Global Metro Monitor

8/2/2019 Global Metro Monitor

1/52

GLOBAL METRO MO THE PATH TO ECONOMIC RE

Global Metro Monitor

PREPAREd BY

METROPOLITAN POLICY PROGRAM, THE BROOkINGs INsTITuTIONLsE CITIEs, LONdON sCHOOL Of ECONOMICs ANd POLITICAL sCIENCE

wITH dEuTsCHE BANk REsEARCHsuPPORTEd BY THE ALfREd HERRHAusEN sOCIETY, THE INTERNATIONAL fORuM Of dEuTsCHE BANk

dECEMBER 2010

A PRELIMINARY OVERVIEw Of 150 GLOBAL METROPOLITAN ECONOMIEs IN THE wAk


2/52

2 G L O B A L M E T R O M O N I T O R

Metropolitan Policy Program, Brookings Institution Alan Berube Alec Friedho Carey Nadeau

LSE Cities, London School o Economics and Political SciencePhilipp Rode Antoine PaccoudJens Kandt

Deutsche Bank ResearchTobias JustReto Schemm-Gregory, Helmut-Schmidt University

Special ThanksMichael Bruninger, The Hamburg Institute o International EconomicsGreg Clark, City Advisor

Toby Irving, Ox ord EconomicsEmilia Istrate, Brookings Anthony Light, Ox ord EconomicsUrs Mller, BAK Basel EconomicsSanjeev Sanyal, Sustainable Planet Institute Vincenzo Spiezia, OECDTony Travers, LSE

Publication

Adam Kaasa, Co-ordination, LSE CitiesMiranda Iossi dis, Design, LSE CitiesGip elgold Werbeagentur GMBH, Art Direction

AuTHORsauthors

First published by The Metropolitan Policy Program, The Brookings Institution and LSE Cities, London School o Economics andPolitical Science, 2010.

While every e ort has been made to ensure the accuracy o the material o the material in this report, the authors, The MetropolitanPolicy Program and/or LSE Cities will not be liable or any loss or damage incurred through the use o this document. I noti ed, TheBrookings Institution and the LSE will recti y any errors or omissions at the earliest opportunity.

All rights reserved. No part o this publication may be reproduced, stored in a retrieval system, or transmitted, in any orm or by any means, without the prior permission o The Brookings Institution and/or the London School o Economics and Political Science.

Disclaimer: Brookings recognizes that the value it provides to any donor is in its absolute commitment to quality, independenceand impact. Activities sponsored by its donors refect this commitment and neither the research agenda, content, nor outcomes areinfuenced by any donation.

LSE CitiesLondon School o Economicsand Political ScienceHoughton StreetLondon WC2A 2AEUnited Kingdom

www2.lse.ac.uk/lsecities

Metropolitan Policy ProgramThe Brookings Institution1775 Massachusetts Avenue NW Washington, D.C.20036United States o America

www.brookings.edu/metro


3/52


4/52


The global nancial crisis o the late 2000s precipitated an economicdownturn o such magnitude and reach that many now re er to theperiod as the Great Recession. According to the International Monetary Fund, global economic output, which had grown at an annual rate o 3.2percent rom 1993 to 2007, actually shrank by 2 percent rom 2008 to2009. A precarious economic recovery is now underway.

Aggregate views o the global economy, however, mask the distinctexperiences o its real hubsmajor metropolitan areas. Metro areas,

which are economically integrated collections o cities, suburbs, ando ten surrounding rural areas, are centers o high-value economicactivity in their respective nations and worldwide. And because metrosorm the undamental bases or national and international economies,understanding their relative positioning be ore, during, and a ter theGreat Recession provides important evidence on emerging shi ts in thelocation o global economic resilience and uture growth. The Global MetroMonitor examines data on economic output and employment in 150o the worlds largest metropolitan economies, located in 53 countries,

rom 1993 to 2010 and makes the ollowing ndings:

The Global Economy is Led by MetropolitanEconomiesThe 150 metropolitan economies pro led in the Global MetroMonitorexhibit highly diverse stages o development. Their per capita measureso Gross Value Added (GVA) range widely, rom under $1,000 inHyderabad and Kolkata, India, to roughly $70,000 in San Jose, U.S.A.and Zurich, Switzerland.

What is consistent about these metropolitan areas, however, is theirunction as locations or high-value economic activity in their respectivenations and world regions. Nearly our in ve boast average incomes (asproxied by per capita GVA) that exceed averages or their nations. Thisis particularly true in rapidly emerging areas o Eastern Europe and Asia, where major metro incomes exceed those or nations by average marginso at least 90 percent.

As a result, these metro areas punch above their weight in national and global economic output. In 2007, they accounted or just under 12percent o global population, but generated approximately 46 percent o world GDP.

The Global Downturn and Recovery are Acceleratinga Shi t in Growth Toward Lower-IncomeMetropolitan Areas in Asia and Latin America Virtually no place completely escaped the e ects o the global nancial crisis and ensuing economic downturn in the late 2000s. Yet impactsacross the 150 global metropolitan areas were highly uneven, asillustrated through the Global MetroMonitors ocus on the combinedincome and employment per ormance o these places during three

distinct economic periods rom the past two decades:

Pre-RecessionBetween 1993 and 2007, roughly hal o the metro areas that achievedthe strongest growth in GVA per capita and employment were located inrising nations o Asia, Latin America, and the Middle East that bene tedrom new heights o global economic integration. Metro areas such asShenzhen, China and Bangalore, India roughly tripled their income, andemployment in Singapore and Belo Horizonte, Brazil grew more than hal over the 14-year period.

Portions o the worlds more industrialized regions, including the UnitedStates and Europe, also registered strong metro per ormers duringthat time. Eastern European metros such as So a and Krakow, as well as Dublin and in Western Europe, achieved rapid growth in income.In the United States, Las Vegas, Phoenix, and Austin posted majoremployment gains over the same period. Overall, however, U.S. metroson average ranked slightly behind their European counterparts, and well behind their counterparts in other regions o the world, on economic

per ormance through much o the 1990s and early- to mid-2000s.

RecessionThe negative impact o the global economic downturn, commencingin 2008, was widespread among the 150 metro areas. Roughly sevenin eight lost either employment or income in at least one year between20072008 and 20092010.

But or several global metropolitan areas, the late 2000s marked moreo a temporary slowdown than a Great Recession. The top-ranked

metro per ormers or the most part experienced no decline in eitheremployment or income rom 2007 to 2010. Fully 28 o the 30 top-rankedmetros during that period were located outside o the United Statesand Europe, with China accounting or the top ve. Australian metros(Melbourne, Brisbane, and Sydney) registered strong per ormance, dueto their important economic linkages with stable East Asian economies.Latin American metros proved resilient as well, with Lima, Buenos Aires,Bogot and three Brazilian metros ranking among the top 30.

ExECuTIVE suMMARYExe ive smm y


5/52

5

By contrast, many o the metros in the United States and Europe thatfew highest be ore the recession experienced tremendous alls. Dublin,Madrid, and the three Baltic capitals (Riga, Tallinn, and Vilnius), along with Las Vegas and Riverside (Cali ornia) in the United States, movedrom the top 30 spots pre-recession to the bottom 30 spots during therecession. These regions exhibited signi cant asset bubbles in the 2000s,as evidenced by the all in home prices in their respective nations inrecent years. Overall, the Great Recession appeared to hit U.S. metroshardest, while it improved the relative position o metros outside the

United States and Europe.

Recovery The most recent year, rom 2009 to 2010, appears to have urtherstrengthened the relative economic standing o metro areas in therising nations o Asia, Latin America, and the Middle East. O the top 30ranked metros in this period, a diverse group o 29 was located outsidethe United States and Europe. China and India alone accounted or 10,Latin America registered seven, and the Middle East and North A ricarecorded our. Most o these metros posted annual growth rates o at least2.5 percent in employment, and 5 percent in income, in the rst year o worldwide recovery.

While the recession hit U.S. metros harder than their Europeancounterparts, the recovery seems slower to take hold in European than American metros. Metros along Europes western, eastern, and northernperipheries, rom Porto and Valencia, to Thessaloniki and So a, toHelsinki and Stockholm, anchor the bottom 30 economic per ormersrom 2009 to 2010. Meanwhile, several U.S. metros that su ered severeeconomic declines during the recession, such as Detroit and Cleveland,posted signi cant rebounds in their rankings on the strength o robustincome growth, even as metros such as Atlanta and Las Vegas await astronger recovery.

The upshot: The past two decades have seen lower-income metro areas inthe global East and South close the gap with higher-income metros inEurope and the United States, and the worldwide economic upheaval hasonly accelerated the shi t in growth toward metros in those rising regionso the world.

National Context and Industrial Patterns ShapeMetro Per ormanceBeyond indicating economic opportunities within broad world regionsand di erent stages o development, metros recent per ormance alsorefects intrinsic actors such as their industrial base, and the impact o national scal, monetary, and trade policies.

First, the presence and magnitude o certain industries within metro areasrelated strongly to economic per ormance, though these di ered by period

and world region. Metros with high shares o their output in constructionper ormed much better than average in the pre-recession period,particularly in the United States, but much worse than average in therecovery, particularly in Western Europe and other high-income regions.Be ore the recession, an energy and manu acturing ocus was associated with strong per ormance o lower-income metro areas, particularly inChina and the Middle East, and weaker per ormance o U.S. metros.Higher-income nancial and business services centers in the Asia/Paci cand North American regions per ormed less well than others in the pre-recession and recession periods. And high output in non-market services,such as government, health, and education, was a boon or European and American metros during the recession, signaling that those industriesremained relatively healthy amid market turmoil.

Second, national context does matter. In any given period, roughly hal to three-quarters o metro economic per ormance was associated withrespective national economic per ormance. For example, the analysesabove point to distinct economic dynamics among U.S. metros that made

their recession generally deeper than in other world regions, but thatmay also account or the stronger rebound some U.S. metros are postingcompared to their European counterparts. Examining national economiesalone, however, overlooks the important variations in metro per ormancethat separated nearby metros such as Leipzig (#77) and Berlin (#144) inthe pre-recession period; Abu Dhabi (#16) and Dubai (#97) during therecession period; and Cleveland (#49) and Bu alo (#120) in the recovery period.

As global metro areas emerge rom the shadow o the Great Recession, they

also nd themselves in markedly di erent places along their own growthtrajectories. Many in Asia and Latin America were scarcely a ected by therecession at all, or have posted a ull recovery. Several in the United Statesand other high-income regions have rebounded to their prior employmentor income level, but not yet both. About hal o the 150 continue to loseground on one o the key measures, in most cases employment, and thebulk o these metros are in Western Europe and the United States. Anda small hand ul o metros, most in Europe, continued to decline in bothemployment and income through 2010 as the recession raged on.

The Global MetroMonitor thus portrays a world economy whose continuedtransition will be driven in large part by the distinct experiences o itspower ul network o major metropolitan economies. As metropolitanleaders worldwide con ront the challenges and opportunities thataccompany continued global economic integration, and many seek new growth models to replace old ones, the shi ting metro map points towardan emerging array o productive metro-based economic relationships thatcould drive regional and national prosperity in the decades to come.

E x E C u T I V E s u M M A R Y


6/52


7/52

7

Oklahoma City New YorkMiamiPragueToulouse

ColumbusChicagoMilwaukeeDenverOrlandoMarseilleNew Orleans ViennaDusseldor Jacksonville Auckland VancouverLyonHamburgSacramentoParisLille

OsloLeipzigCopenhagenZurichPortlandTurinCologneBudapest

RomeRiversideStuttgartMunichMilanBerlinLisbonFrank urt

LjubljanaEdinburghLos AngelesCape Town Vilnius

Glasgow Bu alo

Hart ordKansas City LondonHelsinkiSan FranciscoRotterdamPhiladelphiaRochesterPittsburghStockholmBirmingham AmsterdamBrusselsManchester

NaplesSo aIndianapolisTallinn AtlantaPorto AthensMadrid

JohannesburgRiga ValenciaLas VegasThessalonikiBarcelonaDubaiDublin

7677787980

8182838485868788899091929394959697

9899100101102103104105

106107108109110111112113

114115116117118

119120

121122123124125126127128129130131132133134

135136137138139140141142

143144145146147148149150

13290663352

111115141644084

14812212680556393

12750

114104

10677

10114245

13713157

1162813410212014496

130

4688

1225044

10412713513211943

1074985

12555956361

1155154

34659958

1391058186

6713011376

111565383

8162827513

86117

1381246859

11212511913612887

12111397

108

14010992156

1451922

54114914513526

10670

13764

147

7969

10159

100138133

71787741

108117746296

9480

1261491368747

134

116148140128

5714597

144

PRE-RECEssION1993-2007

RECEssIONYEAR Of MINIMuM GROwTH 2007-2010

RECOVERY2009-2010

PRE-RECEssION1993-2007

RECEssIONYEAR Of MINIMuM GROwTH 2007-2010

RECOVERY2009-2010

uNITEd sTATEs EAsTERN EuROPE OTHER LOwER-INCOMEwEsTERN EuROPE OTHER HIGHER-INCOME

E x E C u T I V E s u M M A R Y


8/52


be ore, during, and a ter the Great Recession provides importantevidence on emerging shi ts in the location o global economic resilienceand uture growth, and the underlying actors that might propel andsustain that growth.

Building on the Brookings Metropolitan Policy Programs MetroMonitor, which tracks the economic per ormance o U.S. metro areas over thecourse o the recession and recovery, the Brookings Institution andLSE Cities at the London School o Economics partnered to produce

this Global MetroMonitor, which examines data on economic outputand employment in 150 o the worlds largest metropolitan economies,located in 53 countries on six continents. 3 They include the 50 largestmetropolitan economies in the United States; the national capital economies o 25 European countries plus the 25 largest other metroeconomies in Europe; and 50 o the largest metro economies in otherregions o the world, including representatives in Asia, Australia/New Zealand, Latin America, Canada, and A rica.

The Global MetroMonitor is by no means the only report to ever study a cross-section o global metropolitan areas. Organizations worldwideproduce various world city rankings on a regular basis. For instance, in2010, Foreign Policy magazine published a Global Cities Index, ranking65 large metro areas worldwide on a variety o economic, social, cultural,and political dimensions. 4 The Partnership or New York City publishesan annual report examining how 21 global cities per orm as centers o business opportunity. 5 Brookings has published research in the pastexamining the position o U.S. cities and others in a world city networkbased on the location o multinational advanced-services rms. And

Boston Consulting Group recently published a report that classi ed citiesin what it calls emerging markets, by their role in the international andregional economies. 6 This report di ers rom those in ocusing purely on key economic outcomes or global metro areas, not just the cities attheir core, and examining a range o actors that may help explain theirrecent economic per ormance. The Global Urban Competitiveness Report,published by a team o Chinese and American researchers, o ers acomprehensive analysis o economic indicators or 500 metros worldwide,but stops short o analyzing the Great Recession and its a termath or

these global centers. 7 In these ways, the Global MetroMonitor makes aunique and important contribution to understanding these 150 metroareas contemporary economic per ormance and position.

Not surprisingly, these metro areas, like their respective nations, exhibithighly diverse stages o development. Their per capita Gross Value Added(GVA), a measure o income per person, ranged widely in 2007, romunder $1,000 in Hyderabad and Kolkata, India, to roughly $70,000 in SanJose, U.S.A. and Zurich, Switzerland.

0.5

1.0

1.5

2.0

0

WesternEurope(n=39)

UnitedStates(n=50)

OtherLower-Income(n=33)

OtherHigher-Income(n=17)

EasternEurope(n=12)

1. INTROduCTIONi d i

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data.

Figure 1-1. Metro Incomes Exceed National Incomes Average Ratio o Metro to National GVA per Capita by World Region, 2007

The global nancial crisis o the late 2000s precipitated an economicdownturn o such magnitude and reach that many now re er to theperiod as the Great Recession. According to the International Monetary Fund, global economic output, which had grown at an annual rate o 3.2percent rom 1993 to 2007, actually shrank by 2 percent rom 2008 to2009.1 A precarious economic recovery is now underway.

Aggregate views o the global economy, however, mask the distinctexperiences o its most important hubsmajor metropolitan areas. These

economically integrated collections o cities and their surrounding areasare centers o high-value economic activity in their respective nationsand worldwide. They play di erent but complementary economic rolesin national and international contexts, by virtue o location, stage o development, industrial base, demographics, and local and national policies that set the conditions or economic per ormance. 2

Because metropolitan areas orm the undamental bases or national and international economies, understanding their relative positioning


9/52


10/52

1 0 G L O B A L M E T R O M O N I T O R

The Global MetroMonitor assesses the economic per ormance o 150metropolitan areas worldwide. It builds on the MetroMonitor, a quarterly Brookings publication ocused on the economic per ormance o the 100largest U.S. metropolitan areas during the recent recession and ongoingrecovery. As the report demonstrates, these global metro areas, whichinclude cities and surrounding rural and urban areas that together ormintegrated regional economies, account or signi cant shares o national and global output and jobs, and represent an important lens through which to view the uneven trajectory o economic growth worldwide.

Selection and Defnition o Metropolitan AreasThe Global MetroMonitor evaluates 150 o the largest metro economies worldwide, as measured by their total economic output, while it alsoportrays metro economic per ormance in a broad cross-section o worldregions (Figure 2-1). The United States and Europe are each represented with 50 metros. 9 An additional 50 metro areas were selected rom otherregions o the world and include 28 in Asia and Australia/New Zealand,

14 in North and South America, and eight in A rica and the Middle East.In each o these three world regions, the priority was to select the largestmetro economies or which complete, comparable data were available. 10

The 50 U.S. metropolitan areas represent the largest regional economies inthe U.S., as measured by gross metropolitan product (GMP) in 2008 (themost recent year or which public data are available). In the United States,metro areas are de ned by the ederal O ce o Management and Budget(OMB) to include one or more large urban cores plus outlying areas thathave social and economic linkages to the urban core(s). The 50 U.S. metroareas in this report vary in size rom just more than 900,000 residents inthe Bridgeport-Stam ord-Norwalk, CT metropolitan area to more than 19million residents in the New York-Northern New JerseyLong Island, NY-NJ-PA metropolitan area, and their average size is 3.4 million residents.

In Europe, there is no o cially accepted metropolitan area standard, asthere is with the OMB standard used in the United States. Among existingde nitions o European metropolitan areas, the approach developed by ESPON was deemed most appropriate.11 It is based on the aggregation o E.U. Tier 3 (NUTS 3) administrative regions which range rom 150,000 to800,000 inhabitants. The Nomenclature o Territorial Units o Statistics(NUTS) is developed by Eurostat based on the administrative divisionso European nations and the NUTS regions are thus comparable acrossEuropean countries. Our sample o European metros includes 25 capital

metro areas and 25 other large metropolitan areas based on population sizeand area. In terms o population, the largest European metropolitan area isLondon with 14.8 million residents, the smallest is Ljubljana with just overa hal million inhabitants, and the average is 3.1 million inhabitants.

The 50 metropolitan areas outside the United States and Europe wereselected with respect to both size and geographic spread in Asia,Oceania, A rica, Latin America, and non-U.S. North America. Here, thenal choice o metro areas was heavily dependent on data availability

and comparability, and the lack o reliable economic data explains theun ortunate absence o some o the worlds largest and astest growingmetropolitan areas including Dhaka, Karachi, Kinshasa, and Lagos.Particularly in A rica and Asia, not all countries have created administrativeareas or at least statistical boundaries that yield reliable economicestimates or metropolitan areas. In certain countries or areas where thisproblem exists (such as India), data rom the administrative city are usedi only a small proportion o the metropolitan areas population is thereby discarded. 12 Similarly, the wider province or region is chosen as a proxy or

the metropolitan area i it is not much larger in terms o population. Theaverage population o this nal set o 50 metropolitan areas is just over 10million, given the much larger average size o Asian metro areas. 13

Data SourcesTo assess the economic per ormance o 150 metropolitan areas, theGlobal MetroMonitor ocuses on the ollowing baseline data: Gross Value Added (GVA), employment, and population (which allows us to assessGVA per capita) rom 1993 to 2010. In addition, GVA and employment are

broken down by major industry sector (see below). Data availability andcomparability precluded expanding the investigation to other economicindicators o interest, such as house prices and unemployment rates.

There are two major technical considerations with respect to the data inthis analysis. The rst s tems rom this reports ocus on the recent impacto the recession and the resulting need to analyze data or 2008, 2009,and 2010 that are not yet available through most national statistical o ces. Three data providers supplied these estimates: Moodys Economy.com or the United States, Cambridge Econometrics or Europe, andOx ord Economics or the rest o the world. 14

By its very nature, relying on orecasted data introduces a measureo uncertainty into any analysis. While the degree o uncertainty involved cannot be known, it is ultimately determined by the quality

2. dATA ANd METHOdsD d Me d

9 The United States and the European Union toget her account or roughly 50 percent o world G.D.P. (IMF WorldEconomic Outlook Database, October 2010). Thus, their metro areas are somewhat over-represented in this analysis, which refects in part the greater availability o comparable metropolitan data within these regions versus ones outsideEurope and the United States. Regional analysis in the report treats Moscow as an Eastern European metro, and Istanbul (in the E.U. candidate country o Turkey) as a lower-income metro.

10 The metros eatured in Global MetroMonitor include , or example, 87 ranked among the 100 largest by GDP in 2008in John Hawksworth, Thomas Hoehn, and Anmol Tiwari, Which Are the Largest City Economies in the World and How Might This Change by 2025? PriceWaterhouseCoopers UK Economic Outlook, November 2009. That list excludes anumber o U.S. metro areas eatured here, such as Sacramento and Kansas City, that by our estimates would also rankamong the 100 largest worldwide.

11 ESPON is the European Observation Network or Territorial Development and Cohesion.


11/52

1 1

o the orecasting model used by each data provider. Furthermore,because each data provider uses a di erent model, our comparisonsmay be a ected by di erences in model only. While each provider usedthe best available data to model the recent economic per ormance o these metropolitan areas, the nature o the exercise demands somecaution when interpreting the results. Findings regarding metropolitanper ormance that are based on estimates or 2009 and 2010 should thusbe treated as preliminary in nature, and subject to urther revision asnational statistical agencies compile and publish o cial metro-level estimates in the coming year(s).

The second major technical consideration results rom the international scope o our analysis. While our data are conceptually consistent acrosscountries, we are limited to the data collection and statistical methodsutilized by each countrys s tatistical agencies. Consequently eachindicator may be calculated slightly di erently on a country-by-country basis.

For U.S. metropolitan areas, employment data come rom the U.S.Bureau o Labor Statistics Local Area Unemployment Statistics (LAUS)program. LAUS data are model-based, relying on data rom the CurrentPopulation Survey (CPS, the source or our U.S. level employmentestimates) as the primary input. Employment is measured as o July in each year. 15 Population data come rom the U.S. Census BureausPopulation Estimates Program, which are model-based estimates that rely on decennial census data as primary inputs; the Census Bureau measurespopulation as o July 1st o each year. 16 Moodys Economy.com suppliesthe GVA data, which are derived rom the Bureau o Economic Analysis

(BEA) gross domestic product by state estimates. They parcel out state-level GDP data to counties on an industry-by-industry basis according toeach countys share o state employment and sum the resulting county totals to arrive at a metropolitan total. 17 The last year o available datarom BEA was 2008 and so data or 2009 and 2010 have been orecasted.Moodys Economy.com also provides GVA by industry data, classi edaccording to the North American Industry Classi cation Standard(NAICS).

For European metros, Cambridge Econometrics relies primarily onthe Eurostat REGIO database or underlying economic data, and theInternational Labour Organization (ILO) or population and labor

12 The metro population used in these cases was the population o the corresponding Urban Agglomeration, as publis-hed in the UNs World Urbanisation Prospects Database, 2009 revision.13 Tokyo, in particular, alone increases the average size o Asian metros by 680,000 inhabitants, given its populationo 35 million.14 Brookings subscribes to a Moodys Economy.com metropolitan economic database or the U.S. MetroMonitor andrelated e orts; LSE subscribes to a Cambridge Econometrics database or similar data in the European context.Brookings and LSE jointly contracted with Ox ord Economics to supply data on 50 metro areas outside Europe and theUnited States.15 At the time o this analysis, July was the most recent month or which revised employment data were available romBLS.

orce data. The orecasting model used by CE is called the EuropeanRegional Economic Model (EUREGM), which has a medium-term ocusand tries to capture a variety o actors that can lead to both regional divergence and convergence. A primary input to this model is a metric o economic potential which can be viewed as closeness to markets andto suppliers, with a high economic potential associated with enhancedproduction, supply and distribution conditions. Forecasts or GVA rely primarily on industry structure, population density, and economicpotential to predict uture output levels. Forecasts or employmentproceed in a similar ashion, but depend on estimates o GVA by sectorand assumptions about technological trends. Cambridge Econometricsprovides GVA by industry data, classi ed according to the ISIC Rev.3.1standard into the ollowing categories: Agriculture (A), Energy and Manu acturing (C,D,E), Construction (F), Distribution, Hotel &Restaurants, Transport, Storage and Communications (G,H,I), Financial Intermediation, Real Estate, Renting and Business Activities (J,K) andNon-Market Services (L,M,N,O,P).18

Ox ord Economics data are based on a wide variety o sources, includingnational statistical agencies or other data providers where available. Where data were not available, Ox ord Economics relies upon its national-level orecasts provided by the Ox ord Economics Global EconomicModel to provide a orecast based on the historical relationship betweenmetropolitan area industry data and the national level gure. Accordingto Ox ord Economics, the orecast or each metropolitan area isessentially shaped by how strong demand is likely to be or each industry in that location.

Time PeriodsThree time periods between 1993 and 2010 were identi ed in order tomeasure the per ormance o all metropolitan areas in three distincteconomic contexts, which the report re ers to as pre-recession, recession,and recovery.

The pre-recession period gives an indication o the long term,underlying economic trend each metropolitan area ollowed priorto the recession. It urther serves as the baseline period rom which to assess the degree to which metros were a ected by thecrisis. For the pre-recession period, a xed time rame rom 1993 to

16 The most recent data available were or July 1, 2009. To estimate the population in 2010, the annual average growthrate rom 2007 to 2009 was applied to the 2009 estimate.17 In the United States, metropolitan areas are aggregations o counties, which are local levels o government below thestate level, but typically above the municipal level.18 In the United States, industries are classi ed according to the North American Industr y Classi cation System(NAICS) while or much o the rest o the world industries are classi ed according to the International StandardIndustrial Classi cation (ISIC). Detailed NAICS-based industry data rom Moodys Economy.com were used toapproximate ISIC sectors. At the national level or most sectors, this strategy works well; the largest error is associated with our NAICS-based approximation o the ISIC construction sector, which is an estimated 10 percent larger than it

otherwise would be i the data were originally de ned using the ISIC scheme.

d A T A A N d M E T H O d s


12/52


2007 was used, based on the availability o data across all metrosextending back to 1993, and the star t o the recession in the UnitedStates in December 2007. 19 There ore, 2007 is treated as the last year in which all countries worldwide were not yet a ected by theGreat Recession, though clearly some metros, countries, and worldregions su ered recessions o their own during this period.

The recession period measures the impact o the recent worldwideeconomic downturn on each metro area. For this period, the year o minimum annual growth rate ( or GVA per capita and employment)between 2007 and 2010 was identi ed or each metro. This method

takes into account di erences in when the recession a ectedeach metro area or world region. Selecting the minimum one-yeargrowth rate means comparing exactly the same thing everywhere,and avoids averaging out the recession drop across the years thatpreceded or ollowed it (as usual peak-to-trough calculations would).20 This method also makes it easier to account or the many metropolitan areas that did not experience actual declines in GVA per capita and/or employment over the recession period, but whosegrowth rates still ell compared to their long-term average.

The recovery period re ers to 2009 to 2010, a period during whichmost o the 150 metro areas analyzed experienced growth in the wake o a downturn, or accelerated growth relative to the recessionperiod. Assessing both recession and recovery periods then allowsor some preliminary conclusions on how the recent recession may have changed the global metro economic landscape. Using thisxed period urther helps identi y metropolitan areas that are still inrecession, and how well metropolitan areas are recovering comparedto their respective nations.

The terms pre-recession, recession, and recovery thus re er to thecondition o the broader global economy during each o these periods,and not necessarily to the experience o all metropolitan areas studiedhere. For instance, some American metropolitan areas such as Detroit were losing jobs and output well be ore the onset o the worldwidedownturn. Large employment losses came a bit later to certain parts o Europe than to the rest o the world, extending into 20092010. Andas the analysis explains, several metro areas in Asia and Latin America(and two Polish metro areas) experienced no decline in either output

or employment in any year rom 2007 to 2010. In this way, the Global MetroMonitor provides a snapshot o metro per ormance at key stagesleading up to, during, and a ter the global economic crisis, but does notattempt to measure the speci c e ects o the crisis on each metro area.

Indicators, Scoring and RanksThe report measures the economic per ormance o metropolitan areasusing two main indicators: the annual growth rate o real GVA per capita;and the annual growth rate o employment. There ore, this study isconcerned with the dynamics o metropolitan economies, and how metros compare in terms o their growth per ormance and potential,rather than their absolute per ormance levels. 21 These two indicatorsrefect the importance that people and policy makers attach to achievingrising incomes and standards o living (GVA per capita), and generating widespread labor market opportunity (employment). GVA per capita,

unlike absolute GVA, controls or contributions to GVA that ollow rom population growth alone (especially over longer time periods).Throughout the report, we re er to GVA per capita as income, andchange in the measure as income growth.

In order to create a ranking o metropolitan areas in each o the threeperiods, the Global MetroMonitor combines calculations o each metroareas per ormance on income and employment growth, giving equal weight to each sub-measure. In order to combine these annual growthrates, each is s tandardized using the inter-decile range standardizationmethod. This method compares each value o a variable (X i) to themedian (X med), which is then divided by the distance between the value o that variable at the 90th percentile o the distribution (X 90) and the 10thpercentile (X 10):

This method was judged more appropriate or these data thanZ-score standardization, which compares each value o a variableto the mean and divides their di erence by the standard deviation,as they do not ollow a normal distribution. It was also pre erred torange standardization (which compares each value o a variable tothe minimum and divides their residual by the distance between theminimum and the maximum) because o the sensitivity o this lattermethod to outliers. Inter-decile range standardization helps to minimizethe infuence o outliers by using the 90th and the 10th percentile valuesinstead o the minimum and maximum values, and best refects the non-normal distribution o metro economic growth rates.

Standardized scores are obtained by applying the inter-decile rangestandardization to annual income and employment growth rates, thenadding those values together to yield a nal score or each period oreach metro area. That score is used to rank the 150 metropolitan areasaccording to their per ormance during each o the three periods.

19 The start o the period in 1993 also refects the end o an initial period o volatility in Europe associated with thecollapse o the Soviet regime and transition o the ormer Eastern Bloc countries.20 Note that we did attempt to calculate peak-to-trough/slowed growth measures or metro areas during the recessionperiod, and that the ultimate ranking o metro areas by this method was very similar to that achieved using the

minimum annual growth rate method.

Standardized score =xi - xmedx90 - x10

21 Employment growth does not take into account changes in the size o the labor orce, in the way that GVA per capitaaccounts or changes in population. This choice refects the lack o comparable data across metro areas on the size o thelabor orce over time, which would enable one to calculate the employment rate.


13/52


14/52


Figure 2-1. 150 Metropolitan Regions

20 m

5 m

So PauloRio de Janeiro

New York

Lima

Mexico City

Chicago

Los Angeles Dallas

Bogot

Buenos Aires

Miami

Santiago

Atlanta

Toronto

Houston

BostonMontreal

Brasilia

Seattle

WashingtonSan Francisco

Portland

Vancouver

The Global MetroMonitor tracks the economic per ormanceo 150 major global metropolitan areas, integratedcollections o cities and surrounding areas that ormunctional regional economies. The 150 metro areas includethe 50 largest economies in the United States, 25 national capitals plus another 25 large metros in Europe, and 50other large metro areas in North and South America, A rica,and the Asia/Paci c region.

Metropolitan Population2010 estimates

Source: Ox ord Economics, Cambridge Econometrics data, and U.S. Census Bureau.


15/52

1 5

TokyoSeoul

Cairo

Jakarta

Osaka

Mumbai

Beijing

Tianjin

Paris

Delhi Shanghai

Kolkata

London

Manila

Istanbul

Moscow

Bangkok

Taipei

ChennaiHyderabad

Bangalore

Hong Kong

Madrid

Guangzhou

Rome

Riyadh

Berlin

Sydney

Athens

Johannesburg

SingaporeKuala Lumpur

Lisbon

Budapest

Melbourne

Dubai

Abu Dhabi

Cape Town

Warsaw

Brisbane

Dublin

Oslo

Bucharest

Stockholm

Auckland

Barcelona

Shenzhen

d A T A A N d M E T H O d s


16/52


The period rom 1993 to 2007 was one o tremendous growth and changein the global economy. Political shi ts, technological trans ormation,reduced barriers to trade, and the emergence o a highly integratedglobal nancial system greatly broadened participation in thateconomy. As the chie hubs o national and international commerceand governance, major metropolitan areas stood at the ore ront o those trends. Some metros witnessed unprecedented levels o growththroughout the 14-year period. For others, these trends challenged theireconomic identity and prosperity as never be ore. This section exploresthe longer-run metro economic backdrop against which the GreatRecession and the recovery thus ar have occurred.

Regional Summary In the decade and a hal be ore the crisis, the 150 metro areas posted amedian employment growth rate o 1.4 percent annually, and a somewhataster growth rate o 2.3 percent in income. Among the 150, sevenregistered a loss in employment between 1993 and 2007, and income

dipped in two. For the most part, however, these metro areas becamelarger and richer in their economic pro le in the long lead-up to the GreatRecession.

The metro growth spectrum was nonetheless wide during this period.The highest per orming metropolitan areas rom 1993 to 2007 achievedtypical employment growth o 3 percent per year, and typical incomegrowth o over 5 percent per year (Figure 3-4). By contrast, employmentrose only 0.5 percent annually, and income only 1.3 percent annually, inthe lowest per orming metro areas over that time.

The top and bottom metro per ormers also refected important economicdistinctions across and within world regions (Figure 3-5). In general, twotypes o metro areas occupied the top spots. The rst included rapidly emerging Asian, Middle Eastern, and Eastern European metro areas thatbene ted rom recent integration into the expanding world economy,in many cases aided by national political and economic policy re orms. All ve Chinese metro areas in the dataset, or instance, ranked amongthe top per ormers, as did our o six Indian metros. Some o thesemetros achieved astonishing rates o growth. Guangzhous economy,or instance, was roughly our times larger per capita in 2007 as in 1993,and Shenzhen more than tripled its employment during that time.Meanwhile, eight o 12 Eastern European metro areas posted scoresamong the 30 highest. Most o those metros experienced relatively anemic employment growth, but underwent sweeping industrial trans ormation that boosted their incomes by rates o 6 percent annually or more.

3. PRE-RECEssION PERIOdpe-ee iThe second type o high-per orming metro in the pre-recession periodcould be ound in portions o the United States, Western Europe, andother high-income areas o the globe. Four U.S. metros in the South and West (Austin, Las Vegas, Phoenix, Riverside) joined Athens, Dublin, andMadrid in Europe, Abu Dhabi and Dubai in the United Arab Emirates,and Brisbane, Australia among the top 30 per ormers. 30 These placesattracted both robust population and employment growth rom 1993 to2007; Dublins income more than doubled in that time.

Like those at the top o the list, the weakest 30 per ormers pre-recessionell into two general categories. The majority were older industrial regions o the United States and Central/Southern Europe. Low-per orming U.S. metros were exempli ed by manu acturing centers suchas Detroit, Cleveland, St. Louis, and Pittsburgh. In Europe, Stuttgart,Turin, Naples, and Porto exhibited similarly weak per ormance; sixo the eight German metros in the dataset ranked among the bottom30.31 A smaller second category o Asian metros struggled during thepre-recession period, too. Weak per ormance by the three Japanesemetros (Tokyo, Nagoya, Osaka) refected their entire countrys decade o economic stagnation, and the late 1990s nancial crisis in Southeast Asiadragged Jakarta and Bangkok toward the bottom o the rankings. With aew exceptions, metros at the bottom o the list experienced increases inboth employment and income rom 1993 to 2007, but growth rates tendedto be anemic, typically 0.5 percent annually or employment and 1.3percent annually or income.

Across the complete set o 150 metro areas, those metros in emergingeconomies out-per ormed others by signi cant margins during the

pre-recession period. Eastern European metros achieved an averageper ormance rank o 29, ollowed by lower-income metros outside o Europe and the United States at 49. U.S. and Western European metrosposted similarly low average rankings, at 91 and 96, respectively (Figure3-1). Notwithstanding these di erences, the period was marked by ameasure o regional diversity among both high and low metro per ormers.

30 As the next section indicates, many o these high per ormer s outside Asia and Latin America experienced house-price bubbles in the lead-up to the recession.

31 The relatively weak per ormance o German metros could be explained in part by the act t hat the analysis period(1993 to 2007) began soon a ter the uni cation boom ended, and the price or restructuring the country (by, e.g.,

allowing or wage convergence between East and West) was paid in somewhat lower rates o economic growth.


17/52

1 7

Metro Per ormance FactorsBeyond regional location, other actors may help explain the disparateeconomic per ormance o global metro areas in the 14 years precedingthe Great Recession.

Population Size and Growth A metro areas population level did not appear to relate to its per ormancelevel in the pre-recession period. No signi cant association existedbetween population and overall per ormance, either or all 150 metroareas or or metro areas within their respective regional groupings. 32 Metros with aster growing populations were stronger per ormers overall,although this was largely a unction o their s tronger employmentgrowth, which naturally accompanies population growth. 33

60

40

20

0

100

80

WesternEurope(n=39)

UnitedStates(n=50)



EasternEurope(n=12)

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data

Figure 3-1. Eastern European Metros Achieved Higher Per ormance Rankings than Other Metros in the Pre-Recession Period Average Rank out o 150

IncomeIn general, lower-income metro areas per ormed better than middle- andhigher-income metro areas. This ollows rom the regional ndings, with Asian, Latin American, Middle Eastern, and Eastern Europeanmetros achieving higher per ormance rankings than their counterpartselsewhere. Metro income levels related more strongly to long-run metroincome (GVA per capita) growth than employment growth, suggestingthat the 1993 to 2007 period was one in which most lower-income metrosnarrowed the wealth gap with middle- and higher-income metros.

Within world regional categories, however, the relationship betweenmetro income and overall economic per ormance was limited. Only inEastern Europe was the relationship statistically signi cant, with lower-income capitals in the Baltic states and Bulgaria generally outpacing wealthier (yet still success ul neighbors) such as Warsaw and Ljubljana.In the United States, higher-income metro areas outpaced others inincome growth, but not employment, pointing to the emergence o deeper regional income inequalities throughout the nation during the14-year period.

National Per ormance Across all global metro areas studied rom 1993 to 2007, the typical metroexperienced slightly aster employment growth than its correspondingnation, and comparable GVA per capita growth (Figure 3-2). Therelationship between metro and national per ormance di ered amongregions, however. In particular, metro areas in Eastern European nations well outpaced national averages on both indicators. So a, or instance,achieved annual growth o 3.3 percent in employment, and 6.2 percentin GVA per capita, compared to Bulgarian averages o 0.5 percent and 3.5percent, respectively. Outside o Europe and the United States, the typical lower-income metro posted slightly greater employment gains than itsnation, while the typical higher-income metro posted slightly smaller GVA per capita gains.

These typical experiences, o course, do not capture the underlying variation in per ormance among metro areas that exist within the samenation. For instance, Munich outpaced German national averages onboth employment and GVA per capita growth, while Stuttgart laggedthe nation on both measures. Given its sheer size, the United Statesexhibited a wide range o metro experience, sometimes even within itsown states. In Tennessee, or example, Nashville exceeded U.S. averageson employment and income growth, while Memphis ell behind.

32 One exception was that larger metros in Eastern Europe tended to per orm worse in the recession than smaller ones.33 Western Europe was the only region in which population growth and income growth at the metro level were signi -cantly related, but this seems largely attributable to Dublin, which experienced 23 percent population growth and 5.9percent annual GVA per capita growth rom 1993 to 2007.

P R E - R E C E s s I O N P E R I O d


18/52


Still, national rates o economic growth do appear to set an importantplat orm or metro-level per ormance. Controlling or the share o national output that each metropolitan area contributes, the averagerate o employment growth at the national level explained a little underhal o the variation in metro employment growth rom 1993 to 2007. Theaverage rate o income growth nationally explained even more o theunderlying metro variation across the period, about three-quarters. 34 Inshort, metro economic per ormance in the lead-up to the recession wasnot independent o national economic per ormance.

Industrial StructureThe contribution o certain economic sectors to overall output was, insome regions, associated with stronger or weaker metro per ormancein the pre-recession period. Because, or example, manu acturingindustries in China are quite distinct rom those in central Europe andthe U.S. Midwest, these sectoral relationships are examined within thespeci c contexts o world regions (Figure 3-3):

In Eastern Europe, metros with large shares o output in logistics,communications, and hospitality per ormed better, perhapsrefecting the rapid growth o trade and tourism in the region overthe period. At the same time, per ormance was weaker in metros with a signi cant ocus in non-market services such as government,health, and education, including Bratislava and Budapest

In Western Europe, metros with relatively high levels o constructionoutput experienced more rapid economic growth pre-recession,perhaps refecting the inward fow o population and investmentto regions including Thessaloniki, Dublin, Toulouse, Valencia, andMadrid. This was also the case in high-income metros outside theUnited States and Europe including Brisbane, Sydney, and Seoul

A similar dynamic prevailed in the United States, where rapidly growing Western metros such as Las Vegas, Riverside, andPhoenix had much o their pre-recession output concentratedin the construction sector. On the other hand, metro areas witha signi cant manu acturing presence underper ormed others,refecting long-run employment struggles o older industrial areasin portions o the U.S. Northeast and Midwest regions

The reverse was true or lower-income energy and manu acturing-specialized metros, which outper ormed their counterparts largely on the strength o Chinese metros rapid emergence in the global trade o manu actured goods, and expanding utility sectors incountries with rapidly developing middle-class consumers

2

-0.5

0

0.5

1

1.5

Annual EmploymentGrowth Difference

Annual GVA per CapitaGrowth Difference

WesternEurope(n=39)

UnitedStates(n=50)



All regions(n=150)

EasternEurope(n=12)

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data

Figure 3-2. Metro and Nations Per ormed Similarly in Most Regions in thePre-Recession PeriodMedian Di erence Between Metro and National Annual Employment and GVA per CapitaGrowth Rate by Region, 1993-2007

Figure 3-3. Construction and Logistics-Focused Areas Per ormed WellPre-Recession

All Metros(n=144)*

+

+

Western Europe(n=39)

Other Higher-Income(n=12)*

Eastern Europe(n=12)

Other Lower-Income(n=31)*

United States(n=50)

+ +

CONSTRUCTION

LOGISTICS,LEISURE,COMMUNICATIONS

ENERGY ANDMANUFACTURING

FINANCIAL ANDBUSINESS SERVICES

NON-MARKETSERVICES

+

+ +

+ +

+ +

+

Notes: Symbols indicate direction o statistically signi cant correlation between metro per ormance score and shareo GVA in industry ; Two symbols indicate strong correlation (r2>= 0.5); * Japanese and South Korean metros, and BeloHorizonte, Brazil excluded rom this analysis due to data quality issues. Source: Analysis o Ox ord Economics, MoodysEconomy.com, and Cambridge Econometrics data.

34 The results are largely the same when the 50 U.S. metros are excluded rom the analysis.


19/52

1 9

Deconstructing Metro Per ormance:Employment versus Income GrowthThe basic measure o metro economic per ormance in the Global MetroMonitor combines indicators o employment and GVA percapita growth, refecting the value that the public and policy makers attach to achieving both outcomes on behal o peopleand places. Although these indicators depend to some degreeon one another, they do not always move in unison. On the onehand, some metros that appear quite good on income growth may not generate new jobs, refecting increased productivity but not

necessarily growing employment opportunities. On the other hand,metros can grow employment, but not the type o employment thatboosts incomes and standards o living or the broader population.

Unlike in the recession and recovery periods examined below, theoverall relationship within the 150 metros between employmentand income growth in the pre-recession period was weak. How di erent would the top and bottom 30 metro areas look i theirper ormances were judged separately on these sub-measures?

Overall, about one-third o the strongest and weakest pre-recession-era metro per ormers change i employment growthand income growth are analyzed separately. On employment,Eastern European metros in particular all out o the topper ormers, as their rapid income increases resulted rom industrial trans ormation, rather than boosts in labor supply. At the sametime, central and southern European metros, particularly inGermany, per ormed somewhat better on employment growthduring this period than their bottom ranks indicate, perhapsrefecting the e ects o an infux o less-skilled labor rom EasternEurope and elsewhere abroad.

On income growth, the bloom is o the rose in high-ranked American metros such as Phoenix and Las Vegas, where much o the baseline employment growth was concentrated in industrieslike construction. The same was true or the ast-growing metroso Brisbane and Madrid, where GVA per capita growth was merely average rom 1993 to 2007. At the same time, many American and Western European metros at the bottom o the ranks, includingBirmingham, Rotterdam, Pittsburgh, St. Louis, Cleveland, andRochester posted somewhat stronger income gains than those

low ranks would indicate. This is likely attributable to long-runproductivity increases in their important manu acturing sectorsthat occurred alongside slow and steady declines in the number o people employed in those sectors.

Summary The decade and a hal leading up to the Great Recession ound airly widespread growth o metropolitan economies across the globe, butparticularly in lower-income regions, most notably Eastern Europe, thatbene ted greatly rom new rontiers in global economic integration.U.S. and Western European metros exhibited a wide range o economicper ormance both across and within their nations, but achieved similaraverage levels o per ormance, generally well below those in other partso the globe.



20/52


Figure 3-4. Pre-Recession Per ormance Ranking 1993-2007

ShenzhenDubaiGuangzhouBeijing Abu Dhabi

DublinBangaloreShanghaiMoscow So aRigaTianjin Vilnius

Las VegasHyderabadLimaKuala LumpurSingapore AthensPhoenixTallinnMadrid

Krakow Mumbai AustinBratislavaBrisbaneRiversideBucharestNew Delhi

ChennaiSeoul PragueManilaBarcelona AlexandriaCairo

8.24.8

10.48.72.2

5.97.59.89.96.2

10.211.88.61.55.44.44.53.85.42.88.12.5

3.54.23.46.62.73.29.02.0

4.64.86.23.42.42.62.5

9.410.94.34.06.9

4.52.91.41.13.30.5

-0.60.8

4.92.32.82.83.12.03.60.33.73.12.73.11.13.32.9

-0.63.6

1.91.81.02.63.13.02.9

12345

678910111213141516171819202122

2324252627282930

31323334353637

5.14.12.23.51.7

3.72.64.11.92.72.72.02.42.82.04.11.91.61.26.02.82.4

2.62.53.21.62.01.82.11.1

3.12.11.62.32.75.02.11.4

1.31.92.92.03.0

1.92.51.62.82.32.22.62.42.12.51.22.52.72.9

-0.11.92.1

1.92.01.52.52.12.32.12.6

1.42.02.31.91.60.21.92.3

3839404142

4344454647484950515253545556575859

6061626364656667

6869707172737475

Warsaw KolkataOrlandoSantiagoBelo Horizonte

San DiegoIstanbul PortlandBogotMelbourneSalt Lake City ValenciaSacramentoThessalonikiToulouseTaipeiJohannesburg Auckland AtlantaBudapestMonterrey Helsinki

Buenos AiresNashvilleEdinburgh VancouverDenverCharlotteMiamiToronto

LondonGuadalajaraSo PauloDallasHong KongSan JoseMontreal Cape Town

RANk RANk

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data; see Data and Methods section or urther details.

ANNuAL CHANGE (%)

EmploymentIncome

ANNuAL CHANGE (%)

EmploymentIncome


21/52

2 1

Sydney LeipzigSan AntonioSeattleJacksonville

LjubljanaLos AngelesTampaMarseille WashingtonGlasgow StockholmMexico City BusanNew YorkHouston Virginia BeachLyonBostonBrasiliaLisbonBrussels

MinneapolisIndianapolisRio de JaneiroCopenhagenMunichRichmondLilleProvidenceOsloMemphisManchesterBaltimoreBridgeportColumbusSan Francisco Amsterdam

2.32.72.12.31.5

4.32.82.42.22.13.02.92.15.42.61.62.42.33.20.52.72.4

2.11.71.62.21.91.32.12.41.21.82.52.13.21.32.22.3

1.71.41.81.62.1

0.41.31.51.51.61.01.01.5

-0.51.11.71.21.30.72.30.91.1

1.21.41.41.01.21.51.00.81.51.10.70.90.21.40.80.7

7677787980

8182838485868788899091929394959697

9899100101102103104105106107108109110111112113

1.91.81.23.01.4

1.81.2

2.21.80.41.42.51.3

1.22.4

-2.61.20.51.31.61.3

1.82.61.01.81.01.01.61.41.00.11.41.00.91.40.8

-1.5

1.00.91.20.11.0

0.81.1

0.50.71.50.90.20.91.00.23.20.91.30.80.60.8

0.4-0.20.80.30.60.60.20.30.40.90.00.20.0

-0.6-0.30.2

114115116117118119120

121122123124125126

127128129130131132133134

135136137138139140141142143144145146147148149150

ParisChicagoRomeBu aloCincinnati

PhiladelphiaMilan

Birmingham ViennaJakartaKansas City RotterdamDusseldor

HamburgPittsburghRiyadhFrank urtCologneOklahoma City St. LouisStuttgart

ClevelandRochesterTurinHart ordLouisvilleNaplesMilwaukeeZurich TokyoBerlinPortoNagoyaDetroitNew OrleansOsakaBangkok

RANk RANk

uNITEd sTATEs EAsTERN EuROPE OTHER LOwER-INCOMEwEsTERN EuROPE OTHER HIGHER-INCOME

ANNuAL CHANGE (%)

EmploymentIncome

ANNuAL CHANGE (%)

EmploymentIncome



22/52


high

average

20 m

5 m

low

New Orleans

Milwaukee Detroit

Rochester Hartford

LouisvilleLas Vegas

Per ormance

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data.

Figure 3-5. Metro Per ormance During Pre-Recession Period (19932007)The strongest-per orming metros in the pre-recessionperiod could be ound in emerging nations o Asia and Latin America, as well as in the American Southwest and EasternEurope.

The weakest-per orming metros were split among thesesame world regions, including many older industrial areas o the United States and Western Europe, as well as Japaneseand Southeast Asian economies that experienced economicdi culties in the 1990s.

Labeled metro areas are the 15 top and bottom rankedper ormers.

Metropolitan Population2010 estimates


23/52

2 3

NaplesTurin

Zurich

Porto

Tokyo

NagoyaOsaka

Bangkok

Berlin

Riga

Vilnius

BangaloreHyderabad

Dubai

Abu DhabiShenzhen

Beijing

Shanghai

Tianjin

Guangzhou

Moscow

Sofia

Dublin



24/52


Shenzhen is located in southeastern China on the South China Sea, very near the Hong Kong and Guangzhou metropolitan regions. The Shenzhenmetropolitan region, home to 9.5 million inhabitants in 2010, is part o the larger Pearl River Delta mega-region o over 120 million people. Likemany other Chinese metropolitan areas, Shenzhen is experiencing rapidpopulation growth, expanding by 22 percent rom 2007 to 2010 alone dueto continued in-migration rom the countrys rural inland region.

Pre-RecessionShenzhen is one o the top per ormers across all periods in the Global MetroMonitor. In the 1993 to 2007 period, it achieved the highestranking among all 150 metro areas, posting annual income growth o 8.2 percent, and annual employment growth o 9.4 percent. The ormermeasure was in line with the national average over this time, whilethe latter ar outstripped growth rates in other major Chinese metros(Beijing, Shanghai, Guangzhou, Tianjin) and the nation as a whole.

Shenzhen became Chinas inaugural Special Economic Zone in the early 1980s, permitting market capitalism to fourish within its borders well be ore much o the rest o the nation. As a result, the region becamean attractive location or manu acturing in China, not only or ormerHong-Kong based industries but also or many Taiwanese and Japaneseelectronics companies. Manu acturing and energy output now accountsor roughly 58 percent o Shenzhens economy. Shenzhens growth overthis period has been nothing short o astonishing, expanding rom a

rural shing village o 20,000 in 1980 to a global metropolis o 10 millionby 2010.35

Recession and Recovery Shenzhen largely avoided exposure to the recession. While its growthrates slowed as compared to long-run averages, neither employment norincome dropped during the worldwide downturn. While income grew ata slower rate than the national average, employment growth remainedpositive and higher than in China as a whole.

Growth sped up in Shenzhen in 20092010, as it did nationwide.Shenzhens income and employment growth rates o 5.9 percent rankedthe metro second overall among the 150 studied. Its growth rates havenot rebounded to their prior levels, but it is unclear how sustainablethose pre-recession growth rates were.

The impact o the worldwide downturn on Shenzhen was muted in part by the areas growing role as a manu acturing and service hub or mainlandChina. From 2006 to 2009, the contribution o exports to Guangdongprovinces economic output declined rom 92 percent to 62 percent. 36

Shenzhen, Pearl River Delta, and China more generally, are upgradingand expanding their industries to achieve a competitive edge inthe international marketplace. Shenzhen is now home to global telecommunications giants Huawei and ZTE, and labor standards in theregion are improving. 37 The Chinese government is building bullet trainsto connect the Pearl River Delta metropolises o Hong Kong, Shenzhen,and Guangzhou, with anticipated travel times between each o 15minutes. 38 These developments could strengthen Shenzhens positionin meeting the needs o a growing internal market, while allowing theregion to remain a leader in international exports.

METRO PERfORMANCE PROfILE

sHENZHENshEnZhEn

Shenzhen

China

9,501

1,354,146

10,598

2,011

8.2%

9.3%

9.4%1 .0 %

Population2010 (million)

GVA per Capita2007 ($)

GVA per CapitaChange

Pre-Recession (19932007)

EmploymentChange

4.6%

8.0%

2.3%0 .0 %

5.9%

8.7%

5.9%1 .3 %


Recession (20082009)

EmploymentChange


Recovery (20092010)

EmploymentChange

35 Janet Carmosky, In Chinas Future, Wheres Hong Kong Fit In? Interview with Alex Fong, CEO o Hong KongChamber o Commerce. The China Business Network, June 4, 2010.36 Deutsche Bank Research: Chinas Provinces. Online at www.dbresearch.com/ser vlet/reweb 2.ReWEB?rwdspl=0&rwnode=DBR_INTERNET_EN-PROD$RMLCHPM&rwsite=DBR_INTERNET_EN-PROD [accessed October

2010]. This refects decreased global demand or exports amid the economic crisis, as well as growing domestic demandor Shenzhens manu actured goods.37 The spirit o enterpr ise ades: The cradle o Chinas start-up rms is showing its age. The Economist, January 21,2010.38 Carmosky, In Chinas Future, Wheres Hong Kong Fit In?


25/52

2 5

The Austin-Round Rock, TX metropolitan area is the capital regionor the state o Texas, located in the southwestern United States. Themetropolitan region is comprised o the core city, Austin, which together with its nearby suburbs has a population o 1,763,000 in 2010. From 1993to 2007, Austins population increased by 68 percent, and it was the sixthastest-growing large U.S. metro area in the 2000s.

Pre-recessionPrior to the recession, Austin outper ormed the United States onindicators o income and employment growth. Both rose by more than3 percent annually during this period, compared with more modest U.S.annual growth o 1.3 percent in income and 2.1 percent in employment.The Austin metro ranked 25th overall in the Global MetroMonitor or thepre-recession period.

The 1990s were a boom period or Austin, with major technology rmssuch as Dell Computer, IBM, and Texas Instruments anchoring theiroperations in the area. They employed tens o thousands o youngand highly educated workers who moved to the region or economicopportunity and the citys renowned cultural are. By 2000 the region hadthe highest share o population between the ages o 25 and 34 among the50 largest metros in the United States. 39

As a result o its technology ocus, the region was hit harder than mostduring the dot-com bust o the early 2000s, but rebounded to postrobust 4 percent annual employment growth rom 2003 to 2007. By 2007,the Austin metro produced 27 patents per 10,000 employees, third-highest among the 100 largest U.S. metro areas. 40 Its innovative capacity owes in part to its high level o college degree attainment (38 percent,8th in the nation) and the act that it is home to the University o Texasfagship campus, one o the nations largest and highest-rated publicuniversities.

RecessionDuring the recession, GVA per capita in Austin dropped precipitously (-3.1 percent rom 2008 to 2009), mirroring the nationwide decline.Employment stagnated, but did not all considerably as it did nationwide.The region was buoyed by its concentration in education and governmentservices (22 percent o jobs, versus 18 percent across all U.S. metros),industries that were not as impacted by the downturn. At the heighto the recession rom 2008 to 2009, Austin still registered net in-migration o more than 25,000 residents, the third-highest metro total in the nation. It also bene ted rom location in Texas, where relatively conservative state lending regulations reduced the prevalence o speculative mortgages that, in other parts o the nation, producedrampant home oreclosures and severe house price and employmentdeclines. 41 Austin ranked 40th among the 150 global metros during therecession, third-highest among American metros.

Recovery During the recovery, Austins income grew (2.7 percent) somewhataster than the national average (2.4 percent). Unlike the United Statesas a whole, which continued to shed employment in 20092010, Austinreturned to employment growth rates comparable to its pre-recessionper ormance (3.2 percent). Sectors such as pro essional and businessservices, education and health, leisure and hospitality, and governmentall posted strong job gains in the Austin metro rom 2009 to 2010. 42 As asign o the areas continued strength, Facebook made its rst major U.S.

expansion outside Cali ornia in Austin in 2010.43

In sum, Austins continued attraction and retention o high-skilledhuman capital, its diverse set o export-based industries, and itsavoidance o the worst U.S. housing market excesses o the 2000s helpexplain its stronger-than-national per ormance throughout the threeperiods.

METRO PERfORMANCE PROfILE

AusTINa i

1,763

309,756

43,860

37,928

3.4%

1.3%

3.1%2. 1%

Population2010 (million)

GVA per Capita2007 ($)


EmploymentChange

-3.1%

-3.3%

0.1%-4. 0%

2.7%

2.4%

3.2%-0. 7%


EmploymentChange


EmploymentChange

Pre-Recession (19932007) Recession (20082009) Recovery (20092010)

Austin

U.S.

39 Joe Cortright , The Young and the Restless in a Knowledge Economy (CEOs or Cities, 2005).40 Brookings analysis o U.S. Patent and Trademark O ce data.41 Alyssa Katz, The Lone Star Secre t: How Texas Avoided the Worst o the Real Estat e Meltdown. The Big Money,March 30, 2010.

42 According to U.S. Bureau o Labor Statistics Current Employment Statistics data.43 Kirk Ladendor , Facebook Friends Austin to Support Its Rapid Growth. Austin American Statesman, April 10, 2010.



26/52


Unprecedented levels o global economic integration that propelledgrowth in major metropolitan economies in the lead-up to 2007ampli ed the worldwide e ects o what transpired soon therea ter. A mild employment downturn began in the United States in early 2008,amid signs o weakness in the housing sector. By autumn o that year,the problem had morphed into a ull-blown nancial crisis implicatingregional institutions and markets worldwide, sparking the deepestglobal recession in over 60 years. While virtually no metropolitan areacompletely escaped the e ects o this Great Recession, the downturn was

ar rom uni orm in its impacts across and within world regions. 44

Regional PatternsThis section measures metro economic per ormance during the GreatRecession based on minimum year-over-year employment and GVA percapita growth rom 2007 to 2010. 45 Most o these measures derive romorecasts based on o cial government metro-level estimates or 2008 orearlier years, and should thus be treated as preliminary in nature, andindicative o metro per ormance rather than precise in their implications.For the vast majority o the 150 metro areas, the minimum year o employment and income growth was between 2008 and 2009. 46 Thirty-one (31) metro areas, however, registered their greatest employmentlosses, or smallest employment gains, in the most recent year rom 2009to 2010.

In a sharp reversal rom baseline per ormance be ore the recession, thetypical metro area in the dataset saw a one-year employment declineo 1.7 percent during the recession period, and an even more dramatic

annual income decline o 4 percent. Yet the variation around this typical per ormance was vast, rom a more than 17 percent employment drop inMoscow to a gain o more than 4 percent in Lima; and rom a 16 percentincome loss in Tallinn to a rise o more than 6 percent in Tianjin. Overall,114 metros shed employment, and 127 lost income, in the year o theGreat Recessions deepest impact.

Moreover, the top and bottom-per orming metros during the recessionperiod refected several dramatic changes rom the pre-recessionperiod (Figure 4-5). China posted the top ve per ormers in this period,led by Beijing, which still managed 4 percent growth in employmentand 5 percent growth in income at the nadir o its growth. The top 12per orming metro areas during this time, and 22 o the top 30, werelower-income metros outside the United States and Europe. All six Indianmetros ranked among the top 30, as did six Latin American metros, uprom just one in the pre-recession period. All o these top-per orming

4. RECEssION PERIOdee i pe id

44 As noted in the Data and Methods section, the recession period did not mark an actual decline inemployment or GVA per capita in all 150 metros; see the Looking Back and Looking Ahead section or urtheranalysis.

lower-income metro areas added employment during the GreatRecession, and just a hand ul experienced dips in GVA per capita.

Among the metro areas rom other world regions that posted top-30ranks were Warsaw and Krakow in Poland, the three Australian metros(Brisbane, Melbourne, and Sydney), the two South Korean metros (Seoul and Busan), and Abu Dhabi. None, notably, came rom the United Statesor Western Europe, which together had placed six metros within the top30 per ormers in the pre-recession period. In addition, Eastern Europesrepresentation among the top-ranked metros was reduced rom eight inthe pre-recession period to just two during the recession period.

Not only did several American and European metro areas all out o thetop-per orming category they had occupied pre-recession, but alsomany ended up among the bottom recession-era per ormers ( gure 4-6).Moscow and the Baltic capitals (Riga, Tallinn, and Vilnius) occupied thebottom our spots in the metro rankings. In Western Europe, ormer topper ormers Dublin and Madrid ell into the lowest-ranking metros, as didLas Vegas and Riverside in the United States.

This pattern o dramatic top to bottom metro per ormance shi ts duringthe recession refected a broader trend in the United States and WesternEurope. While the relative per ormance o a ew older industrial metrossuch as Detroit, Cleveland, and Birmingham changed little rom thepre-recession period, several other metros, particularly in the UnitedStates, altered substantially. San Jose, Charlotte, Portland, Atlanta,Denver, Nashville, and Salt Lake City, all strong growers rom the 1990sthrough the mid-2000s, plummeted at least 50 positions to the bottomo the metro rankings as the recession took hold. Barcelona, Valencia,and Helsinki ollowed a similar trajectory in Western Europe. Overall, U.S.metros occupied 19 o the bottom 30 spots during this period.

Average metro per ormance rankings by region suggest a much deeperimpact o the Great Recession on U.S. metros, in contrast to its relatively light touch on lower-income metros outside the United States and Europe(Figure 4-1). The 50 U.S. metros achieved a very low average rank o 102 during the recession period, down rom 91 pre-recession. WesternEuropean metro areas actually improved slightly relative to others, while

Eastern European metros saw their average ranking plummet rom 29 to89. As a result, the recession appeared to strengthen the relative positiono metro economies outside Europe and the United States, with theirhigher-income ( rom 66 to 54) and lower-income ( rom 49 to 33) metrosmoving up in rank on average.

45 For some metro areas, these minimums occurred in di erent years.46 O the 150 metro areas, 104 experienced their minimum employment change rom 2008 to 2009, and 136experienced their minimum GVA per capita change that year.


27/52

2 7

60

40

20

0

100

80

120

Pre-Recession Period

Recession Period

WesternEurope(n=39)

UnitedStates(n=50)



EasternEurope(n=12)

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data. Some values based onorecasted estimates, please see Data and Methods section or urther details

Figure 4-1. Metro Areas Outside the United States and Europe OutrankedOthers on Economic Per ormance During the Great Recession Average Rank out o 150

The relationship between individual metro per ormance be ore therecession, and metro per ormance during the recession, variedconsiderably by world region. In lower-income non-U.S. and Europeanmetro areas, strong per ormance be ore the recession was associated with strong per ormance during it. The opposite held true in Eastern and Western Europe, where weaker per ormance during the recession wasassociated with stronger per ormance be orehand. 47

Metro Per ormance Factors While region-speci c dynamics clearly contributed to the disparateper ormance o global metro areas during the Great Recession, somestructural metro- and national-level actors appeared to be important as well.

Population Size and Growth While across the 150 metro areas larger places per ormed better duringthe recession than smaller places, this seemed primarily to refect thestronger showing o big Asian metro areas, and the weaker showing o

small Eastern European metro areas. Thus, being big in and o itsel didnot seem to insulate a metro rom the economic downturn. 48 Similarly,population growth across all metros was associated with strongerrecession per ormance, but this was again the product o growthdi erences across regions, not within them. Only in high-income metrosoutside the United States and Europe was population growth associated with stronger recession per ormance, and then only through largeremployment gains (or smaller employment losses).

Income As the metro rankings suggest, lower-income metro areas seemed to weather the recession better than others. This was equally true withrespect to employment and income growth. The very lowest-incomemetros (those with GVA per capita under $10,000) per ormed thebest, but as with population this also seemed to refect region-speci cpatterns, and not necessarily the value o having lower income orsubsequent growth. 49

National Per ormanceHow metros compared to their nations during the recession indicatessomething about their broader role within national economies. Across all global metro areas, the typical metro shed employment at a slightly lowerrate than the national average, but lost income at a slightly higher rate(Figure 4-2). The di erence was even starker within world regions. InEastern Europe, or instance, the typical rate o employment loss in largemetros was over 1 percentage point lower than at the national level, while

47 There was no statistica lly signi cant relationship between pre-recession and recession per ormance (combinedemployment and income growth) or U.S. metro areas.48 Among the ve world regions, only within Easter n Europe was metro population signi cantly associated withrecession-era per ormance, as bigger metros on average did worse than other metros.

49 Within the ve regions, there was no statistically signi cant relationsh ip between metro income and either metrooverall economic per ormance, or growth in employment or income, during the recession period.

R E C E s s I O N P E R I O d


28/52


the typical rate o income loss in other lower-income metros was nearly 1percentage point higher. The typical U.S. metro narrowly outper ormed

the national average on employment, but experienced an income declinenearly 1.5 percentage points greater than the nation. It may be thatthe high value-added nature o jobs in these global centers meant thattheir more modest employment declines during the Great Recessionnevertheless produced larger-than-average income losses. Metrosthemselves may also have shed high-value jobs at a disproportionate rateduring this recession, due to the greater susceptibility o those jobs tothe impacts o the nancial crisis on capital markets.

The contribution o national context to metropolitan per ormanceappeared to be at least as important during the recession as be orehand.Once again, a little under hal o metropolitan employment changecould be explained by national employment change, controlling or eachmetros contribution to national output. An even greater shareover80 percento a metros income change during the recession could beattributed to the national trend than in the pre-recession period. Thestrong relationship between national economic per ormance and metroeconomic per ormance during this period may refect the enhancedinfuence o the health o national nancial and debt status and related

national nancial and scal policies in the context o a severe global downturn, which exist outside the scope and powers o individual metropolitan areas. 50

The condition o national housing markets be ore and during therecession also helps to explain some o the dramatic changes inmetropolitan per ormance between the two periods. Deutsche BankResearch, using data rom the Bank or International Settlements(BIS), classi ed 34 o the 53 countries in which the 150 metropolitanareas sit into two groups based on international house price data: thosethat experienced negative house-price shocks coincident with therecession, and those that did not. Those that did experience a shockcontain many o the metros that were per orming well above-averagein the pre-recession period, but per ormed well below-average duringthe recession (Figure 4-3). From an average per ormance ranking o 77 be ore the recession, metro areas in nations experiencing house-price shocks dropped to an average rank o 93 during the recession. By contrast, those in nations that avoided severe house price declines saw their average rank rise rom 94 to 66 between the two periods. Signi cantdeterioration in housing market conditions thus appears to help explainthe rapid descent o metro areas such as Dublin, Riga, Valencia, Las Vegas, and Phoenix as the Great Recession took hold. 51

50 The stronger relationship may also refect the statistical infuence o national trends in the models that orecastmetropolitan-level per ormance through 2009. That noted, historical government estimates o metro-level output in, orexample, the United States are themselves derived in part rom national GDP statistics. This suggests that the statistical relationship between metro and national economic per ormance, while probably stronger in the orecasted data, is notnecessarily unique to the more recent periods. See Matthew J. McCormick, Sharon D. Panek, and Ralph M. Rodriguez,Gross Domestic Product by Metropolitan Area. Survey o Current Business, October 2009, pp. 100131.

1.5

-1.5

-1

-0.5

0.5

1

0

WesternEurope(n=39)

UnitedStates(n=50)



All regions(n=150)

EasternEurope(n=12)

Annual EmploymentGrowth Difference

Annual GVA per CapitaGrowth Difference

Source: Analysis o Ox ord Economics, Moodys Economy.com, and Cambridge Econometrics data. Some values based onorecasted estimates, please see Data and Methods section or urther details

Figure 4-2. Metros in Most Regions Led Their Nations on Employment Growth,but Lagged Their Nations on Income Growth, During the RecessionMedian Di erence between Metro and National Annual Employment and GVA per CapitaGrowth Rate by Region, Recession Period

51 Countries identi ed as experiencing or having experienced house-price shocks according to BIS data include:Bulgaria, China, Denmark, Estonia, France, Hong Kong, Latvia, Lithuania, Norway, New Zealand, Russia, Slovakia,Spain, United Kingdom, United States. Countries not experiencing signi cant house-price shocks include: Austria, Australia, Belgium, Czech Republic, Finland, Germany, Greece, Hungary, India, Indonesia, Israel, Italy, Korea, Malaysia,Netherlands, Portugal, Sweden, South A rica, Switzerland, Thailand. The observed impact o housing price bubbles onmetro per ormance would likely be greater i one were able to identi y house-price shocks at the sub-national level,especially in the United States; comparable sub-national data on house prices were not available on an international basis, however.

Industrial StructureShares o output in our industry groupings were associated with stronger

or weaker metro economic per ormance during the recession in di erent world regions (Figure 4-4):

In Western Europe, metros with large shares o output inconstruction per ormed worse than others during the recession, areversal rom the pre-recession relationship, and an indication o the impact o house-price shocks and associated sharp declines inconstruction activity on the economy in metros such as Dublin andMadrid

Metros in Western Europe with signi cant energy andmanu acturing output, such as Milan, Stuttgart, and Birmingham,also exhibited weaker recession per ormance, refecting at leastas much the long-run economic di culties o these metros as theimpact o reduced global demand on their export industries


29/52

2 9

Financial and business services centers had di erent trajectories indi erent regions during the recession. In Eastern Europe, metros

with larger shares o output in those industriesparticularly Warsawtended to out-per orm their peers. In high-income metrosoutside Europe and the United States, however, nancial andbusiness services-oriented metros such as Vancouver, Toronto,Osaka, and Tokyo tended to under-per orm other metros, dueperhaps to their higher degree o integration with the strugglingglobal capital markets

In Eastern and Western Europe and the United States, metros with signi cant representation o non-market services per ormed

signi cantly better than others, indicating the greater stability o sectors such as government, health care, and education in the aceo the Great Recession

Period Summary For some major metro economies, the Great Recession rein orced existinggrowth patterns. Lower-income metros in Asia, Latin America, and theMiddle East were much less a ected by the downturn than WesternEuropean and American metros, and posted even stronger relative

per ormance during the recession. In other respects, the crisis markeda dramatic shi t in growth, particularly in Eastern European metros thathad been among the strongest per ormers prior to the recession. Thereand in corners o the United States and Western Europe, steep drops inhousing prices reversed metro trajectories, and challenged the growthmodels that had propelled those economies prior to the downturn. Theseverity o the recession pointed to a long road back or many o thesemetro areas and their respective nations, as explored in the next section.

40

House-PriceShock

No House-PriceShock

Recession

Global Metro Monitor

Documents