México\'s demographic transition: Public policy and spatial process

M~xico's Demographic Transition: Public Policy and Spatial Process Brian J. L. Berry L. Shane Hall Rodolfo Hernandez-Guerrero Patricia H. Martin 1 The University of Texas at Dallas

A logistic growth equation is used to model M~xico's epidemiological and fertility transitions, creating variables used to model the spatial diffusion of demographic change across the states. Consistent with the goals of the L~zaro C~rdenas administration, the epidemiological transition unfolded uniformly across the states, accessible to rich and poor alike, but the urban-oriented family planning programs introduced by Luis Echeverria have favored elites, have diffused selectively, and have ensured that the burdens of the population explosion have borne down most heavily on the poor and the remote.

INTRODUCTION During the demographic transition modernizing societies move from high fertility-high mortality conditions to low fertility-low mortality conditions through a transitional period of population explosion. The explosion occurs because declines in death rates (the epidemiological transition) precede declines in birth rates (the fertility transition), producing rising, then declining, rates of natural increase. Casual inspection of Figure 1 suggests that M~xico is following this path. The initial condition was what Wrigley Please address correspondence to Brian J. L. Berry, University of Texasat Dallas, Social Sciences--UTD GR 31, Richardson, TX 75083-0688. Population and Environment: A Journal of Interdisciplinary Studies Volume 21, Number 4, March 2000 9 2000 Human Sciences Press, Inc.

363

364 POPULATION AND

ENVIRONMENT

60

[

LUIS

I ECHEVERRtA ~ADMINISTRATION I

50

eo

40.

O. 0 Q. 0 30. 0

-

I

I

c~~

't

-

2O

lf)

[ 0

~.,. R.,e / .

.

.

.

.

.

.

.

.

.

.

.

.

.

.

,

,

,

,

FIGURE 1. M~xico's demographic transition.

and Schofield (1981) have termed a "high-pressure equilibrium." Before 1930 there was a sufficient excess of fertility over mortality to press the bulk of the population close to the margin of existence, with episodic reductions in the population caused by sharp mortality spasms. One such spasm is evident in the graph, the M~xican Revolution of 1910-1917 followed by the influenza pandemic of 1918-1920. During this traumatic decade M~xico's population declined by 1 to 1.5 million people. M~xico's epidemiological transition began in the 1930s, and was substantially complete by the early 1970s, by which time the death rate had fallen to 7.5, the level typical of the world's leading urban-industrial economies during the 1960s. The fertility transition did not begin until the 1970s. Between 1930 and 1970, the growing gap between births and deaths drove the rate of natural increase to 3.8 percent annually. With the onset of fertility decline, the rate of natural increase has been pushed down to 2.8 percent. Had the death rate remained at 7.5 after 1970, the rate of

365 B. J. L. BERRY, L. S. HALL, R. HERNANDEZ-GUERRERO, AND P. H. MARTIN

natural increase would have been less than 2.8. Because of the population explosion, however, a large proportion of the population is very young, reducing the crude death rate to 4.9. 2 Reductions in the rate of natural increase in the next several decades therefore will depend upon fertility decline, because increases the death rate are unlikely before the birth cohorts of the 1950s and 1960s reach the end of their lifespans. Thus far, the account is unexceptional. What is distinctive is that in the M~xican case, neither the epidemiological nor the fertility transitions emerged naturally from an ongoing process of economic modernization. Rather, each was initiated by political action, the first by the programs of the L~zaro C~.rdenas administration, 1934-40, and the second by the administration of Luis Echeverria, 1970-76 (Solis, 1981 ). The nature and implementation of these programs appear to have had a significant impact on the spatial diffusion of the demographic transition across the states of M~xico. It is the resulting differences in the spatial diffusion process that are the focus of what follows. Figure 2 maps the course of M~xico's epidemiological transition from 1930 to 1990, by state. In 1930, the pattern was capped by a ridge of very high rates extending south from Coahuila to Jalisco, thence eastward to Oaxaca, Chiapas and Yucatan. As the death rate declined, this ridge persisted. Rather like an upturned canoe slowly sinking into a lake, death rates declined at a uniform pace, and the ridge was the last feature of the spatial pattern to disappear, between 1970 and 1980. Figure 3 provides a similar glimpse of the first two decades of the fertility transition. Neither the initial pattern nor the spatial pattern of change resemble that of mortality decline. What quickly emerged was a distinction between the northern border states and the rest of the country. What forces could be at work? We turn for clues to two bodies of theory. The first, temporal, deals with the process whereby objects are transformed from one state (A) to a second state (B), and leads to the conclusion that such transformations can be modeled by sigmoid (logistic growth) curves. The second, spatial, views the new conditions (B) as innovations, and seeks to unravel the process by which they diffuse geographically during the course of the A--~B transition.

THE TRANSFORMATION PROCESS The process by which transformed conditions B are substituted for initial conditions A is well known. The basic idea is that transformations are

FIGURE 2. M~xico's declining death rate by state, 1930-1990.

FIGURE 2.

(Continued)

FIGURE 3. Initial declines in the M6xican birth rate, by state, 1970-1990.

369 B. J. L. BERRY, L. S. HALL, R. HERNANDEZ-GUERRERO, A N D P. H. MARTIN

propagated by chain reactions and that the rate of reaction is dependent on the relative sizes of the active masses of the reacting conditions A and B. 3 Let PA be the proportion of the mass that is A, and PB the proportion that is B. Then PB = 1 - PA, and the rate at which A is changed to B is -dpA/dt

= d p B / d t = kpB(1 - Ps)

where k is a rate constant that measures the pace of transition. Integrating and setting t = to.s when PB = 0.5 gives In [PB / (1 -- PB)] = - k (t - t0.s) In [PA/(1

-

PA)] -- k (t

-

to.s)

These relations lead to a sigmoid (logistic growth) curve if p is plotted against t, and a linear relationship for In (PB / (1 -- PB)] plotted as a function of t: In [PB / ( 1

-

PB)] =

a + kt

At no time is PB either zero or infinity. There must be some B present for it to progress, and some of condition A must remain for the process to continue.

Application to the M6xican Case How might such a model be applied to the M~xican epidemiological and fertility transitions, where what is observed is the death rate or the birth rate? We begin by postulating that initial conditions (A) are the death rates or birth rates that existed in 1899, prior to transition. The transition is assumed to be towards conditions (B) that existed in the world's urbanindustrial societies that had already moved through the demographic transition by the 1960s: death rate =7.5 and birth rate =10.0 (= fertility rate of 1.2). Now follow through the calculations in Tables 1 and 2. 4 The first column records year and the second column records time t in years from 1899 to 1993. Column three sets down the actual M~xican death or birth rates in the years specified, and column four the cumulative reduction in death or birth rates from 1899 to 1993. For example, the last number in column four of Table 1 is 29.5, or the initial death rate (34.4) minus the ending death rate (4.9). Column five plots the cumulative reduction in col-

370 POPULATION AND ENVIRONMENT

TABLE 1 Fitting a Logistic Curve to a Declining Death Rate-M~xico 1899-1993 Year 1899 1904 1909 1914 1919 1924 1929 1934 1939 1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1993

t

Death Rate

0 34.4 5 33.4 10 32.9 15 (46.6) 20 (48.3) 25 28.4 30 26.7 35 26.7 40 23.5 45 21.8 50 17.8 55 15.4 60 12.5 65 10.4 70 9.4 75 8.6 80 6.9 85 5.8 90 5.1 94 4.9

Cumulative DR Reduction

Cumul. Reduction as Prop. of Maximum (PB)

In[pB/(1-pB)]

1.0 1.5

0.037 0.056

-- 3.259 --2.825

6.0 7.7 7.7 9.9 12.8 16.8 19.2 22.1 24.0 25.0 25.8 27.5 28.6 29.3 29.5

0.223 0.286 0.286 0.368 0.475 0.624 0.713 0.821 0.892 0.929 0.959 0.999 0.999 0.999 0.999

- 1.248 --0.915 -0.915 -0.541 -0.100 0.507 0.910 1.523 2.111 2.571 3.152 6.907 6.907 6.907 6.907

umn four as a proportion of the reduction required to achieve the postulated urban-industrial conditions of the 1960s: 34.4 --> 7.5 = 26.9. Thus, by 1944 the cumulative death rate reduction of 12.8 was 0.475 of 26.9, revealing that almost half the active mass had been converted from condition A to condition B. In the same year, Table 2 shows that the birth rate reduction of 3.5 was barely 0.087 of the needed 47.3 -~ 10.0 of the fertility transition. In computing Ps, we skip over the rate reversals of the M~xican Revolution and the succeeding influenza pandemic, removing 1914 and 1919 from our calculations. From 1979 onward, the cumulative proportional death rate reduction is coded 0.99: because of the population explosion, the urban-industrial death rate had been overshot, but we did not want the overshoot to be part of the analysis. Column six converts column five (PB) to the dependent variable In [PB /

371 B. J. L. BERRY,L. S. HALL, R. HERNANDEZ-GUERRERO,AND P. H. MARTIN

TABLE 2 Fitting a Logistic Curve to the Declining M~xican Birth Rate 1899-1993

Year

t

1899 1904 1909 1914 1919 1924 1929 1934 1939 1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1993

0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 94

Birth Rate

Cumulative BR R e d u c t i o n

47.3 46.5 46.0 (43.2) (40.6) 45.3 44.3 44.1 43.5 43.8 44.5 45.0 45.8 44.1 43.7 43.4 36.3 34.2 32.8 32.6

C u m u l . Reduction as Prop. of M a x i m u m (PB)

In [pJ(1-pB)]

0.8 1.3

0.020 0.032

- 3.891 - 3.409

2.0 3.0 3.2 3.8 3.5 2.8 2.3 1.5 3.2 3.6 3.9 11.0 13.1 14.5 14.7

0.050 0.075 0.080 0.095 0.087 0.070 0.057 0.037 0.080 0.090 0.097 0.276 0.329 0.364 0.369

-2.944 -2.512 -2.442 -2.254 -2.350 -2.586 -2.806 - 3.259 - 2.442 -2.313 -2.231 -0.964 - 0.712 -0.558 - 0.536

(1 - Ps)]. Fitting the logistic e q u a t i o n In [PB / (1 -- PB)] = a + kt to the data in c o l u m n s 6 and 2 of Table 1 yields the expression In [ P B / ( 1

-- PB)] =

--4.84 + 0.121t

w i t h both coefficients significant at the 0.01 level and the adjusted R2 = 0.91. W h e n e x p l o r i n g logistic t r a n s f o r m a t i o n processes, it is c o n v e n i e n t to define " t a k e o f f " as the year in w h i c h PB = 0.1, "saturation" w h e n PB = 0.9, and the " d i f f u s i o n t i m e " At as to.9 - to.l:

Takeoff

In [0.1 / ( 1 - 0.1)] = a + kto.1 t0.1 = [In 0.111 - a ] / k = [ - 2 . 1 9 7 - a] / k

372 POPULATION AND ENVIRONMENT

Saturation

to.9 = [In9 - a ] / k = [2.197 - a ] / k

Diffusion time

At = to.9 to.1 = 4.394 / k -

Takeoff of the M~xican epidemiological transition was to.~ = [ - 2.197 - ( - 4 . 8 4 ) ] / 0 . 1 2 1 = 2.643/0.121 = 21.8 or 22 years after 1899 = 1921. Similarly, saturation was reached at [2.197 - ( - 4 . 8 4 ) ] / 0 . 1 2 1 = 58.1 years after 1899 = 1957, and diffusion time was 4.394/0.121 = 36.3 years, or approximately two cohort generations of + 18-19 years. Because the fertility transition began so much later than the epidemiological transition, there are very few years for which birth rate reductions are observable in Table 2, making it difficult to fit the logistic equation by statistical means. Instead, we rely on the linear relationship between In [PB / (1 - PB)] and t and argue that knowledge of any two points on the line should permit algebraic computation of k. In 1974 the cumulative birth rate reduction was 0.097 and In [ P B / (1 -- PB)] was --2.231. In 1993 the numbers were 0.369 and - 0.536. In 19 years, the dependent variable had declined from -2.231 to - 0 . 5 3 6 , or 0.089 per year. Therefore k = 0.089. This yields a projected diffusion time of 4.394 / 0.089 = 49.4 years from takeoff c.1974. If the trend from 1974 to 1993 continues, saturation should be achieved by 2023AD.

THE SPATIAL DIFFUSION

PROCESS

Now consider a country divided into states. Innovations are introduced thatjchange initial conditions A into transformed conditions B on a state-by-state basis. Is the transformation process universal, in the sense that it unfolds at an equal pace in each of the states from a common takeoff, or is a spatial diffusion process at work? In the latter case, certain states would be observed to innovate early, others would learn from them and join the main bandwagon, and yet others would be laggards, and the pace of transition might vary according to the timing of takeoff. The next step is to address this question of universality vs. spatial diffusion with respect to M~xico's demographic transition. We begin by applying transformation theory, as outlined in the previous section, to each of the 32 M~xican states. Thus, Table 3 records, for each state, the death rate in 1930 as an initial condition, the parameters of the logistic growth equation a and k, the takeoff and saturation years, the

e,~ b .

D

~ ~ .~_ ~

~

0

~

~

~

0

~

~

~

c5 ~=

0

I

t-

I

I

I

I

I

I

I

I

I

I

I

~m rm

0

u , ~ , , ~, ~ ' B

~

~ ~ ~o

Cxl o'1 '~l" u'l ~.0 I",. r

~0 " - ~ "

0"; 0

~

r

~ 0

0 .

o

. o

~

;

~

~

. . o o

~

l

t

~ ~

. . . o o o

o

0 o q o o

0

I

;

I

i

I

o~o~-g~, ~ ~

,~.-~o~= z _ ~ z z

~

0

o

~

;

~ ~

. . . . . . . o o o o o o

~

"0

o ._.~0

~

~

I

I

~

~ . o

~

;

I

. o

q ~ o

~

~

I

t

I

t

;

0

,~=o

8~'~

==~ -~ o ' o ' ~ ~ ~ ~-

•

,~ ~.r

1

375 B. J. U BERRY,U S. HALL, R. HERNANDEZ-GUERRERO,AND P. H. MARTIN

diffusion time in years, and the end-point distribution in 1970. The latter is included to ascertain whether there is any residual pattern. Table 4 records the initial birth rates of 1970, the materials used to calculate the coefficient k, and the years of takeoff. What factors influence the timing of takeoff, the pace of transition, the achievement of saturation, and the incidence of residuals at the level of states within nations? We learn from the literature (Berry, 1972; Haggett, Cliff and Frey, 1977; Cliff, Haggett, Ord and Versey, 1981) that diffusion should spread outwards from core to periphery, from rich and more highly urbanized regions to those that are poor and less urbanized, and down the urban hierarchy from regions centering on large cities to those in which the cities are small. What factors specific to the M~xican situation might be added? The literature on that country suggests differences between the high-altitude interior and the low-lying littoral, and in the case of fertility, between states that are predominantly Catholic and those that have greater non-Catholic populations. Each of these ideas was converted to one or more independent variables to be included in regression equations. To capture core-periphery differences, population potentials were computed for the years 1930, 1970 and 1990. The size of the largest city in each state in each of these three years was recorded, as was the percentage of the population living in urban areas. To capture the interior-littoral variable, altitude was recorded for each state's largest city. Finally, a poverty index was obtained, and the percentage non-Catholic was recorded for the years 1930, 1970 and 1990. Armed with this set of independent variables, we asked which of them were associated with state-to-state variations in each of the measures characterizing the transformation process: initial conditions, date of takeoff, and speed of transition (k), and in the case of mortality diffusion time, date of saturation, and the ending pattern (the fertility transition was barely onethird complete in 1993). This gave us six mortality regressions and three fertility regressions. The first four mortality variables were regressed on 1930 values of the independent variables and the last two on 1970 values. In the case of fertility, the three dependent variables were regressed on 1970 independent variable values. The results are detailed in Tables 5 and 6. The six mortality models are very weak, with coefficients of determination varying from 0.10 to 0.45, suggesting relatively little cross-state spatial variation in the transition process. Initial death rates were high along the central mountainous spine, as we saw in Figure 2, as well as where the level of urbanization was greater, suggesting the unhealthful conditions obtaining in pre-transition cities. To the extent that there were variations in the timing of transition, the takeoff

~

0

v

0 ~

~

0

~ ~

~

~

0

~

~

d d d d d d d d d d d o d

v

89 O. r-

1

89 o.

~ ~

~ ~

0 ~

~ 0

0 ~

o d d d d d d d d d d d d

~ m

v

~

~-E

._o .R

ev

~

0

~

0

0 0 0 0 0 0 0 0 0 0 0 0 0 I I I I I I I I I

I

I

I

I

I

I

o

o x,o~

0 ".~ ~

I (2) r',.

g,o,

I

I

I

I

I

I

I

I

|

C13

-~ .E e~

-o

~o~

~

~

o'5

"F'~.~-

o

.-

~ ~

0 ~

~

~

~

0

~

0

~

0

~

0

0

~

0

ddddddddddddddddddd

~

~

0

~

~

0

~

~

d ~ d d d ~ d d d d d ~ d d ~ d d I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 I I t I I I I I I I I I I I

I

I

I

I

1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

~ ~

~

0

0

~

~

~

~

0 ~

~

~

~

T~ '777' 177777177'77

~

~

C

~

~

=

"0

0

~

8"~-0

~

~=.~

o=_~,g~

~aO~

""-

0

l

.0 o ~-~E

".=

o

o

~

o

~

~

0

E 0

0

0

0

0

I

,7. 0

0

o

0

0

II

o

. . . .

~ ~. m. ~. r,,:. m. r,: ~ m. m. m. ~. ~. ~.

0

0

c-O

-~

C:

.O -

~

.-

o

C=

"G

E

c~d~

x

(D 0... 0

"-

~

e

o

q~.'-- C

8-

~

>

"=

o

379 B. J. L. BERRY, L. S. HALL, R. HERNANDEZ-GUERRERO, AND P. H. MARTIN

TABLE 6 Analysis of the Fertility Transition Dependent Variable

Independent Variables Population potential Size of Igst. City Level of urbanization Altitude Poverty index Percent non-Catholic Constant Coeff. of determination

1 Initial Conditions 1970

2 Date of Takeoff

3 Speed of Transition

6.36e-03"** (4.76) 5.24e-06"** (3.68) - 1.58 (0.27) 1.54e-03" (1.95) 0.13 (0.91) - 38.35 (1.4) 59.88*** (4.23) 0.55

1.36e-03"* (2.16) - 1.54e-06"* (2.29) 0.58 (0.21) 1.35e-04 (0.36) 9.52e-02 (1.37) 4.41 (0.34) 19.74e02"** (295) 0.27

- 1.89e-05"* (2.12) 2.14e-08"* (2.25) 3.44e-03 (0.08) - 2.37e-06 (0.44) - 1.4e-03 (1.43) -0.04 (0.25) 0.31 *** (3.34) 0.27

t-statistics are in parentheses. *significant at 0.1 level. **significant at 0.05 level ***significant at 0.01 level.

date came soonest where the poverty index was greater, was speediest along the country's mountainous spine, but was delayed in the core, contrary to expectations based on spatial diffusion theory. Neither diffusion times nor saturation dates show any association with the spatial diffusion variables. The mortality pattern of 1970 reveals that residual rates are highest where poverty is greatest. In contrast, the fertility models summarized in Table 6, while still of modest power, reveal changes that are consistent with received theory: an initial pattern of high fertility in core regions and large cities, takeoff that occurred first in states with the largest cities but was otherwise retarded in the core regions, and a transition that has been speediest in the large-city states and retarded elsewhere.

380

POPULATIONAND ENVIRONMENT

THE PUBLIC POLICY ROLE

What explains these differences? Public policy, we believe, was critical not only in initiating the epidemiological and fertility transitions, as suggested by Figure 1, but also in structuring the resulting spatial diffusion processes. The M~xican Revolution had been fought over issues of economic and political inequality. The different leaders--Francisco L. Madero, Pancho Villa and Emiliano Zapata--paid special attention to the distribution of land, to education, and to political freedoms. The main contributions of L~zaro C~rdenas after he assumed the presidency of M~xico in 1930 were to implement agrarian reform and to significantly increase the share of the federal budget devoted to social expenditures, including education, health, water supply, sewage disposal, and welfare (Knight, 1991). By 1940, C~rdenas had distributed 18 million hectares of land to some 800,000 recipients; by that year ejidos held 47 percent of all cultivated land, compared to 15 percent in 1930. As government revenues swelled with economic recovery from the depression that preceded his election, C~rdenas also channeled resources into agriculture. Fully 9.5 percent of the budget was devoted to provision of agricultural credit in 1936. Additional resources went into roads, irrigation, and rural electrification. As a result, the standard of living of the campesinos (peasants) increased: in the Laguna region between Coahiula and Durango, rural minimum wages were a third higher in 1939 than in 1934, as the Ejidal Bank provided previously unavailable farm credit. The Laguna experience was repeated in the Mexicali Valley, in Sonora (where the Yagui and Mayo Indians won partial restitution of their lands), and in Michoacan. The reforms were least successful in Yucatan and in Chiapas, where corrupt caciques seized control of the mechanisms of agricultural credit. Setting an example, the Sonorans also built six thousand rural schools and the maestro became the carrier of national and secular values. Educational projects proliferated, and congress approved C~rdenas' plan for a form of socialist state education that was free to everyone. Socialism, it is said, confronted Catholic obscurantism, helped bring literacy and hygiene, and increased productivity (Knight, ibid). C~rdenas' rural-based programs contrast markedly with the urban-centered initiatives of Luis Echeverria. When he came to office in 1970, Echeverria was faced with one of the highest rates of natural increase in the world, 3.8 percent annually. Only 31 percent of the population had reported using any form of fertility limitation in 1966, and of this 31 percent the rhythm method was the preferred alternative (Turner, 1974). Cultural norms accommodated to large families, and the Catholic Church opposed

381

B. J. L. BERRY, L. S. HALL, R. HERNANDEZ-GUERRERO, AND P. H. MARTIN

any form of birth control. President Gustavo Diaz Ordaz had taken the position that population control was not an issue: M~xico's greatest asset, he argued, was its human potential. Initially, Echeverria continued the population policies of the Ordaz administration, in which he had served as Interior Minister, but worsening economic conditions led him to change direction. He began in 1972 with a government public relations program that used "responsible parenthood" as its theme. Its focus was on educating doctors, nurses and the public on family planning. In addition to instruction on family planning techniques, the campaign also emphasized family solidarity and discouraged values that were "foreign to the national culture" (Turner, 1974). Turner notes that these moves were meant to placate opponents in the political left, among nationalists, and in the church. The biggest changes in policy came on January 1, 1973, when the Secretariat of Health and Welfare began opening family planning centers, and in 1974, when the M~xican constitution was amended to guarantee every couple the right to plan their family freely. The program commenced in eight cities, including the Federal District, Toluca, Cuernavaca, Pachuca and Queretaro, spreading to other centers later. The program dispensed contraceptives to those who needed assistance in getting them and included postpartum instruction in family planning in every M(~xican hospital. To fund his efforts, Echeverria significantly increased the share of the federal budget devoted to social expenditures to more than 25 percent, the highest ever; the share had lagged after C~rdenas left office.

CONCLUSIONS L~zaro C~rdenas' rural-based assault on inequality and Luis Echeverria's urban-centered family planning programs were instrumental in sparking M~xico's epidemiological and fertility transitions, but their programmatic initiatives had markedly different consequences for diffusion of the demographic transition. C~rdenas' equity-focused programs initiated a near-universal process of mortality decline that took hold soonest in poverty areas and in the periphery. Echeverria pursued fertility control via a set of urban-centered initiatives that have been slow to diffuse to M~xico's peripheral regions. One of our peer reviewers s comments: the results are also what most campesinoswould recognize in daily life. Death control is partly driven by "magic bullets," such as vaccination and is not controversial, so it gets around quickly and with few regional differences. Birth control, by con-

382 POPULATION AND ENVIRONMENT

trast, is highly controversial, the technologies are imperfect, and the contraceptive choices and abortion (an intrinsic part of fertility regulation in M~xico as in all other countries) are constrained by laws, relatively high cost, the Church and the practices of health professions. It follows that public policy moves slowly and fertility regulation spreads more slowly than death control. The urban elites are able to struggle over the barriers society puts between the individual and the technologies they need to limit family size more readily than are peasants in, say, Chiapas. The ironic consequence is that the population explosion has borne down most heavily on those areas pressed closest to the margins of existence, regions that were the first to see the onset of C~rdenas-induced mortality decline yet are the last to benefit from the Echeverria-initiated fertility transition.

ENDNOTES 1. 2. 3.

4.

5.

Betsy Brewer, Cyprain Engwenyi, Mark Frank and Mark Frost also provided research assistance. The consequences of M~xico's changing age structure are discussed in Weeks (1996). We are indebted to James F. Duncan of the New Zealand Futures Trust for the suggestion that Eyring's rate theory (Glastone, 1954) and Semenoff's chain reaction theory (Semenoff, 1935), both developed in the field of chemical kinetics, provide a conceptual basis for analyzing such transformations. The national birth and death rates used in Figure 1 and Tables 1 and 2 are most readily obtained from Mitchell (1993, p. 81), supplemented by Instituto Nacional de Estadfstica, Geografla e Inform~tica (INEGI). 1996. Anuario Estadfstico de los Estados Unidos M~xicanos 1995. Aguascalientes, AGS: INEGI. The death rates and birth rates by state by year came from INEGI, Anuario Estadfstico, various years. INEGI is also the source of information on city size, altitude, the state populations and latitude-longitude coordinates of major cities used to compute population potentials, percentage of the population nonCatholic, etc. The poverty index was originally developed by Wilkie (1967), and subsequently by INEGI. Readers may want to consult ftp://ftp.ciesin.org/pub/data/Mexico/. To whom we are exceedingly grateful.

REFERENCES Berry, B. J. L. (1972). Hierarchical Diffusion. The Basis of Developmental Filtering and Spread in a System of Growth Centers. In Niles M. Hansen (Ed.), Growth Centers in Regional Economic Development. New York: The Free Press. 108-138. Cliff, A. D., P. Haggett, Ord, J. K., & Versey, G.R. (1981). Spatial Diffusion. Cambridge, UK: Cambridge University Press. Duncan, J. F. (1998). The Chemistry of Social Interactions. Wellington, NZ: New Zealand Futures Trust.

383 B. J. L. BERRY, L. S. HALL, R. HERNANDEZ-GUERRERO, AND P. H. MARTIN

Glastone, S. (1954). Elements of Physical Chemistry. London: Macmillan. Haggett, P., Cliff. A. D., & Frey. A. (1977). Locational Analysis in Human Geography. Second edition. New York: John Wiley and Sons. Knight, A. (1991). The Rise and Fall of Cardenism, c.1930-c.1946. In L. Bethell (Ed.), M~xico Since Independence. Cambridge, UK: Cambridge University Press. Mitchell, B. R. (1993). International Historical Statistics. The Americas 1750-1988. New York and London: M. Stockton Press. Semenoff, N. (1935). Chemical Kinetics and Chain Reactions. Oxford: Clarendon Press. Solis, L. (1981 ). La Realidad Econdmica Mexicana: Retrovisidn y Perspectivas. M~xico: Siglo XXI. Turner, F. C. (1974). Responsible Parenthood: The Politics of M~xico's New Population Policies. Washington DC: The American Enterprise Institute. Weeks, J. R. (1996). Population. An Introduction to Concepts and Issues. Belmont, CA: Wadsworth Publishing Co. Wilkie, J. W. (1967). The M~xican Revolution. Federal Expenditure and Social Change Since 1910. Berkeley and Los Angeles: University of California Press. Wrigley, E. A. & Schofield, R. S., (1981). The Population History of England. Cambridge, MA: Harvard University Press.