a joint initiative of ludwig-maximilians-universität …timeline that ended in 2009. tariffs and...
TRANSCRIPT
Forum W I N T E R
2010
Trends
Focus
NAFTA Carol Wise
Christian Deblock andMichèle Rioux
Robert A. Blecker andGerardo Esquivel
Dominick Salvatore
Philip Martin
Meera Fickling
Specials
TURBULENT WATERS IN THE EMU:TRANSLATION FROM
WIRTSCHAFTSWOCHE
GREENSPAN, DODD-FRANK AND
STOCHASTIC OPTIMAL CONTROL
ESTIMATION OF PRODUCTION COSTS
FOR ENERGY RESOURCES
REPEC – A PLATFORM FOR
MEASURING OUTPUT IN ECONOMICS
Karl Otto Pöhl
Jerome L. Stein
Hans-Dieter Karl
Christian Seiler andKlaus Wohlrabe
STATISTICS UPDATE
A joint initiative of Ludwig-Maximilians-Universität and the Ifo Institute for Economic Research
VOLUME 11, NO. 4
CESifo Forum ISSN 1615-245X (print version)ISSN 2190-717X (electronic version)
A quarterly journal on European economic issuesPublisher and distributor: Ifo Institute for Economic Research e.V.Poschingerstr. 5, D-81679 Munich, GermanyTelephone ++49 89 9224-0, Telefax ++49 89 9224-98 53 69, e-mail [email protected] subscription rate: €50.00Single subscription rate: €15.00Shipping not includedEditors: John Whalley ([email protected]) and Chang Woon Nam ([email protected])Indexed in EconLitReproduction permitted only if source is stated and copy is sent to the Ifo Institute
www.cesifo-group.de
Volume 11, Number 4 Winter 2010_____________________________________________________________________________________
NAFTA
The North American Free Trade Agreement: A RequiemCarol Wise 3
NAFTA – A Model Running Out of Breath?Christian Deblock and Michèle Rioux 9
NAFTA, Trade and DevelopmentRobert A. Blecker and Gerardo Esquivel 17
Measuring the Economic Effects of NAFTA on MexicoDominick Salvatore 31
NAFTA and Mexico-US Migration: What Lessons, What Next?Philip Martin 38
North America’s Uphill Battle on Climate Change and Its Implications for the North American Trading SystemMeera Fickling 45
Turbulent Waters in the EMU: Translation from WirtschaftsWocheKarl Otto Pöhl 52
Greenspan, Dodd-Frank and Stochastic Optimal ControlJerome L. Stein 55
Estimation of Production Costs for Energy ResourcesHans-Dieter Karl 63
RePEc – An Independent Platform for Measuring Output in EconomicsChristian Seiler and Klaus Wohlrabe 72
Statistics Update 78
Focus
Trends
Specials
Forum
CESifo Forum 4/20103
Focus
THE NORTH AMERICAN
FREE TRADE AGREEMENT: A REQUIEM
CAROL WISE*
Despite the emphasis that then-presidential candidateBarack Obama placed on the need to renegotiate theNorth American Free Trade Agreement (NAFTA)during his 2008 US presidential campaign, thispromise has thus far come to naught. Once elected,President Obama’s first foreign visit was to Mexico toconfer with President Felipe Calderón. During thisdiplomatic foray Obama assured Calderón that theUnited States mainly intended to upgrade those partsof NAFTA that were of most concern to US voters,namely, labor standards and environmental protec-tion. Yet, when the ‘Three Amigos’ met for their firstNAFTA summit in 2009, the agenda discussed byObama, Calderón, and Canadian Prime MinisterStephen Harper was dominated by the problem ofundocumented migration of Mexican workers intothe United States and Canada and by the explosion ofdrug trafficking and cartel-related violence along theUS-Mexico border.
As palpable as the domestic debate over the need forNAFTA reform may have seemed at the time of the2008 US presidential election, the fact is that NAFTAis more or less beside the point at this political eco-nomic juncture. This is because 99 percent of all tar-iffs on those goods and services covered by NAFTAhave basically been eliminated, and because politi-cians and policy makers in all three of the membercountries have failed to institutionalize and updatethe agreement in ways that address problems that aremultiplier effects of NAFTA itself. At the top of thislist – apart from the easy flow of drugs and undocu-mented migrants across the border – would be thefailure to promote export competitiveness and to fos-ter the development of NAFTA as a regional projectproper. With NAFTA now fully implemented, the
inability of political leaders in North America to ren-
ovate and expand on the accord has meant its eclipse
by more compelling global forces.
The most remarkable force is the rapidity with which
China has gained a foothold in sectors once consid-
ered ‘North American’, such as computer peripherals,
sound and television equipment, telecoms, electrical
machinery, equipment and parts. Since China’s 2001
entry into the World Trade Organization (WTO) its
exports have steadily surpassed those of Mexico and
Canada in any number of US market niches. What’s
more, US exports, mainly in auto parts and produc-
tion, are quietly being displaced by Chinese investors
in Mexico’s northern export processing, or maquila,
zones. Canada and Mexico remain the most impor-
tant trading partners of the United States overall, and
together represent the largest supply of US energy
imports; however, when it comes to remedies for
today’s economic pain, China, it seems, should be the
main departure point for any debate over the current
sources of job dislocation and associated economic
stress in the United States and larger North American
market.
Economic policy and public discourse across North
America has yet to register the full implications of
China’s rapid ascendance in regional markets. Even
though the prospect of the ‘China threat’ has spawned
a whole cottage industry of academic and popular
books in the West, apart from launching a series of
elite diplomatic dialogues and pressuring the Chinese
to revalue their currency, Washington, Ottawa and
Mexico City have been slow to react.
Nevertheless, although competition from China may
be the most obvious factor in accounting for height-
ened job insecurity in both the United States and
Mexico, NAFTA still bears its own share of the pub-
lic’s wrath. Despite NAFTA’s considerable break-
throughs, the North American project has been
stymied by continued political gridlock over trade
policy in the United States, as well as some bitter dis-
appointments over expectations versus actual out-
comes with regard to labor and the environment.
NAFTA may still carry symbolic weight for the US
NAFTA
* University of Southern California, Los Angeles.
electorate, but at this point it seems safe to say that the
agreement has been steadily relegated to history’s
junk heap.
NAFTA: from glitz to gloom
It is now twenty years since former Mexican trade
minister Jaime Serra Puche and US Trade
Representative Carla Hills sat down at the annual
Davos Forum to explore the possibilities for negotiat-
ing what would later become the North American
Free Trade Agreement. The Uruguay Round was in
limbo at the time and the decision of the US to nego-
tiate a bilateral free trade agreement (FTA) that
included a developing country was unprecedented.
But this is just one way in which NAFTA broke new
ground.
A second benchmark for NAFTA was Mexico’s will-
ingness to forgo any special or different treatment
related to its developing country status at the NAFTA
negotiating table. It is difficult to exaggerate the extent
to which this stance represented a complete U-turn in
Mexico’s approach to foreign economic policy. The
decision marked the advent of a new generation of
more technocratic policy makers within the upper
echelons of the state bureaucracy and the eagerness of
this market-oriented cohort to lock in an entirely new
set of policies based on liberalization, privatization,
and deregulation.
This paved the way for a final agreement that went
well beyond what had been accomplished to date
within the Uruguay Round. On this count, NAFTA’s
key innovations were the protection of intellectual
property rights (IPRs), the liberalization of invest-
ment and trade in services, and the creation of mech-
anisms to resolve investment disputes based on bind-
ing international arbitration. For the first time, ‘old’
issues on the multilateral trade agenda (market access
for agricultural and industrial goods) were combined
in one agreement with the kinds of ‘new’ issues (ser-
vices, investment and IPRs) that the OECD countries
had been pushing for at least since the Tokyo Round
of the General Agreement on Tariffs and Trade
(GATT).
The fourth breakthrough was the negotiation and
attachment of labor and environmental side accords
to the NAFTA agreement, as the prospect of
Mexico’s entry into an FTA with the United States
and Canada had invoked valid worries about environ-
mental dumping and the abuse of labor rights in theselatter countries. Historical in their own right, theselabor and environmental agreements were offered upby the Clinton administration as side payments togarner congressional votes for NAFTA and tocounter the political blowback on the domestic frontthat had arisen with regard to the steep asymmetriesbetween Mexico and the United States.
Along with the side agreements on labor standardsand environmental protection, the NAFTA accordpromoted the free flow of goods, investment, and ser-vices within the North American bloc over a 15-yeartimeline that ended in 2009. Tariffs and non-tariffbarriers were eliminated on 65 percent of NorthAmerican goods by the 5-year point; tariff reductionson automobiles occurred over a 10-year period, withthe rules-of-origin stipulation that such vehicles mustmeet a 62.5 percent local-content requirement inorder to qualify.1 In the agricultural sector, sensitiveproducts were allotted a 15-year liberalization sched-ule that ended in 2009.
In the end, negotiating tensions were such that sugarand dairy products were excluded altogether in tradebetween Canada and Mexico. This is just one way inwhich NAFTA still fell short of its mandate to liber-alize substantially all trade between the three part-ners. First, administered protection persisted in thesetting of hefty percentages for local content underNAFTA’s rules of origin in such sectors as autos,high-tech products and textiles and apparel. Second,little progress was made toward the elimination ofantidumping practices and countervailing duties.Despite the ability of special interests to secure theseprotectionist concessions, hindsight suggests thatNAFTA has been a liberalizing force overall, as eachparticipant was clearly looking to reduce transactioncosts while simultaneously increasing the benefits ofcooperation.
Canada and Mexico saw an opportunity to secureaccess to the US market and establish clearly definedrules and procedures for resolving trade and invest-ment disputes. The United States, while also con-cerned with promoting and rationalizing economicties within the North American bloc, primarily soughtto bolster the rules and norms that constituted theinternational trade regime codified within the GATT.
CESifo Forum 4/2010 4
Focus
1 In order to qualify for NAFTA’s preferences, goods have to: (1) beproduced entirely within the NAFTA bloc; (2) incorporate onlythose non-NAFTA materials that are sufficiently processed in NorthAmerica to qualify for a tariff reclassification; and (3) satisfy a min-imum-content rule.
CESifo Forum 4/20105
Focus
Although pragmatic in the sense that all three parties
sought to strengthen and institutionalize respective
political-economic ties that had long been in place but
had heretofore been managed in an ad hoc manner,
hindsight shows that NAFTA also unexpectedly
unleashed its own share of animal spirits in North
America.
From the launching of the negotiations in September
1991 to the ratification vote in the US House of
Representatives in November 1993, the tone of the
NAFTA debate was counter-intuitive. Economic inte-
gration theory suggests that both the United States
and Canada, as larger, wealthier and more open G8
economies, should anticipate that marginal adjust-
ments would occur. In turn, Mexico, as the smaller,
poorer and more closed economy, should expect to
undergo a more costly adjustment in the short term,
but to realize considerable dynamic gains in the medi-
um to long term.
From this theory it should follow that the debate over
whether to pursue an FTA would be more heated in
Mexico, the country that had the most at stake.
Conversely, given that the United States and Canada
had much less on the line, one would expect a fairly
tame discussion about whether to negotiate an FTA
that included Mexico. Paradoxically, the opposite sce-
nario emerged.
In Mexico, the administration of Carlos Salinas
(1988–1994), which oversaw the NAFTA negotia-
tions, was able to quell open debate over NAFTA by
drawing on the authoritarian clout of its ruling
Revolutionary Institutional Party (PRI) to reinforce
informational asymmetries and marginalize dis-
senters. In Canada and the United States, the
NAFTA debate literally exploded. In the United
States, in particular, a full-scale national controversy
ensued, a main upshot being the emergence of a blue-
green coalition of grassroots labor and environmen-
tal activists that managed to insert non-trade issues
onto the US trade policy-making agenda like never
before.
First was a realistic reaction to the miserable work-
ing conditions and badly polluted maquila factory
sites that lined the US-Mexico border. If NAFTA
signified the free flow of goods, services, and capital
between all three countries, what was to stop the
flow northward of environmental pollution and
sub-standard working conditions? To the chagrin of
free trade purists like Jagdish Bhagwati,2 it was this
coalition that compelled the senior Bush adminis-tration to expend political capital on border cleanupand the enforcement of much higher environmentalstandards. With the election of President BillClinton in 1992, the formal negotiation of labor andenvironmental side agreements to accompanyNAFTA was offered as a quid pro quo for the blue-green endorsement of the 1993 NAFTA-implement-ing legislation.
The second reaction was largely symbolic, wherebyNAFTA came to embody all that was cumulativelywrong with the US political economy at the outset ofthe 1990s. What had started out as an issue-orientedblue-green coalition in 1991 blossomed into a full-blown anti-NAFTA movement that included every-one from job-seeking college graduates to downsizedbusiness executives, laid-off factory workers to teach-ers’ unions. Regardless of the actual effect thatNAFTA would have on any of these constituents,they were united in the perception that they had some-how been excluded from the prosperity that surround-ed them in the late twentieth century, and they wereunderstandably angry about it.
NAFTA’s self-appointed ‘losers’ have thus kept theopposition to further trade pacts alive, as witnessed inthe paper thin margins by which subsequent US bilat-eral FTAs have been ratified by the US Congress,including the 2006 US-Central America FTA. Butbeyond this phenomenon of NAFTA coming to sym-bolize a general sense of downward mobility in theUnited States, it is the tenacity of the blue-green coali-tion and its effect on congressional deliberations thatperhaps best accounts for the testiness of US tradepolicy since the launching of NAFTA.
Although this coalition won the battle in securing theattachment of labor and environmental side agree-ments to NAFTA, the lackluster enforcement ofthose agreements has further prolonged the trade pol-icy war on the domestic side. During the entire Bushadministration in the 2000s, for example, just twolabor complaints against Mexico were accepted forreview, whereas on the environmental side just sevencases involving Mexico have been resolved over thelife time of the agreement. Thus, much of the fighthas centered on correcting the institutional weakness-es in those earlier agreements, namely, the obligationof each country to enforce its own existing nationallaws but with little regard for strengthening and har-
2 See Bhagwati (2008).
monizing North American labor and environmentalstandards overall.
The blue-green opposition has continued its demandsfor more binding commitments in enforcing labor andenvironmental standards. Hence, the US insistence onincorporating labor and environmental stipulationsinto the pending Doha agreement, as well as the sub-sequent FTAs signed in the 2000s. Even seeminglyeasy US bilateral talks with countries that are other-wise ready and willing to constructively adhere to thefull menu of blue-green demands (South Korea,Panama, Colombia) have proved to be quite cumber-some.
NAFTA’s uneven returns
NAFTA’s critics have arguably done a better job ofadvertising its failures than its proponents have donein touting the concrete gains that have underpinnedNorth American integration since the early 1990s. Atleast at the aggregate level, it would be difficult topaint NAFTA as anything but a success. This is espe-cially so when NAFTA is judged according to its owngoals: the creation of a free trade area in which allthree partners have pursued an economic growthstrategy via the liberalization of goods, capital, andservices amongst themselves.
Total NAFTA trade now accounts for some 30 per-cent of all US trade,3 and the number of jobs gainedin the US economy since NAFTA’s implementation in1994 more than compensates for those jobs lost – evenwhen considering the massive job losses that haveoccurred in the wake of the 2007–2009 US financialcrisis. US trade in goods and services with Canadaand Mexico tripled – from 341 billion US dollars in1993 to more than 1 trillion US dollars in 2007 – andinward foreign direct investment quintupled amongthe three countries and increased tenfold in Mexicobetween 1990 and 2005 (Pastor 2008). In terms ofgross product, the NAFTA zone has now surpassedthe European Union (EU); however, the impacts ofregional integration in North America have beenquite uneven.
Again, when viewed from the dictates of economicintegration theory, it was expected that NAFTAwould benefit all three countries, but especiallyMexico, through the deepening of already strong ties
in cross-border production and intra-industry trade.
First, the elimination of barriers at the border would
promote scale economies related to greater specializa-
tion, increased technological capabilities, and more
rapid and efficient deployment of those factors for
which Mexico has a comparative advantage (natural
resources and comparatively cheap labor).
Second, it was argued that the blending of Mexico’s
abundant factors with the capital, technology, and
know-how that Canada and the United States
brought to the table would trigger a dynamic pattern
of income convergence among the three members.
According to this largely neoclassical trade narrative,
Mexico would readily advance up the industrial and
technological learning curve, substantially increase its
per capita income, and more authentically approxi-
mate the economic indicators of its fellow OECD
members.
The data show that NAFTA has delivered rather
erratically on these expectations. At the macroeco-
nomic level, Canada and Mexico have clearly con-
verged toward the more highly developed US stan-
dard in terms of aggregate growth, interest rates,
exchange rate stability and the lowering of inflation to
under five percent annually. But the microeconomic
data tell a different story, which highlights the need
for sound domestic policy reforms to complement and
maximize on the opportunities intrinsic to a regional
integration scheme.
After rebounding from the disastrous 1994 peso crisis,
the growth of per capita gross domestic product
(GDP) in Mexico has hit a virtual standstill in the
2000s and is still 6.3 times lower than that of the
United States. Even Canada, despite its advantage as
a G8 country, has lagged in this regard. Although
Canadian income distribution is the most equitable in
North America, Canada’s per capita income remains
about 20 percent lower than that of the United States
and its productivity and investment ratios are similar-
ly trailing.
While the roots of microeconomic under-perfor-
mance appear to lie somewhere in the gulf between
neoclassical trade theory – which assumes a state of
perfect competition and constant returns to scale
under NAFTA – and the concrete empirical obstacles
that underpinned its launching back in the early
1990s, the persistent divergence between the United
States and its NAFTA partners can also be chalked
up to the nature of political institutions and policy
CESifo Forum 4/2010 6
Focus
3 TradeStats Express, International Trade Administration, USDepartment of Commerce.
CESifo Forum 4/20107
Focus
making in these countries. In short, both Canada and
Mexico, albeit in greatly varying degrees, are still in
need of more proactive competition policies that spur
rather than deter investments, increase ties between
research and development (R&D), universities and
private initiative, and promote the application of
advanced technology to the extraction of natural
resources and the production of goods.
NAFTA’s potential here is limited, as it has delivered
its punch in terms of the role that enhanced levels of
trade and investment can play in catalyzing further
microeconomic change for both Canada and Mexico.
Because of the minimalist institutional framework
that all three members agreed to at the outset,
NAFTA has basically been frozen in place and is sore-
ly out of date when it comes to tackling today’s micro-
economic challenges.
The twilight of North American integration
The importance of Canada and Mexico as US trade
and investment partners is indisputable, and the
impressive growth of North American gross product
is testimony to the depth of these ties. At this point,
however, NAFTA’s operational tendencies are still
more akin to two bilateral deals that have basically
been cobbled together, meaning that the whole is no
greater that the sum of its parts. Whereas the very cre-
ation of NAFTA is testimony to the possibilities of
trilateral coordination based on the national interests
of each trade partner, all three countries adamantly
resisted the option of strengthening this cooperation
via the creation of European Union (EU)-style supra-
national institutions.
Canada and Mexico opposed the institutional for-
malization of NAFTA on the grounds that they
would be pushed around and further disempowered
by the United States if North America were to take
the supranational institutional route. The United
States reacted in its typical Anglo-Saxon fashion,
pejoratively equating the creation of supranational
institutions with the proliferation of the ‘Brussels
bureaucracy’ in North America. This insistence
that NAFTA remain a free trade area in the absence
of sound institutional moorings has thus stunted its
evolution into a more compelling regional project.
Case in point: the growth in total NAFTA trade in
the 2000s has been about 3 percent, versus the
9.8 percent growth rate registered in 1994–2000
(Pastor 2008).
In light of this impasse, it seems safe to say that the
authentic revival of NAFTA as a regional project
would require that the United States, as the hegemon-
ic member and industrial anchor, step forward with
the necessary leadership and provision of public
goods. Yet, the most visible US commitment in the
Bush junior era was the construction of a double-lay-
ered wall and hundreds of miles of vehicle barriers
along the 1,933-mile US-Mexico border meant to halt
the northward flow of undocumented workers.
Mandated by the Secure Fence Act of 2006 in the
wake of failed efforts within the US Congress to reach
bipartisan agreement on any number of sticking
points in the proposed immigration legislation, the
US Department of Homeland Security expects to
complete this project by 2011 at a cost of 7.6 billion
US dollars. Rather than investing in badly needed
improvements in the highway infrastructure that links
the three countries, this hefty US financial commit-
ment to construct further border barriers has under-
standably enraged its NAFTA partners.
A final wedge is China, now a major trade and invest-
ment presence throughout the Western Hemisphere.
Whereas China’s trade relationship with South
America is based on more traditional patterns of
comparative advantage – China’s export of lower-end
industrial goods and its import of primary products
from Argentina, Brazil, Chile and Peru, in particular
– the China-NAFTA relationship is one of export
similarity and fierce competition for manufacturing
market share, especially with regard to Mexico and
the United States (Devlin 2008).
Between 2002 and 2008, Mexico’s share of the US
import market slipped by 11 percent, from 11.6 to
10.3 percent, while China’s share rose by 50 percent,
from 10.8 percent to 16.2 percent (Watkins 2009).
Given that nearly all of Mexico’s manufactured exports
to the United States are goods produced by companies
that operate under the maquila (two-thirds of manu-
factured exports) or Pitex (one-third of manufactured
exports) programs, i.e. programs that were specifically
designed to deepen US-Mexican integration in these
sectors, it is incumbent upon both sides in this partner-
ship to work jointly to combat these intense competi-
tive challenges from China. Remarkably, US and
Mexican leaders in the public and private sector have
been completely passive on this count.
Along with China’s outpacing of all other developing
countries in its growth of manufactured exports from
2000–2006, it is now increasing its competitiveness in
high technology exports at an even greater speed than
in manufacturing as a whole (Gallagher and
Porzecanski 2010). Although Mexico is the only Latin
American country to rank amongst the top twenty
developing countries in terms of the technological
content of its manufactured exports, Kevin Gallagher
and others report that as of 2006, 82 percent of
Mexican exports in this category were under some
degree of competitive threat from China (Gallagher
and Porzecanski 2010).
The irony in Mexico’s case is that the stated purpose
of the country’s 1994 entry into NAFTA was precise-
ly to advance steadily up the industrial learning curve
and to situate domestic producers more securely on
the technological frontier. What went wrong? Prior to
entering NAFTA, trade and investment were liberal-
ized and longstanding industrial policies were dis-
mantled. Under NAFTA, the innovation process,
including technology transfer and R&D, was relegat-
ed to foreign direct investment (FDI) in Mexico’s
assembly plants, and it was envisioned that innova-
tion would result from the dynamic spillovers and
multiplier effects of heightened trade and investment
flows. However, sixteen years later, expenditures on
R&D have actually declined since 1994, and under
the prevailing laissez-faire regime the country’s
Information Technology (IT) sector and firms have
been decimated.
With US policy makers fixated on the completion of a
highly symbolic wall along the Mexican border, and
US public opinion holding on to its longstanding
NAFTA grudge, Chinese investors are also quietly
staking out their claims in the Mexican market. The
overriding goal is to establish manufacturing opera-
tions in Mexico based on integrated global production
chains, with an eye toward exporting to the US market.
For example, while still an incipient trend, the Chinese
computer company Lenovo is establishing supply
facilities in the northern Mexican state of Chihuahua,
the Golden Dragon firm is constructing a plant to
produce copper tubes in the state of Coahuila and, in
the Mexican state of Hidalgo, China’s Giant Engine
Company has invested 50 million US dollars to
acquire an auto assembly plant (Ellis 2009). Through
joint partnerships with companies such as Mexico’s
Grupo Elektra, a major distributor and financier of
infrastructure, Chinese automakers like Zhongxing,
the First Automobile Works, and others have set their
sights on jointly producing some 1.6 million cars per
year in Mexico by 2012.
Thus, the United States could soon be facing theworst-case scenario of all with regard to its mam-moth commercial deficit: the displacement of USsuppliers by Chinese firms in Mexico’s maquila
assembly plants – a trend that is now underway – andChina’s ability to offset Mexico’s higher labor andproduction costs by meeting NAFTA’s regional con-tent requirements and thereby gaining duty-freeaccess to the US market.
Needless to say, this is a far cry from what NAFTA’sarchitects originally had in mind. Unhappily, thedaunting domestic and regional repercussions of the2007–2009 financial meltdown in the United States,combined with President Obama’s considerable lossesin the US 2010 mid-term elections, do not bode wellfor policy innovation vis-a-vis NAFTA. As Canadalooks once again to the EU for answers and Mexicoretreats into a survival-oriented mode, it seems safe tosay that North America has officially entered thepost-NAFTA era.
References
Bhagwati, J. (2008), Termites in the Trading System, New York:Oxford University Press.
Devlin, R. (2008), “China’s Economic Rise”, in: Roett, R. and G. Paz(eds.), China’s Expansion into the Western Hemisphere, WashingtonDC: Brookings Institution, 111–147.
Ellis, E. (2009), China in Latin America: The Whats and Wherefores,Boulder: Lynne Rienner.
Gallagher, K. and R. Porzecanski (2010), The Dragon in the Room,Palo Alto, CA: Stanford University Press.
Pastor, R. (2008), The Future of North America, Foreign Affairs 87,84–98.
Watkins, R. (2009), The China Challenge to Manufacturing inMexico, Paper presented at Global Perspectives Conference: Focuson China, University of Arizona, April.
CESifo Forum 4/2010 8
Focus
CESifo Forum 4/20109
Focus
NAFTA – A MODEL
RUNNING OUT OF BREATH?
CHRISTIAN DEBLOCK AND
MICHÈLE RIOUX*
‘Marriage of convenience’ or ‘default choice’(Weintraub 1990; Brunelle and Deblock 1989)? In anycase, turning to the United States and embracing freetrade, Canada and Mexico abandoned economicnationalism that reached its limits in the 1980s andmade a bet that securing a broadened, safe and prefer-ential access to their main market, they would takeadvantage of North American integration. First, itwould stimulate their stalled economies and restoregrowth on new foundations. It would allow them touse the US market as a springboard for successful andcompetitive insertion into the global economy and itwould attract foreign investments looking for a gate-way to the US market. Other arguments were alsoimportant. In the case of Canada, for example, fear oflosing the advantages gained from the free trade agree-ment (FTA) signed with the United States in 1987,and, in the case of Mexico, NAFTA was an opportu-nity to lock in economic reforms. The United Stateshad more ambitious goals. The first was to establish alarge open market to enable US businesses to takeadvantage of deep integration de jure with their firstand third largest trading partners; the second was tobuild on this agreement and accelerate multilateral ne-gotiations and other negotiations;the third was to meet the chal-lenges of relentless globalization,the revival of European integra-tion and the rise of Asia as aneconomic power. North Americawas to become also a model thatwould pave the way for the cre-ation of other regional groupings(APEC, FTAA, transatlanticFTA and AGOA) (Deblock andTurcotte 2003).
The debate on NAFTA had political, legal, econom-ic, cultural and social dimensions. In Canada andMexico, it was seen as a direct threat to national sov-ereignty, social programs and developmental policies.In the United States fear of deindustrialization, or bymass relocations to Mexico were dangers shapingpublic opinion and debates. Yet, most pessimistic sce-narios did not materialize and the US economy expe-rienced strong growth throughout the 1990s, stimu-lated by globalization, inflows of foreign investmentsand development of new technologies, and this hadpositive impacts on growth in Canada and Mexico.NAFTA, nonetheless, continues to spark debates fordifferent reasons. First, it has lost much of its appealas a model of regionalism, as other models now com-pete with it, and second, NAFTA has become far lessinteresting to investors, and trade inside NAFTA aredwindling. Third, increasing pressure of internation-al competition is accompanied in Canada andMexico with another economic concern: sluggishproductivity. We will examine these three points inthe following pages, and conclude by asking a mostimportant question: “is NAFTA running out ofbreath as a model of regionalism, and if so, whatshould be done about it?”
NAFTA as a model
NAFTA is associated with the emergence of a modelof ‘new regionalism’, referred to as ‘open’ regional-ism as opposed to the previous models of ‘closed’.
* CEIM (Centre d’études sur l’intégra-tion et la mondialisation) and Universitédu Québec à Montréal (UQAM).
Table 1
Growth of real GDP and GDP per capita in Canada, Mexico
and the United States
Average annual GDP growth rate (%)Period
Canada Mexico United States
1980–89
1990–99
2000–09
3.0
2.4
2.1
2.4
3.4
1.9
3.1
3.2
1.8
Average annual growth rate of GDP per capita (%)Period
Canada Mexico United States
1980–89
1990–99
2000–09
1.9
1.3
1.1
– 0.5
1.7
0.9
2.5
1.9
0.8
Source: IMF World Economic Outlook Database.
The formula is well conceived but does not accurate-ly reflect the reality as any RTA (Regional TradeAgreement) is preferential and exclusive. The differ-ence distinguishing new and old ones is, on the onehand, their institutional characteristics: generally,they are trade agreements aiming to liberalize trade,and they are mostly bilateral and they are character-ized by a contractual approach. On the other hand,they are directed toward international markets, i.e.towards the promotion of trade in the broadest sense.They are also negative, yet in-depth integrationagreements. ‘Negative’ in that they aim to removebarriers to trade and competition. And ‘deep’because they also tend to harmonize rules and poli-cies in different areas of trade and investment with-out involving the elaboration of common policies(Lawrence 1996).1
NAFTA is a very comprehensive agreement coveringnot just trade in goods (including agriculture and ser-vices) but also investment, telecommunications, gov-ernment procurement contracting, intellectual prop-erty, standards and the movement of business people.In other words, it seeks as much to open up marketsas to establish common market rules, while avoidingthe negotiation of common policies. NAFTA resem-bles the European model but it does not go as far asfor example, setting a common competition policy.Where NAFTA radically differs from the Europeanmodel because it is essentially a ‘contractual’ agree-ment aiming at the sole objectives (Preamble and Art.102) of establishing a free trade zone subject to a sys-tem of fair competition, resolving disputes effectivelyand promoting trilateral cooperation in various areasrelated to the agreement.
NAFTA binds parties to respect the principle of equaltreatment (national treatment and most favorednation clause, MFN), with exceptions and otherclauses clearly identified. Also, as with any contract,parties must comply with the commitments made andmust submit, in cases of dispute, to procedures ofarbitrage and, where applicable, to a binding disputesettlement system. Its institutions are declined on fourlevels: ministerial (a commission on free trade andcommissions in the areas of environment and laborcooperation), administrative (coordinators and threesecretariats), legal (dispute settlement) and technical
(groups and working committees). The agreement isintended only to create an institutional and legal envi-ronment conducive to free movement of goods and
services, capital and business people. One importantaspect to mention is that, contrary to the ongoingintegration process in Europe, the agreement is notupgradeable. Some proposals were made to improvethe agreement, notably by President Zedillo who pro-posed to establish a regional development fund, andto transform it into a customs union or common cur-rency union, but these proposals were not pursuedand neither seriously considered.2
NAFTA might not have lead to as much debate if theUnited States had not sought to use it for purposes.The United States’ objectives, however, went farbeyond the North American framework. In the firstplace, the agreement was meant to create precedentsand to serve as leverage to advance multilateral nego-tiations in areas as contentious as investment andintellectual property rights. Firstly, for example,Chapter 11 on Investment (Part V) served as a modelfor negotiations, which ultimately failed, of a multi-lateral agreement on investment (MAI) (Gagné 2001).Secondly, Mexico was to serve as an example, demon-strating that trade liberalization coupled with contin-ued reforms could lead to vigorous growth and devel-opment creating jobs and wealth, and this, in return,would contribute to reinforce democracy, greaterrespect for individual rights, greater economic securi-ty and control of migration flows. And thirdly, astrong signal had to be sent to all countries willing tolaunch bilateral negotiations with the United States,in the Americas but also elsewhere in the internation-al community. In this regard, NAFTA symbolized theway forward in trade liberalization.
These expectations were obviously exaggerated, ifnot unfounded, as in the case of the reduction ofmigration flows. Moreover, if negotiations in NorthAmerica did in fact have direct impacts on theUruguay Round and its results, the US trade agendaquickly encountered strong resistance – beginning inthe Americas where negotiations were finallydropped because of a lack of consensus (Summit ofMar del Plata), but also in the case of APEC and atthe WTO (Deblock 2007). Worse, by engaging in abilateral path, the United States somehow freedeveryone the obligation to favor multilateralism,thus opening a true Pandora’s box. In this regard,the disastrous Ministerial Conferences in Seattle(1999) and Cancun (2003) were turning points.Apart from the fact that they revealed the gap
CESifo Forum 4/2010 10
Focus
1 Integration is said to be positive when it is oriented towards com-mon goals and shallow when it is limited to trade.
2 The Security and Prosperity Partnership , established in 2005,helped open discussions on new areas including energy, transportand certain regulations, but its primary intention was to strengthensecurity at the borders and facilitate passage thereof.
CESifo Forum 4/201011
Focus
between US positions and thoseof emerging countries, they de-monstrated the impossibility ofa multilateral consensus in-evitably leads to a multiplicationof bilateral and regional chan-nels as the default solution. TheUnited States is not solelyresponsible for the proliferationof the spaghetti bowl of tradeagreements, but the doctrine of‘competitive liberalization’,which, under the presidency ofGeorge W. Bush, replaced re-gional blocks and their disinter-est in multilateralism, onlyadded fuel to the fire. What isstriking today is not so much therace towards trade agreementsbut rather the new form they take. We have entered athird wave of regionalism, marked both by the pro-liferation of agreements in Asia, a region that longremained outside this trend, and by the emergence ofa different model that we call a ‘partnership’ model,initiated mainly by China, a new player that everyday affirms and reaffirms its ambitions to shape theworld economy (Beeson 2007).
The two periods of NAFTA
NAFTA is often criticized, often severe when it comesto jobs and working conditions (Faux, Salas andSchott 2006), but its overall record clearly remains pos-itive, far more so in fact than was initially anticipated(Courchene 2003; Hufbauer 2008; Pacheco-López2008). First, it established an institutional framework
conducive to trade, to the great satisfaction of the busi-ness world. Some disputes – timber, transportation andsugar for instance – regularly reappear, but they remainfew given the magnitude of trade. Similarly with dis-putes relating to investment: they remain isolated casesand arbitration decisions are too uncertain to allowstrong conclusions to be drawn from them.Furthermore, trade between the three countries experi-enced a sharp rise. Between 1992 and 2008, exportsfrom Canada and Mexico to the United States weremultiplied respectively by 3 and 4.5, while their importsfrom the United States have been multiplied by 2.5 and3. As for trade between Canada and Mexico, it hasbeen multiplied by 6. Finally, even if NAFTA did notbegin auspiciously (Zapatista uprising on 1 January1994 and the peso crisis by the end of that year),Mexico recuperated quickly from the crisis, and experi-enced strong growth led by exports and a booming
maquilas industry. Thus, the num-ber of jobs in these industries dou-bled between 1995 and 2000 from648,263 to 1,291,232.3 Anothercompelling fact is that the share ofthe manufacturing sector in totalexports went from less than halfthe total before NAFTA to morethan 80 percent in the early 2000s(Waldkirch 2008).
One of NAFTA’s most notewor-thy effects was to significantlystrengthen the trade links
Table 2
Exports and imports of goods and services 1970–2009
(% of GDP)
ExportsPeriod
Canada Mexico United States
1970–79
1980–89
1990–99
2000–09
23.3
27.0
34.2
39.0
8.1
17.1
21.5
26.9
7.5
8.5
10.5
10.9
ImportsPeriod
Canada Mexico United States
1970–79
1980–89
1990–99
2000–09
22.8
25.2
32.7
35.2
10.7
13.6
23.1
28.6
7.7
10.3
11.9
15.7
Source: United Nations Statistics Division.
Table 3Growth of US merchandise trade with Canada and Mexico 1980–2009
(average annual growth rate in %) Exports to Period
Canada Mexico 1985–89 1990–94 1995–99 2000–04 2005–09
8.7 7.6 7.9 2.9 2.4
16.6 15.7 12.1 5.8 3.7
Imports from Period Canada Mexico
1985–89 1990–94 1995–99 2000–04 2005–09
6.0 8.0 9.0 5.6
– 0.9
8.9 13.2 17.3 7.9 3.3
Source: Bureau of Economic Analysis (BEA), US International Trans-actions Accounts Data.
3 Source: INEGI.
between the three countries. It isnecessary, however, to distinguishbetween the period prior to andafter 2000. Before 2000, tradegrew rapidly. Thus, the share ofintra-NAFTA trade between1994 and 2000 went from 45 per-cent to 56 percent for exports andfrom 37 percent to 40 percent forimports.4 Thereafter, the share ofintra-NAFTA exports remainedstable until 2005, but then beganto recede slowly before droppingto 48 percent since the 2008 crisis.The decline is felt even moreacutely in the import sector.From 40 percent, the share ofintra-NAFTA imports droppedto 38 percent in 2005 and then to33 percent in 2009. Clearly, tradeis slackening and three reasonsexplain this. First, Canada andMexico benefited from twoadvantages during the 1990s:strong growth in the UnitedStates and weak currencies.These have disappeared as theUS economy has lost momentumand the crisis of 2008–2009 hassharply driven down bilateraltrade while the Canadian dollarachieved near-parity with the USdollar and the peso’s real valuerose sharply. Both Canada andMexico have lost competitive-ness, at least in terms of prices(Morales 2010).
Furthermore, trade conditions have dramaticallychanged since China’s accession to the WTO. Chinesecompetition is felt at two levels: both within theUnited States (the main market for Canada andMexico), and in these two countries as well whereChinese products are challenging local industries andimports from the United States. Thus, between 1996and 2008, we have witnessed the share of importsfrom China growing from 2.1 percent to 10.1 percentof total imports in Canada, from 0.8 percent to10.7 percent in Mexico and from 6.6 percent to16.5 percent in the United States.5 Conversely, whileUS exports to China rose from 1.9 percent to 6.5 per-
cent between 1996 and 2008, this was not the case for
Canada and Mexico: Canada’s exports to China rose
only from 1.1 percent to 2.5 percent and Mexico’s,
from 0.2 percent to 0.7 percent. But what do we have
to offer to China? While Canada has natural
resources and cutting-edge industries in certain sec-
tors, Mexico does not have the advantage of natural
resources, with the exception of petroleum, and its
industries are still very fragile and dependent on inte-
gration with the United States while it is especially
exposed to Chinese competition, with the exception of
the automobile industry (Meza and Salvador 2009;
Rosales and Kuwayama 2007; Arès 2005).
Finally, the third source of concern is the changes in
the direction of direct investments. In Canada, this
phenomenon was observed long ago: in 1994, two
thirds of direct investments (FDI) came from the
CESifo Forum 4/2010 12
Focus
40
50
60
70
80
90
100
1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009
Exports
Imports
Sources: Statistics Canada, Canada's Balance of International Payments, Catalogue 67-001; Mexico, Ministry of Economy, Underministry of Foreign Trade.
CANADA AND MEXICO MERCHANDISE TRADE WITH THE US 1989–2009
% of total trade
Mexico
Canada
60
80
100
120
140
1993 1995 1997 1999 2001 2003 2005 2007 2009
Canadian dollar
Mexican new peso
Source: Federal Reserve Bank of St Louis.
REAL EXCHANGE RATES Mexican new pesos and Canadian dollars to one US dollar
1993 = 100
Figure 1
Figure 2
4 Source: WTO.5 Source: United Nations, ComTrade.
CESifo Forum 4/201013
Focus
United States. By 2009, only half did, whileCanadian FDI in the United States went from 53 per-cent to 44 percent of total Canadian FDI abroad.But it is on the US side that the changes are most
notable. China and Asia general-ly attract few Canadian compa-nies, but this is not the case forUS companies which have longbeen established in this part ofthe world. Figure 3 is very reveal-ing, giving us the number of jobsin US subsidiaries (affiliates) inCanada, Mexico and China forthe manufacturing sector. Itclearly shows that althoughNAFTA originally stimulatedjob creation in both Canada andMexico, that trend has since beenreversed (Blecker 2009).
These contrasting trends revealtwo things. First, NAFTA hasgiven new impetus to economicgrowth in Canada and Mexicobut it has also had the perverseeffect of reinforcing their natur-al dependence on the UnitedStates. Second, changes in theinternational economic environ-ment have forced both countriesto rethink their economicstrategies, so far much toofocused on the United States(Goldfarg 2006).
Productivity – the weak link
Did NAFTA divert the attentionof Canadian and Mexicanauthorities from consideringother markets and partners? Onecan think so. On the one hand,the agreement was most satisfac-tory as trade was extremelyfavorable to Canada and Mexicothat benefitted from large tradesurpluses year after year. On theother hand, even when bothcountries sought to diversifytheir trade, their efforts did notyield the expected results sincecompanies did not really re-
sponded, and because the failure of the FTAA pro-ject revealing the failure of their economic diploma-cy on the continent. This being said, times havechanged. The United States, increasingly tied up with
Table 4US merchandise trade with major partners 1980–2009
(average share of total trade in %)Exports to Period
Canada Mexico Japan China 1980–89 1990–99 2000–09
22.4 22.3 21.9
6.2 9.8 12.9
10.8 10.1 6.2
1.5 1.8 4.6
Imports from Period Canada Mexico Japan China
1980–89 1990–99 2000–09
19.1 19.3 16.9
5.4 8.6 10.9
18.4 15.6 8.5
1.4 6.2
13.9 Source: Bureau of Economic Analysis (BEA), US International Trans-actions Accounts Data.
0
100
200
300
400
500
600
700
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008
Canada
Mexico
China
Source: US Bureau of Economic Analysis.
US MAJORITY-OWNED FOREIGN AFFILIATES, EMPLOYMENT 1997–2008Total manufacturing, 1 000 employees
80
90
100
110
120
130
140
1991 1993 1995 1997 1999 2001 2003 2005 2007
CanadaMexicoUnited States
Source: OECD Factbook 2010: Economic, Environmental and Social Statistics.
LABOUR PRODUCTIVITY GROWTH 1991–20081991 = 100
Figure 3
Figure 4
China, shows little interest in the Americas whileChina has become a major player in the Americas inonly a decade. Both Canada and Mexico are now upagainst a wall: how can they preserve their marketshares within North America while gaining shares inother markets, if not through innovation and invest-ment in the future?
Low productivity has indeed become a major concernin Canada and Mexico; no doubt does it also explainto a large extent the problems of competitiveness thatboth countries are faced with today. In this regard,Figure 4 and Table 5 are most revealing about threethings. First, the hopes for convergence have not beenrealized by NAFTA: Mexico has not reduced its pro-ductivity gap with the United States and, since 2000,Canada has seen this gap widening. Second, the rateof productivity growth has been falling dramaticallysince 2000, far more than in the United States. Andthird, at least for Canada, this gap is mainly due to theMFP (multi-factor productivity). Many reasons havebeen given to explain this fact, but they all converge inthe same directions: (1) low exchange rates combinedwith large trade surpluses have not encouraged firmsto invest as much as they should have, and have led topoor economic decisions; (2) companies have laggedbehind in implementing new technologies and invest-ing capital, largely because of weak demand for inno-vation and insufficient pressure from competition;and (3) political authorities have not been sufficientlyreactive in terms of research and development, inno-vation and education (Ibarra 2010; Ito 2010; Dion2007; Fujii, Candaudap and Gaona 2005; Sharpe andArsenault 2008).
One of the great objectives of Canada and Mexicowas to use the US market as a springboard, and they
both expected free trade to boost competitiveness.Both economies have opened outwards significantlyas the share of trade in GDP increasing from 26 per-cent to 37 percent between the 1980s and 2000s inthe case of Canada, and from 15 percent to 27 per-cent for Mexico. But, this opening has largely beendue to integration trade with the United States andit has not spurred much growth as was expected. Oneshould be cautious in this respect. It would be disin-genuous to say that NAFTA did not producedynamic effects. Quite the contrary, as evidenced bythe very high ranking of Canada’s CompetitivenessIndex of the World Economic Forum (10th) or, inthe case of Mexico, the technological composition ofexports, higher than that of Canada. Even though,the fact remains that both countries suffer from seri-ous weaknesses in innovation and investment, withMexico even receding on the scale of global compet-itiveness (66th in 2010). All in all, if a cause must befound for the difficulties faced by both countries indiversifying their trade, it lies in the shortcomings ofboth governments and businesses and not inNAFTA primarily.
This is indeed one of the great lessons to be learnedfrom the experience of NAFTA. Trade liberalizationoffers windows of opportunity and in this sense,NAFTA, by creating an environment favorable totrade, has allowed Canada and Mexico to seize thisopportunity, and turn North American to theiradvantage. Of course, we cannot compare Canadaand Mexico, still an emerging country which, despitethe reforms, still suffers from economic rigidities,structural weaknesses and severe inequalities. But itis also clear that in order to lead to results a policyof export-led growth must be supported by invest-ments and endogenous growth policies which priori-
CESifo Forum 4/2010 14
Focus
Table 5
Labour productivity, levels, growth rates and contribution to GDP growth
Levels of GDP
productivity 2008
GDP
productivity
growth rate
Contribution to
GDP growth
per
capita
per hour
worked
per hour worked
US dollars, current
prices and PPPs,
US = 100
1995–
2000
2001–
2008
GDP
growth
1985–
2008
Labour Capital Multi-factor
productivity
United States
Canada
Mexico
Chile
OECD
100
83
31
31
71
100
78
34
28
76
2.38
2.26
1.95
–
2.33
1.96
0.72
0.45
1.44
1.70
2.89
2.65
2.80
5.60
–
0.94
1.19
–
–
–
0.86
1.10
–
–
–
1.09
0.37
–
–
–
Source: OECD Factbook 2010: Economic, Environmental and Social Statistics.
CESifo Forum 4/201015
Focus
tize innovation, research and education, without
neglecting the domestic market (Zeda, Wise and
Gallagher 2009).
Conclusion: time for a change?
NAFTA will turn twenty in 2013. It should be given
credit for having significantly strengthened economic
ties among Canada, the United States and Mexico
and thus to have contributed to the revitalization of
national economies. It is, however, far from having
achieved all the objectives that the governments were
aiming at when they signed the agreement. One can
identify gaps and deficiencies in this integration
model, especially in terms of economic and political
convergence as well as in terms of inequalities, which
remains a problem. One can also point to the limits of
a trilateral cooperation that is far too circumscribed
for adequate management of the agreement and to
the fact that it triggered a proliferation of trade agree-
ments that led to a messy international trading sys-
tem. But the focus of this article is not so much to
address these shortcomings of NAFTA as to empha-
size the fact that while well adapted to the context of
the 1990s, NAFTA is much less adapted to the reality
of the 2000s. Is the NAFTA model running out of
breath? One can readily answer ‘YES’.
To put it simply, there is a before 2000 and an after
2000. The institutional model of NAFTA is now
much less convincing than it was in the 1990s. As a
model of integration, it is also weakened by changes
in international economic trends. Both in Canada and
Mexico, the stimulating effects of regional integration
seem exhausted or, at least, insufficient to stimulate
investment and productivity. Economic conditions
have changed: neither Canada nor Mexico can now
count on the US locomotive, let alone surf on a favor-
able exchange rate as before. In both Canada and
Mexico, the dramatic drop in bilateral trade with the
United States in 2009 was cause for concern.
Although trade has indeed resumed vigorously in
2010, these worries remain. Fear, as they say, is the
beginning of wisdom. Or at least, in these spreading
concerns, the crisis should be credited for reminding
us not only that competitiveness is built first at home,
but also that regional integration cannot be reduced
to a simple exercise in trade diplomacy. NAFTA is
certainly out of breath and even outdated, but too
many interests are at stake and the current challenges
are too important to let things go astray (Schott 2008;
Alexandroff, Hufbauer and Lucenti 2008). In this
sense, the crisis offers an opportunity to redefine the
framework of regional cooperation, and to rethink
terms of integration that do not rely primarily on
market forces, as it has been the case until now.
References
Alexandroff, A.S., G.C. Hufbauer and K. Lucenti (2008), “StillAmigos: A Fresh Canada-US Approach to Reviving NAFTA”,Commentary 274, C. D. Howe Institute, Toronto.
Arès, M. (2005), “Du triomphe à l’inquiétude. L’industrie maquiladans la tourmente”, in: Rioux, M. (ed.), Mondialisation et pouvoir desentreprises, Collection Économie politique internationale, Montréal:Éditions Athéna, 215–242.
Beeson, M. (2007), Regionalism and Globalization in East Asia:Politics, Security and Economic Development, Houndsmill: PalgraveMacmillan.
Blecker, R.A. (2009), “External Shock, Structural Changes, andEconomic Growth in Mexico, 1979–2007”, World Development 37,1274–1284.
Brunelle, D. and C. Deblock (1989), Le libre-échange par défaut,Montreal: VLB éditeur.
Courchene, T (2003), “FTA at 15, NAFTA at 10: A CanadianPerspective on North American Integration”, The North AmericanJournal of Economics and Finance 14, 263–285.
Deblock, C. (2007), “Il était une fois dans les Amériques... Le projetenvolé de zone de libre-échange”, Europa & America Latina 2,105–132.
Deblock, C. and S. Turcotte (2003, eds.), Suivre les États-Unis ouprendre une autre voie? Diplomatie commerciale et dynamiquesrégionales au temps de la mondialisation, Bruxelles: Éditions Bruylant.
Dion, R. (2007), “Interpreting Canada’s Productivity Performance inthe Past Decade: Lessons from Recent Research”, Bank of CanadaReview, Summer, 19–32.
Faux, J.,C. Salas and R.E. Scott (2006), Revisiting NAFTA: Still NotWorking for North America’s Workers, EPI Briefing Paper, EconomicPolicy Institute, Washington DC.
Fujii, G., E. Candaudap and C. Gaona (2005), “Salarios, productivi-dad y competitividad de la industria manufacturera Mexicana”,Comercio Exterior 55, 16–28.
Gagné, G. (2001), “The Investor-State Provisions in the AbortedMAI and in NAFTA: Issues and Prospects”, The Journal of WorldInvestment 2, 481–505.
Goldfarg, D. (2006), “Too Many Eggs in One Basket? EvaluatingCanada’s Need to Diversify Trade”, Commentary 236, C. D. HoweInstitute, Toronto.
Hufbauer, G. (2008), “NAFTA’s Bad Rap”, The InternationalEconomy, Summer, 19–23.
Ibarra, C.A. (2010), “Exporting without Growing: Investment, RealCurrency Appreciation, and Export-Led Growth in Mexico”, TheJournal of International Trade & Economic Development 19, 439–464.
Ito, T. (2010), “NAFTA and Productivity Convergence betweenMexico and the US”, Cuadernos de Economia 47/May, 15–55.
Lawrence, R.W. (1996), Regionalism, Multilateralism, and DeeperIntegration, Washington DC: The Brookings Institution.
Meza, L. and J. Salvador (2009), “La dinámica del comercio exteriorde México y China”, Comercio Exterior 59, 615–628.
Morales, I. (2010), Post-NAFTA North America, Houndmills:Palgrave Macmillan.
Pacheco-López, P. (2008), “Foreign Direct Investment, Exports andImports in Mexico”, World Economy 28, 1157–1172.
Rosales, O. and M. Kuwayama (2007), “Latin America Meets Chinaand India: Prospects and Challenges for Trade and Investment”,CEPAL Review 93, 81–103.
Schott, J.J. (2008), The North American Free Trade Agreement: Timefor a Change?, Washington DC: Peterson Institute for InternationalEconomics.
Sharpe, A. and J.F. Arsenault (2008), The Canada-US ICT InvestmentGap: An Update, CSLS Research Report 2008-1, Ottawa.
Waldkirch, A. (2008), The Effects of Foreign Direct Investment inMexico since NAFTA, MPRA Paper 7975.
Weintraub, S. (1990), A Marriage of Convenience: Relations betweenMexico and the United States, New York: Oxford University Press.
Zeda, E., T.A. Wise and K.P. Gallagher (2009), “Rethinking TradePolicy for Development: Lessons from Mexico under NAFTA”,Policy Outlook, Carnegie Endowment for International Peace,Washington DC.
CESifo Forum 4/2010 16
Focus
CESifo Forum 4/201017
Focus
NAFTA, TRADE AND
DEVELOPMENT
ROBERT A. BLECKER* AND
GERARDO ESQUIVEL**
Introduction
In 1990, Mexican President Carlos Salinas de Gortariand US President George H.W. Bush announced theirintention to sign a free trade agreement (FTA)between their two countries. After the government ofCanada joined this effort a year later, the three coun-tries began negotiations to establish what was thencalled the largest free trade area in the world. Themere announcement of this effort, which led to theformation of the North American Free TradeAgreement (NAFTA) in 1994, immediately sparked aseries of debates along multiple dimensions. InMexico and the United States, the domestic discus-sion focused on the economic impact of engaging inan FTA with a highly asymmetrical partner that couldeither destroy domestic industries due to its techno-logical superiority (as was expected in Mexico) or gen-erate massive job losses in response to lower wagesacross the border (as was expected in the UnitedStates). Of course, proponents of NAFTA dismissedthese fears and instead emphasized the benefits of freetrade in terms of efficiency, productivity and greatervariety of products for consumers.1
At a more general level, the announcement of the cre-ation of NAFTA also generated a heated debateabout regionalism versus multilateralism in tradenegotiations, since it was seen as a step forward in thecreation of a large trading bloc that could affect nego-
tiations leading towards full multilateral trade liberal-
ization. From a slightly different perspective, NAFTA
was seen as a geopolitical move that was a natural US
response to European integration efforts, and
although it was probably not going to have a pro-
found impact on the US economy, it could have one
on the Mexican. NAFTA was also seen by many as a
radical departure from the protectionist policies tradi-
tionally followed by Mexican policy makers that
could eventually lead to closing the historical devel-
opment gap between Mexico and its northern neigh-
bors. For that reason, NAFTA generated huge expec-
tations among analysts interested in understanding
the economic impact of an FTA between highly
asymmetrical partners.
In this paper, we analyze the expectations and the
realities about the economic impact of NAFTA on
Mexico in terms of economic convergence, trade,
investment, employment, wages, and income distrib-
ution. We show that NAFTA has basically failed to
fulfill the promise of closing the Mexico-US devel-
opment gap, and we argue that this was due in part
to the lack of deeper forms of regional integration or
cooperation between Mexico and the United States.
We also explore other factors that could explain this
negative outcome, and we briefly discuss the oppor-
tunities for both Mexico and the United States to
mutually benefit from a further economic integra-
tion process.
NAFTA expectations and realities
The Mexican government had pinned its hopes on
NAFTA not merely to boost exports to the US and
Canadian markets, but also to attract large amounts
of foreign direct investment (FDI), create a significant
number of new industrial jobs, and give the Mexican
economy the growth stimulus it had been lacking
since the tepid recovery from the debt crisis of the
1980s (Lustig 1998). President Salinas famously pre-
dicted that NAFTA would permit Mexico to ‘export
goods, not people’ and to join the ranks of ‘first-
world’ nations. NAFTA’s critics in the United States
predicted that it would cause a massive relocation of
* American University, Washington DC.** El Colegio de México, Mexico City.An earlier version of this paper was presented at the workshop onMexico and the United States: Confronting the Twenty-First Centuryheld at the Center for US-Mexican Studies, University of California,San Diego, July 2009. The workshop was co-sponsored by theMexico Institute of the Woodrow Wilson Center (Washington), ElColegio de la Frontera Norte (Tijuana), and El Colegio de México(Mexico City), along with the Center. The authors would like tothank the workshop participants for their helpful comments andsuggestions.1 For a discussion of the US debate about NAFTA around the timeof its passage, see Cohen, Blecker and Whitney (2003).
US industries and jobs to Mexico, while fosteringgreater inequality in both societies by creating a ‘raceto the bottom’ in social and labor standards. NAFTAsupporters in turn promised that the agreement wouldstimulate US employment via trade surpluses with agrowing Mexican market. Paradoxically, NAFTA’soriginal supporters and opponents seemed to agreethat, whatever else it would do, this agreement wouldgive a major impetus to Mexico’s industrial develop-ment and job creation.
Of course, NAFTA did not go into effect in a vacu-um, and it is perilously difficult to disentangleexactly what were the effects of this trade agreementrelative to other factors in the post-1994 evolutionof the two economies. NAFTA built upon the baseof the much larger tariff reductions and more far-reaching market-opening measures that Mexico hadalready adopted unilaterally after it joined theGeneral Agreement on Tariffs and Trade (GATT) in1986, so not all of the effects of trade liberalizationcan be attributed to NAFTA.2 In addition, macro-economic factors such as financial crises, exchangerates, oil prices, and business cycles were importantdeterminants of what actually occurred.3 Sub-sequent trade agreements, both multilateral (the for-mation of the World Trade Organization) and pref-erential (the many other FTAs entered into sepa-rately by Mexico and the United States), reducedthe significance of the tariff preferences containedin NAFTA. China’s emergence as a global econom-ic power and the rapid increase in its share of NorthAmerican markets have also had an enormousimpact on the region.
The fact that NAFTA was never supplemented bydeeper forms of regional integration, social policies,or economic cooperation probably limited the bene-fits and exacerbated the costs (Pastor 2001; Studerand Wise 2007). Domestic policies in both nationsmattered, as did underlying geographic and demo-graphic realities. US efforts to stem unauthorizedimmigration, coupled with post-September 11 securi-ty measures, have made the border tougher, not easi-er, to cross, even for legal goods and services. As aresult of all these factors, it is safer to analyze whathappened after NAFTA than what happened because
of NAFTA, but we will try to draw some inferencesabout causality where the evidence permits.
In fact, the trajectories of the US and Mexicaneconomies after NAFTA bear little resemblance toany of the more exaggerated forecasts on either side.NAFTA did not solve Mexico’s employment prob-lems, raise its average real wages, or reduce migrationflows, and it seems to have done little to raise thecountry’s long-run average growth rate, although itcontributed to a strong recovery from the 1994–1995peso crisis and a short-lived boom in 1996–2000.Mexico did reap gains in exports, FDI, and other indi-cators, especially in the late 1990s, but NAFTA didnot turn out to be the panacea promised by theSalinas administration.
The United States did not suffer a catastrophic lossof manufacturing employment immediately afterNAFTA went into effect, although it began to hem-orrhage manufacturing jobs more severely after theAsian financial crisis of 1997–1998 and the surge inChinese imports beginning around 2001. US work-ers made significant real wage gains in the late 1990sin spite of increasing US trade with Mexico, whileMexican workers suffered a sharp decline in realwages following the 1994–1995 peso crisis that wasonly barely reversed by the early 2000s. During thefirst seven years of NAFTA (1994–2000), NorthAmerica showed signs of becoming a more integrat-ed and competitive regional market area, but muchof the progress on the regional front was reversed inthe next eight years (2001–2008), as we shall seebelow.
NAFTA and economic convergence
When NAFTA was signed, one of the main objectivesof the agreement (at least from the Mexican perspec-tive) was to achieve a reduction in the historical gap ofeconomic development between Mexico and theUnited States. Despite of all the anti-Americanrhetoric traditionally displayed by Mexican politi-cians4, the truth is that many Mexicans have longaimed to benefit from being close to one of the biggestand richest markets in the world. Of course, thisexplains not only the large flows of migrants fromMexico to the United States, but also the close tradeties that have been established historically between thetwo countries. In that sense, when NAFTA was signed
CESifo Forum 4/2010 18
Focus
2 Furthermore, NAFTA contained many provisions that wentbeyond trade liberalization, such as guarantees of property rights forforeign investors, and was also intended to lock-in Mexico’s previousmarket reforms. In this sense, NAFTA may have had consequencesbeyond the direct effects of the reductions in trade barriers it con-tained.3 For an analysis of the impact of these macroeconomic factors onMexico, see Blecker (2009).
4 Remember, for example, the famous expression attributed toMexican dictator Porfirio Díaz: “Poor Mexico, so far from God, andso close to the United States!”
CESifo Forum 4/201019
Focus
there were huge expectations thattrade and FDI could help toreduce the Mexico-US economicgap. From the perspectives ofboth countries, this could bringabout multiple benefits for every-one involved in the agreement:for Mexican workers, this wouldimply higher wages and a betterstandard of living; for Ameri-cans, this would imply having amore stable and economicallysound neighbor that could alsobecome a good client for US-made products. Under this sce-nario, Mexican workers wouldhave lower incentives to migrateand, since migration has always generated a heateddebate in some segments of the US population, thiscould also help to ease tensions in the Mexican-USrelationship. All in all, NAFTA would be a win-winsituation.
The relevant question, then, is what has happened tothe historical Mexico-US economic development gapsince (or as a result of) NAFTA? Has there been eco-nomic convergence between Mexico and the UnitedStates since (or as a result of) NAFTA?5
Figure 1 provides the answer to these questions.The graph shows alternative long-term measures ofincome per capita or income per worker in Mexicoas percentages of the corresponding measures inthe United States. The data are shown in relativeterms to better capture the idea of economic con-vergence: if income per capita in Mexico increasesrelative to that in the United States, the relativevariables will rise and we would then conclude thatthere was a process of economic convergencebetween the two countries. Otherwise, we would saythat there was no convergence. Indeed, if the rela-tive variables decline, we would then say that therewas a process of economic divergence between thecountries.
Figure 1 shows data from two different sources: theWorld Bank (WB) and the Penn World Tables v. 6.3(PWT), and it shows two indicators from each source.From the WB we use the series on GDP per capita inpurchasing power parity terms (PPP), which adjustsfor price differentials across countries (this is theseries WB GDP per capita, PPP) as well as the serieswithout adjustment (WB GDP per capita).6 Bothvariables are measured as ratios of data in currentprices. From the PWT we also use two series: the firstis the Mexico-US ratio of real income per capita(PWT real GDP per capita) and the second is the ratioof real GDP per worker (PWT real GDP per worker).Both variables are measured as ratios of data in con-stant prices.
All four series show essentially the same result: the levelof economic development in Mexico relative to theUnited States has been remarkably stable since 1995,which means that there has been no economic conver-gence between these two countries as a result of (orassociated with) NAFTA. Notice that, even after therecovery from the 1994–1995 crisis, the level of eco-nomic development in Mexico relative to the UnitedStates (in either per capita or per worker terms)remained slightly below what it was before the passageof NAFTA. The data in Figure 1 show that Mexico’sincome per capita is about one fourth (WB, PPP terms)or one third (PWT) of the US income per capita,depending on which source we use. The figure alsoshows that output per worker in Mexico relative to theUnited States has steadily declined since 1981 and thatit is now just slightly above 30 percent. For comparison
5 Note that the term economic convergence as used in this paper isdifferent from how the term has been used in some other studies, par-ticularly Lederman, Mahoney and Servén (2005), and Haber, Klein,Maurer and Middlebrook (2008). These studies utilize a counterfac-tual analysis of the type: is Mexico better-off with NAFTA thanwithout it? Or, could the Mexico-US gap have been greater in theabsence of NAFTA? The former study includes a time-series analy-sis that investigates whether Mexico is converging toward a constantper capita income differential with the US (i.e. 50 percent of the USlevel). Economic convergence for us means a reduction in theabsolute Mexico-US gap in terms of the variables that affect eco-nomic well-being, such as income per capita, average wages, andlabor productivity, and the eventual approach of Mexico to US lev-els of these variables.
0
10
20
30
40
50
60
70
80
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005
WB GDP per capita WB GDP per capita, PPP PWT real GDP per capitaPWT real GDP per worker
Sources: Penn World Tables (PWT) 6.3; World Bank, World Development Indicators.
MEXICO S ECONOMIC PERFORMANCE RELATIVE TO THE US% NAFTA
,Figure 1
6 We include the latter variable only as a reference. It is more accurateto use PPP data when making international comparisons of livingstandards.
purposes, note that this ratio wasclose to 40 percent in 1993, beforeNAFTA went into effect.
In sum, the data show that therehas been no economic conver-gence whatsoever between Mexi-co and the United States sinceNAFTA’s enactment. As a result,the historical Mexico-US eco-nomic gap in percentage termshas not been reduced after15 years of free trade, and theincentives to migrate are proba-bly even greater than before sincethe income gap in absolute termsis now larger that it was 15 yearsago. Furthermore, since therecent international financial crisis has affectedMexico more than any other country in the WesternHemisphere7, we can anticipate that the Mexico-USgap increased even further in 2008–2009.
Trade and investment flows
This lack of convergence did not occur because of afailure of trade to grow faster after NAFTA went intoeffect. On the contrary, Table 1 shows that US non-petroleum imports from Mexico accelerated to anaverage annual growth rate of 19.5 percent in the firstseven years of NAFTA (1993–2000), after growing atan already rapid clip of 13.9 percent in 1987–1993 fol-lowing Mexico’s unilateral liberalization. As a resultof this faster growth, Mexico’s share of US non-petroleum imports climbed from 6.7 percent in 1993to 11.4 percent in 2000. The accelerated growth in1993–2000 should not be attributed entirely toNAFTA, however, but also resulted from two otherfactors: the ‘new economy’ boom in the United Statesin the late 1990s, which led to an enormous explosionof US demand for imports generally; and the depreci-ation of the Mexican peso following the 1994–1995peso crisis, which left the peso at a more competitiveexchange rate for the next several years.
However, US import growth from Mexico slowedconsiderably after 2000. US non-petroleum importsfrom Mexico grew only at a 4.9 percent annual rate in2000–2008, while US imports from China continued
to soar at a torrid 16.4 percent annual pace duringthat period. To be sure – and this is where bothNAFTA and geography may have helped – Mexicosucceeded in maintaining its US market share betterthan other global regions in the 2000–2008 period.The 11.3 percentage point increase in the Chineseshare of US non-petroleum imports during this peri-od came mostly at the expense of other countries,while Mexico’s share dipped only slightly.Nevertheless, it is likely that US imports from Mexicowould have grown much faster and increased theirshare further after 2000 in the absence of the rapidinflux of imports from China.8
The disappointing growth of Mexican exports to theUnited States in 2000–2008 occurred after Chinajoined the World Trade Organization (WTO) andobtained ‘permanent normal trade relations’ (former-ly known as ‘most favored nation’) status from theUnited States in 2001. However, other factors werealso at work. As part of its inflation-targeting mone-tary policy, the Mexican government allowed thevalue of the peso to rise significantly in the early2000s.9 The end of the Multifibre Arrangement(MFA) in 2005 led other developing countries (large-ly, but not exclusively, China) to increase their sharesof global textile and apparel production, therebydestroying a large part of the vertically integratedNorth American textile-apparel complex that flour-ished briefly under NAFTA’s rules of origin in the late
CESifo Forum 4/2010 20
Focus
Table 1
US non-petroleum imports from Mexico, China and other countries
Percentage share in total US non-petroleum imports
1987 1993 2000 2008
Mexico 4.5 6.7 11.4 11.1China 1.7 5.9 9.0 20.3Other countries 93.8 87.4 79.6 68.6Total 100.0 100.0 100.0 100.0
Growth (average annual percentage rates)
1987–
1993
1993–
2000
2000–
2008
Mexico 13.9 19.5 4.9China 30.8 17.9 16.4Other countries 5.4 9.4 3.3Total 6.6 10.9 5.2
Sources: US Bureau of Economic Analysis, International Transactions
Accounts; Petróleos Mexicanos (PEMEX), Anuario Estadístico (various
years); authors’ calculations.
7 Mexico’s per capita GDP in 2009 is estimated to fall by about 8 percent, whereas US per capita GDP is expected to fall by at most3 percent.
8 For evidence of significant displacement of Mexican exports byChinese exports, see Gallagher, Moreno-Brid and Porzecanski (2008),Hanson and Robertson (2009) and Feenstra and Kee (2006).9 See Galindo and Ros (2008) for evidence that the Banco deMéxico’s monetary policy was biased toward permitting the peso toappreciate in the early 2000s.
CESifo Forum 4/201021
Focus
1990s. High-tech producers alsodiscovered that they could findlower wages and more supportivegovernment policies in variousEast Asian countries (Gallagherand Zarsky 2007).
Mexico’s trade data show a simi-lar pattern of regional integrationincreasing during the 1993–2000period and then diminishingthereafter (see Table 2). On theexport side, the largest increase inthe US share of Mexican exportsoccurred in 1987–1993; this sug-gests the natural pull of geogra-phy in stimulating intra-regionaltrade even when Mexico openedup its own economy unilaterally.10
In spite of efforts by Mexico todiversify its export outlets, espe-cially through the signing ofnumerous other bilateral FTAs,80.2 percent of Mexican exportswere still sold in the US market asof 2008.
In contrast, the US share ofMexican imports remained rela-tively stable at around 70 percentfrom 1987 through 2000, andthen fell abruptly to 49 percent in2008. There were several causesof this sharp reduction in intra-regional trade post-2000. First,since both the US dollar andMexican peso were at relativelyhigh values during the first sever-al years of the 2000s, producersthroughout North America hadstrong incentives to source prod-ucts (both final and intermediategoods) outside the continent. Second, the penetrationof Chinese and other Asian imports into the US mar-ket not only displaced Mexican exports to the UnitedStates, but also displaced US exports of intermediategoods that would otherwise have been shipped intoMexico for assembly. Third, Mexican trade policyactively encouraged imports of intermediate goods
from outside the region through the Pitex program of
tariff exemptions.
Mexico did succeed in attracting a notably increased
average level of FDI inflows after NAFTA went into
effect in 1994 (Table 3). FDI inflows did not increase
continuously, however, but rather stabilized at an
average level of about 3 percent of GDP in the post-
NAFTA period. Interestingly, the proportion of US
FDI outflows that go to Mexico has not varied much
since the pre-NAFTA period (1987–1993), especially
Table 2
Country composition of Mexico’s external trade
(% share of total trade)a)
1987b)
1993 2000 2008c)
Exports: destination country
United States 69.2 82.7 88.7 80.2Canada 1.1 3.0 2.0 2.4China na 0.1 0.1 0.7Rest of world 29.7 14.2 9.1 16.7
Imports: country of origin
United States 74.0 69.3 73.1 49.0Canada 1.7 1.8 2.3 3.1China 0.2 0.6 1.7 11.2Other Asia 4.5 10.7 10.0 16.7Rest of world 19.6 17.6 13.0 20.0
a) Including maquiladora industries. –
b) The US percentages for 1987
were taken from Hufbauer and Schott (1992), Table 3.1, based on IMF,
Direction of Trade Statistics; 1987 data for other countries were esti-
mated using data from INEGI, Anuario Estadístico de los Estados
Unidos Mexicanos 95 (Aguascalientes: INEGI 1996) in combination
with Hufbauer and Schott’s percentages for the United States. –c)
Pre-
liminary figures.
Sources: Instituto Nacional de Estadística, Geografía e Informática
(INEGI) except for 1987; authors’ calculations.
10 It may seem paradoxical that Mexico’s liberalization of importsmade its exports to the United States grow so rapidly, but this seem-ing paradox is readily explained by the fact that the export productswere very intensive in imported intermediate goods, and also becauserestrictions on FDI were liberalized around the same time.
Table 3
Average inflows of foreign direct investment into Mexico
1987–
1993
1994–
2000
2001–
2007
Total inflows of FDI into Mexico
in billions of US dollars 3.2 12.4 22.4
as a share of Mexico’s GDP (%) 1.1 3.0 2.9
Inflows of FDI from
the United States into Mexico
in billions of US dollars 1.6 4.6 8.7
as a share of total US outflows
of FDI (%) 3.6 3.7 4.4a)
as a share of total FDI inflows
into Mexico (%) 61.0 61.7 54.7 a)
Excluding 2005 when the total was very low due to a large
adjustment for exchange rate changes; if we also exclude 2001 (when
Citibank bought Banamex) this figure would be 3.3%. If we include
both 2001 and 2005, the average for all years 2001–2007 amounts
to 7.5%.
Sources: IMF, International Financial Statistics; US BEA; INEGI; author’s
calculations.
if we discount one unusually high year (2001, when Citibankacquired Banamex). The propor-tion of Mexican FDI inflowscoming from the United Statesfell in 2001–2007 compared withthe earlier periods shown, evenincluding 2001. Thus, Mexicohad more success in attractingadditional FDI from countriesoutside North America than itdid in attracting a larger share ofUS FDI outflows – and this mayalso have contributed to the fall-off in the US share of Mexico’simports after 2000.
Effects on manufacturing employment
US manufacturing employment did fall off a cliff –but not until after 2001, seven years after NAFTAwent into effect (see Figure 2). Roughly three millionmanufacturing jobs disappeared following the 2001recession and China’s accession to the WTO in thatyear, and another two million vanished in the finan-cial crisis and steep recession of 2008–2009. None ofthese events had anything to do with NAFTA orMexico, however, and, as we shall see below,Mexican manufacturing employment also fell inboth periods.
Nevertheless, it does not follow that NAFTA or US-Mexican trade had no negative impact on US manu-facturing employment, which might have beenexpected to have grown more during the economicboom of 1994–2000 than it actually did. US manu-facturing employment rose very little during thatperiod, in spite of GDP growth that averaged3.9 percent per year at that time. However, the high-est credible estimate of the cumulative US manufac-turing job losses that can be attributed to US-Mexican trade during (roughly) the first decade ofNAFTA is about 500,000, and other estimates arelower (some even claim net gains).11 Even taking thehigh-end estimate of about a half million jobs lostover a decade, it is a relatively small amount in acountry where payroll employment totaled 114 mil-lion in 1994 and reached 138 million in 2007, andsmaller than the monthly job losses during the worstof the recession of 2008–2009.12 Moreover, the500,000 figure is an estimate of job losses due to theincreased US trade deficit with Mexico, not effects ofNAFTA specifically.
If the US job losses that can credibly be attributed totrade with Mexico (if not to NAFTA per se) are rela-tively small, by the same token the employmentincreases that Mexico achieved in its tradable goodsindustries were much more modest than the moreoptimistic ex ante predictions. Total payroll employ-ment in Mexican manufacturing increased from2.5 million in 1989 to 2.9 million in 1994, and rose fur-ther to 3.8 million in 1999, but then declined to3.4 million in 2004.13 Overall, the net increase in man-ufacturing payroll employment in Mexico in the firstdecade after NAFTA (1994–2004) was roughly500,000 – perhaps coincidentally, just about the highend of the estimates of US job losses over the sameperiod. This a far cry from an amount of job creationthat could have put a serious dent in Mexico’s employ-ment needs (given that the labor force grows by near-ly 1 million workers annually) or stem the flow of emi-gration (which is estimated to have been in the rangeof about 350,000 to 580,000 per year in the 1990s andearly 2000s).14
CESifo Forum 4/2010 22
Focus
11 The high-end estimate comes from Scott, Salas and Campbell(2006), who calculate a net loss of 559,564 US jobs between 1993 and2004 as a result of the increased US trade deficit with Mexico. Incontrast, Hufbauer and Schott (2005) report that 366,000 US work-ers received certification of NAFTA-related job losses under theNAFTA Trade Adjustment Assistance (TAA) program between1994 and 2002; this number also includes jobs lost to Canada.Hufbauer and Schott (2005) argue that these job losses were morethan offset by gains in ‘jobs supported by exports’, but the latter arenot estimated by the same methodology used to calculate the joblosses. Neither study isolates effects of NAFTA as opposed to otherfactors.12 Total payroll employment data are from US Council of EconomicAdvisers (2009), Table B-46. The estimated job losses do loom larg-er relative to manufacturing employment, which was about 17 mil-lion in 1994 (see Figure 2).13 These data are from the Mexican Economic Census, which is con-ducted every five years; data for 2009 had not been released at thetime of this writing. These data were obtained from INEGI. OtherINEGI data, which are available on a monthly basis, show that mostof the job creation in manufacturing in the 1990s occurred in theexport-oriented maquiladora plants.14 The migration estimates are from Hanson (2006).
10
12
14
16
18
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Source: US Bureau of Labour Statistics.
TOTAL EMPLOYMENT IN US MANUFACTURUNG SECTORJanuary 1990–May 2009
seasonally adjusted Million
NAFTA
Finacial crisis
ChinajoinsWTO
Figure 2
CESifo Forum 4/201023
Focus
In retrospect, it should have been more obvious thattrade liberalization would not have had an enormousimpact on total industrial employment in Mexico.Trade liberalization increases imports as well asexports, and increased imports displace domestic jobsjust as much as increased exports create them. Thus,an important perspective on the disappointing jobgains in Mexico’s manufacturing industries can beobtained by examining the country’s trade balanceswith the United States and the rest of the world.While for the United States its growing deficit withMexico was part of a much larger increase in its over-all deficit, for Mexico its increasing surplus with theUnited States was completely offset by rising deficitswith other countries, primarily in Asia (see Figure 3).Furthermore, many Mexican export industries areessentially assembly operations that rely heavily onimported parts and components, and which lack‘backward linkages’ to domestic industries (Ruiz-Nápoles 2004; Moreno-Brid, Santamaría and Rivas
Valdivia 2005). As a result, theincreases in the gross value ofexports are an exaggerated indi-cator of value added and employ-ment generation in the exportindustries.
Income distribution, relative
wages and inequality
Figure 4 shows one of the mostwidely cited indicators of wageinequality, the skilled-unskilledwage gap, measured by the ratioof salaries of employees (non-production workers, in the USterminology) to wages of produc-
tion workers, from the monthly survey of non-maquiladora industries in Mexico. The sharp rise inthis measure of wage inequality in the first decade oftrade liberalization (1987–1997) surprised most econ-omists, since they had assumed that trade liberaliza-tion would boost the wages of less-skilled workers inMexico due to a supposed abundance of less-skilledlabor. One explanation for the rise in this ratio at thattime is that the initial tariffs which were lowered in thetrade liberalization of the late 1980s were higher in theindustries that were most intensive in less-skilled labor(Revenga and Montenegro 1998). Another explana-tion is that skill-biased technological change duringthis period boosted demand for more educated work-ers – although this shift may have been at least par-tially an effect of trade liberalization rather than anindependent cause.15
Of course, a rise in wage inequality that began sever-al years before NAFTA cannot be attributed to this
trade agreement. After NAFTAwent into effect, this measure ofwage inequality stopped increas-ing and turned gradually down-ward from 1997–2007, althoughas of 2007 it remained 34 percentabove its 1987 level. While thereare probably several causes ofthis reversal, the leading explana-tion is an increase in the relative
- 100
- 75
- 50
- 25
0
25
50
75
100
1993 1995 1997 1999 2001 2003 2005 2007Source: INEGI.
MEXICO S TRADE BALANCE: TOTAL AND WITH THE US, ASIA AND ALL NON-US COUNTRIES
1993–2008Billion US dollars
US
Asia
Total
All non-UScountries
,Figure 3
1.5
2.0
2.5
3.0
3.5
1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007
Sources: INEGI; authors’ calculations. (The old survey, based on 129 classes of economic activity, and the new one, based on 205 classes, were spliced together in 1994, which was the one year of overlap.)
RATIO OF SALARIES OF EMPLOYEES TO WAGES OF PRODUCTION WORKERS IN THE NON-MAQUILADORA MANUFATURING INDUSTRIES OF
MEXICORatio of salaries to wages
Figure 4
15 See Esquivel and Rodríguez-López(2003) and Verhoogen (2008). Feenstra(2006) argues that trade liberalization‘selects’ for more efficient industrial firmsand plants, resulting in increases in aver-age productivity and decreases in the useof less-skilled labor as less efficient firmsand plants are eliminated and more effi-cient ones expand.
supply of more-skilled labor dueto the rising levels of educationof the Mexican labor force.16
The changes in wage inequality inMexico also have important re-gional and gender dimensions.17
Census data reveal that regionalinequality between workers in thenorthern and southern Mexicanstates increased between 1990 and2000. For the more recent period,studies have found that thedecreases in the skill gap in thelate 1990s and early 2000s wereconcentrated in the northern bor-der states, which have the highestdegree of ‘globalization’ according to various indica-tors of exports and FDI. Furthermore, the decrease inthe skill gap in the last decade occurred almost exclu-sively among women workers in those states. In the restof the country, where the effects of imports are likely todominate the effects of exports and where there hasbeen relatively less FDI, less-skilled workers (of eithergender) do not appear to have benefited as much fromtrade liberalization either pre- or post-NAFTA.
Thus, it is difficult to generalize about the effects oftrade liberalization or NAFTA on Mexico’s wagestructure, as there were many effects that went in dif-ferent directions for different groups of workers andregions of the country at different times (and not allof the distributional changes were caused by tradepolicy). If anything, the evidence seems more clear-cut that the initial liberalization contributed to therise in wage inequality from 1987–1997, whileNAFTA’s effects are more muted and mixed. This isnot surprising, since the earlier liberalization involveda more drastic opening of Mexico’s economy com-pared to NAFTA.18
However, there are other dimensions of income distri-bution that can be affected by trade policy beyond therelative wages of more- and less-skilled workers,which have received perhaps disproportionate atten-
tion from economists on both sides of the border.What Mexico hoped for when it opened its economyand joined NAFTA was not merely a reduction ininequality among different groups of workers, butmore importantly a significant increase in the averagewage level for all Mexican workers. This, in turn,would have contributed to a rising standard of livingfor most citizens and a diminution of outward migra-tion. This simply has not come to pass, especially inthe tradable goods industries that are most impactedby trade.
Figure 5 shows an index of Mexico’s average realcompensation per person in manufacturing since1980.19 Evidently, this index has followed the cycles inthe Mexican economy, as real compensation collapsedduring the debt crisis of the early 1980s, partiallyrecovered in 1988–1994, collapsed again following thepeso crisis in 1995–1996, and recovered once more inabout 1998–2003. However, average real compensa-tion stagnated in the last five years shown(2003–2008), and at the end of this period was barelyback to its pre-crisis level of 1994. In the long run,average real labor compensation in Mexican manufac-turing has not increased since the debt crisis of theearly 1980s. Since average US wages rose during thisperiod, the wage gap with the United States increasedrather than decreased.
In hindsight, the expectations of significant overallwage gains for Mexican workers as a result of tradeliberalization alone were surely unrealistic. The pre-diction that Mexican workers in general – and less-skilled workers in particular – would benefit from
CESifo Forum 4/2010 24
Focus
16 See Esquivel, Lustig and Scott (2010). In addition, López-Córdova(2004) shows that average Mexican tariffs remained higher on goodsthat were more intensive in less-skilled labor after trade liberalizationand NAFTA. Also, Robertson (2007) cites the rising proportion ofmaquiladora employment in total manufacturing employment asindicating an increase in the relative demand for less-skilled labor.17 This paragraph draws on the following sources: Hanson (2004),Borraz and López-Córdova (2007), and Chiquiar (2008).18 According to USITC (1991), the average tariffs in effect at thetime NAFTA was adopted were about 3.4 percent for US importsfrom Mexico and 10 percent for Mexican imports from the UnitedStates.
0
20
40
60
80
100
120
140
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Source: Banco de México. Data wer seasonaly adjusted by the authors.
1993 = 100 NAFTA
REAL COMPENSATION PER WORKER IN MEXICO S NON-MAQUILADORA MANUFATURING INDUSTRIES
,Figure 5
19 Compensation (‘remuneraciones’ in Spanish) includes fringe bene-fits in addition to wages or salaries.
CESifo Forum 4/201025
Focus
trade liberalization hinged on the assumption that
Mexico had a relative abundance of (less-skilled)
labor compared with its trading partners. Although
this is true in regional terms, i.e. in comparison with
Canada and the United States, it is not true in global
terms, i.e. in a world economy that includes the much
more labor-abundant countries of South and East
Asia. Mexico is close to the world average in terms of
labor abundance, in-between highly labor-abundant
countries like China and India on the one side, and
relatively labor-scarce countries like the United States
and Canada on the other (Blecker 2010). Similarly,
although Mexico is the low-wage country in North
America, it is a medium-wage country globally
(Leamer 1998). Thus, Mexico does not have a global
advantage in labor costs and should not have been
expected to reap large gains in wages from opening up
to trade, except in those sectors where the country can
parlay its geographic proximity to the US market into
special competitive advantages.
Why Mexico is not converging?
In addition to what has already been mentioned, there
are a number of domestic factors that explain why
Mexico is not converging to US levels in terms of
income per capita, income per worker, or average
wages. Among other aspects, we can mention the fol-
lowing: (1) badly implemented economic reforms,
which instead of promoting economic growth have
actually been a drag on it; (2) lack of other important
economic reforms in areas such as rule of law, compe-
tition, financial sector, education, infrastructure, etc.;
(3) lack of a domestic engine that could complement
the external one (mainly represented by the US indus-
trial sector and consumer market); and (4) restrictive
macroeconomic policies. Let us review each of these
aspects in more detail.
Badly implemented economic reforms
In the second half of the 1980s and the early 1990s,
Mexico undertook a series of economic reforms
(trade opening, financial reform, and privatization of
banks, highways, etc.) that were supposed to radically
transform the semi-closed, inward-looking Mexican
economy into a more modern and export-oriented
one. Some of these reforms, however, were badly
implemented and led to disastrous outcomes that in
some cases were the opposite of what the policies were
supposed to achieve (Esquivel and Hernández-Trillo
2009). The privatization of banks, for example, was
done without having a proper institutional and regu-latory framework, which then led to an unsustainablecredit boom that exacerbated the costs associatedwith the currency crisis of December 1994 (the so-called Tequila crisis). Something similar happenedwith the privatized highways, which were subsequent-ly bailed-out by the Mexican government at anextremely high cost. Other privatizations, such as thatof the state telephone company Telmex, only replaceda public monopoly with a private one, which has sincethen extracted huge rents from a captive and mostlyuncontested domestic market (Del Villar 2009).
Lack of other important economic reforms
The negative outcomes of some of the previous eco-nomic reforms, together with political gridlock in thenewly multi-party Congress (since 1997), have led to areform paralysis in the country. In fact, since the mid-1990s there have been no new important economicreforms in Mexico, despite the fact that everyoneacknowledges the importance of undertaking certainchanges in the economy. Of course, some of thesereforms are highly controversial and there wouldhardly be a consensus on some of them, as in the caseof fiscal or labor reform, where the approaches andproposed solutions of different political parties arecompletely different. However, there are certainreforms that could be easily approved and implement-ed and that would not engender ideological con-frontation among the different political parties,although they would undoubtedly affect some specialinterests groups. So far, these groups have been suc-cessful in blocking or even avoiding discussion ofthese reforms, which include the rule of law, competi-tion policy, and financial regulation.
Lack of a domestic engine
One thing that has definitely changed since NAFTA isthe increasing correlation of Mexican and US busi-ness cycles, presumably reflecting greater sensitivity ofthe Mexican economy to short-run fluctuations in theUS economy. Several studies using a variety of statis-tical methodologies have found large and significantincreases in the ‘synchronization’ of Mexican outputgrowth and industrial production with the corre-sponding US variables since NAFTA.20 Figure 6 con-firms graphically that Mexican GDP growth has beenhighly correlated with US GDP growth since 1994,except for 1995 when Mexico suffered a steep reces-
20 See, for example, Blecker (2009) and the references cited therein.
sion during the peso crisis, while no significant corre-lation can be seen in the prior years. The large impactof US growth on Mexico benefited the latter duringthe United States’ boom of the late 1990s, but had aless favorable impact during the slower-growth yearsof the early 2000s and especially in the financial crisisand global recession of 2008–2009.
The strong correlation between the Mexican and USeconomies is partly behind the remarkably steadyMexico-US ratios of income per capita and income perworker shown in Figure 1. Indeed, the fact that botheconomies have been growing at similar rates since1996 (as shown in Figure 6) explains why those ratioslook practically unchanged since NAFTA’s enactment.Of course, such a strong correlation can only beexplained by the lack of a domestic engine in Mexico.This result is rather surprising considering that Mexicois one of the largest economies in the world and pre-sumably would have a relatively large domestic market.However, Mexico’s transformation into an outward-looking, export-oriented economy probably went toofar and may have reached the point where the domesticmarket becomes almost irrelevant, thereby aggravatingthe country’s external vulnerability especially to eco-nomic conditions in the US market.
Macroeconomic policy restrictions
In addition to the reforms already described, there havebeen two other important reforms in the conduct ofmacroeconomic policy in Mexico in recent years: on theone hand, the Central Bank is now independent and hasprice stability as its single objective; on the other hand,fiscal policy is conducted according to a highly pro-cyclical rule, which mandates a zero deficit regardless of
the state of the business cycle. Thiscombination of policies, togetherwith the strong correlation of theMexican and US economies,implies a straitjacket for the con-duct of macroeconomic policythat severely limits the ability ofMexican policy makers to respondto external shocks in a counter-cyclical manner (Esquivel 2010).This means that the Mexicaneconomy absorbs all the externalshocks and has no ability to pur-sue independent stimulus policies.Furthermore, the institutionaldesign of macroeconomic policyin Mexico may even exacerbate
negative shocks by inducing fiscal, monetary, andexchange rate policies that end up increasing exchangerate and output volatility. The profound economicimpact of the financial crisis of 2008–2009 on theMexican economy is a case in point.
New opportunities and US interests
After the eventual recovery from the financial crisis andglobal recession of 2008–2009, Mexico and the UnitedStates are likely to enjoy certain opportunities forrenewing their economic cooperation in their mutualinterest. One positive development on the Mexican sideis that the crisis left the peso at a more competitiveexchange rate than it had been at for more than adecade. The peso depreciated from about 10 per dollarin August 2008 to 15 in March 2009, before recuperat-ing to 13 in October 2009. This represents a multilater-al real depreciation close to 20 percent since the begin-ning of the crisis. If the peso is allowed to remain atsuch a competitive level going forward, Mexican indus-tries could get a leg up in attracting FDI and exportingto the US market and elsewhere.
Just before the financial crisis worsened in September2008, the business press was noting a trend toward thereturn of some manufacturing production from Asiato both the United States and Mexico, as a result ofthe high energy prices and transportation costs thathad emerged at that time coupled with the then-lowervalue of the dollar and concerns over quality controlin China.21 The financial crisis and recession tem-
CESifo Forum 4/2010 26
Focus
- 8
- 6
- 4
- 2
0
2
4
6
8
10
1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009
Source: IMF, World Economoc Outlook Database.
ANNUAL GROTH RATES OF REAL GDP, US AND MEXICO1970–2009
%
Mexico
NAFTA
US
Figure 6
21 Regarding the impact of exchange rates and transportation costs,see Mui (2008), and Rubin and Tal (2008). In regard to quality con-cerns about Chinese imports, see Engardio and Smith (2009).
CESifo Forum 4/201027
Focus
porarily interrupted this process, as energy prices
tanked, transportation costs fell, and the dollar tem-
porarily recovered (not only against the peso, but
against most currencies) in the fall and winter of
2008–2009. However, as the global economy began to
revive in the second half of 2009, energy and com-
modity prices started to recover and the dollar
resumed its previous downward course against the
major currencies such as the euro. If the dollar and
peso both stay low and transportation costs again rise
when global demand recovers, the potential for a
revival of both Mexican and US manufacturing is
enormous.
Press reports also indicate that existing foreign
investment in Mexico has been remarkably resilient
in spite of the increased violence resulting from the
government’s crackdown on narcotrafficking
(Engardio and Smith 2009); success in the latter
effort could help the country attract yet more FDI
inflows. Furthermore, although both the US and
Mexican automobile industries took a big hit in the
crisis, as the auto companies begin to focus on small-
er and more fuel-efficient cars for the US market,
there is significant potential for a recovery of region-
al trade in automobiles and auto parts. One sign of
this potential is the (pre-crisis) announcement by
Ford Motors that it would produce a new (low-cost,
fuel-efficient) Fiesta model at its plant in Toluca,
Mexico (Roig-Franzia 2008).
The Ford example reminds us of why US-Mexican
trade relations can be fraught with conflict, since the
jobs that will be supported at the Toluca plant are
jobs that will not be found in Detroit or elsewhere in
the United States. Indeed, the likelihood of US auto
companies increasing their outsourcing was a major
point of controversy in regard to the government
bailout of the US automakers in early 2009. Never-
theless, there are many reasons why expanded trade
with Mexico and efforts to promote Mexican conver-
gence are in the US interest.
First, trade with Mexico is more of a two-way street
for the United States than trade with most Asian
countries. Although the United States has a large
overall trade deficit, its deficit with Mexico is rela-
tively smaller in proportional terms. The average
ratio of US imports to US exports in 2008 was
1.6:1; this ratio was only 1.4:1 for US trade with
Mexico but 4.7:1 for US trade with China. Thus,
even though some Mexican production displaces
some US jobs, Mexico is a better customer for US
exports than most other countries, and hence trade
with Mexico also supports relatively more US jobs.
Hence, a growing Mexican economy would be an
opportunity for, not a threat to, the United States.
Second, the primary economic driver of migration
from Mexico to the United States is the persistently
large wage gap between the two countries, i.e. the lack
of convergence in wages. Hence, policies that could
foster convergence between the two countries via
increased wages in Mexico are the one and only thing
that can, in the long run, stem the tide of Mexican
workers seeking to cross the US border. Instead of
building walls, regional efforts to promote Mexican
growth and convergence would be the best way to
reduce migration pressures.
Third, there are special opportunities for mutual
gains from US-Mexican cooperation in the areas of
health care and elder care services. Given the aging of
the US population and the high and rising costs of
health and elder care in the United States, it would
make sense to allow US Medicare benefits and pri-
vate insurance payments to flow to Mexican
providers of medical care and elder services (e.g.
assisted living or nursing homes), who can provide
those services at significantly lower cost. In fact,
some US senior citizens are already taking advantage
of the lower cost of retiring and seeking medical
treatments in Mexico, but their numbers could be
vastly expanded if Medicare and insurance benefits
were allowed to be spent there (subject, of course, to
adequate quality controls). This could provide enor-
mous numbers of jobs for Mexicans not only in
health and elder care directly, but also in various sup-
plier industries. Given that the manufacturing sector
does not seem capable of supplying adequate num-
bers of jobs in Mexico, for the reasons discussed ear-
lier, Mexico needs to focus on other sectors, such as
services and construction, to solve its employment
problems. Since rising health care costs are threaten-
ing both the private and public sectors of the US
economy, both countries could reap enormous gains
from such an arrangement.
This area of opportunity, however, will not be per-
manent since demographic complementarities
between Mexico and the United States will eventual-
ly disappear. To grasp an idea of how important the
Mexico-US demographic complementarities are,
Figures 7 and 8 show the age structure and the old-
age dependency ratio, respectively, for both countries.
The age structure is shown for 2005, whereas the pro-
jected old-age dependency ratio is shown for the2005–2050 period. This ratio is defined as the numberof people aged 65 and over as a percentage of theproductive segment of the population, which isdefined as people aged 15 to 64. The figure clearlyshows two important elements: first, the old-agedependency ratio in the United States is currentlytwice as high as it is in Mexico and it will be greaterthan the Mexican ratio at least for the next 40 years;second, the gap in old-age dependency ratios willsteadily increase until the mid 2020s, when the gapwill slowly start to decline. This means that the nextten or fifteen years will be the best time for exploitingthe demographic complementarities between Mexicoand the United States. For that reason, this area ofopportunity is one that needs to be explored immedi-ately in order to reap the largest possible benefits forboth countries.
Conclusions
The decision to convert NorthAmerica into a free trade areawith the adoption of NAFTAconcealed a deeper clash ofvisions over what kind of eco-nomic integration was intended.On the one hand, some econo-mists supported it reluctantlybecause of its preferential nature.These economists wanted aNAFTA that would keep NorthAmerica wide open to trade withother global regions and that, ineffect, would be little but a waystation on the road to multilater-al trade liberalization. On theother hand, some advocates of‘industrial policy’ sought aNAFTA that would function as atrue trading bloc, transformingNorth America into a more inter-nally integrated and externallycompetitive region. The industri-al policy advocates were con-cerned mostly about competitionfrom Japan, the four Asian tigers,and the European Union (EU) inthe early 1990s; China was notyet on their radar screens.
In reality, NAFTA – in spite of itsmany exceptions to pure free trade– ended up functioning more like
a globally open regional market than a self-containedtrading bloc, and this had a profound impact on whatthe agreement did and did not accomplish for theMexican and US economies in the long run. NAFTAwas neither the panacea promised by the Mexican gov-ernment nor the disaster predicted by some US oppo-nents. Although the agreement did have a significantimpact on trade and investment flows, it had at most amodest impact on the variables that matter most, suchas employment, income distribution and growth. Thebiggest problem is not what NAFTA did, but what itdidn’t do, namely, to foster a regional integrationprocess that could have lifted up the Mexican economyand produced a convergence in Mexican per capitaincome or average wages toward US levels.
The point is not that NAFTA should have been aneconomic ‘fortress’ defended by high protectionist
CESifo Forum 4/2010 28
Focus
6 4 2 0 2 4 6
0-45-9
10-1415-1920-2425-2930-3435-3940-4445-4950-5455-5960-6465-6970-7475-7980-8485-8990-94
95-100
Men
15 10 5 0 5 10 15
0-45-9
10-1415-1920-2425-2930-3435-3940-4445-4950-5455-5960-6465-6970-7475-7980-8485-8990-94
95-100
WomanMexico
MEXICO AND US POPULATION PYRAMID 2005United States
Million MillionSources: Comisión Nacional de Población for Mexico; US Census Bureau for the Unted States.
Figure 7
- 10
0
10
20
30
40
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050
OLD-AGE DEPENDENCY RATIOS IN MEXICO AND THE US2005–2050
%
Mexico
US
Sources: Comisión Nacional de Población for Mexico; US Census Bureau for the Unted States; authors' calculations.
US - Mexico Gap
Figure 8
CESifo Forum 4/201029
Focus
barriers. Rather, the problem was that neitherMexico nor the United States ever adopted the com-plementary policies that could have promoted amore successful regional integration effort. Thesepolicies would have included promulgating adequateeducation and industrial policies, making the neces-sary infrastructure investments, and maintainingcompetitive exchange rates. Furthermore, theNAFTA countries did not adopt policies to pro-mote convergence of the less developed regions ofthe sort used in the EU, such as its regional andsocial cohesion funds (Pastor 2001). Although theUnited States extended some additional TradeAdjustment Assistance for US workers displaced bytrade with Canada or Mexico, overall the NAFTAcountries did not implement adequate social safetynets for groups adversely impacted by the agree-ment’s adjustment costs. Mexico eventually adoptedcertain redistributive policies, i.e. the Procampo andProgresa/Oportunidades programs, but these werepoorly designed (in the case of Procampo) and cametoo late or on too small a scale to assist during theinitial liberalization of trade or the first few years ofNAFTA.
Although NAFTA did promote increasing regionalintegration in the late 1990s, in the early 2000s thistrend was partially reversed as the lower trade barrierswithin North America were overwhelmed by otherdevelopments, including the lowering of global tradebarriers under the WTO, the tightening of US borderrestrictions, and the emergence of China as an eco-nomic powerhouse. In effect, the vision of NAFTA asa globally open trading region rather than a morecompetitive trade bloc won out, but the goal of pro-moting economic convergence of Mexico to US andCanadian levels of per capita income lost out. Thechallenge for the US and Mexican governments goingforward is to see if they can find a way to rejuvenatethe process of regional integration that can movetoward that goal while serving the mutual interests ofthe US and Mexican economies.
References
Blecker, R.A. (2009), “External Shocks, Structural Change, andEconomic Growth in Mexico, 1979-2007”, World Development 37,1274–1284.
Bleker, R.A. (2010), “Comercio, Empleo y Distribución: Efectos de laIntegración Regional y Global”, in: Lustig, N., A. Y. Naude and A. C. Sabido (eds.), México 2010: Volumen Economía, Mexico City: El Colegio de México, forthcoming.
Borraz, F. and J.E. López-Córdova (2007), “Has Globali-zation Deepened Income Inequality in Mexico?”, Global Economy Journal 7/1, Article 6,http://www.bepress.com/gej/vol7/iss1/6.
Chiquiar, D. (2008), “Globalization, Regional Wage Differentials andthe Stolper-Samuelson Theorem: Evidence from Mexico”, Journal ofInternational Economics 74, 70–93.
Cohen, S.D., R.A. Blecker and P.D. Whitney (2003), Fundamentals ofU.S. Foreign Trade Policy, 2nd edition, Boulder: Westview.
Del Villar, R. (2009), “Competition and Equity inTelecommunications”, in: Walton, M. and S. Levy (eds.), No Growthwithout Equity, Washington DC: World Bank/Palgrave MacMillan,321–369.
Engardio, P. and G. Smith (2009), “Business Is Standing Its Ground”,Business Week, 20 April, 34–39.
Esquivel, G. (2010), “De la Inestabilidad Macroeconómica alEstancamiento Estabilizador: El Papel del Diseño y Conducción de laPolítica Económica en México”, in: Lustig, N., A. Y. Naude and A. C. Sabido (eds.), México 2010: Volumen Economía, Mexico City:El Colegio de México, forthcoming.
Esquivel, G. and J.A. Rodríguez-López (2003). “Technology, Trade,and Wage Inequality in Mexico before and after NAFTA”, Journal ofDevelopment Economics 72, 543–565.
Esquivel, G. and F. Hernández-Trillo (2009), “How Can ReformsHelp Deliver Growth in Mexico?”, in: Rojas, L. (ed.), Growing Painsin Latin America, Washington DC: Center for Global Development,192–235.
Esquivel, G., N. Lustig and J. Scott (2010), “Mexico: A Decade ofFalling Inequality. Market Forces or State Action?”, in: López-Calva,L. F. and N. Lustig (eds.), Declining Inequality in Latin America. A Decade of Progress?, Washington DC: Brookings Institution Press,forthcoming.
Feenstra, R.C. (2006), “New Evidence on the Gains from Trade”, Reviewof World Economics (Weltwirtschaftliches Archiv) 142, 617–641.
Feenstra, R.C., and H.L. Kee (2009), “Trade liberalization andExport Variety: A Comparison of Mexico and China”, in:Ledermann, D. et al. (eds.), China’s and India’s Challenge to LatinAmerica: Opportunity or Threat?, Washington DC: World Bank,245–263.
Galindo, L.M. and J. Ros (2008), “Alternatives to Inflation Targetingin Mexico”, International Review of Applied Economics 22, 201–214.
Gallagher, K.P., J.C. Moreno-Brid and R. Porzecanski (2008), “TheDynamism of Mexican Exports: Lost in (Chinese) Translation?”,World Development 36, 1365–1380.
Gallagher, K.P. and L. Zarsky (2007), The Enclave Economy: ForeignInvestment and Sustainable Development in Mexico’s Silicon Valley,Cambridge, MA: MIT Press.
Haber, S., H.S. Klein, N. Maurer and K. J. Middlebrook (2008),Mexico since 1980, New York: Cambridge University Press.
Hanson, G. (2004), “What Has Happened to Wages in Mexico sinceNAFTA? Implications for Hemispheric Free Trade”, in:Estevadeordal, A. et al. (eds.), Integrating the Americas: FTAA andBeyond, Cambridge, MA: Harvard University Press, 505–537.
Hanson, G. (2006), “Illegal Migration from Mexico to the UnitedStates”, Journal of Economic Literature 44, 869–924.
Hanson, G. and R. Robertson (2009), “China and the RecentEvolution of Latin America’s Manufacturing Exports”,Ledermann, D. et al. (eds.), China’s and India’s Challenge to LatinAmerica: Opportunity or Threat?, Washington DC: World Bank,145–178.
Hufbauer, G.C. and J.J. Schott (1992), North American Free Trade:Issues and Recommendations, Washington DC: Institute forInternational Economics.
Hufbauer, G.C. and J.J. Schott (2005), NAFTA Revisited:Achievements and Challenges. Washington, DC: Institute forInternational Economics.
Leamer, E.E. (1998), “In Search of Stolper-Samuelson Linkagesbetween International Trade and Lower Wages”, in: Collins,S. M. (ed.), Imports, Exports, and the American Worker, WashingtonDC: Brookings Institution, 141–214.
Lederman, D., W.F. Mahoney and L. Servén (2005), Lessons fromNAFTA for Latin America and the Caribbean, Washington DC: WorldBank.
López A G. (2006), Mexico: Two Decades of the Evolution of Educationand Inequality, World Bank Policy Research Working Paper 3919.
López-Córdova, E. (2004), Economic Integration and ManufacturingPerformance in Mexico: Is Chinese Competition to Blame?, WorkingPaper 23, Latin America/Caribbean and Asia/Pacific Economics andBusiness Association (LAEBA).
Lustig, N. (1998), Mexico: The Remaking of an Economy, 2nd Edi-tion, Washington DC: Brookings Institution.
Moreno-Brid, J.C., J. Santamaría and J. C. Rivas Valdivia (2005),“Industrialization and Economic Growth in Mexico after NAFTA:The Road Travelled”, Development and Change 36, 1095–1119.
Mui, Y.Q. (2008), “Ikea Helps a Town Put It Together: Manu-facturing Jobs Come Back to Southern Va”, Washington Post, 31 May.
Pastor, R.A. (2001), Toward a North American Community: Lessonsfrom the Old World for the New, Washington DC: Institute forInternational Economics.
Revenga, A.L. and C.E. Montenegro (1998), “North AmericanIntegration and Factor Price Equalization: Is There Evidence ofWage Convergence between Mexico and the United States?”, in:Collins, S. M. (ed.), Imports, Exports, and the American Worker,Washington DC: Brookings Institution, 305–347.
Robertson, R. (2007), “Trade and Wages: Two Puzzles from Mexico”,World Economy 30, 1378–1398.
Roig-Franzia, M. (2008), “Ford’s ‘Global Car’ to Roll Out in Mexico:Small, Efficient Auto Designed to Be Sold Anywhere”, WashingtonPost, 31 May.
Rubin, J. and B. Tal, (2008), “Will Soaring Transport Costs ReverseGlobalization?”, StrategEcon, CIBC World Markets, 27 May, 4–7,http://research.cibcwm.com/economic_public/download/smay08.pdf.
Ruiz-Nápoles, P. (2004), “Exports, Growth, and Employment inMexico, 1978-2000”, Journal of Post Keynesian Economics 27,105–124.
Scott, R.E., C. Salas and B. Campbell (2006), Revisiting NAFTA: StillNot Working for North America’s Workers, EPI Briefing Paper 173,Economic Policy Institute, Washington DC.
Studer, I. and C. Wise (2007, eds.), Requiem or Revival? The Promiseof North American Integration, Washington DC: BrookingsInstitution.
US Council of Economic Advisers (2009), Economic Report of thePresident 2009, Washington DC: Government Printing Office,www.gpoaccess.gov/eop/download.html.
US International Trade Commission (USITC, 1991), The LikelyImpact on the United States of a Free Trade Agreement with Mexico,USITC Publication 2353, Washington DC: USITC.
Verhoogen, E.A. (2008), “Trade, Quality Upgrading, and WageInequality in the Mexican Manufacturing Sector”, Quarterly Journalof Economics 123, 489–530.
CESifo Forum 4/2010 30
Focus
CESifo Forum 4/201031
Focus
MEASURING THE ECONOMIC
EFFECTS OF NAFTA ON
MEXICO
DOMINICK SALVATORE*
Introduction
The economic effects of NAFTA on Mexico havebeen discussed and measured, for the most part, interms of its effect on employment on both sides of theborder and by the increase in Mexico-US trade andinvestments before and after NAFTA. Measured inthis way, some economists predicted that NAFTAwould create hundred of thousand of new jobs onboth sides of the border, reduce wage inequalitiesbetween the United States and Mexico, and sharplyreduce the migration of Mexicans to the UnitedStates (see, for example, Hufbauer and Schott 1992).Others, such as Ross Perot (the third party candidatein the 1992 US Presidential election), believed insteadthat the United States would lose thousand of jobs toMexico as US firms migrated south attracted by muchlower wages and fewer labor and environmental regu-lations (the ‘giant sucking sound’). Of course, bothpredictions were wrong, but what is more important isthat measuring the effect of NAFTA based on itseffect on employment and comparing trade beforeand after the Agreement is incorrect.
More reasonable economists attempted to measurethe effects of NAFTA by comparing Mexican growthbefore NAFTA and with NAFTA. This also is wrongbecause, as we know, growth is affected by many otherfactors, independently of NAFTA. For example,Mexico faced a serious financial and economic crisisin 19941995 (just when NAFTA took effect), theUnited States faced recession in 2001 and slow growthin 2002, and China joined the WTO in December2001, thus becoming a more powerful competitor ofMexico in the United States and on world markets(and also leading to the diversion of some foreign
direct investments from Mexico to China – see also
Salvatore 2010a). Thus, measuring the economic
effects of NAFTA on Mexico based on the net num-
ber of jobs created or destroyed, or by comparing
Mexican growth before and after NAFTA, is not
appropriate.
Measuring the economic effects of free trade areasand NAFTA
Until Jacob Viner wrote his classic work The Customs
Union Issue in 1953, economists believed that a free
trade area (FTA), by removing trade restrictions
among member states, increased specialization in pro-
duction and invariably benefited the member coun-
tries and it also had positive spillover effects on the
rest of the world. Viner, however, showed that this was
not necessarily the case. Member nations and the rest
of the world would benefit only if the FTA created,
rather than diverted, trade. This is an application of
the theory of the second best, which postulates that if
all the conditions for first best or maximum social
welfare (free trade) are not satisfied, trying to satisfy
as many of the conditions as possible (partial trade
liberalization) is not necessarily or usually second
best. Lipsey (1961) then specified the conditions
under which a FTA was more likely to be trade creat-
ing and thus welfare enhancing.
Meade (1955) extended Viner’s analysis showing that
by considering not only the production effects of FTA
(as Viner did) but also the consumption effects of a
FTA, then even a trade-diverting FTA could improve
members’ and world’s welfare. Meade pointed out
that the smaller the relative inefficiency of FTA mem-
bers in relation to non-members, the greater was the
probability that even a trade-diverting FTA or cus-
toms union could lead to net benefits to members and
non-member nations. Although important, this
method of measuring the economic effects of a FTA
concentrates on the static effects of economic integra-
tion without considering its more significant dynamic
effects, such as the increased competition, economies
of scale, stimulus to investment, and the generally bet-
ter utilization of economic resources. * Fordham University, New York. This paper is a revision andupdate of Salvatore (2007).
According to theory, therefore, the appropriatemethod of measuring the effects of a FTA, such asNAFTA, is by the increased efficiency and produc-tivity resulting from specialization in production,trade, investments, and competition. This increasedefficiency will certainly affect wages, employmentand growth. But welfare can increase in a membernation even if the FTA creates few or no jobs, aslong as it increases efficiency and productivity in thenation. Such an increase in efficiency and productiv-ity will stimulate growth in member nations but sincegrowth also depends on other crucial concomitantfactors, it is impossible to identify the specific effectsof a FTA by comparing its growth before and afterits formation. But this is exactly how the effects ofFTAs, in general, and of NAFTA, in particular, havebeen mostly measured – that is, by its effect onemployment and on growth before and after the for-mation of the FTA (see, for example, Hufbauer andSchott 1992, 2005 and 2008; Kose, Meredith andTowe 2004).
The correct way to measure the effects of a FTA onmember nations, instead, is by counterfactual simula-tion of its effect on intra-FTA trade and growth.That is, by how much intra-FTA trade and growth ishigher with the FTA as compared with the situationwithout the FTA. Performing valid counterfactualsimulations are fraught with difficulties, however. Ageneral equilibrium model of trade and growthwould have to be constructed, the model would haveto be validated by in-sample dynamic simulation, andthen used for counterfactual simulation. In our case,this would involve estimating how high trade and
growth would have been among the FTA memberswithout the FTA.
I will begin, instead, with a more down-to-earth andless elegant but still legitimate method of estimatingthe economic effects of NAFTA on Mexico by com-paring (1) the growth of intra-Mexico-US trade, onthe one hand, to the growth of total Mexican trade,on the other, during the first dozen years (i.e. until2005) of NAFTA’s operation and also until 2008(when the NAFTA accord became fully operational)and (2) by comparing the flow foreign direct invest-ment (FDI) from the United States to Mexico, to totalFDI to Mexico, over the same periods. I will then pre-sent the results of a counterfactual simulation of theeffect of NAFTA on trade, FDI, growth, inflation,and other economic and financial variables.
The effects of NAFTA on Mexican trade
Table 1 shows the level and growth of Mexican exportsto the United States as well as total Mexican exports tothe rest of the world from 1984–1993 (i.e. during thedecade before the formation of NAFTA) and duringthe first dozen years since the creation of NAFTA (i.e.until 2005) and also until 2008. The table shows thatfrom 1984–1993, total Mexican exports grew fromUSD 29.4 billion in 1984 to USD 51.9 billion in 1993,or at an average yearly rate of 5.5 percent. On the otherhand, Mexican exports to the United States grew fromUSD 18.1 billion in 1984 to USD 40.4 billion in 1993,or at a yearly average rate of 7.6 percent. Thus, from1984–1993 (i.e. in the decade before NAFTA), Mexican
CESifo Forum 4/2010 32
Focus
Table 1
Mexico’s exports: 1984–1993, 1994–2005 and 1994–2008
(billion US dollars)
Year 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993
Average
growth
1984–1993
Total 29.4 27.2 22.0 28.0 30.8 35.3 41.0 42.9 46.2 51.9 5.5%
To US 18.1 19.1 17.7 20.3 23.3 27.1 30.5 31.5 35.6 40.4 7.6%
Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Total 60.9 79.5 96.0 110.4 117.5 136.4 166.1 158.8 161.0 164.8
To US 50.1 62.8 75.1 86.7 95.4 110.6 137.0 132.6 136.1 139.8
Year 2004 2005 2006 2007 2008
Average
growth
1994–2005
Average
growth
1994–2008
Total 188.0 214.2 249.9 271.9 291.3 9.3% 8.7%
To US 158.3 173.5 202.0 214.8 220.3 9.2% 8.4%
Sources: WTO (2009); US Department of Commerce.
CESifo Forum 4/201033
Focus
exports to the United States grewmore rapidly that total Mexicanexports.
From Table 1, we see that in thedozen years of NAFTA’s opera-tion (i.e. until 2005) and alsountil 2008, the total exports ofMexico increased at about thesame rate as Mexican exports tothe United States. We wouldhave expected, instead, thatintra-FTA trade would growfaster than extra-FTA tradebecause of the elimination oftrade barriers among membercountries but not with respect tononmembers. Actually, the in-crease in Mexican exports to theUnited States jumped by an aver-age of 28.3 percent per year from1991–1993 in anticipation of NAFTA, as comparedwith an average increase of 21.0 percent for totalMexican exports over the same period. But Mexicanexports to the United States had been growing fasterthan the total exports of Mexico from 1984.
Similarly, Mexican exports to the United States grewmuch faster from 1994–2000 than from 2001–2005 orfrom 2001–2008, but over the entire 1994–2005 and1994–2008 Mexican exports to the United States grewat a slightly lower rate than total Mexican exports.Thus, NAFTA seems to have benefited Mexico (andthe United States) by leading to a more rapid expan-sion of trade in general for the three years precedingand for the seven years following the creation ofNAFTA as a result of the general liberalization oftrade (and the signaling of Mexico’s commitment tofurther liberalization and reform) rather than by amore rapid increase of Mexican-US trade, as such, inrelation to total Mexican trade.
The pattern of trade increase as a result of NAFTAalso seems somewhat different from that arising inthe European Union. Table 2 shows the value oftotal exports, intra-regional-trade-agreement(RTA) exports, and intra-RTA exports as a percent-age of the total RTA exports of the EuropeanUnion (EU) and NAFTA in 1990, 1995, 2000, 2005,and 2008. The table shows that the EU has a largerpercentage of intra-RTA trade than NAFTA, butintra-RTA increased only slightly for the EUbetween 1990 and 2005 and between 1990 and 2008,
while for NAFTA it increased significantly. The bigjump in intra-NAFTA trade, however, occurredduring the first five years of NAFTA.1 Intra-EUtrade also increased in the years immediately afterits creation (not shown in Table 2), but afterwardsintra-EU trade expanded at about the same rate astotal EU trade. Also note that there seems to havebeen no similar increase in intra-EU trade in antic-ipation of and immediately after the EU-expansionfrom 15 to 25 members in 2004 and from 25 to27 members in 2008.
The effects of NAFTA on the inflow of FDI toMexico
Another way by which a FTA can benefit a membernation is by encouraging an inflow of FDI, therebystimulating the growth of the nation. Table 3 com-pares the inflow of FDI to Mexico, in total and fromthe United States, before and since the creation ofNAFTA. The table shows that from 1984–1993, totalFDI into Mexico averaged USD 2.8 billion per year ascompared to USD 1.1 billion, or 39 percent of thetotal, coming from the United States. The big jump intotal FDI to Mexico during this period occurred in1991, perhaps in anticipation of NAFTA.
Table 2
Total and intra-EU and intra-NAFTA trade:
1990, 1995, 2000, 2005 and 2008
(billions US dollars and percentages)
NAFTA exports (billion US dollars)
Year Total Intra-NAFTA Intra-NAFTA as
percentage of total
1990 562 240 42.8
1995 857 394 46.0
2000 1,225 682 55.6
2005 1,478 824 55.8
2008 2,036 1,015 49.9
EU exports (billions US dollars)
Year Total Intra-EU Intra-EU as
percentage of total
1990 (EU15) 1,482 980 66.1
1995 (EU15) 1,937 1,295 66.9
2000 (EU15) 2,251 1,392 61.9
2000 (EU25) 2,437 2,523 67.5
2005 (EU25) 4,001 2,673 66.8
2008 (EU27) 5,898 3,974 67.4
Source: WTO (2009).
1 The sharp decline in Mexican intra-NAFTA exports as a percent-age of the total exports from 55.8 in 2005 to 49.9 in 2008 was theresult of the sharp reduction in US imports from Mexico as a resultthe economic crisis in the United States. In 2007, the correspondingpercentage was 51.6 (still lower that the value of 55.8 in 2005) as aresult of the approaching US crisis.
Form 1994–2005, total FDI to Mexico averaged USD16.9 billion per year as compared with USD 5.7 bil-lion, or 34 percent of the total coming from theUnited States. The corresponding figures for1994–2008 are, respectively, 18.2, 6.4 and 35. The bigjump in FDI to Mexico in total and from the UnitedStates occurred in 1994 (the first year of NAFTA’soperation) and then again in 2001 (eight years afterthe start of NAFTA). As in the case of trade, there-fore, FDI to Mexico seemed to have increased as thegeneral result of the liberalization of foreign tradeand investments that accompanied the creation ofNAFTA and not as a direct and specific result ofNAFTA itself (since US FDI to Mexico did notincrease more rapidly that total FDI to Mexico dur-ing the NAFTA period).
NAFTA’s effects on Mexico’s growth
Table 4 shows the growth of real GDP in Mexicobefore and after the formation of NAFTA. Theexpectation was that by stimulating the flow of tradeand investments, a FTA would increase specializa-
tion in production and efficiency in general through-out the economy of a member nation, and thusspeed up its rate of growth. The specific effect ofNAFTA on Mexico’s growth, however, can only bemeasured by estimating what growth in Mexicowould have been without NAFTA and comparingthat with its actual growth (i.e. by counterfactualsimulation). As pointed out earlier, the reason forthis is that growth depends on many other factorsbesides NAFTA and comparing actual growthbefore and after NAFTA simply cannot identifyNAFTA’s contribution.
For example, Table 4 shows that the growth of realGDP averaged 2.5 percent per year in Mexico from1984–1993 (i.e. before NAFTA) and 2.8 percent fromper year from 1994–2005 (i.e. after the creation ofNAFTA) and 2.9 percent from 1994–2008. If weexclude the 1995 recession year in Mexico, yearlygrowth averaged 3.6 and 3.5 percent, respectively. Butwe cannot tell how much, if any, of the higherMexican growth since 1994 can be attributed toNAFTA. That can only be measured by counterfactu-al simulation.
CESifo Forum 4/2010 34
Focus
Table 3
Foreign direct investments inflows to Mexico:
1984–1993, 1994–2005 and 1994–2008
(billion US dollars)
Year 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 Average
1984–1993
Total 1.5 2.0 2.0 1.2 2.0 2.8 2.6 4.8 4.4 4.4 2.8
from US 0.3 0.5 – 0.1 0.3 0.6 1.4 1.9 2.3 1.3 2.4 1.1
Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Total 11.0 9.5 9.2 12.8 12.7 13.9 18.1 29.8 23.6 16.6
from US 4.5 3.0 2.4 5.6 4.7 8.2 4.2 14.2 5.2 3.7
Year 2004 2005 2006 2007 2008
Average
1994–2005
Average
1994–2008
Total 23.8 22.3 19.8 27.3 23.2 16.9 18.2
from US 6.4 6.8 10.6 8.8 7.2 5.7 6.4
Sources: IMF (2010); US Department of Commerce.
Table 4
Growth of Real GDP: 1984–1993, 1994–2005 and 1994–2008 (%)
Year 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993Average
1984–1993
GDP 3.5 2.5 – 3.6 1.8 1.3 4.2 5.1 4.2 3.6 1.9 2.5%
Year 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
GDP 4.5 – 6.2 5.5 7.2 5.0 3.6 6.0 – 0.9 0.1 1.4
Year 2004 2005 2006 2007 2008
Average
1994–2005
Average
1994–2008
GDP 4.0 3.2 4.9 3.3 1.5 2.8 2.9
Source: OECD (2010).
CESifo Forum 4/201035
Focus
Table 5 shows long-run simulations results ofNAFTA’s impact on Mexico to the year 2005 andcompares these to the actual outcome using theUnited Nations LINK Model of the world economy.During the 1995–2005 period, Mexican real GDP wasestimated to grow at a rate of 5.2 percent per yearwith NAFTA, as compared with 3.8 percent withoutNAFTA. NAFTA was also expected to (1) reduce theMexican inflation rate from 14.5 percent to 9.7 per-cent per year and the short-term interest rate from18.3 percent to 13.0 percent, (2) increase the inflow offoreign direct investments (FDI) from USD 6.0 billionto USD 9.2 billion per year and the growth of exportsfrom 8.3 to 10.4 percent, and (3) raise the Mexicantrade deficit from USD 9.7 billion to USD 14.9 billionand net financial inflows from USD 10.6 billion toUSD 14.7 billion per year.
The actual results, as yearly averages from 1994–2005,were as follows: a growth rate of real GDP of 2.8 per-cent per year, a rate of inflation of 13.9 percent, ashort-term interest rate of 18.7 percent, an inflow ofFDI of USD 16.9 billion, a growth of exports of9.2 percent, a trade deficit of USD 7.7 billion, and netfinancial inflows of USD 16.8 billion. The actualresults for 1994–2008 were similar to those for1994–2005 (see the last column of Table 5).
Mexico did not realize more of the expected benefitsfrom NAFTA because of its deep economic crisis in1994–1995, the US recession in 2001 and slow
growth in 2002, and, more importantly, because ofweak economic institutions and inadequate structur-al reforms, which limited Mexico’s internationalcompetitiveness (Mexico ranked 47th out of58 economies evaluated according to the 2010 World
Competitiveness Report) in the face of increasedcompetition from China. If we removed from thedata 1995 (the recession year in Mexico) and also2001 and 2002 (the years of recession and slowgrowth in the United States, which reduced USimports from Mexico), the average annual growth ofreal GDP in Mexico would be 4.5 percent for1994–2005 and 4.1 for 1994–2008, which were hard-ly adequate to achieve a rapid increase in the stan-dard of living of the nation (and less than half ofChina’s growth rate over the same time period).
General effects of NAFTA on member nations
We now examine in more general terms the effect ofNAFTA on Mexico, the United States and Canada.As we have seen, NAFTA benefited Mexico by indi-rectly leading to greater export-led growth resultingfrom increased access to the huge US market and byincreasing inward foreign direct investments – all ofwhich increased production efficiency and competi-tion. Perhaps more significant than the change inthe total volume of trade was the change in the com-position of Mexican trade. As a result of verticalintegration, Mexican trade shifted much more
toward intra-industry and intra-firm trade with the UnitedStates and Canada. Mexico suf-fered a net loss of jobs and in-comes in agriculture, but theselosses were more than matchedby net increases in industry.With time, increasing employ-ment opportunities and risingwages in industry are expectedto reduce the pressure onMexicans to migrate to theUnited States. Mexico’s abilityto benefit from NAFTA hasbeen limited, however, by weakeconomic institutions and inad-equate structural reforms of theeconomy.
The implementation of NAFTAbenefited the United States byincreasing competition in prod-
Table 5
Simulation of NAFTA’s impact on the Mexican economy
(yearly averages: 1994–2005 and 1994–2008)
Estimates
with
NAFTA
Without
NAFTA Difference
Actual
results
1994–2005
Actual
results
1994–2008
Growth of
real GDP (%) 5.2 3.8 1.4 2.8 2.9
Inflation rate
(%) 9.7 14.5 – 4.8 13.9 12.0
Short-term
interest rate
(%) 13.0 18.3 – 5.3 18.7 16.5
Inflow of FDI
(bill. USD) 9.2 6.0 3.2 16.9 18.2
Growth of
exports (%) 10.4 8.3 2.1 9.2 8.4
Trade deficit
(bill. USD) 14.9 9.7 5.2 7.7 9.6
Net financial
inflows
(bill. USD) 14.7 10.6 4.1 16.8 16.2
Sources: Klein and Salvatore (1995); Hufbauer and Schott (2005);
IMF (2010); OECD (2010).
uct and resource markets, as well as by lowering the
price of many commodities to US consumers.
Because the US economy is more than 15 times larger
than Mexico’s economy the US gains from NAFTA
as a proportion of its GDP were much smaller than
Mexico’s, however. Furthermore, with wages more
than six times higher in the United States than in
Mexico, NAFTA led to a loss of unskilled jobs, but an
increase of skilled jobs, for an overall net increase in
employment in the United States of between 90,000
and 160,000 (see Inter-American Development Bank
2002). A more recent study by Hufbauer and Schott
(2005), however, concluded that net gain in US jobs as
a result of NAFTA may have been much smaller (and
may even have resulted in a small net loss). Some
States (such as Alabama and Arkansas) suffered while
high-wage areas gained, but with a 15-year phase-in
period and about USD 3 billion assistance to dis-
placed workers, the harm to workers in low-income
areas in the United States was minimized.
Free trade access to Mexico allows US industries to
import labor-intensive components from Mexico
and keep other operations in the United States
rather than possibly losing all jobs in the industry
to low-wage countries. Some of the jobs that
Mexico gained during the years immediately pre-
ceding and following the creation of NAFTA did
not, in fact, come from the United States but from
other countries, such as Malaysia, where wages are
roughly equal to Mexico’s. As a condition for con-
gressional approval of NAFTA, the United States
also negotiated a series of supplemental agreements
with Mexico governing workplace and environmen-
tal standards (to prevent US firms from moving all
of their operations to Mexico to take advantage of
much more lax labor and environmental regula-
tions), as well as to protect some American indus-
tries against import surges that might threaten
them.
NAFTA did not directly affect Canada in a signifi-
cant way because Canada had already negotiated a
free trade agreement with the United States in 1988,
and so most of its economic effects on the two coun-
tries had already taken place by the time NAFTA
came into effect in 1994. Indeed, one could say that
the primary reason for Canada joining in the NAFTA
negotiations was to protect its trade interests with the
United States. Canada was and remains the largest
trade partner of the United States, Mexico was sec-
ond until 2006 when it was displaced by China, and
Japan is fourth.
Summary and conclusions
The economic effects of NAFTA on Mexico have
been discussed and measured mostly in terms of their
effect on employment on both sides of the border and
by the increase in Mexico-US trade and investments,
before and after NAFTA. These are not the appropri-
ate ways to measure the effects of a free trade area on
a member state. The theoretically correct way of mea-
suring the economic effects of a FTA on a member
nation is through its effects on trade, investments,
competition and efficiency. It is through these effects
that the growth and employment in the nation are
affected. To measure the direct effects of a FTA on
member nations requires using a counterfactual simu-
lation. That is, comparing trade, investments, compe-
tition, and efficiency in general and, through them,
their effect on growth and employment in the nation,
with and without the FTA.
In the case of NAFTA, the benefits flowing to Mexico
seem to have resulted more from the general liberal-
ization of trade and investments than directly from
NAFTA, as such. That is, the general liberalization of
trade and investments that accompanied NAFTA led
to a general increase in Mexican exports and inflows
of FDI, which increased specialization, competition,
productivity and efficiency in Mexico. But the
increase in total Mexican exports and FDI inflows
from the rest of the world was as large as or larger
than that from the United States. Furthermore, most
of the (indirect) benefits that Mexico received from
NAFTA occurred in the years immediately preceding
the creation of NAFTA rather than in the years soon
after its creation. Mexico was unable to capture more
of the potential benefits from NAFTA or for a longer
period of time because of the economic crisis that
afflicted Mexico in 1994–1995, the US recession in
2001 and slow growth in 2002, increased competition
from China, but, most importantly, because Mexico
failed to adequately restructure and liberalize its econ-
omy and improve the education and training of its
labor force.
The implementation of NAFTA benefited the
United States by increasing competition in product
and resource markets, as well as by lowering the
prices of many commodities to US consumers.
Because the US economy is so much larger than
Mexico’s, however, US gains from NAFTA as a pro-
portion of its GDP have been much smaller. Canada
was the least affected by NAFTA because Canada
had already negotiated a free trade agreement with
CESifo Forum 4/2010 36
Focus
CESifo Forum 4/201037
Focus
the United States in 1988, and so most of its eco-nomic effects had already taken place by the timeNAFTA came into effect in 1994.
References
Fatemi, K. and D. Salvatore (1994), The North American Free TradeAgreement, New York: Pergamon.
Hufbauer, G.C. and J. J. Schott (1992), North American Free Trade:Issues and Recommendations, Washington DC: Institute forInternational Economics.
Hufbauer, G.C. and J.J. Schott (2005), NAFTA Revisited, WashingtonDC: Institute for International Economics.
Hufbauer, G.C. and J.J. Schott (2008), “Nafta’s Bad Rap”, TheInternational Economy 12, 19–23.
IMD (2010), World Competitiveness Report, Lausanne: IMD.
IMF (2010), International Financial Statistic, Washington DC: IMF.
Inter-American Development Bank (2002), Integration in theAmericas, Washington DC: Inter-American Development Bank.
Klein, L. and D. Salvatore (1995), “Welfare Effects of NAFTA”,Journal of Policy Modeling 17, 163–176.
Kose, M.A., G.M. Meredith and C.M. Towe (2004), How HasNAFTA Affected the Mexican Economy? Review and Evidence, IMFWorking Paper WP/04/59.
Lipsey, R.G. (1961), “The Theory of Customs Unions: A GeneralSurvey”, Economic Journal 71, 498–513.
Meade, J. (1955), The Theory of Customs Union, Amsterdam: North-Holland.
Mendoza, G.E. (2010), “The Effect of the Chinese Economy onMexican Maquiladora Employment”, The International TradeJournal 24, 52–83.
OECD (2010), Economic Outlook, Paris: OECD.
Salvatore, D. (2006), “Can NAFTA Be a Steppingstone to MonetaryIntegration in North America?” Economie Internationale 3, 135–148.
Salvatore, D. (2007), “Economic Effects of NAFTA on Mexico”Global Economy Journal 7, 1–12.
Salvatore, D. (2009), “The Challenge to the Liberal Trading System”,Journal of Policy Modeling 31, 593–599.
Salvatore, D. (2010a), “China’s Financial Markets in the GlobalContext, The Chinese Economy 43, 8–21.
Salvatore, D. (2010b), “Globalization, International Competitiveness,and Growth”, Journal of International Commerce, Economics andPolicy (JICEP) 1, 21–32.
Viner, J. (1953), The Customs Union Issue, New York: The CarnegieEndowment for International Peace.
WTO (2009), International Trade Statistics, Geneva: WTO.
CESifo Forum 4/2010 38
Focus
NAFTA AND MEXICO-USMIGRATION: WHAT LESSONS,WHAT NEXT?
PHILIP MARTIN*
One hoped-for side effect of the North American FreeTrade Agreement, which lowered trade and invest-ment barriers between Canada, Mexico and theUnited States beginning in 1 January 1994, was fastereconomic and job growth in Mexico that would slowunwanted Mexico-US migration. Instead, Mexico-USmigration has surged: the number of Mexican-bornUS residents almost tripled to 12 million between1990 and 2010. The NAFTA-related migration hump– the surge in Mexico-US migration to over 500,000 ayear – turned out to be far larger and to persist farlonger than expected, reflecting supply-push condi-tions in Mexico, demand-pull factors in the UnitedStates, and networks that link Mexican workers to USjobs. This paper explains the factors in Mexico andthe United States that allowed Mexico-US migrationto increase rapidly and why it will be hard to reduceMexico-US migration in the short term.
Mexico-US migration
Between 1820 and 2010, over 75 million immigrantsarrived in the United States. About half were fromEurope, including a tenth from the leading country ofimmigration, Germany. However, only four percent ofGerman immigrants arrived since 1980, comparedwith three-fourths of Mexican immigrants, which iswhy Mexico surpassed Germany as the leading coun-try of immigration in 2007. A quarter of the legalimmigrants in the United States and over half ofunauthorized foreigners, were born in Mexico.
How did Mexico become the major source of USimmigrants? In 1800, Mexico and the United Stateshad populations of roughly equal size, six million,
and Mexico’s per capita GDP was about half that ofthe United States. Northern Mexico (now the south-western US states) was transferred to the UnitedStates by the Treaty of Guadalupe Hidalgo in 1848,ending a war that began when American settlersmoved into Mexican territory and rebelled againstMexican authority. The relatively few Mexican resi-dents of what is now the southwestern United Statesbecame Americans, and there was relatively littlemigration and trade between these DistantNeighbors1 for the next century (Riding 1989).
The US population grew with immigration and birthsand the United States industrialized in the first half ofthe 20th century; Mexico had high birth and deathrates, a much slower growing population, and anagrarian economy marked by large landowners andpoor peasants. Early in the 20th century, whenMexico was undergoing a civil war to break up its lat-
ifundia, the US government approved the recruitmentof Mexican workers to come to the United States asguest workers. These so-called Braceros (arms) wereyoung Mexican men admitted legally between 1917and 1921 and again between 1942 and 1964 to workon US farms and railroads.
Both of the Bracero programs that brought Mexicanmen into the United States as guest workers began dur-ing wartime emergencies, when farmers said that theyfaced a shortage of labor. Both got larger and lastedlonger than expected because of distortion and depen-dence. Distortion reflects the assumptions of US farm-ers that Braceros would continue to be available,prompting investments in orchards and vineyards thatmay not be profitable if a reduced inflow of Bracerosraised wages, which prompted farmers to resist effortsto reduce or end the Bracero program (Martin 2009).
Meanwhile, some Mexicans became dependent onhigher wage US jobs, and moved their families to theborder to increase the chance of being selected as aBracero.2 When the Bracero program ended in 1964,
* University of California, Davis.
1 Mexico’s desire to avoid closer ties with the United States was sum-marized in the aphorism, ‘Poor Mexico, so far from God, so close tothe US’.2 Farmers had to pay transportation costs from the place of recruit-ment in Mexico. By moving to border cities, Mexicans improvedtheir chances of being selected by US farmers.
CESifo Forum 4/201039
Focus
during the height of the Civil Rights movement and
amidst concern about poor Americans, thousands of
Braceros and their families in Mexican border cities
had no way to earn a living. In a bid to discourage
them from migrating to the United States, the
Mexican and US governments modified their trade
laws to allow the creation of maquiladoras, factories
in Mexican border cities that imported components,
hired Mexican workers to assemble them into goods
such as TVs, and re-exported the finished products.
The maquiladoras never provided many jobs for ex-
Braceros, who were almost all men, because they pre-
ferred to hire young women.
Even though many rural Mexicans had experience
working in US agriculture, there was relatively little
illegal Mexico-US migration during the 1960s and
early 1970s, the so-called golden era for US farm
workers. About 110,000 Mexicans were apprehended
just inside the Mexico-US border in FY65, the year
after the Bracero program ended. The number of
apprehensions gradually rose, reaching 420,000 in
FY71 and topping a million in FY77 and FY78, when
the leader of the United Farm Workers, Cesar
Chavez, complained that Mexican ‘wetbacks’ were
making it hard for US farm workers to win wage
increases (Martin 2003).
Mexico had extremely high population growth rates
during the 1960s, when Mexican women averaged
7.2 children. After oil was discovered in the Mexican
part of the Gulf of Mexico in 1978, the Mexican
government, anticipating ever-higher oil revenues,
went on an unsustainable spending spree that ended
with a sharp peso devaluation in 1982 and increased
illegal migration to the United States. Appre-
hensions of Mexicans just inside the Mexico-US
border topped 1.2 million in FY83, and averaged
about 1.5 million a year for the next 15 years. The
Border Patrol handled internal enforcement within
the United States, and agents seeking unauthorized
foreigners would surround farms and workplaces
and try to catch workers who ran away. However,
employers were not punished for hiring unautho-
rized workers, and there was not enough enforce-
ment to prevent their spread from farms to con-
struction sites, factories and restaurants.
The US government responded to rising illegal
Mexico-US migration with the Immigration Reform
and Control Act (IRCA) of 1986, which introduced
penalties on US employers who knowingly hired
unauthorized workers. The theory was that ‘closing
the labor market door’ would discourage Mexicansfrom attempting illegal entry over the loosely guarded2,000 mile border, since it would not be worth takingthe risk to cross the border illegally if employersshunned unauthorized workers to avoid fines. How-ever, unauthorized Mexicans soon learned that theycould still get US jobs by showing forged documentsor documents belonging to legal workers, enablingthose who eluded border agents to find jobs as before.Meanwhile, IRCA legalized 2.7 million unauthorizedforeigners, 75 percent Mexicans, creating millions ofnew links between legal Mexicans in the United Statesand their relatives and friends in Mexico.
In 1990, there were 4.3 million Mexican-born US res-idents, and most were legally in the United States as aresult of the legalization in 1987–1988. The number ofMexican-born US residents tripled over the next twodecades to 12 million, and today over half are notauthorized to be in the United States. This means thattwo-thirds of Mexican-born US residents arrivedsince 1990, and most are unauthorized, despite legal-ization to wipe the slate clean and enforcement effortsaimed at deterring their entry. The third element in theeffort to deter unauthorized Mexico-US migrationwas NAFTA, a trade and investment agreement thatwas expected to speed up economic and job growth inMexico keep Mexicans there.
NAFTA and migration
Mexico in the mid-1980s changed its economic poli-cies from inward-looking to outward-looking, shiftingfrom a policy that protected local industries fromimports to a policy that lowered trade barriers toattract foreign investors. Mexico joined the GeneralAgreement on Tariffs and Trade in 1986 (now theWorld Trade Organization). A Canada-US Free TradeAgreement (FTA) went into effect with little fanfarein 1989, prompting the Mexican government to pro-pose its own FTA with the United States in 1990. Theresult was the North American Free TradeAgreement, which went into effect 1 January 1994.Mexico’s then President, Carlos Salinas, asserted thatlowering trade and investment barriers in NorthAmerica would allow Mexico ‘to export goods, notpeople’.3
NAFTA put Mexico on the economic map for USand other foreign investors. Foreign investment
3 Quoted in Elisabeth Malkin, “Nafta’s Promise, Unfulfilled”, New York Times, 24 March 2009.
CESifo Forum 4/2010 40
Focus
poured in and trade increased, making Mexico the
second-largest trading partner of the United States,
after Canada. But NAFTA got off to a rough start in
Mexico, with a sharp peso devaluation in 1994 and a
deep recession that sharply lowered wages and
increased unemployment. NAFTA hastened changes
already underway in Mexico, such as accelerating
rural-urban migration. Many young women left rural
areas to seek jobs in border-area maquiladoras, and
many young men headed to the United States. For
both men and women, the mantra was ‘go north for
opportunity’.
Rural Mexico presented a special challenge. When
NAFTA went into effect in 1994, a quarter of
Mexicans lived in rural areas, three fourths of
Mexico’s poverty was in rural areas, and over half of
the Mexicans in the United States were from rural
Mexico. During the Mexican Revolution of
1910–1917, the battle cry was land for the peasants,
and Article 27 of the Mexican constitution redistrib-
uted large landholdings into ejidos whose farmer-
members could farm their land, and pass it on to their
heirs, but not rent or sell it. In 1992, after decades of
rural poverty, the Mexican government modified
Article 27 in anticipation of NAFTA to allow the
rental and sale of ejido land, assuming that ejido
members agreed (Cornelius and Myhre 1998).
The combination of allowing ejido land to be rented
or sold and the employment and wage changes asso-
ciated with NAFTA signaled especially young people
in rural Mexico that their future lay elsewhere. One
dimension of the displacement challenge in Mexico is
evident in corn production. In the early 1990s, the US
state of Iowa produced twice as much corn as the
country of Mexico, and at about half the price the
Mexican government offered to farmers who had corn
to sell. With about half of the days worked in
Mexican agriculture devoted to corn production, free-
ing up trade in corn meant that millions of rural
Mexicans would have to find alternative jobs. The
hope was that many Mexican farmers would shift
from producing corn to labor-intensive fruits and veg-
etables, but lack of knowledge, investment, and infra-
structure, plus the continued availability of Mexican
farm workers in the United States, meant that much
of the growth in labor-intensive agriculture occurred
in the United States rather than Mexico.
Mexicans could have stayed in Mexico and found
non-farm jobs. However, the creation of formal sec-
tor jobs has been too slow to absorb the growing
labor force and make a dent in widespread underem-ployment. Mexico’s GDP growth has been less thanthree percent over the past three decades, includingthe 15 years since NAFTA went into effect; with thepopulation growing by over one percent a year, percapita GDP growth averaged about 1.5 percent peryear. The gap between GDP per capita in the UnitedStates and Mexico, about five to one according tothe World Bank, has narrowed only slightly sinceNAFTA went into effect in 1994, when the gap wasabout six to one.
Most experts say that the Mexican economy mustaverage six percent economic growth to create enoughjobs to reduce the gap in GDP per capita and the eco-nomic incentive to migrate (Weintraub 2010).4 TheMexican education system must also be reformed tojustify higher Mexican wages. Mexicans adults in2010 have an average eight years schooling. TheMexican government via the Opportunidades(Progresa from 1997–2002) is trying to break theinter-generational transmission of poverty by educat-ing children, providing payments to the mothers ofpoor children every two months if the child was inschool at least 85 percent of the time. The paymentsincrease as children get older, reflecting the risingopportunity cost of going to school, since 15 and16 year olds could work and earn more than youngerchildren (Levy 2006). In 2010, Opportunidades pro-vided support to five million poor Mexican families(there are about 26 million families in Mexico) at acost of 3.6 billion US dollars. About 85 percent ofOpportunidades recipients are in rural Mexico.
By 2010, Mexico had a population of 110 million anda labor force of 45 million. Both the population andlabor force are expanding by a million a year (OECD2010). Mexico offers too few formal-sector jobs tokeep Mexicans at home. Only half of Mexico’s work-ers are wage and salary employees, according to theOECD, and many of those who appear to have formalsector jobs are employed only part time, making themineligible for these benefits. The number of privatesector jobs enrolled in IMSS, Mexico’s social securitysystem, has remained at about 15 million since 2000.There are about eight million Mexican-born workersin the United States, which means that a third ofMexicans with formal-sector jobs are in the UnitedStates. Until Mexico can add formal-sector jobs at afaster pace over time, especially young Mexicans with
4 Weintraub (2010) characterizes the Mexico-US relationship asdependent-dominant, that is, the Mexican economy is dependent onthe dominant United States to accept its exports and its migrants.
CESifo Forum 4/201041
Focus
friends and relatives already in the United States will
continue to migrate north for opportunity.
The migration hump
NAFTA accelerated changes already underway in
Mexico, adding to rural-urban migration, creating
higher wage jobs near the Mexico-US border, and
displacing workers from jobs in previously protect-
ed industries in central Mexico. These NAFTA-
linked changes meant that some Mexicans who
were previously protected behind trade barriers in
agriculture and industry had to find new jobs, and
made it clear to many small farmers in rural
Mexico that a better life lay elsewhere. However,
poor education systems in rural Mexico meant that
many of those leaving rural Mexico were not pre-
pared for good jobs in Mexico or the United States
(Levy 2008).
NAFTA provides an example of an economic policy
that produces pain before gain, that is, the same
trade and investment policies that should reduce
unwanted international migration in the long term
can increase such migration in the short term.
Analyses of free trade agreements generally find that
countries that are more fully integrated into the
world economy grow faster than countries that are
more closed to the global economy. However, these
comparative static analyses are based on before and
after comparisons; they often ignore the migration
that occurs as richer and poorer economies with a
pre-existing migration relationship lower barriers to
trade and investment.
International migration, legal and unauthorized, can
rise as especially poorer economies already experienc-
ing out-migration adjust to economic integration, cre-
ating a migration hump. Reducing trade barriers can
displace workers in both poorer and richer countries,
as when garment and auto parts factories closed in
the United States when these factories shifted to
Mexico and Mexican farmers and workers making
heavy equipment lose their jobs as imports from the
United States rise. The displaced US workers rarely
migrate to Mexico, but some of those displaced in
Mexico migrate to the United States, so that migra-
tion and trade can increase together, especially if freer
trade displaces Mexican workers faster than foreign
investment creates new jobs in Mexico. Eventually,
faster economic and job growth in Mexico should
reduce Mexico-US migration.
The migration hump is pictured in Figure 1. The
solid line through B represents the rising migration
expected without economic integration as a result of
rapid labor force and slow formal sector growth in
Mexico. The extra migration in the hump is pictured
above A, which shows how economic integration
leads to an increase in migration over the status quo
trajectory. However, economic integration should
also speed up economic and job growth in Mexico,
so that migration falls below the status quo level at
B, after 15 years in the figure. As faster economic
and job growth in Mexico continues, migration is
projected to fall even more, and area C represents the
migration avoided by economic integration.
Eventually, some migrants may return from abroad
as net emigration turns into net immigration, as
occurred in Ireland, Italy and Spain.
Table 1
Mexico-US population and labor force 1970–2010
Year Mexico United States
1970 Population (million)
Labor force (million)
Labor force as a share of population (%)
53
15
28
203
83
41
Population (million)
Labor force (million)
Labor force as a share of population (%)
110
45
40
310
155
50
2010
Employment
Total formal sector jobs (million)
• Filled with Mexicans (million)
Employment in Agriculture (million)
• Filled with Mexicans (million)
15
15
7
7
125
6
3
2
2050 Population (million)
Labor force (million)
Labor force growth 2010–2050 (%)
130
65
45
425
212
37
Sources: US Census and Conapo for the United States, 2050 projections from Population Reference Bureau
(PRB); Instituto Mexicano del Seguro Social (IMSS), Subdirección General de Finanzas for Mexico.
CESifo Forum 4/2010 42
Focus
The critical policy parameters are A, B, and C – howmuch additional migration results from economicintegration (A), how soon does the hump disappearas migration returns to the status quo level (B), andhow much migration is avoided by economic integra-tion and other changes (C)? Generally, there must bea pre-existing migration relationship and three morefactors for economic integration between poorer andricher countries to lead to a migration hump: a con-tinued demand-pull for migrants in the destinationcountry, an increased supply-push in the country oforigin as freer trade displaces workers and accelerateschanges under way such as rural-urban migration,and migration networks that can move workersacross borders.
These conditions were present in the case of Mexicoand the United States. Migration was the major rela-tionship between Mexico and the United States formost of the past half century, which enabled migra-tion networks to make rural Mexicans aware of jobopportunities in the United States. The US govern-ment did not develop an effective policy to keep unau-thorized workers out of US jobs and, when Mexicangovernment policies and NAFTA accelerated dis-placement in rural Mexico, more Mexicans migratedto the United States. Today, 10 percent of the 120 mil-lion people born in Mexico live in the United States.With their US-born children, the United States hasabsorbed at least 25 million people who would other-wise live in Mexico.
There are three central policy messages of the migra-tion hump. The first is the need for policy makers toacknowledge the potential for a pain-before-gain
increase in migration with closereconomic integration, requiringe.g. the Mexican government tocooperate to reduce the openmassing of Mexicans on theMexican side of the border toattempt illegal US entry. The sec-ond policy message is that tradeand other policies should beimplemented in ways that makethe migration hump smaller andshorter, such as phasing in freertrade slowly in areas with themost labor displacement and notchanging policies that add tomigration, as when Mexicoallowed the rental and sale of
ejido land at about the same timethat NAFTA went into effect.
The third policy message is to link economic integra-tion with enhanced cooperation to manage migrationso that the migration hump migration does not lead toa backlash that slows freer trade and investment.Mexican and US trade negotiators, saying that it washard enough to negotiate NAFTA, took migration offthe table. Except for NAFTA’s Chapter 16, which cre-ates freedom of movement for college educated resi-dents of Canada, Mexico, and the United States whohave job offers in another NAFTA country5, migra-tion is not dealt with in NAFTA.
Mexican and US responses
The Mexican government’s response to risingMexico-US migration in the 1990s was to emphasizethat most Mexican migrants found US jobs, suggest-ing that they were needed because US employershired them. The Mexican government argued that themigration hump should be turned into legal migra-tion with a guest worker program negotiated cooper-atively. Then Mexican Foreign Relations SecretaryJose Angel Gurria in 1996 said: “the phenomenon ofmigration can be beneficial and offers great potentialadvantages to both the migrants’ original and newcountries, but only when a common vision on thisphenomenon is reached”.6
0 5 10 15 20 25 30 35 40
Source: Author's conception.
THE MIGRATION HUMP Migration flow
Status quo pattern
Migration avoidedC
A B
Additional migration
Migration pattern witheconomic restructuring
Year ofeconomic
restructuring
DYears
Figure 1
5 The major effect of Chapter 16 has been to move Canadian profes-sionals into US jobs with TN visas. About 20 percent of Mexicanswith doctorates are in the United States, and half of the20,000 Mexican PhD holders in the United States entered since 1990.6 Quoted in Migration News (1996), Mexican Views onImmigration, Vol. 3, No. 11,http://migration.ucdavis.edu/mn/more.php?id=1072_0_2_0
CESifo Forum 4/201043
Focus
The efforts of commissions and task forces to designmutually acceptable guest worker programs were notsuccessful, even during the late 1990s when the USunemployment rate dipped below five percent. Therewas hope for a new era in Mexico-US migration in2000, when Mexicans elected Vicente Fox President inJuly 2000, breaking the 70-year hold of theInstitutional Revolutionary Party (PRI) on power,and George W. Bush, who advocated a new guestworker program during the campaign, was elected USPresident in November 2000.
During the spring and summer of 2001, Fox pressedBush for a comprehensive immigration reform. Foxwas especially bold in calling for freedom of move-ment. In January 2001, Fox said: “when we think of2025, there is not going to be a border. There will bea free movement of people just like the free move-ment of goods”.7 Some US and Mexican advisorsurged Fox to press for more than just a guest workerprogram, and Mexican Foreign Minister JorgeCastaneda in June 2001 summarized Mexico’s four-pronged immigration agenda to include legalizationfor unauthorized Mexicans, a large guest-worker pro-gram, a reduction in border violence and an exemp-tion of Mexico from visa quotas, ending with whatbecame a memorable phrase: “it’s the whole enchila-
da or nothing”.
Fox and Castenada were in Washington DC justbefore the 11 September 2001 terrorist attacks, whenFox said that he expected the United States to enactcomprehensive immigration reform before the end ofthe year.8 The terrorist attacks turned attention else-where and accelerated the addition of Border Patrolagents and fences on the Mexico-US border so thatterrorists could not use the migration infrastructuredeveloped to move Mexican workers into the UnitedStates. Smuggling fees rose from 200 to 300 US dol-lars in the early 1990s to 2,000 to 3,000 US dollars adecade later, and migrants who paid more to enter theUnited States illegally were more likely to settlebecause of the difficulty and expense of re-entry.
In 2005, when the US unemployment rate droppedtoward five percent and unauthorized Mexicans dif-
fused throughout the United States in agriculture,construction and services, Congress once again tack-led Mexico-US migration. Under Republican leader-ship, the House approved the Border Protection,Antiterrorism, and Illegal Immigration Control Actin December 2005, which took an enforcement-onlyapproach to unauthorized migration. The House billcalled for mandatory screening of newly hired as wellas existing employees to ensure they are legally autho-rized to work in the United States and would havemade ‘illegal presence’ in the United States a felony,which would have made harder for unauthorized for-eigners to eventually become legal immigrants. Thisso-called Sensenbrenner bill prompted massivedemonstrations against an enforcement-onlyapproach and in favor of legalization in spring 2006,culminating in a May 1 ‘day without immigrants’protest.
Under Democratic leadership, the Senate took a dif-ferent approach, approving the ComprehensiveImmigration Reform Act (CIRA) in May 2006.CIRA included more enforcement, as in the Housebill, as well as new earned legalization and guestworker programs. Unauthorized foreigners in theUnited States at least five years could become ‘pro-bationary immigrants’ by proving they had worked inthe United States, paid any back taxes owed and aUSD 1,500 fee, and passed English and backgroundtests. After six years of continued US work and taxpayments and another USD 1,500 fee, they couldapply for immigrant visas, a process called fees are‘earned legalization’.9 This so-called Kennedy-McCain bill approved by the Senate was not consid-ered by the House and died.
The Senate debated a revised Comprehensive Immi-gration Reform Act (CIRA) in 2007 that was tougheron unauthorized foreigners, but opponents blocked avote on the bill despite the active support of PresidentGeorge W. Bush. During the 2008 presidential cam-paign, Barack Obama expressed support for CIRA,but John McCain reversed his previous support andsaid that enforcement must precede legalization.Obama has several times repeated his support for com-prehensive immigration reform, but Congressionalaction has been blocked by opponents of legalizationand opponents of new enforcement efforts that couldincrease discrimination against minorities.
7 Quoted in Migration News (2001). Mexico: Guest Workers, Vol. 8,No. 2, http://migration.ucdavis.edu/mn/more.php?id=2297_0_2_08 Fox said: “the time has come to give migrants and their communi-ties their proper place in the history of our bilateral relations … wemust, and we can, reach an agreement on migration before the endof this very year ... [so that] there are no Mexicans who have notentered this country legally in the United States, and that thoseMexicans who come into the country do so with proper documents”– quoted in Migration News (2001), Fox Visits Bush, Vol. 8, No. 10,http://migration.ucdavis.edu/mn/more.php?id=2463_0_2_0.
9 Unauthorized foreigners in the United States for two to five yearswould have to satisfy the same requirements, but in addition returnto their countries of origin and re-enter the United States legally, andthose in the United States less than two years would be expected todepart, although they could return legally as guest workers.
CESifo Forum 4/2010 44
Focus
The stalemate in Congress has encouraged somestates and cities to enact laws to deal with illegalmigration. Some have been friendly to unauthorizedforeigners, such as issuing them ID cards or prohibit-ing police from asking about the immigration statusof persons they encounter, but most have aimed tomake life difficult for unauthorized foreigners.Arizona in April 2010 enacted the Support Our LawEnforcement and Safe Neighborhoods Act (SB 1070),which makes it a state crime for unauthorized for-eigners to be in the state. A federal court injunctionblocked SB 1070 from going into effect, agreeing withopponents that regulating immigration was strictly afederal responsibility. Mexican President FelipeCalderón, who reportedly has relatives living illegallyin the United States, said that SB 1070 “introduces aterrible idea: using racial profiling as a basis for lawenforcement”.
Legal and unauthorized Mexico-US migration con-tinues with no end in sight. Economic and jobgrowth in Mexico was slowed by the 2008–2009recession and by longer term factors that range fromthe rising preference of some foreign investors forChina and other Asian countries to a governmenteffort to deal with drug gangs that left over30,000 people dead in four years. The United States,which has about 15 million unemployed workers, isprojected to have an unemployment of over nine per-cent for the next several years, and to experienceespecially slow growth in the residential constructionemployment that employed many Mexican-bornworkers.10 Sluggish growth and drug wars in Mexicomay encourage Mexicans to migrate northward,while more border enforcement and high unemploy-ment may counter this migration, making migrationa continued irritant in Mexico-US relations even asNAFTA promotes economic integration.
References
Cornelius, W. and D. Myhre (1998, eds.), The Transformation of RuralMexico: Reforming the Ejido Sector, Boulder: Lynne Rienner Pub-lishers.
Levy, S. (2006), Progress Against Poverty: Sustaining Mexico’sProgresa-Oportunidades Program, Washington DC: Brookings Insti-tution.
Levy, S. (2008), Good Intentions, Bad Outcomes: Social Policy,Informality, and Economic Growth in Mexico, Washington DC:Brookings Institution.
Martin, P. (2009), Importing Poverty? Immigration and the ChangingFace of Rural America, New Haven: Yale University Press.
Riding, A. (1989), Distant Neighbors: A Portrait of the Mexicans,New York: Vintage.
Weintraub, S. (2010), Unequal Partners: The United States and Mexi-co, Pittsburgh: University of Pittsburgh Press.
10 Employment in residential building construction (NAICS) peakedat a million in 2006 and fell below 600,000 in 2010.
CESifo Forum 4/201045
Focus
NORTH AMERICA’S UPHILL
BATTLE ON CLIMATE CHANGE
AND ITS IMPLICATIONS FOR
THE NORTH AMERICAN
TRADING SYSTEM
MEERA FICKLING*
Introduction
As the debate over climate change legislation brewedin the US House and Senate in 2009 and early 2010,its implications for heavily traded manufacturing werehotly contested. Concerned that higher energy priceswould cause firms to produce these goods outside theUnited States instead of complying with US regula-tions, lawmakers offered free emissions allowances tocompensate domestic firms and threatened to requireimporters from countries that did not meet US cli-mate standards to purchase emissions allowances atthe border.
A year later, passage of US climate legislation nowseems all but impossible. As the United States hasbacked away from its initial ambitions, NorthAmerica as a whole seems likely to follow suit. Asidefrom anemic federal regulation under the US CleanAir Act, climate change legislation has been left up tostates and provinces for the time being.
The fragmentation of climate change policy has itsown implications for North American trade andinvestment. Though the scale is now far smaller, regu-lation at the provincial level makes it more likely thatemissions could migrate to other areas. States andprovinces also have less capacity to adjust their trans-mission infrastructure to ramp up renewable electrici-ty production at least cost. This does not mean thatstate and provincial policy is not crucial to North
America’s climate change mitigation goals – for now,it seems that state policy is the only feasible means ofworking toward climate change mitigation in NorthAmerica. However, this is not the best possible situa-tion. Federal climate policy, though jettisoned due toconcerns about jobs and competitiveness, would actu-ally be more effective at reducing emissions withoutadverse trade impacts.
The volume of North American energy trade is large.The United States derives about a fifth of its oil fromCanada, and in 2008 about two thirds of the crude oilproduced in Canada was shipped to the UnitedStates.1 Canada is the largest supplier by far of ener-gy-intensive manufactures to the United States,including steel (20 percent of US imports), cement(53 percent of US imports), paper (52 percent of USimports), and aluminum (55 percent of US imports).In total, Canada exported USD 44 billion of highlytraded, energy-intensive products to the United Statesin 2008 (Hufbauer and Kim 2009).2
As a consequence, it is important to understandwhat state legislation means for trade in energy andenergy-intensive goods. This paper explores thisissue below.
Prospects for North American action
Federal
Prospects for comprehensive federal climate changelegislation in North America are not bright. Thoughthe United States seemed promising in 2009, with thepassage of the Waxman-Markey bill through theHouse of Representatives, legislation stalled in theSenate in 2010 as partisan acrimony intensified andthe political mood soured on cap-and-trade. To cap itoff, the Democratic Party, which is far more support-ive of climate change legislation, experienced heavylosses in Congressional midterm elections. There is lit-tle likelihood in the near future that Republicans in
* Peterson Institute for International Economics, Washington DC.This article draws heavily on the forthcoming book, NAFTA andClimate Change by Meera Fickling and Jeffrey J. Schott, to be pub-lished by the Peterson Institute for International Economics inearly 2011.
1 Data obtained from the Energy Information Administration andStatistics Canada, 2009.2 Based on products scheduled to receive allowance rebates under theWaxman-Markey and Kerry-Lieberman bills.
the House and the Senate will rally around a keyObama initiative that most of them campaignedagainst. Even initiatives such as a renewable portfoliostandard that once seemed politically safe now seemless likely.
Environmental advocates have turned to new chan-nels. One of the chief tools in the ObamaAdministration’s arsenal is the Clean Air Act, which isadministered by the Environmental ProtectionAgency (EPA). In 2007, the Supreme Court ruled thatthe EPA had the authority to regulate greenhousegases (GHGs), which cause climate change, under theAct.3 Technically, the EPA could achieve substantialGHG emissions reductions through this approach(Bianco and Litz 2010) – but this would involve imple-menting a much more far-reaching regime thanCongress is likely to allow. Already, a number of res-olutions have been introduced in Congress to limit theEPA’s powers. These measures have so far been kept atbay, but the incoming Congress promises to be evenless sympathetic to the EPA than before.
So far, EPA regulation has been modest. The EPA hasannounced that it will require new sources and majormodifications producing more than 75,000 tons ofCO2e per year – in other words, new projects thatwould produce a large amount of emissions – to obtainPrevention of Significant Deterioration (PSD) permits.The PSD permitting system is administered on a state-by-state basis. In November 2010, the EPA released itsproposed guidance to the states for PSD permitting.The document indicates that a relatively lenientapproach will be taken toward coal-fired power plants,relative to the possibilities. Energy efficiency improve-ments are on the table, but the guidance document dis-courages states from requiring coal-fired power plantsto switch to coal or biomass or implement carbon cap-ture and sequestration.4 These energy efficiencyimprovements are not expected to produce substantialemissions reductions; Richardson et al. (2010) estimatethat modest energy efficiency improvements in coalplants could reduce GHG emissions by 3 percent.
If the United States is not able to pull together a pro-gram to substantially reduce GHGs, Canada isunlikely to do so either. Over the past couple of years,the Canadian government has transitioned from atleast a nominal policy-maker on climate change to a
taker of policy from the United States. In 2007, theCanadian government released Turning the Corner, arelatively ambitious plan that aimed to reduce emis-sions by 20 percent from 2006 levels by 2020.Subsequently, citing fears of lost industrial competi-tiveness, it has backed away from this plan and insteadpromised to emulate whatever the US action turnedout to be. By early 2010, Canada had shifted so fartoward this stance that its Copenhagen pledge read,“17 percent, to be aligned with the final economy-wide emissions target of enacted US legislation”(UNFCCC 2010a).
The host of the 2010 COP 16 in Cancun, Mexico hasshown international leadership on climate change. Atthe 14th session of the UNFCCC Conference ofParties (COP 14) in Poznan, Poland in December2008, it announced its intention to reduce emissions50 percent from 2002 levels by 2050, contingent upondeveloped-country assistance. In the run-up toCopenhagen, Mexico proposed a ‘Green Fund’, awell-received plan to provide 10 billion US dollars peryear toward mitigation, adaptation, and technologytransfer. And its Copenhagen submission committedto reduce emissions 51 MMT below business-as-usuallevels by 2012 (equivalent to about a 6.4 percent cut)and 30 percent below business as usual by 2020(equivalent to about 250 MMT of carbon dioxide-equivalent) – see UNFCCC (2010b).5
Mexico has been active at home as well. Its 2009Special Climate Change Program (PECC) lays out anumber of actions toward meeting its 2012Copenhagen target, as well as the agencies responsiblefor their implementation, the Secretariat ofEnvironment and Natural Resources (SEMARNAT)and the National Energy Secretariat (SENER).Proposed actions include management of landfill gas;expansion of sustainable forest management, includ-ing expansion of the payment for environmental ser-vices scheme; self-supply schemes for renewable ener-gy; and wind power generation by the FederalElectricity Commission (CFE).
However, all of Mexico’s ambitious undertakings areconditioned upon adequate financing from developedcountries.6 And foreseeable funding sources depend at
CESifo Forum 4/2010 46
Focus
3 Commonwealth of Massachusetts v. Environmental ProtectionAgency. 4 Carbon capture and sequestration technology is not yet commer-cially viable.
5 The 2012 percentage emissions reduction estimate and the 2020absolute emissions reduction estimate are calculated by the authorsbased on business-as-usual emissions projections in Centro MarioMolina (2008). 6 Mexico’s commitment to reduce emissions 30 percent by 2020 isconditioned upon “the provision of adequate financial and techno-logical support from developed countries as part of a global agree-ment” (UNFCCC 2010b), and its Poznan commitment is also con-tingent upon developed country financing.
CESifo Forum 4/201047
Focus
least in part on the United States. US Secretary of
State Hillary Clinton provided the impetus behind the
USD 100 billion climate change fund promised at
Copenhagen, and the United States will be expected
to make a substantial contribution to this fund. A US
cap-and-trade program with offset and allowance
trading provisions could provide additional financial
incentives for Mexican firms to use cleaner technolo-
gies and production methods. While Mexico might
benefit from selling renewable electricity and carbon
offsets to California, which is implementing climate
change legislation on a far smaller scale, it seems
unlikely that the United States Congress will make cli-
mate change a high priority in the near future. As a
consequence, the United States cannot be counted on
as a key source of funding in the near future.
State
Despite the dismal outlook described above, climate
change regulations cannot be written off just yet.
Action is occurring at the sub-federal level. The
Regional Greenhouse Gas Initiative (RGGI), a cap-
and-trade system for electricity emissions operated by
a group of Northeastern US states, has been active
since September 2008. The RGGI aims to reduce
emissions produced from electricity by 12 percent by
2020. While emissions reductions are not required
until 2015, auctions have been taking place since the
program became active.
California is creating a cap-and-trade program and
recently released its proposed regulations. This regu-
lation is part of the Western Climate Initiative
(WCI), a group of US states and Canadian provinces
that have promised to reduce their economies’ emis-
sions by 15 percent from 2005 levels by 2020. The
program will cover 85 percent of the state’s emis-
sions and will take effect in 2012 for electricity gen-
erators and large industrial sources and 2015 for
transportation, residential and commercial fuels.
Other parties that do not fall into these categories
can voluntarily opt into the emissions trading pro-
gram (Fickling 2010).
A few Canadian provinces, together comprising
about half of Canada’s emissions, also appear on
track to implement cap-and-trade programs under
the WCI banner by 2012. British Columbia, Que-
bec, and Ontario have all passed bills authorizing
cap-and-trade systems, and British Columbia
released draft regulations in October 2010 (Fickling
2010).
In addition, states have implemented a number of per-
formance standards. One of the most prominent is the
renewable portfolio standard, which requires utilities
to procure a certain percentage of total retail electric-
ity sales from renewable sources. This standard has
been implemented in thirty US states and a handful of
Canadian provinces. Target renewable percentages
range from California’s standard, which aims to meet
a third of the state’s electricity demand though renew-
able resources by 2020, to Texas’s standard, which
requires utilities to procure 5 percent of electricity
from renewable sources by 2015 (Fickling 2010).
Sources that are most often eligible to fulfill state
renewable energy standards include wind, solar,
geothermal, landfill gas, and ocean energy. Hydro-
power, biomass, and municipal solid waste tend to
be given lower priority under a renewable portfolio
standard, if such sources are eligible at all.
Renewable portfolio standards are often combined
with energy efficiency resource standards, which
require utilities to reduce demand through energy
efficiency measures (Fickling 2010).
Another measure that has gained some traction in the
United States is the low-carbon fuel standard (LCFS),
which is currently being piloted by California. The
LCFS is intended to replace the state’s ethanol blend-
ing requirement, which has been criticized for failing
to discriminate between relatively GHG-intensive bio-
fuels such as corn-based ethanol and lower-impact
biofuels such as sugarcane and cellulosic ethanol.
Instead, the LCFS requires the California Air
Resources Board (CARB) to assign a GHG intensity
value to each transport fuel sold in California.
To obtain these values, CARB must quantify the
GHGs emitted over the lifecycle of a fuel, from the
extraction of the raw materials used to produce the
fuel to the burning of the fuel in one’s car. The
standard then requires fuel retailers to reduce the
GHG intensity of their fuel sales by 10 percent by
2020. Proponents of the standard argue that this
process emphasizes reducing carbon emissions
rather than promoting American corn growers,
although critics argue that CARB uses a flawed
methodology to calculate lifecycle intensity values
(Fickling and Schott 2011).
Trilateral
At least until Canadian Environment Minister Jim
Prentice left office in November 2010, trilateral dia-
logue on climate change policy benefited from a close
working relationship among the three environment
ministers, as well as a commitment to trilateral energy
cooperation among heads of state. After nearly a
decade of inaction on climate change policy, the
North American Leaders’ Summit in August 2009
refocused attention on sustainable energy issues and
instructed officials to develop a trilateral working
plan for cooperation on energy science and technolo-
gy. North American environmental ministers also
committed to improving the comparability of data
gathering and inventories for mitigation and adapta-
tion projects at the NAFTA Commission for
Environmental Cooperation (CEC) ministerial in
August 2010.
Bilateral arrangements were also advanced. In
2009, US President Barack Obama and Canadian
Prime Minister Stephen Harper established a Clean
Energy Dialogue to coordinate on carbon capture
and storage research and modernization of the elec-
tric grid. Obama and Mexican President Felipe
Calderon created the US-Mexico Bilateral
Framework on Clean Energy and Climate Change,
agreeing to collaborate on low-carbon technology
development and capacity building, as well as adap-
tation to climate change.
A substantial roadblock is the low level of funding for
climate change activities within NAFTA environmen-
tal institutions. The CEC amasses environmental
information, provides recommendations on trilateral
environmental issues, and promotes environmental
law enforcement – all with a 9 million US dollar bud-
get. The amount of money allocated to the CEC has
remained unchanged since its creation in 1994, despite
inflation and exchange rate movements (the budget is
expressed in dollars, while the CEC is physically locat-
ed in Canada) that have decreased the real value of
this amount. Although the CEC has been effective for
its size, its budget constrains the scope of its opera-
tions (Hufbauer and Schott 2005).
The North American Development Bank (NADB)
and Border Environment Cooperation Commission
(BECC), two interrelated institutions that fund envi-
ronmental projects on the US-Mexico border, receive
far more money in the form of cash and loan guaran-
tees. The capital base of the NADB is 3 billion US
dollars per year (Hufbauer and Schott 2005). Lending
started out far below this capacity due to high interest
rates and a cumbersome application process, but the
NADB/BECC has improved markedly in recent years,
offering subsidized loans to needy communities and
grants from paid-in capital (Kass and McCarroll
2008). Total loans disbursed have skyrocketed from
USD 11 million as of 2002 to USD 1.1 billion as of
September 2010 (NADB 2010). While the scope of
the NADB/BECC has officially expanded to include
renewable energy and energy efficiency, however, these
issues still make up a tiny portion of the institutions’
overall lending portfolio (NADB 2010).
What does this mean for North America?
State policies are an improvement over no policy.
State environmental regulation has historically served
as a ‘laboratory for innovation’, paving the way for
more stringent federal environmental policy than
would otherwise be adopted and allowing ideas to be
tested on a small scale before they are implemented
nationally. A prominent recent example is California’s
‘Pavley’ automobile tailpipe emissions standards,
which required automakers to achieve a fleet average
fuel efficiency of 36 miles per gallon. After fourteen
other US states and four Canadian provinces adopted
these requirements, the federal governments of the
United States and Canada implemented them on a
national scale.
However, the fragmentation of climate policy is ideal
neither for the climate nor for the North American
trading system. A robust network of state cap-and-
trade regimes could generate substantial reductions.
But the group of states likely to adopt a cap-and-
trade approach – a handful of Northeastern states
and Canadian provinces, plus California – cannot
produce anything close to the amount of reductions
promised by the United States and Canada in
Copenhagen.
State climate policy is more susceptible to so-called
‘carbon leakage’. Leakage can occur at either the pro-
ducer or the consumer level. Climate change policy
aims to make it more expensive to produce and con-
sume carbon-intensive goods. Because climate policy
is not globally integrated, firms that produce these
goods could simply move to unregulated jurisdictions
rather than reduce their emissions. Likewise, envision
a firm selling two comparable products, one clean and
one dirty, to two places, one with regulations and one
without. The firm could simply sell its dirtier good to
consumers in the unregulated area and its cleaner
good to customers in the regulated area, without nec-
essarily changing its overall production of each good.
CESifo Forum 4/2010 48
Focus
CESifo Forum 4/201049
Focus
The extent to which leakage could occur depends
upon a host of factors, including the extent to which
energy prices increase, the energy intensity of in-state
industries, and the ease with which firms can shift
production and sales. States and provinces that are
likely to adopt comprehensive climate change legisla-
tion do not tend to have high concentrations of ener-
gy-intensive, trade-exposed industries. However,
states are particularly economically integrated with
each other, so production and consumption may shift
more easily to unregulated areas than if policy were
adopted at a national level. Modeling of state cap-
and-trade policies produces mixed results as to the
probability of leakage according to RGGI (2007) and
WCI (2008).
In addition, climate policy that is limited to only a
handful of states and provinces could worsen the
already existent disparities in energy consumption
between leading and laggard regions of North
America. The difference between the GHGs emitted
by California and Wyoming per capita is large, partic-
ularly when it comes to the type of electricity used by
the two states. Carbon-intensive production is also
concentrated in certain states and provinces, particu-
larly in the American Midwest and Southeast and the
western Canadian provinces. These regional differ-
ences already pose frictions for North American poli-
cymakers, who have to construct national climate
change regimes without imposing a disproportionate
economic burden on any one region (Fickling and
Schott 2011). If some states make significant progress
before others catch up, this could exacerbate the diffi-
culty of passing national legislation.
There are difficulties coordinating various state stan-
dards. Whereas some standards such as automobile
and fuel standards are modeled off California’s, oth-
ers such as the renewable portfolio standard are
designed on a state-by-state basis, with little agree-
ment as to what constitutes renewable energy or
where it must come from in order to meet a particu-
lar standard. As a result, it is difficult to trade credits
across regions – even though energy from a different
region might achieve the common goal of reducing
emissions.
Economies of scale for renewable generation are less
easily captured through state action. When renewable
electricity is generated over a wide area, rather than in
a single locality, it is less susceptible to changes in
weather patterns. States also cannot single-handedly
create the infrastructure necessary to take full advan-
tage of renewable energy. Expanded transmissionwould allow renewable electricity to be sold fromareas with high potential to areas with lower poten-tial. A recent US Department of Energy report pointsout that wind power could comprise 20 percent ofnational electricity generation if, among other things,transmission capacity were significantly expanded(Department of Energy 2008). A smart grid thatmoves some electricity usage from peak hours to off-peak hours can also help ‘soak up’ excess wind powergenerated during off-peak hours. Such a project mustbe coordinated among many states; one state alonecannot revamp the current antiquated grid.
State standards, trade politics and trade law
Recent US federal cap-and-trade proposals includedwidely debated measures to protect the competitive-ness of domestic manufacturing. Most federal legis-lation introduced after 2007 included a so-calledinternational reserve allowance program (IRAP),which required importers to purchase allowances atthe border to compensate for the difference betweenthe cost of production at home and the cost of pro-duction in unregulated jurisdictions. The IRAP wasimposed based on country of origin – imports fromcountries that had not taken sufficient action on cli-mate change would be subject to the border measure,with exceptions only for least developed countriesand de minimis emitters. The measure was a signifi-cant concern for Canada and Mexico, whose officialsfeared that their large volumes of carbon-intensiveexports could be put at risk.7 It was also of question-able WTO legality.
Even though federal US cap-and-trade legislation isstalled for the time being, Canada and Mexico are notoff the hook. Carbon tariffs are a response to domes-tic political pressures from carbon-intensive industriesthat anticipated a high cost of reducing emissions andtherefore feared a loss of competitiveness. As notedabove, these pressures were particularly potentbecause most carbon-intensive industries are dispro-portionately concentrated in politically importantstates. If these states fall even further behind others in reducing emissions, this will both increase the to-tal amount of emissions that the United States must eventually eliminate in order to meet its interna-tional goals and increase the gap between the cost ofnational climate policy to leading states and the cost
7 For more information, see Schott and Fickling (2009).
of national climate policy to laggard states. Con-sequently, protectionist pressures could return with avengeance if and when the United States musters thepolitical will to give cap-and-trade another try.
Although state legislation avoids some of the frictionscaused by recent national cap-and-trade proposals,some standards are particularly controversial.Notably, California’s low carbon fuel standard sepa-rates petroleum into two categories: ‘conventional’and ‘unconventional’. Whereas conventional fuels areall assigned the same lifecycle carbon intensity value(equivalent to the weighted average carbon intensityof the fuels consumed in California), unconventionalfuels such as oil sands crude, etc. are given a separatelifecycle analysis. Unsurprisingly, Canada, whichexports 2.5 million barrels of oil sands crude to theUnited States every day, has expressed its concern thatthis measure violates WTO rules.8
At issue is whether the distinction between oil sandscrude and conventional crude is arbitrary or justifi-able. In order to defend its standard before a WTOdispute settlement panel, California would have toprove that environmental considerations require oilsands crude to be treated differently from cruderecovered via conventional drilling processes. Thepanel’s decision could hinge upon whether oil sandscrude is sufficiently more GHG-intensive to merit aseparate category. While oil sands crude is about 15to 20 percent more GHG-intensive than the averageconventional crude to produce and use, the differ-ence between it and heavier crudes is narrower(Toman et al. 2008).
Other standards have created somewhat less contro-versy but have nevertheless proved to be sources offriction between states and provinces. Some renewableportfolio standards require qualifying electricity to beproduced in state, or exclude certain sources of elec-tricity from qualifying as renewable. Many US staterenewable electricity programs have excluded largehydropower, a decision that United States andCanada have long disputed (Rowlands 2009).Ontario’s feed-in tariffs for renewable electricity areconditioned upon a domestic content requirement of25 percent for wind turbines and 50 percent for solarpanels. This provision has been challenged in WTOdispute settlement by Japan, the United States, andthe European Union. Some states also have require-
ments that are transparently engineered to favor
locally important industries; an example is North
Carolina, which requires utilities to generate a certain
percentage of electricity from swine and poultry waste
(Fickling 2010).
Conclusion and policy recommendations
While coordinated federal legislation in the United
States, Canada and Mexico would be ideal, this paper
offers some modest steps that the two countries could
take to coordinate state initiatives. The CEC could
function as a clearinghouse for climate change-related
data. With modest budgetary increments, this institu-
tion could play a significant role in NAFTA climate
change initiatives by expanding its database on North
American emissions and reporting on new climate ini-
tiatives and regulations in each country. In so doing,
the CEC could become a North American clearing-
house for monitoring, reporting, and verification
(MRV) of carbon credits issued under provincial or
regional carbon regimes, which could lower transac-
tion costs of offset projects among the three North
American countries. The scope of the NADB could
also be expanded to include more projects related to
clean energy and energy efficiency.
States and provinces are already discussing the possi-
bility of mutual recognition of carbon credits gener-
ated by various regional cap-and-trade schemes, and
they should continue to study options for coordinat-
ing or integrating these evolving carbon regimes.
Policy coordination could facilitate carbon credit
trading by ensuring that carbon credits in all jurisdic-
tions represent similar kinds of carbon reductions. In
addition, greater coordination among carbon trading
regimes could help address concerns regarding ‘car-
bon leakage’ that have plagued the implementation of
cap-and-trade programs in many states.
Somewhat more controversially, states and provinces
may work toward coordinated renewable electricity
policies. All parties should agree on how imported
electricity should be credited and certified under
renewable portfolio standards, both at the federal and
state levels. To the extent feasible, states and provinces
should harmonize definitions of renewable electricity
in order to stimulate development by increasing the
fungibility of RECs. Harmonization and expansion
of renewable energy credit tracking systems could
also widen the geographic area from which renewable
credits could be purchased.
CESifo Forum 4/2010 50
Focus
8 See Energy Information Administration (2010) and letter fromCanadian Ambassador Michael Wilson to Mary Nichols,14 November 2008 (http://www.canadainternational.gc.ca/washing-ton/events-evenements/LCFS_Nichols.aspx?lang=eng.).
CESifo Forum 4/201051
Focus
The North American countries should shield climatechange taxes and regulations from claims under theindirect takings provisions of NAFTA Chapter 11.Chapter 11 requires governments to provide com-pensation to investors for measures that are ‘tanta-mount to expropriation’. To date, Chapter 11 caseshave assumed a limited scope for environmentallaws’ constituting expropriation. Climate changelaws will most likely have much broader economiceffects than prior environmental legislation, and thescope of potential claims under NAFTA Chapter 11due to climate change laws and regulations could beorders of magnitude greater than those filed in thepast. The potential for such Chapter 11 litigationagainst climate change laws could slow the imple-mentation of measures designed to mitigate GHGemissions and adversely flows of trade and invest-ment in the region.
These measures would increase the efficiency of stateand regional climate change regulations. However,they are no substitute for a comprehensive nationalapproach to climate change in both countries. Inorder to ensure the best policy outcome, Obama andHarper should work toward a national cap-and-tradeor carbon tax bill.
References
Bianco, N. and F. Litz (2010), Reducing Greenhouse Gas Emissions inthe United States Using Existing Federal Authorities and State Action,Washington DC: World Resources Institute.
Centro Mario Molina (2008), Low Carbon Growth: A Potential Pathfor Mexico, Discussion Draft presented at Poznan Climate ChangeConference, Poznan, 1–12 December.
Department of Energy (2008), 20 Percent Wind Energy by 2030:Increasing Wind Energy’s Contribution to US Electricity Supply,http://www.osti.gov/bridge.
Energy Information Administration (2010), Crude Oil and TotalPetroleum Imports Top 15 Countries, http://www.eia.doe.gov.
Fickling, M. (2010), US and Canadian Climate Legislation by Stateand Province, Washington DC: Peterson Institute for InternationalEconomics.
Fickling, M. and J.J. Schott (2011), NAFTA and Climate Change.Washington DC: Peterson Institute for International Economics,forthcoming.
Hufbauer, G. and J. Kim (2009), U.S. Climate Legislation Impli-cations and Prospects: Challenges for Canada, Conference Board ofCanada, http://www.conferenceboard.ca/documents. aspx?did=3300.
Hufbauer, G. and J.J. Schott (2005), NAFTA Revisited, WashingtonDC: Peterson Institute for International Economics.
Kass, S.L. and J. McCarroll (2008), Environmental Enforcement andProtection under NAFTA, New York Law Journal,http://www.clm.com/publication.cfm/ID/191.
North American Development Bank (NADB, 2010), QuarterlyStatus Report, http://www.nadbank.org/pdfs/status_eng.pdf.
RGGI Emissions Leakage Multi-State Staff Working Group (RGGI,2007), Potential Emissions Leakage and the Regional Greenhouse GasInitiative (RGGI): Evaluating Market Dynamics, Monitoring Options,and Possible Mitigation Mechanisms,http://rggi.org/docs/il_report_final_3_14_07.pdf.
Richardson, N., A.G. Fraas and D. Burtraw (2010), Greenhouse Gas Regulation under the Clean Air Act: Structure, Effects, and Impli-cations of a Knowable Pathway, RFF Discussion Paper 10-23,http://www.rff.org.
Rowlands, I.H. (2009), “Renewable Electricity Politics acrossBorders”, in: Selin H. and S.D. VanDeveer (eds.), Changing Climatesin North American Politics: Institutions, Policymaking, and MultilevelGovernance, Cambridge MA: MIT Press, 181–198.
Schott, J.J. and M. Fickling (2009), Setting the NAFTA Agenda onClimate Change, Policy Brief 09-18, Washington DC: PetersonInstitute for International Economics.
Toman, M. et al. (2008), Unconventional Fossil-Based Fuels: Eco-nomic and Environmental Trade-Offs, RAND Corporation,http://www.rand.org/pubs/technical_reports/TR580/.
UNFCCC (2010a), Submission of Canada to the Copenhagen Accord,http://unfccc.int/files/meetings/application/pdf/canadacphaccord_app1.pdf.
UNFCCC (2010b), Mexico, National Mitigation Actions Submitted tothe Copenhagen Accord, 31 January 2010, http://unfccc.int/files/meet-ings/application/pdf/mexicocphaccord_app2.pdf.
Western Climate Initiative (WCI, 2008), Design Recommendations forthe WCI Regional Cap-and-Trade Program, http://www.westerncli-mateinitiative.org/component/remository/func-startdown/18/.
CESifo Forum 4/2010 52
Special
TURBULENT WATERS IN THE
EMU: TRANSLATION FROM
WIRTSCHAFTSWOCHE
KARL OTTO PÖHL*
In September 2010 the German business weekly,WirtschaftsWoche, interviewed Mr. Pöhl on old andnew conflicts in the European Monetary Union, thefuture of the euro and necessary structure reforms forthe ECB and the Bundesbank.
WirtschaftsWoche: Mr. Pöhl, the Bundesbank is caught
up in difficult times. How did you react to the recent
debate over Mr. Sarrazin, and what lessons should we
learn from it?
I do not want to comment on this issue in detail, but itwould surely make sense to change the Bundesbankstatute. One should give thought to reforming the proce-dures for appointments to the executive board. As it nowstands it is too federalist. The influence of the federalstates on appointments to the executive board has his-torical reasons that are no longer valid today. What tiesdoes the Bundesbank still have with the federal states?My suggestion would be that only the federal govern-ment make appointments to the executive board, and itshould include the Bundesbank in this decision making.
Another important issue is the European Monetary
Union. How do you evaluate the current debt crisis of
the euro countries?
I put no stock in the rescue programme for the euroarea. It gives the countries the wrong incentives andviolates the spirit and letter of the monetary union.We never wanted a transfer union – but now we arecaught up in one. The rescue programme is, after all,a traditional French position. Already in 1979 Francewanted a common European fund. ChancellorSchmidt supported President Giscard d’Estaing onthis matter. Then the wish was to stabilise theexchange rates in the European Monetary System
(EMS). The idea was that the member states could
borrow money from the fund with relative ease. Italy,
for example, would have been able to take out a loan
in deutschmarks and pay it back in lira. That would
have been a real inflation machine. The Bundesbank
was strictly opposed to this at the time, and a strong
confrontation with the federal government was the
result. Helmut Schmidt even threatened us with a
reform of the Bundesbank statute to deprive us of our
power. However, in the end we prevailed and the fund
did not come about. This time the French managed to
get their idea through, which could prove to be very
expensive for us. If countries such as Greece, Portugal
or Ireland do in fact make use of the rescue pro-
gramme, it will become prohibitively expensive.
Would there have been other options in your opinion?
Of course. Greece should never have been accepted
into the monetary union. But it was. Now it must be
possible to leave the union. Then, a haircut should
have been carried out (partial debt forgiveness by the
creditors), the Greeks would have devalued in order to
improve their competitiveness. But Mr. Sarkozy was
no doubt fearful of the French banks, which were
strongly committed in Greece.
Apart from the problem of technical feasibility – would-
n’t that have resulted in incredible turbulence?
That’s what many say, but I’m not so sure. There
might have been the risk of the monetary union col-
lapsing, but the risk was much smaller than most
believed. Greece is not really that important after all.
The core countries would have remained in the euro
area. I do not believe that there would have been
major deformations.
ECB President Jean-Claude Trichet has said again and
again that the Europeans are bound together by fate.
You were involved – as he is – in the process of Euro-
pean unification for many decades. Do Europeans share
a common fate?
We are indeed a community. But I do not see things
as dramatically as Mr. Trichet does. I have known* Former Bundesbank President.
CESifo Forum 4/201053
Special
him for decades, he is an honest man, and he would
never act against his convictions. It was certainly a
difficult situation at the beginning of May, when the
decision was made on the rescue programme. But I
don’t believe that Europe itself was threatened
because of Greece. The ECB unfortunately has
moved into turbulent waters by throwing major prin-
ciples overboard. It is now purchasing government
bonds from euro countries and is thus weakening its
independence.
Can this we undone?
In my opinion it is irreversible. They can’t ‘do this’ one
minute and ‘do that’ the next. The bond purchases
and the rescue programme are a breach of the
Maastricht Treaty by mutual agreement. Of course it
was an emergency, but now it has also become a
precedent, and that is the great danger.
Now the situation on the financial markets has sta-
bilised somewhat and the euro is also on the road to
recovery ...
… which is still no justification for throwing your
principles overboard.
What can the ECB do to restore its credibility?
One possibility would be to make Axel Weber its
president in autumn next year. Mr. Weber stands
for stability policies that the Bundesbank still holds
high. It would be even more important, however, to
change the ECB statutes. At the beginning of the
1990s when we drafted the statutes, I strongly
favoured the principle of ‘one country one vote’.
That surprised many at the time. Weighting the
votes of a country according to economic strength
or population size, for example, would not have
succeeded politically. But the times have now
changed. Then I would not have considered it pos-
sible that countries such as Malta and Cyprus
would be included in the euro. I thought it would be
limited to the six founding members. Now the mon-
etary union has became larger than we originally
believed. And therefore the principle of ‘one coun-
try one vote’ is no longer in keeping with the times.
It is not acceptable for the central banks of Malta
or Cyprus to have the same voting power in the
ECB as the Bundesbank. This waters down the
decisions of the European Central Bank. Voting
rights in the ECB should be changed and weighted
votes should be assigned according to the strength
of the countries. This would help the ECB manage
future crises more convincingly.
Will there still be the euro in ten years?
I believe so, yes. The euro is irreversible. Perhaps there
will be a different euro, perhaps limited to fewer coun-
tries. This I consider possible and it would also be
desirable. But it will not cease to exist.
There is always tension between monetary policy and
government policy, as was the case during German
reunification. At the time, as President of the
Bundesbank, you spoke out against an immediate
German monetary union. How do you see this now –
20 years later?
The main issue at the time was the D-mark/GDR-
mark exchange rate. The experts at the time agreed
that a 1:1 exchange rate would be false economi-
cally. The correct rate would have been perhaps
three eastern marks to one D-mark. But this was
unacceptable for Mr. Kohl and Mr. Waigel. They
wanted to push through the monetary union and
simply overruled me and the Bundesbank. This
was one of the reasons I resigned, since it was a
policy I couldn’t support. Today I realise that there
had been enormous pressure. The east German
population wanted to finally have the prosperity
they had been dreaming about for decades. Just
imagine that wages in the east, with an exchange
rate of 3:1, would have fallen by around two thirds
– there would have been a popular uprising. In this
respect political reality was stronger than econom-
ic logic.
Looking back on your career, you came a long way from
a convinced social democrat to a market liberal ...
I have always been a liberal. Karl Schiller was my
great model at the SPD, and he was in fact a liberal.
That was certainly possible in the SPD of the 1970s
in the social-liberal coalition. The liberals (FDP)
had Otto Graf Lambsdorff, who was a friend of
mine. I always worked closely with Mr. Lambsdorff
and Mr. Genscher, much more so than with some
SPD ministers.
The liberal wing of the SPD does not exist anymore, at
least not since Wolfgang Clement’s withdrawal ...
… which is why I also left the party in 2005. I could
no longer identify with the course of the SPD.
CESifo Forum 4/2010 54
Special
With the policies of Mr. Beck and now Mr. Gabriel
and Ms. Nahles, it will never win another election.
The SPD is out of touch, its programme is no
longer in accord with the times. It should be more
open economically; it should take a stand on glob-
alization and international competition. But it has-
n’t done that.
Let’s go back again to the 1980s. This was also the time
that the course was laid out for the euro in the context
of European integration ...
Correct. Helmut Schmidt and Giscard d’Estaing had
just set up the European Monetary System (EMS),
which called for fixed exchange rates between the
countries of the European Community. However, it
was a difficult birth. In the 1980s there were eleven
realignments in the EMS that is re-evaluations of the
exchange rates. In spite of these setbacks, integration
made further progress. The Single Market and the
Schengen Agreement were realised. This was backed
by the insight and political will that Europe can be
only be advanced as a joint effort. It’s like riding a
bike – you have to keep on pedalling because if you
stop you fall over.
At the time there were also financial crises ...
I remember there always being crises, since my
involvement in government. My first day in the
Chancellor’s office in 1970 began with Karl Schiller
freeing the Bundesbank of its intervention obligation
vis-à-vis the dollar. I almost had a stroke. I had to
explain to Willy Brandt why that was necessary. Then
came the European ‘currency snake’, the forerunner
of EMS. It failed because there was always someone
who couldn’t maintain the exchange rate parity. Then
came the EWS with its eleven revaluations. And in
1992, when Mr. Soros was speculating against the
pound, the British unfortunately had to leave. There
was always some crisis.
Is the current financial crisis fundamentally different?
This crisis has a new quality – only because of its
dimensions. This is due to increasing globalisation.
The states depend so strongly on each other that
undesirable developments are expanded exponential-
ly. Fluctuations become greater. This crisis also accel-
erates the shifts in the international balance of power.
China and India have increased their weight in world
politics. The United States with its gigantic mountain
of debt has been weakened.
You studied economics in the 1950s. Then John
Maynard Keynes was the guru of the economists. Are
we experiencing his comeback in the financial and eco-
nomic crisis?
I don’t think that the return to Keynesianism in eco-nomic policy will be permanent. At least not to thesimple state dirigisme of his time. Just image, in myfinal examination at the University of Göttingen in1955 I defended the thesis that the amount of indebt-edness is limited by tax revenue. I was given the grade‘good’ for this (laughs). Today of course I would notput it that way – and John Maynard Keynes probablywouldn’t either.
CESifo Forum 4/201055
Special
GREENSPAN, DODD-FRANK
AND STOCHASTIC OPTIMAL
CONTROL
JEROME L. STEIN*
Dodd-Frank financial reform bill
The Dodd-Frank (D-F) bill establishes the FinancialServices Oversight Council. The bill authorizes theFederal Reserve Board to act as agent for the Councilto monitor the financial services marketplace to iden-tify potential threats to the stability of the US finan-cial system and to identify global trends and develop-ments that could pose systemic risks to the stability ofthe US economy and to other economies.
Is the Fed capable of fulfilling these requirements?Shojai and Feiger (2010), in their article Economists’
Hubris – The Case for Risk Management – write thatthe tools that are currently at the disposal of theworld’s major global financial institutions are notadequate to help them prevent such crises in thefuture and that the current structure of these institu-tions makes it literally impossible to avoid the kind offailures that we have witnessed. I evaluate whatGreenspan has learned and develop the StochasticOptimal Control approach that should be used toimplement the D-F bill.
Greenspan’s retrospective
Greenspan’s paper (2010) presents his retrospectiveview of the crisis. His theme has several parts. First,the decline and convergence of world real long-terminterest rates – not Federal Reserve monetary policy –led to significant housing price appreciation, a hous-ing price bubble. Second, this bubble was leveraged bydebt. There was a heavy securitization of subprimemortgages. In the years leading to the current crisis,financial intermediation tried to function on too thin
layer of capital – high leverage – owing to a misread-
ing of the degree of risk embodied in ever more com-
plex financial products and markets. Third, when the
bubble unraveled, the leveraging set off a series of
defaults. Fourth, the breakdown of the bubble was
unpredictable and inevitable, given the ‘excessive’
leverage – or unduly low capital – of the financial
intermediaries.
Prior to the subprime crisis of 2007, there was a false
sense of safety in financial markets. Alan Greenspan
said in 2004 that “the surge in mortgage re-financings
likely improved rather than worsened the financial
condition of the average homeowner”. Moreover
“overall, the household sector seems to be in good
shape, and much of the apparent increase in the
household sector’s debt ratios in the past decade
reflects factors that do not suggest increasing house-
hold financial stress” (Greenspan 2004a and 2004b).
The market and the Fed did not consider these mort-
gages to be very risky. By 2007 a measure of risk, the
yield spread (CCC bonds – 10 year US Treasury), fell
to a record low.
When the crisis came in 2008, Greenspan said: “those
of us who have looked to the self-interest of lending
institutions to protect stockholders’ equity, myself
included, are in a state of disbelief”. The lesson for the
future that he has learned is that it is imperative that
there must be an increase in regulatory capital and liq-
uidity requirements by banks (Greenspan 2010).
My basic questions are: what is an optimal leverage or
capital requirement that balances the expected growth
against risk? What are theoretically founded – not ad
hoc empirical – early warning signals of a crisis? I
explain why the application of stochastic optimal con-
trol (SOC) is the effective approach to determine the
optimal degree of leverage, the optimum and exces-
sive risk and the probability of a debt crisis.
I show that the theoretically derived early warning sig-
nal (EWS) of a crisis is the excess debt ratio, equal to
the difference between the actual and optimal ratio.
The excess debt of households starting from
2004–2005 indicated that a housing crisis was most* Brown University, Providence. I thank Peter Clark and ShlomoMaital for excellent suggestions.
likely. This SOC analysis should be used by those
charged with surveillance of financial markets. It is
hoped that the Fed will not be like the ancien régime:
“Ils n’ont rien appris, ni rien oublié”.
Stochastic optimal control (SOC)/dynamic risk management
The financial crisis was precipitated by the mortgage
crisis and spread through the financial sector due to
high leverage. I focus upon the housing sector, which
was has been at the origin of the crisis. At the begin-
ning of the financial chain are the mortgagors/debtors
who borrow from financial intermediaries – banks,
hedge funds, government sponsored enterprises. The
latter are creditors of the mortgagors, but ultimately
are debtors to banks or to institutional investors at
the other end. For example, the Federal National
Mortgage Association (FNMA) borrows in the world
bond market and uses the funds to purchase or later
resell packages of mortgages. If the mortgagors fail to
meet their debt payments, the effects are felt all along
the line. The stability of the financial intermediaries
and the value of the traded derivatives – CDO and
CDS – ultimately depend upon the ability of the
mortgagors to service their debts. When the mort-
gagors default, the whole leveraged financial structure
collapses.
SOC is dynamic optimization where key variables are
stochastic. A sketch of the SOC approach will facili-
tate understanding the analysis below. Technical
mathematical details are in my recent papers (Stein
2010 and 2011). The criterion/object is to maximize
the expected logarithm of household net worth at a
future date. This is a risk-averse strategy because the
logarithm is a concave function. Declines in net worth
are weighted more heavily than increases in net worth.
In fact very severe penalties are placed upon bank-
ruptcy – a zero net worth. This criterion is the growth
variable that will optimally balance expected growth
and risk.
The growth of net worth is affected by leverage. An
increase in debt to finance the purchase of assets
increases net worth by the return on investment, but
decreases the growth of net worth by the associated
interest payments. The return on investment has two
components. The first is the productivity of assets and
the second is the capital gain on the assets. An increase
in leverage will increase expected growth if the return
on investment exceeds the interest rate. The produc-
tivity of assets is observed, but the future capital gain
and the interest rates are unknown when the invest-
ment decision is made. The true stochastic process is
unknown. One must specify the stochastic process on
the capital gain and interest rate if one wants to select
the optimal leverage – to maximize the expected loga-
rithm of future net worth.
The SOC approach derives an optimal debt ratio con-
ditional upon the stochastic processes. Alternative
stochastic processes imply different optimal debt
ratios. My standard of optimality is based upon sus-
tainable stochastic processes concerning the capital
gain and interest rate. By contrast, the market opti-
mized on the basis of unsustainable stochastic
processes, which led to the bubble and its subsequent
collapse.
A sustainable stochastic process is as follows. Call this
the Prototype Model. Reasonable variations imply
similar qualitative but not quantitative results. The
capital gain is the sum of two terms: a constant drift
and a Brownian motion term. The interest rate has a
similar structure: a constant drift plus a Brownian
motion term. The capital gain and interest rate are
negatively correlated. The drift of the capital gain is
constrained not to exceed the drift of the interest rate,
to preclude the ‘free lunch’ described below.
Given the stochastic process, an optimal leverage or
capital requirement is derived as follows. The expect-
ed growth of net worth is a concave function of the
leverage. It is maximal when the optimal leverage is
chosen. As the leverage exceeds the derived optimal,
the expected growth declines and the variance/risk
rises. If the debt ratio is less than the optimal, expect-
ed growth is unduly sacrificed to reduce risk. Leverage
is equal to one plus the debt ratio; and the capital
requirement is the inverse of the leverage. I focus
upon the debt ratio, and the other ratios follow.
The main theoretical results are as follows. (1) The
optimal debt ratio is not a number, but a function. It
is proportional to: the drift of the capital gain less the
drift of the rate of interest plus the current productiv-
ity of capital less a risk premium. The factor of pro-
portionality is the reciprocal of risk elements.
Therefore the optimal debt ratio or capital require-
ment will vary among sectors and over time. One size
does not fit all. (2) Define the excess debt as the actu-
al debt ratio less the optimal ratio. For a sufficiently
high excess debt, the expected growth is zero or nega-
tive and the variance is high. The probability of a
CESifo Forum 4/2010 56
Special
CESifo Forum 4/201057
Special
decline in net worth or a debt crisis is directly related
to the excess debt ratio.
The market selected a debt ratio based upon an illu-
sion of a ‘free lunch’ – an unsustainable stochastic
process. The market estimated the drift of capital
gains on the basis of recent price changes. The recent
capital gains exceeded the interest rate so that the
mortgagors thought that they were getting a free
lunch. The rises in housing prices and in owner equi-
ty induced a demand for mortgages by banks and
funds. The mortgagors borrowed at an interest rate
below the capital gain. In about 45-55 percent of the
cases, the purpose of the subprime mortgage taken
out in 2006 was to extract cash by refinancing an
existing mortgage loan into a larger mortgage loan.
They expected to repay the loan plus interest from the
higher value of the home, due to the capital gain. The
quality of loans declined. The share of loans with full
documentation substantially decreased from 69 per-
cent in 2001 to 45 percent in 2006 (see Demyanyk and
Van Hemert 2007). The ratio of debt/income rose
drastically. The only way to service or refinance the
debt was for the capital gain to exceed the interest
rate. This is an unsustainable situation since it implies
that there is a ‘free lunch’ or that the present value of
the asset diverges to infinity.
Is the Fed capable of implementing the D-F bill? The
Fed, IMF, Treasury and the ‘Quants’/market lacked
the appropriate tools of analysis to answer the follow-
ing questions: what is an optimal leverage or capital
requirement that balances the expected growth against
risk? On the basis of the SOC analysis, I derive the
Early Warning Signals of the crisis. The excess debt
starting from 2004–2005 indicated that a crisis was
most likely. This SOC analysis should be used by those
charged with surveillance of financial markets.
The basic equations: Prototype Model
The formal structure of the Prototype Model is the
subject of this part. The reader is referred to Stein
(2010 and 2011) and to Fleming and Stein (2004) for
the mathematical details. The empirical implications
for an EWS are in the later sections.
Criterion function
As my criterion of performance, I consider maximiz-
ing the expected logarithm of net worth of the mort-
gagors. I focus upon the net worth of the mortgagors
for two reasons. First, the entire structure of the
derivatives rested upon the ability of the mortgagors
to repay their debts. Hence I ask what the optimal
debt ratio of the mortgagors is. Second, I derive an
Early Warning Signal (EWS) that a bubble, the hous-
ing price bubble, is likely to collapse.
Let W(X,T) be the expected logarithm of net worth
X(T) at time T relative to its initial value X(0). The
stochastic optimal control problem is to select debt
ratios f(t) = L(t)/X(t) during the period (0,T) that
will maximize W(T) in equation (1). The maximum
value is W*(X,T). The optimal debt/net worth
ratio f*(t) plus one is the optimal leverage, and will
vary over time. The solution of the stochastic opti-
mal control/dynamic risk management problem
tells us what an optimal and what an ‘excessive’
leverage is.
(1) W*(X,T) = maxf E ln [X(T)/X(0)],
f = L/X = debt/net worth; leverage = assets/net
worth = 1 + f
The logarithm ln(X) is a concave function of X(T). As
the expectation E[X(T)] goes to zero, the logarithm ln
[E(X(T)] goes to minus infinity. Therefore the expec-
tation E[ln X(T)] would go to minus infinity as
E[X(T)] goes to zero. Low values of net worth close to
zero may not be likely, but they have large negative
utility weights. Hence the criterion function reflects
strong risk aversion. Bankruptcy X = 0 is severely
penalized.
Dynamics of net worth
The mortgagors have a net worth X(t) equal to the
value of assets A(t) less debt L(t), see equation (2).
The value of assets A(t) = P(t)Q(t) is the product of a
deterministic physical quantity Q(t), for example an
index of the ‘quantity’ of housing, times the stochas-
tic price P(t) of the capital asset which is the housing
price index.
(2) X(t) = A(t) – L(t) = P(t)Q(t) – L(t), while
A(t) = P(t)Q(t).
The control variable is the debt ratio. The next steps
are to explain the stochastic differential equation for
net worth, relate it to the debt ratio, and specify what
are the sources and characteristics of the risk and
uncertainty.
In view of equations (1) and (2), focus upon the
change in net worth dX(t) of the mortgagors. It is
the equal to the change in the value of assets dA(t)
less the change in debt dL(t). The change in the
value of assets dA(t) = d(P(t)Q(t)) shown in equa-
tion (3) has two components. The first is the change
due to the change in price of capital asset, which is
the capital gain or loss term, A(t)(dP(t)/P(t)). The
second is investment in housing I(t) = P(t) dQ(t),
the change in the quantity times the price.
(3) dA(t) = d(P(t)Q(t)) = Q(t)dP(t) + P(t)dQ(t) =
A(t)dP(t)/P(t) + I(t)
The change in debt dL(t), equation (4), is the sum
of expenditures less income. Expenditures are the
debt service i(t)L(t) at interest rate i(t), plus invest-
ment I(t) = P(t) dQ(t) plus C(t) the sum of con-
sumption, dividends and distributed profits.
Income Y(t) = ß(t)A(t) is the product of assets A(t)
times its productivity. Variable ß(t) corresponds to
the imputed rental income from housing divided by
the value of housing. This equation can be
expressed as (4a) where saving S(t) = ß(t)A(t) – C(t)
and investment is I(t). Thus the change in debt is
the sum of interest payments plus investment less
saving.
(4) dL(t) = i(t)L(t) + P(t)dQ(t) + C(t) – ß(t)A(t).
(4a) dL(t) = i(t)L(t) + I(t) – S(t).
Combining these effects, the change in net worth
dX(t) = dA(t) – dL(t) can be shown
(5) dX(t) = dA(t) – dL(t) = A(t)[dP(t)/ P(t) + ß(t) dt]
– i(t)L(t) – C(t) dt.
Since net worth is the value of assets less debt, equa-
tion (6) describes the dynamics of net worth equation
(5) in terms of the ratio f(t) = L(t)/X(t) of debt/ net
worth and an arbitrary consumption ratio c(t) =
C(t)/X(t) ≥ 0. Since leverage A(t)/X(t) = (1+f(t)), the
control variable could be either f(t) the debt ratio or
the leverage.
(6) dX(t) = X(t) {(1 + f(t)) [dP(t)/P(t) + ß(t) dt] –
i(t) f(t) – c(t) dt}.
The mortgagors borrow at interest rate i(t) and bene-
fit from the capital gain dP(t)/P(t). Both variables are
stochastic/unpredictable. What is the optimum debt
ratio, leverage or capital requirement?
The optimization of equation (1) subject to equation
(6) depends upon the stochastic processes underlying
the capital gain dP(t)/P(t), productivity of capital ß(t)
and interest rate i(t) variables. The productivity of
capital ß(t) is deterministic and observable but
changes over time. However the change in price dP(t)
from t to t+dt and future interest rates are unpre-
dictable, given all the information through present
time t. The derived optimal debt ratio, leverage or
capital requirement will depend upon the specifica-
tion of the stochastic processes of the capital gain and
interest rate.
Optimization in the Prototype Model
The Prototype Model that I use for optimization
describes the stochastic process of the capital gain as
equation (7) and the interest rate as equation (8). The
capital gain dP(t)/P(t) has a constant drift or mean
π dt and a diffusion or stochastic term σpdwp. The
expectation of the stochastic term is zero and its vari-
ance is σp2dt. Similarly the interest rate has a mean or
expectation of i dt and a variance of σi2dt. The corre-
lation between the capital gain and interest rate is
E(dwpdwi) = ρ dt, 1 ≥ ρ ≥ –1.
(7) dP(t)/P(t) = π dt + σp dwp.
(8) i(t) = i dt + σidwi
E dwp = E dwi = 0, E(dwi2) = dt, E (dwp2) = dt,
E(dwidwp) = ρ dt.
The maximization of expected net worth, equation
(1), subject to the stochastic processes, equations
(6)–(8), implies equation (9) for f*(t) the optimal ratio
of debt/net worth. Since leverage is equal to one plus
the debt ratio and capital requirement is the recipro-
cal of leverage, equation (9) is the key theoretical
result:
(9) f*(t) =[[(π + ß(t) – i) – (σp2 – ρσiσp)]/[σp2 + σi2 –
2 ρσiσp] ≥ 0.
The economic meaning and implications of f*(t)
are explained in the next part. An empirically use-
ful upper bound on the optimal debt ratio in the pro-
totype model f**(t) occurs when the two distur-
bances are independent and the drift of the capital
gain is equal to that of the interest rate, equations
(9a) and (9b):
(9a) f*(t) < [(ß(t) – σp2]/[σp2 + σi2] ≥ 0.
CESifo Forum 4/2010 58
Special
CESifo Forum 4/201059
Special
(9b) f**(t) = [(ß(t) – σp2]/[σp2 + σi2]
The ‘fundamentals’ that determine the optimal debtratio f*, leverage (1+f*) or capital requirements1/(1+f*) are in the right hand side of equation (9).
Economic implications of the optimum debtratio/leverage/capital requirements in the PrototypeModel
There are several important implications of equation(9). First, the optimum debt ratio f*(t) is proportion-al to the expected return (π + ß(t) – i) less a risk pre-mium (σp2 – ρσiσp), where the factor of proportional-ity is 1/[σp2 + σi2 – 2 ρσiσp], the risk elements. Optimalleverage and capital requirements follow. Second, debtwill only be optimal if the expected return exceeds therisk premium. Third, define excess debt Ψ(t) = f(t) –f*(t) as the difference between the actual debt ratio f(t)and the optimal f*(t). As the debt ratio exceeds theoptimum f*(t) there is an ‘excess debt’ and the expect-ed growth declines. For sufficiently high debt ratio f-max, the expected growth is zero. A warning signal
that too much risk has been undertaken is that theexcess debt Ψ(t) = f(t) – f*(t) is large. Alternatively,leverage is excessive when the debt ratio exceeds f*(t).The probability of a crash increases with the excessdebt, and is very likely when f(t) > f-max. The capitalrequirement A/X = 1/[1+f(t)] is optimal when f(t) =
f*(t) and is too low for debt ratios above f*(t). This isgeneral formulation that can be applied to any sector.Equations (9) or (10) imply the optimal capitalrequirement X*(t)/A(t).
Application to housing sector: estimates of excess debtas an early warning signal of a crisis
The financial crisis was precipitated by the mortgagecrisis. First, from 1995–2005 the decline in the 30-yearfixed rate mortgage interest rates (Figure 1) led to cap-ital gains CAPGAIN (Figure 2). Second, the mort-gagors and the financial intermediaries deluded them-selves in thinking that the mean of the capital gain –based upon the recent price experience – could con-tinue to exceed the mean of the interest rate. Hencethe market thought that the ‘optimum’ debt ratio,based upon π – i > 0, exceeded f**(t) in equation (9b).They thought that the ‘free lunch’ could continue.Third, a whole structure of financial derivatives wasbased upon the ultimate debtors – the mortgagors.Fourth, the financial intermediaries, whose assets andliabilities were based upon the value of derivatives,were very highly leveraged. Percentage changes in thevalues of their net worth were large multiples of per-centage changes in asset values. Fifth, the financialintermediaries were closely linked – the assets of onegroup were liabilities of another. The whole structureof derivatives rested upon the mortgagors being able
Figure 1THIRTY-YEAR CONVENTIONAL MORTGAGE RATE IN THE US
Source: Board of Governors of the Federal Reserve System.
to service their debts. Sixth, the collapse occurredwhen the capital gain fell below the rate of interest:the ‘free lunch’ was over. Defaults and bankruptciesoccurred. A cascade was precipitated by the mortgagedefaults.
The application of the Prototype Model/SOC analy-sis is done in several steps. First, on the basis of theanalysis, I derive estimates of the excess debt Ψ(t) = f(t) – f*(t) that lowered the expected returnand raised risk. Early warning signals (EWS) arethereby derived. An Early Warning Signal of a debtcrisis is a series of excessive debts Ψ(t) = f(t) – f*(t)> 0. When the debt ratio f(t) exceeds f-max, theexpected growth is negative and the risk is high. Thenext question is: what are the appropriate measuresof the actual and the optimal debt ratio to evaluateexcess debt Ψ(t)?
The debt ratio that I use in empirical work is the ratioof household debt as a percent of disposable income,since I do not have estimates of household net worth.In order to make alterative measures of the debt ratioand key economic variables comparable, I use nor-malized variables where the normalization (N) of avariable Z(t) called N(Z) = [Z(t) – mean Z]/standarddeviation. The mean of N(Z) is zero and its standarddeviation is unity. The normalized debt ratio is equa-tion (10) and is graphed in Figure 3:
(10) DEBTRATIO = N[f(t)] = [debt/disposable income – mean]/standard deviation.
One cannot be sure what the correct stochasticprocesses on the capital gain and interest rate are.Therefore there is ambiguity concerning the exact
value of the optimal debt ratiof*(t). For this reason I work withf**(t) which is an upper bound ofthe optimum debt ratio basedupon equation (9b). A justifica-tion is as follows. In the case ofthe housing sector, historicallythe mean capital gain 1980–2007was π = 5.4 percent, with a stan-dard deviation of 2.9 percent.The 30-year conventional mort-gage rate of interest from 1998 to2007 ranged between 7.5 percentand 6 percent. If we assume thatthe difference (π – i) between themean interest rate and the meancapital gain is not significant, and
the correlation ρ = 0, then an upper bound of the opti-mal debt ratio f** for the housing sector is (9b). Thisformulation is qualitatively, but not quantitatively,consistent with alternative theoretical measures of theoptimum debt ratio implied by alternative stochasticprocesses.
(9b) f** = L*/X = [ß(t) – σp2]/[σp2 + σi2] ≥ 0.
The term [ß(t) – σp2]/[σp2 + σi2] represents the ‘funda-
mental’ determinants of the optimal debt ratio. Wemust estimate ß(t), the productivity of assets. The pro-ductivity of housing assets is the (implicit net rentalincome/value of the home) plus a convenience yield inowning one’s home. Assume that the convenienceyield in owning a home has been relatively constant.The productivity of assets ß(t) is rental income/valueassets = Y(t)/A(t) = Y(t)/Q(t)P(t), where Y(t) is rentalincome, P(t) is an index of housing prices and Q(t) isan index of the physical quantity of housing.Therefore ß(t) is proportional to a ratio of rentalincome to an index of housing prices.
(11) ß(t) ~ Y(t)/P(t)
An empirical proxy for f**(t) an upper bound ofthe optimal debt ratio is RENTPRICE defined inequation (12). Since the units of numerator anddenominator differ, it makes sense to use normal-ized variables to estimate ß(t) the productivity ofassets. The term [(ß(t) – ß)] is the deviation of thecurrent return on assets from its mean value ß overthe entire period.
In Figure 3 and equation (12) variable RENTPRICEis the normalized return, measured in units of stan-
CESifo Forum 4/2010 60
Special
0
2
4
6
8
10
12
14
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
Source: Office of Federal Housing Enterprise Oversight (OFHEO).
%
APPRECIATION OF SINGLE-FAMILY HOUSING PRICES CAPGAIN4Q appreciation of US housing prices (HPI)
Figure 2
CESifo Forum 4/201061
Special
dard deviation from the mean ß. It is equal to the ratio of (rental income/index of housing prices –mean)/standard deviation.
(12) N(f**(t)) = [Y(t)/P(t) – mean]/st. dev. ~ [(ß(t) – ß)] / σ(ß) = RENTPRICE
Variable N(f**(t)) in equation (12) is proportional toan upper bound of the optimal debt ratio in equation(9b). Both the actual (DEBTRATIO) and optimal(RENTPRICE) are graphed in normalized form inFigure 3.
The next question is how to estimate the excess debtΨ(t). I estimate excess debt Ψ(t) = (f(t) – f**(t)) byusing the difference between two normalized variablesN(f(t)) – N(f*(t)), see equation (13). This difference ismeasured in standard deviations.
(13) Excess Debt Ψ(t) = N[f(t)] – N[f*(t)] = DEBTRATIO – RENTPRICE.
Excess Debt Ψ(t) corresponds to the difference betweenthe two curves DEBTRATIO and RENTPRICE inFigure 3. The probability of a decline in net worth ispositively related to Ψ(t) the excess debt because, as theexcess debt rises, the expected growth declines.
In the most general way, Figure 3 should be viewed asfollows. Assume that over the entire period 1980–2007the debt ratio was not excessive. Both variables arenormalized to make them comparable. When the
DEBTRATIO is above (below) its mean, the RENT-PRICE should be above (below) its mean. When thedebt ratio rose significantly above a proxy for anupper bound of the optimal debt ratio, the RENT-PRICE declined below the mean. From 1996 and by2007 it was 1.5 standard deviations below the mean.The actual debt ratio DEBTRATIO grew steadilyabove the mean from 1998, and by 2007 was 2 stan-dard deviations above the mean. Thus the excess debtgrew to 3 standard deviations above the meanfrom1998 to 2007.
The normalized actual debt ratio got out of line withthe normalized proxy for an upper bound of the opti-mal debt ratio. The latter reflects the ‘fundamentals’.The sequence of excess debts Ψ(t) is a clear measureof a bubble. The actual debt was induced by capitalgains in excess of the interest rate. The debt couldonly be serviced from capital gains. This situation isunsustainable. When the capital gains fell below theinterest rate, the debts could not be serviced fromincome. A crisis was inevitable.
The advantages of using excess debt Ψ(t) in Figure 3as an Early Warning Signal compared to just the ratioof housing price/disposable income are that Ψ(t)focuses upon the fundamental determinants of theoptimal debt ratio as well as upon the actual ratio.The probability of declines in net worth, the inabilityof the mortgagors to service their debts and the finan-cial collapse and a crisis due to leverage, are directlyrelated to the excess debt.
Conclusions
The Jackson Hole Consensus(JHC) has been the prevailingregulatory approach taken by theFed. It is based upon three prin-ciples. Central banks (i) shouldnot target asset prices; (ii) shouldnot try to prick an asset pricebubble; and (iii) should follow a‘mopping up’ strategy after thebubble bursts by injecting enoughliquidity to avoid serious effectsupon the real economy. A justifi-cation for this policy was seen inthe period 2000–2002 with thecollapse of the dot.com bubble.
Issing (2010) objects to the JHCbecause it constitutes an asym-
Notes: Excess debt Ψ(t) = N[f(t)] – N[f*(t)]; N[f(t)] = DEBTRATIO = (household debt as per-cent of disposable income – mean)/standard deviation; N[f**t)] = RENTPRICE = (rentalincome/housing price index – mean)/standard deviation.Sources: FRED, Federal Reserve Bank St. Louis; Office of Federal Housing EnterpriseOversight (OFHEO).
10
11
12
13
14
15
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008
DEBTRATIO
RENTPRICE
EARLY WARNING SIGNALS
3
2
1
0
-1
-2
Figure 3
metric approach. When asset prices rise without infla-tionary effects measured by the CPI, this is deemedirrelevant for monetary policy. But when the bubblebursts, central banks must come to the rescue. This, heargues, produces a moral hazard. He wrote: “did wereally need a crisis that brought the world to the brinkof a financial meltdown to learn that the philosophywhich was at the time seen as state of the art was infact dangerously flawed? We must conduct a thoroughdiscussion as to appropriate strategy of central bankswith respect to asset prices”.
Greenspan argues that the crisis was unpredictable. Itis ironic that the Fed claims that it can use the FederalFunds rate to target inflation or to stabilize the econ-omy but asserts that the financial crisis was unpre-dictable and inevitable. On the other hand, on thebasis of the SOC analysis in this paper, the sequenceexcess debts Ψ(t) from 2003 in Figure 3 was an earlywarning signal of a crisis.
The Dodd-Frank bill authorizes the Fed to performmarket surveillance. I explain why the application ofstochastic optimal control (SOC) is an effectiveapproach to determine the optimal degree of leverage,the optimum and excessive risk, the optimumrisk/expected return trade-off and EWS of the proba-bility of a debt crisis. A similar analysis was applied tothe Asian crisis in Stein (2006). This SOC analysisshould be used by those charged with surveillance offinancial markets.
References
Demyanyk, Y. and O. Van Hemert (2007), Understanding theSubprime Mortgage Crisis, Federal Reserve Bank of St. Louis Super-visory Policy Analysis Working Paper 2007–05.
Fleming, W.H. and J.L. Stein (2004), “Stochastic Optimal Control,International Finance and Debt”, Journal of Banking & Finance 28,979–996.
Greenspan, A. (2004a), Understanding Household Debt Obligations,Remarks at Credit Union National Association, Washington DC,23 February.
Greenspan, A. (2004b), The Mortgage Market and Consumer Debt,Remarks at America’s Community Bankers Annual Convention,Washington DC, 19 October.
Greenspan, A. (2008), Testimony on Sources of Financial Crisis,House Committee on Government Oversight and Reform,Washington DC, 23 October.
Greenspan, A. (2010), “The Crisis”, Brookings Papers, second draft.
Issing, O. (2010), “Some Lessons from the Financial Crisis”, Inter-national Finance 12, 431–444.
Shojai, S. and G. Feiger (2010), “Economists’ Hubris – The Case ofRisk management”, Journal Financial Transformation 28, 25–35.
Stein, J.L. (2006), Stochastic Optimal Control, International Financeand Debt Crises, New York: Oxford University Press.
Stein, J.L (2010), “Greenspan’s Retrospective of Financial Crisis andStochastic Optimal Control”, European Financial Management 16,858–871.
Stein, J.L. (2011) “The Crisis, Fed, Quants and Stochastic OptimalControl”, Journal of Economic Modelling, forthcoming.
CESifo Forum 4/2010 62
Special
CESifo Forum 4/201063
Special
ESTIMATION OF PRODUCTION
COSTS FOR ENERGY RESOURCES
HANS-DIETER KARL*
Introduction
The use of fossil fuels largely depends on the costs ofmaking these fuels available. Among these costs, theproduction expenses, which are influenced by numer-ous factors, are of prime importance. The aim of thisstudy is to determine the unit costs of productionfrom the currently used deposits and to estimate theunit costs of production from deposits that will betapped in future. An exact definition of productioncosts is exceedingly difficult to make. Therefore, theaverage production costs for mineral oil, natural gasand black coal determined in this study must be seenas approximations of reality. For the determination ofthe production costs of currently used deposits andfor the estimates of the costs of future deposits,recourse is made primarily to existing studies and esti-mates in the literature.
Energy production costs of great importance
In the present and future use of fossil energy sources,the costs of making them available play a key role,among which the production costs, which are influ-enced by numerous determinants, are crucial. In addi-tion to the geological, technical and economic factors,the political conditions of extraction are also ofmajor importance. This study seeks to determine theunit costs of production at currently used depositsand to estimate the unit costs of production fromdeposits that will be tapped in future. An exact defin-ition of production costs is very difficult for severalreasons. Since the production costs are an importantcompetitive factor for individual enterprises, they areusually not made public. In addition the costs arehighly dependent on the factors of the individualdeposits such as their location and size, and can there-
fore differ greatly. Finally, during extraction of the
raw materials additional, unanticipated expenditures
such as strikes, storms or subsequently imposed regu-
lations may occur so that the actual production costs
can often only be precisely determined after the fact.
In the following the production costs for the domi-
nant fossil energy sources – mineral oil, natural gas
and black coal – will be identified. These costs include
the costs for the extraction of the energy resources, i.e.
for the operation of the extraction facilities (especial-
ly wage and intermediate material costs) and the cap-
ital costs allocated to the service life of the facility,
which result from the expenditures for the exploration
and the construction of the facilities for the extraction
of the energy sources from the site of the deposits.
Included in these capital costs are also the expendi-
tures that accrue in the course of the depletion of the
raw materials in order to counter the gradual decline
in the production output and to be able to keep
extraction at a high level. This applies especially to oil
and gas deposits, for example, by the application of
secondary and tertiary procedures to increase the
degree of exploitation.
Not included are the so-called user costs, i.e. the costs
from the loss in value of the remainder of the in-situ
resources that arise as a result of the extraction of the
resource itself, as well as the royalties that precisely
reflect these users costs if the extracted resources are
not the sole property of the extracting company.
For the determination of the production costs of the
currently used deposits, recourse is made to several
currently available studies and estimates in the litera-
ture, where the production costs are given in the form
of unit costs. The information available in the litera-
ture for various years is converted uniformly to a price
basis of 2009. The estimation of the unit costs of
deposits exploited in the future is also carried out by
means of the information in the literature and on the
basis of forecasts of the requirements for the individ-
ual energy sources in the coming decades as well as by
using estimates of the probable development of
investments for accessing new deposits or for the con-
tinuation of production.* Ifo Institute for Economic Research.
Mineral oil: the dominant energy source
Mineral oil continues to be the most important
energy source worldwide. It clearly stands in first
place in terms of both consumption and traded vol-
umes. The high trading volume of mineral oil is
attributable mainly to the fact that the worldwide
concentrations of consumption and the areas with
the highest crude oil production, which frequently
have a high export potential, are at a comparably
far distance from each other. Furthermore, the
most important exporting countries are also those
with the most extensive conventional oil reserves.
At the end of 2008 the worldwide reserves amount-
ed to ca. 182 billion toe (tonnes of oil equivalent).
With an annual consumption of currently ca.
3.9 billion toe, a static range of 46 years of oil
reserves can be calculated. Of total reserves,
101 billion toe are concentrated in the Middle East,
and the annual consumption of these countries in
2008 amounted to 0.3 billion toe (ExxonMobil
2009). The leading position of mineral oil is based
on its specific advantages, especially its broad
application spectrum as well as its favourable trans-
portation and storage. Because of its overriding
importance, mineral oil is a strategic raw material.
Most analyses and forecasts of developments on
the world energy markets conclude that demand for
mineral oil will continue to increase also in the
coming years. However, smaller growth than only a
few years ago is expected, for example by the
International Energy Agency (IEA), because of the
increasing difficulties in accessing deposits and the
associated high costs.
The great number of analyses and forecasts are
understandable in light of the extraordinary impor-
tance of mineral oil. However, the great majority of
studies deal with questions of available quantities,
of the technical possibilities of extraction, price
developments and the political implications of
greater oil consumption when faced with the insecu-
rities of the supply. Studies of the costs of explo-
ration, accessing and extracting crude oil are con-
siderably less numerous. There are objective reasons
for this, especially the geological differences and the
size of the individual deposits as well as the differ-
ent technical efforts for oil extraction. In addition,
companies often treat their data on the costs of oil
production confidentially. Under these circum-
stances, the published numbers on production costs
are usually estimates of the respective authors.
Nevertheless, these estimates are probably fairly
accurate. Because of the considerable differencesbetween the oil fields, however, the production costscover a fairly wide range. This is also a result of thestrong increase in the price of oil in recent years thatmade it possible to extract oil from sites with com-paratively high extraction costs. In the following,the present and expected future production costs arepresented as a whole as well as their two compo-nents: capital costs and operating costs. The focuswill be particularly on the presentation of averagevalues.
A recent study was conducted by Deutsche Bank(Deutsche Bank 2009), containing estimates of cur-rent production costs (OPEX) for 90 percent of worldoil production without capital costs, royalties andexploration costs for 2009. For the large oil fields,which account for ca. 70 percent of worldwide pro-duction, these average costs stand at 6.20 US dol-lars/bbl (bbl = barrel; 1 bbl = 159 litres). The range ofthese costs extends from 1 US dollar/bbl in theArabian Gulf states to about 26 US dollars/bbl crudein the Canadian oil sands, the latter not being ofmuch importance in current production volume. Thedirect production costs of the smaller fields – about asixth of worldwide production – amount to 8.30 USdollars/bbl, on average. The production costs for allthe fields examined in the study, which determine thedevelopment of the worldwide oil market, average6.60 US dollars/bbl.1
For 2007, the Deutsche Bank study shows in additionthe total costs of oil production (technical costs).These costs are derived from information – also with-out royalties or user costs – of large mineral oil com-panies and display a considerable range. On average,these total costs without taxes amounted to 15.20 USdollars/bbl; they contain the direct production costs(share: approximately 39 percent), the capital costs(about 47 percent) as well as the costs for exploration(14 percent). Total costs increased in the last five yearsby an annual average of 14 percent. Since for the esti-mate of these production costs only the data of themineral oil companies were used, the calculation doesnot include the corresponding data of the national oilcompanies, which are of great importance for the oilmarket and determine, for example, the supply of theOPEC countries. It is, however, to be assumed thatmany of these national providers are able to produceat relatively low capital and exploration costs. For thisreason it can be presumed that global average produc-
CESifo Forum 4/2010 64
Special
1 The average is calculated by weighting the production costs with therespective shares of the production volume.
CESifo Forum 4/201065
Special
tion costs for 2007 were below the 15.20 US dol-
lars/bbl mentioned above.
For comparison reference is made to a study by
Dresdner Kleinwort (2006), in which also the oil pro-
duction costs of large oil companies are listed for
2005. Average production costs of these companies of
10.5 US dollars/bbl can be derived from a graph in
this study; for 2007, with the above-mentioned yearly
rate of increase, production costs of 13.5 US dol-
lars/bbl result. According to the estimates of
Dresdner Kleinwort (2006), the average production
costs of the Russian companies examined in the study
were less than half as high. These estimates are an
indication of the oft-times favourable production con-
ditions of national oil companies.
Informed by these studies, we turn to another approach
to estimating current oil production costs. The study by
Aguilera et al. (2009) calculates total production costs in
US dollar prices of 2006 for crude oil extracted in the
future. In this approach, which is based on a study by the
United States Geological Survey in 2000, the expected
costs for all relevant oil producing areas are given, and
for most of the deposits total production costs are divid-
ed into (average) capital and operating costs. An
amount-weighted assessment of these total costs showed
that 56 percent of the total costs are accounted for by the
operating costs. Taking this percentage as approximately
valid also for current production, then global operating
costs – as listed in the study of Deutsche Bank (2009) –
of 6.6 US dollars/bbl result in total production costs of
11.8 US dollars/bbl; this corresponds to a share of the
world market price for crude oil in the fourth quarter of
2009 of approximately 16 percent. This value for total oil
production costs is comparably close to the result found
in the study of Jojahrt (2008), in which for 2006 an aver-
age value of 9 US dollars/bbl was indicated (2009:
almost 11 US dollars/bbl) in a range of between 3.38 and
20.79 US dollars/bbl. The production costs estimated by
Remme, Blesl and Fahl (2007) are at about the same
level; a weighted average of ca. 8.8 US dollars/bbl in
prices of 2000 can be derived from the data of this study.
This corresponds to production costs for 2009 of nearly
11 US dollars/bbl crude oil. Altogether, average produc-
tion costs are currently approximately 11 to 12 US dol-
lars/bbl for crude oil.
Rise in oil production costs expected
For estimating the unit costs of crude oil production
from future deposits, there are different possibilities.
On the one hand, the study of Aguilera et al. (2009)
can be used directly, in which the oil still available in
the future and the additional accessible reserves are
indicated at a total of 3.561 trillion bbl; this corre-
sponds to approximately 486 billion toe and thus
almost 2.7 times the reserves that were listed by
ExxonMobil (2009) for the end of 2008. These total
reserves are comprised of the 2.713 trillion bbl from
the amounts derived from the United States
Geological Survey and 848 billion bbl from addi-
tional fields. For the production of the quantities
found in the United States Geological Survey, aver-
age costs are calculated at 6 US dollars/bbl in 2006
prices; the range of these costs extends from 1.08 US
dollars/bbl for deposits in the Arabian Gulf states to
15.27 US dollars/bbl for on- and offshore oil from
Kazakhstan. The costs for the additional oil produc-
tion quantities average 18.9 US dollars/bbl with a
range of between 15.41 US dollars/bbl in Europe
and 26.72 US dollars/bbl in Sub-Saharan Africa and
in Antarctica.
For the total amounts of oil under consideration,
average production costs at 2006 prices can be calcu-
lated amounting to approximately 9.1 US dollars/bbl.
At annual price increases of 14 percent, as listed by
Deutsche Bank (2009) for 2002 to 2007, nominal aver-
age production costs for 2007 are 10.4 US dollars/bbl.
Since the cost increases have slowed down consider-
ably, production costs for 2009 are estimated to total
11 US dollars/bbl. This appears to be comparably lit-
tle in light of the above estimated production costs at
deposits used today of 11.8 US dollars/bbl, and can
only be explained by the fact that the production costs
listed by Aguilera et al. (2009) are to be understood as
static; they include neither the technical progress that
leads to a reduction of costs over time nor do they
take cost increases, as witnessed particularly between
2006 and 2008 into consideration, since they are
regarded to a considerable extent as cyclical.
In the study by Remme, Blesl and Fahl (2007) average
weighted production costs of between 8.6 and
22.9 US dollars/bbl in prices of 2000 – or an average
of 15.8 US dollars/bbl – are calculated for oil deposits
to be tapped in the future; in prices of 2009 this
amounts to 20 US dollars/bbl. Within this range of
production costs lie the costs for the extraction from
oil sands and heavy oil deposits. The production costs
for the extraction of the oil resources, however, are
somewhat below the production costs of enhanced oil
recovery (EOR) of known deposits. Extraction from
oil shale is by far the most expensive.
In addition there is also the possibility of estimatingthe unit costs of deposits to be tapped in future bymeans of the 2008 projections of the IEA.2 These pro-jections see the worldwide supply increasing from84.3 mb/d (million barrels a day) in 2007 to 94.4 mb/din 2015 and 106.4 mb/d in 2030. Via the upstreaminvestments in this period, however, not only theincreasing oil requirement must be satisfied but alsooffset with the natural decline in production of thepresently exploited deposits. The IEA assumes adecline in production of present oil wells of an aver-age of 6.7 percent per annum; this rate will increase inthe coming years and, according to the estimates ofthe IEA, will amount to ca. 10.5 percent per annum in2030. This means that, on the one hand, because ofthe decline in production of present deposits, oil wellsproducing 64 mb/d must be tapped by 2030 and, onthe other hand, because of the increase in the rate ofdecline, the production capacity must be raised by anadditional 23 mb/d. With an increase in the supply ofca. 22.1 mb/d, production capacities must be in-creased by ca. 110 mb/d by 2030. For this, the IEAestimates necessary investments of 5.036 trillion USdollars in 2007 prices; this breaks down to 4.604 tril-lion US dollars for conventional oil and 432 billionUS dollars for non-conventional oil.
Thus, in this period, 45,780 US dollars (2007) will beinvested, on average, for the creation of an addition-al production capacity of 1 b/d. Under the assump-tion of a typical service life of 25 years for an oilfield (see International Energy Agency 2010), invest-ments of 5 US dollars/bbl can be calculated. Thereare, however, considerable differences depending onthe deposits: in the Middle East investments, accord-ing to IEA information, amount to ca. 12,000 USdollars/b/d; converted to the total amount of crudeoil produced, this corresponds to ca. 1.3 US dol-lars/bbl. For oil extraction in the Gulf of Mexico,30,000 US dollars/b/d (3.3 US dollars/bbl) must becalculated, in the North Sea 40,000 US dollars/b/d(4.4 US dollars/bbl) and for the oil sands approxi-mately 70,000 US dollars/b/d (7.7 US dollars/bbl).As a comparison, reference is made to recent esti-mates by OPEC (2009), according to which invest-ments for the development of additional productioncapacity of 1 b/d are much lower. For the period of2010 to 2030, average investment costs of ca. 20,000US dollars/b/d in 2008 prices are listed, whereby thelowest are in the OPEC countries at 12,000 US dol-
lars/b/d and the highest required investments are inWestern Europe at ca. 26,000 US dollars/b/d.
The 5 US dollars/bbl calculated on the basis of theIEA figures stand for the linear write-offs on theinvestments; the interest that accrues in additionduring the life span of the projects is determined viathe annual annuities. For this, starting from the totalinvestment costs up to 2030 amounting to 5.036 tril-lion US dollars, capital costs for the producedamount of approximately 9.8 US dollars/bbl3 are cal-culated at an interest rate of 6 percent.4 Since theprice level for upstream investments from 2007 to2008 only increased by 1.5 percent, it is assumed thatthe price increase between 2007 and 2009 wasapproximately 3 percent, so that the capital costs in2009 should amount to 10.1 US dollars/bbl.Together with the average operating costs of 6.6 USdollars/bbl, total average production costs in 2009prices were 16.7 US dollars/bbl.
The following production costs – in each case in 2009prices – are calculated using this method for depositsopened up in the future in selected regions:
– Middle East: 4.1 US dollars/bbl,– Gulf of Mexico: 11.7 US dollars/bbl,– North Sea: 20.3 US dollars/bbl,– Canadian oil sands: 41.2 US dollars/bbl.
With the current projections of the IEA in the World
Energy Outlook 2009 there is a slight modification ofthe results. Because of the global financial market cri-sis, worldwide economic growth and with it also ener-gy demand will be lower than initially predicted. Theentire supply of oil in 2015, according to the currentIEA estimate, will only increase to 88.4 mb/d – 6 mb/dless than in the estimate of 2008. In 2030, the supplyof oil will be 105.2 mb/d, only 1.2 mb/d less than theearlier forecast. With this oil production costs willalso tend to be lower since the capital productivity willbe higher because of the higher share of oil extractedat lower expense. Hence, a production capacity of ca.105 mb/d is to be attained by 2030 and investments
CESifo Forum 4/2010 66
Special
2 Because of the recent recession, demand for oil will be lower thanestimated in the report at the end of 2008. Current development istaken into account in the IEA report of 2009.
3 These capital costs can only be given approximately because theunderlying annuity refers to a time period, whereas actual oil pro-duction has a different chronological structure. Oil production overtime generally follows an asymmetrical bell curve, with much of theactivity occurring at the beginning of extraction. By placing capitalcosts in relationship to the extracted amount, it is assumed that theoil production is constant each year in the period examined and thatno further investments are necessary in this period.4 The interest rate of 6 percent results from the long-term real capi-tal market interest rate of approximately 4 percent, as calculated forGermany (Bundesbank 2001), and a risk premium as is derived inthe Goldman Sachs (2008) study from the average of the total risk,comprised of the technical and the political risk.
CESifo Forum 4/201067
Special
are likely to be around 8 percent lower. For a produc-tion capacity of 1 mb/d, on average approximately44,000 US dollars are to be invested; this is on thebasis of a total volume of crude oil production of4.8 US dollars/bbl. The capital costs that accrue forthe amounts produced are 9.4 US dollars/bbl in 2007prices. This corresponds to ca. 9.7 US dollars/bbl in2009 and, together with average operating costs of6.6 US dollars/bbl, leads to total average productioncosts of 16.3 US dollars/bbl. It is implicitly assumedhere, however, that not least due to technologicalprogress in oil extraction no strong increase in extrac-tion costs will occur, as was observed in the years upto 2008.
In addition to the average oil production costs, con-sideration must also be given to the marginaldeposits (e.g. oil sands, deep sea deposits), which areof key importance for pricing in the oil market. Animpression of these marginal deposits can be gainedfrom the reports of the major US oil and gas pro-ducers to the Energy Information Agency (2009) ontheir own production costs. For the period 2006 to2008 they indicated that their average worldwideproduction costs (2008 prices) were approximately34 US dollars/bbl.
The results of the research and considerations of aver-age oil production costs as well as the shares of oilproduction costs in the average prices in recent yearsare given in Table 1. Future production costs refer inparticular to the period up to 2030. In light of the
averages listed in Table 1, it must not be overlookedthat oil production costs vary considerably dependingon the different deposits. Current production costs asa share of average oil prices for 2005 to 2009 liebetween 3 percent and 67 percent.
Natural gas production influenced by the oil marketand transport costs
The use of natural gas has expanding strongly world-wide in recent decades, and according to IEA esti-mates demand will continue to grow also in the com-ing decades. Natural gas is used predominantly forheating; its use as a fuel for transport is still of limit-ed importance worldwide. The main reasons for theincreasing use of natural gas are its favourable com-bustion qualities, its versatility and its better environ-mental compatibility in comparison with other com-bustibles. The extraction of natural gas is closely con-nected to mineral oil production, and often it is pro-duced simultaneously. For this reason extensiveanalogies to mineral oil exist with regard to produc-tion conditions and production costs. However, thereare far fewer studies available on the production costsfor natural gas.
For the estimation of the production costs for naturalgas from currently used deposits, the study byRemme, Blesl and Fahl (2007) can be used. Accordingto this study, the cost of natural gas extraction (in USdollar prices of 2000) as a weighted average of all
world areas is between 4 and5.5 US dollars/boe (barrel oilequivalent). The mean valuedetermined by this interval fornatural gas production costs (inprices of 2009) is nearly 6 US dol-lars/boe. In contrast, an IEAstudy of 2009 gives the range ofcosts for natural gas productionfrom conventional deposits; therange is between 2.7 and 32.5 USdollars/boe. Since it can beassumed that there is a tendencyto use the better gas sourcesintensively first, for 2009 produc-tion costs of between 5 and10 US dollars/boe are assumed;for the period up to 2030 a costrange of 8 to 20 US dollars/boeseems realistic. Here it must bepointed out, however, that the
Table 1
Production costs for crude oil and share of production costs
in oil prices (in 2009 prices)
Aguilera
et al.
Remme,
Blesl
and Fahl Jojahrt
Deutsche
Bank/Ifo IEA/Ifoa)
Oil production costs
of currently used
deposits in USD/bbl 11.0 11.0 11.8
Oil production costs
of future deposits in
USD/bbl 11.0 20.0
16.7
(16.3)
Percentage of
production costs in
the total oil priceb)
of
current deposits 16.0 16.0 17.2
Percentage of
production costs in
the total oil priceb)
of future deposits 16.0 29.1
24.3
(23.7)a)
IEA (2008) in brackets IEA (2009). – b)
Average prices from 2005 to
2009.
Source: Listed studies; Ifo Institute.
transport costs for gas comprise up to 80 percent of
total costs of producing gas. For this reason it is plau-
sible to assume that a gas source that is cheaply acces-
sible but far from the markets will only be tapped
after an expensive field that is closer to the markets, if
prices are high.
For the determination of future production costs for
natural gas, recourse can also be made to the num-
bers of Remme, Blesl and Fahl (2007) and of
Aguilera et al. (2009), and again, as with crude oil,
to our own calculations made on the basis of the
IEA reports of 2008 and 2009. In the course of the
increasing depletion of the deposits classified as
reserves, the extraction of natural gas from known
resources takes on increasing importance. The costs
of the natural gas production from these deposits
are comparably low compared with the extraction
from more remote and difficult to access deposits.
The average production costs given by Remme, Blesl
and Fahl (2007) in US dollar prices of 2000 are
between 8.9 and 13 US dollars/boe. In 2009 prices,
average production costs for natural gas stood at
about 13.7 US dollars/boe.
In addition, according to Remme, Blesl and Fahl
(2007), for natural gas production even further in the
future, technologies and sources that are combined
with considerably higher costs will be likely. Among
these, ranked according to ascending costs, are the
increased exploitation of known deposits (EGR –
enhanced gas recovery), the extraction of gas from
coal, tight gas (gas in rock strata), aquifer gas or gas
hydrates. The costs of the natural gas extraction from
gas hydrates is at the upper limit and lies between
55 and 100 US dollars/boe (in 2000 prices); this corre-
sponds to a price range (in 2009 prices) of about 70 to
125 US dollars/boe.
The estimates of natural gas production costs by
Aguilera et al. (2009) are much lower; for future nat-
ural gas production and the growing reserves they cal-
culate average production costs of 4.8 US dollars/boe
(in 2006 prices) on the basis of amounts estimated by
the United States Geological Survey (2000). For addi-
tional deposits with higher extraction expense they
calculate average production costs of 12.6 US dol-
lars/boe. Thus, for the total amount of gas, produc-
tion costs amount to 6.6 US dollars/boe (in 2006
prices); on the basis of 2009 prices this means costs of
approximately 7.5 US dollars/boe. The gas quantities
accessible at these costs are, as for crude oil, consider-
ably higher than in other available estimates. Aguilera
et al. (2009) use a gas production volume of 3,375 bil-lion boe, which is three times the reserves of 1,115 bil-lion boe that ExxonMobil (2009) indicated.
As was done for crude oil, we estimate the unit costsof deposits to be accessed in future by means of the2008 IEA projections. Accordingly, worldwide natur-al gas production will increase from 2,959.3 in 2006 to3,512 billion m3 (2015) and 4,434 billion m3 in 2030.With the upstream investments in this period, com-pensation must be made not only for the growingdemand for gas but also for the natural productiondeclines of currently exploited deposits. We assume adecline in the production of present gas wells of 5 per-cent per annum on average; this rate will increase inthe coming years and will amount to approximately8 percent per annum in 2030. This means that on theone hand because of the production declines of pre-sent extraction sites, gas deposits with an annual out-put of 2,000 billion m3 to 2030 must be accessed andon the other hand that because of the rise in the ratesof decline of production capacity, a further increaseof approximately 1,225 billion m3 will be needed.With an increase of the supply of 1,475 billion m3, theextraction capacities must be expanded by 2030 by ca.4,800 billion m3 per annum, or converted into oil unitsby 30,258 mb/year. The IEA estimates investments of3,322 billion US dollars will be needed for the periodup to 2030 (2007 prices).
For an additional production capacity by 1 b/d,40,070 US dollars (2007) must be invested, on average.Assuming a typical service life of a gas deposit is25 years, investments are calculated at 4.4 US dol-lars/boe. There are, however, considerable differencesdepending on the deposits: In the OECD countriesinvestments are estimated to be around 67,500 US dol-lars/b/d (7.4 US dollars/boe), whereas in the non-OECD countries only ca. 30,000 US dollars/b/d(3.1 US dollars/boe) needs to be invested. Assumingtotal investment costs of 3,322 billion US dollars andan interest rate of 6 percent, capital costs with referenceto the extracted amount of 8.6 US dollars/boe result.In 2009 prices, this means capital costs of ca. 8.9 USdollars/boe. Together with the average operating costs5
of approximately 4.6 US dollars/bbl, total average pro-duction costs of 13.5 US dollars/bbl result for 2009.
Here too, using the 2009 IEA projections there is asmall modification of future production costs. Total
CESifo Forum 4/2010 68
Special
5 Here, the average operating costs of Aguilera et al. (2009) for 2009were used.
CESifo Forum 4/201069
Special
natural gas production, according to current IEA esti-mates, will only increase to 3,395 billion m3 in 2015 –117 billion m3 less than in the forecast for 2008. In2030 gas production at 4,313 billion m3 will be 121 bil-lion m3 below the earlier forecast. Accordingly, alsothe gas production costs will tend to be lower sincecapital productivity will be higher because of thehigher share of natural gas extractible at a smallerexpense. With this in mind, an annual gas productioncapacity of ca. 4,600 billion m3 must be attained,which in oil units corresponds to approximately29,000 mb/year, by 2030; investments should bearound 10 percent lower. For achieving a productioncapacity of 1 mb/d, ca. 37,700 US dollars must beinvested on average; in terms of the total extractedamount of natural gas this amounts to investmentcosts of 4.1 US dollars/boe. Capital costs calculatedon the basis of the total amount of gas producedamount to 8.1 US dollars/boe in 2007 prices. This cor-responds to ca. 9.3 US dollars/boe for 2009 andtogether with average operating costs of 4.6 US dol-lars/boe amounts to total production costs of 12.9 USdollars/boe. As is the case for oil, there is also a largerange of production costs worldwide for gas; becauseof the high share of transport costs and the prevailinglink to oil prices, the marginal sources have a weakerinfluence on gas price formation.
The results of the available cost estimates and ourconsiderations are summed up in Table 2. With refer-ence to the average costs listed in Table 2, it must bepointed out that because of the distribution of the
production costs, their share in prices currently fluc-tuates between 4 percent and 35 percent.
Large black (bituminous) coal reserves dampen costincreases
According to IEA information, coal’s share of world-wide primary energy consumption was 26.5 percent,making it the most important energy source aftermineral oil (34 percent). The importance of the coalas a source of energy has increased since 1990; itsshare in energy consumption grew in this period byone percentage point. Included in the different sortsof coal are brown coal and black coal. Whereasbrown coal is primarily used locally because of itslower energy content, for black coal there is an activeworldwide trade that has continuously expanded inrecent decades. Here, a distinction must be madebetween coking coal, which is mainly employed insteel production, and steam coal, which is used in theheating market. The by-far largest share in the world-wide coal market is for steam coal, with a share ofmore than three quarters. For this reason we only listthe production costs for steam coal, and only forthose countries that dominate the world coal market.This country group includes Indonesia, Colombia,Venezuela, South Africa, China, Australia, the UnitedStates and Russia. Here, it must be noted than inEurope and the United States, coal is used mainly inlarge plants for the production of electricity and heat;direct final consumption only plays a subordinate
role. For this reason, the highdegree of competition is oftengreater in the coal markets thanfor oil and gas.
In supply costs found in the liter-ature, the haulage costs to theconsumers or to the export portsare often included. This is be-cause haulage is very importantfor determining total productioncosts or the differences in pricefor coal. As a result the proximityof coal deposits to ports is a con-siderable site advantage for coal.Here, however, we only take intoconsideration the costs that areimmediately associated with pro-duction, that is the costs for capi-tal, labour, operations and pro-cessing. These are the costs free at
Table 2
Production costs for natural gas and share of production costs
in the gas price (in 2009 prices)
Aguilera
et al.
Remme,
Blesl and
Fahl
IEA
(2009)/Ifoa)
IEA/Ifob)
Natural gas production
costs for current
deposits in USD/boe 6.0 7.0
Natural gas production
costs for future
deposits in USD/boe 7.5 13.7 13.0
13.5
(12.9)
Percentage of
production costs in the
total gas pricec)
for
current deposits 14.1 16.5
Percentage of
production costs in the
total gas priceb)
for
future deposits 17.6 32.3 30.6
31.8
(30.4)a)
Conventional deposits. – b)
IEA (2008) in brackets IEA (2009). –c) Gas
prices from 2005 to 2009.
Source: Listed studies; Ifo Institute.
the site of deposit, as they have been analysed bySchulz (1984a and 1984b) for different types of open-cast and below-surface mining operations.
The average production costs for steam coal from pre-sent deposits, in prices of 2006/2007, according toRitschel and Schiffer (2007) were between 5.5 and10.3 US dollars/boe, which in prices of 2009 is ca. 6 to11 US dollars/boe or, on average, ca. 8 US dollars/boe.The production costs calculated by the IEA (2009) areof a similar order of magnitude. Adjusted approxi-mately for the haulage costs contained in these num-bers, production costs of 7.2 US dollars/boe result;this number also includes the production costs ofexport-relevant mines in China and the United States,in contrast to Ritschel and Schiffer (2007). FollowingRemme, Blesl and Fahl (2007), production costs forsteam coal of 8.5 US dollars/boe for the current pro-duction can be derived; for future production theyindicate costs of about 13.5 US dollars/boe.
Currently approximately 5.6 Gt coal is produced perannum at an average cost of about 8 US dollars/boe.Production costs for additional black coal willincrease to a relatively small extent in the comingdecades (International Energy Agency 2010). Thecosts for the development and production of thefirst 100 Gt of the global coal reserves amount tobetween 5 and 25 US dollars/boe, according to cur-rent estimates. For the next 500 Gt, production costsare estimated to fall in a range of 8 to 40 US dol-lars/boe. During the production of the last 110 Gt
of the known reserves, a strong increase in costs islikely, the estimates ranging between 13 and 123 USdollars/boe. These coal reserves together wouldenable the present annual output to continue for aperiod of more than 120 years. Table 3 shows theproduction costs for steam coal from currently useddeposits and for the mined amounts in the period upto about 2030. In contrast to the average values inTable 3, the share of production costs of coal in theaverage price of steam coal is currently between28 percent and 61 percent.
That the conditions for an expansion of the world-wide coal supply are quite favourably can be seen inthe 2009 IEA report. An expansion of coal consump-tion of around 1.9 percent per annum between 2007and 2030 would require investments – in prices of2008 – of 661 billion US dollars. The demand for oiland gas in these years will increase less strongly, butwill require total investments of 5,919 billion and5,149 billion US dollars, respectively. Thus it wouldnot be surprising if the share of coal in the totalworldwide energy supply increased disproportionatelyon a long-term basis and if mineral oil and naturalgas is used even more strongly in areas, e.g. fuel fortransport, in which they have comparative advan-tages. Currently the use of coal is increasing stronglyin Asia, in particular in China. However, the compa-rably high carbon dioxide emissions connected withcoal argue against the increasing use of this source ofenergy. Since carbon dioxide is regarded as the prima-ry cause of the worldwide climate change, the increas-
ing used of coal can only be re-garded as responsible if it is donewith high combustion efficiencyor if the sequestration of emis-sions prevents the carbon dioxidefrom entering the biosphere.
Conclusion
In light of the great differences inextraction and the qualitative dif-ferences of energy resources fromthe individual deposits, the deter-mination of average productioncosts for the total production ofan energy source can only beunderstood as an approximationof actual costs. Table 4 containsthe average values of present andfuture production costs for crude
CESifo Forum 4/2010 70
Special
Table 3
Production costs for steam coal and share of production costs
in the coal price (in 2009 prices)
Ritschel
und Schiffer
Remme,
Blesl and
Fahl
IEA
(2009)/Ifoa)
IEAb)
Coal production costs
for current deposits in
USD/boe 7.9 8.5 7.2 8.2
Coal production costs
for future deposits in
USD/boe 11.5 10.0
Percentage of
production costs in
total coal pricec)
for
current deposits 44 47 40 46
Percentage of
production costs in
total coal priceb)
for
future deposits 64 56a)
IEA (2009); selected countries. –b)
IEA (2010) (expected), up to
100 Gt. – c)
Coal prices between 2005 and 2009.
Source: Listed studies; Ifo Institute.
CESifo Forum 4/201071
Special
oil, natural gas and steam coal, expressed in each casein barrels of oil or oil equivalent, as well as the respec-tive share of the average production costs in the aver-age energy source prices for 2005 to 2009.
For crude oil there are similarly clear circumstances,since world market prices exist for the different oilqualities. For 2009 average production costs crude oilof ca. 11.4 US dollars/bbl were determined. This cor-responds to a 17-percent share in the average oil priceof 2005 to 2009 of ca. 69 US dollars/bbl. The produc-tion costs for oil that will be produced in the period upto 2030 is likely to amount to 16 US dollars/bbl in2009 prices, on average.
With regard to natural gas, the situation is somewhatless clear especially because of the existence of sev-eral regional markets. The weighted average produc-tion costs of natural gas amounted to 7 US dol-lars/boe in 2009. With an average natural gas price ofca. 42.5 US dollars/boe, a ca. 16-percent share ofproduction costs can be calculated. For natural gasprovided in the period up to about 2030, productioncosts will probably amount to 11 to 12 US dol-lars/boe, in 2009 prices.
The production costs for steam coal refer only to theconditions in the most important exporting coun-tries. For 2009 and based on available studies, currentcosts of ca. 8 US dollars/boe can be calculated. Basedon the current average price of steam coal of ca.18 US dollars/boe, this amounts to a 45-percent shareof production costs, whereby the great differencesamong the providers must be kept in mind. Thefuture production costs for steam coal will increase ata comparably lower pace and will probably be of anorder of magnitude of 10 US dollars/boe in the peri-od up to 2030.
References
Aguilera, R.F., R.G. Eggert, G. LagosC.C. and J.E. Tilton (2009), “Depletionand the Future Availability of PetroleumResources”, The Energy Journal 30,141–174.
Deutsche Bank (2009), The Cost ofProducing Oil, London: Deutsche BankGlobal Markets Research.
Deutsche Bundesbank (2001), “Real-zinsen: Entwicklung und Determi-nanten”, Monatsbericht Juli, 33–50.
Dresdner Kleinwort (2006), Russian Oiland Gas: Key Trends and InvestmentOpportunities, November.
Energy Information Administration(2009), Performance Profiles of MajorEnergy Producers 2008,http://www.eia.doe.gov/emeu/perf-pro/0206(08).pdf.
ExxonMobil (2009), Oeldorado, http://www.exxonmobil.de/unternehmen/service/publikationen/downloads/files/oeldorado09_de.pdf.
Goldman Sachs (2008), Global: Energy, 190 Projects to Change theWorld, London.
International Energy Agency (2003), World Energy InvestmentOutlook 2003, Paris.
International Energy Agency (2005), Resources to Reserves, Oil & GasTechnologies for the Energy Markets of the Future, Paris.
International Energy Agency (2008), World Energy Outlook 2008,Paris.
International Energy Agency (2009), World Energy Outlook 2009,Paris.
International Energy Agency (2010), Resources to Reserves, forth-coming.
Jojahrt, C. (2008), The End of Easy Oil: Estimating AverageProduction Costs for Oil Fields around the World, Working Paper 72,Program on Energy and Sustainable Development, StanfordUniversity.
OPEC (2009), World Oil Outlook 2009, Vienna.
Remme, U., M. Blesl and U. Fahl (2007), Global Resources andEnergy Trade: An Overview for Coal, Natural Gas, Oil and Uranium,IER-Forschungsbericht 101, Universität Stuttgart, Institut fürEnergiewirtschaft und Rationelle Energieanwendung (IER),Stuttgart.
Ritschel, W. and H.W. Schiffer (2007), Weltmarkt für Steinkohle,Essen-Köln: RWE Power.
Schulz, W. (1984a), “Die langfristige Kostenentwicklung derSteinkohle am Weltmarkt (Teil I)”, Zeitschrift für EnergiewirtschaftMarch, 8–20.
Schulz, W. (1984b), “Die langfristige Kostenentwicklung derSteinkohle am Weltmarkt (Teil II)”, Zeitschrift für EnergiewirtschaftJune, 108–117.
United States Geological Survey (2000), World Petroleum Assess-ment, Washington DC.
Table 4
Production costs for energy sources and share of production costs
in the price of the energy source (in 2009 prices)
Oil Gas Steam coal
Average production costs for current
deposits in USD/boe 11.4 6.8 8.0
Average production costs for future
deposits in USD/boe 16.0 11.6 10.0
Percentage of production costs in
the total price of the energy source
for current deposits 17 16 45
Percentage of production costs in
the total price of the energy course
for future deposits 23 27 56
Source: Listed studies; Ifo Institute.
CESifo Forum 4/2010 72
Special
REPEC – AN INDEPENDENT
PLATFORM FOR MEASURING
OUTPUT IN ECONOMICS
CHRISTIAN SEILER AND
KLAUS WOHLRABE*
Introduction
Research assessments have become an importantpart of the academic world. Since they often formthe basis in appointment procedures and financingdecisions within the research world, the quality ofthe assessment depends significantly on the qualityof the underlying measurements (see Combes andLinnemer 2010). A well-known procedure inGerman-speaking countries for measuring researchoutput is the ranking of the daily business newspa-per Handelsblatt, which has gained a high degree ofattention in the area of economics (see Hofmeisterand Ursprung 2008). In addition to the rankingsfor authors there is also one for institutes. However,for the latter only economics departments at uni-versities are included and not research institutions(such as Ifo Institute or DIW Berlin). At the inter-national level there are the Times Higher WorldUniversity Ranking and the Shanghai Ranking.The focus of these studies is on a comparison ofuniversities.
In this article we present the RePEc network as anindependent platform for current research assess-ment, in particular in the area of the economics. Firstwe describe how RePEc functions, then we presenthow the rankings are calculated in the network, whichincludes both institutions (faculties, think-tanks, etc.)as well as authors. We also indicate how these rank-ings can be interpreted and some points that must bekept in mind. The advantages of this network are thegreat number of ranking criteria and that they arevery up-to-date.
The RePEc network
The RePEc network (Research Papers in Economics, )is a bibliographic service for economic research andits adjunct fields such as statistics. The goal of thisnetwork consists in constructing as complete a collec-tion as possible of all research results that have beenpublished in some form. Also, by using this informa-tion, various evaluations or rankings can be pro-duced. An important difference from many otherranking methods is that RePEc is based on the ‘wiki’principle and the relevant information is not compiledby an individual author or institution. This principleapplies both to authors as well as publishers, which tosome extent are dependent on each other. On the onehand, publishers must make available the meta-infor-mation of their publications (journal articles, books,book contributions, working papers) such as authornames, titles, editions, number of pages or citations.On the other hand, scholars must register themselvesat RePEc and classify their works. This enables a clearallocation to the authors. With the help of the infor-mation available in the network, rankings can be com-puted for authors and institutions. A potential disad-vantage, however, is that some information (e.g. par-ticular journals or citations) may not have been madeavailable to the network or, because of the concentra-tion on economic research, publications from otherdisciplines may not be included. For the researchcommunity there is thus a strong incentive to make asmuch information as possible available in order tofully exhaust the network effect.1
Ranking criteria
On basis of the bibliographic information available inthe network, RePEc releases monthly 34 differentrankings for registered authors as well as institutions.2
The basis is comprised of five main categories: thenumber of (registered) works as well as the numbersof pages of the journal articles, the number of cita-tions, the number of cited authors and access statistics
* Ifo Institute for Economic Research.
1 For further information on how the RePEc networks functions, seeZimmermann (2007).2 The RePEc network itself considers all its rankings to be experi-mental.
CESifo Forum 4/201073
Special
of the RePEc network. For some of these categories,
additional weightings with different quality features
are carried out (see Table 1 for an overview). In the
following the criteria are described in detail.
Number of works
For tallying the number of the works, all scholarly
publications, i.e. working papers, journal articles,
book contributions, books and software components,
are considered. Since articles can be published again
in different series (especially working papers) this can
lead to distortions in the evaluation. Therefore differ-
ent publications of the same article are counted only
as one individual work.
Number of pages
For the number of the published pages only the arti-
cles that have appeared in academic journals and as
book contributions are counted. The pages in work-
ing papers and books are not counted. Since books
and working papers as a rule are not subject to a
review process, the author has a strong influence on
the number of the pages, which in the case of articles
in academic journals is mostly determined externally.
With this criterion it must be noted however that the
number of characters per page can vary between jour-
nals and is not adjusted by RePEc.
Citations
For this criterion all citations registered in the net-
work are counted and assessed. In order to avoid an
artificial increase because of self-citations, these are
not taken into consideration. In addition to a simple
counting, the citations are also adjusted by the year of
citation in order to reflect the current scholarly dis-
course. This means that an article that was quoted in
recent years has more weight than an article that is
quoted just as frequently but further back in the past.
On basis of the citations Hirsch’s h-Index (2005) is
also calculated. This is defined as follows: a scholar
has an index of h if from his n articles h was quoted at
least h times. The remaining (n-h) articles do not have
more than h citations. This index reflects both the
quality and the breadth of research results. Scholars
with few but frequently cited works tend to have a
lower h-index. A stricter variant of the h-Index is the
Wu-index (see Wu 2008). This is calculated in a simi-
lar way as the h-Index, but one needs ten times as
many citations to reach a value of w.
Cited authors
A problem with the simple counting of citations isthat so-called ‘citation clubs’ can emerge in which afew authors quote themselves mutually. In order toavoid this, the number of different authors making thecitations is also counted. Here, a weighting is also car-ried out that depends on the rank of the author inRePEc. If a well-ranked author quotes a less well-ranked author, the latter profits more from this thanwhen he is quoted by an author whose ranking is notas high.
RePEc access assessment
In the RePEc network all abstract views and down-loads are registered. Automated accesses, e.g. byrobots, are registered by the system and filtered out.In addition a control is made for suspicious accesspatterns in order to avoid individual manipulations.All rankings based on these access assessmentsinclude only the last twelve months.
Quality weighting
In order to register the quality of scholarly works, inaddition to the citations, further quality features areconsidered: the published articles are weighted withthe impact factors of the publication series. These arecomputed by the network itself. In addition an adjust-ment is made for the number of authors.
Impact factors
The impact factors computed by RePEc differ fromthe semi-official impact factors of Thomson Scientificin that the latter uses only the number of articles fromthe past two years.3 The impact factor is calculated bydividing the number of the citations in year t of thearticles in the years t–1 and t–2 by the number of arti-cles that appeared in years t–1 and t–2. An impact fac-tor larger than 1 means that there are more citationsof the articles from the last two years than articlespublished in this period. There are also impact factorsthat include more than two years, but the above-men-tioned calculation method is the most widespread.The main difference between the RePEc method andthe semi-official impact factors is that a correspond-ing value is computed for every journal and workingpaper series. A comparison is only partially possible,
3 A good survey of the historical development and interpretation ofthe impact factor is given by Garfield (2006).
CESifo Forum 4/2010 74
Special
however, since the impact factors of RePEc are notchronologically limited and since citations from non-listed sources in RePEc or in other disciplines are notregistered. Another difference in the calculation of theimpact factors is that the self-citations of journals(which need not necessarily be by the same author)are not considered.
A possible disadvantage of the simple impact factorsis that all citations are weighted equally, i.e. citationsfrom top journals are treated the same way as cita-tions from journals of a lower standing. In order toadjust for this problem, the recursive impact factor iscalculated in addition. Here, however, there is an opti-misation problem since the weighting of the citationsin turn influences the underlying impact factor of thejournal. In its calculations RePEc uses algorithmsthat start with the simple impact factors and iterative-ly adjust the recursive impact factors. These recursiveimpact factors are to be interpreted in such a way thatthey measure the importance of the journals relativeto each other whereas the simple impact factors regis-ter the absolute importance.
Number of the authors
In addition to the qualitative weighting of an articlevia the impact factor of the series, the number ofauthors of a publication is also taken into consider-ation. The more authors an article has, the fewerpoints an individual author receives. Differences
between main and co-authors are not observed byRePEc; every author thus has an equal share in apublication.
Calculation of an overall ranking for authors
As evident in Table 1, not all combinations of thediscussed criteria and weighting are calculated. Thisis not always sensible (e.g. the weighting of down-loads according to impact factors) and in additionparticular criteria are to be assigned a greater weightin the calculation of the average ranking. This is par-ticularly the case for citations that form the basis for13 of the 33 rankings and are thus the main qualityfeature.
For every combination listed in Table 1, a value is cal-culated for every author, and based on this a rankingis made. These rankings form the basis for the overallranking. For every author the best and worst place-ment is removed and from the remaining placementsthe average ranking is formed with the help of har-monious averaging.4 Here neither the total number ofarticles nor the Wu index are taken into account. Theaverage rankings are sorted in ascending order andthus display the overall ranking.
Table 1
Academic rankings in RePEc
Wit
ho
ut
any
fu
rth
er
wei
gh
tin
g
Sim
ple
im
pac
t
fact
or
Rec
urs
ive
imp
act
fact
or
Nu
mb
er o
f au
tho
rs
Nu
mb
er o
f au
tho
rs
+ s
imp
le i
mp
act
fact
or
Nu
mb
er o
f au
tho
rs
+ r
ecu
rsiv
e im
pact
fact
or
Overalls XWorks
Distinct X X X X X X
Overalls X X X X X X Citations
Discounted by citation year X X X X X X
Overalls XCiting authors
Weighted by authors slim X
Journal pages X X X X X X
Abstract views X X Access via RePEc
Downloads X X
h-Index XIndices
Wu-Index* X
* Only for authors.
Source: www.repec.org.
4 Harmonious averaging is preferred over the arithmetic averaging bythose authors and institutions that are far in front in a few rankings,as is evident in Box 1.
CESifo Forum 4/201075
Special
Ranking of institutions
The idea of an institutional ranking is based on thefact that every institution can be seen as an individualauthor to which all the articles of affiliated personsare assigned. Every author who has registered worksin RePEc contributes an added value to the corre-sponding institution. Building on this, the total rank-ing is calculated similarly to that of the author rank-ing. This procedure is unproblematic as long as everyauthor can be assigned precisely to an institution. Butwhat happens if an author assigns himself to severalinstitutions?5 In such cases RePEc carries out a distri-bution to the individual institutions. The ‘main insti-
tution’ receives 50 percent, andthe remaining 50 percent is dis-tributed to all the other listedinstitutions, with a correctionbeing made for the number ofaffiliated persons. The main insti-tution is not indicated explicitlybut is determined by the regis-tered e-mail address or the indi-cated Website of the author.
A European ranking for authorsand institutions
Table 2 lists the 25 best institu-tions in Europe. A comparison ofthe worldwide ranking with theEuropean one shows, for exam-ple, that the Institute for FiscalStudies in London is in17th placein the worldwide ranking ofEuropean institutions but is in10th place in the Europe ranking.This inconsistency is a result ofthe formation of averages via therankings and is explained inBox 1 by means of an example.Table 2 also includes the numberof authors assigned to this insti-tution in RePEc. Here it is evi-dent that a high number does notnecessarily lead to a high place-ment, as can be seen in particularwith the Centre de Recerca enEconomia Internacional (CREI)in Barcelona. In general, itshould be noted for the institu-tional rankings that due to volun-
tary registration, the number of authors of a particu-lar institution in RePEc does not necessarily have tocorrespond with the actual number.
Table 3 shows the Europeans rankings for authors.The large discrepancy to the worldwide placementarises from the fact that all authors of the worldwideranking who are at least affiliated with one Europeaninstitution are taken into consideration in this rank-ing. This is problematic in particular if the relevant
Box 1
Illustration of the shifts in regional rankings
In addition to the ranking of all registered authors and institutions in the
RePEc network, regional rankings are also made, for example for
Germany and the European Union. Due to the affiliation of many authors
to institutions from different areas, it can be the case that an author’s name
appears in several regional rankings. Although institutions, unlike authors,
can be assigned unambiguously to a region, there can be inconsistencies
also in the institutional rankings similar to those in the Europe ranking.
The following simplified example briefly illustrates this problem:
Institutions A and B exist in a specific region. For these institutions,
rankings were made according to five different criteria (I-V) in the table
below. In the worldwide ranking, institution A is particularly well
positioned in Ranking I and II. Institution B does not stand out in any
criterion but is somewhat better than A in Rankings III–V. The average
ranking of Institution A, however, is less than that of Institution B because
of the clear advantage in I and II, both on the basis of the harmonic as well
as the arithmetic mean. If the rankings are transferred to the regional
assessments, the advantage of Institution A over B no longer exists. To be
sure, the underlying scores of A are still clearly better than those of B, but
since B is the second-best institution in this region according to criterion I
and II, the difference in the ranking is only that of one place. Since,
however, B is better than A in the other three rankings, a lower average
ranking for A results and with this a better placement in the regional
ranking. This is known as Simpson’s Paradox (see Simpson 1951) and
can only be resolved in this particular case if one were to calculate the
average scores instead of the average ranking. Since the scores between
the ranking criteria are not comparable, however (for the best placement
in I a score of 5 000 points is necessary, but for II only three points),
considerable distortions can result in such a calculation method.
Worldwide ranking
I II III IV V Harmonic
mean
Arithmetic
mean
A 9 11 202 234 198 23.1 130.8
B 175 182 135 152 178 162.3 164.4
Regional ranking
I II III IV V Harmonic
mean
Arithmetic
mean
A 1 1 2 2 2 1.4 1.6
B 2 2 1 1 1 1.3 1.4
5 A prominent example is the affiliation in academic networks, suchas the CESifo network or NBER. In most cases, this affiliation is thesecond institution next to the main institution.
CESifo Forum 4/2010 76
Special
authors are mainly active abroad but have assignedthemselves to a well-known network such as CESifoor CEPR. In the worldwide ranking for all authorsaffiliated with European institutions, Peter Phillips(Yale University, University of Auckland and
University of York) is in first place but in theEuropean ranking only in place 785. This poor place-ment is because less than 50 percent of his outputwent into the European ranking whereas 100 percentwas included for the worldwide ranking.
Table 2
Ranking of institutions in Europe: top 25
Wor
ldw
ide
Wor
ldw
ide
Eur
opea
nin
stitu
tion
s
Eur
ope
Inst
ituti
on
Cou
ntry
Num
ber
ofre
gist
ered
auth
ors
8 2 1 London School of Economics (LSE), London United Kingdom
216
7 1 2 Department of Economics, Oxford University, Oxford United Kingdom
142
19 3 3 Toulouse School of Economics (TSE), Toulouse France 128 35 5 4 Department of Economics, University College London
(UCL), London UnitedKingdom
69
36 6 5 Department of Economics, University of Warwick,Coventry
UnitedKingdom
66
40 8 6 European Central Bank, Frankfurt am Main Germany 122 33 4 7 CentER for Economic Research, Universiteit van
Tilburg, Tilburg Netherlands 139
38 7 8 Faculty of Economics, University of Cambridge,Cambridge
UnitedKingdom
68
43 9 9 Paris School of Economics, Paris France 171 69 17 10 Institute for Fiscal Studies (IFS), London United
Kingdom53
46 10 11 Organisation de Coopération et de DéveloppementÉconomiques (OCDE), Paris
France 144
56 12 12 Wirtschaftswissenschaftliche Fakultät, UniversitätZürich, Zürich
Switzerland 66
72 18 13 Institute for International Economic Studies (IIES),Stockholms Universitet, Stockholm
Sweden 20
67 15 14 Faculteit Economie en Bedrijfskunde, Universiteit van Amsterdam, Amsterdam
Netherlands 77
68 16 15 ECORE, Louvain-la-Neuve/Bruxelles Belgium 93 53 11 16 Faculteit der Economische Wetenschappen en
Bedrijfskunde, Vrije Universiteit, AmsterdamNetherlands 80
66 14 17 Solvay Brussels School of Economics and Management, Université Libre de Bruxelles, Bruxelles
Belgium 132
85 20 18 Innocenzo Gasparini Institute for Economic Research(IGIER), Università Commerciale Luigi Bocconi,Milano
Italy 39
87 21 19 Centre for Economic Policy Research (CEPR), London United Kingdom
258
91 23 20 Centre de Recerca en Economia Internacional (CREI),Barcelona
Spain 14
76 19 21 School of Business and Economics, MaastrichtUniversity, Maastricht
Netherlands 105
65 13 22 ifo Institut für Wirtschaftsforschung e.V., München Germany 123 89 22 23 DIW Berlin (Deutsches Institut für
Wirtschaftsforschung), BerlinGermany 119
99 25 24 Institut National de la Statistique et des ÉtudesÉconomiques (INSEE), Government of France, Paris
France 56
95 24 25 School of Economics, University of Nottingham,Nottingham
UnitedKingdom
65
Source: www.repec.org (Status: October 2010 ranking).
CESifo Forum 4/201077
Special
Closing remarks
This article has discussed how the RePEc network
operates. In addition to the aspect of the registration
and dissemination of economic research, the rankings
it makes play an important role in the evaluation of
authors, institutions and journals. The main advantage
is in the dissemination of networks, i.e. a major por-
tion of research output in economics (including work-
ing papers) is registered. This is, however, only guaran-
teed if bibliographic information is regularly main-
tained and the authors keep their profiles updated. In
order to fully exhaust the network effects, it is impor-
tant to list authors and journals that are still missing in
RePEc. The independent preparation of 34 individual
rankings in addition provides a transparent assess-
ment of the research output on the basis of different
criteria both for authors and for institutions.
References
Combes, P.-P. and L. Linnemer (2010), Inferring Missing Citations: AQuantitative Multi-Criteria Ranking of All Journals in Economics,www.vcharite.univ-mrs.fr/pp/combes/Journal_Ranking.pdf.
Garfield, E. (2006), “The History and Meaning of the Journal ImpactFactor”, Journal of the American Medical Association 295, 90–93.
Hirsch, J.E. (2005), “An Index to Quantify an Individual’s ScientificResearch Output”, Proceedings of the National Academy ofSciences 102, 16569–16572.
Hofmeister, R. and H.W. Ursprung (2008), “Das HandelsblattÖkonomen-Ranking 2007: Eine kritische Beurteilung”, Perspektivender Wirtschaftspolitik 9, 254–266.
Simpson, E.H. (1951), “The Interpretation of Interaction inContingency Tables”, Journal of the Royal Statistical Society,Series B, 13, 238–241.
Wu, Q. (2008), The w-Index: A Significant Improvement of the h-Index, http://arxiv.org/abs/0805.4650v1.
Zimmermann, C. (2007), Academic Rankings with RePEc,Department of Economics Working Paper 2007-36R, University ofConnecticut.
Table 3
Ranking of economists in Europe: top 25
Worldwide Worldwide European authors Europe Author
9 2 1 Jean Tirole
30 5 2 Lars E.O. Svensson
35 8 3 Jordi Gali
38 9 4 Richard Blundell
79 16 5 Timothy J. Besley
80 17 6 Stephen John Nickell
87 20 7 Guido Tabellini
65 13 8 Florencio Lopez-de-Silanes
19 4 9 Nicholas Cox
103 26 10 Ernst Fehr
67 14 11 Bruno S. Frey
32 6 12 Peter Nijkamp
126 32 13 Torsten Persson
139 38 14 Christopher A Pissarides
140 39 15 Andrew J. Oswald
50 12 16 M. Hashem Pesaran
152 41 17 Soren Johansen
132 35 18 David F. Hendry
149 40 19 Anthony J. Venables
182 47 20 John Moore
153 42 21 Gilles Saint-Paul
188 48 22 Stephen Roy Bond
180 46 23 Assar Lindbeck
121 30 24 Hans-Werner Sinn
203 56 25 Athanasious Orphanides
Source: www.repec.org (Status: October 2010 ranking).
CESifo Forum 4/2010 78
Trends
FINANCIAL CONDITIONS
IN THE EURO AREA
The annual growth rate of M3 stood at 1.0% in October 2010, compared to1.1% in September 2010. The three-month average of the annual growthrate of M3 over the period from August to October 2010 rose to 1.1%, from0.8% in the period from July to September 2010.
Between April and November 2009 the monetary conditions index re-mained rather stable after its rapid growth that had started in mid-2008.The index started to grow again since December 2009, signalling greatermonetary easing and reached its peak in June 2010. In particular, this hasbeen the result of decreasing real short-term interest rates. However, theindex started to decline again since June 2010.
In the three-month period from September to November 2010 short-terminterest rates increased. The three-month EURIBOR rate grew from anaverage 0.88% in September to 1.04% in November. The ten-year bondyields also increased from 3.50% in September to 3.73% in November. Inthe same period of time the yield spread increased from 2.62% (September)to 2.69% (November).
The German stock index DAX grew in November 2010, averaging6,688 points compared to 6,229 points in September. The EuroSTOXX also increased from 2,766 in September to 2,810 in November.The Dow Jones International grew as well, averaging 11,198 points inNovember compared to 10,598 points in September.
CESifo Forum 4/201079
Trends
According to the first Eurostat estimates, GDP increased by 0.4% in theeuro area (EU16) and by 0.5% in the EU27 during the third quarter of2010, compared to the previous quarter. In the second quarter of 2010 thegrowth rates were 1.0% in both zones. Compared to the third quarter of2009, i.e. year over year, seasonally adjusted GDP increased by 1.9% in theeuro area and by 2.2% in the EU27.
In November 2010, the Economic Sentiment Indicator (ESI) continued toimprove in both the EU27 and the euro area (EU16). The indicatorincreased strongly, by 1.3 of a point in the EU27 and, more significantly,by 1.5 of a point in the euro area, to 105.2 and 105.3 respectively. In boththe EU27 and the euro area the ESI stands above its long-term average.
* The industrial confidence indicator is an average of responses (balances) to thequestions on production expectations, order-books and stocks (the latter with invert-ed sign).** New consumer confidence indicators, calculated as an arithmetic average of thefollowing questions: financial and general economic situation (over the next12 months), unemployment expectations (over the next 12 months) and savings (overthe next 12 months). Seasonally adjusted data.
In November 2010, the industrial confidence indicator improved by0.7 points in the EU27 and by 0.9 points in the euro area (EU16). On theother hand, the consumer confidence indicator increased by 0.5 points in theEU27 and, more strongly, by 1.5 points in the euro area.
Managers’ assessment of order books improved from – 16.3 in September to– 13.5 in November 2010. In August the indicator had reached – 18.4.Capacity utilisation increased to 77.9 in the fourth quarter of 2010, from77.3 in the previous quarter.
EU SURVEY RESULTS
CESifo Forum 4/2010 80
Trends
The exchange rate of the euro against the US dollar averaged 1.37 $/€ inNovember 2010, an increase from 1.31 $/€ in September. (In August therate had amounted to 1.29 $/€.)
The Ifo indicator of the economic climate in the euro area (EU16) rose onceagain marginally in the fourth quarter of 2010, and is now only slightlybelow its long-term average. The assessments of the current economic situ-ation were considerably more favourable than in the third quarter of 2010.The expectations for the next six months, however, were once again weakeralthough they remain positive on the whole. The survey results indicate thatthe economic recovery in the euro area will continue in the coming sixmonths, albeit at a slower pace than before.
Euro area (EU16) unemployment (seasonally adjusted) amounted to 10.1%in October 2010, compared to 10.0% in September. It was 9.9% in October2009. EU27 unemployment stood at 9.6% in October 2010, also unchangedcompared to September. The rate was 9.4% in October 2009. In October2010 the lowest rate was registered in the Netherlands (4.4%) and Austria(4.8%), while the unemployment rate was highest in Spain (20.7%) andLatvia (19.4% in the second quarter of 2010).
Euro area annual inflation (HICP) was 1.9% in October 2010, comparedto 1.8% in September. A year earlier the rate had amounted to – 0.1%. TheEU27 annual inflation rate reached 2.3% in October 2010, up from 2.2%in September. A year earlier the rate had been 0.5%. An EU-wide HICPcomparison shows that in October 2010 the lowest annual rates wereobserved in Ireland (– 0.8%), Latvia (0.9%) and Slovakia (1.0%), and thehighest rates in Romania (7.9%), Greece (5.2%) and Estonia (4.5%). Year-on-year EU16 core inflation (excluding energy and unprocessed foods)rose to 1.1% in October 2010 from 1.0% in August.
EURO AREA INDICATORS