48 Journal of Reviews on Global Economics, 2016, 5, 48-58

Is the Economic Convergence of the Mexican States Possible?

Estimates for the 1940-2013 Period

Mario Camberos C.* and Joaquín Bracamontes N.*

Department of Economic Development at the Center for Research on Food and Development (CIAD), México
Abstract: In this paper we study -convergence: absolute and conditional, also  convergence, using cross-state
regression models, based in a long term period of time for the Federated States of Mexico. When absolute convergence
is estimated (Solow Model, 1956) we found the negative sign expected but the result is not statistically reliable; while,
estimates for decades only show absolute convergence for the period of 1960-1970, known with the term of “Mexican
Miracle”, a third regression including population growth rate and physical capital investment per capita, variables as
considered by Solow models, confirm that there is not absolute convergence like the first result obtained. The estimate
including human capital index (HCI) and human development index (HDI 2), shows a number of outliers, suggesting the
introduction of proxy variables which capture the political effects and explore conditional convergence. When panel
heteroskedastic is considerate, convergence is observed, but  2% any case was estimated.

Keywords: Convergence, human capital, regional economics.

1. INTRODUCTION hypothesis or beta convergence (); the second part is

related with conditional convergence. Both Lucas, and
Mexican states low and uneven economic growth Barro and Sala-i-Martin insight is related to the
and inequality, registered last four decades, throw foundation of the Endogenous Economic Growth
worries about development path, for which it is Modern Theory.
necessary to research economic growth factors that
have guided failure Mexican state development. In this  convergence implies that developing countries
context it is important drive the economic convergence grow faster than rich ones steadily to an estimated rate
idea, as central subject of this paper, and review at of 2%. (Barro and Sala-i-Martin 1992). Conditional
light of our results, whether factors such as capital convergence relates convergence towards different
physic and labor, supported on Solow and human steady states. Implies that different countries achieve
capital on Lucas models of economic growth theory, different income levels while in a steady state.
and the Barro and Sala-i-Martin hypothesis to prove ß-
convergence, which has been tested for countries and Another basic concept is sigma convergence (),
regions such as the United States, the European Union suggested by Barro and Sala-i-Martin (1992), and it is
using econometric methods, and new members of the defined as income dispersion reduction during
European Union, explain Mexican state economic convergence period; not always observed through the
growth. process. This kind of convergence (as suggested by
Barro and Sala-i-Martin 1992) is very important in the
In this though line convergence hypothesis as
Mexican case study, which is the objective of this
enunciated by Lucas (1988), developed a fruitful
paper, because economic differences of development
debate that has prevailed for the last three decades, on
levels among regions and states have been persistent
what forces guide economic growth towards a steady
along its modern economic history; moreover, the
state. Lucas introduced in the Cobb-Douglas functional
Mexican case has been recognized in the world by its
form of production, human capital, as the most
important factor on economic growth. Then Barro and economic and social inequality, and requires a long
Sala-i-Martin questioned whether poor countries or term research with solid statistics foundations. In this
regions tend to grow faster than rich ones. Are there paper convergence
is estimated.
automatic forces that lead to convergence over time in
The major factors that join to the economic growth
the levels of per capita income and product? (Barro
process are capital and work (Solow, 1956), also
and Sala-i-Martin 1992). Decomposition is possible in 2
known as the exogenous growth model, with free
parts: the first one known as the absolute convergence
access to technology; population growth rate, human
capital and depreciation rate (Lucas 1988; Barro and
*Address correspondence to these authors at the Department of Economic Sala-i-Martin 1992) also known as the endogenous
Development at the Center for Research on Food and Development (CIAD),
México; Tel: 6622-892400 ext. 319; Fax: 6622-800057; growth model, this because the human capital makes
E-mail: [email protected], [email protected] the difference. Beta conditional regards other factors
Thanks Christian Millan and Abel Castro.

E-ISSN: 1929-7092/16 © 2016 Lifescience Global

Is the Economic Convergence of the Mexican States Possible? Journal of Reviews on Global Economics, 2016, Vol. 5 49

such as economic policies, openness, money pattern of
across periods can reflect aggregate
exchange, savings, life expectancy, foreign inversion, disturbances that have differential effects on state
financial dependency and government consumption income as represented by term iSt that catch sectorial
(Sachs 1996; Rodrik 2014; Dobrinsky and Havlik 2014) composition.
regarding state intervention in order to achieve speed
of convergence. Estimation of convergence for 98 counties using the
same methods, income rate annual as dependent
All models are mainly based on Cobb-Douglas variable, and growth per capita income, Barro and
function and econometric methods; Ordinary Least Sala-i-Martin do not find convergence
. Then several
Squares cross-section and panel data are used to proxy variables are added to human and political
measure convergence, which will be described in the capital and measure of market distortion, and
methodology section and when doing its application. conditional convergence is estimated
= 0.0184, no
longer below US state estimation
= 0.0236 same time
Hypothesis has been tested for counties and states period 1960-1985. One of the most important
in the United States (Barro and Sala-i-Martin 1992; conclusions found is that
-convergence around 2%,
Sachs and Warner 1996; Rodrik 2014), European became a convergence pattern and it is considered for
Union (Dobrinsky and Havlik 2014; Dvorokova 2014); further research in the next 25 years.
OECD (Arnold, et al. 2010); New European Union
members (Sikic 2013; Oblath et al. 2015); World Later another authors applied the same model to
regions (Sala-i-Martin 1995; Mathur, 2005); counties in estimate convergence conditional to test economic
America (Young et al. 2008); Mexican case (Diaz- policies and politics facts impact. In this though stream,
Bautista 2000). Sachs (1995) lying augment: Economic growth and
therefore economic convergence require reasonably
Hypotheses that have been proved include specific economic institutions. Poorly managed economies are
subjects too, such as public policies (Sachs and unlike experience convergence no matter what the
Warner 1996); infrastructure investment (Button 1998); underlying production technology or initial level of
education role (Diaz-Bautista 2000); agricultural sector human capital. Economies that are poorly managed are
(Somasekharan 2011). improbable of experimenting convergence no matter
what the underlying technology or level of human
In this scheme the objective is to search ß- capital are.
convergence, and
-convergence over specific regions
in Mexico’s federated states in a long term fashion. That is why as convergence condition a certain
amount of dummy variables are introduced and copes
The next step in this paper’s strategy is to show with two types of countries: qualifying and non-
convergence evidences for U.S country and states, EU qualifying. Non-Qualifying if one of the next features is
and new members, next make a discussion about the present: 1) a socialist economic structure; 2) civil war o
convergence basic theory foundation and major external war during the period 1970-1989 and 3)
methodological approach, after showing some stylized extreme deprivation of civil or political rights, or do not
facts on Mexico’s federated states during the last 70 pass the test of openness (market).
years, followed by analysis results and conclusions at
last. Under this criterion 81 countries were excluded of a
total of 117, Latin America as a whole, Mexico due to
2. CONVERGENCE HYPOTHESIS EVIDENCE black market’s 20% premiums and due to
government’s domestic protection policies. The
Barro and Sala-i-Martin (1992) estimates found regression considers this division and parameter ß is
absolute convergence for the 1880-1988 period,
= estimated in a separated way.
0.0175, and for any nine period
values interval
[0.0119-0.0.0373], correct sign, value significance The author test convergence hypothesis 1970-1989,
statistic and errors below 5%, except 1920-1930 initially with only income showing the simple bivariate
decade
= -0.0122 value meaning divergence. When relationship between initial income in 1970 and
other explicative variables are included (9 economic subsequent growth between 1970 and 1989, but does
sectors) model estimates are mixes: before 1920-30 not registers convergence for the entire sample and
decades, divergence is observed, and convergence on coefficient is positive and is statistically insignificant;
later decades. Authors argument that this unstable separating both samples, qualifying countries show
50 Journal of Reviews on Global Economics, 2016, Vol. 5 Camberos and Bracamontes

convergence, but non-qualifying do not (results are integrated in the Cobb-Douglas production function,
statistically insignificant). Then 5 politics are included included Solow residual, and known as Exogenous
using 5 dummies. Results confirm the effectiveness of Development Model and 2. Lucasian (Lucas, 1988) and
politics on convergence. Signs are negative, which his concern to include some endogenous factors such
contributes to reduce income difference between as human capital and technological innovation, like
countries through time. On the other hand, Young et al. development foundation.
(2008) estimated convergence for the United States
and 32 states, based on 3058 counties information Both Solow and Lucas models are basic for Barro
between the 1970-1998 period, using OLS and 3SLS and Sala-i-Martin (1992) convergence theory that
and getting contrasting results. So OLS estimate for guarantees that the economic growth in developing
USA gives  = 0.0239 value and wide states interval countries, will be faster than in rich countries in long
[0.0044 West Virginia, -0.08 Idaho]; while using 3SLS  term. Considering as an outset different neoclassicals
= 0.0658 value and states interval [California 0.0375, - as Ramsey (1925) and Solow (1956), per capita growth
Louisiana 0.015] rate tends to be inversely related to initial production or
income per person. Particularly, if economies are
Authors as Dvorokova (2014) used the ordinary similar according to preferences and technology, then
least square test to measure absolute convergence in the poor economies grow faster than rich ones. Thus
the period from 2001-2012 of the European Union and there is a force that promotes convergence in levels of
estimated  = 0.027 and by specification model, per capita product and income, as outlined by Barro
coefficient has a negative sign, as outcome and Sala-i-Martin. So speed growth poorer countries,
convergence was observed. In this framework analysis, or regions, in combination with slow growth richer
Dobrinsky and Havly (2014) found in a study for 27 countries outcome convergence in the long run toward
European Union Countries and ten new members, steady state, defined a situation that guarantees a
using an univariate and multivariate cross-country constant growth rate of population, capital, as
regression model of per capita income growth as consequence the income-product growth was
dependent variable, that the 27 European economies considered as stylized facts.
throughout the last 35 years (ending 2012) estimates 
= 2% value convergence, similar Barro and Sala-i- It is possible to distinguish three concepts of
Martin (1992) finding twenty year ago. convergence. Two -convergence: absolute and
conditional convergence, and at last,
-convergence.
Results of
-divergence were found by Barro and Barro and Sala-i-Martin (1992) define original
Sala-i-Martin for the United States: increasing 1880- convergence as a process of long term growth income
1930, divergence; decreasing 1930-1970, convergence speed of the developing countries in comparison to the
and soaring 1970-1988, divergence; when sectorial rich ones towards a steady state. According to Sikic
composition variables are included, income dispersion (2013), absolute convergence assumes that the
of the 1930 and 1988 peak years eases up (Young et countries of analyzed groups have the same
al., 2008) estimating sigma during 1970-1998, but characteristics. Dvorokova (2014) argue in a recent
found divergence. The 1998 standard deviation for full paper for the EU, that initially poorer countries
sample grew about 5.8 percent; meanwhile for 24 evidence more dynamic growth. This means that
states sigma was higher at 10 percent level. poorer countries converge to initially richer countries
Conclusion is that for many individual states as well as because those do not have such dynamic growth.
for the entire U.S.
-divergence occurred from1970 to
1998, similar to Barro and Sala-i-Martin as stated Conditional convergence, opposite to absolute
above. convergence, postulates that convergence is based on
endogenous factors promoted by mean economic
3. THEORY AND CONCEPTUALIZATION policies such as: investment in human capital and
infrastructure, in addition to capital equipment;
The standard convergence theory is based on the economic liberalization and money exchange policies
neoclassical concept of long-term economic growth (Lucas 1988; Sachs 1996; Young et al. 2008;
(Dvorokova, 2014), that proposes two problems to Rodrik2014), all together achieve faster convergence.
solve: 1. Solovian (Solow, 1955) per capita income
growth to constant rate in the long term assuming However convergence is not an ineluctable fact; by
capital free access, labor and another external factors the time, dissident authors as Romer, (cited by Sachs
Is the Economic Convergence of the Mexican States Possible? Journal of Reviews on Global Economics, 2016, Vol. 5 51

1996) noted that his theoretical [Solow] model with sectional dispersion of log (yit) decline over time. The
increasing returns to scale seemed to be broadly effect of -convergence, which tends to reduce
consistent with cross-country growth experience of the dispersion, is balanced by random shocks, which tend
postwar era, in which there was no discernible trend of to increase this dispersion.
poorer nation to converge with the richer nations. In the
same line of thought Baumol (1994 Sachs cited) Authors demonstrate by means of this formulation:

( )
suggested that only countries with an adequate initial
level of human capital endowments could take  t2 = e
2   t
1
2
+  μ2 (2)
advantage of modern technology to enjoy the
possibility to convergent growth.  μ2  2  μ2 
2 t
 =
2
+  0
e (3)
t
1
e
2   1
e
2 
After thirty years of discussion about convergence,
several questions may arise from this initial observation
1. What growth rate guarantees convergence to steady As it is observed in
t2
both equations
state? 2. Does Convergence guarantees income monotonically approaches the steady state value,
dispersion reduction of the poorest in relation to the which rises with
μ2 but decline with .
richest countries?
Sigma convergence has been converted now day in
Barro and Sala-i-Martin model analyses questions
important subject, regarded as one of relevant
and propose the average growth rate of Y over the
problems because the inequality has reached high
interval between dates 0 and T:
levels within, as between world countries, in
 Y (T ) 
consequence, is very important to search Mexican
1 1
e
T  Ŷ * 
 log  = x+  log   states case sigma convergence too.
 Y ( 0 ) 
(1)
T T  Ŷ (0)
4. METHODOLOGY AND DATA
*
Where Y = income, Ŷ =income growth rate, Ŷ =
steady state income growth rate, T = years,  = In this part, steps to measure  and
convergence
parameter measure convergence rate and X = will be explained. Econometric methods have been
progress technological exogenous. used to operate the economic growth and per capita
income relation to measure convergence, principal
Special interest represents  parameter, in which its cross-section, pooled and panel models.
negative sign demonstrates if convergence really
exists, and its value, growth rate, to steady state. In this 4.1. ß and
Convergence
point of analysis most authors refuse Solow model of
convergence as a natural result of economic growth Dvorokova (2014) found out that the methodology to
and exogenous factors, and prefer to include measure -convergence, comes from original Baumol
endogenous factors (Lucas 1988; Barro and Sala-i- (1986) real convergence study; but Barro and Sala-i-
Martin 1992), economic and social policies (Sach1996, Martin (1992) elaborated an equation that became the
Rodrik 2014). standard one using a cross-section country method to
estimate -absolute convergence:
In relation to the second question, Dvorokova
y 
(2014), based on a recent European Union research,
claims that it is not entirely clear that -convergence
log  it
 y i,t
1
 ( ) ( )
= a
1
e
  log yi,t
1 + μi,t (4)

guarantees income dispersion reduction. On the other

hand, Young et al. (2008) claims that despite literature Where per capita real GDP is used, t represents the
emphasizes on -convergence, economists have time (in years), i represent the nation or region and

stochastic error term. The left-hand side is the growth
acknowledged that is not a sufficient condition for
-
in per capita income and explanatory variable on the
convergence. Quah and Friedman, following the
right-hand side is the initial period per capita income, 
author, both suggest that
-convergence is of most
and  coefficients are estimated by no lineal least
interest because it directly tells whether distribution of
squares techniques. If  is estimate positive
income across economies is becoming more equitable
convergence is implied (negative exponent). A larger 
or not; while Barro and Sala-i-Martin (1992) explain that
represents faster convergence (Pfitzner and Lang,
-convergence does not implies that the cross -
2014).
52 Journal of Reviews on Global Economics, 2016, Vol. 5 Camberos and Bracamontes

Conditional convergence is measured, following 1 y 1
e
T

ln i,t = 
 ln yi,t
T +  X i,t + uit (8)
Ling, et al., (2003) using a model that includes a T yi,t
T T
number of other variables In order to keep the steady
state characteristics of economics constant: With variables defined as (5), added Xi,t ,that include
a group of 2 new variables: k(capitalper capita) and
1 yi,t 1
e
T
ln =
ln yi,t +  X i,t + uit (5) n(population growth rate). 3). Convergence
T yi,t
T T endogenous model like Lucas Model, that add to (7), a
group of variables Zi,t definited as:
The dependent variable is the average rate of
growth for a country i, T is interval of years, Xi,t is a set 1 y 1
e
T
of exogenous variables maintaining the steady state ln i,t = 
 ln yi,t +  X i,t +  Z i,t + uit (9)
T yi,t
T T
characteristics and  is a set of unknown coefficients.
To operate model econometric and estimate -
convergence Barro and Sala-i-Martin (1992) used a New variables defined as l (labor returns originating
cross-state regression model, assuming that the from human capital) and h (human capital index
coefficient a, is the same for all i; they assume that the composed of education and health.
steady state value, ŷi* (equation 1, above) and the rate
In this paper is assumed, as Barro and Sala-i-Martin
of technological progress, xi, do not differ across
(1992), that the coefficient
is the same for all i; that is,
states.
they assume that the steady-state value, ŷi* (equation
However, it includes variables which they think 1) and the rate of technological progress do not differ
could hold the effects of aggregate shocks on economy across Mexican states. This model is regulated through
i´s growth rate to achieve cross-sectional OLS regression cross-sectional data of the states and
independence of error terms vit in equation the hypothesis is tested by using the Bayesian
information method.
uit =
i st + vit (st is used by Barro and Sala-i-Martin
1992). Sigma convergence is measured with standard
deviation  as in (6). Finally speed convergence is
Sigma convergence has been measured by mean estimated by using the following formula:
standard deviation as the simple variance of log
income on t time (Young et al., 2008): ()
T = ln 2
(10)

 1
 ( )
2
 ln yit
μt
N
 t2 =  (6) Where variables are defined as above.
N i=1

4.3. Data
4.2. Econometric Model to Prove Mexican States ß-
Convergence Table 1 shows some indicators and index that
describes the growth and development in Mexico for
To operate the econometric model and estimate - the 1940-2013 period, specifically the annual average
convergence using a cross-state regression is used. It rates of population growth, GDP per capita and capital
is necessary to prove both, exogenous and formation, the human capital index (HCI) and the
endogenous models according to the next proceeding: human development index (HDI 2). These growth rates
and development index were estimated from the
1). general model of convergence, following Barro
databases devised and constructed in the effort to
and Sala-i-Martin (1992) and Dvorokova (2014):
obtain homogeneous and comparable statistics series
y  over time of the states in Mexico in order to facilitate
1
T  yi,0 
( )
log  i,t  =
+  log yi,0 +  i (7) long-term studies on issues such as economic growth,
regional inequality, convergence, etc. (German-Soto
2005; German-Soto 2008;German-Soto et al. 2013).
Where  parameter to measure convergence,
is a
constant parameter, i is the index of the Mexican 5. THE STYLIZED FACTS
states (32 states registered for this paper), 0 and T are
the indexes for the time (0 = 1960, T = 2013). The growth rates of GDP per capita help to identify
two stages on the path of economic growth in the
2). Convergence exogenous as Solow Model: country. The first stage comprises boom era and the
Is the Economic Convergence of the Mexican States Possible? Journal of Reviews on Global Economics, 2016, Vol. 5 53

Table 1: Mexico. Economic Development Statistics 1940-2013

1940-1950 1950-1960 1960-1970 1970-1980 1980-1990 1990-2000 2000-2013

Growth Rate Population 2.41 2.76 3.06 2.88 1.81 1.79 1.51
Growth Rate GDP/Percápita 2.81 2.72 2.93 2.78 -0.58 0.71 0.71
1
GDP/Percápita 4,671.7 6,226.5 8,353.0 11,259.0 13,427.3 13,436.2 22,067.2
Growth Rate Stock/Percápita nd nd 4.91 4.61 8.78 -0.86 0.12
1
Stock/Percápita nd nd 11,491.7 19,815.5 48,686.9 74,545.1 63,252.2
Labor-Income and HCI nd 3.52 2.75 3.67 4.52 6.69 8.56
HDI(2) nd 0.582 0.655 0.739 0.811 0.867 0.883
1
Both amounts are in Mexican pesos, based 1993.
Source. Own estimates based on:
1. Germán-Soto, Vicente (2005): "Generación del producto interno bruto mexicano por entidad federativa, 1940-1992", El Trimestre Económico, vol. 72(3): 617-653.
2. Germán-Soto, Vicente (2008): “El stock de capital industrial medido a través de la relación inversión-empleo: estimaciones para los estados mexicanos”,
Ensayos, 27(1): 53-80.
3. German-Soto, Vicente; Reyna E. Rodríguez Pérez and Carmen N. Escamilla Jiménez. 2013.“Acumulación y desigualdad del capital humano entre los estados
mexicanos durante 1960-2008”, Paradigma Económico, Revista de Economía Regional y Sectorial, 5(2): 5-31.
4. Germán-Soto, Vicente (2015): "Population Statistics by Mexican federal entity", from the Selected Works of Vicente German-Soto.

process of import substitution industrialization (Blanco while the average annual growth rate of population was
1979 and Huerta 1986), which covers the first three 1.51%.
decades (1940-1970), where the economic growth
remains at relatively low rates: 2.81%, 2.72% and Despite a steady increase seen on the national
2.93%, which are lower than the rate of population GDP per capita, the trend observed in the average
growth. rates of economic growth per decade do not realize
that the country was holding a process of economic
The second stage includes the last four periods of convergence, a situation confirmed by analyzing the
analysis (1970-2013), which comprises for a significant behavior growth rates in the enclosed states,
drop in the rate of national economic growth. In this particularly those states that have historically been the
way, we can see that in the decade of the 1970's the most underdeveloped states in the Mexican economy:
average GDP per capita annual growth rate resulted in Chiapas, Guerrero and Oaxaca, which for the last
2.78%-less than population growth rate-, despite the period of analysis showed growth rates of -0.14 %,
inequality in the balance of payments (BOP), the public 0.26% and 0.67%, respectively. Moreover, all states
deficit and price increases that forced the Mexican experienced a drop in the growth rate of GDP per
peso devaluation of the peso in August of 1976 (Ros capita throughout the study period, as same as the
1987 and Huerta 1986). country itself (see Table 2).

For the decade of the 1980’s, in the context of the In terms of capital stock per capita accumulated a
crisis and renegotiation of the foreign debt, it can be continuous increase over the period is also observed,
seen an abrupt drop of economic growth in the country but when analyzing the mean annual growth rates of
due to a negative rate of -0.58% recorded over this capital stock per capita a process of less capitalization
period, while the average annual population growth of the Mexican economy in long term can be seen,
was 1.81%, whereby this period has been named the which shows a process of divergence rather than
lost decade of economic dynamics in Mexico. economic convergence in the country. The apparent
contradiction between the high rate of growth in stock
In the decade of the 1990's economic growth rate per capita and the sharp drop in the growth rate
was only of 0.71%, in a context of current account observed in the eighties, is explained by the high inflow
deficit due to the high inflow of capital invested in short- of capital invested in short-term financial instruments
term financial instruments rather than productive rather than productive investment, which led to the
investment (Guillén 2000), which led to the devaluation crisis of mid- 1990's.
th
of December 29 , 1994, which in turn became in
inflation and economic stagnation. Similarly, in the On the other hand, human capital index based on
global financial crisis of the century, the economic labor income (Labor-Income based HCI) measures the
growth rate was just 0.71% in the Mexican economy, skills and abilities of people who are involved in
54 Journal of Reviews on Global Economics, 2016, Vol. 5 Camberos and Bracamontes

Table 2: Mexico. Economic Development Statistics for States 1940-2013

Growth
Rate Growth Combined
GDP/ Rate index of
GDP/ GDP/ Capital/ Capital/
Growth Per Stock Education
Rate Per Per capita
1
Per Per K PC Labor- and
1 1 1 1
Population Population Population capita capita 1940- capita capita 1960- incomebased Health
States 1940 2013 1940-2013 1940 2013 2013 1960 2013 2013 IHC (2008) (2008)

Aguascalientes 161,693 1,252,265 2.80 4,135.3 22,512.3 2.32 2,441.0 89,454.2 6.90 8.396 0.898

Baja California 78,907 3,381,080 5.21 16,392.5 18,247.5 0.14 10,848.3 53,308.8 2.99 7.741 0.905

Baja California
51,471 718,196 3.63 3,782.9 17,838.7 2.12 5,364.5 41,413.7 3.86 7.600 0.904
Sur

Campeche 90,460 880,299 3.12 3,443.6 15,247.5 2.03 11,760.4 290,962.5 6.12 20.202 0.878

Coahuila 550,717 2,890,108 2.27 6,083.5 23,706.3 1.86 11,544.0 157,994.6 4.96 8.564 0.897

Colima 78,806 698,295 2.99 5,072.7 15,862.2 1.55 3,190.2 70,650.3 5.90 6.382 0.892

Chiapas 679,885 5,119,186 2.77 1,532.5 6,292.0 1.93 339.6 12,391.3 6.89 10.059 0.834

Chihuahua 623,944 3,635,966 2.41 4,626.0 24,278.1 2.27 9,822.9 55,086.9 3.24 8.043 0.898

Distrito Federal 1,757,530 8,893,742 2.22 14,684.1 44,625.4 1.51 19,652.8 73,597.2 2.48 9.417 0.911

Durango 483,829 1,728,429 1.74 5,526.5 14,717.3 1.33 5,197.6 41,096.8 3.90 5.962 0.887

Guanajuato 1,046,490 5,719,709 2.32 1,901.0 13,453.0 2.68 2,653.7 32,587.6 4.75 6.348 0.880

Guerrero 732,910 3,523,858 2.14 1,289.9 8,124.9 2.52 1,140.1 49,876.0 7.25 4.695 0.837

Hidalgo 771,818 2,806,334 1.76 1,968.1 9,365.9 2.13 5,012.2 90,199.5 5.50 7.962 0.874

Jalisco 1,418,310 7,742,303 2.32 2,472.2 15,925.2 2.55 7,425.7 57,340.9 3.86 7.533 0.892

México 1,146,034 16,364,210 3.66 1,887.4 13,080.2 2.65 21,039.9 45,705.7 1.45 8.875 0.899

Michoacán 1,182,003 4,529,914 1.83 1,437.3 9,411.7 2.57 3,326.3 30,466.2 4.19 5.588 0.866

Morelos 182,711 1,874,188 3.20 3,267.4 15,304.9 2.11 4,299.5 20,923.2 2.97 8.522 0.892

Nayarit 216,698 1,178,403 2.31 2,562.6 9,431.8 1.78 1,708.7 47,267.9 6.34 5.837 0.886

Nuevo León 541,147 4,941,059 3.03 6,645.5 31,341.6 2.12 40,064.0 168,274.1 2.69 9.012 0.903

Oaxaca 1,192,794 3,959,042 1.63 808.8 6,980.4 2.96 671.8 16,774.3 6.14 9.352 0.847

Puebla 1,294,620 6,067,607 2.11 1,584.3 11,597.1 2.73 7,507.5 59,119.6 3.90 7.419 0.871

Querétaro 244,737 1,943,889 2.84 4,424.2 20,960.5 2.12 5,045.7 102,006.2 5.73 8.774 0.889

Quintana Roo 18,752 1,484,960 6.09 7,897.7 22,315.8 1.41 917.5 10,407.4 4.60 8.156 0.900

San Luis Potosí 678,779 2,702,145 1.88 2,171.9 14,310.2 2.58 2,635.8 44,149.4 5.36 7.970 0.883

Sinaloa 492,821 2,932,313 2.44 3,693.3 13,395.0 1.76 6,330.0 26,892.3 2.71 6.500 0.887

Sonora 364,176 2,851,462 2.82 5,165.5 22,061.9 1.98 6,014.4 48,497.6 3.94 7.705 0.899

Tabasco 285,630 2,334,493 2.88 2,543.7 10,155.0 1.89 936.6 211,137.1 10.55 22.407 0.880

Tamaulipas 458,832 3,461,336 2.77 6,029.4 17,077.9 1.42 5,465.9 60,706.5 4.56 8.923 0.895

Tlaxcala 224,063 1,242,734 2.34 1,724.0 8,235.0 2.14 4,910.2 43,506.6 4.12 5.646 0.898

Veracruz 1,619,338 7,923,198 2.17 3,533.3 10,137.8 1.43 8,426.8 88,520.9 4.45 13.826 0.859

Yucatán 418,210 2,064,151 2.18 4,411.1 13,496.3 1.52 8,464.1 32,425.1 2.52 4.877 0.879

Zacatecas 565,437 1,550,179 1.37 1,718.8 11,555.7 2.61 2,742.2 52,022.5 5.60 7.301 0.892

National 19,653,552 118,395,054 2.46 3,919.8 16,697.6 1.98 9,200.7 63,569.4 3.64 8.556 0.883
1
Both amounts are in Mexican pesos, based 1993.
Source. Own estimates based on:
1. Germán-Soto, Vicente (2005): "Generación del producto interno bruto mexicano por entidad federativa, 1940-1992", El Trimestre Económico, vol. 72(3): 617-653.
2. Germán-Soto, Vicente (2008): “El stock de capital industrial medido a través de la relación inversión-empleo: estimaciones para los estados mexicanos”, Ensayos,
27(1): 53-80.
3. German-Soto, Vicente; Reyna E. Rodríguez Pérez and Carmen N. Escamilla Jiménez. 2013.“Acumulación y desigualdad del capital humano entre los estados
mexicanos durante 1960-2008”, Paradigma Económico, Revista de Economía Regional y Sectorial, 5(2): 5-31.
4. Germán-Soto, Vicente (2015): "Population Statistics by Mexican federal entity", from the Selected Works of Vicente German-Soto.

production, generating income for themselves and the HCI allows quantifying progress in human capital
economy as a whole (IMF 2015). In other words, the performance and includes many of the productive
Is the Economic Convergence of the Mexican States Possible? Journal of Reviews on Global Economics, 2016, Vol. 5 55

aspects of the human being, to the extent in which they only nine states exceed the national long term ICH:
are incorporated in the wages of workers (German- Campeche, Chiapas, Mexico City, Nuevo Leon,
Soto et al. 2008). For example, job training or Oaxaca, Queretaro, Tabasco, Tamaulipas and
experience contribute to human capital, the more these Veracruz; whereas states with the lowest HCI were
characteristics are incorporated into wages, more Guerrero and Yucatan, followed by Tlaxcala, Nayarit,
easily they are captured by Labor-Income based HCI. Michoacán, Durango, Guanajuato and Colima (see
Table 2).
Henceforward, the analysis shows a significant
improvement in performance of long term human capita The basic methodology of combined education and
in the country, as it went from a 3.52 HCI in the early health index (HDI 2) lay on the United Nations
1950's an HCI 8.56 for the last decade. In this regard, Development Programme (UNDP) proposal for the

Table 3: Estimates of  and
Convergence for Federal States of Mexico

-regress 1 -Equation 2 -Equation 3 -Equation 4

lngdppercap
1940-2013 -0.43 0.68
(-0.432)
1940-1949 0.61

1950-1959 0.24 0.56

(-0.513)
1960-1969 -3.07 0.51
(-0.499)
1970-1979 5.75 0.44
(-1.557)
1980-1989 7.10 0.44
(-1,969)
1990-1999 0.66 0.42
(-0,696)
2000-2013 0.48 0.44
(-0,391)
1960-2013 3.10 3.71
(-0.615) (-0.616)
Capit per cap -2.10 -1.41
(-0.258) (-0.285)
Growthpopulation -0.05 -0.42
(-0.154) (-0.171)
Human capital index (ICH ) -0.09
(-0.111)
Human developmentindex (HDI 2) -16.29
(-3.483)
1
Both amounts are in Mexican pesos, based 1993.
Source. Own estimates based on:
1. Germán-Soto, Vicente (2005): "Generación del producto interno bruto mexicano por entidad federativa, 1940-1992", El Trimestre Económico, vol. 72(3): 617-653.
2. Germán-Soto, Vicente (2008): “El stock de capital industrial medido a través de la relación inversión-empleo: estimaciones para los estados mexicanos”,
Ensayos, 27(1): 53-80.
3. German-Soto, Vicente; Reyna E. Rodríguez Pérez and Carmen N. Escamilla Jiménez. 2013.“Acumulación y desigualdad del capital humano entre los estados
mexicanos durante 1960-2008”, Paradigma Económico, Revista de Economía Regional y Sectorial, 5(2): 5-31.
4. Germán-Soto, Vicente (2015): "Population Statistics by Mexican federal entity", from the Selected Works of Vicente German-Soto.
56 Journal of Reviews on Global Economics, 2016, Vol. 5 Camberos and Bracamontes

creation of the human development index of the Figure 1 confirms the findings, because is not observed
countries; therefore, this index is based on the any tendencies through period, but values
measured educational performance with the following concentration zero economic growth; even more, if
data: a) adult literacy b) school attendance rate and life statistic confidence was not considered, absolute
expectancy at birth as a proxy variable for health convergence Mexican states because lower economic
(German-Soto et al. 2013). It can be seen that in the growth during last 73 years, which outcomes, if
country this measure steadily increases and does not economic growth span is estimated as T = ln 2/ ß =
decreases, from 0,582 in the early 1950' s to 0.883 in 0.0067, convergence it is possible in the next 100 year,
the last decade. bad news for Mexican states economic growth.

In all the included years the Federal District stands A second step estimates convergence Mexican
out with the highest combined index, followed by states states, for any one decade using regression 2. Results,
that are mainly located in the northern border region: on base Table 3, register no convergence
, except for
Baja California, Baja California Sur, Nuevo Leon, 1960-1970 decade, by the time named Mexican
Sonora, Coahuila and Chihuahua; while the southern economic miracle; as in this time was recognized, by
states of Guerrero, Oaxaca and Chiapas are invariably media, like Mexican moment.
classified with the lowest composite index, followed by
other entities such as Hidalgo, Puebla, Michoacán (see Finding above described are caused by states
Table 2). heterogeneity economic and social characteristic of
Mexican development along Twenty Century: North
5.1. Mexican States Convergence Analysis Results close to U.S., which economic activity is related
industrial and modern services employment and South
Table 3 register all results described in this apart. tied to poor Central America region sustained on
Firstly relationship economic growth rate and per capita agricultural sectorial economic and artisanal works. In
income OLS all period is analyzed. By the way, this framework economic growth states findings reflex
regression 1 estimate absolute convergence ß = - an uneven Mexican regional development. Results
0.4901 for all period T, annual 0.0067, negative sign is perhaps evidence no operate neoclassical assuming
the expected, but standard error is higher than 5 state homogeneity, as it is observed U.S. states (Barro
2
percent. Tests t and F register reduced values and R and Sala-i-Martin, 1992; Young et al., 2008), EU
tends zero; therefore statistic confidence is not proved.

Figure 1: Economic Growth and GDP per capita Mexican States long term relationship.
Is the Economic Convergence of the Mexican States Possible? Journal of Reviews on Global Economics, 2016, Vol. 5 57

members (Dvorokova 2014) and Western Europe 5.2.
-Convergence

countries (Dobrinsky and Havly2014), basic for
success and speed convergence process. The 0.677 sigma value for the entire 1940-2013
period, is larger compared to the United States at its
A third regression including population growth rate very beginning in 1940. Nevertheless, trends are
and physical capital invest per capita, variables similar (Table 3): decreasing between 1940-1970 as
regarded by Solow models. In this case data the
value decreased from 0.612 to 0.437, which
considerate period 1960-2013, capital stock per capita reflects an income dispersion markdown between the
variable data is registered. In the same Table 3, ß- entities in Mexico. Therefore, convergence was
convergence parameter has positive sign and higher observed, but throughout the last three decades of the
th
value; in this way as estimates above report divergence 20 century, it sustained a slight increase to
= 0.441.
and seems confirm states uneven capital access as
before spot lighted.
-convergence results coincide with a Mexican
economic slow growth, situation observed since 35
At last regression that considers labor human years ago. This is a very important finding, because
capital index, education and health mixed human convergence hypothesis does not admit this result, but
capital index. Results reported of the other explicative it is shown in the Mexican state economic growth over
variables Solow and Lucas models, register negative the twentieth century and the beginning of the present
sign, that could mean convergence contribution; but one.
coefficient ß-convergence positive sign, say a different
history, refute convergence hypothesis. Finally is necessary, to contribute convergence, say
that in comparative with our finding, Diaz-Bautista
It is not interesting Bayesian test to choose better (2000) study, estimated ß-convergence 1970-1993
model, both Solow and Lucas do not back results. Two period, model include, as explicative variable, different
points is necessary take a count: 1. a lot of outliers in education levels. The study finds that the annual rate of
time series were observed; quality data should be convergence of human capital per capita in Mexico
better by dropped out. 2. Introduce proxy variables that ranges from 3.55% to 4.58%. It also estimates the
catching up policies effects and explore conditional convergence of income per capita conditional on
convergence. human capital variables, but economic growth rate and
per capita income relationship, as convergence
5.1.1. Panel Analysis
hypothesis suggest, is not tasted, and so our results
Now 32 states information is organized as panel are not comparative at all.
analysis heteroskedastic: 2336 observations, 32
groups and 73 periods. Results are registered in Table 6. CONCLUSIONS
4. It show an important coefficient  = -1.0461, that
At light of slow and uneven regional economic
mean convergence, in addition statistic confidence is
development registered last four decades, and worried
good, because standard error lesser 5%, as
about a new economic growth path, it is necessary to
heterogeneous differences among states is eliminated
Look for factors that have guided Mexican state
by using heteroskedastic panel balanced. As outcome,
development. In this context central subject of this
convergence path T = ln 2/ ß = 0.0143 = 48.5 years, is
paper was to prove Mexican state ß-convergence,
reduced at half years, related one hundred years OLS supported on Solow and Lucas economic growth
estimates; therefore, if the state heterogeneous was theory, and the Barro and Sala-i-Martin model to prove
eliminated convergence long term is possible. ß-convergence hypothesis, which has been tested for
countries and regions such as the United States, the

Table 4: Estimates of Panel  Convergence for Federal States of Mexico

ecogrorate Coef. Std. Err. zP>|z| [95% Conf. Interval]

lngdpcap -1.0461 0.1569 -6.67 0.000 -1.3537 -.73852

_cons 11.5537 1.4158 8.16 0.000 8.7786 14.328
Waldchi2(1) = 44.43 Prob> chi2 = 0.0000
Source: own estimates.
58 Journal of Reviews on Global Economics, 2016, Vol. 5 Camberos and Bracamontes

European Union using econometric methods, and new German-Soto, Vicente. 2005. “Generación del Producto Interno Bruto
Mexicano por Entidad Federativa, 1940-1992”. El Trimestre
members of the European Union. In the present script, Económico. 72(3): 617-653.
Mexican states observed no convergence. German-Soto, Vicente. 2008. “El Stock de Capital Industrial Medido
Explanations are of different nature. a través de la Relación Inversión-Empleo: Estimaciones para
los Estados Mexicanos. Ensayos. 27(1): 53-80.
First, per capita income and outcome data of the German-Soto, Vicente. Reyna E. Rodríguez Pérez and Carmen N.
Escamilla Jiménez. 2013. “Acumulación y Desigualdad del
economic growth dispersion showed an important Capital Humano entre los Estados Mexicanos Durante 1960-
aberration (outlier), which constitutes an obstacle for 2008,” Paradigma Económico, Revista de Economía
convergence. Second, the so-called free access to Regional y Sectorial. 5(2): 5-31.
Guillén, R. Arturo. 2000. México hacia el Siglo XXI. Crisis y Modelo
capital for developed countries does not apply for most
Económico Alternativo. México: Plaza y Valdés Editorial-
of the states in Mexico. Capital access is more UAMI.
expensive, and bound to resources limitation of they Huerta, Arturo. 1986. Economía Mexicana más allá del Milagro.
both explain the observance of low levels of capital, México: Instituto deInvestigaciones Económicas-Ediciones
de Cultura Popular, pp. 63-100.
yielding a reduced impact on the income growth per
Lucas, Robert. 1988. “On the Mechanics of Economic Development”,
capita, and thereby, non-convergence. Third, when Jounal of Monetary. (22): 3-42.
human capital is included, it does not support
- Mathur, Somesh K. 2005. “Economic Growth & Conditional
convergence hypothesis. The lack of sustained Convergence: Its Speed for Selected Regions for 1961-
2001”, Indian Economic Review. 40(2):185–208.
economic development shown over the analyzed
Oblath, Gábor. Éva Palócz. Dávid Popper and Ákos Valentinyi. 2015.
period of time could reflect a scarce human capital “Economic Convergnce and Structure Change in the New
resource demand. Therefore, human and physical Member States of the European Union”, MT-DP 2015/44.
capital does not seem to affect the economic growth. Ramsey, Frank P. 1926. "The foundations of mathematics."
Fourth, regression panel heteroskedastic results reflex Proceedings of the London Mathematical Society. 2(1):338-
384.
reality heterogeneous related Mexican states economic http://dx.doi.org/10.1112/plms/s2-25.1.338
growth and it is necessary to be corrected; perhaps Rodrik, Dani. 2011. “The Future of Economic Convergence.”Harvard
Government support should be considered to lead Kennedy School Goverment. Faculty Research Working
Paper Series.
developing forces towards economic convergence.
Ros, Jaime. 1987. “La Economía y la Política Macroeconómica
Durante el Auge Petrolero: 1978-1982” in Cordera, Rolando
REFERENCES and Tello, Carlos (eds) El Auge Petrolero de la Euforia al
Desencanto. México: UNAM, pp. 13-33.
Arnold, Jens., Andrea Bassanini and Stefano Scarpetta. 2010.
“Solow or Lucas? Testing Speed of Convergence on a Panel Sachs, Jeffrey. and Andrew Warner. 1995. “Economic Convergance
of OECD Countries”.IZA DP No. 5261. and Economic Policies”. NBER Working Paper No. 5039,
February 1995.
Barro, Robert J., and Xavier Sala-i-Martin 1992. “Convergence.”
Journal of Political Economy 100(2):223-251. Sikic, Luka. 2013. “Long Term Economic Convergence among Ten
http://dx.doi.org/10.1086/261816 New Members States in the Light of the Economic Crisis”.
Blanco, José. 1979. “Génesis y Desarrollo de la Crisis en México Financial Theory and Practice. 37(4):361-381.
1962-1979”. Investigación Económica. 38(150): 21-88. http://dx.doi.org/10.3326/fintp.37.4.2
Button, Kenneth. 1998. “Infraestructure Investment, Endogenous Solow, Robert M. 1956. “The Production Function and the Theory of
Growth and Economic Convergence.” The Annals of Capital”. The Review of Economic Studies. 23(2): 101–108.
Regional Science, (32): 145-162. http://dx.doi.org/10.2307/2296293
http://dx.doi.org/10.1007/s001680050067 Somasekharan, Jayasechar. Syam Prasad. and V.P. Roy Nirmal.
Diaz-Bautista, Alejandro. 2000. “Convergence and Economic Growth 2011. “Convergence Hypotesis: Some Dynamics and
in Mexico.” Frontera Norte, 13(24):85-110. Explanations of Agricultural Growth across Indian States”,
Dobrinsky, Rumen. and Peter Havlik. 2014. “Economic Convergence Agricultural Economics Research Review. (24):211-216.
and Structural Change: the Role of Transition and EU World Economic Forum. 2015. The Human Capital Report.
Accession.” Wiener Institut für Internationale Employment, Skills and Human Capital. Global Challenge
Wirtschaftsvergleiche. Insight Report. Switzerland: World Economic Forum.
Dvoroková, Katerina. 2014. “Sigma Versus Beta-convergence in Young, Andrew; Matthew Higgins and Levy Daniel. 2008. “Sigma
EU28: Do They Leand to Differente Results?”, Mathematical Convergence versus Beta Convegrence: Evidence from U.S.
Methods in Finance and Business Administration. County-Lavel Data”. Journal of Money, Credit and Bank.
Proceedings of the 1st WSEAS International Conference on 40(5): 1083-1093.
Pure Mathematics (PUMA '14): Tenerife, Spain, January 10- http://dx.doi.org/10.1111/j.1538-4616.2008.00148.x
12, 2014, pp. 88-94.

Received on 08-03-2016 Accepted on 15-04-2016 Published on 06-05-2016

DOI: http://dx.doi.org/10.6000/1929-7092.2016.05.05

© 2016 Camberos and Bracamontes; Licensee Lifescience Global.

This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License
(http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in
any medium, provided the work is properly cited.