Determinants of Regional Growth by Manufacturing Sector in Mexico,

2004-2014
Objective: Identifying the determinants of regional growth in Mexico by manufacturing sector in
the period 1988-2008.

Introduction
Figure 1. MA’s Share of National Employment in the Sector

Apparel:
2004 2014

Electrical apparatus:
2004 2014

Note: The sectors shown were selected as an example of each tech group. The sum of all areas for each year
is 100%. Source: Author’s calculations using data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 Hanson (1998) emphasize transportation costs as the driver of industrial location and
agglomeration. However, as shown in Figure 1, during the last ten years, low technology
 sectors (e.g. apparel) have moved from the north of the country to the center and south,
while medium-high technology sectors (e.g. electrical appliances) have tended to move
north.

 The literature identifies three different kinds of agglomeration effects:

i) Jacobs economies, which are externalities generated by other industries or
urbanization and under which competition is positive for growth;
ii) Marshall-Arrow-Romer (MAR) economies, which are intra-industry externalities or
specialization benefits which can be internalized by monopolies that are therefore
beneficial for growth; and
iii) Porter externalities, in which specialized industries incentivize growth and local
competition enhances the adoption of technology.

Econometric Specification

(1.1)

lij,t = Personal ocupado en la industria i en la ubicación j en el periodo t

= Personal ocupado en la industria i en la ubicación j en el periodo t-τ

Crecimiento del personal ocupado en la industria i en el resto de las zonas

metropolitanas

Salario del sector i en la zm j en el periodo inicial

Índice de concentración o especialización del sector i en la zm j en el periodo

inicial

Índice de competencia del sector i en la zm j en el periodo inicial

Índice de no-diversidad del sector i en la zm j en el periodo inicial

Participación de la zona metropolitana j en el empleo de la industria maquiladora

en el periodo t-τ

Variables dummy de región.

The construction of the agglomeration indicators is as follows:

 Specialization index: the relative share of sector i in location j versus the same sector’s share
in the national industry.

 Competition: Number of establishments per worker in industry i and location j with respect
to the number of establishments per worker in that same industry for the whole country

 Diversity: As a measure of diversity, the indicator suggested by Henderson et al. (1995) is

calculated, which is based on a Hirschman-Herfindahl Index (HHI). This index takes the sum
of the squares of the shares of all the other sectors in local employment.

Parameters:

 1 is the convergence parameter, widely used in growth equations, and captures the
effect of national demand on industry i.

 The parameters of interest are 3 - 7. The effect of wages (initial market conditions) on
labor demand growth, which is captured by 3 is expected to be negative.

 4 is the effect of specialization and is expected to be positive in the case of MAR or

Porter economies. is the competition parameter and under Porter or Jacobs economies it
should have a positive sign, while under MAR it should be negative or non significant.

 The effect of diversity, which is related to Jacobs economies should be captured by 6 and
its sign is expected to be negative as an increase of the HHI is related to a reduction in
diversity.

 Rj are regional dummies, which allow to control for region-specific policies or

transportation costs.

 Equation 1.1 is estimated for two values of : 1) , in which only information from the 2004
and 2014 censuses is used; and 2) , where data from the three last censuses is used (2004,
2009 and 2014).

Data

 The last three Economic Censuses (2004, 2009 and 2014) at the municipality level.

 MA level according to the INEGI/ CONAPO/SEDESOL MA classification of 2010 and 3

important MA´s for the maquiladora: Acuña, Agua Prieta and Nogales (following Soloaga
and Pereira, 2008).

 Regional Classification of INEGI

  OECD classification by technological intensity:

Medium-high
Low intensity Medium-low intensity intensity
Food, beverages Petroleum and coal Machinery and
and tobacco products equipment
Transportation
Textiles Plastic and rubber equipment
Non-metallic mineral Computers and
Apparel products electronics
Leather and Primary metal
footwear manufacturing and metal Electric apparatus
Wood Chemical products
Paper
Printing
Furniture and
mattresses

Other industries

Table 1. Share of MAs in Manufacturing Employment

Year Employment (%) of total
2004 3,153,100 76.3%
2009 3,353,160 76.7%
2014 3,728,279 74.7%
Source: Author’s calculations using data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 As Table 1 shows, restricting the sample to 58 MA’s (INEGI/ CONAPO/SEDESOL MA

classification of 2010), more than 70% of the manufacturing employment reported in
economic censuses is covered. 8 sectors of the manufacturing industry were considered.
Descriptive statistics

Figure 3. Kernel Density of Total Manufacturing Employment by MA

Source: Author’s calculations using data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 The density of employment by MA shows that most of the observations take low values.
That is, a few MAs concentrate most of the employment. It is necessary to control for the
initial value of employment.

Figure 4. Share of Manufacturing Employment by Technological Intensity

Source: Author’s calculations using data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 The shares of each intensity group in total employment have remained relatively
stable, indicating that even though low-technology sectors have the largest share,
the three groups have shown similar growth patterns.
 Figure 5. Growth in Employment Against the Rest of the Regions

North Capital

Gulf Pacific

South Central-north
Center
1 "Food, beverages and tobacco"
2 "Textile"
3 "Apparel"
4 "Leather and footwear"
5 "Wood"
6 "Paper"
7 "Printing"
8 "Petroleum and coal products"
9 "Transportation equipment"
10 "Computer and electronics"
11 "Electric apparatus"
12 "Machinery and equipment"
13 "Chemical industry"
14 "Plastic and rubber"
15 "Non-metallic mineral products"
16 "Primary metal manufacturing and metal products"
17 "Furniture and mattresses"
18 "Other industries"
Source: Author’s calculations using data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 Last figure shows whether employment in an MA is growing in the same sectors in which
national demand is increasing (using employment growth in the rest of the regions as a
proxy):
o For the north region, employment grows in the same sectors as national demand.
However, in the case of medium-high technology sectors, such as non-metallic
mineral products (11), machinery and equipment (13), computer and electronics
(14) and transportation equipment employment in this region is growing, while in
other regions there is little change.

Figure 6. Correlation Between Initial Wage and Employment Growth in the Industry MA
T=10 (2004-2014) T=5 (2009-2014)

Source: Author’s calculations using data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 Last figure presents the relationship between market conditions (initial wage) and
employment growth by sector-MA for the two different periods analyzed. The data exhibits
the expected negative relation (higher wages create negative incentives for the location of
industries in an MA).
 Table 2. Descriptive Statistics of Independent Variables
2004 2009 2014
Mean Std. Dev.CV Mean Std. Dev.CV Mean Std. Dev.CV
Whole Sample
Logarithm of employment 5.52 1.78 0.32 5.53 1.76 0.32 5.53 1.85 0.33
Logarithm of nominal
wage 2.24 2.39 1.07 2.26 2.50 1.11 2.45 2.60 1.06
Non-diversity index 0.28 0.15 0.52 0.26 0.13 0.50 0.27 0.14 0.51
Competition index 3.46 8.78 2.54 3.39 9.04 2.66 3.64 11.56 3.17
Specialization index 1.16 1.94 1.68 1.11 1.86 1.67 1.14 1.88 1.66
N= 845 886 880

Low technology
Logarithm of employment 5.17 1.68 0.32 5.19 1.59 0.31 5.08 1.67 0.33
Logarithm of nominal
wage 1.60 2.25 1.41 1.56 2.36 1.51 1.71 2.43 1.43
Non-diversity index 0.29 0.15 0.50 0.27 0.13 0.50 0.28 0.14 0.50
Competition index 2.79 3.98 1.42 2.27 2.80 1.23 2.58 3.57 1.38
Specialization index 1.08 1.89 1.75 1.04 1.78 1.71 1.06 1.91 1.81
N= 513 531 530

Medium-low technology
Logarithm of employment 5.74 1.38 0.24 5.83 1.45 0.25 5.92 1.49 0.25
Logarithm of nominal
wage 2.68 1.97 0.73 2.84 1.93 0.68 2.89 2.13 0.74
Non-diversity index 0.28 0.15 0.52 0.26 0.14 0.52 0.27 0.14 0.51
Competition index 1.68 1.28 0.76 1.96 2.63 1.34 1.88 2.05 1.09
Specialization index 1.02 0.85 0.84 0.98 0.82 0.83 1.01 0.86 0.86
N= 183 186 181

Medium-high technology
Logarithm of employment 6.45 2.15 0.33 6.27 2.23 0.36 6.53 2.24 0.34
Logarithm of nominal
wage 3.87 2.45 0.63 3.78 2.66 0.70 4.27 2.57 0.60
Non-diversity index 0.23 0.13 0.55 0.22 0.11 0.48 0.23 0.11 0.49
Competition index 7.92 18.92 2.39 8.50 19.12 2.25 8.84 24.92 2.82
Specialization index 1.61 2.83 1.76 1.49 2.69 1.81 1.54 2.47 1.61
N= 149 169 169
Source: Author’s calculations with data from the 2004, 2009 and 2014 Economic Censuses, INEGI.

 Last table presents that the specialization index by technological intensity does not show
much specialization (mean values close to unity), but standard deviations for each group
indicate that there are MAs where industries are overrepresented, especially for medium-
low technology industries.
Results

Table 3. SUR Estimates by Technological Intensity, T=10 (Standardized Coefficients)

 E1
Equation Obs Parms
RMSE R-sq chi2 P
growth_sam~N 790 6 2.13 0.08 71.72 0.00
<
Dependant variable: MA-industry employment growth Coef. Std. Err. z P>z [95% Conf. Interval]

Employment growth same industry in rest of the MAs 0.93 0.36 2.57 0.010 .2202898 1.64
Logarithm of nominal wage 2004 0.21 0.11 2.04 0.041 .0086275 0.42
Logarithm of employment 2004 -0.37 0.15 -2.56 0.010 -.6574525 -0.09
Specialization index 2004 -0.07 0.04 -1.62 0.105 -.1543953 0.01
Competition 2004 0.07 0.01 5.47 0.000 .0454312 0.10
Non-diversity 2004 -0.02 0.57 -0.04 0.966 -1.141558 1.09
_cons 1.99 0.62 3.22 0.001 .7786044 3.20

 E2
Equation Obs Parms
RMSE R-sq chi2 P
growth_sam~N 790 12 2.11017 0.1023 89.99 0.00

Dependant variable: MA-industry employment growth Coef. Std. Err. z P>z [95% Conf. Interval]

Employment growth same industry in rest of the MAs 1.03 0.36 2.87 0.004 .3256035 1.73
Logarithm of nominal wage 2004 0.18 0.10 1.75 0.080 -.0217531 0.39
Logarithm of employment 2004 -0.43 0.14 -2.99 0.003 -.7168635 -0.15
Specialization index 2004 -0.03 0.04 -0.75 0.456 -.1194534 0.05
Competition 2004 0.07 0.01 5.33 0.000 .0436191 0.09
Non-diversity 2004 -0.09 0.59 -0.15 0.884 -1.234102 1.06
Regional effects
Capital -0.85 0.37 -2.33 0.020 -1.565868 -0.13
Gulf -0.76 0.26 -2.96 0.003 -1.263576 -0.26
Pacific -0.52 0.30 -1.71 0.087 -1.105777 0.07
South -0.36 0.31 -1.17 0.240 -.9675906 0.24
Central-North -0.18 0.24 -0.73 0.465 -.6494479 0.30
Center -0.86 0.27 -3.23 0.001 -1.384271 -0.34
_cons 2.72 0.65 4.21 0.000 1.454004 3.98
  E3
Equation Obs Parms
RMSE R-sq chi2 P
growth_sam~N 790 13 2.10956 0.1028 90.5 0.00

Dependant variable: MA-industry employment growth Coef. Std. Err. z P>z [95% Conf. Interval]

Employment growth same industry in rest of the MAs 1.03 0.36 2.86 0.004 .3231691 1.73
Logarithm of nominal wage 2004 0.19 0.10 1.78 0.075 -.019008 0.39
Logarithm of employment 2004 -0.43 0.15 -2.94 0.003 -.710955 -0.14
Specialization index 2004 -0.04 0.04 -0.82 0.412 -.1234987 0.05
Competition 2004 0.07 0.01 5.36 0.000 .04401 0.09
Non-diversity 2004 -0.21 0.61 -0.34 0.730 -1.416828 0.99
Share of the MA in employment of the maquiladora
industry -2.66 3.92 -0.68 0.497 -10.34077 5.02
Regional effects
Capital -0.72 0.41 -1.76 0.078 -1.527223 0.08
Gulf -0.78 0.26 -3.01 0.003 -1.285207 -0.27
Pacific -0.50 0.30 -1.67 0.096 -1.094216 0.09
South -0.38 0.31 -1.22 0.224 -.9827759 0.23
Central-North -0.18 0.24 -0.75 0.450 -.6555502 0.29
Center -0.87 0.27 -3.27 0.001 -1.39789 -0.35
_cons 2.76 0.65 4.26 0.000 1.490093 4.03

 E4
Dependant variable: MA-industry employment growth Coef. Std. Err. z P>z [95% Conf. Interval]

Employment growth same industry in rest of the MAs 1.06 0.36 2.95 0.003 .356531 1.76
Logarithm of nominal wage 2004 0.18 0.10 1.75 0.080 -.0220948 0.39
Logarithm of employment 2004 -0.45 0.15 -3.08 0.002 -.7328991 -0.16
Specialization index 2004 -0.03 0.04 -0.57 0.568 -.1137908 0.06
Competition 2004 0.07 0.01 5.32 0.000 .0434282 0.09
Non-diversity 2004 0.03 0.64 0.04 0.966 -1.217157 1.27
Share of the MA in employment of the maquiladora
industry 11.16 10.16 1.10 0.272 -8.749055 31.07
Square of the share of the MA in employment of the
maquiladora industry -118.09 80.09 -1.47 0.140 -275.0701 38.89
Regional effects
Capital -0.47 0.44 -1.07 0.285 -1.342678 0.40
Gulf -0.70 0.26 -2.65 0.008 -1.215585 -0.18
Pacific -0.49 0.30 -1.62 0.105 -1.079645 0.10
South -0.30 0.31 -0.96 0.338 -.914095 0.31
Central-North -0.19 0.24 -0.77 0.439 -.6592844 0.29
Center -0.83 0.27 -3.08 0.002 -1.353989 -0.30
_cons 2.64 0.65 4.05 0.000 1.361568 3.92
  E5 (Regional effects and Competition deciles)
Equation Obs Parms RMSE R-sq chi2 P
growth_sam~N 790 21 2.03 0.17 158.73 0.00

Dependant variable: MA-industry employment growth Coef. Std. Err. z P>z [95% Conf. Interval]

Employment growth same industry in rest of the MAs 1.33 0.35 3.81 0.000 .6445102 2.01
Logarithm of nominal wage 2004 0.27 0.10 2.61 0.009 .066986 0.47
Logarithm of employment 2004 -0.40 0.14 -2.87 0.004 -.6784965 -0.13
Specialization index 2004 0.00 0.04 -0.06 0.953 -.090361 0.09
Non-diversity 2004 -0.67 0.60 -1.11 0.266 -1.854861 0.51
Share of the MA in employment of the maquiladora
industry -3.29 3.81 -0.86 0.388 -10.76339 4.18
Regional effects
Capital -0.58 0.40 -1.46 0.144 -1.361679 0.20
Gulf -0.87 0.26 -3.39 0.001 -1.374283 -0.37
Pacific -0.65 0.30 -2.15 0.031 -1.232714 -0.06
South -0.39 0.31 -1.29 0.199 -.9931045 0.21
Central-North -0.25 0.24 -1.05 0.293 -.7300545 0.22
Center -0.78 0.26 -3.01 0.003 -1.295546 -0.27
Competition deciles
2 0.11 0.33 0.35 0.728 -.5292484 0.76
3 0.36 0.34 1.06 0.287 -.3014106 1.02
4 0.35 0.35 1.00 0.319 -.3363625 1.03
5 0.42 0.36 1.15 0.252 -.2954089 1.13
6 0.38 0.37 1.02 0.306 -.3435518 1.09
7 1.21 0.38 3.16 0.002 .4625369 1.97
8 0.80 0.40 1.99 0.046 .0135961 1.58
9 0.72 0.41 1.78 0.074 -.0711996 1.52
10 3.32 0.45 7.35 0.000 2.433333 4.20
_cons 2.10 0.69 3.03 0.002 .7389739 3.45

Whole Sample (without sectoral or tech intensity breakdowns):

 The results for the variables of interest (wages and agglomeration economies variables)
are robust to the inclusion of regional effects, share of the MA in the maquiladora industry
employment and MAs fixed effects.
 Initial wages have a positive, though small, effect in employment growth for an MA-sector.
This is different for the results of Soloaga and Pereira (2008)
 Regarding the variables associated with knowledge spillovers, the specialization index that
indicates MAR or Porter economies is not significant in any specification.
 The competition variable is positive and significant, but low; if a sector increases its
competition level against country average by 10%, it will result in an increase of half a
percentage point in employment growth in the period of analysis.
 The non-diversity variable is non-significant in all specifications.
Robustness Checks
 Measurement error in the competition variable: INEGI does not report the number of
firms, when the number of firms is as small as to easily identify. Therefore, the number of
establishments for these cases is considered as cero. Thus, competition is underestimated.
In order to ensure the robustness of the results, equation (Table 3 E5) was estimated
including dummy variables for each decile of competition. Results indicate that employment
growth rate between 2004 and 2014 increases with each decile of competition, which shows
that results are robust.
 Regional labor markets: We check for the assumption of a national labor market. If the
ratios of regional wages against the north did not change much during the period of analysis,
the assumption of a national labor market, in which wages growth rates are the same, can
be made.

Figure 11. Nominal Wages by Region (North average wage=1)

Source: Author’s calculations with data from the 2004, 2009 and 2014 Economic Censuses, INEGI.
  Sectoral labor markets:

Figure 12. Sectorial Average Wage Growth 2004-2014

a. Sectorial Average Wage Growth 2004-2014
1 Food, beverages and tobacco
2 Textiles
3 Apparel
4 Leather and footwear
5 Wood
6 Paper
7 Printing
8 Petroleum and coal products
9 Chemical industry
10 Plastics and rubber
11 Non-metallic mineral products 12 Primary metal
manufacturing and metal products
13 Machinery and equipment
14 Computers and electronics
15 Electric apparatus
16 Transportation equipment
17 Furniture and mattresses
18 Other industries

b. Sectorial Average Wage per Worker Growth 2004-2014

1 Food, beverages and tobacco
2 Textiles
3 Apparel
4 Leather and footwear
5 Wood
6 Paper
7 Printing
8 Petroleum and coal products
9 Chemical industry
10 Plastics and rubber
11 Non-metallic mineral products 12 Primary metal
manufacturing and metal products
13 Machinery and equipment
14 Computers and electronics
15 Electric apparatus
16 Transportation equipment
17 Furniture and mattresses
18 Other industries

Still working on:

 Data: Checking the Non-diversity variable. Non-diversity was estimated taking the
HHI of the MA.
 Results with the whole sample: Non-diversity indicator: Correlation between the
Theil Index and the HHI
 Next Results: Technological intensity breakdown estimates.