2
Why the Twentieth Century
Was So Remarkable
Research during the past two decades has produced significant
advances in the description and explanation of the secular
decline in mortality. Although many of these findings are still tentative,
they suggest a new theory of evolution that Dora Costa
(an economist and biodemographer at MIT) and I call “technophysio
evolution.” Studies of the causes of the reduction in mortality
point to the existence of a synergism between technological and
physiological improvements that has produced a form of human
evolution that is biological but not genetic, rapid, culturally transmitted,
and not necessarily stable.1
This process is still ongoing in
both rich and developing countries. In the course of elaborating
this theory, thermodynamic and physiological aspects of economic
growth will be defined, and their impact on economic growth rates
will be discussed.
Unlike the genetic theory of evolution through natural selection,
which applies to the whole history of life on earth, technophysio
evolution applies only to the past 300 years of human history
and particularly to the past century.2
Despite its limited scope,
20
why the twentieth century was so remarkable 21
technophysio evolution appears to be relevant to forecasting likely
trends over the next century or so in longevity, the age of onset
of chronic diseases, body size, and the efficiency and durability of
vital organ systems. It also has a bearing on such pressing issues of
public policy as the growth in population, in pension costs, and in
health care costs.
The theory of technophysio evolution rests on the proposition
that during the past 300 years, particularly during the past century,
human beings have gained an unprecedented degree of control over
their environment – a degree of control so great that it sets them
apart not only from all other species, but also from all previous generations
of Homo sapiens. This new degree of control has enabled
Homo sapiens to increase its average body size by over 50 percent
and its average longevity by more than 100 percent since 1800,
and to greatly improve the robustness and capacity of vital organ
systems.
Figure 2.1 helps to point up how dramatic the change in the
control of the environment after 1700 has been. During its first
200,000 or so years, Homo sapiens increased at an exceedingly
slow rate. The discovery of agriculture about 11,000 years ago
broke the tight constraint on the food supply imposed by a hunting
and gathering technology, making it possible to release between
10 and 20 percent of the labor force from the direct production of
food and also giving rise to the first cities. The new technology of
food production was so superior to the old one that it was possible
to support a much higher rate of population increase than
had existed prior to c. 9000 b.c. Yet, as Figure 2.1 shows, the advances
in the technology of food production after the second Agricultural
Revolution (which began about a.d. 1700) were far more
dramatic than the earlier breakthrough, since they permitted the
population to increase at so high a rate that the line of population
appears to explode, rising almost vertically. The new technological
breakthroughs in manufacturing, transportation, trade, communications,
energy production, leisure-time services, and medical
services were in many respects even more striking than those in
22 the escape from hunger and premature death, 1700–2100
-9000 -6000 -4000 -2000 0 2000
-5000 -3000 -1000 1000
Time
(in years)
0
1,000
2,000
3,000
4,000
5,000
6,000
Population
(in millions)
B. 1st Ag. Rev.
B. Pottery
I. Plow
1st Irrigation Works
1st Cities
B. Metallurgy
B. Writing
B. Math
Peak of Greece
Peak of Rome
Black Death
Disc. New World
B. 2nd Ag. Rev.
B. Indust. Rev.
I. Watt Engine
B. Railroads
Germ Theory
Electrification
I. Telephone
I. Automobile
Penicillin
Disc. DNA
Nuclear Energy
PCs
I. Airplane
War on Malaria
High-Speed Computers
Man on Moon
B. Genome Project
Dolly, the Cloned Sheep
Stem Cell Research
Approved in U.S.
Figure 2.1 The Growth of World Population and Some Major Events
in the History of Technology.
Sources: Cipolla 1974; Clark 1961; Fagan 1977; McNeill 1971; Piggott 1965; Derry
and Williams 1960; Trewartha 1969. See also Allen 1992, 1994; Slicher van Bath
1963; Wrigley 1987.
Note: There is usually a lag between the invention (I) of a process or a machine and its
general application to production. “Beginning” (B) usually means the earliest stage
of this diffusion process.
agriculture. Figure 2.1 emphasizes the huge acceleration in both
population and technological change during the twentieth century.
The increase in the world’s population between 1900 and 1990
was four times as great as the increase during the whole previous
history of humankind.
why the twentieth century was so remarkable 23
The Relationship between Body Size and the Risk
of Death at Middle and Late Ages
Although Figure 2.1 points to changes in technology that permitted
a vast increase in population, it does not reveal a connection
between technological changes and physiological benefits. To get
at that question, we need to consider a number of recent studies
that have demonstrated the predictive power of height and body
mass with respect to morbidity and mortality at later ages. The results
of two of these studies are summarized in Figures 2.2 and 2.3.
Figure 2.2 plots the relationship between relative mortality risk
and height found by Hans Waaler among Norwegian men aged
40–59 measured in the 1960s and among Union Army veterans
measured at ages 23–49 and at risk between ages 55 and 75.3
Short men, whether modern Norwegians or nineteenth-century
Americans, were much more likely to die than tall men. Height
Relative
Risk
1.5
1.0
0.5
62 64 66 68 70 72 74 76 78 80
Union Army Veterans
Height (inches)
Waaler
Figure 2.2 Relative Mortality Risk among Union Army Veterans and
among Modern Norwegian Males.
Note: A relative risk of 1.0 means that the risk at that height was equal to the average
risk of death in the entire population of males of the specified ages. Also note that the
tallest data point, in both the Norwegian and Union Army cases, is not statistically
significant.
24 the escape from hunger and premature death, 1700–2100
2.14
0.88
Modern
Norwegian
Males
Union Army
Veterans
17 19 21 23 25 27 29 31 33 35
BMI
RelativeRisk
Figure 2.3 Comparison of Relative Mortality Risk by BMI among
Men 50 Years of Age, Union Army Veterans around 1900 and Modern
Norwegians.
Source: Reprinted from Costa and Steckel 1997, p. 54, with the permission of The
University of Chicago Press. c 1997 by the National Bureau of Economic Research.
Note: In the Norwegian data, BMI for 79,084 men was measured at ages 45–49, and
the period of risk was 7 years. BMI of 550 Union Army veterans was measured at ages
45–64, and the observation period was 25 years.
has also been found to be an important predictor of the relative
likelihood that men aged 23–49 would have been rejected by the
Union Army during 1861–65 because of chronic diseases.4
Despite
significant differences in ethnicity, environmental circumstances,
the array and severity of diseases, and time, the functional relationship
between height and relative risk is strikingly similar in the
two cases.5
Waaler has also studied the relationship in Norway between
body mass, measured by BMI, and the risk of death.6
A curve summarizing
his findings for men aged 45–49 is shown in Figure 2.3.
why the twentieth century was so remarkable 25
The curve for Union Army veterans measured at ages 45–64 and
followed for 25 years is also shown in Figure 2.3. Among both
modern Norwegians and Union Army veterans, the curve is relatively
flat within the range 22–28, with the relative risk of mortality
hovering close to 1.0. However, at BMIs of less than 22 and over 28,
the risk of death rises quite sharply as BMI moves away from its
mean value.
It is important to understand that, as used in this discussion,
“risk” or “risk of death” refers to the likelihood of dying during
any defined period of time; the risk period in Figure 2.4, for example,
is 18 years. Mortality risk is most commonly presented as
the crude death rate (CDR). In this diagram, a relative risk of 1.0
is the average risk of dying in the population as a whole over all
heights and weights (i.e., the average crude death rate – the total
deaths during a year divided by the midyear population). Greater
or lower risks that vary with height and weight are all expressed
relative to the average risk in the population as a whole. For example,
a relative risk of 2.0 means having a risk that is twice the
average CDR.
Although Figures 2.2 and 2.3 are revealing, they are not sufficient
to shed light on the debate over whether moderate stunting
impairs health when weight-for-height is adequate. To get at the
“small-but-healthy” issue, one needs an iso-mortality surface that
relates the risk of death to height and weight simultaneously.
Such a surface is presented as a three-dimensional diagram in the
frontispiece. For some purposes, it is more convenient to represent
a three-dimensional surface in two dimensions, as is done in
topographical maps. Such a two-dimensional surface, presented in
Figure 2.4, was fitted to Waaler’s data. Figure 2.4 combines three
different types of curves. The solid curves are iso-mortality risk
curves, each of which traces out all the combinations of height
and weight that represent a given level of risk. Transecting the
iso-mortality map is a set of iso-BMI curves, represented by dashed
lines. Each iso-BMI curve is the locus of all of the combinations
of height and weight that yield a specific level of BMI, ranging from
26 the escape from hunger and premature death, 1700–2100Height(m)
2.2
1.9
1.6
1.3
1.0
0.7
16
19
22
25
28
31
34
1.55
1.60
1.70
1.80
1.90
1.95
1.65
1.75
1.85
Weight (kg)
40 50 60 70 80 90 100 110
Iso-Mortality-Risk Curves Iso-BMI Curves Minimum-Risk Curve
(0.7~2.2) (16~34)
⊕
1705
⊕
1785
⊕
1867
⊕
1967
Key:
Figure 2.4 Iso-Mortality Curves of Relative Risk for Height and
Weight among Norwegian Males Aged 50–64, with a Plot of the
Estimated French Height and Weight at Four Dates.
Sources: Data points for 1705 and 1785 are from Fogel, Floud, and Harris, n.d.; the
point for 1867 is from Baxter 1875, 1: 58–59; the point for 1967 is from Eveleth and
Tanner 1976.
Note: For a brief description of the procedure used to estimate the 1705 and 1785
points, see Fogel 1997; a more extensive explanation appears in Fogel, Floud, and
Harris, n.d. This figure supersedes versions that have appeared previously.
why the twentieth century was so remarkable 27
16 to 34. The heavy black curve running through the minimum
point of each iso-mortality curve gives the weight that minimizes
risk at each height.
Figure 2.4 shows that even when body weight is maintained at
what Figure 2.3 indicates is an “ideal” level (BMI = 25), short men
are at substantially greater risk of death than tall men. Figure 2.4
also shows that the ideal BMI varies with height. A BMI of 25 is
ideal for men in the neighborhood of 176 cm (69 inches), but for
tall men the ideal BMI is between 22 and 24, while for short men
(under 168 cm or 66 inches) the ideal BMI is about 26.7
Superimposed on Figure 2.4 are rough estimates of heights and
weights in France at four dates. In 1705 the French probably
achieved equilibrium with their food supply at an average height of
about 161 cm (63 inches) and a BMI of about 18. Over the next 290
years the food supply expanded with sufficient rapidity to permit
both the height and the weight of adult males to increase. Figure 2.4
implies that while factors associated with height and weight jointly
explain virtually all of the estimated decline in French mortality
rates over the period between c. 1785 and c. 1867, they explain
only about 35 percent of the decline in mortality rates between
c. 1867 and c. 1967.8
The analysis in this section points to the misleading nature of
the concept of subsistence as Malthus originally used it and as it
is still widely used today. Subsistence is not located at the edge of
a nutritional cliff, beyond which lies demographic disaster. Rather
than one level of subsistence, there are numerous levels at which
a population and a food supply can be in equilibrium in the sense
that they can be indefinitely sustained. However, some levels will
have smaller people and higher normal mortality than others.9
The Relevance of Waaler Surfaces for Predicting
Trends in Chronic Diseases
Poor body builds increased vulnerability to diseases, not just
contagious diseases but chronic diseases as well. This point is
28 the escape from hunger and premature death, 1700–2100
1.0
1.5
2.0
2.5
160 170 180 190 200
Height (cm)
RelativeRisk
Figure 2.5 Relationship between Height and Relative Risk of Ill
Health in NHIS Veterans Aged 40–59.
Source: Fogel, Costa, and Kim 1993.
Note: This curve is similar to the curve in Figure 2.2, except that Figure 2.5 gives
the average risk of reporting poor health by height, whereas Figure 2.2 gives the
average risk of dying by height. Also note that the tallest data point is not statistically
significant.
demonstrated by Figure 2.5, which shows that chronic conditions
were much more frequent among short young men than among
tall men in the U.S. National Health Interview Surveys (NHIS)
for 1985–88. Virtually the same functional relationship between
stature and chronic diseases was found in the 1860s among young
adult and middle-aged men examined by the surgeons of the Union
Army. Stunting during developmental ages had a long reach and
increased the likelihood that people would suffer from chronic diseases
at middle and late ages.10
American males born during the second quarter of the nineteenth
century were not only stunted by today’s standards, but
their BMIs at adult ages were about 10 percent lower than current
U.S. levels (see Figure 2.6).11
The difference in average BMI between
adult males today and those born in the nineteenth century
widened with age, perhaps because of the accumulated effects of
why the twentieth century was so remarkable 29
27
26
25
24
23
22
21
19 22 27 32 37 45 57 72
1991
1961
1944
1864
1894
1864
1944
1961
1991
1894
1864
1944
1961
1991
1991
1961
1944
1864
1894
1991
1961
1944
1864
1894
1894
1864
1961
1991
1991
1961
1900
1910
1961
1991
Age
BMI
Figure 2.6 Mean BMI by Age Group and Year, 1864–1991.
Source: Reprinted from Costa and Steckel 1997, p. 55, with the permission of The
University of Chicago Press. c 1997 by the National Bureau of Economic Research.
Note: The age groups are centered at the marks and are ages 18–19, 20–24, 25–29,
30–34, 35–39, 40–49, 50–64, and 65–79. For some years BMI is not available for a
specific age group.
differences in nutritional intake and physical activity and because
of the increased prevalence of chronic conditions at older ages
(see Figure 2.6). The implication of combined stunting and low
BMI is brought out in Figure 2.7, which presents a Waaler surface
for morbidity estimated by John Kim (1993) from NHIS data for
1985–88 that is similar to the Norwegian surface for mortality (see
Figure 2.4).
Figure 2.7 also presents the coordinates in height and BMI of
Union Army veterans who were 65 or over in 1910 and of veterans
(mainly of World War II) who were the same ages during 1985–88.
These coordinates imply that increases in height and BMI should
have led to a decline of about 35 percent in the prevalence of chronic
diseases between the two cohorts.12
30 the escape from hunger and premature death, 1700–2100
1.55
1.60
1.65
1.70
1.75
1.80
1.85
1.90
1.95
Height(m)
40 50 60 70 80 90 100 110
Weight (kg)
BMI=20
Opt.WeightCurve
BMI=25
BMI=30
Risk=1.0
Risk=1.0
Risk=1.5
Risk
=
1.5
1985 - 1988 Males
(Height = 1.77 m, BMI = 26.4)
1910 Males (Height = 1.727 m, BMI = 22.8)
Optimum Weight Curve
Iso-Risk Curves (0.5−2.2)
Iso-BMI Curves (13−39)
Key:
Figure 2.7 Health Improvement Predicted by NHIS 1985–88 Health
Surface.
Source: Kim 1993.
Note: All risks are measured relative to the average risk of morbidity (calculated over
all heights and weights) among NHIS 1985–1988 white males aged 45–64.
why the twentieth century was so remarkable 31
Table 2.1 Comparison of the Prevalence of Chronic Conditions
among Union Army Veterans in 1910, Veterans in 1983 (Reporting
Whether They Ever Had Specific Chronic Conditions), and
Veterans in NHIS, 1985–88 (Reporting Whether They Had Specific
Chronic Conditions during the Preceding 12 Months), Aged 65
and Above, Percentages
1910 Union Age-Adjusted NHIS
Army 1983 1983 1985–88
Disorder Veterans Veterans Veterans Veterans
Musculoskeletal 67.7 47.9 47.2 42.5
Digestive 84.0 49.0 48.9 18.0
Hernia 34.5 27.3 26.7 6.6
Diarrhea 31.9 3.7 4.2 1.4
Genitourinary 27.3 36.3 32.3 8.9
Central nervous, 24.2 29.9 29.1 12.6
endocrine,
metabolic,
or blood
Circulatorya
90.1 42.9 39.9 40.0
Heart 76.0 38.5 39.9 26.6
Varicose veins 38.5 8.7 8.3 5.3
Hemorrhoidsb
44.4 7.2
Respiratory 42.2 29.8 28.1 26.5
Notes: Prevailing rate of Union Army veterans are based on examinations by physicians.
Those for the 1980s are based on self-reporting. Comparison of the NHIS rates with
those obtained from physicians’ examinations in NHANES II indicates that the use of
self-reported health conditions does not introduce a significant bias into the comparison.
See Fogel, Costa, and Kim 1993 for a more detailed discussion of possible biases and their
magnitudes.
a Among veterans in 1983, the prevalence of all types of circulatory diseases will be underestimated
because of underreporting of hemorrhoids.
b The variable indicating whether the 1983 veterans ever had hemorrhoids is unreliable.
Source: Fogel, Costa, and Kim 1993.
This exercise is consistent with what actually occurred.13
Table 2.1 compares the prevalence of chronic diseases among Union
Army men aged 65 and over in 1910 with two surveys of veterans
of the same ages in the 1980s.14
That table indicates that musculoskeletal
and respiratory diseases were 1.6 times as prevalent,
heart disease was 2.9 times as prevalent, and digestive diseases
32 the escape from hunger and premature death, 1700–2100
were 4.7 times as prevalent among veterans aged 65 or over in
1910 as in 1985–88. Young adults born between 1822 and 1845
who survived the deadly infectious diseases of childhood and adolescence
were not freer of degenerative diseases than persons of
the same ages today, as some have suggested, but were more afflicted.
Hernia rates at ages 35–39, for example, were more than
three times as high in the 1860s as in the 1980s. Of special note
is the much higher incidence of clubfoot in the 1860s – a birth
anomaly that suggests that the uterus was far less safe for those
awaiting birth than it is today. The provisional findings thus suggest
that chronic conditions were far more prevalent throughout
the life cycle for those who reached age 65 before World War I than
is suggested by the theory of the epidemiological transition.15
Reliance
on cause-of-death information to characterize the epidemiology
of the past has led to a significant misrepresentation of the
distribution of health conditions among the living. It has also promoted
the view that the epidemiology of chronic diseases is more
separate from that of contagious diseases than now appears to be
the case.
What is the basis for the predictive capacity of Waaler surfaces
and curves? Part of the answer resides in the realm of human physiology.
Variations in height and weight appear to be associated with
variations in the chemical composition of the tissues that make up
these organs, in the quality of the electrical transmission across
membranes, and in the functioning of the endocrine system and
other vital systems.
Research in this area is developing rapidly, and some of the new
findings have yet to be confirmed. The exact mechanisms by which
malnutrition and trauma in utero or in early childhood are transformed
into organ dysfunctions are still unclear. What is agreed
upon is that the basic structure of most organs is laid down early,
and it is reasonable to infer that poorly developed organs may break
down earlier than well-developed ones.16
The principal evidence so
far is statistical and, despite agreement on certain specific dysfunctions,
there is no generally accepted theory of cellular aging. Much
of this evidence is described in Chapter 3.
why the twentieth century was so remarkable 33
Thermodynamic and Physiological Factors
in Economic Growth
So far, I have focused on the contribution of technological change to
physiological improvements. However, the process has been synergistic,
with improvements in nutrition and physiology contributing
significantly to the process of economic growth and technological
progress.
I alluded to the thermodynamic contribution to economic
growth when I pointed out that individuals in the bottom 20 percent
of the caloric distributions of France and England near the
end of the eighteenth century lacked the energy for sustained work
and were effectively excluded from the labor force. Moreover, even
those who participated in the labor force had only relatively small
amounts of energy for work.
Since the first law of thermodynamics applies as much to human
engines as to mechanical ones, it is possible to use energy cost
accounting to estimate the increase in energy available for work
over the past two centuries. In the British case the thermodynamic
factor explains 30 percent of the British growth rate since 1790.17
The increase in the amount of energy available for work had two
effects. It raised the labor force participation rate by bringing into
the labor force the bottom 20 percent of consuming units of 1790
who had, on average, only enough energy for a few hours of slow
walking. Moreover, for those in the labor force, the intensity of
work per hour has increased because the number of calories available
for work each day increased by about 50 percent.18
The physiological factor pertains to the efficiency with which
the human engine converts energy input into work output. Changes
in health, in the composition of diet, and in clothing and shelter
can significantly affect the efficiency with which ingested energy is
converted into work output. Reductions in the prevalence of infectious
diseases increase the proportion of ingested energy available
for work both because of savings in the energy required to mobilize
the immune system and because the capacity of the gut to absorb
nutrients is improved, especially as a consequence of a reduction
34 the escape from hunger and premature death, 1700–2100
in diarrheal diseases. Thermodynamic efficiency has also increased
because of changes in the composition of the diet, including the
shift from grains and other foods with high fiber content to sugar
and meats. These dietary changes raised the proportion of ingested
energy that can be metabolized (increased the average value of the
“Atwater Factors,” to use the language of nutritionists). Improvements
in clothing and shelter have also increased thermodynamic
efficiency by reducing the amount of energy lost through the radiation
of body heat.19
Moreover, individuals who are stunted but otherwise healthy at
maturity will be at an increased risk of incurring chronic diseases
and of dying prematurely. In other words, when considered as work
engines, they wear out more quickly and are less efficient at each
age. The available data suggest that the average efficiency of the
human engine in Britain increased by about 53 percent between
1790 and 1980. The combined effect of the increase in dietary
energy available for work, and of the increased human efficiency in
transforming dietary energy into work output, appears to account
for about 50 percent of the British economic growth since 1790.20
Making Economic Sense of the Conflicts between
Economic and Biomedical Measures of Inequality
As already noted, traditional economic measures of the standard
of living, such as per capita income and indexes of real wages,
sometimes conflict with biomedical measures such as stature, BMI,
and life expectancy. What are we to make of a situation in which
real wages were rising, as apparently occurred in England during
the last three quarters of the nineteenth century, while workingclass
heights and BMI remained at relatively low levels, showing
little increase over half a century? How should we characterize
conditions of workers in the United States between 1820 and 1860
if real wages were generally constant or rising, sometimes quite
rapidly, but heights and life expectancy were decreasing?21
During
an era in which 50 to 75 percent of the income of workers was spent
why the twentieth century was so remarkable 35
on food, is it plausible that the overall standard of living of workers
was improving if their nutritional status and life expectancy were
declining? Although these questions are not yet resolved, they are
now being vigorously investigated.22
It may be fruitful to consider
some of the new issues about the course of the standard of living
and their implications for the measurement of inequality that are
suggested by the anthropometric and demographic data.
If cholera and other diseases that afflicted the United States
during the nineteenth century were acts of God, unrelated to the
functioning of the economic system, they would pose no special
problem for the resolution of the standard-of-living controversy.
However, economic growth, the spread of disease, and the concomitant
increase in morbidity and mortality rates were intricately
intertwined. Not only was internal migration responsible for as
much as 50 percent of the increase in measured per capita income
during the antebellum era,23
it was also a principal factor
in the spread of cholera, typhoid, typhus, malaria, dysentery, and
other major killer diseases of the era.24
Increasing population density,
another concomitant of economic growth, also increased the
prevalence of various diseases, raising the level of malaria, enteric
diseases, and diseases of the respiratory system.25
The increase in mortality between 1790 and 1860, therefore,
indicates that a downward adjustment is necessary even if wage
rates in high-disease localities fully reflected the extra wage compensation
(which economists refer to as a “bribe”) that workers
demanded for the increased risks of living in these areas, since national
income accounting procedures treat the bribe as an increase
in national income when it is merely a cost of production. Different
ways of correcting estimates of the unmeasured cost of mortality,
and of adjusting the national income accounts accordingly, are discussed
in the notes. They show that much of what appears to have
been a rise in real wages between 1790 and 1860 is spurious, and
that the apparent growth in average real wages over these years
needs to be reduced by at least 40 percent.26
So far, I have stressed that measures of per capita income
exaggerate economic growth because they fail to remove costs
36 the escape from hunger and premature death, 1700–2100
of production from the measure of real income. This point is
akin to Simon Kuznets’s correction of national income for wages
paid to police because crime is not a benefit but a cost of urban
production.27
However, even when average real wages are appropriately
adjusted, the bearing of this line of argument on the measurement
of trends in inequality during the nineteenth century is
obscure because we lack the detail needed to correct the variations
of income among wage earners as well as between wage income
and other types of income. The veil is lifted somewhat, however,
if we switch from the conventional economic measures of inequality
to the biomedical measures. Data on life expectancy in Great
Britain reveal that although the life expectancy of the lower classes
remained constant or declined in some localities during much of
the nineteenth century, the life expectancy of the upper classes rose
quite sharply. From the beginning of the Industrial Revolution to
the end of the nineteenth century, the gap in life expectancy between
the upper and lower classes increased by about 10 years.
Similarly, the gap in stature between the upper and lower classes
appears to have increased between the end of the Napoleonic wars
and the beginning of the twentieth century.28
In other words, the biomedical data suggest that the disparity
between the upper and lower classes increased during much if not
most of the nineteenth century. This is a different finding than calculations
based on income distributions, which suggest that during
most of the nineteenth century the inequality of the English income
distribution remained constant. After considering the discrepancies
between the traditional economic measures and the biomedical
measures, I lean toward the conclusion that for the nineteenth
century the biomedical measures are more laden with economic information
than the traditional economic measures, at least where
assessing secular trends in inequality is concerned.29
A preference for biomedical measures over conventional economic
measures of inequality seems even more warranted when
interpreting trends in the twentieth century. In both the British and
the U.S. cases, life expectancy increased dramatically between 1890
and 1930, by about 14 years in Britain (a 31 percent increase) and
why the twentieth century was so remarkable 37
by about 16 years (a 36 percent increase) in the United States.30
Over the same period, U.S. stature increased by about 6 cm. However,
in both the British and American cases, measures such as the
share of income held by the top 5 or 10 percent of the income
distribution show that inequality was relatively constant over this
period or that it might have increased slightly.31
The experience
of the Depression Decade is even more paradoxical. In the United
States the unemployment rate between 1931 and 1939 varied but
was never less than 16 percent; for half of the period, unemployment
ranged between 20 and 25 percent. Yet life expectancy between
1929 and 1939 increased by 4 years and the heights of men
reaching maturity during this period increased by 1.6 cm.32
The resolution of the paradox turns on the huge social investments
made between 1870 and the end of World War I, whose
payoffs were not counted as part of national income during the
1920s and 1930s even though they produced a large stream of
benefits during these decades and continue to do so down to the
present. I refer, of course, to the social investment in public health
and in biomedical technology, whose largest payoffs came well after
the investment was made. Included in this category are not
only the direct federal investments in biomedical research, which
remained modest before 1950, but also the expansion of clinical
medicine practiced in a vastly expanded network of hospitals established
on scientific principles, the quadrupling of higher education
in medicine and the increase in the quality of that education, and
the international expansion of the stock of knowledge of the biology,
chemistry, and epidemiology of disease. Also included in this
category are such public health investments as the construction of
facilities to improve the supply of water, the purification of the
milk supply, the development of effective systems of quarantines,
and the cleaning of the slums.
The point is not merely that these benefits are often excluded entirely
from national income accounts, and from the measures of real
wages, but that they are greatly undervalued even when some aspects
are included, because they are measured by inputs rather than
by benefits (which are outputs). Moreover, these benefits accrued
38 the escape from hunger and premature death, 1700–2100
disproportionately to those with modest incomes. That those who
occupy the lower rungs of society have gained more from certain
forms of unmeasured income is visible in the biomedical measures
because they show by how much the gap in life expectancy, in
stature, and in BMI that once existed between the upper and lower
income classes has been reduced.33
The discussion of omitted variables so far indicates that much
less progress was made by the lower classes during the nineteenth
century than is shown by conventional measures and that, as some
have argued, the relative condition of the working class may have
deteriorated during major parts of the century. The implication
for the twentieth century is the reverse: omitted variables lead to
an underestimate of the absolute and relative gains of the lower
classes. Would these conclusions still hold if another omitted variable,
leisure, was brought into consideration?
Although there were some gains in leisure for the lower class in
the United States and Britain during the nineteenth century, they do
not appear to have occurred until well past the middle of the century.
Of the roughly 25-hour reduction in the work week between
1860 and 1990, perhaps 5 or 6 hours were eliminated before 1890.
Moreover, the scope of leisure-time activities was narrow, limited
primarily to frequenting bars and attending church. Drama, opera,
ballet, concerts, literature, and visual arts were usually too expensive
to be readily accessible to the poor. Although there were antecedents
during the nineteenth century, public libraries, movies,
radio, television, and the like are mainly products of the twentieth
century.34
Kuznets, who was the leading designer of the U.S. national income
accounts, recognized the large underestimate of economic
growth occasioned by the omission of leisure from these accounts.
Valuing the increased daily hours of leisure of workers at the average
wage, he pointed out, would raise per capita income in
the late 1940s by about 40 percent. Today, the figure would be
closer to 120 percent. If such a computation was undertaken for
each decile of the income distribution, it would be apparent that
those in the top decile experienced much less of a gain in leisure,
why the twentieth century was so remarkable 39
since the highly paid professionals and businessmen who populate
the top decile work closer to the nineteenth-century standard of
3,200 hours per year than to the current middle-income standard
of about 1,800 hours. Improvements in the variety and quality of
leisure-time activities have also been less for the upper than the
lower classes. The upper classes still have a proclivity for those
expensive amusements that are most fully measured – opera, concerts,
drama, literature. That proclivity, combined with their longer
work week, means that they spend far less of their time in those
forms of leisure activity for which the unmeasured gains have been
greatest.35
The Remarkable Reduction in Inequality during
the Twentieth Century
The twentieth century contrasts sharply with the record of the
two preceding centuries. In every measure that we have bearing
on the standard of living, such as real income, homelessness, life
expectancy, and height, the gains of the lower classes have been
far greater than those experienced by the population as a whole,
whose overall standard of living has also improved.
The “Gini ratio,” which is also called the “concentration ratio,”
is the measure of the inequality of the income distribution most
widely used by economists.36
This measure varies between 0 (perfect
equality) and 1 (maximum inequality). In the case of England,
for example, which has the longest series of income distributions,
the Gini ratio stood at about 0.65 near the beginning of the eighteenth
century, at about 0.55 near the beginning of the twentieth
century, and at 0.32 in 1973, when it bottomed out, not only in
Britain but also in the United States and other rich nations.37
This
measure indicates that over two-thirds of the reduction in the inequality
of income distributions between 1700 and 1973 took place
during the twentieth century. The large decrease in such inequality,
coupled with the rapid increase in the average real income of the
English population, means that the per capita income of the lower
40 the escape from hunger and premature death, 1700–2100
classes was rising much more rapidly than those of the middle or
upper classes.38
A similar conclusion is implied by the data on life expectancies.
For the cohort born about 1875, there was a gap of 17 years
between the average length of life of the British elite and of the population
as a whole. There is still a social gap in life expectancies
among the British, but today the advantage of the richest classes
over the rest of the population is only about 4 years. Thus about
three-quarters of the social gap in longevity has disappeared. As
a consequence, the life expectancy of the lower classes increased
from 41 years at birth in 1875 to about 74 years today, while
the life expectancy of the elite increased from 58 years at birth to
about 78 years. That is a remarkable improvement. Indeed, there
was more than twice as much increase in life expectancies during
the past century as there was during the previous 200,000 years.
If anything sets the twentieth century apart from the past, it is this
huge increase in the longevity of the lower classes.39
Data on stature also indicate the high degree of inequality during
the nineteenth century. At the close of the Napoleonic wars, a
typical British male worker at maturity was about 5 inches shorter
than a mature male of upper-class birth. There is still a gap in
stature between the workers and the elite of Britain, but now the
gap is only on the order of 1 inch. Height differentials by social
class have virtually disappeared in Sweden and Norway but not
yet in the United States. Statistical analysis across a wide array of
rich and poor countries today shows a strong correlation between
stature and the Gini ratio.40
Weight is another important measure of inequality. Despite the
great emphasis in recent years on weight reduction, the world still
suffers more from undernutrition and underweight than from overweight,
as the World Health Organization has repeatedly pointed
out. Although one should not minimize the afflictions caused by
overnutrition, it is important to recognize that even in rich countries
such as the United States, undernutrition remains a significant
problem, especially among impoverished pregnant women, children,
and the aged.
why the twentieth century was so remarkable 41
The secular increase in body builds is due primarily to the great
improvement in socioeconomic conditions over the past several
centuries, rather than to genetic factors, as can be seen by considering
Holland. The average height of young adult males was
only 64 inches in that country during the middle of the nineteenth
century. The corresponding figure today is about 72 inches. An
increase of 8 inches in just four generations cannot be due to natural
selection or genetic drift because such processes require much
longer time spans. Nor can it be attributed to heterosis (hybrid
vigor) because Holland’s population has remained relatively homogeneous
and because the effects of heterosis in human populations
have been shown both empirically and theoretically to have
been quite small. It is hard to come up with credible explanations
for the rapid increase in heights that do not turn on environmental
factors, especially improvements in nutrition and health. These
environmental factors appear to be still at work. Stature is still increasing,
although at a somewhat slower rate, and nations have
not yet reached a mean height that represents the biological limit
of humankind under current biomedical technology.41
Homelessness is another indicator of the dramatic reduction in
inequality during the twentieth century. Down to the middle of the
nineteenth century, between 10 and 20 percent of the population
in Britain and on the Continent were homeless persons whom officials
classified as vagrants and paupers. Estimates of vagrancy and
pauper rates in the United States during the nineteenth century are
less certain, but these rates appear to have reached European levels
in the major cities during the middle decades of that century.
When we speak of homelessness in the United States today, we
are talking about rates below 0.4 percent of the population. Many
of the homeless today are mentally ill individuals prematurely released
from psychiatric institutions that are inadequately funded.
Many others are chronically poor and inadequately trained for the
current job market.42
The relatively generous poverty program developed in Britain
during the second half of the eighteenth century, and the bitter
attacks on that program by Malthus and others, have given the
42 the escape from hunger and premature death, 1700–2100
unwarranted impression that government transfers played a major
role in the secular decline in beggary and homelessness. Despite the
relative generosity of English poor relief between 1750 and 1834,
beggary and homelessness fluctuated between 10 and 20 percent.
Despite the substantial reduction in the proportion of national income
transferred to the poor as a result of the poor laws of 1834
and later years, homelessness declined sharply during the late nineteenth
and early twentieth centuries.
The fact is that government transfers were incapable of solving
the problems of beggary and homelessness during the eighteenth
and much of the nineteenth centuries, because the root cause of the
problems was chronic malnutrition. Even during the most generous
phases of the relief program, the bottom fifth of the English
population was so severely malnourished that it lacked the energy
for adequate levels of work.43
At the end of the eighteenth century British agriculture, even
when supplemented by imports, was simply not productive enough
to provide more than 80 percent of the potential labor force with
enough calories to sustain regular manual labor. It was the huge
increases in English productivity during the later part of the nineteenth
and the early twentieth centuries that made it possible to feed
even the poor at relatively high caloric levels. Begging and homelessness
were reduced to exceedingly low levels, by nineteenthcentury
standards, only when the bottom fifth of the population
acquired enough calories to permit regular work. The principal
way in which government policy contributed to that achievement
was through its public health programs. By reducing exposure to
disease, more of the calories that the poor ingested were made
available for work.