Download - Determinantes urbanos, económicos y estructurales en la generación de empleo en Argentina
Programa de Pasantías InternacionalesBuenos Aires - Argentina
Observatorio New School - SES
“Determinantes urbanos, económicos y estructurales en la generación de empleo en
Argentina”
Director de GPIA de The New School University
Michael Cohen
Coordinador Académico del Programa de Pasantías en Bs. As.
Alberto Minujin
Pasantes del programa 2012 en ACUMAR
Emily Miller
2012
Urban, Economic, and Structural Determinants of Employment Generation in Argentina
Emily Miller
The New School
Argentina IFP, 2012
1
Acknowledgements
First and foremost, I would like to thank Martín Abeles at CEPAL for his insight, patience, and
thought-provoking discussions. I deeply grateful for Roxana Maurizio’s help in conceptualizing
the work in its initial stages and for making the work with microdatos possible. Thanks to the
entire Buenos Aires CEPAL office and Alberto Minujín for supporting me in this internship, with
a special thanks to Daniel Vega for his encyclopedic knowledge of data, and to Pascual
Gersetenfeld for ensuring a smooth transition. Finally, I am indebted to Michael Cohen for
planting the seed both of this work and of the relationship with CEPAL.
2
INTRODUCTION
Employment has become an increasingly important topic for both developing and
developed nations. Developed nations such as the United States struggle with the effects of
recent financial crises, and debates continue over how to beckon back down elevated
unemployment rates. Developing nations such as Argentina boast relatively more financial
stability compared to past decades, but continue to struggle with generating quality employment
for all sectors of the population. This analysis focuses on employment generation on a macro
level, integrating the fields of macroeconomics and urban policy/planning. While employment
has been discussed in both economic and urban studies, such analyses have largely remained in
their respective silos. In contrast, this study seeks to understand the city as an economic entity,
linking macroeconomic regimes with urban factors in order to provide a more integrated analysis
of the determinants of employment generation in Argentina.
The distinct repercussions in the labor market of the opposing macroeconomic regimes
Argentina, specifically the Convertibility period (1991-2001) and the post-crisis model (2003-
2010), are the focus of the first section of this paper. Illustrating the fundamental link between
macroeconomic regimes and employment generation, this first section outlines the basic tenants
of the two macroeconomic models, the resulting trajectories of the real exchange rate and gross
domestic product (GDP), and the mechanisms of transmission to the labor market. This section
shows that without GDP growth, there cannot be employment, but that growth alone does not
necessarily ensure employment growth as happened in the 1990s. Rather, economic growth must
be accompanied by a favorable set of exchange rate and institutional (labor and social) policies
that help ensure parallel growth in the labor market.
While employment generation was much greater in the post-Convertibility period, such
growth was not homogenous across cities. The second part of this study therefore enters into the
3
urban policy dynamic to hypothesize why under a fixed macroeconomic regime, different areas
have been able to generate more employment than others. Urbanization is a salient topic given
the high urbanization rates in Latin America and the Caribbean – at around 80%, they are some of
the highest in the world1.
Argentina is no exception, with an
urbanization rate at 89% of the
population (Prud’homme,
Huntzinger, & Kopp, 2004).
Moreover, there is a positive
relationship between urbanization
and per capital GDP growth,
which shows that while urbanization does not guarantee economic growth, it does go hand in
hand with economic development (Buckley, 2012; CEPAL, 2012). In this second section,
theories of economies of agglomeration are used to trace the transmission mechanisms between
size and density and employment generation.
The third section introduces two additional determinants of employment generation: the
production structure and the provision of infrastructure. Using the discourse of structuralist
economics, it is hypothesized that a more complex production structure (less structural
heterogeneity) should facilitate higher levels of employment generation. The role of
infrastructure is presented as having a facilitating role in the reduction of structural heterogeneity,
and in the realization of benefits from urban size and density.
The fourth section outlines two econometric models used the test the relative weights of
1 When urban areas are measured more liberally as areas over 2,000 people, the urbanization rate in Latin America and the Caribbean has grown from 40% to 80% of the population between 1950 and 2000. Even when urban areas are measured with a more restrictive value of 20,000 people or more, the urbanization rate has still more than doubled, increasing from 32% of the population in 1950 to 62% of the population in 2000 (CEPAL, 2012).
4
macroeconomic regimes and urban factors in employment generation. The first model is a macro
level estimation of aggregate labor demand, using time series data on GDP, city size and density.
The second model is a micro level estimation of the probability of being employed, analyzing the
effects of size, density, and production structure complexity, after controlling for demographic
variables such as age, gender, and education. The results are presented in section five.
Section six presents three categories of policy implications as well as topics for future
research. The first deals with the importance of macroeconomic regimes, reiterating the
importance of a competitive real exchange rate which increases the profitability of tradable
sectors and promotes industrialization. The second relates to production structure complexity and
argues for policies which reduce structural heterogeneity via education and technology policies,
and support to small and medium enterprises. The third relates to size and justifies the use of
industrial policies in cities which are too small to benefit from the endogenous employment-
generation capacity of large cities stemming from agglomeration economies. Finally this study
argues that while the models show a positive effect of city size on employment generation, one
must be wary of “too much of a good thing.” While size and density generate positive
employment dynamics, there is a saturation point past which congestion and other negative
effects counterbalance the earlier benefits. This leads into a discussion on regional spatial
planning and argues for the use of policies to encourage the development of medium-sized cities.
5
1. THE ROLE OF THE INTERNATIONAL CONTEXT AND MACROECONOMIC
REGIMES
Following Damill and Frenkel (2011), the macroeconomic evolution of Argentina should
be understood as a result of the interaction between changes in the international context, the
particular structural configuration of the local economy, and the main characteristics of the
national macroeconomic policy. This section therefore seeks to link these three areas in a broad
review of the distinct macroeconomic regimes implemented in Argentina between 1990 and 2010.
1.1 INTERNATIONAL CONTEXT
As a consequence of the financial integration implemented in the second part of the
seventies, many Latin American countries had accumulated high and persistent current account
deficits and external debt; Argentina was no exception. However in the 1980s, the erosion of
foreign finance and sudden spike in international interest rates due to contractionary policies in
the US triggered massive balance of payments crises in Argentina, Brazil, Chile, Ecuador, Peru,
Uruguay and Venezuela. Unlike the inflation situation Chile and Colombia in which levels never
rose above 35%, inflation in Argentina stubbornly remained in triple digits, only to be disciplined
by a strict fixed peg exchange rate regime implemented in 1991 (Frenkel & Rapetti, 2012).
Unlike the 1980s, the international environment in the 1990s was more favorable in terms
of capital availability. Due in part to the Brady plan, the 1990s initially featured high liquidity
and low interest rates, a marked change from the “credit rationing” of the previous decade
(Frenkel & Rapetti, 2012). Nevertheless, the nineties were a time of high financial volatility for
emerging economies. Rapid and unrestricted financial and commercial openings, exposure to
international liquidity cycles (and consequent increases in external debt), dismantling of labor
institutions via labor flexibilization policies, and significant real exchange rate appreciations led
to cyclical dynamics, which resulted in frequent financial crisis (Pastrana, Toledo & Villafañe,
6
2012). These included the Tequila crisis in Mexico in 1994, and the Asian and Russian crises in
1997 and 1998, the sizable Brazilian devaluation in 1999, and of course the Argentina crisis of
2001.
1.2 DOMESTIC MACROECONOMIC POLICIES
The Convertibility Plan
Following the economic stagnation, triple digit inflation, and massive balance of
payments crisis in the 1980s, in 1991 the Argentine peso was pegged to the US dollar at a ratio of
1:1. This marked the beginning of the Convertibility regime, in which the fixed exchange rate
functioned as the nominal anchor necessary to control inflation. The implementation of a fixed
nominal exchange rate (NER) was done concurrently with an almost complete liberalization of
trade flows and a full deregulation of the capital account (Frenkel & Rapetti, 2012). In addition
to this stabilization program, a variety of structural reforms including the privatization of state-
owned enterprises,2 modification of the pension system, and labor market flexibilization and
deregulation policies were implemented. Important labor contract regulation changes included a
reduction in social security contributions of employers, the authorization of part-time contracts
with lower firing and social security contributions, and a reduction in severance payments, among
other policies (Damill, Frenkel, & Maurizio, 2011). As a result of these widespread market-
oriented reforms for which the World Bank and the International Monetary Fund advocated
heavily, Argentina was dubbed the “poster child” of neoliberalism as embodied in the Washington
Consensus (Cohen, 2012). These bold actions successfully stabilized prices and combated high
inflation.
Nevertheless, under this exchange rate regime, Argentina became ever more dependent
on external capital, making it more prone to financial crises. Moreover, the inflexibility of the
2 Privatizations of public services were wide ranging and included water, electricity, transport, telecommunications, garbage collection, and mail services (Cohen, 2012).
7
currency board meant that different exogenous shocks were transmitted directly to the labor
market meaning. As a result volatility in external markets had direct ramifications on the local
credit market and domestic level activity, and consequently employment (Cohen, 2012); yet
policy space remained limited as the fixed exchange rate completely annulled the role of
exchange rate policy in correcting for an appreciated exchange rate (although the use of such
policy later proved to be key in Argentina’s macroeconomic configuration post-Convertibility)
(Maurizio, Perrot, & Villafañe, 2009). Space for countercyclical policy, e.g. expansionary
policies during moments of crisis, was also restricted due to the prioritization of debt reduction
(fiscal consolidation).
Argentina’s Macroeconomic Regime Post-Convertibility
The economics goals outlined by the Argentina government in the post-Convertibility
regime were three-fold: domestic financing for economic growth, re-industrialization (i.e.
supporting non-traditional sectors, ensuring profits for the manufacturing industry), and
employment generation (Pastana et al., 2012). The macroeconomic regime adopted after the
2001-2002 crisis differed fundamentally from the Convertibility regime in its preservation of a
stable and competitive real exchange rate (SCRER). While the Convertibility regime relied on a
fixed peg, the regime in the 2000s was based on a managed floating arrangement. This was
similar to the Chile’s crawling bands strategy used from 1984-1999, and aimed to combine short-
run NER volatility with the preservation of a SCRER in the medium run. Argentina´s Central
Bank followed quantitative money targeting in which targets were announced at the beginning of
every year and money aggregates acted as the nominal anchor. Intervention by the Central Bank
in the FX market (to control the NER and thereby maintain the SCRER) was done using
sterilization operations to prevent the money supply for expanding outside the target range. Such
a managed floating regime provided the same flexibility to absorb unexpected external shocks as
a floating regime, but intervention in the FX market made possible the important prevention of
8
distortionary effects of large capital inflows on the NER (Frenkel & Rapetti, 2012). This new
macroeconomic regime allowed for high levels of employment generation, resulting in sustained
falls in unemployment and underemployment. Unemployment decreased from 19.7% in the
fourth quarter of 2003 to 7.6% in the same period of 2008 (Abeles, 2012). Moreover, real wages
recovered, poverty declined, and the income distribution improved (Damill et al., 2011).
The post-crisis regime also contrasted with the Convertibility regime with respect to its
healthy dose of social policies. To deal with the skyrocketing unemployment due to the crisis, the
National Economic Emergency Law suspended unjustified layoffs for 180 days and doubled the
severance pay. Plan Jefes y Jefas de Hogar Desocupados (Program for the Unemployed Male and
Female Heads of Housholds) provided an income equal to 75% of minimum wage in 2002 to
about 2 million beneficiaries, equal to 11% of the active population. During the economic
recovery between July 2003 and September 2009, more than 600,000 beneficiaries found jobs in
the formal sector. By September 2009, the program covered less than 400,000 people and was
replaced at the end of 2009 by the cash-transfer program Asignación Universal por Hijo
(Universal Allowance per Child) (Damill et al., 2011).
Labor market policies were steered the opposite direction from the deregulation and
flexibilization trends of the nineties. The institutional configuration in the post-Convertibility
period reconstructed a space for collective and individual labor market relations, reinstigating
workplace inspections, promoting collective bargaining, and recovering state capacity to mediate
labor conflicts. The number of people in collective bargaining agreements rose from 3.5 million
in 2002 to 5.8 million in 2008, and the minimum wage rose by 520 percent between 2002 and
2009 (Abeles, 2012). A new labor law in 2004 reversed some of the severance pay reductions
passed in the 1990s and also simplified the labor registry system to promote the formalization of
workers (Damill et al., 2011).
9
1.2.2 GDP GROWTH TRAJECTORIES
The GDP trajectory of the 1990s followed a cyclical pattern which was repeated twice
during the decade; the first cycle took place between 1990 and 1995, resulting in a financial crisis
and recession; the second took place between 1996 and 2001, ending in the collapse of the
Convertibility regime. The stylized pattern is as follows. Initially, capital inflows spurred by
high interest rate differentials begin an expansionary phase. Foreign reserves are accumulated,
and domestic credit and aggregate demand expand. A RER appreciation follows due to increases
in prices and wages for non-tradables caused by demand pressures, and inertial inflation. The
RER appreciation and expansionary effects of capital inflows provoke a current account deficit
and growing external debt (Frenkel & Rapetti, 2012). The worsening of the current account
balance weakens the credibility of the exchange rate rule while the increase in external
vulnerability and probability of devaluation is reflected in a slowdown of capital inflows, and a
rise in domestic interest rates and the country risk premium. Reserve accumulation stops and a
contraction begins; higher interest rates and capital outflows give way to an illiquid financial
scenario ending in a financial crisis and possible collapse of the exchange rate regime as occurred
in 2001 (Frenkel & Rapetti, 2012; Damill et al., 2011).
The Argentine economy underwent expansionary growth periods between 1991 and 1994,
and 1996 to 1998. The recession in 1994-1995 was triggered by the Tequila crisis in Mexico,
10
which prompted a sudden stop of capital inflows to Argentina. With the help of a sizeable
financial assistance package by the International Monetary Fund (IMF), Argentina was able to
preserve the Convertibility regime in 1995, however economic activity contracted by 5.6% due to
capital outflows. The currency crisis in Brazil and the contagion from the Asian and Russian
crises in 1997-1998 triggered a prolonged recession in Argentina starting in 1998, which in
conjunction with an unsustainable external debt, culminated in financial and external crises in
2001-2002 (Frenkel & Rapetti, 2012). The collapse of the convertibility regime caused a
substantial 19.9% contraction in GDP and an increase in the unemployment rate to 21.5%,
leaving more than half of the population below the poverty line. This increase in poverty resulted
from the combination of the jump in prices immediately after the devaluation and the already
precarious previous social situation.
After the crisis, GDP continued to fall in the first quarter of 2002, stabilizing in the
second quarter, and increasing starting from the third quarter of 2002. Rapid and sustained
growth was maintained until 2008. Importantly, the post-Convertibility regime showed a marked
reduction in volatility compared to the cyclical nature of GDP in the nineties. In contrast with
international predictions that such a large crisis would essentially wipe Argentina off the map,
Argentina underwent one of its most successful growth episodes in its history (mid-2002 to mid-
2008) during which the economy grew at an average annual rate of 8.5%. By mid-2005, GDP
had already reached the maximum level achieved in 1998 (Maurizio, 2009) and for the first time
in modern history, Argentina boasted simultaneous external and fiscal surpluses for an extended
period (2003-2008) – an important contrast from the twin deficits experienced between 1994 and
2001(Frenkel & Rapetti, 2012). It should be noted that unlike in past growth episodes, a current
account surplus was maintained during the economy expansion, therefore growth did not come at
the expense of increased indebtedness. Nevertheless, the dynamism of output started to slow in
11
beginning in late 2007.3 Argentina’s country-risk premium increased and capital flight became an
“almost permanent feature of the Argentina balance of payments” from the third quarter of 2007
to the second quarter of 2010 (Pastana et al., 2012). Still, Argentina was in a relatively strong
position to face the global economic crisis in 2008-2009 given the sizable stock of international
reserves it had accumulated and the liquidity and solvency of its financial system. For the first
time in decades, such a significant external shock did not trigger a balance of payments crisis or
wreak havoc on the local financial system in Argentina (Abeles, 2012). The two channels of
influence were the financial channel and the trade channel, as private capital outflows increased
substantially and exports plummeted by 21% in 2009 (Damill et al., 2011). These effects are
evidenced by the drop in GDP growth rates in 2008-2009.
1.3.1 EXCHANGE RATE REGIMES
As previously mentioned, the nominal exchange rate was fixed at a ratio of 1 to 1 under
the Convertibility plan. The inability of the peso to adjust via devaluation eventually undermined
the sustainability of the currency board rule which was formally abandoned in January 2002,
however it is important to note that the real exchange rate was already significantly appreciated
when the nominal exchange rate was pegged to the dollar (Damill et al., 2011).
After the partial default on the public debt and collapse of the Convertibility exchange-
rate parity, the peso underwent a rapid devaluation reaching levels close to 4 pesos to 1 dollar.
3 Damill et al. (2011) attribute this in large part to the acceleration of inflation, a highly debated topic. According to the official statistics published by the United Nations Economic Commission on Latin American and the Caribbean, inflation has remained within the range of 7 to 10% since 2006. The Ministry of Labor, Employment and Social Security (Ministerio de trabajo, empleo y seguridad social) has made little reference to inflation. In fact, in its report reviewing the post-crisis period published for the bicentennial entitled Trabajo y empleo en el bicentenario: Cambio en la dinámica del empleo y la protección social para la inclusión, the topic of inflation is avoided altogether. Nevertheless, official inflation rates from the provinces denote worrisome signs of signs of accelerating inflation.
Frenkel and Rapetti (2012) propose that inflation was partly due a combination of rising commodity prices, an increase in wages beyond productivity growth, and an uncoordinated macroeconomic policy. While monetary and exchange rate policies were focused on maintaining an SCRER, fiscal policy was not used to moderate aggregate demand when inflationary pressures arose. Instead, public spending reached a level significantly above what was gained via increased tax revenues since 2006, serving to further boost aggregate demand.
12
The magnitude of this devaluation is evidenced by the sharp spike in the graph. July 2002 proved
to be an important turning point given the implementation of policies to stabilize prices after the
sharp devaluation and the imposition of a new exchange rate regime, both of which served to
reestablish macroeconomic equilibriums. Controls on foreign exchange were strengthened and
interventions in the exchange market began to form part of a systematic policy intended to
stabilize the foreign exchange market (Damill et al., 2011). After the exchange rate bubble was
stopped, a nominal appreciation began, which by mid 2003 the Central Bank began to counter,
thus beginning the new exchange rate policy.
A strong increase in the level of internal prices, for once not initially due to inflation,
followed the rapid depreciation. However unlike previous instances, this devaluation did not
result in an inflationary process due to i) weak domestic demand which limited the pass-through
to consumer prices, ii) high unemployment which eliminated potential wage indexation, iii) a lack
of liquidity due to bank deposit restrictions, iv) the government decision to freeze prices of public
utilities, and v) taxes on exports which put a ceiling on the domestic prices of exportable goods
(Maurizio, 2009; Damill et al., 2011). After its peak April 2002, the monthly inflation rate tended
to be lower than 1% for the following two years (Damill et al., 2011). The graph below plots the
real exchange rate deflated by wages from 1995 to 2010.
13
The exchange rate is a fundamental part of the macroeconomic regime due its wide-
ranging effects. Frenkel and Taylor (2006) outline five channels of influence: resource allocation
(via macroprice ratios), economic development (via an enhancement of competitiveness and
thereby a boost productivity and growth), finance (via control of expectations and behavior in
financial markets), external balance (via ‘substitution’ responses and changes in effective demand
which influence trade and consequently the current account), and inflation (an important
transmission mechanism for monetary policy effects). Out of the five areas of influence of the
RER listed by Frenkel and Taylor, the role of the exchange rate in resource allocation is most
relevant to this discussion of employment as the RER determines macroprice ratios between
tradable and non-tradable goods, capital goods and labor, and even exports and imports (via the
cost of intermediate inputs and capital goods, for example).
Frenkel and Taylor (2006) outline the destruction of employment brought about via an
RER appreciation as follows. Producers of importables will face tougher foreign competition and
cost cutting is often done by shedding labor. If they are forced to close down altogether, jobs will
be destroyed and non-tradable sectors will have to absorb labor displaced from the tradable
sector, however jobs are less likely to open up there, as cheaper foreign inputs (i.e. intermediates
and capital goods) are substituted for domestic labor. Appreciated RERs are also “an invitation to
disaster” as they can provoke destabilizing capital flow cycles that contribute to fiscal and
external crises. It must also be remembered that employment destruction happens much faster
and easier than employment generation. In contrast, a relatively weak RER can help boost
employment by increasing the profitability of the tradable sector and decreasing labor unit costs.
The expansion of exports reduces the balance of payments constraint to growth and also allows
for a systematic accumulation of FX reserves which serves to protect the economy from external
shocks (Damill & Frenkel, 2011).
As outlined by Frenkel and Taylor (2006), the exchange rate had an important negative
14
impact on the local production structure. Given that the fixed exchange rate prohibited corrective
depreciation, the resulting RER appreciation in conjunction with the radical opening to imports
obligated firms to become more competitive or shut down. Firms responded by increasing the
use of imported inputs, importing products previously produced domestically (as imports were
favored by the exchange rate), and substituting labor with capital – all measures aimed at
reducing costs and increasing productivity. The result was an intense loss of jobs in tradable
sectors, particularly manufacturing (Damill et al., 2011). Industrial sector employment fell from
28% of total employment in 1995 to 23% of total employment in 2000, amounting to the loss of
approximately 57 million manufacturing jobs (Castillo et al., 2002 as cited in Mazorra &
Becarria, 2007). The impact on manufacturing is important given that structural economists
(such as Hirshmann, Rosentein-Rodan, and Gerschenkron) cite the manufacturing sector as the
engine of development via its linkages with productivity increases [which via forward and
backward linkages, knowledge spillovers and technological externalities give way to increasing
returns] (Kaldor 1996 as cited in Cimoli, Novick, & Palomino, 2007). Pieper (2000) (as cited in
Cimoli et al., 2007) that low industrial sector growth results in few virtuous cycles as countries
remain on the track of low-road development in which there is a trade-off between productivity
growth and employment growth. Moreover short-term adjustments via unemployment, lower
salaries and reduced public spending can have long term effects on the “dynamic efficiency” of
the economy via hard to reverse destruction of human capital, thereby inhibiting future growth
possibilities (Van der Hoeven, 2000 as cited in Cimoli, Novick & Palomino, 2007).
1.3 MANIFESTATION IN THE LABOR MARKET
Having reviewed the exchange rate and output trends of the two regimes, the following
section presents their cumulative effects on employment. The main channels of transmission
from the macroeconomic regime to the labor market are via aggregate demand (and consequently
level of activity) and relative price ratios (RER, real wage, etc.) which determine labor utilization.
15
The Convertibility period was characterized by a low dynamism of aggregate
employment demand and high levels of unemployment. In general, unemployment rose, labor
market segmentation and informality increased, and wages and employment became more
volatile (Pastrana, et al., 2012). Both the direct transmission of exogenous shocks to labor
markets and the decrease in prices of capital goods prices relative to that of labor served to drive
down the elasticity of labor demand. The employment rate reached its peak for the decade in
1992, showing a marked fall afterwards down to its lowest point in 1996. Between the maximum
of 34.1% and the minimum observed in the second half of 1996, the employment rate fell
approximately 5.2% (Damill et al., 2011). As with GDP, the employment rate showed two clear
cycles.
Strong correlation between GDP and employment rate (R=0.8662)
Initially, capital inflows fed aggregate demand and the economy grew quickly. High
growth rates encouraged job creation, most intensely in non-tradable sectors. Nevertheless, the
trade opening and appreciated exchange rate deterred employment in the industrial sector which
explains why even the rapid GDP growth present in the early nineties was not accompanied by
increased labor demand (Maurizio, 2009). The low elasticity of labor demand (0.15) in
16
conjunction with an increased labor supply resulted in increased unemployment, which reached
double-digit levels even before the Mexican crisis of 1994. After the 1995 recession induced by
the Mexican crisis, the economy recovered and employment creation was in tandem with GDP
growth. The employment rate recovered by about 2.4%, reaching a new peak in the first semester
of 1998. The elasticity of labor demand increased to 0.62 which lead to a significant generation
of new jobs. Moreover, the participation rate began to stabilize which helped reduce
unemployment. However during this period, employment creation in the non-tradable sectors
was insufficient to counterbalance the negative effects in employment generation in the
manufacturing sector, thus neither employment nor unemployment rates reached pre-Tequila
levels (Damill et al., 2011). As shown in the previous graph, October 1998 to October 2001 was
a period of severe labor market and macroeconomic worsening. From 1998-2002, GDP
experienced strong contractive phases, dropping around 20% with 60% of this fall in 2001 alone
(Maurizio, 2009). The recession beginning in 1998 induced additional unemployment pressures.
It is important to note that during this period, the increase in unemployment was due fully to the
reduction in available positions, not the participation rate as the latter had actually decreased. At
the end of 2001, the open unemployment rate neared 20% (Maurizio et al., 2009).
Such instability has negative effects on the labor market as it signified a lower use of
productive capacity (restricting labor force potential) and an inferior productivity than in a stable
situation. In other words, liberalization increased financial activity without an increase in
national savings, leading to a very low investment rate in fixed capital and intense fluctuations in
economic activity and employment. Volatility gave way to financial and exchange rate crises
whose recessive effects discouraged the formation of national capital, and resulted in a
deteriorated labor market with increased formality (Ffrench-Davis, 2011). Thus while the
Convertibility plan succeeded at curbing four decades worth of inflation, it did not trigger
significant job growth. The instability caused by the volatility of macro variables – due to a
17
combination of external shocks, capital flows, and domestic policies – and deterioration of
capacity due to the appreciated exchange rate rather served to discourage employment growth
(Ffrench-Davis, 2011; Maurizio, 2009).
Between October 2001 and May 2002, there was a significant contraction in employment.
Between July 2001 and July 2002, according to INDEC 755,000 jobs were lost; by June 2002,
Argentina had the fourth highest unemployment rate in the world (Cohen, 2012). However by the
fourth quarter of 2002, employment growth was enough to stop the fall in net employment.
During the first years after the devaluation of the peso, the creation of formal jobs was so rapid it
even surpassed high GDP growth levels. The implementation and expansion of the Programa
Jefes de Hogar facilitated this rapid employment generation. Nevertheless, the program was
relatively short-lived and a reduction in beneficiaries was observed starting as early as the middle
of 2003, by which time the substantial number of jobs created in the private sector more than
compensated for such reduction. In October 2002, the employment rate (including employment
programs) was already higher than the last observation of the convertibility period one year
before. If the effects of the employment plan are excluded, the full recovery was completed by
the second quarter of 2003 (Damill et al., 2011).
Between the third quarter of 2003 and the same period in 2008, the employment rate rose
substantially from 38% to 42%. By the third quarter of 2002, the employment level had
surpassed than the 1998 level – the maximum level of the nineties. Likewise, the unemployment
rate fell by more than 50%, from 16.3% to 7.8% over the same period. Employment dynamism
started to weaken slowly since mid-2004, nevertheless the response of employment to the
domestic activity level remained very elastic. This is in direct contrast with the 90s in which high
levels of GDP growth were not necessarily accompanied by increasing employment levels
(Damill et al., 2011). As of late 2006, employment generation has stagnated as evidenced by the
relative flattening-out of the employment rate curve in the late 2000s.
18
It should be noted that the increase in jobs post-2003 was led by an increase in full-time
employment (78% of total employment in 2003, 90% at the end of 2008) (Damill et al., 2011).
The improved job quality (i.e. social security, health benefits, job safety, job stability, recognition
of collective bargaining agreements, job training and job mobility) was also an important change
(Cohen, 2012). With regard to informality and job precariousness – a key feature of the labor
market in the nineties – registered job positions explained 80% of new employment in the period
2003-2010, increasing by 53% while non-registered jobs only rose by 10% during the same
period (explaining 11% of job creation) (Damill et al., 2011). Nevertheless, precariousness
remains a problem as informal workers (equal to 43.1% of the workforce in the 4 th quarter of
2010) receive an hourly wage equivalent to only 40% that of registered workers.
Since 2003, commerce and construction together with industrial activities and financial
services have made the highest contribution to employment generation, explaining about 70% of
new employment (Maurizio, 2009). The manufacturing sector, to which the fixed exchange rate
and commercial opening were detrimental, was finally able to break out of the downward trend
experienced in the 1990s, growing at an average rate of 9.5% for 2003-2008. This is the highest
growth rate of manufacturing in the previous 60 years. Manufacturing increased its share in GDP
from 17.4% (1993-2001) to 21% (2003-2008), and employment in the sector grew by 42%
between 2002 and 2008 (Abeles, 2012; Cohen, 2012). Nevertheless due to the sustained
contraction of the nineties, employment in manufacturing activities remained 22% lower in 2008
than it was at the beginning of the nineties.
The negative effects on labor demand due to the global economic crisis further served to
weaken employment generation in the late 2000s, however government-sponsored programs such
as the Productive Recovery program (REPRO4) that preserved some jobs in the formal sector and
4 This program established government-provided subsidies to companies (government pays part of workers’ wages to maintain employment levels). In 2009, the program covered 143,000 workers (equivalent to 1 percentage point of the unemployment rate) belonging to 2750 companies (Damill et al,
19
an increase in public employment have helped to mitigate falls in employment rates (Damill et
al., 2011). Unemployment rates increased and employment rates fell as a result of the global
financial crisis in 2008-2009. Nevertheless due to its growth strategy which maintained the
competitiveness of Argentine products and therefore allowed for sizable reserve accumulation,
Argentina confronted this crisis with much greater ability and space to implement countercyclical
policy (Pastana et al., 2012). As a result rises in unemployment were nowhere near the
magnitude of that experienced in the 2001 crisis.
1.4 CONCLUSION
The chart below presents the averages for the real exchange rate level and GDP growth in
the two periods discussed. On average GDP growth was much greater in the post-Convertibility
regime, which was facilitated by a depreciated real exchange rate. This boosted the
competitiveness of the tradable sector and allowed for greater employment generation.
Convertibility Post-Convertibility1995-2001 2002-2010
RER 1.331 2.314GDP growth 0.896 5.576
Overall, the economic regime in the 90s with its single focus on nominal stability and full
reliance on market mechanisms to promote growth and employment (rapid trade opening,
currency appreciation and structural reforms) negatively affected job creation. Employment
generation stagnated even before the first recession in 1994. The manufacturing industry was hit
particularly hard as the change in relative prices due to the appreciated RER prompted a
substitution of labor with capital, leading to the destruction of many jobs and firms. The deep
crisis in 2001 caused a severe contraction in the level of domestic activity, with lasting negative
2011).
20
impacts on employment and income. In contrast, 3 million jobs have been created since 2003 and
2010. The maintenance of the SCRER has led to the expansion of the whole tradable sector,
turning the domestic/internal market into engine of growth.5,6 As a result the economy has
exhibited high GDP growth rates and an elastic labor demand. Positive trade balances have led to
the accumulation of voluminous amounts of reserves and a reversal of current account deficits to
surpluses, thereby providing greater room for countercyclical policies that have served to mitigate
the transmission of external shocks to the labor market. Nevertheless, concerns over inflation
have prompted many to argue that the macroeconomic regime has been losing coherence since
2006. Due to the global financial crisis in 2008 and the recent RER appreciation trend,
employment generation in the late 2000s has lost its former dynamism. Still in terms of
employment generation, the macroeconomic setting in the 2000s remains a marked improvement
over the cyclical boom-and-bust nature of employment generation in the 1990s.
5 Other important factors commonly cited in explaining Argentina’s recovery include the renegotiation of the debt and the re-institutionalization of labor policies. 76% of the debt was accepted in default, which reduced the public debt to US$ 63,700 million, while 40% of new titles were expressed in national money which was significantly more favorable to Argentina given the depreciation. By the end of 2005, Argentina was able to play the full debt of US$ 10,000 million to the IMF (Maurizio et al, 2009). Labor policies such as the minimum wage, and collective bargaining facilitated an improvement in the income distribution, which was a key factor in the growth of domestic demand (Ministerio de Trabajo, Empleo y Seguridad Social, 2011).
6 While commodity prices certainly played a role in Argentina’s dynamic growth, Frenkel and Rapetti (2012) argue it was the expansion of the whole tradable sector, facilitated by the SCRER, that put the economy on the path towards rapid growth. Maurizio et al. concur that the initial recovery was demand led (i.e. due to domestic factors), which was reinforced by an increase in export prices and a decrease in international interest rates.
21
2.1 URBAN FACTORS IN EMPLOYMENT GENERATION
While the macroeconomic regime is of fundamental importance, this paper argues that it
is not the only factor. Rather, employment generation is also contingent upon urban factors such
as size and density. This is evidenced by the heterogeneity of employment rates in various cities
as illustrated in the graph below.
Ideally, this graph would show elasticities of employment as employment rates show a static
picture rather than a dynamic ability of larger cities to generate more employment under a given
macroeconomic regime. Nevertheless, elasticity of employment graphs by city size when not
estimated econometrically include much noise such as widespread privatizations in the 1990s as
well as the effects of other
variables such as the
production structure. Still,
in a snapshot view presented
in the accompanying scatter
plot, the positive relationship
between size and elasticity
of employment is noticeable.
22
2.1.1 SIZE
Why is it that larger cities have shown higher levels of employment? The benefits of city
size stem from economies of agglomeration. This is the notion that “spatial proximity of
activities makes resources more efficient than if such activities are spatially dispersed” (Goldstein
& Gronberg, 1984); this entails an outward shift in the production function as industry size
increases. Unlike economies of scale which are internal to the firm (large fixed costs and smaller
variable costs that create increasing returns to scale at the firm level), agglomeration economies
have been termed “external economies of scale” and are most relevant to this discussion of
employment. Quigley (1998) explains, “The economies are external in the sense that the firm
obtains them from outsiders, and they are economies in the sense that the firm can satisfy its
variable or part-time needs in this manner more cheaply than it could satisfy them from within.
The outsider, in turn, can afford to cater to the firm’s fractional needs because he also caters to
many other firms.” Marshall is arguably the best known and most influential of the early analysts
of agglomeration and is credited with the exposition of the concept of external economies.
Nevertheless Smith (1776) can be credited with the first analysis of the benefits from
agglomeration, albeit with a more narrow argument focusing on the division of labor (Duranton
& Puga, 2003).
There are two types of external economies of scale: localization and urbanization
economies. Localization economies refer to economies of scale arising from the spatial
concentration of activity within industries; they are external to a firm at a given location but
internal to the industry at that location (Goldstein & Gronberg, 1984; Rosenthal & Strange,
2001). Urbanization economies accrue to a firm from the level of overall economic activity in the
area, and are therefore external to the firm and to the specific industry at a given location
(Nakamura, 1985). They reflect the benefits from operating in a large, urban environment in
which there is a large overall labor market, and a large service sector interacting with the
23
manufacturing sector (Henderson, 1986). Using case studies of Brazil and the US, Henderson
finds much stronger evidence of the productivity effects of localization economies than
urbanization ones, but nevertheless concludes that urbanization remain important for city growth
(Murphy et al., 1989 and Krugman, 1991 as cited in Glaeser, Kallal & Scheinkman, 1992).7 In
this study, I include density as a third factor (the first two being spatial concentration within an
industry and the overall level of economic activity in the area) which should facilitate greater
employment generation in urban areas. While similar to localization economies, the benefits of
density arise from the spatial concentration of people and activity within an area as opposed to
only within an industry. Density is defined here as the intensity of labor and capital relative to
physical space. The following section outlines the specific links between localization and
urbanization economies, and density on employment generation.
Localization economies
Examples of localization economies include increased communication between firms,
decreased search costs, greater specialization, and scale in public intermediate inputs (Henderson,
1986). With localization economies, firms benefit from a closer flow of information to suppliers
and a proximity to final customers, therefore choosing to locate in a city in part because demand
is higher there (Dumais, Ellison & Glaeser, 2002). It should be noted that while producers prefer
to concentrate to economize on transport costs and prefer production sites that are close to large
markets, markets are large precisely where large numbers of producers have chosen to site their
facilities. Thus the location of production and location of demand are interdependent (Krugman,
1991), which suggests self-sustaining growth past a certain threshold.
A larger labor market reduces search costs and increases the probability of matches
between differentiated skilled workers and differentiated job requirements (Hesley & Strange,
7 The existence and considerable magnitude of localized scale externalities is well documented empirically, see Henderson, 1988; Ciccone and Hall, 1995; and Glaeser et al, 1992, as cited in Henderson, 2000.
24
1990). Concentration of several firms in a single location offers a pooled labor market for
workers with industry specific skills (Krugman, 1991).8 This ready availability of workers for
firms alleviates hold-up problems, as asset specificity is less likely to be an issue where the
number of potential partners is large (Duranton & Puga, 2003). Proximity to specialized labor
such as workers in accounting, law, advertising, and other technical fields, can also reduce costs
for businesses (Quigley, 1998).
In addition, greater industry size permits greater specialization among firms (Hesley &
Strange, 1990). This is illustrated by Adam Smith’s famous pin factory example in which an
expansion in a firm’s workforce increases output more than proportionately because it allows
existing workers to specialize. There are resulting productivity gains due to workers’ increased
dexterity at a particular task, savings on fixed costs by not having workers switch tasks, and
labor-saving innovations made possible by simpler tasks that can more easily be mechanized
(Duranton & Puga, 2003). Given that the division of labor is limited by the extent of the market,
which is determined by transportation efficiency, cities are the “natural” market where division of
labor takes place (Duranton & Jayet, 2011).
Concentrating industrialization also allows the economy to conserve on “economic
infrastructure” that can be shared among firms. Examples include physical infrastructure capital
such as transport and telecommunication – inputs that are not costlessly mobile – and managerial
resources (Henderson, 2000; Glaeser, Kallal & Scheinkman, 1992). The ability to share
transportation infrastructure is able to transform production with constant returns to scale to one
with increasing returns to scale via transport costs that rise with distance. Firms therefore save on
transport costs both due to proximity to input suppliers, and due to the ability to sell some of their
output without incurring transport costs.
8 As agglomeration economies improve the expected quality of matches between job requirements and skills of workers for all workers, there is a positive ex-ante relationship between wages, productivity, and city size (Hesley & Strange, 1990).
25
Urbanization economies
As previously mentioned, urbanization economies reflect the benefits of operating in a
large, urban environment. These are particularly relevant as giving the rise in large cities, as
illustrated by the accompanying map. Currently, one in three Latin Americans lives in a city of 1
million or more (CEPAL, 2012).
Large markets facilitate the existence of highly specialized products. As Adam Smith
initially explained using the example of a porter, there are various specialized businesses and
consumer services for which the per business or per capita demand is so small that a large city is
needed to support even a few suppliers (Mills, 1967). Thus, not only are there gains from
narrower specialization sustained by larger production, but there are also gains from a wider
variety of input suppliers that can be sustained by a larger final-goods industry (Duranton &
Puga, 2003). In general it is found that a higher degree of differentiated intermediate inputs leads
26
to a higher productivity in the production of final goods (Ciccone & Hall, 1993). Thus large
cities provide public intermediate inputs that can be tailored to the needs of a particular industry
(Hesley & Strange, 1990), such as specialized local producer services like repair and
maintenance, legal support, transportation, and financial resources (Abdel-Rahman & Fujita,
2006).
Additionally, large urban markets are likely to smooth employment fluctuations, giving
way to more stable aggregate level employment.
Mills argues that the most important aspect of agglomerative economies “is statistical in nature, and is an application of the law of large numbers. Sales of output and purchases of inputs fluctuate in many firms and industries for random, seasonal, cyclical, and secular reasons. To the extent that fluctuations are imperfectly correlated among employers, an urban area with many employers can provide more nearly full employment of its labor force than can an urban area with few employers” (Mills, 1967 cited in Goldestin & Gronberg, 1984).
Thus, employment can be stabilized as some firms are hiring while others are not. This extends
to sales of output as well meaning firms can carry less inventory (Quigely, 1998), which amounts
to real savings for businesses. Those producing specialized products subject to widely fluctuating
demand but more stable industry demand will also benefit from agglomeration (Glaeser, Kallal &
Scheinkman, 1992).
Density
Arguably the most important benefit from density relates to the diffusion of knowledge.
Knowledge spillovers are particularly effective in cities as communication between people is
more extensive. Jacobs argued that the cramming of individuals, occupations, and industries into
close quarters provides an environment in which ideas flow quickly between people, and that
these interactions between people in cities facilitate learning and innovation (Glaeser, Kallal &
Scheinkman, 1992). With respect to firms, the greater “communication” enables the adoption of
new technologies/innovations. These informational spillovers therefore give clustered firms a
27
better production function than isolated producers (Krugman, 1991).
While innovation has been found to be lower in low-dense areas (Carlino, Chatterjee &
Hunt, 2006 ), various studies have illustrated a positive link between innovation and employment.
Novick et al. (2009), Roitter, Erbes & Trajtenberg (2010), and Robert et al. (2010) as cited in
Barletta et al. (2012) have found that those firms which have generated the most employment in
the upward phase of the cycle and lost the least in downswings were firms with high levels of
technology and organization capacity. Barletta et al. (2012) confirms these results, finding that
firms with complex innovation strategies and links to other firms and services with the goal of
developing technical and organizational capacity have had a superior performance in terms of
employment. Innovative activities are defined by Schumpeter as those relating to the creation
and improvement of products and processes, and organizational change (Novick et al, 2009 as
cited in Barletta, Castillo & Yoguel, 2012).
In summary, agglomeration economies arising from size and density engender a plethora
of benefits: reduced search costs, greater specialization, scale in intermediate inputs, more stable
aggregate employment, and increased knowledge and innovation. All of these suggest that size
and density give larger cities an increased ability to generate employment under a given
macroeconomic regime. However, it is worth noting that there is not a one-size-fits-all optimal
city size. Rather, the efficient city size depends on the particular set of interrelated activities and
goods produced within its bounds, otherwise various city-sizes would not exist in a market
economy (Goldstein & Gronberg, 1984). As Henderson (1986) explains, “if these goods involve
different degrees of scale economies, cities will be of different sizes because they can support
different levels of commuting and congestion costs.” In general, the equilibrium city size will
increase as the degree of localization economies of the industry the city is specialized in increase,
nevertheless there is a limit to benefits from size as will be discussed in the policy implications
section.
28
3.1 PRODUCTION STRUCTURE AS A DETERMINANT OF EMPLLOYMENT
GENERATION
The production structure also has important consequences on employment generation. In
general, regions with a lower productive diversification are those that registered a larger negative
impact in terms of production and employment in the recessive phase at the end of the 1990s and
a slower recovery in the subsequent upwards cycle (Tomada, 2007). The role of the production
structure has predominantly been the domain of structuralist economics (Hirschmann, Rosentein-
Rodan, Gerschenkron, Chenery y Sirquin) which argues that development implies the
reassignment of factors (capital and labor) from low-productivity sectors to high-productivity
sectors, leading to productive convergence. Productive structure heterogeneity therefore refers to
large productivity differences both between and within economic sectors. This produces a
segmentation of the production system and labor markets, with asymmetrical technological
conditions and wages (Porcile, 2011). Structural heterogeneity stems from a variety of factors,
including unequal access to technology, the structural organization of firms, the nature of the
product's demand, and unionization (Cimolo, Novick & Palomino, 2007). Prebisch famously
outlined the problem of slow and unequal technological diffusion, in which technological
progress is mainly absorbed only by sectors related to exports while other sectors remain
unchanged (Porcile, 2011). Unfortunately many sectors with high productivity such as those
related to natural resources (ex. copper) account for a relatively small quantity of total
employment. Moreover, the expansion of natural resource production generates limited positive
externalities given the large reliance on imported products, thus presenting less opportunities for
contracting out, learning processes, etc.
Structural heterogeneity reflects negatively on employment generation. With high levels
of structural heterogeneity, externalities and positive spillovers cannot be as widespread given the
absence of linkages between sectors. Investment levels also tend to be low and incentives for
29
productivity are weak. Moreover, economies with high levels of structural heterogeneity have a
lower capacity to respond to demand changes via diminished innovation. This implies a loss of
market share and less competitive exports, which taken farther, can pose risks for exchange rate
crises due to worsening of the trade balance, and can lead to more volatile growth patterns
(CEPAL, 2010). Moreover, structural heterogeneity consequently leads to the creation of
“enclave sectors” in which only the formal sector benefits, as it bases its modernization on the
adoption of innovations from the exterior thereby inhibiting endogenous development of
technological capacity. With this production structure, productivity increases do not lead to
higher wages and greater employment, but rather lead to the expulsion of workers into
informality.
Over the past 50 years, the high productivity share of firms has increased its contribution
weight to GDP without a corresponding increase in employment. Due to the insufficient growth
of the intermediate strata in terms of production and employment, the majority of the labor force
has been absorbed by lower productivity sectors, which register a negligible contribution to GDP
(Porcile, 2011). Importantly, structural heterogeneity is reproduced even under conditions of
rapid economic growth as high-productivity firms drive the economy (Infante & Sunkel, 2009).
Crises are thus likely to hit hardest on the low-productivity sectors, which do not greatly
influence growth yet absorb the bulk of the workforce, causing rises in unemployment and
informality.
30
4.1 INFRASTRUCTURE
Finally, it is worth spending a moment on infrastructure. While the exact impact of
infrastructure on employment is hard to measure, this paper argues that infrastructure is likely to
act as “grease on the wheels,” facilitating the benefits of size, density, and a complex production
structure on employment generation. Infrastructure’s relationship with employment is therefore
mediated through its ability to elevate productivity and lower production costs. I use
infrastructure in this paper to refer to “economic infrastructure,” namely public services (i.e.
energy, telecommunications, water and sanitation, solid waste management), public works (i.e.
highways, drainage, etc.), and transportation (i.e. railways, urban transportation systems, ports,
airports).
With respect to urban factors, infrastructure can play an important role in the realization
of benefits of agglomeration economies. For example the benefits of shared transport costs
cannot be realized without an adequate highway and/or railway system. Increased
telecommunications infrastructure decreases search costs and increases the flow of information.
In general, increased infrastructure provision decreases costs for businesses which facilitates
increases in employment. With respect to the productive structure, a lack of urban services and
spatial segregation can give way to vicious cycles of production segregation in which large
proportions of the population remain in areas of low productivity. Low productivity sectors have
a harder time innovating, adopting new technology, and inciting learning processes (CEPAL,
2010). The presence of low productivity sectors then feeds back negatively on infrastructure
provision as the concentration of informal (or less dynamic) activities in low-income areas
reduces the land value which reduces the income from taxes that the municipal government
receives, thereby limiting their ability to provide public services and invest in infrastructure
(CEPAL, 2010). In a more direct relationship, the formation of physical capital generates
employment in the short-term. Moreover benefits extend to the long-term as well, as increased
31
capital formation provides/permits permanent occupations in the higher level of production of
goods and services it sustains. In this way, infrastructure provision may act as a way of reducing
structural heterogeneity.
While infrastructure may play the role of a helping hand in economic growth, it is also
likely to be an effect of increased economic growth as increased tax revenues allow for the
realization of more infrastructure projects.9 According to the 1994 World Development Report, a
one percent increase in infrastructure capital is associated with a one percent increase in GDP
(World Bank, 1994). For example, a quick look at water access data for the period 2003 to 2010
as measured by the EPH household databases (access to an in-house water source),10 shows that
all cities with the exceptions of Gran Parana, Rio Gallegos and Ushuaia-Rio Grande exhibit
improvements in water access. The decreases in these three cases are minimal: 1.1% or less.
However such positive trends are to be expected given the high levels of GDP growth and
employment growth experienced in the 2000s. Given that increases in infrastructure provision
are likely to occur in times of economic growth, and economic growth is often correlated with
employment generation, it is hard to clarify/distill out the relationship between infrastructure and
employment. Furthermore, infrastructure is difficult to measure. While some inventory stock
data exists - such kilometers of roads, number of airports and ports, etc. – the benefits of
infrastructure are not always realized in the same area where the infrastructure is located. A
bridge for example may have a larger impact on the towns they connect, rather than the area in
which it was built. Due to the complexity, multidirectionality, and measurement difficulty of
infrastructure, this study recognizes the indirect effects of infrastructure via the production
9 Nevertheless, even in periods of growth in the 1990s, infrastructure spending was notoriously low due to the fiscal discipline imposed by the IMF. Moreover, the widespread privatizations weakened the national government’s capacity to provide basic services. During the most vigilant Washington Consensus years, the investment rate was near the lowest rate registered in the lost decade of the 1980s (Ffrench-Davis, 2011).
10 As this question is only included in the household, not the individual survey, water access data is not included in the micro model estimation of the probability of being employed.
32
structure, growth, and urban factors but does not include infrastructure as a direct variable in the
models that follow. In general, such linkages between infrastructure and employment are
complex and difficult to trace, nevertheless they are an important area of future research.
33
5.1 ECONOMETRIC MODELS
This study uses two models to measure the respective weights of macroeconomic
regimes, the production structure, size, and density on employment. The first model is a “macro”
level model which estimates aggregate labor demand. The second model is a “micro” level
model which estimates the probability of being employed. The following section presents the
design and results of each of the two models.
5.1.1 Macro model
The macro model uses time series data to estimate aggregate labor demand at they city
level. The number of employed workers is used as the dependent variable, and the explanatory
(independent) variables include gross domestic product, size, complexity, density, wages, and
region. The model is run twice, once for the period October 1995 to May 2001, and the second
for the period 2003 to 2010. This distinction is necessitated by the methodological change in the
household survey (EPH) in 2003, yet is also useful given the very distinct nature of the two
macroeconomic regimes.
5.1.2 Variable descriptions
NUMBER OF EMPLOYED PERSONS
This variable was calculated using EPH city-level estimates of the total number of
employed people. After re-categorizing people participating in employment plans as
unemployed, employment rates were re-calculated and then multiplied by the population
estimates to get the total number of employed workers. Quarterly population estimates were
calculated using the 2010 4th quarter urban agglomeration population totals from the EPH
database, and applying the provincial population growth rate published by the census agency (one
growth rate for 2010-2001, and a second from 2000-1995). While an imperfect measure of
34
population as provincial growth rates may differ from city growth rates, it was more important to
avoid short-term fluctuations present in the EPH quarterly population data stemming from the
expansion of sample results to population estimates. This variable enters into the regression in
log form to measure the change in employment over time. As with all variables based on data
from the EPH, data reflects October and May values for the period 1995 to 2001, and quarterly
values from 2003 to 2010.
TAX COLLECTION
Ideally gross domestic product would be used to proxy the effects of the macroeconomic
regime. However, such a measure is not available on a city nor provincial level for the entire
duration of the series. National GDP cannot be used in this regression given that it is attempting
to account for differences between cities (thus putting the same values for all cities will explain
little). Total tax collection by province is used instead. Gross geographic product does exist by
province until 2005 which
allows a comparison of gross
geographic product with the
tax collection proxy. When
tested in the 1995-2001
regression, the coefficient for
tax collection is only 0.03
lower than that of gross
geographic product and the
adjusted R-square change
falls minimally (by 0.0017).
This shows that tax
collection is an excellent
35
proxy of GDP. This variable enters into the regression equation in log form to reflect growth
rather than level effects.
SIZE
Argentina has a very unequal distribution of city size, as illustrated by the graph on the
previous page. Apart from the metropolitan area of Buenos Aires, most cities in Argentina are on
the order of 300,000 to 400,000 inhabitants. Population estimates are taken from the 4th quarter
of the 2010 EPH.
Size enters the regression as a dichotomous variable, reflecting large, medium, and small
cities. To create the variable, I divide the urban agglomerations into three categories based on a
“cumulative percent approach” in which category division lines are drawn at 80, 90 and 100
percent11. This is illustrated below. Such an approach illustrates how the population is
distributed. In other words, out of every 80 people that are born, 80 will reside in large cities, 10
will reside in medium ones, and 10 in small ones. Appendix 1 presents more detailed information.
The resulting categorization is as follows:
11 An exception is made for Sante Fe, which is taken in the large city category due to intuition, given that it is the capital city of the province.
36
Large: Área metropolitana de Buenos Aires, Gran Córdoba, Gran Rosario, Gran Mendoza, Gran Tucumán-Tafí Viejo, Gran La Plata, Mar del Plata-Batán, Salta, Gran Santa Fe
Medium: Gran San Juan, Gran Resistencia, Santiago del Estero-La Banda, Corrientes, Bahía Blanca-Cerri, Jujuy-Palpalá
Small: Posadas, Gran Paraná, Neuquén-Plottier, Formosa, Gran Catamarca, San Luis-El Chorrillo, La Rioja, Río Cuarto, Concordia, Comodoro Rivadavia-Rada Tilly, Ushuaia-Río Grande, Santa Rosa-Toay, Río Gallegos
While some may argue that there are only three truly large cities in Argentina (AMBA,
Córdoba and Rosario), including more cities in this first group is important for the regression.
With only three large cities, density as a dichotomous variable becomes highly correlated with the
size dichotomous variables. With nine large cities, there exists heterogeneity with respect to
density within the category, thereby allowing the model to differentiate between the effects of size
and density.
DENSITY
The series on density data is unique to this study as there are no series published by
INDEC which measure density on a city level. To create this series, I calculated size in
kilometers squared of each city by summing the corresponding “departamento” area data
published on the 1991 census page and divided them by quarterly population estimates
(calculated as explained in the number of employed section), to give a measure of inhabitants per
kilometer squared. This is then converted into a dichotomous variable by giving the value of 1 to
cities in which density is greater or equal to the median density for that quarter or year, and 0 if it
is below. Appendix 2 presents density in the year 2005 ranking cities from most to least dense.
COMPLEXITY
To measure the effects of production structure, a complexity indicator was constructed
based on the proportion of city-level employment in sectors with high value added compared to
national employment in such sectors. This indicator not only shows which cities have a more
37
complex production structure than others, but also how each city compares to the national
average as cities with a complexity indicator value greater than 1 are more complex than the
national average. This variable is not constructed as a time series, but rather constructed on the
basis of 2005 data. This is not problematic though as the production structure is not something
that undergoes short-term fluctuations as evidenced by the persistence of structural heterogeneity
over the past 50 years. Employment data comes from the year 2005 as it is a relatively neutral
year, avoiding potential distortionary effects of the crisis years and widespread privatizations in
the 1990s. Appendix 3 presents the equation and indicator values.
REGION
Regional dummy variables are included to control for inter-regional differences based on
the relationship between sectoral production and macroeconomic regime. Agricultural,
manufacturing, and tourism occur in relatively distinct parts of the country, thus it is important to
recognize that which goods and services are favored by the macroeconomic regime and external
conditions (such as high commodity prices) will partly explain different levels of employment
generation. The regions are: Gran Buenos Aires (1), Cuyo (2), Noreste (3), Noroeste (4),
Pampeana (5) and Patagónica (6).
MODEL LIMITATIONS: Excluded Employment Determinants
This model does not seek to incorporate all determinants of employment. Many previous
studies have been embarked on exploratory analyses of the effect of the size of firms - finding the
while small and medium enterprises are fundamental in explaining employment generation, the
majority of employment remains explained by large firms12 - and sector - finding that
employment is inversely related with primary goods production. Local policies also play a key
role in explaining heterogeneous city employment rates (Rafeala is a prime example of this).
12 For more see, Birch, D. L. (1981). Who Creates Jobs? The Public Interest,65 (Fall), 3-14.
38
While further discussion of these themes is outside the scope of this paper, it suffices to say that
some of the unexplained variance in the model is due to such excluded variables.13
The regression equation for the macro model is therefore:
Employed workers = α + β1*tax collection + β2*large + β3*medium + β4*complexity +
β5*density
5.1.3 Macro model results
In the 1990s macro model (1995-2001), all variables are significant at the 1% level
except for density which is statistically significant at the 5% level. The macroeconomic effect is
very important, with a 13% increase in tax collection associated with a 1% increase in
employment. In addition, size and complexity are both significant predictors of employment
generation. In large cities, aggregate labor demand is 65% higher than in small cities. Likewise,
employment demand is 28% higher in medium cities in comparison with small cities. Density
shows a positive coefficient, however its effect is markedly smaller than all other variables. The
model has a good fit, with an adjusted R2 value of 0.74, meaning that 74% of the variance in
number of employed workers is explained by the variables in the model.
13 This model also attempted to include a proxy of wages in order to avoid overestimating the impact of GDP. The proxy was created as follows. The average wage for each city was taken from the EPH microdata and then for each quarter or semester (1990s and 2000s model, respectively), divided by the average/median wage for that quarter or semester. Results greater than 1 were given the value of 1, and results less than 1 were given a value of 0. The creation of a dichotomous variable was important in order to avoid problems with inflation which would occur if the variable was used as a time-series. This was a better option than dividing each city-level wage by the median value calculated across all periods, because the large between-city differences and changes over time forced the majority of values in the first half of the series to be zero, and those in the second half to take the value of one; this would eliminate variation between cities within a given quarter or semester. Thus, this vertical calculation was chosen instead to give an idea of where wages were the highest.
Nevertheless, this measure had many limitations. Most importantly, businesses choose to contract labor where labor is cheap on a sectoral level. In other words, it does not matter if average wages are higher or lower in Bahia-Blanca than Mendoza, but rather how the wages compare for the specific sector the worker is in. Secondly, this estimate includes wages for the informal sector which biases the wage estimate downwards. This is a likely reason why wages and the complexity indicator were not collinear. In the end, wage proxy turns out to have near zero explanatory power and is therefore excluded from the model.
39
Due to collinearity between complexity and region, I ran a second regression including
regional dummies and eliminating the complexity indicator. This model has additional
explanatory power, with an adjusted R2 value of 0.92, however it is important to note that the
regional dummies likely hide other omitted variables. Size continues to be the most important
variable, followed by complexity, tax collection, and lastly, density. The size of the coefficients
for tax collection, size, complexity, and density are slightly reduced, however that is likely due to
some correlation with the regional dummies (i.e. some of their effect is captured instead by the
regional dummy). When Patagonia is taken as the base region, all regional dummies show a
positive value, meaning that employment in every other region is higher than employment in
Patagonia. Region 1 has the largest value, which is to be expected given high rates of
employment in Buenos Aires.
When the model is run for the 2000s (2003-2010), the only difference is that density
becomes significant at the 10% rather than the 5% level. Significance tables for all regressions
can be found in Appendix 4.
5.2.1 MICRO MODEL
The micro model estimates the probability of being employed to test the effects of
structural and urban determinants on employment while controlling for demographic
characteristics of workers. It should be noted that the findings in this section are very preliminary
and should be treated with caution. This model works directly with the EPH databases
(individuals).
5.2.2 Variable descriptions
The dependent variable in this model is a dichotomous one, “estadon,” which takes the
value of 1 for employed persons within the active age and 0 for unemployed persons within the
active age. The active age range is defined as 15 to 65 for men and 15 to 60 for women. The
40
independent variables include gender, education, age, and an interaction between gender and head
of household. Education enters into the regression equation as five dichotomous variables:
incomplete primary education, incomplete secondary education, completed secondary education,
incomplete tertiary education and completed tertiary education. Age is broken up into there
groups, 15-24, 25-49, and 50-65 (60 for women). The interaction between gender and head of
household is important women are more likely to work if they are the head of the household as
well. I also include a dummy variable for public sector employment, given the lower volatility of
the public sector, often used as a countercyclical mechanism. In addition to these demographic
variables, I include the variables size, complexity, and density defined the same way as in the
macro model. The regression equation is therefore:
Probability of being employed = α + β1*gender + β2*ip + β3*is + β4*cs + β5*it + β6*ct +
β7*age2 + β8*age3 + β9*gen*headhousehold + β10*large + β11*medium + β12*publicsector +
β13*density
5.2.3 Micro model results
As expected, all of the demographic variables are significant. Age is the most important
(a greater age meaning a greater probability of being employed), followed by gender, education,
and status as head of household. Production complexity continues to be a highly important
variable with respect to important, showing that a more complex production structure is
associated with a higher probability of being employed. Density in this case is statistically
significant. In contrast with the other model, the coefficient is negative. Although statistically
significant it appears to have little effect in reality given the very small magnitude of the
coefficient.
Both of the dichotomous size variables take a positive value, however only “large” not
“medium” is significant, meaning that once demographic characteristics are controlled for, the
41
benefits of size are not felt in medium-sized cities. This could suggest that the macro results
reflect different compositions of employment (larger cities having more people that have a higher
probability of being employed) rather than an endogenous capability of large cities to generate
employment. However, this is a very preliminary result and requires much additional testing.
The significance table for the second quarter of 2006 is presented in Appendix 4.
In general, the model has a good fit (R2 equals 0.45). However, it is important to mention
that the non-demographic variables add very little to the model in terms of explanatory power. In
the absence of size, complexity, and density, the R2 is only 0.0011 lower. Moreover, the
significance of the complexity and size variables greatly depend on the year used from the EPH.
Thus, the results of this model are not robust and should be taken as an invitation for future
research rather than as conclusive results.
42
6.1 POLICY IMPLICATIONS
Macroeconomic regimes
This study has illustrated the fundamental importance of macroeconomic regimes with
respect to employment generation. As evidenced by the 1990s, economic growth by itself is a
necessary but insufficient condition for employment generation. The overvalued peso during the
Convertibility plan reduced the competitiveness of Argentine exports, undermined the domestic
economy, and contributed to the vulnerability of the economy to external shocks, all over which
led to low levels of employment generation (Cohen, 2012). In contrast, the post-Convertibility
regime illustrates the positive effects of a stable and competitive real exchange rate which
boosted the profitability of tradable goods and incentivized industrialization, and was
accompanied by an array of social and labor policies which encouraged the formalization of
workers and supported those with low incomes. The exchange rate is arguably the most
important transmission mechanism between the macroeconomic model and the labor market, thus
it must be managed with care. The problem with the fixed exchange rate in the 1990s was not
necessarily that it was fixed (although that became problematic in that it prevented adjustment to
changing external circumstances incited by crises in Mexico and Russia and the devaluation in
Brazil), but rather that it was fixed at an appreciated rate. This gave way to a profound
deindustrialization as businesses substituted capital for labor and many manufacturing firms were
forced to shut down. The competitive real exchange rate on the other hand, in conjunction with
fiscal policies and social policies of redistribution, increased the profitability of the
manufacturing sector and turned the domestic market into the engine of growth.14 This reduced
volatility as the economy became less susceptible to external shocks, and the improving trade
balance via the increased competitiveness of exports allowed for a significant accumulation of
reserves which relaxed the balance of payment constraint and created more fiscal policy space to
14 A devalued exchange rate can also help reduce structural heterogeneity via an enhancement of the profitability of non-traditional tradable sectors.
43
stimulate demand, growth, and employment generation.
Structural heterogeneity
This study has also shown that a more complex production structure has positive effects
on employment. The reduction of structural heterogeneity implies a growth rate in parallel to
productivity growth which guarantees the creation of employment in the high productivity strata.
Such policies include those that increase education access, promote innovation and increase in
technological capabilities via research and development and production linkages that integrating
small and medium enterprises into production chains to incite a rise in productivity in the lower
strata. Industrial policies therefore play a key role in changing predominating patters by altering
prices so that they favor technologically intensive production – sectors most capable of bringing
forth a dynamic competitive advantage due to the presence of high externalities (CEPAL, 2010:
Abeles, 2012). Infrastructure provision also plays a key role in ensuring that the growth of
capacity to provide goods and services. Nevertheless, it must be recognized that structural
change is a process which takes time and requires a high level of coordination between various
actors. Implementing policies in just one area can serve to reinforce rather than reduce productive
heterogeneity.
Size: Industrial policies and regional spatial planning
This study has shown a positive relationship between city size and employment,
illustrating that larger cities have a positive endogenous quality which allows them to generate
more employment under a given macroeconomic regimes. Large, dense urban areas (i.e.
concentration of population and productive activities) reduces costs for businesses, increase the
profitability of investment, encourage knowledge spillovers, and promote the division of labor
(CEPAL, 2012). The policy implication is thus the necessary use of industrial policies in smaller
cities in order to artificially create the benefits afforded by economies of agglomeration in larger
44
cities. Such policies will allow smaller cities to reap a greater employment benefit from
economic growth on the national level.
While there are many benefits of large and dense urban areas, it is important to recognize
that there can exist too much of a good thing.15 Large populations, extensive geography, and
intensive productive activities, all characteristic of large cities, lead to congestion costs, the
generation of increasingly large quantities of waste, the rising cost of territorial management, and
rising prices of buildable land. With larger cities, failures in institutional coordination and
strategic metropolitan governance (structural and functioning inefficiencies) become more
common (Prud’homme et al., 2004). CEPAL (2012) terms the costs of large cities “urban
deficits”, which relate to the themes of housing, infrastructure, public services, transportation, the
environment, security and city management, among others. With respect to employment
specifically, increased migration to cities can lead to an increase in informality if this growth in
the labor force is not accompanied by a growth in available positions. According to CEPAL
(2012), urban deficits result from the historical inability to absorb productively, coherently and
with dignity the rapid growth of the population, the combination of a limited surface area and
intense activity of the cities, scarce and unequally allocated resources, weak urban institutions,
the absence of a strategic vision and the lack of technical and administrative tools for designing
and applying appropriate city policies.
Framed in terms of a cost benefit analysis, Quigley (1998) shows that as city size and the
area devoted to housing increase spatially, the average distance a worker must commute to get to
the central business district for work necessarily increases, as does congestion. Thus, increased
costs arising from higher rents due to competition for space and higher commuting costs due to
more distant residences will eventually offset the production and consumption advantages of
15 Williamson 1965, Wheeton and Shishido 1981, and Henderson 2000 (cited in CEPAL, 2012) have found an inverted U-shape relationship between economic development and urban concentration, finding that economic growth drives urban concentration until a saturation point at a distinct per capital income level, after which urban concentration begins to fall.
45
large cities.
In both cases there appears a saturation point past which increasing per person resource
costs negate the resource savings due to scale economies in good production, or more broadly,
that the costs of congestion and other urban deficits outweigh the benefits of size. Henderson
(2000) finds very robust growth losses from significantly non-optimal concentration, listing
Argentina as having excessive primacy.
This implies a large role for regional spatial planning in incentivizing the growth of
medium sized cities, of which there is an absence as shown initially with the distribution of
Argentine cities. The benefits of the creation of medium sized cities are two-fold. One the one
hand, it would reduce migration to the already saturated cities, thereby slowing the deepening of
urban deficits. On the other hand, it would allow other cities to surpass the necessary threshold of
size and density, beyond which they can reap the benefits of agglomeration. Regional spatial
planning also relates to the coordination of production, given that all cities cannot specialize in
the same thing. Rather, there needs to be some oversight with respect to which industries to
subsidize and where. This leads into the first area of future research, namely the need for work
on specific industrial policies that can be used in small cities to help artificially create the benefits
of large cities. Case studies on policies used in cities whose populations have doubled, tripled, or
quadrupled in the past two decades will be of key importance as they can help specify the
mechanisms of transmission between policies and their effects on city size.
The second area of future research relates to thresholds, beyond which the benefits of
agglomeration economies are realized. Thresholds will depend on a variety of factors, ranging
from the goods produced, proximity to other cities, relationship with the hinterland, provision of
infrastructure, social protection nets, among other things. While it is clear that there is not a one-
size-fits-all threshold, research on the different factors will help give a better idea of how and
46
when to promote such growth in the size of cities in order to realize the benefits of the large
cities, without the costs of megalopolises. Such research will ensure that the theoretical benefits
of size and density do not remain that way, but rather bear fruit in reality, generating employment
and improving the lives of the population.
47
Bibliography
Abdel-Rahman, H., & Fujita, M. (2006). Product variety, Marshallian externalities, and city sizes. Journal of Regional Science, 30 (2), 165-183.
Abeles, M. (2012). Argentina and the 2008-09 international economic crisis: Unconventional resilience. In Cohen, M. (Ed.), The global economic crisis in Latin America: Impacts and responses (59-84). New York: Routledge Studies in the Modern World Economy.
Buckley, R. (2012). The Economics of Cities: Feb. 1, 2012 [PowerPoint Slides].
Carlino, G.., Satyajit, C. & Hunt, R. (2006). Urban density and the rate of invention (Working Paper 04/16-R). Federal Reserve Bank of Philadelphia.
CEPAL. (2012). Población, territorio y desarrollo sostenible. Naciones Unidas: Santiago.
CEPAL (2010). La hora de la igualdad: Brechas por cerrar, caminos por abrir. Naciones Unidas: Santiago.
Ciccone, A. & Hall, R. (1993) Productivity and density of economic activity (Working Paper No. 4313). National Bureau of Economic Research.
Cimoli, M., Novick, M & Palomino, H. (2007). Introducción: Estudios estratégicos sobre el trabajo y el empleo para la formulación de políticas. In Novick, M. & Palomino, H. (Cds). Estructura productive y empleo: Un enfoque transversal. Ministerio de Trabajo, Empleo y Seguridad Social: Buenos Aires.
Cohen, M. (2012). Argentina’s economic growth and recovery: The economy in a time of default. New York: Routledge Studies in the Modern World Economy.
Damill, D., & Frenkel, R. (2011). Macroeconomic policies and performance in Latin America 1990-2010 (Centro de Estudios de Estado y Sociedad). Retrieved from http://www.itf.org.ar/pdf/documentos/84-2011.pdf
Damill, M., R. Frenkel & R. Maurizio (2011). “Macroeconomic policy for full and productive employment and decent work for all. An analysis of the Argentine experience”, Employment Working Paper Series No. 109, OIT, Geneva.
Dumais, G., Ellison, G. & Glaeser, E. (2002). Geographic concentration as a dynamic process. The Review of Economics and Statistics, 84 (2), 193-204.
Duranton, G., & Jayet, H. (2011). Is the division of labour limited by the extent of the market? Evidence from French cities. Journal of Urban Economics, 69 (1), 56-71.
Duranton, G. & Puga, D. (2003). Micro-foundations of urban agglomeration economies (Working Paper No. 9931). Retrieved from National Bureau of Economic Research website: http://www.nber.org/papers/w9931
Ffrench-Davis, R. (2011). Macroeconomia para el empleo decente en America Latina y el Caribe. (Employment Working Paper No. 79). Retrieved from Organizacion Internacional del Trabajo website: http://www.ilo.org/employment/about/news/WCMS_156123/lang--es/index.htm
48
Frenkel, R., & Rapetti, M. (2012). Exchange rate regimes in the major Latin American countries since the 1950s: Lessons from history. Revista de Historia Economica (Second Series), 30, 157-188. doi:10.1017/S0212610911000292
Frenkel, R. & L. Taylor (2006). “Real Exchange Rate, Monetary Policy, and Employment”, DESA Working Paper No. 19, United Nations, Department of Economic and Social Affairs.
Glaeser, E. L., Kallal, H. D.,& Scheinkman, J. A. (1992). Growth in cities. Journal of Political Economy, 100 (6), 1126-1152.
Golstein, G.. S. & Gronberg T. J. (1984). Economies of scope and economies of agglomeration. Journal of Urban Economics, 16 (1), 91-104.
Henderson, J. V. (2000). How urban concentration affects economic growth (Working Paper No. 2326). The World Bank Policy Research Working Paper Series.
Henderson, J. V. (1986). Efficiency of resource usage and city size. Journal of Urban Economics, 19, 47-70.
Hesley, R. W. & Strange, W. C. (1990). Matching and agglomeration economies in a system of cities. Regional Science and Urban Economics, 20 (2), 189-212.
Infante, R. B. & Sunkel, O. (2009) Chile: Towards inclusive development. CEPAL Review, 97. Retrieved from http://www.eclac.cl/publicaciones/xml/0/36660/RVI97InfanteSunkel.pdf.
Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99 (3), 483-499.
Krugman, P. (1991). Cities in space: three simple models (Working Paper No. 3607). National Bureau of Economic Research.
Mazorra, X. & Beccaria, A. (2007). Especialización productiva y empleo en Areas Económicas Locales. In Novick, M. & Palomino, H. (Cds). Estructura productive y empleo: Un enfoque transversal. Ministerio de Trabajo, Empleo y Seguridad Social: Buenos Aires.
Maurizio, R. (2009). Macroeconomic regime, trade openness, unemployment and inequality: The Argentine experience (Working Paper No. 03/2009). Retrieved from International Development Economics Associates (IDEAs) website: http://www.networkideas.org/working/apr2009/wp23_03_2009.htm
Maurizio, R., Perrot, B., & Villafañe, S. (2009). Políticas públicas y empleo: Desafíos y oportunidades en una economía global. In OIT-Ministerio de Trabajo, Empleo y Seguridad Social, Informe nacional sobre el impacto Social de la Globalización en Argentina (chapter 2). Buenos Aires: Argentina.
Mills, E. S. (1967). An aggregative model of resource allocation in a metropolitan area. The American Economic Review, 57 (2), 197-210. Retrieved from http://www.jstor.org/stable/1821621.
Ministerio de Trabajo, Empleo y Seguridad Social (2011). Trabajo, ocupación y empleo: La complejidad del empleo, la protección social, y las relaciones laborales, Serie de Estudios/10.
49
Nakamura, R. (1985). Agglomeration economies in urban manufacturing industries: A case of Japanese cities. Journal of Urban Economics, 17 (1), 108-124.
Pastrana, F., Toledo, F., & Villafañe, S. (2012). El modelo económico ante la crisis internacional sostenimiento de las condiciones laborales y sociales en Argentina. In Macroeconomía, empleo e ingresos: debates y políticas en Argentina frente a la crisis internacional 2008-2009 (63-108). Ministerio de Trabajo, Empleo y Seguridad Social and Organización Internacional del Trabajo. Argentina.
Porcile, G. (2011). La teoría estructuralista del desarrollo. In Infante, R. (Ed.) El desarrollo inclusivo en América Latina y el Caribe: Ensayos sobre políticas de convergencia productiva para la igualdad. CEPAL, Naciones Unidas: Santiago.
Prud’homme, R., Huntzinger, H., & Kopp, P. (2004). Stronger Municipalities for Stronger Cities in Argentina. Inter-American Development Bank.
Quigley, J. M. (1998). Urban diversity and economic growth. Journal of Economic Perspectives, 12 (2), 127-138.
Rosenthal S., & Strange, W. C. (2001). The determinants of agglomeration. Journal of Urban Economics, 50, 191-229.
Tomada, C. (2007). Prólogo. In Novick, M. & Palomino, H. (Cds). Estructura productive y empleo: Un enfoque transversal. Ministerio de Trabajo, Empleo y Seguridad Social: Buenos Aires.
World Bank. (1994). Informe sobre el desarrollo mundial 1994: Resumen. World Bank: Washington D.C.
50
Appendix 1: Size
Population Individual % Cumulative %LARGEPartidos del GBA 10,089,033 41.028% 41.028%Ciudad de Buenos Aires 2,995,050 12.180% 53.207%Gran Córdoba 1,410,132 5.734% 58.942%Gran Rosario 1,264,321 5.141% 64.083%Gran Mendoza 904,771 3.679% 67.762%Gran Tucumán - Tafí Viejo 814,915 3.314% 71.076%Gran La Plata 747,361 3.039% 74.115%Mar del Plata - Batán 620,095 2.522% 76.637%Salta 540,319 2.197% 78.834%Gran Santa Fe 507,908 2.065% 80.900%MEDIUMGran San Juan 470,562 1.914% 82.813%Gran Resistencia 391,962 1.594% 84.407%Santiago del Estero - La Banda 372,639 1.515% 85.922%Corrientes 360,255 1.465% 87.387%Bahía Blanca - Cerri 311,085 1.265% 88.653%Jujuy - Palpalá 310,756 1.264% 89.916%SMALLPosadas 300,200 1.221% 91.137%Gran Paraná 277,902 1.130% 92.267%Neuquén - Plottier 266,733 1.085% 93.352%Formosa 241,801 0.983% 94.335%Gran Catamarca 206,235 0.839% 95.174%San Luis - El Chorrillo 204,800 0.833% 96.007%La Rioja 183,413 0.746% 96.752%Río Cuarto 164,833 0.670% 97.423%Concordia 152,986 0.622% 98.045%Comodoro Rivadavia - Rada Tilly 143,669 0.584% 98.629%Ushuaia - Río Grande 123,193 0.501% 99.130%Santa Rosa - Toay 122,007 0.496% 99.626%Río Gallegos 91,911 0.374% 100.000%
51
Appendix 2: Density
DENSITY (2005)Inhabitants/km2
Ciudad de Buenos Aires 14431.97Mar del Plata - Batán 6416.69Partidos del GBA 2582.95Corrientes 694.65Gran La Plata 598.85Gran Córdoba 428.74Gran San Juan 345.17Gran Rosario 325.02Posadas 289.12Gran Tucumán - Tafí Viejo 238.94Gran Santa Fe 160.57Salta 156.70Gran Catamarca 130.02Bahía Blanca - Cerri 126.37Gran Resistencia 108.09Santiago del Estero - La Banda 62.26Gran Mendoza 58.30Gran Paraná 53.86Concordia 40.04Formosa 37.21Neuquén - Plottier 33.37Santa Rosa - Toay 15.46San Luis - El Chorrillo 14.27La Rioja 12.44Comodoro Rivadavia - Rada Tilly 9.13Río Cuarto 8.59Jujuy - Palpalá 8.13Ushuaia - Río Grande 5.02Río Gallegos 2.26
52
Appendix 3a: Complexity
Complexity Indicator
SAN LUIS - EL CHORRILLO 1.41357412
LA RIOJA 1.36355623
PARTIDOS DE GBA 1.33860146
GRAN ROSARIO (2) 1.19966323
GRAN CATAMARCA 1.12247919
USHUAIA - RIO GRANDE 1.10060513
GRAN MENDOZA 1.02785197
RIO CUARTO (5) 0.99544474
GRAN LA PLATA (1) 0.99515698
GRAN CORDOBA 0.97374744
CIUDAD DE BUENOS AIRES 0.92978562
MAR DEL PLATA, SAN NICOLAS Y BAHIA BLANCA (6) 0.89236583
RAWSON - TRELEW 0.88094691
GRAN SAN JUAN 0.86639664
CONCORDIA (4) 0.86046041
GRAN SANTA FE 0.83351074
GRAN PARANA 0.8221811
SALTA 0.73762679
CORRIENTES 0.72924316
POSADAS 0.71406286
GRAN TUCUMAN 0.70314884
GRAN RESISTENCIA 0.66743791
SGO DEL ESTERO 0.65776745
53
SANTA ROSA - TOAY 0.63547483
RIO GALLEGOS 0.62741332
CDRO RIVADAVIA (3) 0.59530381
NEUQUEN - PLOTTIER 0.57277489
FORMOSA 0.5646834
JUJUY 0.54404319
VIEDMA - CARMEN DE PATAGONES 0.41755434
Aclaraciones sobre algunos aglomerados
(1) Dato aproximado a partir del Empleo declarado en la Zona Fiscal denominada Tercer Cinturon del GBA
(2)Dato aproximado a partir del empleo declarado en la Zona fiscal del Resto de Santa Fe (Sin Ciudad de Santa Fe)(3)Dato aproximado a partir del empleo declarado en la Zona fiscal del Resto de la Provincia de Chubut (Sin Ciudad de Rawson y Trelew)
(4)Dato aproximado a partir del empleo declarado en la Zona fiscal del Resto de la Provincia de Entre Rios (Sin Ciudad de Parana)
(5)Dato aproximado a partir del empleo declarado en la Zona fiscal del Resto de la Provincia de Cordoba (Sin Ciudad de Córdoba)(6)Dato aproximado a partir del empleo declarado en la Zona fiscal del Resto de la Provincia de Buenos Aires (Sin Ciudad de Bs As, Partidos de GBA y Tercer Cordon de GBA)
Fuente: Observatorio de Empleo y Dinámica Empresarial, MTEySS en base a SIPAAppendix 3b: Complexity
Where X is registered employment, and
i = high value added sectors
j = city
Xij= workers in the high value added sectors in the specified city
Xj= total workers in the specified city
Xi= total workers in the high value added sectors in Argentina
X = total workers in Argentina
54
High value added sectors refer to the manufacturing industry, and the service sectors: transportation, communication, and financial intermediation.
55
Appendix 4: Significance Tables
MACRO MODEL
1995-2001
locupados
limpuestos 0.131
(5.20)**
grande 0.649
(14.52)**
mediana 0.278
(6.91)**
complejidad 0.298
(5.16)**
densidadd 0.079
(2.22)*
_cons 4.160
(62.75)**
R2 0.74
N 348
* p<0.05; ** p<0.01
locupados
limpuestos 0.064
(2.44)*
grande 0.613
(22.99)**
mediana 0.245
(9.62)**
complejidad 0.113
(2.94)**
region1 0.954
(19.84)**
region2 0.177
(4.88)**
region3 0.225
(7.60)**
region4 0.085
(2.79)**
region5 0.081
(2.57)*
_cons 4.379
(78.79)**
R2 0.92
N 348
56
MACRO MODEL2003-2010
locupados
limpuestos 0.130
(8.89)**
grande 0.684
(26.77)**
mediana 0.294
(11.97)**
complejidad 0.323
(9.20)**
densidadd 0.039
(1.75)
_cons 4.184
(88.98)**
R2 0.75
N 841
locupados
limpuestos 0.079
(6.24)**
grande 0.615
(41.61)**
mediana 0.247
(15.94)**
complexity 0.142
(5.76)**
region1 0.859
(30.26)**
region2 0.129
(5.33)**
region3 0.166
(8.73)**
region4 0.064
(3.26)**
region5 0.031
(1.73)
_cons 4.422
(118.71)**
R2 0.92
N 841
57
MICRO MODEL
2006 Quarter 2
estadon sexo 0.630(32.58)**
hpi -0.698(27.26)**
si 0.213(9.03)**
sc 0.388(16.77)**
ti 0.309(11.27)**
tc 0.553(17.63)**
mediana 0.019(0.71)
grande 0.088(3.60)**
edadt2 1.586(88.47)**
edadt3 1.381(57.33)**
complexity 0.175(5.10)**
publico 1.868(38.68)**
densidadd -0.071(3.27)**
sex_jef 0.264(11.10)**
_cons -1.769(44.92)**
N 47,646
58