1 Introduction
In
1990, the United Nations Development Program (UNDP) proposed an alternative measure of development, the Human Development Index (
HDI), a composite indicator of health, education, and income. Since then, researchers have increasingly used
HDI estimates to explain the long-run evolution of standards of living in different countries and periods. In this chapter, we use the
HDI to study long-run regional development in Mexico at subnational levels.
The literature using the
HDI to measure well-being over time is extensive. Crafts (
1997,
2002) calculated estimations for a set of 24 countries in the period 1870–1999, concluding that by the end of the twentieth century the gaps between Western Europe and developing regions were smaller than those present in the middle of the century. In a similar study, Prados de la Escosura (
2010) concludes that the gap between developed and developing countries has narrowed over time, although the degree of convergence has slowed since 1970. Astorga et al. (
2005) construct an
HDI series for Latin American countries and conclude that the region experienced the largest gains in living standards in the middle decades of the twentieth century. Bértola et al. (
2010) compute an inequality-adjusted historical human development index for Argentina, Brazil, and Uruguay in the twentieth century. These authors compare the results to developed countries (France, Germany, the UK, and the USA) and find divergence in Argentina from the early decades of the twentieth century, while Uruguay began the same pattern in the 1950s. Brazil was steadily catching up until 1980, when it finally began to diverge from the core countries.
1 It is worth noting that these papers compare the indices
across countries over time, and that little research exists on well-being
within countries using
HDI estimates.
In Mexico, the study of living standards has been developed using different approaches, considering income and non-income components. Bortz and Águila (
2006) analyze real wages from the Porfiriato (1876–1911) to the early twenty-first century and conclude that the largest increases took place between 1952 and 1970, while the economic crises of the 1980s and 1990s caused a drastic reduction in real wages. López-Alonso (
2007,
2012) uses anthropometric measures (adult heights) drawn from military and passport data. These studies identify a deterioration in the standard of living of the laboring classes over time, in contrast to an increase for the elite and upper middle classes.
2 Recently, Challú and Gómez (
2015) found that real wages of non-skilled workers in Mexico City exhibited a secular decline since the mid-eighteenth century and only partially recovered from the late 1860s to 1930. Esquivel (
1999) finds convergence in
GDP per capita at state level from 1940 to 1960, and a relative constant dispersion thereafter.
Campos-Vazquez and Velez (
2012) calculate a human development index to study human development at the state level from 1895 to 1910. Their results show important improvements in the 1895–1910 period, but these improvements are not distributed proportionately among states. Márquez (
2012) applies the methodology developed by these two authors to study state-level human development in 1930. Her study shows an improvement in human development in comparison to the 1895–1910 period. Interestingly, the regional development pattern found by Campos-Vazquez and Velez is also found in this study, suggesting an important structural difference in development between the northern and southern regions of the country.
In this chapter, we analyze the development of Mexico over the long run (1895–2010) using an
HDI measurement at the subnational level to facilitate a comparison across regions. Our methodology provides a consistent standard of living measure based upon proxy variables to those generally used for recent periods.
3 The chapter is organized as follows. Section
2 discusses the historical background on Mexico’s history since 1895 and the literature that have applied human development concepts to the Mexican experience. Section
3 presents the methodology and data sources used in our estimates. Following, in Sect.
4, we offer main results of our study. Section
5 summarizes our findings and offers some conclusions.
2 Historical Background
Mexico experienced significant economic changes along the twentieth century that enabled the country to overcome past obstacles to growth while facing new challenges to improve the well-being of the population. In the last quarter of the nineteenth century Mexico grew at a sustained rate, in sharp contrast with the performance of the first 50 years after independence. Indeed, from 1821 to around 1870, Mexico stagnated, with episodes of weak recoveries and decline.
4 Although economic historians have debated the pace of the slow secular growth, there is consensus on the recovery from the 1870s to the outbreak of the Mexican Revolution in 1910. During the Porfiriato (1876–1911) railroad construction, foreign demand for mineral and agricultural products, and enactment of regulations to secure property rights all contributed to the modernization of crucial sectors of the economy.
By 1910, Mexico was a wealthier economy than 30 years earlier. Ports, major cities, mines, and industrial centers were all connected by railroads, larger banks had branches throughout the country, and internal taxes increased their share of total federal revenue. However, economic growth did not reach all regions of the country with the same intensity. Mining districts, industrial centers, and export-oriented agriculture became engines of growth, but they coexisted with low-productivity agriculture and isolated regions with few connections to international markets.
The Mexican Revolution had uneven consequences on the economy. The areas disputed by different leaders were hit hard by the civil war. But even in this situation economic production did not come completely to a halt: the production of foodstuffs continued, in many instances promoted by revolutionary leaders; oil fields increased their output in response to growing international demand; and the geographic isolation of the Yucatan peninsula allowed henequen haciendas to supply the American market. By the end of the armed phase of the revolution a new institutional arrangement, centered in the 1917 Constitution, set in motion a land reform program and changes in labor legislation. Also, the government acquired a more interventionist stance in the economy which shaped social and economic policy in the decades to follow.
The economic disturbances caused by the revolutionary upheaval were neither generalized nor permanent, but their consequences differed across regions and sectors. In the 1920s mineral and agricultural exports continued to drive economic growth. New investment in manufacturing also contributed to the recovery of the economy while the banking system slowly regained its basic functions. In contrast, Mexico was in arrears, without access to foreign loans, and in difficult negotiations with its creditors. Private investors were more reluctant to channel resources to agriculture than to manufacturing and mining, in part because of uncertainty created by the land reform program implemented by postrevolutionary governments.
Also, certain policies changed radically in the new institutional environment. In restructuring the economy the federal government introduced direct taxation (income tax), created the central bank, and established development banks in support of agriculture and industry. The provision of education and health services became priorities for the federal and state governments. Vaccination campaigns, eradication of epidemics, and sewage projects gained support among public officials. The School of Public Health was founded in 1922, and national health campaigns emerged as part of the social policy of the 1920s. The establishment of the Ministry of Public Education (
Secretaría de Educación Pública,
SEP) included an ambitious literacy program and the establishment of rural schools, and the federal government increased its efforts to centralize education policy (Fierro et al.
2009).
By 1926, as foreign demand for Mexican products tapered off, an economic slowdown was evident, and the international crisis in 1929 aggravated an already delicate situation. A sharp decline in international prices and the collapse of external markets constrained Mexico’s capacity to grow, and that of other countries in Latin America, until 1932. However, the resumption of growth in that year initiated a 50-year period of sustained economic expansion. The closure of foreign markets during the 1930s opened up favorable conditions for the production of consumer and intermediate goods, while the early recovery from the external shock guaranteed domestic demand. Under the Lázaro Cárdenas administration (1936–1940) social expenditure soared and land reform policy broadened. Although the pace of land distribution leveled off subsequently, the government performed an active role in the economy through public investment in infrastructure, provision of cheap energy, and subsidies of raw materials and intermediate goods. Industrialization reached its stride with protectionist policies, fiscal incentives, and restrictions on foreign capital. The 1950s and early 1960s were characterized by low inflation, exchange rate stability, and average GDP yearly growth of more than 6%. In addition to high growth rates and price stability the expansion of health services and public education all contributed to improvements in social welfare and a betterment of income distribution.
In the mid-1960s the so-called Mexican miracle showed the first signs of profound economic woes. A growing trade and public deficits, lack of competitiveness of the industrial sector, and falling productivity in agriculture reveal serious obstacles to sustain sound economic growth in the years to come. By the early 1970s, structural problems mounted. The government chose fiscal expansion as the means to deal with its economic difficulties, resulting in rising inflation, an exploding public deficit, and an external deficit that required increased foreign borrowing. Inflation and devaluation ensued. The discovery of rich oil deposits coupled with the rise in international oil prices changed the prospects for the Mexican economy at the end of the 1970s. Growth resumed without resolution of the structural problems. Foreign debt was initially devoted to building an oil export platform, but the overvaluation of the peso and the deindustrialization associated with the so-called Dutch disease soon forced the government to increase borrowing. In 1981, international oil prices declined and interest rates rose, limiting the possibility of foreign financing. A year later, neither the devaluation of the peso nor the exchange rate controls had proven sufficient and, in consequence, the government declared a 3-month moratorium on the external debt and the nationalization of private banks. Mexico shared with other Latin American countries the problem of not only overindebtedness, but also an industrial base dependent on imported inputs, overprotected domestic markets, and a lack of international competitiveness. Severe cuts in public investment, a reduction in bureaucracy, elimination of subsidies, and price controls failed to recover a path of sustained growth or control inflation. Overall, in the 1980s, growth was mediocre.
5
In the midst of the external debt crisis Mexico took a risky gamble. Instead of tightening the protectionist barriers, President Miguel de la Madrid (1992–1998) decided to liberalize the economy. A first step towards this goal was to join the General Agreement on Tariffs and Trade (
GATT), which required the elimination of quantitative import controls and the lowering of import duties. To avoid future policy reversals, President Carlos Salinas (1988–1994) promoted the North American Free Trade Agreement (
NAFTA) with the USA and Canada. Opening up the economy had initially aimed at increasing the flow of foreign direct investment; yet actual capital flows to the Mexican economy fell short of expectations (both in the mid-1980s and in the early 1990s). However, liberalization became the cornerstone for a new model of growth, one that placed significant emphasis on exports as the driving force of the economy. Completing the shift to this model also required a change in the role of the government. The sale of public enterprises at the start of the foreign debt crisis began as a measure to reduce the public deficit, but later became one of the pillars supporting a shift to a market-oriented economy.
6
The changes in the Mexican economy were so profound that some authors did not hesitate to describe them as a revolution.
7 In addition to
NAFTA, President Salinas (1988–1994) launched ambitious economic reforms including the landholding regime of the
ejido to allow peasants to sell their plots, recasting the Bank of Mexico as an autonomous agency with a mandate to control inflation, passage of an antitrust law with establishment of a Federal Competition Commission, as well as privatization of banks, steel plants, the telephone company, television stations, and railroads, among other industries. However, these reforms were insufficient to recover high rates of growth needed to compensate for the “lost decade” of the 1980s.
8
A currency crisis in late 1994 led the Mexican economy to a severe drop in output equivalent only to that of the 1930s. Despite a quick recovery, the effects on well-being were not easy to overcome. After a decade of mediocre growth in the 1980s, the new crisis deepened the vulnerabilities of the middle- and low-income strata of society. In the first decade of this century, social programs coupled with macroeconomic stability had ameliorated the effects of years of slow growth. However, the international financial turmoil of 2008 caused another large dip in the growth trajectory, losing the gains in well-being brought about by social programs.
Over the course of more than a century the Mexican economy experienced a profound transformation. At the end of the nineteenth century the majority of the population lived and worked in rural areas, the railroad grid was crucial to the transportation of mineral and agricultural exports, and education and health services were limited and provided mostly in urban areas. In 2010, most Mexicans lived in cities, most children had access to primary education, and the country’s export basket combined raw materials and manufactures. Between these two moments of Mexican history, the country changed its patterns of growth from a primarily export model to an import substitution industrialization strategy, and finally to a market-oriented economy. How did well-being develop over time? Can we observe any continuity in regional patterns? These questions are addressed in the following sections, where we present the results of estimates of human development indexes from 1895 to 2010.
3 Methodology and Data
In its original version the
HDI was calculated as a simple average of the health, education, and income indexes. The health index is related to life expectancy, the education index to enrollment and literacy rates, and the income index to per capita income. Since 2010, the
HDI has been calculated as a geometric mean of its components:
$$ \mathrm{H}\mathrm{D}\mathrm{I}={\mathrm{Health}}^{1/3}\cdot {\mathrm{Education}}^{1/3}\cdot {\mathrm{Income}}^{1/3} $$
(1)
This new methodology has several advantages. First, it avoids the perfect substitutability problem of the simple average aggregation. In other words, if one subindex gains at the expense of another, the previous formula would not reflect a changed
HDI, while the new formula does. Second, as a consequence of imperfect substitutability, uneven progress can be accounted for as a negative effect on the overall index. Finally, the maxima chosen in the normalization of each subindex are irrelevant (Klugman et al.
2011).
9
In order to construct the
HDI in Eq. (
1), we use proxy variables for each subindex following Campos-Vazquez and Velez (
2012) and Ian Morris (
2013). First, the calculation of the education index takes into account the variables and weights proposed by UNDP (
2010). The only difference is that we restrict the upper bound of the age range of literacy rates for individuals to 12 years instead of 24 years as specified by UNDP (
2010). We thus calculate the education index as a weighted average of literacy rates (LR) for individuals older than 15 years of age and enrollment rates (ER) for individuals aged 6–12 years such that
$$ \mathrm{Education}=\frac{2}{3}\mathrm{L}\mathrm{R}+\frac{1}{3}\mathrm{E}\mathrm{R}. $$
Second, instead of using life expectancy, we use the number of physicians for each 10,000 inhabitants. Our proxy variable is a good measure of the health standard if we considered the effect of access to medical services in overall health. There is a large range of studies documenting the importance of the number of physicians on the health status of the population. For example, Shi (
1992,
1994) has shown that US states where the number of physicians per capita is higher have better health outcomes. This is in line with results of Robst and Graham (
1997,
2004) also for the USA. They show a positive influence of the number of physicians on the health status of the population. Moreover, they find a higher correlation in rural areas across the USA than urban areas. Vogel and Ackermann (
1998) have shown a positive relation between a higher supply of physicians and life expectancy. Finally, Pièrard (
2009) shows that a higher supply of general practitioners in Canada is correlated with better health outcomes, including self-reported measures of health status.
Our proxy variable for life expectancy, the number of physicians per 10,000 inhabitants, is a good measure of the health standard if we consider the effect of access to medical services on overall health.
10 To construct this subindex, we use the number of physicians (NP) per 10,000 inhabitants relative to 35, which was the corresponding figure for 2001 in Switzerland, the country in Europe with the highest life expectancy. The health index is thus constructed as
$$ \mathrm{Health}=\frac{\mathrm{NP}}{35}. $$
Third, instead of using income, we use urbanization rates. There is a long tradition in economic history of using urbanization rate as a proxy for economic development. Paul Bairoch (
1975,
1988) presents a compelling argument that urbanization is correlated with economic development for both developed and developing regions. A book by Komlos (
1994) reaches a similar conclusion. Moomaw and Shatter (
1996) formally test the previous argument using a panel data of countries and find that indeed urbanization increases with
GDP per capita. More recently, Acemoglu et al. (
2002) convincingly argue that population density and urbanization are a good proxy for economic development. For a sample of Latin American countries, Astorga et al. (
2005) find that urbanization is positively correlated with living standards in the period 1900–2000. In the study of how culture affects economic development in Europe, Tabellini (
2010) argues that urbanization rates are a good proxy for economic development. Nunn and Qian (
2011) argue that the introduction of potatoes to the Old World improved nutrition and agricultural productivity which lead to higher urbanization rates and higher economic development. Finally, Cali and Menon (
2013) find that urbanization causes rural poverty to decline through increased demand for agricultural products in the case of Indian districts. Furthermore, disaggregated urbanization data for developing countries is generally more easy to find whereas
GDP per capita estimates were often inexistent or lacked precision as the analysis moved back in time. In so doing we attempt to provide a basis of comparison with cases or historical periods with little information. Nevertheless, we should keep in mind that urbanization is a good proxy for income in the long run, but urbanization may not follow the same short-run fluctuation as income. The positive results of the consistency test (see next section) support our choice of urbanization as a proxy for income in the long run. As urbanization is already a number between zero and one, we do not need to transform it into an index; we simply introduce it in Eq. (1) as proxy of the subindex income.
Table
1 summarizes the definitions of the subindices explained above. In sum, as a proxy for per capita income we use urbanization rate. For health, we use the number of physicians per 10,000 inhabitants. For education, we can closely mimic the official definition in the HDI calculation.
Table 1
Components of HDI as in UNDP (
2010) and variables used in the study
Income |
GDP per capita | Urbanization rate |
Education | Enrollment rate of individuals older than 15 years | Enrollment rate of individuals older than 15 years |
Literacy rate in individuals aged 6–24 years | Literacy rate in individuals aged 6–12 years |
Health | Life expectancy rate | Physicians per 10,000 inhabitants relative to 35 (physicians in Switzerland in 2001) |
3.1 Construction of the Education Index
Education variables are the best measured variables in our dataset. Literacy rates for the population aged 15 years and older are available at the state level for every year of interest in
Estadísticas Históricas de México (INEGI
2009) and
Censo General de Población y Vivienda (INEGI
2010a,
b). Enrollment rates are defined as the share of population aged 6–12 years who are enrolled in school. From 1920 to 2010, enrollment rates are calculated using data available in the population censuses. Rates from 1900 to 1910 are calculated using data from
Estadísticas económicas del Porfiriato (Colegio de México
1964) and
Estadísticas sociales del Porfiriato (Secretaría de Economía
1956). Only the enrollment rates from 1895 have to be estimated.
11 Both literacy rates and enrollment rates are also calculated using microdata from the Population Censuses from 1990 onwards provided by IPUMS (Minnesota Population Center
2011).
It would be preferable to use enrollment rates for the population aged 6–24 years, but there is not enough data available for 1920, 1930, and 1940 to calculate them. Correlations between enrollment rates for persons aged 6–12 years and those aged 6–24 years for 1950, 1970, and 1990 are 0.98, 0.96, and 0.81, respectively.
12
3.2 Construction of the Health Index
The number of physicians per 10,000 inhabitants is the best proxy variable that can be used for the whole period of study. Although the origin of the data is varied, the number of physicians per state is available for most of the years under study. The number of doctors at the state level is available in
Estadísticas Sociales del Porfiriato (Secretaría de Economía
1956) for the period 1895–1910, in Mendizábal (
1947) for 1930, in Huerta Maldonado (
1960) for the mid-1950s, in N. Myers (
1971) for 1960, and in Frenk et al. (
1995) for 1970 and 1990. Numbers for 1980were interpolated from the latter source, and yearbooks published by the National Statistical Institute (
INEGI) provided the observations from 2000 to 2010 in a more homogenous manner.
13
There is no data available for the number of physicians at the state level for 1920 and 1940, so we impute these values using econometric methods. Trends of physician concentration were clearly different before and after 1930, so two estimations are needed. The number of physicians for 1920 is estimated using data from 1895, 1900, 1910, and 1930, and the number for 1940 using data from 1950, 1960, and 1970.
14
Statistical evidence also supports the proposition that the number of physicians is a good proxy variable for the health dimension. The correlations between the number of physicians and life expectancy for 1950, 1990, and 2000 are 0.78, 0.61, and 0.54, which implies a strong positive relation.
15 Furthermore, we expect that the relationship between physicians and life expectancy should be stronger for earlier years of the study, given that transportation costs were higher.
3.3 Construction of the Income Index
The urbanization rate is defined as the share of the population living in communities with more than 2500 inhabitants.
16 We need to rely on an estimation of the urbanization rate only for the year 1895.
17 Data available in
Estadísticas económicas del Porfiriato (El Colegio de México
1964) is used to calculate urbanization rates for 1900 and 1910, and population census data is used for the period 1920–2010. Statistical evidence also supports that urbanization is a good proxy variable for the income dimension; correlations between urbanization and per capita income for 1970, 1990, and 2000 are 0.76, 0.78, and 0.84, respectively.
18
5 Conclusions
In this chapter, we estimate a measure of long-run development in Mexico by calculating a human development index from 1895 to 2010 at subnational level. This is the longest homogenous series to date that tracks this phenomenon at the national and state levels. We discuss Mexico’s well-being in terms of multidimensional characteristics, but we also focus on measurements at the subnational level in order to disentangle regional patterns.
We find that there is a significant increase in human development over the period studied. We identified absolute convergence in the
HDI and each of its components with a trend of sharp convergence in the period 1940–1980. Our index begins at 0.13 in 1895 and ends at 0.70 in 2010, representing an acceleration rate of 1.45 %. For the period from 1895 to 1910, well-being grows at a rapid rate but slows down by the end of the period. For the revolutionary and postrevolutionary period (1910–1930), standards of living increase, although the increase is not homogenous across regions. The largest increases in human development are from 1940 to 1980. From 1950 onward, the gap across states decreases substantially, and after 1980, disparities increased again. Health contributes close to 65 % of total growth before 1930, and 55 % thereafter. These patterns of results are similar to those found by Prados de la Escosura (
2010) in an international sample.
We also find a large persistence in regional development patterns across states. Northern states are wealthier than the rest from the beginning of the period, while southern states are the poorest. States surrounding Mexico City were as poor as southern states at the beginning of the century, but they developed more rapidly from 1940 to 1980. The expansion of the welfare state thus failed to modify the spatial distribution of relative levels of well-being in the country.
In sum, this chapter makes an important contribution to the long-run measurement of well-being in Mexico. Future studies for other countries may take advantage of our methodology to analyze regional patterns, and deepen our understanding of the determinants of change in well-being across regions within countries.