried to explain US trade policy by looking at the role of political lobbying. The thesis is simple: industries that arerepresented by stronger lobbies onCapitol Hill are more likely to beprotected. Our work investigates anotherpotential channel: the strong incentivesfor presidential candidates to try to winover certain states.

Our hypothesis is that industrieswhich employ many people in the keyswing states are more likely to receive

Hawaii

Alaska

California

ColoradoUtahNevada

Illinois

Missouri

Texas

Kentucky

TennesseeKansas

New Mexico

Nebraska

Wyoming Iowa Indiana

MichiganWisconsinMinnesotaMontana

7.Delaware

9.Washington, DC

Pennsylvania

6.New Jersey

8.Maryland

Virginia

West Virginia

Ohio

North DakotaWashington

South DakotaOregon

North Carolina

South Carolina

Maine

Georgia

Florida

Arizona

Idaho

ArkansasOklahomaAlabama

Mississippi

Louisiana

5.Connecticut

4.Rhode Island

New York

3.Massachusetts

1.New Hampshire

2.Vermont

12

3

45

6

7

89


22

Figure 1:

2004 US presidential election: Bush versus Kerry

� ‘Safe’ Republican states

� ‘Safe’ Democrat states

� Potential swing states in the 2004 election

Subtle non-tariff barriersallow presidents to get aroundWorld Trade Organisationrules on protection


23

to be both swing and decisive are morelikely to be protected.

Using data for 1983 (the yearpreceding the poll), we construct ameasure of which industries wereconcentrated in which states. Ourmeasure is based on employment, asworkers are voters. In particular, thehigher the proportion of employment inany given industry in a swing state, the higher we expect its electoralinfluence to be.

We combine these employment datawith a measure of each state’s jointprobability of being both swing anddecisive in the 1984 election. At thattime, Texas, with 29 electoral votes, wasthe most contentious state, with thehighest joint probability of being bothswing and decisive, followed closely byPennsylvania and California, while Floridaemerges as the seventh most swing anddecisive state.

The White House could not easilyintroduce tariffs to protect the relevantindustries, as such barriers would beillegal under WTO rules. But there areother, more subtle non-tariff barriers toprotect industries. These can take manyforms, such as anti-dumping measures or countervailing duties, as well asrequirements on how a good is produced,or its quality, preventing the import ofgoods that do not meet suchspecifications.

Our results show that while lobbiesare important in determining the ‘industrynon-tariff barrier coverage ratio’ (ourmeasure of protection), industries thatemploy many workers in swing states(especially those that carry a larger weightin the electoral college) enjoy higher levelsof protection than others.

So it appears that the electoral collegesystem that determines the outcome ofthe US presidential election createsincentives for presidents to intervene infavour of industries in swing and decisivestates. As the campaign season reaches itsclimax, it appears that politics couldtrump economics in the setting of tradepolicy. John McCain, a supporter of freetrade, acknowledges that ‘globalisationwill not automatically benefit everyAmerican (worker)’, while Barack Obama has stated that if electedpresident, he ‘will not sign another tradeagreement unless it has… protections forAmerican workers’.

The electoral college systemcreates incentives forpresidents to intervene in favour of industries inswing states

Mirabelle Muûls is an associate of CEP’s

globalisation programme and a research

associate at the Grantham Institute for

Climate Change, Imperial College London.

Dimitra Petropoulou is a stipendiary

lecturer in economics at Hertford College,

University of Oxford.

This article summarises ‘A Swing-state

Theory of Trade Protection in the Electoral

College’ by Mirabelle Muûls and Dimitra

Petropoulou, CEP Discussion Paper No. 849


dp0849.pdf)


By influencing market expectations of futureinterest rates, central bank communicationshave become a key tool of monetary policy-making. Research by Carlo Rosa finds thatannouncements by the US Federal Reservemove market interest rates significantly morethan comparable announcements from theEuropean Central Bank.

24

Words that work:comparing the effectiveness of central bank communications


25

Over the past few years,financial marketparticipants – especiallythose active in the bondmarkets – have benefited

from increased transparency from centralbanks about how they set monetarypolicy. Alongside their interest ratedecisions, central banks increasinglypublish detailed explanatory documents,such as the Bank of England’s quarterlyInflation Report.

Central banks have tried to becomemore transparent about their futuremonetary policy intentions in part becauseexpectations of future interest ratedecisions are priced into today’s moneymarkets – and so, in theory at least,communication can be a tool to tighten orloosen policy, as much as actually alteringthe key policy rate.

Of particular note are central banks’‘balance-of-risk’ statements –announcements about the likelihood of afuture increase or decrease in the targetrate. These are used by marketparticipants to predict the future directionof interest rates.

The fundamental question for centralbankers is whether (and to what extent)financial markets react to theirannouncements over and above anychange in the interest rate itself. Forexample, the statement following theFederal Open Market Committee (FOMC)meeting on 28 January 2004 led to one ofthe largest reactions in the bond market inthe past 15 years: no less than a 25 basispoint increase (one quarter of onepercentage point) in the five-year Treasurybond rate in the half-hour surrounding the announcement.

Contrary to conventional wisdom, thisreaction was spurred by the Fed’s wordsrather than its deeds. As marketparticipants expected, the FOMC did notchange its target interest rate. Instead, itwas the decision to replace the phrase‘policy accommodation can be maintainedfor a considerable period’ in itsaccompanying statement with ‘theCommittee believes it can be patient inremoving its policy accommodation’ thatled market participants to revise upwardstheir expectations of future interest rates.

Are the words of central bankersalways heeded in this way? And are thewords of some central banks more effectivethan others? To answer these questions, my

research looks at the communicationstrategies of the European Central Bank(ECB) and the Fed. I also directly comparethe ability of the ECB and the Fed to affectmarket rates using deeds (changes in thepresent target rate) or words (balance-of-risk statements).

I find that the Fed is much moreeffective than the ECB at steering marketinterest rates on bonds of all maturities.Figure 1 provides a graphical comparison ofthe average total effects of central bankwords and deeds on interest rates both inEurope and the United States. It is strikingto note that overall the Fed is able to movemarket rates twice as much as the ECB.

I also find that long-term US interestrates react much more strongly to newinformation from the Fed than theequivalent reaction of European long-termyields to new information in ECBannouncements. This discrepancy can beexplained in two ways.

One possibility is that the Fed’s long-term inflation objective is less explicit thanthe ECB’s objective: the Fed’s mandate isto pursue stability of both prices andeconomic activity while the ECB’s mandateassigns overriding importance to pricestability. Hence, long-term inflationexpectations (and hence interest rates) inthe United States may be more sensitive toFed statements than are those in the euroarea. In other words, Fed statements maycontain information not only about itsfuture policy intentions but also about itsopaque inflation target, which may varyover time.

An alternative explanation might bethat Fed statements are more informativethan ECB statements, and as a result theformer move interest rates in the moneymarkets more than the latter.

My analysis suggests that the greatersensitivity of US long-term yields to centralbank communication is intimately related

Figure 1:

The effects of central banks’ decisions and announcementson interest rates, 1999-2006

Note: This figure plots the average total effect of the surprise components of central banks’

words and deeds on interest rates for the period 1999-2006. The horizontal axis is maturity of

interest rates – from three months to 10 years. The vertical axis is interest rates in basis points.

■ ECB■ Fed

0

2

4

6

8

10

10 year7 year5 year3 year2 year12 month6 month3 month

The Fed’s wordshave a muchgreater effect onmarket interestrates than thoseof the ECB

powerful than others. My researchsuggests that the Fed’s words are treatedas a more accurate guide to futuremonetary policy than those of the ECB –and so a change in tone by the Fed ismore likely to move markets. What’s more,the prevalence of dollar-denominated debtmeans that Fed words even have thecapacity to move European interest rates.

So why does the Fed move the marketmore than the ECB? The Fed differs fromthe ECB in at least two respects. The Fedhas not only been much more active thanthe ECB, which could mean that themarket understands better what it saysand does – a transparency effect. But theFed also has a different institutionalmandate compared to the ECB – the Fedis not ‘inflation targeting’.

Theoretical research has shown thatunder the Fed's mandate, it is optimal tocommunicate more information to thepublic than under an inflation targetingregime. But I find empirically that thereason why the market responds more tothe Fed's announcements compared withECB statements is not due to its mandate;rather, it is a pure transparency effect.

to the higher informational content of Fedstatements compared with those of theECB rather than to any difference in their institutional mandate. Figure 2 showsthat the Fed’s announcements predict itsfuture actions more precisely than thecorresponding announcements from the ECB.

Both the ECB and the Fed areconsistent (they match words with deeds)but there remain some differences in theircommunication policies. The ECB is fullytransparent in the very short run – what itwill do in the next month. For example,when the keyword ‘strong vigilance’ isused, the market understands that theECB will increase rates at its followingGoverning Council meeting. The Fed ismore transparent in the short and mediumrun – beyond the next meeting – and thisis clear in Figure 2.

There are also notable differences inthe word length of communications: theFed's balance-of-risk statement containsabout 220 words on average while theECB's statement contains 1,163 words (or 4,533 if the ‘Questions and Answers’section is also considered). Buttransparency is not a matter of thenumber of words. Indeed, the ECB needsto explain the content of its statementduring the Questions and Answers. Forexample, at the press conference on 5 June 2008, the ECB’s president Jean-Claude Trichet explained the meaningof the expression ‘heightened alertness’(http://www.ecb.int/press/pressconf/2008/html/is080605.en.html).

If central bank words can movedomestic market interest rates, then it islikely that these words may affect interestrates in bonds denominated in othercurrencies as well. I find that since 1999,the Fed has been more able to moveEuropean interest rates of all maturitiesthan the ECB to move US rates.

What drives this asymmetricrelationship? During the early years of theeuro, the ECB’s likely conduct of monetarypolicy was not well known, and financialmarket participants seemed to useinformation from the Fed to forecastfuture ECB behaviour on the assumptionthat ECB monetary policy would beinfluenced by Fed policy. Moreover, euro-denominated money and bond marketswere much smaller than dollar-denominated fixed income assets. It is notsurprising, therefore, that the causal effectcomes from the United States to Europe,and not vice versa.

But it is not clear whether the effect ofthe Fed’s behaviour on European interestrates is a simple consequence of globalfinancial integration, or whether financialintermediaries think that the ECB really isgoing to mimic the Fed’s behaviour. My results indicate that the ability of theFed to move euro- as well as dollar-denominated debt seems to be tied to the predominance of dollar-denominatedfixed income assets rather than to anattempt by the ECB to follow the Fed’smonetary policy.

It is clear that central banks’ words arepowerful tools – but some are more


26

This article summarises ‘Talking Less and

Moving the Market More: Is this the Recipe

for Monetary Policy Effectiveness? Evidence

from the ECB and the Fed’ by Carlo Rosa,

CEP Discussion Paper No. 855


dp0855.pdf).

Carlo Rosa is a research associate at the

Center for Operations Research and

Econometrics (CORE), Université Catholique

de Louvain, and an associate in CEP’s

macroeconomics programme.

Figure 2:

Testing central bank transparency: the predictive ability of central bank actions using statements

Note: This figure plots the predictive ability of future policy rates using centr

centrepiece - welcome to lse research online - lse...

Documents