No.1 CEIDIR’S REVIEW: Electronic Publication on Local Economic Development
“Regional Policies” and Income Disparities in Mexico: A Review of Convergence Aspects
Eduardo Rodríguez Oreggia*
*PHD Graduate
E-mail:
Back
Abstract
Since the beginning of the 1970s Mexico has experienced a change in its model of economic policy: until 1985 promoting import substitution, and after starting a period of economic reform and openness. Before 1985 Mexico experienced higher rates of growth, and disparities among regions were closing. However, the recurrent economic crises and the transition to economic reform and liberalisation have brought new problems. Among these problems is the widening gap in per capita GDP among the Mexican regions since 1985. In spite of this, Mexico still suffers from a lack of co-ordinated regional policies as such (OECD, 1997c).
The purpose of this paper is to make a brief review of the called regional policies in Mexico, but mainly to focus on aspects derived in convergence and divergence among the states in Mexico since the 1970s. The paper is organised as follows. The first section introduces the regional policies. The second part analyse the evolution of disparities and it is divided itself in three parts. The first subsection explains the data used. The second subsection analyse the convergence and divergence process. The third subsection considers the migration effect. Finally some conclusions are drawn.
I. Is there any regional policy in Mexico?
Although the OECD points out that in Mexico there are no regional policies as such (OECD, 1997), it does not mean that attempts to set a planning system in which regional characteristics play an important role never existed. All the studies about the regional policies in Mexico divide it in two periods. The first period is before 1970, and the second policies after 1970.
Before 1970 planning policies in Mexico were aimed towards economic planning without concern for regional development, moreover several schemes were successful in their own right (Aguilar, 1989). Previous to the 1940s most efforts were directed to encourage decentralisation and regional development, consisting in colonisation programmes with the immediate aim of favouring agriculture, and at the same time to foster colonisation in abandoned areas outside central Mexico (Palacios, 1989).
During the period 1940-1970 there was a higher commitment to achieve the objectives settled before, complemented with policies aimed to substitute imports to foster industrialisation, supplying with raw inputs and electricity. However, policies adopted to decentralise industry were limited in their scope and did not have any link to regional programmes.
The weight of the economic policies during the period 1940-1970 was directed to create conditions to achieve a faster industrialisation and higher economic growth. Nevertheless, the main feature of this period is the concentration of the industry in few regions, especially in Mexico City and the central region of Mexico. In a study of how government policies affected the pattern of industry concentration in the Mexican’s states, Hernández Laos (1985) shows that the incidence of the protective policies on the concentration indices is highly significant. He also found that the most geographically concentrated industries were those with high effective protective level.
At the beginning of the 1970s, the government was more conscious about regional disparities due to growing diseconomies of congestion in big cities. This congestion led to a rise in the production prices absorbed through subsidies, and to the increase of social costs in industrial agglomeration and higher costs in infrastructure provision (Palacios, 1989).
Three main programmes were implemented by the government: the industrial parks and cities programmes, including tax incentives; the regional implications included in the National Development Plans; and administrative decentralisation measures (Aguilar, 1999). Additionally, the maquila[1] programme was extended from its initial area of 20 kilometres from the border, to the whole country.
Nevertheless, the most successful industrial parks where located near to Mexico City, confirming that the programme benefited mainly to major cities in Central Mexico that were attractive to companies willing to move out of, but remaining close to, Mexico City (Aguilar, 1999). Moreover, the tax incentive scheme was not complemented with disincentives to agglomerations, perhaps because of the negative of government to impose any restriction to industrial location, contributing more to the agglomeration effect.
Therefore, nowadays we can realise that performance and concentration of industry between states has been deeply marked by the policies implemented in the past. As Aguilar (1989) analysed, most regional policy attempts have been incoherent, contradictory and partial, in conflict with non-spatial policies. Moreover, regional policy instruments did not consider the complex social structures that have being altered through time but considered regions as single homogeneous units. Thus, regional distribution was merely circumscribed to the raise of real income of larger segments of the population and not to additional geographical redistribution of resources at the aggregate inter-regional level (Aguilar, 1989).
II. Regional Disparities
II.1 Per capita GDP and data description
Economic disparities between regions are traditionally measured by reference to per capita GDP. This measure is said to be ambiguous given that just one part of a region’s GDP, composed of the sum of the total of its value added, goes to the income of the inhabitants in such region. Additionally, in a country like Mexico with a wide income polarisation, per capita GDP is at risk of not being representative.
Additionally, some analysts (OECD, 1997) point out the existence of a type of statistical fallacy in the case of Mexico using per capita GDP as indicator. This fallacy derives from the fact that oil and petrochemicals have a strong weight in the total GDP, while at the same time these sectors are highly localised (mainly in the states of Campeche and Tabasco), massive generators of value added, and with small impact on the economic development of the region in which they are localised. However, there is no choice in the use of per capita GDP since there are no alternative measures.
The database of state's GDP to be used in this work is from the National Institute for Statistics, Geography and Information (INEGI) available at www.inegi.gob.mx. The methodology for calculation of GDP was based on the input-output matrix for 1970 and 1980. In 1998 INEGI changed its methodology to measure state's GDP, updating the figures from year 1988 and subsequent years to 1993 base year. The reasons argued for such a change are the progress in technology and demand structure as some of the factors affecting the rise or extinction of economic activities. Therefore, there is a considerable variation in the weight of each sector in the generation of national wealth, phenomenon called structural change (Banamex, 1999).
With this updated database variations in each state’s GDP can be summarised in three groups (Banamex, 1999). In the first group there is an increase in the participation of northern states or with a market supported on a foreign market. INEGI argues that this modification mirrors in a more accurate way the effects of trade opening and export dynamic. In the second group the oil states experience a reduction in their weight, mainly due to changes in commercial activities related to this industry. Finally, the third group, including those states with a performance highly vulnerable to neighbours market, experiences a drop in their weight.
II.2 Regional per capita GDP growth
Map 1 shows the regional distribution of annual average growth rates of per capita GDP during the period 1970-1985. The upper group of states comprises a wide range of growth because the two states growing more than 10% in real terms are Campeche and Tabasco due to the oil boom in such period. The period showed is when the country was closed to free international trade. The map shows that, excluding the oil states, growth is concentrated in central states, and few of the Northern states, most of those next to the border with the US are in the lower group of growth.
Map 1
Regional distribution of per capita GDP growth 1970-1985
Map 2 shows the regional distribution of annual average growth rates of per capita GDP during the period 1988-1996. The period showed is after the crises of 1985 and 1987, where annual inflation very high. The country was now open to free international trade and northern states are comprised in the upper groups of growth. Oil producer states of Campeche and Tabasco experienced negative growth in this period. These growth rates are reflecting also the change of methodology of INEGI to calculate the GDP, giving more weight to Northern states because of free trade activity and less weight to oil producer states. Even that someone could argue that year 1996 could be replicating the 1994 crisis, 1996 marks the year of recovering in terms of GDP and the effects of the crisis were suppose to be behind (or almost) in terms of growth rates.
Map 2
Regional distribution of per capita GDP growth 1988-1996
Following the work of Barro and Sala-i-Martin (1991) on the pattern of growth at regional levels there have been flows of works interested in investigating the common speed at which economies converge to their own steady state. These ideas are supported by the neo-classical growth theory, supposing diminishing returns to capital.
The implication behind diminishing returns to capital in neo-classical models is that each addition to capital will generate a more than proportional addition in output when the capital is small, and small addition when the capital is large. Consequently, if the only difference across economies is the initial capital stock, poor regions (with small capital stock) will grow faster than rich regions (with large capital stock), creating a convergence effect.
The literature uses the b coefficient to measure the speed of convergence. There is b convergence if, on average, poor regions are growing faster than rich regions. In a cross-section set of areas, we can find b convergence if there is a negative relation between the growth rate of per capita income during a period and the initial level of income. The speed of convergence is estimated through the equation 1 (Sala-i-Martin, 1996):
Equation (1)
Where the right side is the average growth rate on the initial level of income of a set of regions between time t0 and t0+T. b coefficient is the absolute b convergence coefficient, without conditioning on any other characteristic of states. The model can be modified to include some variables to calculate conditional beta convergence.
Table 1 depicts the non-linear square results for equation 1. Oil producer states of Campeche and Tabasco have been excluded from the sample as they experienced an explosive rate of growth at the end of the 1970s and beginning of the 1980s as a consequence of the boom of the oil prices, then their inclusion in the sample creates distortion in the coefficients.
Table 1
Regression of b coefficient in the Mexican states
(Excluding Campeche and Tabasco)
1970-80 / 1970-85 / 1970-96 / 1985-96 / 1988-96a / 0.0782
(5.74) / 0.0639
(7.99) / 0.0147
(1.83) / -0.0707
(2.92) / 0.0019
(0.171)
b / -0.0192
(-2.53) / -0.0218
(-4.29) / -0.0015
(-0.40) / 0.0228
(3.06) / 0.00569
(1.266)
Thalf / 36 / 32 / 448 / 30* / 122*
R2 / 0.231 / 0.498 / 0.006 / 0.220 / 0.055
T (years) / 10 / 15 / 26 / 11 / 8
N / 30 / 30 / 30 / 30 / 30
*Correspond to the concept of Tdouble, meaning the number of years that it would take to
double the current gap.
t-statistics in parentheses.
According to the results, for the whole periods starting in 1970 the b coefficient is negative, indicating an absolute convergence process in per capita GDP across states, although is only 0.15% per annum in the period 1970-96, and the estimator is statistically insignificant, suggesting weak absolute b convergence. Figure 1 shows a scatter plot with the relation between the log of per capita GDP in 1970 and the average rate of growth during the period 1970-1996, excluding Campeche and Tabasco. From the scatter plot is not possible to establish a relationship between initial per capita GDP and growth in the period, in line with the lack of significance of the beta coefficient in Table 1.
Figure 1
The period before the liberalisation of the economy (1970-1985) shows a negative and significant beta coefficient, with the coefficient ranging between 1.9 and 2.2%, that is in tune with the international findings of rates of convergence around 2% (Barro and Sala-i-Martin, 1995). The period for 1970-85 shows especially a high R square, 49%. Figure 2 shows the scatter plot with the average annual rate of growth for the period 1970-85 and the per capita GDP in 1970, excluding Campeche and Tabasco. The scatter plot shows a negative relation, the lower income regions at the beginning of the period having a higher rate of growth in the period.
Figure 2
For the period 1985-96 the coefficient is statistically significant but with positive sign in Table 1, while in the period 1988-96 the coefficient is positive and statistically insignificant, that is during these periods is possible to find a process of b divergence, the gap between rich and poor regions tended to widen, opposite to the findings of convergence in the previous period 1970-85. Figure 3 displays the relationship between the per capita GDP in 1985 and the average rate of growth during the period 1985-96, excluding again Campeche and Tabasco. The points show a positive relation, the states with a higher initial income having higher rates of growth, while the states with lower initial income have even negatives rates of growth during the period.
Figure 3
Table 1 also displays the Thalf, also called half-life, indicates the number of years that would take to reduce by half the gap between the logarithm of the initial and the steady state GDPs and is calculated according to the formula Thalf = In(2)/b = 0.69/b. During the periods of positive growth, until 1985, the years estimated in Thalf is diminishing from 36 during 1970-80 to 32 during 1970-85, but it experiences a hug rise for the period 1970-96 to 448 years. For periods after 1985 and due to the negative sign of the estimator, Thalf corresponds to the number of years to double the original gap. As table 1 shows, the number of years is increasing from 30 during the period 1985-1996 to 122, meaning that it will take more time to close the original gap between rich and poor states.