Embed Size (px)
Citation preview
This article was downloaded by: [Moskow State Univ Bibliote]On: 15 February 2014, At: 09:11Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Educational ResearchPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/rere20
The role of education in economicgrowth: theory, history and currentreturnsTheodore R. Breton aa Universidad EAFIT , Medellin , ColombiaPublished online: 03 Jun 2013.
To cite this article: Theodore R. Breton (2013) The role of education in economicgrowth: theory, history and current returns, Educational Research, 55:2, 121-138, DOI:10.1080/00131881.2013.801241
To link to this article: http://dx.doi.org/10.1080/00131881.2013.801241
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.
This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
The role of education in economic growth: theory, history andcurrent returns
Theodore R. Breton*
Universidad EAFIT, Medellin, Colombia
(Received 13 May 2012; final version received 5 December 2012)
Background: This paper was prepared to address the issue of whether current levelsof public expenditures on education are cost-effective in countries with widely differ-ing average levels of education.Purpose: The paper examines the role of education in economic growth from a theo-retical and historic perspective, addresses why education has been the limiting factordetermining growth, discusses why certain countries have provided education to themasses and others have not, provides estimates of the quantitative importance of thedirect and the indirect effects of education on the economy, calculates the marginalnational return on investment for 60 countries, and examines the implications ofthese results for government policy.Methodology: The paper presents the results from other studies and estimates the mar-ginal product of human capital and of physical capital and the relative importance ofpost-secondary education in 2005 using cross-country estimates of national incomeand the stocks of human capital and physical capital. The estimates of the stocks ofhuman capital were developed from historic rates of public and private investment inschooling, the cost of capital during schooling, and students’ foregone earnings.Results: The paper presents evidence that education has direct and indirect effects onnational output. Educated workers raise national income directly because schoolingraises their marginal productivity. They raise national income indirectly by increasingthe marginal productivity of physical capital and of other workers. In highly edu-cated countries the spillover effect on other workers is minimal, but in less-educatedcountries the spillover effect appears to be much larger. In all countries, the positiveeffect of rising human capital on the productivity of physical capital is required tooffset the diminishing returns to investment in physical capital and make risinginvestment in physical capital financially viable in the growth process. The empiricalresults indicate that investment in schooling is subject to diminishing returns but thatthe marginal return at the national level is still considerable in highly educated coun-tries, over 10% in 2005. In the least educated countries, the marginal return is muchlarger, in excess of 50%, but since most of this effect is indirect, its magnitude isnot generally appreciated. Achievement of these returns requires public investmentin education because the direct return to the educated individual is insufficient toovercome the high cost of private financing. The results also indicate that investmentin post-secondary education does not provide any additional effect on nationalincome beyond the effect of investment in education generally. The implication isthat governments may allocate their limited funds to primary and secondary school-ing of the poor without suffering a loss in GDP growth.
*Email: [email protected]
Educational Research, 2013Vol. 55, No. 2, 121–138, http://dx.doi.org/10.1080/00131881.2013.801241
� 2013 NFER
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
Conclusions: These very high macro-marginal returns to education make it possiblefor poor countries to grow very rapidly if they make a major public commitment toraising the average level of schooling of the masses.
Keywords: education; schooling; world; financial returns; investment; history
I. Introduction
Nations invest a considerable portion of public and private funds in formal schooling.Although resources are always scarce, in the aftermath of the financial crisis of 2008,many governments are seeking to cut public expenditures for programmes that do notprovide a sufficient return. Governments are legitimately asking whether public invest-ment in schooling should be reduced. Economic theory indicates that these expendituresshould be cut back if their returns are low compared with the nation’s cost of capital,or if private investment in schooling is a feasible substitute for public investment.
Most estimates of the economic return on public investment in schooling quantifythe returns that accrue to workers through increases in wages or salaries. In this article,I denote this return as the ‘direct’ return. Such estimates are useful, but they underesti-mate the return on investment to the nation because they do not include the indirecteffects of an individual’s schooling on national income. Neither do such estimates pro-vide any indication of whether public investment is required to obtain these returns, orwhether they could be obtained through reliance on private investment.
In this article, I estimate the national returns on investment in schooling, whichquantify the increase in national income associated with increases in schooling. Sincethis return captures the direct and the indirect effects of schooling, it is the right mea-sure to determine whether the level of national investment is optimal. In this article, Idenote the difference between the return on investment to the nation and the directreturn as the ‘external’ return.
Economists now accept that investment in education, or human capital, is an impor-tant element in the economic growth process. Econometric and other studies of labourmarkets provide very strong and consistent evidence that more educated workers aremore productive and earn higher salaries (Psacharopoulos and Patrinos 2004). Thesestudies indicate that education is not just a signalling device to identify more capableworkers.
Estimating the effect of education on national income requires a model of nationaleconomic output (GDP) that can distinguish between the effects of schooling and otherfactors. Economists have had a difficult time quantifying the effect of these differentfactors because their effects often do not occur immediately and the indirect effects ofschooling may occur through many different mechanisms. As a result, there continue tobe doubts about whether the very large effects of education on GDP found in somestudies are an indication that education has large indirect effects on the economy orwhether these results are only an indication that important factors were omitted in thesestudies.
If important factors are not included in an analysis of the effect of education, thenthe effects of these factors may be mistakenly attributed to education and erroneouslyraise the estimate of education’s total effect. Since more-educated countries often havesuperior institutions, some economists believe that some of the effect on nationalincome attributed to education should be attributed to superior institutions. Glaeser, LaPorta, Lopes-de-Silanes, and Shleifer (2004) review this literature and argue that educa-tion rather than institutions is the more fundamental cause of higher national income.
122 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
A nation’s average level of schooling and its level of income generally rise together.But these simultaneous increases may occur because education drives economic growth,or because people demand more education as they acquire more income, or for bothreasons. In the United Nations’ (2009) latest version of the System of National Accounts(SNA), the treatment of education is highlighted as an accounting issue because expertshave not agreed on whether education is primarily a form of capital or primarily a com-ponent of consumption.
So the dilemma for public policy is clear. If education is primarily consumption,then public funds for education should be cut in difficult times. But if it is primarilyinvestment, then any cuts could have serious future repercussions. And if education isTHE primary determinant of economic growth, then in poor countries particularly,expenditures on education should be increased even in difficult times.
In this paper, I elaborate on Schultz’s (1961) theories that education plays a criticalrole in the economic growth process and that it is likely to be the limiting factor in thisprocess. I present evidence that supports these theories, and I offer an explanation forwhy historically certain nations provided education to the masses much sooner than oth-ers. Subsequently I present the empirical results from a model of the effect of humancapital on GDP that is consistent with Schultz’s theories, and I show the quantitativeimportance of these effects in rich and poor countries. I then use the estimated parame-ters from this model to estimate the marginal product of schooling in 60 countries in2005, and I investigate whether investment in post-secondary education has a largereffect on national income than investment in lower levels of schooling. Finally, I dis-cuss the policy implications of these results.
II. Evidence for Schultz’s Theory of Economic Growth
After the Second World War, the International Bank for Reconstruction and Develop-ment (IBRD) was created to finance physical capital projects in countries damaged bythe War and in poor countries. At the time economists believed that countries were poorbecause they lacked physical capital. The IBRD provided financing for capital projectsin these countries, but many of these projects were unsuccessful.
Schultz (1961) observed that the rich countries devastated in the Second World Warquickly employed massive amounts of new physical capital, while the poorest countriesseemed unable successfully to utilize even small amounts. He theorized that a nation’scapability to productively use physical capital is a function of its level of human capitaland that if its human capital does not increase along with its physical capital, then humancapital becomes the factor limiting its economic growth. Schultz further observed thathuman capital is more likely to be the type of capital limiting growth because foreigninvestors are eager to invest in physical capital, but not in human capital.
Figure 1 shows the stocks of human capital and physical capital in 2005 for 60countries. These countries are representative of the range of rich and poor countries.1 Ihave estimated these stocks using data from Penn World Table 6.3 (Heston, Summers,and Aten 2009), so the values y are adjusted to equalize purchasing power acrosscountries.2 In the case of human capital, the investment includes public and privateexpenditures on formal schooling, the implicit financing cost during students’ schooling,and students’ foregone earnings while they are in school. As shown in the figure, thedifferences between rich and poor countries are enormous. The levels of total capital/adult (human capital+physical capital) differ across rich and poor countries by a factorof over 100.
Educational Research 123
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
The observed relationship between the two capital stocks is consistent withSchultz’s theory that human capital and physical capital are complementary. There issome variation in the relative amounts of the two types of capital, but no countries havehigh levels of only one type. For example, the USA has more human than physical cap-ital, while Japan has more physical than human capital, but both countries have highlevels of both. The correlation coefficient between the two kinds of capital in this dataset is 0.87.
These data show that economic growth does not occur automatically. If it did, therewould not be such large differences in the magnitude of the capital stocks betweencountries. Clearly some characteristic(s) of the more developed countries, not present inthe less developed countries, facilitated historic investment in both types of capital.Whatever these characteristics are, they clearly vary widely across countries. If humancapital and physical capital are complementary in their effect on investment returns,then historically either type of capital or both could have been the factor limiting invest-ment in the other type of capital.
All of the countries in Figure 1 historically have had a market economy, andnational statistics show that investment has been flowing between them for some time(Obstfeld and Taylor 2004). So it is not a priori evident that a shortage of privatefinancial capital has limited investment in the poor countries. Caselli and Feyrer (2007)show that the marginal product of reproducible physical capital in 1996 was very simi-lar in 43 developed and undeveloped countries. Implicitly local and global investorsprovided financing for those capital projects that had attractive returns, so any recentfailure to invest in poor countries apparently has not been due to a shortage of financialcapital.
But as Schultz (1961) observed years ago, there is no evidence that local and globalinvestors invested in human capital in these countries. So even though human capitaland physical capital appear to be similar in their effect on economic output, they appar-ently are not similar from a private investment standpoint. Why not?
Phys
ical
Cap
ital/A
dult
(200
5 U
S$)
Human Capital/Adult (2005 US$)0 25000 50000 75000 100000 125000 150000 175000
0
25000
50000
75000
100000
125000
150000
175000
200000
Argentin
AustraliAustria
Bolivia
Brazil
Canada
Chile
ColombiaCongo, R
Costa Ri
Cote d`I
Denmark
DominicaEcuador
EgyptEl Salva
Ethiopia
FinlandFrance
Ghana
Greece
Guatemal
India
Iran
IrelandItaly
Jamaica
Japan
Jordan
Korea, R
Malawi
Malaysia
Mali
Mexico
Morocco
Netherla
New Zeal
Niger
Norway
Pakistan
Panama
ParaguayPeru
Philippi
Portugal
Senegal
Singapor
Spain
Sri Lank
Sweden
Switzerl
Syria
Thailand
Togo
TunisiaTurkey
UK
Uruguay
USA
Zambia
Figure 1. Stocks of human capital and physical capital in 2005.
124 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
In a poor country, human capital is created by investing in the education of a child,which is very different from investing in a factory. The factory is a transferable fixedasset, and a child is not. Centuries ago, private investors could and occasionally didinvest in children’s education, with a contractual guarantee of repayment through theindentured servitude of the child (Clark 1977). Today such arrangements typicallyare illegal, and without them the private financing of a poor child’s education is notfeasible.
Even if private financing were feasible, in a poor country the parents’ incentives tofinance their child’s schooling are weak or even negative for several reasons. First, ifthe child is working, enrolment of the child in school immediately reduces the family’sincome. Second, the period over which the parents would have to borrow continually isquite long, and the period they would have to carry the loan before it could be paid offwould be considerably longer. Such long loan periods substantially increase the financ-ing risk and cost. Third, if the investment were to pay off in higher income for thechild, the parents might not benefit, since they would have no legal right to this incomeafter the child reached maturity.
Precisely because the parents have no right to the future income from the invest-ment in the child’s schooling, they cannot collateralize the investment, so they wouldhave to pay a very high rate of interest for a high-risk loan. For this reason, Mincer(1984) argues that historically only the children of the rich have been educated inresponse to market forces.
III. History of mass schooling and economic growth
So how have some countries managed to create a highly educated population? Easterlin(1981) observes that historically the schooling of the masses has occurred only whenideological or political forces made it a priority.
The Jews appear to have been the first people to commit to mass schooling. Afterthe destruction of their Temple in Jerusalem in 70 CE, the religious leaders requiredevery Jewish family to educate their male offspring to enable them to study the Torah.Botticini and Eckstein (2007) argue that this religious obligation created the firsteducated community, but the members of this community had to disperse to put theireducation to economic use. The Jews became a wealthy people, but no single countrydeveloped as a result.
The first national commitment to the schooling of the masses apparently occurredafter the Protestant Reformation in the 16th century. The leaders of several Protestantsects in northern Europe promoted literacy to enable their members to read the Bibleand study religious catechism. This religious obligation was the basis for the first signif-icant efforts to create schools for the poor (Bowen 1981).
Numerous reports document the increase in literacy that occurred after the ProtestantReformation of Christianity (Cipolla 1969). Competition between Protestant andCatholic religious groups to attract believers further spurred the provision of free orsubsidized schooling for the poor in regions where both groups were active (Houston1988). For the next three centuries literacy increased steadily in Europe, largely throughthe use of religious catechisms. By 1700, 35–40% of the population in ProtestantEurope could read, while in Southern Catholic Europe less than 20% were literate(Johansson 1977).
During the course of the 19th century, Europeans and their descendents in formercolonies adopted nation-building as their dominant political ideology, and as part of this
Educational Research 125
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
process, the state increasingly imposed obligatory public schooling on the masses(Ramirez and Boli 1987). In the struggle over control of the educational system, PopePius IX issued an encyclical in 1864 in which he forbade Catholics from accepting civileducation (Johnson 1976).
The Catholic Church’s opposition to secular schooling slowed the provision of massschooling in southern Catholic Europe and in the Iberian colonies. As national levels ofeducation increased from 1850 to 1940, northern Protestant Europe maintained its his-toric advantage relative to southern Catholic Europe, with particularly large differencesrelative to Spain and Portugal (Benavot and Riddle 1988). In 1940, primary schoolenrolment ratios were about 70% in northern Europe and its settlements, 60% in Italy,and 35% in Iberia and its settlements.
No comparable commitment to mass schooling occurred outside of Europe andsome of its settlements until much later (Craig 1981). Japan is the principal exception,as it consciously adopted European schooling practices as part of its national develop-ment strategy. By 1870, Japan had achieved levels of primary enrolment that were com-parable with those in southern Europe (Benavot and Riddle 1988). Average schoolinglevels in Africa, Asia and the Middle East at this time were extremely low (Morrissonand Murtin 2009). Subsequently, in the early and mid-20th century, the European modelof a national society, including state funding for mass schooling, spread throughout theworld (Ramirez and Boli 1987).
Christian missionaries, and particularly Protestant missionaries, were instrumental increating schools for the poor in many parts of the world, often in conflict with Euro-pean governments and settlers, who feared the empowering effect that schooling mighthave on slaves and the native peoples. Education rates are higher today in areas whereProtestant missionaries were more active historically and where more people convertedto Protestant Christianity (Woodberry 2004). Each nation’s historic cultural and politicalpropensities to accept missionary schooling or to provide its own charitable or statefunds for mass schooling determined their level of human capital today.
Anecdotal data suggest that rising incomes have been linked to literacy and school-ing since the 16th century, but comprehensive data on national levels of education areonly available for a subset of the more educated countries since the mid-19th century.Figure 2 shows the relationship between the average schooling attainment of the popu-lation age 15 to 64 and national income for 43 countries in 1910 and in 2000.3 In
Figure 2. Average schooling attainment and GDP/capita across countries in 1910 and 2000.
126 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
1910, the highest average level of schooling in any of these countries was less thaneight years. In 2000, the highest average level of schooling had increased to 13 years.The exponential trend lines in 1910 and 2000 show that over this period therelationship between average schooling and GDP/capita did not change for countrieswith average schooling levels below five years, but at higher levels of schooling, theassociated level of GDP/capita has increased.
The trends in the figure suggest that there are increasing returns to schooling, butthis is not the case. Higher levels of schooling have considerably higher unit costs thanlower levels of schooling (Breton 2010). In addition, as nations increase their averagelevel of schooling, they spend more on schooling at all levels.4 The increasing level ofGDP/capita associated with rising levels of schooling occurs because national expendi-tures per average year of schooling rise even more. If the relationship between GDPand education is examined in financial terms, then the diminishing returns to incremen-tal investment become evident.
Figure 3 shows the relationship between human capital/adult (calculated frominvestment in schooling) and GDP/adult in 2005, using the same human capital datashown in Figure 1.5 The strong relationship between these variables and the diminishingreturns as human capital increases are both evident in the data.
IV. Methodology for estimating the returns to education
The standard methodologies used to estimate the returns to education include only thedirect effect on workers’ salaries. If Schultz’s theory that human capital and physicalcapital are complementary is correct, then education has both direct and indirect effectson national income, and an estimate of the returns to investment in education shouldtake both into account.
Mankiw, Romer, and Weil (1992) created a model of national income (Y) that isconsistent with Schultz’s theory of capital-skill complementarity. This model includesthree factors of production, human capital (H), physical capital (K) and labour (L):
GD
P/Ad
ult (
2005
US$
)
Human Capital/Adult (2005 US$)
0 25000 50000 75000 100000 125000 150000 175000
0
15000
30000
45000
60000
Argentin
AustraliAustria
Bolivia
Brazil
Canada
Chile
ColombiaCongo, R
Costa Ri
Cote d`I
Denmark
DominicaEcuadorEgyptEl Salva
Ethiopia
FinlandFrance
Ghana
Greece
Guatemal
India
Iran
Ireland
Italy
Jamaica
Japan
Jordan
Korea, R
Malawi
Malaysia
Mali
Mexico
Morocco
Netherla
New Zeal
Niger
Norway
Pakistan
PanamaParaguayPeruPhilippi
Portugal
Senegal
Singapor
Spain
Sri Lank
Sweden
Switzerl
Syria
Thailand
Togo
TunisiaTurkey
UK
Uruguay
USA
Zambia
Figure 3. Human capital/adult and GDP/adult in 2005.
Educational Research 127
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
Y ¼ KaHbL1�a�b ð1Þ
In this model, workers receive income as labourers and as a return on their humancapital. The model is a Cobb–Douglas production function, in which each factor ofproduction intrinsically raises production directly and indirectly by increasing theproductivity of the other two factors (Breton 2013). Human capital has a direct effecton the salary received by the educated worker and has indirect effects on the productiv-ity of physical capital and on the productivity of other workers (labour). The Cobb–Douglas production function with β<1 exhibits the diminishing returns to investment inhuman capital shown in Figure 3.
Figure 4 shows the dynamics of this model in response to an increase in schooling.The increase in schooling increases the nation’s human capital. Human capital then hasa direct effect on national income (the solid line) and indirect effects on the productivityof the other two factors (the dotted lines). The figure also shows a third indirect effect(labelled ‘4’), which is the positive feedback that rising income has on the society’sdemand for education.
This model simulates the effect on physical capital that follows a national decisionto increase the level of schooling. After a lag, the increase in schooling increases thestock of human capital of the work force, which raises the marginal productivity ofphysical capital and the expected return on investment. Private investors then increasetheir investment in physical capital, which has a direct effect on national income and anindirect effect on the productivity of human capital and labour. As these various effectswork their way through the economy, economic output rises and national incomeincreases along with it.
Conveniently, the marginal product of human capital (MPH) in this model (i.e. theincrease in national income associated with an increase in human capital) provides anestimate of the full marginal return on investment in schooling to the nation, includingthe direct and the two indirect effects:
MPH ¼ @Y=@H ¼ bY=H ð2Þ
Labour
Schooling
PhysicalCapital
Market Economy
NationalOutput /NationalIncome
1
2
HumanCapital
4
Figure 4. Effect of schooling on national income in a market economy.
128 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
Given an estimate of β for the model in Equation (1) and estimates of GDP/adult andhuman capital/adult, the marginal return on investment in schooling can be estimatedfor each country.
V. Estimates of national returns to investment in education
Breton (2013) has estimated the model in Equation (1) using data for 61 countries in1990. He limits the estimation to market economies because there is no reason to sup-pose that countries with non-market economies (such as China) would have similarmodel parameters. He shows that the model provides estimates of the total effect ofschooling on national income that are consistent with available studies of the directeffect of schooling on workers’ salaries in 36 of these countries.6
As mentioned earlier, I denote the difference between the national return and thedirect return the ‘external’ return. The trends in the direct and the external returns toinvestment are shown in Figure 5.7 They indicate that in 1990 the marginal product ofschooling in workers’ earnings studies varied from 8% in the most educated countriesto 13% in the least educated countries. The external marginal product varied from 6%in the most educated countries to 40% in the least educated countries. These resultsindicate that the indirect effects of education are less than the direct effects in highlyeducated countries, but they are much larger than the direct effects in countries withlow average levels of education. In 1990, the combined direct and external return in thecountries with the lowest educational levels was over 50%.
The estimates of the total external returns to schooling in Figure 5 can be allocatedto physical capital and other workers (labour) using the data provided in Breton’s(2013) study. These two external returns are shown in Figure 6. Their implicationis that the spillover effect of a more educated individual on the productivity of others isvery small in a highly educated country, but it is considerably larger in countries wherethe population is less educated. These results are consistent with the few studies thatestimate the indirect effect of more educated workers on the wages of other workers inspecific countries (Breton 2013).
0
10
20
30
40
50
60
0 20000 40000 60000 80000 100000
Mar
gina
l Pro
duct
(%)
Human Capital/Adult (2000$)
External Estimates External Trend Market Trend
Figure 5. Direct (market) and external returns to education in 1990.
Educational Research 129
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
The marginal effect of human capital on the productivity of physical capital issubstantial in countries with little human capital, but it diminishes noticeably as thelevel of human capital rises. The diminishing return evident in all of these marginalproducts indicates that it will be increasingly difficult for highly educated countries toincrease national income by investing more resources in schooling.
Most countries have substantially increased their level of human capital since 1990,and in this model, these increases reduce the marginal return to investment in education.Figure 7 shows the national marginal products of human capital in 2005, which areanalogous to the sum of the direct and the external returns shown in Figure 5. In thisset of estimates, Denmark has the lowest marginal product, which is 10.8%. This figurecompares to Denmark’s estimated marginal product of 13.5% in 1990 (Breton 2013).This decline is due to the diminishing returns associated with the large increase inhuman capital in Denmark from $85,000 per adult (2005$) in 1990 to $145,000 peradult (2005 US$) in 2005.
Mar
gina
l Pro
duct
of H
uman
Cap
ital (
%)
Human Capital/Adult (2005 US$)0 25000 50000 75000 100000 125000 150000 175000
0
10
20
30
40
50
60
70
Argentin
AustraliAustriaBolivia
Brazil
Canada
Chile
ColombiaCongo, RCosta Ri
Cote d`IDenmark
Dominica
EcuadorEgyptEl Salva
Ethiopia
FinlandFrance
GhanaGreece
Guatemal
India
Iran Ireland
ItalyJamaica
JapanJordan
Korea, RMalawi Malaysia
Mali
MexicoMorocco
NetherlaNew Zeal
Niger
Norway
Pakistan
Panama
ParaguayPeru
Philippi
Portugal
Senegal
Singapor
Spain
Sri Lank
SwedenSwitzerl
Syria
Thailand
Togo
Tunisia
Turkey
UK
Uruguay
USA
Zambia
Figure 7. Marginal product of human capital (macro) in 2005.
Figure 6. External returns to education in 1990.
130 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
VI. Relative national returns at different levels of schooling
These estimates measure marginal returns to investment in education in the aggregate,but not for different levels of schooling. In highly educated countries primary and sec-ondary schooling is virtually universal, so incremental investment is more likely tooccur at the post-secondary level. But it also could be increased at lower levels ofschooling in an attempt to raise the quality of schooling.
In countries with low average levels of education, schooling is not universal at theprimary level and is infrequent at the secondary level. Governments must make choiceswith respect to funding priorities. Should incremental public funds go to expanding uni-versity opportunities, or to improving either coverage or quality at lower levels ofschooling? This is a complex issue, which has serious implications for income inequal-ity, social mobility and political stability.
Numerous studies have estimated the direct returns to investment in education atdifferent levels of schooling. Older studies have found that the direct returns to educa-tion are higher at lower levels of schooling (Psacharopoulos and Patrinos 2004). Recentstudies show that the salary premium for post-secondary education in most OECDcountries has risen in real terms since 1997 (Psacharopoulos 2009). These data mayindicate that direct returns have risen for post-secondary education compared with lowerlevels of schooling.
Estimates of returns based on workers’ salaries ignore the external returns fromschooling, which, according to Figure 5, are very large in poor countries. Implicitlythese returns are associated with increases in schooling at lower levels since this iswhere most schooling occurred in these countries. The total returns to the nation maybe larger for investment at lower levels of schooling even if the direct returns are largerfor post-secondary schooling.
I examine this issue by estimating the effect of the share of the population withsome post-secondary schooling on national income in 2005 in 50 countries, controllingfor the level of human capital/adult. I exclude the sub-Saharan African countries in thisanalysis, since they have very little post-secondary schooling. Although the post-secondary share and the level of human capital/adult are positively correlated in these50 countries (ρ=0.70), the post-secondary share varies substantially in countries withsimilar levels of human capital/adult.
The share of the population with post-secondary education is from Cohen andSoto’s (2007) data set for the population over 25. I use the estimate in 2010 to repre-sent the share of the working population (over 20) with post-secondary education in2005. Figure 8 shows these data.
I estimate two permutations of the Mankiw, Romer, and Weil model in log form:
logðY=LÞ ¼ cþ a logðK=LÞ þ b logðH=LÞ ð3Þ
logðY=LÞ ¼ cþ ða=1� a� bÞ logðK=Y Þ þ ðb=1� a� bÞ logðH=Y Þ ð4Þ
I estimate the model in 2005 with and without physical capital because in a marketeconomy the complementary nature of the two forms of capital implicitly makes physi-cal capital an endogenous variable that is affected by the level of human capital.
The bidirectional causality between schooling and income could bias ordinary leastsquare (OLS) estimates of the effect of these variables on national income, so an instru-mental variable is required to estimate the effect of these schooling variables on
Educational Research 131
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
national income. As described earlier, there is considerable evidence that Protestantaffiliation has an exogenous effect on a nation’s willingness to support schooling forthe masses. Since virtually every country has some share of Protestant affiliation, thelevel of national affiliation is a good instrument for schooling in cross-sectional analy-ses. I use the Protestant share of the population in 1980 from Barrett (1982) as aninstrument for the education variables in 2005.
Figure 9 shows the cross-country relationship between human capital and Protestantaffiliation. Even in countries with other predominant religions, the establishment ofProtestant mission schools appears to have increased these countries’ propensity toprovide non-Protestant schools for the poor (Woodberry 2004).
Table 1 presents the results for the estimates of the various income models. Column1 shows the OLS statistical results for the model in Equation (3) with human capital/
Shar
e of
Pop
ulat
ion
with
Pos
t-Sec
onda
ry E
duca
tion
(%)
Human Capital/Adult (2005 US$)0 25000 50000 75000 100000 125000 150000 175000
0
10
20
30
40
50
Argentin
Australi
Austria
Bolivia
Brazil
Canada
ChileColombia
Costa Ri Denmark
DominicaEcuador
EgyptEl Salva
Finland
FranceGreece
GuatemalIndia Iran
Ireland
Italy
Jamaica
JapanJordanKorea, R
Malaysia
Mexico
Morocco
Netherla
New Zeal
Norway
Panama
Paraguay
PeruPhilippi
Portugal
Singapor
Spain
Sweden
Switzerl
Syria
Thailand
TunisiaTurkey
UK
Uruguay
USA
Figure 8. Population share with post-secondary education vs. human capital/adult in 2005.
Log(
Hum
an C
apita
l/GD
P)
Log(Protestant Share of Population)-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0
-.6
-.3
0
.3
.6
.9
1.2
1.5
Argentin
AustraliAustria
BoliviaBrazil
Canada
Chile
Colombia
Costa Ri
Denmark
Dominica
EcuadorEgypt
El Salva
Finland
France
Greece
Guatemal
India
Iran Ireland
Italy
Jamaica
JapanJordan
Korea, RMalaysia
MexicoMorocco
NetherlaNew Zeal
Norway
Pakistan
Panama
ParaguayPeruPhilippi
Portugal
Singapor
Spain
Sri Lank
Sweden
Switzerl
Syria
Thailand
Tunisia
Turkey
UK
Uruguay
USA
Figure 9. Human capital/GDP vs. protestant share of the population.
132 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
Table1.
Effectof
human
capitalandpo
st-secon
dary
scho
olingon
natio
nalincomedepend
entvariable
islog(natio
nalincome/adult).
12
34
56
78
OLS
OLS
OLS
OLS
2SLS
2SLS
2SLS
2SLS
Log
(H/adu
lt)0.67
⁄(0.03)
0.71
⁄(0.04)
Log
(H/Y)
1.07
⁄(0.17)
0.70
⁄(0.23)
1.50
⁄(0.28)
1.33
(0.63)
0.82
⁄(0.29)
0.73
(0.44)
Post-secondaryshare
�0.11(0.45)
2.73
⁄(0.84)
1.05
(3.56)
0.47
(2.21)
Log
(K/Y)
1.06
⁄(0.21)
1.13
⁄(0.20)
R2
0.90
0.90
0.37
0.42
0.31
0.31
0.65
0.66
Note:
OLS,ordinary
leastsquare;2S
LS,two-stageleastsquares.Robuststandard
errors
inparentheses.
⁄ Significant
at1%
level.
Educational Research 133
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
adult alone, which explains 90% of the variation in log (GDP/adult). Column 2 showsthat when the post-secondary share is added, it has no effect on national income. Thistest could be unfair, however, because the human capital variable is estimated usingeducation’s historic shares of national income, which is highly correlated with currentincome.
Columns 3 and 4 show the same results for the model in Equation (4) that uses log(H/Y) as the explanatory variable. In this model (column 4), the post-secondary sharehas a highly statistically significant, positive effect on national income, but the esti-mated coefficients in these models may be biased because while education affectsnational income, national income also affects a nation’s level of education.
Columns 5 and 6 show the results for the same model using the Protestant share ofthe population in 1980 and the log of this share as instruments for the education vari-ables.8 With these instruments the effect of human capital/adult is larger and the effectof the post-secondary share is smaller and not statistically significant. Columns 7 and 8show the two-stage least squares (2SLS) results when physical capital is included in themodel. Again the estimated coefficient on the post-secondary share is small and not sta-tistically significant.
The implication of the different outcomes using OLS and 2SLS is that a nation’scumulative investment in education raises national income and this increase in incomeraises the demand for post-secondary education. The small, statistically insignificantcoefficient on the post-secondary share indicates that post-secondary investment has lit-tle or no incremental effect on national income beyond the effect from investment ineducation generally. Implicitly there is no loss in national income if the governmentchooses to extend primary and secondary schooling to more of the population insteadof increasing post-secondary schooling for those that already have secondary schooling.
VII. Implications for public policy
This empirical evidence indicates that at the national level the marginal return to invest-ment in education is very large in countries with less-educated populations but is muchlower in countries with highly educated populations. What are the implications of thisevidence for government policy related to the public funding of education?
Given the inapplicability of private financing at the primary and secondary levels ofschooling, state support at these levels must be very substantial to achieve an optimalnational level of education. All countries with high levels of human capital provide free,obligatory public schooling or public funds for private schooling through the secondarylevel. The historic evidence indicates that if poor countries wish to grow, they must pro-vide this financial support for primary and secondary schooling.
At the post-secondary level, the optimal level of state support is less obviousbecause this education can be privately financed to some degree. In countries with lowaverage levels of education, the external returns from incremental post-secondaryschooling are so great that government support for this schooling is justified, unless thegraduates from these schools frequently leave the country. However, in the event thatpublic funds are limited, the empirical results indicate that these funds are best spentproviding primary and secondary education for all. Once students have achieved thislevel of schooling, those that wish to continue their education may be able privately tofinance these studies. In contrast, if government support is not forthcoming at lower lev-els of schooling, it is unlikely that poor students will complete this schooling.
134 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
Some highly educated countries currently provide very substantial subsidies forpost-secondary education. Given the diminishing returns from investment in human cap-ital, are these subsidies too extensive? Should governments reduce these subsidies andrequire students to finance a larger share of their education themselves?
This question has two aspects. The first is whether the relative subsidies (or taxes)for human capital and physical capital are optimal. Both types of capital are requiredfor income growth, and the rate of growth will be suboptimal if government policiesresult in very different marginal returns for the two types of capital. The second ques-tion is: how low does the marginal return have to fall before government support is nolonger justified?
The Mankiw, Romer, and Weil model provides information useful for answering thefirst question, because it provides estimates of the marginal product of both human cap-ital and physical capital. If the marginal product of human capital is less than the mar-ginal product of physical capital, then a nation may be overinvesting in human capitalrelative to physical capital, and vice-versa.
Figure 10 shows the ratio of the marginal products MPH/MPK for 50 countries in2005. This analysis shows that the marginal return to education is equal or greater thanthe return to physical capital for most countries. Some countries with low levels ofhuman capital apparently have a higher return on physical capital, but this indicateseither that their macro-economic data are not accurate or that they are under-investingin both kinds of capital.
Sweden is the only country with a high level of human capital (above $125,000/adult) that appears to have a return to physical capital that exceeds the return to humancapital. If the data are accurate, the implication is that tax and subsidy policies in Swe-den are not optimal in their encouragement of investment in the two kinds of capital.As the marginal return to education continue to decline in the countries with the mosthuman capital, these countries should re-examine their policies for promoting invest-ment in these two kinds of capital.
Assuming the aggregate marginal returns are similar for both kinds of capital, isthere a marginal return below which incremental investment in either kind is counter-productive? Implicitly, yes, but that point does not appear to have been reached in anycountry. Denmark had the lowest estimated return on investment in human capital in
MPH
/ M
PK (r
atio
)
Human Capital/Adult (2005 US$)0 25000 50000 75000 100000 125000 150000 175000
0
.5
1
1.5
2
2.5
3
3.5
Argentin AustraliAustria
Bolivia
Brazil CanadaChileColombia
Costa Ri
Denmark
Dominica
Ecuador
Egypt
El Salva
Finland
France
GreeceGuatemal
India
Iran
IrelandItaly
Jamaica
Japan
Jordan
Korea, R
MalaysiaMexico
Morocco
NetherlaNew Zeal
Norway
Pakistan
PanamaParaguayPeru
Philippi
Portugal
Singapor
Spain
Sri Lank
Sweden
Switzerl
Syria
Thailand
Tunisia
Turkey
UK
Uruguay
USA
Figure 10. Relative returns to human capital and physical capital in 2005.
Educational Research 135
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
2005, and this return is 10.8%. This return still substantially exceeds the cost offinancial capital. But this is an aggregate return on all investment in education, andreturns on particular types of education are undoubtedly more and less than this rate.As the aggregate return continues to fall, the returns to more types of education arelikely to fall below the cost of financial capital. Optimal policy requires that subsidiesfor education be increasingly targeted on those types of education that offer the higherreturns.
VIII. Conclusions
This paper presents evidence that both human capital and physical capital are requiredfor economic growth and that each has a positive effect on the productivity of the other.The evidence also indicates that human capital is more likely to be the limiting factorin a nation’s economic growth.
Human capital is created initially by providing children with primary and then sec-ondary schooling. Private financing of this type of investment is not feasible for poorchildren. Countries that are highly developed today have a long history of providingfree or highly subsidized elementary education to the poor. A review of their historysuggests that the initial impetus for this schooling had a religious basis, but that as thepublic’s level of education and income rose, their demand for schooling rose, and thefinancial support from private donors was replaced or greatly augmented with publicfunds. If poor countries wish to achieve high levels of national income, they need toprovide public funding for the universal education of the poor, at least at the primaryand secondary levels of schooling.
The evidence on returns to education indicates that investment in schooling is sub-ject to diminishing returns, but at the national level the marginal return on all educationis still considerable in highly educated countries, over 10% in 2005. In less-educatedcountries, the marginal return at the national level is much larger, in excess of 50%, butsince most of this return is indirect, its magnitude is not generally appreciated. Thesevery high marginal returns make it possible for poor countries to grow very rapidly ifthey make a major commitment to raising their average level of schooling.
The evidence also indicates that educated workers raise the marginal productivity ofphysical capital and of other workers. In all countries, the positive effect of rising humancapital on the productivity of physical capital is required to offset the diminishing returnsto investment and make rising investment in physical capital financially viable in thegrowth process. In highly educated countries, the spillover effect on other workers isvery small, but in less-educated countries this effect appears to be much larger.
Notes1. China is not included in the data set, but if it were, it would be in the group of countries with
relatively little human capital.2. I estimated these stocks using the standard OECD (2001) methodology, which sums each
nation’s cumulative investment in each type of capital and then depreciates this investmentover its expected useful life. For physical capital, the investment period is 1960 to 2004,while for human capital the period is 1960 to 2000. Physical capital has an assumed geomet-ric depreciation rate of 6%. Human capital has an assumed linear depreciation rate of 2.5%.Physical capital investment rates and income data were obtained from the series ci andrgdpch in PWT 6.3. Human capital investment rates in public schooling were obtained fromUNESCO data. The detailed methodology for the calculation of the human capital stock ispresented in Breton (2013).
136 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
3. The schooling data are from Morrisson and Murtin (2009). The GDP/capita data are fromMaddisson (2003), but the units were converted from 1996 US$ to 2005$ using data fromPWT 6.3.
4. The increase in GDP/capita with increases in schooling at higher levels of income may bedue in part to the effect of income on the demand for schooling.
5. The data for GDP/adult are from PWT 6.3, but they include my estimate of students’ fore-gone earnings. Since students did not receive these earnings, they are not included in GDPstatistics, but these earnings are the value of the time that students have invested in their edu-cation, so in theory they should be included in GDP.
6. Breton (2013) investigates whether estimates of the effect of human capital on income in theMankiw, Romer, and Weil model are sensitive to the inclusion of other variables in themodel. He shows that these estimates are robust, which indicates that the levels of physicalcapital and human capital are the proximate causes of economic output and that othernational characteristics affect income through their influence on the levels of capital.
7. The curves in Figures 5 and 6 are fitted with a power function, which is consistent conceptu-ally with the form of the Cobb–Douglas function in Equation (1).
8. The first stage regressions are:
Log(H/Y) = 0.66 + 0.05 log(ProtSh) + 0.65 ProtSh R2 = 0.40(0.16) (0.03) (0.25)
PostSecSh = 0.28 + 0.023 log(ProtSh) � 0.01 ProtSh R2 = 0.27(0.04) (0.007) (0.05)
ReferencesBarrett, David B. 1982. The World Christian Encyclopedia. New York: Oxford University Press.Benavot, Aaron, and Phyllis Riddle. 1988. “The Expansion of Primary Education, 1870–1940:
Trends and Issues.” Sociology of Education 61 (3): 191–210.Botticini, Maristella, and Zvi Eckstein. 2007. “From Farmers to Merchants, Conversions and
Diaspora: Human Capital and Jewish History.” Journal of the European Economic Associa-tion 5 (5): 885–926.
Bowen, James. 1981. A History of Western Education; Volume III: The Modern West. London:Methuen & Co. Ltd.
Breton, Theodore R. 2010. “Schooling and National Income: How Large Are the Externalities?”Education Economics 18 (1): 67–92.
Breton, Theodore R. 2013. “Were Mankiw, Romer, and Weil Right? A Reconcilation of the Microand Macro Effects of Schooling on Income”. Macroeconomic Dynamics 17 (5): doi.org/10.2139/ssrn.1455161
Caselli, Francesco, and James Feyrer. 2007. “The Marginal Product of Capital.” Quarterly Jour-nal of Economics 122 (2): 535–568.
Cipolla, Carlo M. 1969. Literacy and Development in the West. Baltimore, MD: Pelican Books.Clark, Dennis. 1977. “Babes in Bondage: Indentured Irish Children in Philadelphia in the Nine-
teenth Century.” The Pennsylvania Magazine of History and Biography 101 (4): 475–486.Cohen, Daniel, and Marcelo Soto. 2007. “Growth and Human Capital: Good Data, Good
Results.” Journal of Economic Growth 12 (1): 51–76.Craig, John E. 1981. “The Expansion of Education.” Review of Research in Education 9:
151–213.Easterlin, Richard A. 1981. “Why isn’t the Whole World Developed?” The Journal of Economic
History 41 (1): 1–19.Glaeser, Edward L., Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2004. “Do
Institutions Cause Growth?” Journal of Economic Growth 9: 271–303.Heston, Alan, Robert Summers, and Bettina Aten. 2009, Penn World Table Version 6.3.
Philadelphia, PA: Center for International Comparisons of Production, Income and Prices atthe University of Pennsylvania (CICUP).
Houston, R. A. 1988. Literacy in Early Modern Europe; Culture and Education 1500–1800. NewYork: Longman.
Educational Research 137
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
Johansson, Egil. 1977. “The History of Literacy in Sweden in Comparison with Some OtherCountries.” In 1981, Literacy and Social Development in the West: A Reader, edited byHarvey J. Graff, 151–182. New York: Cambridge University Press.
Johnson, Paul. 1976. A History of Christianity. New York: MacMillan Publishing.Maddison, Angus. 2003. The World Economy: Historical Statistics. Paris: Development Centre of
the Organisation for Economic Co-operation and Development.Mankiw, N. Gregory, David Romer, and David N. Weil. 1992. “A Contribution to the Empirics
of Economic Growth.” Quarterly Journal of Economics 107 (2): 407–437.Mincer, Jacob. 1984. “Human Capital and Economic Growth.” Economics of Education Review 3
(3): 195–205.Morrisson, Christian, and Fabrice Murtin. 2009. “The Century of Education.” Journal of Human
Capital 3 (1): 1–42.Obstfeld, Maurice, and Alan Taylor. 2004. Global Capital Markets: Integration, Crisis, and
Growth. Cambridge: Cambridge University Press.OECD. 2001. Measuring Capital; OECD Manual: Measurement of Capital Stocks, Consumption
of Fixed Capital, and Capital Services. OECD Publication Services, www.SourceOECD.org.Psacharopoulos, George. 2009. Returns to Investment in Higher Education: A European Survey.
A Contribution to the Higher Education Funding Reform Project, Unpublished.Psacharopoulos, George, and Harry Patrinos. 2004. “Returns to Investment in Education: A Fur-
ther Update.” Education Economics 12 (2): 111–134.Ramirez, Francisco O., and John Boli. 1987. “The Political Construction of Mass Schooling:
European Origins and Worldwide Institutionalization.” Sociology of Education 60 (1): 2–17.Schultz, Theodore W. 1961. “Investment in Human Capital.” American Economic Review 51 (1):
1–17.United Nations. 2009. System of National Accounts 2008. New York: UN.Woodberry, Robert D. 2004. The Shadow of Empire: Christian Missions, Colonial Policy, and
Democracy in Postcolonial Societies. PhD Dissertation, University of North Carolina atChapel Hill, NC.
138 T.R. Breton
Dow
nloa
ded
by [
Mos
kow
Sta
te U
niv
Bib
liote
] at
09:
11 1
5 Fe
brua
ry 2
014
