Content uploaded by Walter W. McMahon
Author content
All content in this area was uploaded by Walter W. McMahon on May 10, 2021
Content may be subject to copyright.
eory, Evidence, and Policy Implications 399
e External Social Benets of Higher Education: eory,
Evidence, and Policy Implications
Walter W. McMahon
Laws and institutions must go hand in hand with the progress of the human
mind. As that becomes more enlightened, new truths discovered, and opinions
change, institutions must advance also to keep pace… Establish the law for
educating the common people. is is the business of the state.
omas Jeerson, ltr. to S. Kercheval, 1816,
and Jeerson Memorial, Washington, DC
is paper provides a conceptual framework for the articles that follow in this
special issue by exploring how the external social benets of higher education
contribute to higher sustainable rates of per capita development of families and
nations. It presents a fundamental rethinking of economic development that
explains the major role of education, and of higher education and its relation
to new ideas, as well as of education nance both intuitively and using a formal
model. Empirical estimates of social externalities specic to higher education are
summarized. e paper also develops guidelines for policies. It concludes that
there is substantial underinvestment in higher education and that privatization
of higher education nance has reached or exceeded its economic eciency
limits in the US and probably in England.
e central question addressed in this paper – and in the whole special issue – is
how the external social benets of higher education contribute to sustainable
development of individuals and countries. ‘Development’ is a broad concept:
it includes growth of earnings and Gross Domestic Product (GDP), and also
encompasses many other contributions to quality of life and the well-being of
people and society.
More specically, this paper has a double purpose: to provide a conceptual
framework and backdrop for the other papers in this issue; and to summarize
and extend my earlier research in this eld. It starts with a brief review of what
the external social benets of higher education are. It then explains how these
benets contribute to ‘endogenous development’ – a technical phrase with a
common-sense meaning, namely that human progress or ‘development’ based
on ideas and knowledge is self-perpetuating within families and within nations,
because what is achieved at each step motivates, helps to nance, and enables
the next step. e paper presents endogenous development in two ways: using
intuition and history, and using mathematical logic (in a way that should be
accessible to all in education nance).
Next, the paper summarizes empirical evidence on the external social
benets of higher education – their scope, their size and their value to society.
e evidence is extensive and compelling, though with some gaps that need to
be lled by future research. e nal section of the paper draws both on this
evidence and on the mathematical logic of endogenous development to analyze
policy issues in educational nance. Its conclusions are that there is substantial
underinvestment in 2- and 4-year higher education degrees, and that the shi
from public to private funding has reached its eciency limit at the bachelor’s
level and gone too far in 2-year institutions in the US and England.
e discussion of policy extends beyond economic eciency to cover
inequality, social well-being, and politics. e need for more education beyond
high school has recently been emphasized by Piketty (2020). is text addresses
the plight and the political reaction of the 64 percent or so of the population
in the US and UK without a college education who have missed out on the
benets of growth since 1980. In cutting public funding, governments have also
underestimated the importance of higher education disciplines that contribute
less to economic growth and more to non-monetary development outcomes
beyond growth: the humanities (e.g. an understanding of morals and ethics), the
social sciences (e.g. how democratic and economic institutions function), law
(e.g., constitutional law, human rights), social work and education nance (e.g.
equity), and teacher education (e.g. service to future generations).
A scientic contribution of this paper is to extend the author’s earlier work
on the theory of endogenous development (2002, 2017, 2018) to include an
analytic proof of the crucial role of higher education externalities in sustaining
development. It is important because the current 2.2 percent rate of sustainable
supply-side growth in the US and UK is generally regarded as too low, and is
likely to be lowered further by the damage to skills and other things from the
covid-19 pandemic shock.
e extended model also shows how higher education externalities contribute
directly to ‘total factor productivity growth’ as an important life-long source of
the well-being of graduates and their families. is link is at the root of whether
the younger generation will do better than their parents did, as was the case for
most in the 1960s through 1980 in the US, but now looks unlikely for many.
It also explains most dierences in well-being among families, communities,
states, and nations.
In its review of evidence on the social benets of higher education, this paper
improves on earlier surveys (e.g. Münich and Psacharopoulos 2018) and others
by avoiding overlaps between benets for dierent outcomes (e.g. voting, an
| : - 398
Walter W. McMahon, University of Illinois at Urbana-Champaign
eory, Evidence, and Policy Implications 401
400
intermediate outcome, and democracy, a nal outcome; or a better functioning
court system, an intermediate outcome, and greater human rights for blacks and
others, a nal outcome). e present paper can thus add the estimated value of
the separate nal benets to obtain a consistent estimate of total social benets.
Its assessment of research gaps will hopefully also contribute to greater attention
to issues such as the endogeneity of new ideas and estimates of their value.
e essential feature of external social benets of education is that they are
benets to others in society and to future generations, not private benets to
individuals who are educated (such as higher earnings). e existence of external
benets means that from the perspective of society as a whole, individuals (who
cannot capture all of the social benets privately) have too little incentive to
invest enough, and hence they tend to underinvest in education including
higher education. is provides the main rationale for public support of basic
and higher education, and helps to achieve greater overall economic eciency
that includes these external social benets.
Later in this paper, estimates will be provided of the size and value of the
external benets of higher education, but at the outset it will help to indicate
briey what these externalities are. Some are monetary, such as the higher taxes
paid by those who earn more, higher wages when college graduates generate
social benets that raise the productivity of others, and contributions to a
stable political and economic environment that raises earnings. However, most
externalities are non-monetary.
Direct non-monetary social benets include new and adapted ideas vital to
technical change. Currently, these are contributed largely by people with PhD
or Masters’ degrees who work with complex technologies. ere are of course
others without advanced degrees who have contributed important innovations
such as the Wright brothers, Bill Gates, and Mark Zuckerberg, though most have
had some higher education, especially in recent years.
Direct social benets also include democratization: college graduates are more
likely to support a free press and a viable opposition, to vote, to be more tolerant,
to contribute to democratic institutions such as Congress, state Legislatures,
local boards and commissions, and to hold leading roles in the civil service. Civil
liberties (or human rights) and political stability are other direct social benets:
colleges training lawyers and judges improve the judicial system, and graduates
serve on juries and in other judicial institutions important to the rule of law and
political stability.
Other direct social benets from higher education include contributions to
lower crime rates as graduates remain productively employed leading to lower
criminal justice system costs, better health leading to lower public health and
welfare costs for all levels of government. College graduates also contribute to
local community support for a better environment. Increased access to college
leads to more of these social benets generated by those who otherwise would
not have enrolled.
ere are also private non-monetary benets from higher education –
enjoyed by the graduate and his or her family – such as better own-health, better
child health, better spousal health, greater longevity, and greater happiness.
ese benets, which have been more extensively studied than the external
social benets (see Haveman and Wolfe 1984), Wolfe and Haveman (2003),
Lochner 2011, McMahon 2018, and other surveys), are mainly classiable as
private because the family or household is regarded as the basic decision unit
for nancing investment in undergraduate education.1 ese private benets
to graduates and their families, however, can lead to additional social benets,
such as lower tax rates because they need less Medicaid or other state welfare
assistance.
e preceding discussion has been careful to describe various social benets
of higher education as ‘direct’. In addition, it is necessary to include indirect
benets to measure the total benet. For example, higher education benets
democracy both directly as described above and indirectly as democracy
and greater political stability raise national per capita income, making higher
education’s total eect larger. e indirect eects then are compounded further
over time in a virtuous circle: democratization and higher earnings can increase
political stability, further increasing income and also opportunities for further
education. Although education’s immediate direct eects on democratization
may be modest, the total long-term eects can be substantial. ere are many
other indirect eects like this.
Examples of these relationships have been seen in South Korea and Taiwan,
which started in the 1950’s as poor and authoritarian countries. With their strong
emphasis on education, supporting their export-oriented growth strategies, and
with the build-up of education’s direct and indirect eects, they both became
democracies by 1980. Both countries then continued to have political stability,
growth, and development over the next 40 years, with close to the fastest per
capita growth rates in the world accompanied by greater equality. e eects of
education on democratization, political stability, and development are missed
1. If the individual is regarded as the decision unit with their own budget constraint, as it is in
Haveman and Wolfe (1984) and Wolfe and Haveman (2003), or for graduate students, then bet-
ter child and spousal health, for example, would become external social benets increasing the
proportionate size of the external benets. Nothing else is affected.
eory, Evidence, and Policy Implications 403
402
in most models of growth. Such externalities are also unlikely to be considered
by households in making educational investment decisions, and are probably
underestimated by legislators and voters, too.
:
Endogenous development, focusing on saving and investment in human
capital with externalities that include new ideas, is a fundamental rethinking
of economic development that oers a more general explanation of the
underlying source of economic and social progress. It encompasses, more partial
explanations which are nevertheless useful in elaborating on various aspects.
ese include “export-oriented growth” (that cannot succeed without a skilled
labor force), Schumpeterian “invention and innovation” (that now typically
requires education-based new ideas and skills), the role of “institutions” (that
must be invented and operated based on new ideas and human capital skills), and
“technical change” (oen a label for what is ‘unexplained’). Joseph Schumpeter
rst introduced his invention, innovation, and creative destruction in his book
Capitalism, Socialism, and Democracy, 1943, (6th Ed. pp. 81-4).
Endogenous Development in a Robinson Crusoe Economy
e story told with the formal model and data in later sections of this paper
can be introduced informally in a simple one-person economy where the key
elements of endogenous development of saving and investment in human
capital with externalities are also present. On Robinson Crusoe’s island there
was no money, no markets, no trade, and no institutions like ‘schools’ or the
‘rule of law’. But there was still investment in human capital skills, done then
(and still now largely) by families or individual one-person households (in this
case by Robison Crusoe). It was only later, aer the invention of schools and
universities, rms, and government, that this investment and most production
was done with the help of these institutions.
Robinson Crusoe alone on his island saved by reducing his consumption
of leisure and goods to invest his time in producing a hoe, a physical capital
good. He had also saved time and resources when he was younger in England
to invest in developing human capital skills through education – acquiring the
skills necessary to produce the hoe. He then could use the hoe and his human
capital to produce more food with the higher productivity of his time raising his
standard of living. e new idea for the hoe is ‘technical change’ or greater ‘total
factor productivity’ and is endogenous because it depended on insights due to
his human capital skills. e greater ‘total factor productivity’ is that beyond
the contribution to his productivity made by more physical capital (the hoe)
and more human capital (the skills from England). Additional new ideas might
help him later to oset the diminishing returns that would eventually arise from
investing only in more hoes.
During the time he was not at work, that same human capital helped him
maintain his health, increase his longevity, and improve his environment. Its
contribution thus went beyond increasing his production of food, clothing, and
shelter (the island’s GDP) to improve his overall total well-being or development.
So endogenous development explains all of the progress on Robinson’s island.
Even without trade, markets, money, rms, or government, the basic processes
of saving, investment, technological progress, per capita economic growth,
and per capita development are present. e key factor is how the household
allocates its time.
Endogenous Development Over Very Long Periods
Endogenous development also helps explain development throughout human
history. Going back to 300,000 B.C., homo sapiens saved and invested their
time in producing the human capital skills that included externalities involving
new ideas about how to control re and use it for light, warmth, driving away
wild animals, and for cooking otherwise undigestible food. Also, human capital
formation with externalities resulted in production of new tools for cooking and
hunting. As standards of living began to improve, according to Harari’s Sapiens
(2015), brain size increased. is in turn led to more new ideas and greater
human capital skills that were passed on within the family as parents educated
their children – the rst teachers in the rst ‘school’. ese spread to others and
future generations leading to the invention and production of boats, bows and
arrows, and shhooks. By 70,000 BC according to Harari (op. cit.), some homo
sapiens le East Africa for Europe and by 45,000 BC they landed in Australia.
Soon thereaer they drove all other humans including the Neanderthals from
the face of the earth.
As the cognitive idea revolution continued, homo sapiens settled the rest of
the planet. New human capital formation with externalities generated further
technological change. Later, when the institutions of schools and universities
were invented in the late Middle Ages, younger generations no longer had to
follow in the footsteps of their parents, and development accelerated. Eventually,
following the Renaissance, the industrial revolution began.
Still later, America later the Land Grant Act signed by Abraham Lincoln
sharply increased human capital formation through higher education including
its generation of larger external social benets. In the UK, the Robbins Report
later led to the expansion of access to the “new universities” with more human
capital formation at advanced levels and greater relevance to the UK economy.
In both countries economic growth and development accelerated. Aer WW
eory, Evidence, and Policy Implications 405
404
II, both the GI Bill and Community College expansion in the US and Further
Education Colleges in the UK further expanded student access, human capital
formation, and post-secondary education’s social benet externalities. e
development mission of public colleges and universities – with public support
for larger enrollments, student nancial aid, and government, philanthropic,
and industrial support for research – has since spread across much of the globe.
:
e basic idea of endogenous development started with the work on endogenous
growth of GDP by Romer (1983, 1994), Lucas (1988), Barro and Sala-i-Martin
(1995). McMahon (2002, 2018) extended their analysis of pure GDP growth to
include the non-monetary private and social benet outcomes beyond earnings
and the endogeneity of new ideas. is leads to the broader concept identied as
endogenous development.
e mathematical approach set out below is the one that is most useful for the
design of policy, and most especially policies in education nance. Rather than
just describing the process of development, it analyses the conditions necessary
for per capita development to proceed at the maximum possible sustainable speed.
ese conditions depend on exactly how development is desired as expressed
by the preferences of households and societies, simplied into an algebraic
‘objective function’ for the ‘typical household’. is is an equation that species,
for example, how much the typical household cares about various economic
outcomes such as higher earnings relative to non-economic improvements such
as better health, and how much consumption the household would be willing to
forgo now in order to invest in education to create the human capital skills that
would enable it to obtain a higher level of well-being (or development) in the
future.
e endogenous development ‘model’ contains four other equations, each
describing in a simplied way aspects of how these economic and non-economic
outcomes are produced, including the production of the human capital skills and
the new ideas and knowledge that aect the production of the desired outcomes.
e rst equation describes the production of goods and services by rms using
the human capital of individuals in their jobs. e second equation describes
the production of new or adapted ideas and technologies. e third equation
describes the production by households of nal development outcomes above
and beyond the aforementioned goods and services using their human capital
skills during the hours that they spend at home and in the community. e
fourth equation describes the production of human capital as students invest
their time and family resources using the services of educational institutions.
In summary, the objective function describes what individuals and the society
wants, while the four production equations describe what might be available
(and therefore dene the constraints on the production of economic and non-
economic outcomes). ese ve equations are then combined into one large
equation – known as a Hamiltonian such as that used and explained in Lucas
(1988) – that can be solved using calculus to show the maximum sustainable
speed of per capita development. What makes the solution useful from a policy
perspective is that it reveals which features of the economy and society are
helping to speed up development, what is holding it back, and what levels and
types of investment in human capital are economically the most economically
ecient in leading to earnings, non-monetary outcomes, and the improved
well-being of individuals, families, and the nations in which they live.
Among other things, the solution shows that education externalities are
crucial to sustainable productivity growth. It also shows that there should be
more investment in higher education by households and governments if (but
only if) the rate of return from more investment in education is higher than
the rate of return on alternative investments available to households (or to
governments acting on their behalf). Both these conclusions from the model
have very important implications for education nance and policy.
is endogenous development model focuses on eciency and thus omits
issues of equity, distributive justice, and redistribution. Equity nevertheless is
very important, is formally a part of Samuelson’s (1955) “social bliss”, and must be
taken into account when considering policy implications. For further discussion
of judgements about equity and their implications for education nance see
McMahon (1982).
e Model of Endogenous Development
Readers who are content to accept the conclusions of the model may proceed
directly to the empirical evidence in section V. en there will be no equations
except for one paragraph in Section V with a few simple equations (that also can be
skipped) that explain how non-monetary social outcomes like democratization
are valued. ose who want to know more about the logical underpinning
oered by the theory of endogenous development should read on. ey will
be rewarded by seeing an analytic proof applied within the eld of education
nance, and by seeing the very important implications of that proof for policy in
the eld of education nance. e exposition written here in the deliberate eort
to make it accessible to readers who are not mathematically inclined starts with
four production equations, followed by the objective function equation, and
then turns to solving each equation and deriving their implications for policy.
(More technical detail is available in the endnotes and in McMahon, 2018.)
Production of Goods and Services. In any given year (t), the production of
eory, Evidence, and Policy Implications 407
406
output by rms that uses human capital skills and knowledge, h, is described by
(1) Yt = It[AKtβ(µ1thtNt)1-β]hatγ
Here Yt is output of goods and services as measured by GDP. It is new ideas and
technologies. A is the level of technology, treated as constant in the absence of
new ideas. Kt is physical capital. µ1t is the fraction of time spent at work. (e
remaining fractions of time sum to 1, which is all of the individual’s time in
each period throughout his or her life cycle. Later the remaining fractions of
time spent at school, 2t, or at home or in the community (1-µ1t - 2t) where
the individual’s same human capital skills and knowledge are also used will be
considered.) In equation (1), ht is average human capital or skill per person
(including knowledge acquired through higher education of new technologies
for earlier periods). Nt is the number of persons. e exponents β and (1 – β)
sum to one, meaning that within the square brackets there are constant returns
to scale. For example, a doubling of capital and labor inputs as specied in (1)
would double the amount of output generated by the production function that
has exponents β and (1–β).
e nal term in equation 1, hat, in which human capital appears for a second
time, is educational externalities realized by the community. It is placed outside
the square brackets, meaning that when there are such externalities (i.e. when
γ > 0), there are increasing returns to scale. Doubling the amount of human
capital more than doubles the amount of output, contributing to what growth
accounting calculations label as “total factor productivity growth” or “technical
change”.
Production of Ideas and Technologies. e “new ideas” term in equation1,
It, is also outside the square brackets, raising the amount of output obtainable
from given capital and labor inputs and contributing to technical change. How
new or adapted ideas and technologies (including those used by government
agencies and other institutions, as well as by rms) are generated is described in
a simplied way by,
(2) It = α ht η,
in which the key driver is again the existing level of human capital, ht. Idea
generation is thus a separate education externality that is above and beyond
education externalities contained in hat. New ideas and adaptations do not come
out of the air. Instead, they are endogenous (determined within the system by
past investment in human capital) and accelerated by this investment as reected
in ht, as was rst emphasized by Paul Romer (1986, 1994).2
In advanced economies, the ow of relevant ideas increasingly depends on the
number of people with PhDs from research universities, or at least with some
2. The endogeneity of new ideas distinguishes equations 1 and2 from Lucas’ (1988) production
function, but in other ways they build on his work.
higher education, who can work with and adapt complex technologies. Examples
are adapting Bardeen’s invention of the transistor to modern computers; creating
more eective electric cars (e.g., the Tesla); creating the ideas in economics,
political science, and public service that led to the founding of the United
Nations, the World Bank, and the World Health Organization;. ere was
substantial investment in the education of the microbiology PhD’s that were
essential for development of the coronavirus vaccine, a powerful recent example.
In less developed countries, where universal basic education has not been
achieved, investment in higher education in equation 2 has to be interpreted
more cautiously. Because of great shortages of basic skills at the K-12, or even
K-9 levels (needed to use foreign technologies in local production), and because
basic education is far less costly than higher education, social rates of return to
investment in primary, then junior, and then senior secondary education tend
to be larger than in higher education (Mingat and Tan, 1996). is is true for
the social rates even though private rates of return to a college degree, especially
in Africa, may be high as a result of public subsidies to tuition, and room and
board.
Household Production of Development Outcomes. Households produce desired
non-monetary outcomes during the fraction of their time (1- µ1- µ2) not spent
on the job, µ1, or in human capital formation through education, µ2. Human
capital skills increase the productivity of this time at home or in the community
in a way described by an equation similar to equation 1 for rms,
(3) Cnmt = It’ [ACtβ (1-m1t-m2t) htNt1-β] hatγ’
where Cnmt are non-monetary private and social benets to well-being.
Consumption goods and services which includes durables Ct, appears as an
input rather than physical capital (as in equation 1).3 Household production of
non-monetary outcomes is aided not only directly by human capital, ht, but also
by new ideas, It’ , and by education externalities, hatγ’
.
Equations 1, 2, and 3 are expressed in aggregate terms to show the community
or nationwide eects but can easily be re-expressed in per capita terms by
dividing them by Nt (number of people). On this basis, equation 3 shows that
non-monetary outcomes such as the individual’s better own-health depend
on years of education ht, per capita consumption ct, the level of education in
the community hat, and the fraction of time devoted to household production
including better health (1-m1t-m2t). Equations such as equation 3 have oen been
estimated by multiple regression methods based on micro-economic data for
individuals or households, results from which will be discussed in a subsequent
section. However, the logic of the theory in the equation is helpful in inferring
3. The exponents are not necessarily the same as for production by rms. This is a simplication.
eory, Evidence, and Policy Implications 409
408
causation from education to better health, in this illustration.
Production of Human Capital. Households also produce human capital,
assisted by government funding of schools and universities, as described by,
(4) ∂ hg /∂ t = Gt/Yt δ [µ2] ht
where ∂ hg /∂ t denotes the addition to human capital per unit of time, Gt/Yt is
the fraction of GDP invested by the government in education, µ2 is the fraction
of household time devoted to human capital production, δ is the amount of
human capital acquired per unit of time input, and ht is the initial human capital
of students, parents, and faculty.4 Human capital production is thus a dynamic
process: the human capital of parents and faculty creates skills for the future.
is intergenerational transmission process can be observed by comparing
households of diering education levels within any community.
Gt is treated for simplicity here simply as an exogenous government choice.5
It includes public support of colleges and universities, and student nancial aid.
Larger government support does not drive out private investment, but instead
induces more total saving and investment by households as it lowers the tuition
paid by families and students, inducing larger enrollments and therefore larger
investment in room, board, and in the now lower tuition. at is, matriculation
requires saving as households, including the student, refrain from consumption
and invest more of their resources and forgone earnings in human capital. (e
Gt/Yt ratio is the same on a per capita basis so Gt/Yt = gt/yt.). Parenthetically, for
those not familiar with the economics of education, foregone earnings is the
amount the student could have earned if he or she were not in college, typically
measured by the average earnings of a high school graduate of the same sex. It
is the true economic cost of this part of the investment to which must be added
tuition and fees. e amount spent on room and board and transportation is
a crude approximation of foregone earnings, but the earnings of high school
graduates is a better measure.
Formally, equation 4 refers to all education, for which the rates of return
dier between levels of education as shown in McMahon (2018). However, if the
average return to lower levels of education is treated as a constant, the equation
can be interpreted as a theory of higher education. It stresses that what higher
education produces is not graduates or instructional units but human capital
4. Below the limit, δ, where all effort is devoted to human capital production, there are no
diminishing returns in the production of human capital. Lucas (1988) stresses that this is a social
process that has no counterpart in the accumulation of physical capital.
5. Behind the scenes, Y = C + I + G + F, and G = tY. As such, government expenditure requires
taxes, reducing income and hence consumption. However, if rates of return in which whatever
the government invests are higher than investing in physical capital, I, then Y will increase even
though this budget constraint holds.
skills that benet graduates and society.6 As such it leads to a more meaningful
denition of “productivity” in higher education and exposes policy errors that
arise from cruder measures of productivity – for example, dening the output
of higher education as instructional units (IUs) irrespective of their quality,
or separating research faculty from teaching faculty and thus not exposing
undergraduate students to new knowledge or how it is developed.
If the quality of the higher education is good, it stimulates capacities for
analysis, knowledge of the scientic method, understanding of how truth is
established, and the capacity to evaluate sources and their truthfulness. It creates
the ability to contribute relevant advances because it relates to what is already
known. ese kinds of human capital skills make time more valuable at work
and at home throughout life, reecting the true productivity of colleges and
universities.
Gross vs. Net Investment in Human Capital. ∂ hg /∂ t in equation 4 is gross
investment in human capital formation. Some of the new graduates replace
older graduates who retire or die, but even they increase society’s human capital
because they embody newer knowledge and technologies in each eld. is
replacement investment must be netted out when measuring the average per
capita stock of human capital:
where the rst term on the right hand side adds up all past investment in human
capital and the second term is this period’s reduction (at a rate d) of human
capital through depreciation, obsolescence, and retirement.7 is distinction is
important because much empirical research has understated the true eect of
human capital investment on growth and development by measuring changes in
human capital simply as changes in the stock of graduates and ignoring the new
knowledge embodied in educational investment in graduates who replace those
who retire or die.8
e Objective Function. Policy-makers and the public are assumed to care
about the amount of future per capita total consumption, or well-being, including
of non-monetary outcomes that aect the quality of life, as well as goods and
6. As such it is a more general theory of what higher education does as distinguished from
partial explanations that deal with competition among universities or from what Clotfelter (1999)
insightfully has referred to as “the familiar but curious economics of higher education.”
7. There is an analogous stock accumulation equation for physical capital, K, behind the scenes.
This discussion of both human and physical vintage capital ignores the Cambridge controversies
(cite) and problems with summing over different vintages of capital with different embodied
technologies.
8. This means of transmission of new technologies through their embodiment in new human
capital as it is formed also counters the critique that endogenous growth (and development)
models fail because they do not provide for technology dissemination.
eory, Evidence, and Policy Implications 411
410
services (not just earnings). is is described by,
which sums up the total consumption (ct) of all future periods while discounting
(at a rate ρ) the value of more distant consumption. is means that including all
future periods in the equation rather than just the normal human life span makes
no signicant dierence to the results.9 e coecient σ reects preference for
steady rather than uctuating consumption. Consumption of goods and services
of non-monetary (i.e. subscript nm) outcomes, i.e. cnmtζ, are combined as total
consumption i.e., —ct = ctε cnmtζ, in which ct is the usual personal consumption, a
part of monetary GDP expenditures as before, and the Greek-letter parameters
express the relative strength of people’s preferences for these monetary and non-
monetary forms of consumption.
A high discount rate ρ would imply that returns to education in the distant
future had little inuence on decisions. is could reect the oen relatively
myopic or short-term view of students, households, and legislators in the real
world. However, households and legislators making decisions in the real world
about how much to invest in education, I think, more realistically also behave
in an iterative fashion by rst looking forward and making initial decisions as
the model suggests, then move forward through time, and at decision points
repeat this process while re-evaluating and revising their saving and investment
decisions.
A low ζ would imply that decision makers put little weight on non-market
outcomes.10 A possible reason could be that non-market outcomes are not well
understood or perceived, due to poor information. Universities and US or UK
national and state education ocials should thus try to better educate students,
parents, and legislators on the true non-monetary private and external social
benets of education, similar to ‘truth in labeling’ laws for foods and drugs,
‘truth in lending’ laws for banks, and eorts to improve information in many
other markets where information is poor.
Optimal Development
Since the objective function is dened in per capita terms, it has been convenient
in solving the endogenous development model to do the same for all the other
9. It is known that there is no signicant difference in the rate of return between using, say,
a typical 65-year life span vs. using an innite time horizon. See “Innite Planning Horizons”
(Lang and Marino, 1993).
10. If (ε + ζ) = 1 there are constant returns and, in the absence of risk aversion, the intertempo-
ral elasticity of substitution is unity leading to explicit equilibrium paths here as in Lucas (1988)
and Hu (2008). However, it is intuitively more realistic not to impose this gratuitous assumption
which then requires a numerical solution.
equations by dividing through by Nt (number of people). e current value
Hamiltonian equation which is standard in that it combines the foregoing
equations but without time subscripts, is:
e variables are all as previously dened. What is to be maximized is the
discounted present value of the typical individual’s well-being (the rst term
on the right hand side), subject to three constraints: following θ1 is the amount
of investment in physical capital (per capita production minus consumption);
following θ2 is per capita investment in human capital through education; and
following θ3 is the per capita human capital cost of producing non-monetary
outcomes, net of the goods and services involved in their production. Once
human capital is created (in θ2), however, it may be used to produce non-market
development outcomes at no additional cost in which case the term following θ3
is zero. All of the processes described in equation 7 also apply (as does the whole
model) to much earlier times or to single-person Robinson Crusoe economies
as discussed earlier.11
With the endogenous determination of the volume of new ideas, I = αhη,
substituted into equation 7, the rst order conditions that must hold for
maximizing total well-being are obtained by dierentiating this Hamiltonian
equation with respect to the decisions that the typical household has to make.
ese are decisions about: market consumption, c; non-monetary consumption,
cnm; the fraction of time spent working, µ1; and the fraction of time spent in
education, µ2 (which then also determines (1-m1-m2), the fraction of waking
time spent in household production of nal satisfactions spent at home or in the
community). It is then possible to derive the optimal path of development. is
is the path along which well-being is maximized at a sustainable rate and along
which market consumption and non-market consumption, as well as per capital
income, y, physical capital, k, and the individual’s human capital, h, all grow at
the same constant rate.
One important result from this analysis is a statement of the conditions
required for optimal per capita economic growth, (∂y/∂t) /yt. is is the same as
maximizing per capita market consumption over time, (∂c/∂t) /ct, namely,
(8) (∂c/∂t) /ct = (∂y/∂t) /yt = MPPKt – ρ = MPPHt – ρ
11. However, for early history applications, since there was no government, g/y would be
dropped by setting it equal to 1. In earlier millennia per capita growth was also very slow.
eory, Evidence, and Policy Implications 413
412
is says that the optimal rates of growth of per capita consumption12 and
per capita income are equal to the marginal productivities of physical capital,
MPPKt , and of human capital, MPPht , discounted at rate ρ.13 ese marginal
productivities are rates, which are pure numbers, so although they are shown
here in capital letters which designate aggregates, they are the same rates at a
micro level, and can be compared to any and all rates of return. In a competitive
economy without major distortions by monopoly or long lags these marginal
productivities become equal through the competitive process to the average
rates of return on investment in physical and human capital. us it can be
concluded that if the rate of return to investment in higher education, MPPHt –
ρ, is higher than the rate of return to physical capital, MPPKt – ρ, as measured by
a benchmark such as the average return on bonds or Standard and Poor’s (S&P)
500 index funds, and stays higher, then there is underinvestment in higher
education. It can also be concluded that increasing the rate of investment in
higher education would result in a higher sustained optimal rate of per capita
economic growth. e latter is the economically ecient rate of growth, and
also reects the economically ecient rate of investment. is is supply side
economic growth, not just the cyclical recovery from recessions, and in the
longer run it is not limited by diminishing returns to physical or human capital.
For these two reasons it is sustainable without bounds!
is brings us to the central focus and special uniqueness of this paper, which
is the conditions for attaining economically ecient rates of development (or
improvement in human well-being) that include the non-market outcomes
of higher education above and beyond earnings. ese are the private non-
monetary benets such as better health and the social benets such as
improvements in democracy, human rights, and political stability, lower crime,
and increased tax revenue as earnings rise.14 If the social value of these non-
market consumption outcomes is not fully understood, (ζ <1), investment in
education would be lower and the rate of development of the family and the
12. Strictly, preferences for per capita consumption using a coefcient for risk aversion, σ.
There are also standard human and physical capital accumulation equations behind the scenes.
13. The optimal sustainable growth of per capita income is the same as the optimal growth of
per capita consumption along the balanced growth path as in all standard endogenous growth
models, i.e., (∂c/∂t)/ct = (∂y/∂t)/yt . To get to this, the values of two of the endogenous variables,
per capita consumption, ct, and the fraction of time devoted to production of human capital, µ2,
are rst selected from the rst order conditions ∂H/∂c and ∂H/∂θ1 obtained from differentiation
of the Hamiltonian in solving to achieve the optimal path for investment in human and physical
capital (the state variables).
14. The optimum rate of per capita growth of only development outcomes beyond earnings,
cnmt , which is also determined endogenously, is equal to the optimum rate of growth on market
consumption, ct ,and to the optimal rate development (δ ̅yt/δt)/ ̅yt , in balanced growth. This uses
̅ct = ctε cnmtζ from equation6 and the denitional identity ̅y = ̅c + i + g + f [should these be capi-
talized as above in footnote 5?] as before, but including the non-market development outcomes in
total consumption and total income.
nation would be slowed. e additional investment cost of producing these non-
monetary outcomes is typically zero, since they are produced using the same
human capital that is being used as on the job. As such, achieving the optimal
rate of development requires that the marginal productivity (or rate of return
to investment) in physical capital be the same as the marginal productivity (or
rate of return to investment) in human capital. e cost of producing some non-
monetary outcomes may not be zero. For example, governments may expand
investment in higher education explicitly to generate additional non-monetary
outcomes, or spend more on academic disciplines whose graduates contribute
mainly to external social benets, so that the details of the term following θ3 in
equation 7 are not zero and enter the solution.
In either case, the monetary rate of return to investment in higher education
can now be added the rate of return from these non-monetary development
outcomes. is can be done because the cost of producing these two outcomes
which is in the denominator is nearly identical. If this total return is higher than
the return on physical capital, then more investment in higher education will
raise the per capita rate of development toward an optimal rate for each family,
and in the aggregate, for each nation. is is again sustainable without bounds!
Higher Education Externalities. e rate of growth of per capita or per family
sustainable development and of total consumption (monetary plus non-
monetary) on the le in equation 9 is related to the rate of growth of human
capital on the far right by,15
Since β, η, γ, η’, and γ’ are all positive, equation 9 is an analytic proof that the rate
of growth of per capita development (and of total consumption) is larger when
the rate of growth of human capital investment is larger. is is unsurprising,
since investment in human capital is an important cause of development. More
important (and the third very important result to highlight in this article), it
is also a proof that any given investment in human capital raises total factor
productivity growth. is is as a result of the positive education externalities in
the numerator of equation 9 making the total coecient of (∂h/∂t)/ht greater
than 1. ese social benet externality eects are from the production by rms
of goods and services (γ and η) and production by households of non-monetary
15. To reveal the role of higher education externalities, ones should differentiate the Hamiltonian
in equation 7 with respect to time. It is of the form F = f(Kt,ht) = X, where X is a constant and
therefore goes to zero. From this, the common growth rate of per capita sustainable development,
(∂ ̅y/∂t)/ ̅yt which is equal to the growth rate of non-monetary consumption, (∂ ̅c/∂t)/ ̅ct, is related
to the rate of growth of human capital.
eory, Evidence, and Policy Implications 415
414
outcomes (γ’ and η’).
e fourth important result is a corollary of the foregoing. Since private
incentives are too weak to induce enough private investment in human capital to
generate enough externalities to achieve a socially optimal level of development
(since most of the benets are not received privately by those who invest), the
total factor productivity growth in equation 9 is largely dependent on public
investment or charitable endowment-supported investment in higher education.
Conclusions from the logic of the model are that public investment in higher
education including investment in generating PhD’s and others that produce
new technologies throughout their lives, leads to higher rates of sustainable per
capita growth and development. However, for applications in poor countries,
as mentioned earlier and repeated here for emphasis, caution is needed. In less
developed nations, monetary social rates of return are oen higher for basic
education because basic skills are relatively scarce when basic education is not
yet universal, and because unit costs for basic education are lower. Per student
costs of higher education are relatively higher, especially in Africa, making social
rates lower (and private rates of return higher). Because data for calculating
the social rates include the full social costs, such as the high public subsidy for
room and board with little cost recovery from wealthier households, the private
rates of return do not include these full costs to the society. Also, as previously
mentioned, policies oen are not implemented to reduce the incentives of college
graduates to emigrate. As such the social costs are high and returns are zero
except for remittances. In places where this is true, it is usually more ecient to
invest in basic education rst while expanding higher education more slowly.
Nevertheless, some higher education externality benets may help the
productivity of others and raise real wages if wealthy graduates do not
manipulate the political system to their own advantage. ere is some evidence
that for workers of any given skill, real wages in developed countries where more
individuals have higher education are higher than wages for those same skills in
poor countries (see Moretti, 2004).
Finally, the implication of the model is that if states or national governments
give greater support to public colleges and universities, the external social benets
are larger. is leads to higher sustainable rates of economic development –
higher than those in competitive economies that do not invest as much in higher
education. is is largely due to higher total factor productivity growth, which
public investment in higher education with its externalities supports. e model
also implies that those academic disciplines that generate proportionately larger
external social benets make very important contributions to productivity
growth and to higher rates of per capita development.
e empirical questions now become 1. How large are these external social
benets and what is their value? 2. Should there be more public and household
investment in higher education for economic eciency and optimal per capita
economic development? and 3. How far should privatization in higher education
go? ese all lead to important policy guidelines for the eld of education nance.
e empirical estimates sum up the results of existing research including that
by the author (McMahon, 2017, 2018) on the size and value of the non-monetary
social benet outcomes of higher education. e results by the author are new
in avoiding overlaps and are expressed in common units of measurement, so
they can be summed up to obtain a total. is paper builds on McMahon’s
(2018) refereed journal article and books (2002 and 2017) to which the reader
is referred for more specic detail on how each social benet is measured. is
allows the present paper to be new beyond my prior work in explaining how
public goods outcomes are valued, to focus on higher education externalities,
and to discuss later the degree of privatization in higher education at dierent
types of institutions that is optimal. is paper is also new beyond my prior
work in that it develops important guidelines for education nance such as
the size of the investment in higher education that is optimal for development,
whether or not there is underinvestment in higher education separately for
optimal development as well as for optimal GDP growth, and when decreasing
the public investment per student is appropriate or has gone too far in increasing
privatization (in nancing) for economic eciency.
How Are the Direct External Social Benets of Higher Education Measured?
e direct external social benets of higher education in the summary that
follows are measured separately for each benet by standard (multiple regression)
methods by many authors in articles published in refereed journals. Authors in
this special issue additionally add to this literature by measuring new aspects of
many social benets.
To be included in this overall summary, studies must use logical specications
that are linear approximations of the household production function in equation
3 in the model above. ey must also use appropriate lags and other controls
that meet scientic standards including obtaining education coecients that are
statistically signicant at the .05 level or smaller. For example, with a specic
social benet such as lower crime rates as the dependent variable on the le,
per capita income (or consumption) must appear as a control variable on the
right as required by the theory of household production in equation 3. is is
a necessary control for when the intent is to add up the total benets of higher
education and to avoid double counting the earnings benets. Years of higher
eory, Evidence, and Policy Implications 417
416
education, or merely years of education that include two to four years of higher
education, must also appear on the right with other control variables.
For community, or economy-wide, benets such as crime rates,
democratization, human rights, or political stability, lags of about 15-20 years are
logical such as those used by Keller and Oketch in this special issue or McMahon
(2002, 2007). is time lag is long enough for a college graduate to settle on his
or her lifetime age-earnings pattern with earnings and non-monetary outcomes
that are typical of a graduate’s life cycle.16 e other control variables mentioned
must control for all other signicant eects on the dependent variable if there
are to be controls for the ceteris paribus conditions to be able to infer causation,
an inference that is made from the logic of the theory. It is a causal relation
that is sought, never merely a simple correlation without appropriate controls or
without an underlying theoretical rationale. Other insignicant eects collected
in the disturbance term average out.
It is new in McMahon (2018) and in this paper is to use regressions for each
specic social benet outcome in per capita terms rather than an aggregate
inference of externality eects. is is a better, and probably the only, way of
estimating the size of non-monetary externalities. is approach therefore
improves on most macro growth equations in the economics literature which
focus only on GDP. ese growth equations typically ignore all non-monetary
direct private and direct social benets of higher education, which are the primary
focus of this paper. is macro growth equation literature does reect indirect
eects of non-monetary outcomes on GDP, but they omit all indirect eects
on non-monetary outcomes and as a result grossly underestimate education
externalities. For an excellent survey of macro growth equation regressions see
Davies (2003).
Direct Social Benet Externalities. Empirical estimates of the amounts of each
social benet from four years of college beyond high school and their value are
shown in Columns 1, 2, and 3 respectively in Table 1. e rst 16 listed are direct
external social benets, and the 17th below the sub-total covers indirect benets.
All quantitative estimates of direct benets are made by regression methods (as
explained above) except for prison costs and additional taxes paid, which are
monetary estimates converted to 2016 dollars. Since all control for per capita
earnings (or income), these are values of non-monetary outcomes over and
above earnings. ese results, therefore, can be added to earnings to get the total
value of higher education outcomes.
To get the total eect of higher education on each separate social benet listed,
it is necessary to add the relevant portion of the indirect eects shown near the
16. A time lag of this length also means that the relation is recursive which reduces possible
bias due to simultaneity.
bottom of Table 1, which averages 37 percent of total direct eects (McMahon,
2002) and is assumed to be similar for earnings and social benets. It is also
necessary to adjust downward for net ability bias (about 6 percent) and upward
for longitudinal trends (about 1.35 percent, because age-earnings proles shi
upward over time (see Arias and McMahon, 2001 and their comparison of
estimates with and without trends). e data and sources for direct social benets
and for valuations including these two adjustments are in an Excel spreadsheet
and Technical Appendix and are downloadable, providing for reproducibility
(McMahon and Godinsky, 2020).
Democracy, Civil Rights, and Political Stability. A preliminary estimate of
the size of the democratization (or political rights) eect in Table 1 Column
1 of .0032 using the Freedom House civil rights index is based on an average
of four regressions that use cross country and sometimes panel data. e one
from McMahon (2002, p.98), for example, estimates eects from four years of
secondary education, while controlling for the log of per capita GDP lagged 7
years, and military expenditure as a percent of government expenditure lagged
1 year (which has a highly signicant negative eect on democratization). It has
an R2 of.49. e other three equations have a similar but not identical structure.
ese details are discussed in a Technical Appendix by McMahon and Godinsky
(2020) where they are averaged in the spreadsheet cited there. An analysis of
the contributions of secondary education as compared to higher education to
democratization also using worldwide panel data by Katrina Keller follows as
the next article in this special issue. Her deeper analysis of this specic point
has worked out well, as was hoped. She has similar specications but was able to
better sort out eects as between secondary and higher education, and also as
between the developed and less developed countries. Relevant to the estimates
here, she nds the eects from higher education to be even larger and more
highly signicant than the eects from secondary education (her Table 2) in all
of her models that include both.
So, the estimate of four year of higher education on democratization in Table
1 may be as large or larger than the rst pass estimate of .0032, which then is
valued at $2,274 per graduate per year.17.
is value is estimated as the cost of producing the same result by means other
than education. It is an adaptation of the method developed by Haveman and
Wolfe (1984) to the problem of valuing public goods since individuals normally
cannot purchase these directly in the market. Using the coecient on income in
the regression that explains democratization, one can estimate how much of an
increase in average per capita income it would take with education held constant
17. This $2,274 connects to the slightly smaller $2,207 on the spreadsheet (McMahon &
Godinsky, 2020).
eory, Evidence, and Policy Implications 419
418
for democratization to increase by the same amount.
Since this method of valuing public goods is important, and goes a step beyond
Haveman and Wolfe (2007) in adapting their method to valuing public goods
like democratization and political stability (i.e., not benets to other members
of the same family which are non-monetary but can be addressed by purchasing
things in the market), it is useful to spell it out more specically. Consider the
regression estimated for democratization, D, in per capita terms:
(8) D = αy + βe + u, so the dierentials,
(9) ∆D = β∆e = .0008 where ∆e is not the contribution from four years in Table 1
(i.e., .0032), but instead the contribution from one additional year of education.
To produce this change in democratization by other means, it can be done by
raising average per capita income according to the income (y) coecient (with
education held constant) by:
(10) ∆D = α ∆ —y.
From (9) and (10), by solving (9) for ∆e, and replacing ∆D in this using (10):
∆e = (α/β) ∆ —y, the value in dollars of one additional year of education that we
seek. If e = four years, the value of a four year degree for producing the .0032
improvement in democratization is $2,274 as shown in Table 1.18 e values
of other communitywide public good direct benets in Table 1 such as human
rights and political stability are estimated in the same fashion. For further
details on the calculations, numbers and sources of regressions underlying each
number in Column 1 of the table, data, and discussion of standardization of
all the coecients (all needed for replication) see the Technical Appendix by
McMahon and Godinsky (2020).
Although democratization has not been found by the author to contribute
directly to growth, it does contribute signicantly in worldwide data to political
stability over longer time periods. is in turn contributes to growth as an
indirect eect of democratization (McMahon, 2002). Other studies which
the estimate the eects of education on democratization, civil rights, political
18. To show how this is equivalent to the valuation method developed by Haveman and Wolfe,
but adapted here to value public goods, Haveman and Wolfe (1984, 2007) assume that in private
markets, where the exclusion principle prevails, households typically equate the ratios of the mar-
ginal products of the inputs, β for education and α for other goods represented here by per capita
income, to the ratio of their prices: β/α=P(E)/P(Y)
Here, the value of P(Y), i.e. ∆D, represents the change in average per capita income needed
for a 1-year change in the dependent variable, in this case democratization, D. With values for
P(Y) and β/α, we can calculate the value of P(E), the “price” of one additional year of education,
which is β/α ∆e, and multiply by four years to obtain the value of a four year bachelor’s degree
in producing this .0032 improvement in democratization, the same as discussed in the text. Since
∆e = (α/β) ∆ ̅y from equations 9 and 10, if e = four years, it is easier to use this ratio α/β for valu-
ations in the spreadsheets. Where e is not four additional years of education for everybody but
instead is the higher education enrollment rate, an additional conversion is needed to account for
the non-universality of higher education.
Table 1. e External Social Benets of Higher Education*
Social Benets 4-Years Units of Measurement Estimated Value**
Democracy, Civil Institutions 0.0032 Democracy Index, 1-7 $2,274***
Civil Rights (Judicial) 0.0054 Human Rights, 1-7 $1,209***
Political Stability 0.0135 Political Stability, 1-100 $3,769***
Fertility, Life Expectancy -0.336 Net Growth Eects ($1,268)
Less Inequality Depends on 2-Yr & Grants
Less Poverty & Fertility 0.424 Poverty Index $518
Lower Murder Rates -4.52 Homicides per 100,000 $3,823
Less Other Crimes -315.6 Crimes per 100,000 $8,037
Lower Prison Costs $2,508 Muennig (2000) in 2016 $ $2,630
Additional Taxes Paid Annually, Present Value $3,824 A Monetary Return in $ $3,824
Less Water Pollution 1,216 Water Pollution Index $378
Less Air Pollution 0.56 Air Pollution Index $4,655
Less Forest Destruction 0.00002 % Change in Acres $3,922
Social Capital & Cohesion - Overlap Items above $0
Increased Wages of Others 1.28 Moretti (2004, p.28) & McMahon (2002) $302****
TOTAL DIRECT EXTERNAL SOCIAL BENEFITS $34,073
+Indirect Eects (37% of Direct Benets (Earnings + NM Social & Private) [e 37% is from simulations in McMahon (2002)] $43,050
TOTAL DIRECT & INDIRECT EXTERNAL BENEFITS $77,123
EXHIBIT: PVT EARNINGS BENEFITS, Beyond HS, AV.M&F $32,102
EXHIBIT: PRIVATE NON-MON. BENEFITS (McMahon, 2018) $50,176
TO TAL $159,401
Notes: *Source: McMahon (2018, p.106); **In 2016 dollars; ***ese reect reduction for nonlinearity; ****Approximated as median salary in 2016 ($45,762) x
1.32% (from Moretti)/2(for urban only).
eory, Evidence, and Policy Implications 421
420
stability, and on other social benets in Table 1 are discussed in McMahon
(2018). Other new articles by Keller, Oketech, and Teixeira et.al. that follow in
this special issue further explore the above types of social benets of higher and
secondary education in dierent areas as was indicated in the Introduction to
this special issue.
Lower Fertility Rates and Longer Life Expectancy. Increased education
contributes to lower fertility, but also to lower infant mortality and longer life
expectancy. e net eect initially, however, is that more primary education
for females contributes to higher net population growth rates, which lowers
economic growth expressed in per capita terms as shown in Table 1. ese are
the more immediate direct eects, important particularly in less developed
countries. Lower fertility rates later swamp the eects of primary and junior
secondary in lowering infant mortality aer females complete about 9th grade
(McMahon, 2002). is lowers net population growth which increases per capita
income. is eect is included in Table 1 as longer run positive indirect benets
through lower population growth, which shows up through higher per capita
income and lower household poverty. ese indirect eects are included in the
total indirect benets of $43,050 toward the bottom of the Table 1. Additional
evidence specic to higher education as these indirect eects on family size do
continue through the college years in this special issue reinforce earlier work and
should help these eects to become more widely recognized. e eect of female
education beyond 9th grade in lowering net population growth rates is especially
important in Sub-Saharan Africa and South Asia where population growth rates
remain high. e UN’s ‘Education for All’ policy of lowering high population
growth rates has helped in the turnaround in per capita growth in Africa since
2000, but this fertility eect on per capita income is oen overlooked.
In developed countries, there are positive indirect eects from lower fertility
rates and family size as well as slower population growth that continue throughout
the years females are in higher education and raise per capita income. ese
are indirect eects on growth and so they are a part of the total indirect eects
of college valued at $43,050 toward the bottom of Table 1. But with respect to
direct eects, higher education also contributes to greater life expectancy. is
generates higher social security costs. So, in the regressions, aer controlling for
per capita income, the direct eects have a net negative -$1,268 estimated eect
on per capita GDP growth per year as shown in Row 3 of Table 1.
Inequality and Poverty. With inequality high and rising in the US, UK, and
elsewhere, there is considerable interest in the eects on inequality and on
poverty due to higher education. ese eects are complex and dier due
to government policies. However, two patterns can be observed. First, as
mentioned earlier, the onrush of new technologies and automation has le the
majority of the population in the US and UK that have high school education
or less with real wages that have fallen since 1980. However, those with college
are about 47 percent better o, according to BLS earnings data by education
level. is failure of human capital skills and knowledge to keep up with
technology due to insucient education and continuing automation contributes
to inequality and dissatisfaction. is strongly suggests that expanding access to
community college associate degrees through lower tuition and more lifelong
learning would reduce inequality and poverty while also increasing economic
growth. e dissatisfaction and political reactions to automation and the loss
of manufacturing jobs and the failure of skills and knowledge to keep up is not
unique to the US but has led to political reactions also in the UK, Hungary,
Poland, Brazil, and elsewhere.
But beyond this role, especially for community colleges, there is little evidence
available to use in Table 1 on the overall net eects of higher education on
inequality and poverty. is is a research gap. However, there are eects as
higher education reduces fertility, as cited above, and from two-year degrees
in lowering poverty that lead to the modest net estimates of $$518 in Table 1.
But the overall eect (since most student loans and some colleges benet the
wealthier families) depends heavily on whether two-year public community
college degree programs are fully funded, as well as on increased funding of
Pell Grants and other need-based scholarships. Keller’s paper in this special issue
newly reports signicant but modest reductions in poverty rates from increased
higher education enrollment rates in less developed countries, although Oketch’s
paper nds this does not happen in Africa for reasons he discusses. Also, Keller
also nds even stronger eects on reducing poverty from increases in completion
of secondary education.
Crime Rates and Prison Costs. Reductions in murder rates, other crime rates,
and prison costs have been shown to follow higher enrollment rates in 2-year
community college programs by Lochner (2011). Muennig (2000) estimates of
these savings of $3,823, $8,037, and $2,630, respectively, are shown in Table 1.
Additional Taxes Paid. e typical college graduate is estimated to pay $3,824
more in Federal and state income, and sales taxes annually than the typical
high school graduate. ere are additional indirect benets to others from this
included at the bottom of Table 1.
Benets to the Environment, to Other’s Wages, and to the Flow of New Ideas.
Environmental warming continues. In regressions, higher community college and
sometimes, university enrollment rates are associated with better environmental
outcomes aer controlling for growth eects. For example, regressions show
less water pollution, less air pollution, and less forest and wildlife destruction.
(McMahon 2002). ese are valued at $8,955 per year in Table 1. is is hopeful
eory, Evidence, and Policy Implications 423
422
in a world facing serious problems from global warming.
Wages of others in the community are raised by an average of $302 per college
graduate per year, an estimate by Moretti (2004) in Table 1. Hermannsson
et.al.2016) nds analogous eects.
New ideas generated over the life cycle by college graduates are very important
to new technology and per capita development. But more research is needed.
Hermannsson et. al (2016) nd signicant eects on technologies from PhDs in
rms in Scotland. Also, given that most of the literature other than this focuses
on specic examples but little systematic work on the overall net impacts needed
for Table 1, it is important to mention the indirect connection to work on the
eects from investment in research and development (R&D). For example, Keller
in this special issue does nd signicant eects of higher education enrollments
on investment in R&D which now typically requires prior investment in the
advanced training of PhD’s who use and adapt new ideas. is positive eect is
signicant in Keller’s separate regressions for developed, less developed, and all
countries globally.
Total s. e value of all direct social benets in Table 1 totals $34,073 per
graduate per year, an annual stream of benets that continues throughout
the graduate’s life cycle. is is a conservative estimate because it includes no
estimate of the contribution made (largely by college graduates) to the volume of
new ideas and their adaptations. ese benets and all of the other direct social
benets in Table 1 all totally omitted by studies that use the growth equation
approach that tests for the eects of education and/or higher education on GDP
with the latter as the only dependent variable.
Indirect benets, all of which are externalities, estimated as $43,050 toward the
bottom of Table 1, are generated through education’s direct eect through any
intervening variable on both monetary (GDP) and non-monetary development
outcomes. ese latter non-monetary direct development outcomes are also
ignored by the growth equation approach just dened briey at the end of the
last paragraph and explained earlier, which does, however, pick up the indirect
eects of education externalities on GDP. e value of these indirect eects in
Table 1 is estimated by simulations of the dynamic dierence equation system
determining education outcomes involving interaction among the various
eects of education that accumulate over time as shown in McMahon (2002).19
e $43,050 is calculated as the 37 percent of the total direct benets (dened
shortly) of $116,351. e total direct benets are calculated as the sum of the
direct social benets of $34,073, plus the direct private monetary benets of
$32,102 plus the direct private non-monetary benets of $50,176, as shown with
sources at the bottom of Table 1.
e grand total of all benets from higher education above those realized by
high school graduates, direct and indirect, private and social, monetary and
non-monetary, is $159,401 per yr. Of this total, only $32,102 is additional annual
earnings. A couple components are approximations based on a small number of
studies, and the estimated total is conservative in the sense that it must omit any
value for new ideas and adaptations. Nevertheless, this suggests that education
nancing policies at the campus and state levels should be based on far more
than just jobs and earnings outcomes.
Has Privatization Gone Too Far?
Reduction of public support per student by most states in the US and England at
Associate, or Short Degree, and Bachelors’ levels has forced institutions to shi
toward charging much higher tuition and fees to students and their families.
is has led to accumulating student loans and large debt, and now constitutes a
much larger degree of privatization in the nancing of public higher education
than has existed for a long time. e cut in per-student public support has also
led to larger classes, more classes taught by adjuncts rather than regular faculty,
and department budget allocations based on the number of majors (so students
gravitate toward elds with higher initial earnings and faculty salaries reduces
support per student where social benets are highest).
e endogenous development model and the results in Table 1 provide a
useful guideline for the optimal degree of privatization of higher education,
which is the percentage share of the total costs of higher education that should be
supported by taxation for there to be economic eciency. is share that external
social benets are of the total value of all of the benets of higher education.
Households have insucient incentives to pay for the external social benets,
19. To measure indirect effects from investment in education, a system of difference equations
is estimated by multiple regression methods, one for each non-monetary education outcome. Af-
ter the regression coefcients are obtained, these equations can be solved recursively, generating a
dynamic time path for each of the non-monetary outcomes. It includes both the direct and indirect
effects from the initial investment in education. To obtain the indirect effects, it is possible to rst
suppress the indirect effects by setting their coefcients to zero, and then subtract the results for
each year along the time path from the total effects. This is done in McMahon (2002) with simula-
tions over 40 years shown for 14 advanced and less developed countries worldwide and in Appia
and McMahon (2002) for 13 countries in Sub-Saharan Africa. It is assumed that the ratio of direct
to indirect effects is the same for higher education as it is for all education.
eory, Evidence, and Policy Implications 425
424
direct or indirect, since they are largely received by others, which includes future
generations. So if these are not supported publicly, the external social benets
will be lost, productivity growth will be lower (based on the model), and the
rate of economic development for the household and the country will be sub-
optimal (i.e., inecient).
More specically, the direct plus indirect social benets are estimated in Table
1 to be $77,123. e total benets are $159,401, which also include the private
benets households receive and for which those households can be expected
to pay. e private benets consist of the annual increment above high school
earnings of $32,102, plus direct private non-monetary benets of $50,176 (or
to the extent they are perceived).20 Social benets are thus 48 percent of total
benets, and private benets 52 percent, implying that privatization of up to 52
percent of the total nancing costs at public colleges and universities (which are
not just the institutional costs) would be appropriate for achieving an optimal
rate of development.
Considering how this guideline applies to dierent types of institutions. At US
public research universities, privatization has reached about 60 percent of the
total costs of the investment in human capital formation (which include foregone
earnings costs) for in-state students, a bit above the 52 percent guideline.21
Privatization at the four state comprehensive universities and many local
public two-year community colleges (as average tuition and fees plus foregone
earnings costs paid by students and their families as a percent of the total costs),
surprisingly, is drastically above the 52 percent guideline. is has occurred as
state government support per student has fallen nationwide and in Illinois, but
also because foregone earnings, (i.e., earnings of comparable HS graduates),
that are also approximately the room, board, books, and transportation costs),
largely borne by students and their parents, even when students live at home, are
a larger fraction of the total costs22.
In England, privatization has gone still farther, changing in the last two
decades from a place where tuition was free, to private nancing approaching 100
20. Remembering that the indirect private benets are social benets and are included there.
21. For Bachelor’s degrees at Big 10 public research universities, for example, tuition and
foregone earnings costs each average about $80,000 per student for a four-year degree. Since
institutional direct costs are about $142,992, this means households are paying privately about
68 percent of the total investment costs. This percentage could be higher because not all costs
are instructional costs. However, this is offset by student nancial aid. About 25 percent of the
students receive need-based federal grants (generally Pell grants), need-based state government
grants, or other nancial aid awards that reduce the privatization percentage by about 8 percentage
points to about 60 percent for in-state students.
22. Further explanation and data sources are available in McMahon (2017) and on my website
https://publish.illinois.edu/mcmahon/ , to download the 2016 Spreadsheet for calculating 2016
rates of return, and see the columns for institutional direct costs and foregone earnings costs at the
different education levels. The data shown there and updates are at www.census.gov/ , click on
CPS>Income Tabulations>Persons>PINC-04.
percent, far above the 52 percent guideline. ere is a large income-contingent
loan system, complex to evaluate, but any loans repaid are still private costs.
Universities now charge full cost, or £9,250 in tuition and fees, the highest in the
world – except for Scotland where tuition is free for residents. Further education
colleges approach this high privatization rate as well, but also students benet
from income contingent loans. Again, in these colleges foregone earnings costs
are a larger percent of the total costs. Public support for teaching in higher
education in the UK has been cut from 29 percent of higher education’s direct
costs in 2000 to 7 percent currently. Tuition and fees rose from 8 percent to
29 percent, an almost exact replacement of public support of tuition and fees
with private support. Average student debt at graduation is currently just over
£50,000 in current prices, more than twice the average student loan debt for US
graduates even from the most selective institutions. For UK gures and further
analysis including positive aspects of the degree of privatization in England see
Murphy, Scott-Clayton, and Wyness (2019). However, all this has led to strikes
and discontent in UK institutions. Some colleges are struggling, as are some
academic elds where social benets tend to be relatively larger, in both the UK
and US.
Is ere Over- or Under-Investment in Higher Education?
e model and the evidence have important implications also for the question of
whether there is currently over- or under-investment in higher education, at each
degree level. For the rate of economic growth per capita to be optimal, investment
in higher education must be large enough for its social rate of return, based
only on earnings, to equal the rate of return on physical capital. Empirically, the
earnings-based social rate of return to investment in a bachelor’s degree is 10.9
percent in 2016 dollars. is is the average for males and females corrected for
net ability bias and longitudinal trends. is 10.9 percent is higher than the 7.22
percent average return on physical capital using the 10-year average return on
S&P 500 index funds available to households (e.g. Fidelity Investments or other
companies online) as a benchmark. is indicates under-investment at the four-
year level in higher education. At the Associate degree level, the earnings-based
rate of return is a higher at 12.9 percent (from McMahon 2018), so the under-
investment is larger than at the bachelor’s level. is indicates even larger under-
investment at the Associate’s degree level. Taken together, this reveals that, when
the sustainable pure economic growth rate is the goal, one that largely ignores
the value of non-monetary returns beyond earnings, there is underinvestment
in all undergraduate higher education. In recessions, rates of return tend to be
higher because of lower foregone earnings costs.
If the goal is total well-being, or development, the non-monetary private and
eory, Evidence, and Policy Implications 427
426
social development outcomes summed up at the bottom of Table 1 must also be
considered. In this case, the total return is 39 percent or so, much farther above
the benchmark rate.23 So the underinvestment at the bachelor’s and Associate
Degree levels is even more serious.
e pattern is similar in England, based on UK data (See McMahon &
Oketch, 2013, and OECD data). e returns for US community college dropouts
are lower, as they probably are in UK Further Education colleges. But the US
evidence is that they are still far enough above the margin that some college
is better than none. e data also reveal that even though at graduation job
markets for liberal arts graduates are not as good as for engineers, by age 55
the typical liberal arts graduate is earning more than the engineer because of
the greater exibility the former has in changing jobs – unless the engineer gets
an MBA. Following the COVID-19 pandemic recession, job markets for college
graduates should still be better than for those with high school or less. e job
markets for high school graduates or less have been even worse throughout, so
the foregone earnings costs for the college graduates are lower making rates of
return for college graduates relatively higher.
What About Public Investment?
Another major conclusion from the logic of the model and the data is that total
investment in higher education is sub-optimal, and that this is largely because
public investment is sub-optimal. at is, privatization of higher education
nance has gone too far for eciency, especially at the two-year degree level.
is matters because public expenditure on higher education is the main
policy tool for inducing greater private total savings and investment, induced
by lowering tuition costs, and/or by grants covering living costs, as with Pell
Grants in the US and grants in England. More families are then encouraged to
enroll their children in colleges and hence invest more as they save by reducing
consumption to invest in room, board, and tuition. In public investment in 2
and 4-year higher education, there is no “crowding out” of private investment by
public investment – quite the opposite, there is “crowding-in”.
What Are the Political Eects of is Underinvestment?
Human capital has been losing the race with technology. Millions of people
with inadequate skills continue to be displaced by automation. ere has
23. The 10.9 percent narrow rate is based on calculations using the US annual BLS household
earnings survey data by McMahon (2018). The 39 percent total return is based on the non-mone-
tary estimates of direct social and private outcomes as 262 percent of the earnings benets all of
which are shown in Table 1. The indirect benets are omitted from this computation because they
arise from interactions over a long period of time, 40 years or so, and because the implicit error
margins are larger. As such, the result is a conservative estimate of the total return.
been shrinkage of manufacturing employment, no more elevator or telephone
operators, and fewer workers in agriculture, mining, typing pools, retailing,
and elsewhere. ese are elds where the number of jobs will never return. As
mentioned, this has kept the real earnings of the 64 percent of the US population
with no more than high school skills and knowledge below where they were in
1980, leaving them, justiably, very discontented as evidenced by how they have
voted in the US (and elsewhere), as well as continuing loyalties as of 2021. By
contrast, and even omitting the top 1 percent, the earnings of college graduates
have risen 49 percent. e unemployment rate for college grads remains low, and
earnings remain higher, in relation to those with high school or less according to
the Bureau of Labor Statistics (2021).
e discontent of mostly this 64 percent of the population has fueled political
demands in the US and elsewhere for replacement of ruling elites by more
authoritarian nationalists. Voters with low earnings, oen in rural areas, are
reacting against a political system that is not serving them well. is political
response has led to new-right leaders like Donald Trump in the US, Boris
Johnson (with Brexit) in the UK, Viktor Orban in Hungary, Andrzej Duda in
Poland, Tayyip Erdogan in Turkey, Jair Bolsonaro in Brazil, and the growth
of extremist parties in France and Germany. In developed and less developed
countries alike, manufacturing has become more skill-intensive and has
employed fewer workers (Wood, 2019). is displacement of many lower skilled
jobs by technology worldwide makes it not fully explainable by forces relatively
unique to the US such as trade decits with China or Mexican immigration cited
by ex-President Trump.
Advanced university graduates generating new ideas are a major driver of this
technical change. However, the number receiving undergraduate 2- and 4-year
college education and the capacities it creates for using these new technologies
has not kept up. Likewise, labor with less skills and knowledge (i.e., high school
graduates, those with some high school, or less) is no longer in high demand.
Ironically, the US community college system is well equipped to extend lifelong
learning, but public support per student has been cut limiting its eectiveness.
Can Optimism Prevail?
Endogenous development oers a conceptual framework identifying a more
optimistic way forward for families, communities, and nations. Investment in
human capital skills and knowledge that last a lifetime yield are at the core of the
more productive use of time at work but also at home yielding not just money
earnings but also non-monetary benets to the sustainable growth of wellbeing
of the family and the society. With one important addition, the introduction of
equity into the objective function (while interpreting the model to apply to all
eory, Evidence, and Policy Implications 429
428
levels of education), the model also becomes a theoretical framework for the
eld of education nance. It links education nancing to education outcomes
and points toward the maximization of social well-being that includes equity, or
Samuelson’s (1955) solution for “Social Bliss”.
e model cannot forecast major exogenous shocks, such as the COVID-19
pandemic shock or the timing of the political reaction to widening inequality.
But it can guide rational political strategies and also short-term policies. One
of the latter that is economically ecient (and likely equitable) is could be
zero-tuition for lifelong learning and for all students at US community colleges
and UK further education colleges. is could be supplemented with Pell
Grants or UK maintenance grants for some. It would help to lower inequality
while fostering higher regional and national growth and development. ese
institutions have the highest rates of return on investment in human capital,
given their lower costs, and a record of contributing to regional development
since their graduates tend to remain local including in areas of the country
where many of the currently dissatised voters reside.
Lower tuition at the bachelor’s level due to better public support would be
rational state-level and federal policies that would raise total factor productivity
and sustained rates of development. is is because the return in the form of
development is very high relative to the opportunity cost of the funds, even if
the money is borrowed. Zero-tuition at the higher cost bachelor’s level at public
institutions, however, goes too far because of the need for resource recovery
from wealthy families (see Murphy et.al. op.cit), the need it creates to severely
ration access because of high public costs it then would take to maintain quality,
and greater inequity due to the selective admissions that result and due to any
support coming from regressive state taxes
Beyond pandemics and rising inequality, climate change is also a possible
external shock to sustained per capita development. However, the preliminary
evidence on external social benets is that college educated individuals are
more willing to support the necessary political policies needed to reverse global
warming such as international treaties and a carbon tax. Other external benets
from research largely by PhD level graduates over their lifetimes are among the
most eective ways to reduce global warming. is research by PhD’s has been
and is absolutely crucial to nding vaccines and other medical means necessary
to end COVID-19, future pandemics, and will be to adjust to climate change and
other potential external shocks.
e optimism of the endogenous development perspective is conditional
however on political and voter support, i.e. on governments being willing and
able to invest adequate amounts in higher education. is latter conclusion is
derived from the theoretical model and the evidence in this paper that develops
the relation of higher education and its externalities to wider economic and
human development, and with these externalities shown to be central to total
factor productivity growth that must be supported largely by governments
and charitable giving. is support is then augmented by additional private
investment by students and their families because government support lowers
private costs which induces (“crowds in”) additional private saving and
investment by households. Since charitable giving overall is inadequate, public
investment is necessary both for achieving optimal well-being of individuals and
the society and for sustaining it through the research needed to mitigate current
and future external shocks to the development process.
References underlying the empirical estimates in Table 1 are in McMahon (2018).
Appiah, Elizabeth, and W. McMahon (2002). e social outcomes of education and feedbacks on
growth in Africa. Journal of Development Studies, 20902, 27-68.
Arias, Omar, and W. McMahon (2001). Dynamic Rates of Return to Education in the US.
Economics of Education Review, 20(2), 121-38.
Barro, Robert J., and X. Sala-I-Martin (1995, 2007). Economic Growth. McGraw Hill, NY. Bureau
of Labor Statistics (2021). For earnings by education level see BLS Current Population Surveys
over many years at www.census.gov., for relative unemployment rates see https://www.bls.gov/
opub/ted/2020/unemployment-rate-2-percent-for-college-grads-3-8-percent-for-high-school-
grads, and for jobs data for 2021 and later see www.bls.gov
Clotfelter, Charles (1999). e familiar but curious economics of higher education”, Journal of
Economic Perspectives, 13, 3-12
Davies, Jim (2003). Empirical evidence on human capital externalities. Economic Policy Research
Institute Working Paper 20035. University of Western Ontario, Canada.
Harari, Yuval Noah (2015). Sapiens: A brief history of humankind, Harper-Collins: NY.
Haveman, Robert and B, Wolfe (1984). Schooling and economic wellbeing: e role of nonmarket
eects. e Journal of Human Resources, 19(3), 377-407.Haveman, Robert and B, Wolfe (2007).
“Valuing the non-market and social benets of higher education”, Wider Benets of Learning,
Institute of Education, Univ. of London, www.learningbenets.net
Hermannsson, Kristinn, K. Lisenkova, P. Lecca, P. McGregor, and J. K. Swales. (2016). e external
benets of higher education. Regional Studies. http://www. tandfonline.com.loi/cres.20
Hu, Yunfang (2008). ‘Human capital accumulation, home production, and equilibrium dynamics’,
e Japanese Economic Review, 59(3), 292-311.
Keller, Katrina R. I. (2006). Investment in primary, secondary, and higher education and the eects
on economic growth. Contemporary Economic Policy, 24(1), 18–34.
Lang, Hans J. and D. Merino (1993). e selection process for capital projects. Wiley: NY.
Lockner, Lance (2011). Non-production benets of education: Crime, health, and good citizenship.
NBER Working Paper 16722, National Bureau of Economic Research: Cambridge, MA.
Lucas, Robert E. (1988). On the mechanics of economic development. Journal of Monetary
Economics, 22(1), 3–42.
Lucas, Robert E. (2009). Ideas and Growth. Economica 76(301), 1–19.
McMahon, Walter W. (1982). Eciency and equity criteria for educational budgeting and nance
in Walter McMahon and Terry Geske, Financing Education: Overcoming Ineciency and
Inequity, University of Illinois Press: Urbana, IL.
McMahon, Walter W. (2002). Education and Development: Measuring the Social Benets. Oxford
University Press: Oxford, UK.
McMahon, Walter W. (2007). An analysis of education externalities with applications to
development in the deep south. Contemporary Economic Policy, 23(3), 459-82.
McMahon, Walter W. and M. Oketch. (2013). Education’s eects on life chances and development:
An overview. British Journal of Education Studies 61(1), 79–107.
McMahon, Walter W. (2017). Higher learning, greater good, the private and social benets of higher
Empirical Evidence from Worldwide Data 431
430
education. Johns Hopkins University Press: Baltimore, MD.
McMahon, Walter W. (2018). e total return to higher education: Is there underinvestment for
growth and development? Quarterly Journal of Economics and Finance, 70, 90-111.
McMahon, Walter W., and M. J. Godinsky (2020). “Technical Appendix to Total Return to Higher
Education” and Excel spreadsheet: “Social Benets of Education”, downloadable at: https://
publish.illinois.edu/wmcmahon/
Mingat, Alain, and J.-P. Tan (1996). e full social returns to education: Estimates based on
countries’ economic growth performance, Human Capital Working Paper HCD73, Washington,
DC: World Bank. Moretti, Enrico (2004). “Human Capital Externalities in Cities”, Handbook
of Regional and Urban Economics, pp. 2243-2291.Murphy, Richard, J. Scott-Clayton, and G.
Wyness (2019). e end of free college in England: Implications for enrolments, equity, and
quality, Economics of Education Review, 71 (Aug, 2019) pp 7-22.
Munich, David and G. Psacharopoulos (2018). Education externalities – What they are and what
we know. European Expert Network on Economics of Education Analytical Reports, European
Commission.
Piketty, omas (2020). Capital and Ideology. Harvard University Press, Cambridge, MA.
Romer, Paul (1994). e origins of endogenous growth. Journal of Economic Perspectives, 8(1),
3-22
Samuelson, Paul (1955). A diagrammatic exposition of a theory of public expenditure. e Review
of Economics and Statistics, 37(4), 350-6.
Wolfe, Barbara, and R. Haveman (2003). Social and non-market benets from education in an
advanced economy”, in Y. Kodrzycki (Ed.) Education in the 21st Century. Federal Reserve Bank:
Boston, MA.
Wood, Adrian (2019). Globalization and structural change around the world, 1985-2015. e
World Bank Research Observer, 34(1), 6594.
e External Social and Economic Benets of Higher Education:
Empirical Evidence from Worldwide Data
Katarina R. I. Keller
is paper analyzes the eects of enrollment rates in higher education on its
external social and economic benets across countries globally since the 1960s.
Global panel data regressions over time are used. e eects of enrollment rates
in higher education have a multitude of benets to society. Highly statistically
signicant eects are found, such as reducing poverty rates, infant mortality
rates, and fertility rates. Enrollment rates in higher education augment political
rights (i.e. democratization), as well as R&D expenditures. Equivalent regressions
for sample splits of developed countries and less developed countries separately
indicate similar results.
e benets of education to the economy and society are abundant. Its importance
to economic growth is theoretically and empirically well established. Education’s
eects on other variables are less researched, and this study adds to this area by
considering the external social benets of higher education to variables that do
not directly accrue back to the individual investing in his or her education. e
external social benets considered in this work are higher education’s eects on
poverty rates, infant mortality rates, the natural log of fertility rates, political
rights, and research and development expenditures (R&D). Some of these
variables are related to improved income per person, such as poverty and family
size. Other variables are also benecial to economic growth and can channel
the eects of education on economic growth, such as political rights and R&D.
Some improvements to health and well-being such as decreased infant mortality
rates are benets to society as well.
e multitude of benets to society are oen beyond the benets to the
individual, and this supports why it is important for governments to nance
the expansion of quality education beyond what an individual is willing to
pay. Public expenditures and the adequate amount spent per student are other
imperative areas that are addressed in Keller (2006a, 2006b) as they relate to
promoting economic growth, as well as inuencing some indirect eects such as
increasing political rights and decreasing the natural log of fertility rates.
Some of the variables are closely related to the United Nations’ (2016)
Katarina Keller is Associate Professor, Department Of Economics, and Executive
Director of International Programs, Sigmund Weis School Of Business, Susquehanna
University, Selinsgrove, PA.
| : - 431
