1. Composition of Human Capital, Distance to the
Frontier and Productivity
Marianna Papakonstantinou
(University of Groningen, Netherlands)
Paper for the 33rd IARIW General Conference
Rotterdam, Netherlands
August 24-30, 2014, Session 2D
3. I. Background:
• Studies examining role of human capital on productivity
growth produced mixed results
• Could the effect of human capital vary with stages of
economic development?
4. II. Objectives:
• To examine effect of dissimilar types of human capital on
productivity growth for countries at different distances from
world technology frontier
• Focus: Externality Effects of Productivity Growth
• Hypothesis: Whether skilled human capital contributes more
to productivity in countries closer to frontier and better suited
to innovation?
• Empirical Support:
○ Vandenbussche et al [2006] for OECD countries
○ Ang et al [2011] for high & middle-income countries
• What is Important? : Types of Human Capital need to be
taken into account.
5. • Argument:
○ Average & upper-tail human capital had different impacts on
growth before & after industrialization
(Squicciarini & Voigtländer, 2014).
○ Interaction between education & distance to frontier has been
significant determinant of growth (Madsen, 2013)
○ Effect of high-skilled human capital turns negative for countries
very far from the technology frontier
(Vandenbussche et al, 2006)
• Motivation of the Study: Examine the mechanism on the
relationship between high-skilled human capital and productivity
growth in countries far (close) from (to) the frontier
6. III. Model:
Vandenbussche et al [2006] Model:
TFP Growth = f (Proximity to TFP frontier, High-skilled Human
Capital, Interaction Variable)
TFP Growth = Productivity growth in 5-year interval
Proximity to TFP frontier = Ratio of country TFP level to that of
USA at current PPP
High-skilled Human Capital = Population aged 25 and over with
higher education
7. IV. Data:
• Barro and Lee [2013] dataset on educational attainment in 5-year
intervals
• Revised Penn World Tables (Feenstra et al., 2013) on productivity
measures
○ Why Revised PWT better:
* Accounts for differences in asset composition across countries
and over time
* Do not use same labor share across countries
* Human capital of workers approximated by years of
schooling & rate of return
8. V. Results:
• Restricted Sample of 19 OECD countries: 1960-2000
• Extended Sample of 109 countries: 1950-2010
• OLS Results
IV Results (using past regressors as instruments to control
endogeneity)
V.1 Summary Statistics:
• Productivity growth higher in OECD economies (0.056)
compared to extended sample (0.025) → OECD countries lie
closer to the world technology frontier (US)
• Tertiary & Secondary human capital higher in OECD sample
than in extended sample
9. V.2 Findings of Baseline Regressions: 19 OECD
Countries:
• Proximity to TFP frontier has negative & significant effect on
productivity growth
• Tertiary human capital almost has a positive & significant impact
on productivity growth
• Interaction term enters positively & significantly when country
effects are omitted but with insignificant coefficient when they are
included
• Country fixed effects are not jointly significant
• Overall OECD (high-income) results are consistent with
Vandenbussche et al [2006] & Ang et al [2011] that effect of tertiary
human capital on productivity growth increases as countries get
closer to frontier.
• Use of sophisticated productivity measure has not altered
Vandenbussche conclusions
10. V. 3 Findings of Extended Sample Regressions:
• Could effects of high-skilled human capital turn negative for
countries with low productivity levels?
• Results from extended sample of countries (109 countries) and years
(1950-2010):
• Country fixed effects are now highly significant
• Proximity to TFP frontier has negative & significant effect on
productivity growth
• Tertiary human capital yields positive & insignificant coefficient
• Interaction term enters with negative & significant coefficient
Tertiary human capital bears an overall positive effect on
productivity growth that decreases as countries move closer to the
frontier ↔ Support to the paper’s hypothesis
11. • Could there be non-linear forces that work at distances from the frontier
↔ Proximity to TFP frontier squared & its interaction with tertiary
human capital are added in regression
○ Effect of proximity variable is negative and significant
○ Level effect of human capital is positive and significant
• What is the marginal effect of tertiary education at the frontier?
○ Plot of marginal effect of tertiary human capital on productivity growth
shows larger effects for countries far from the frontier compared to those
closer to it ↔ Large externalities in low-productivity countries due to
higher incidence of learning from others but the effect diminishes as
higher education leads to unemployment/rent seeking
=> Greater impact of human capital on productivity growth in ‘imitating’
countries. The impact diminishes as country benefits from technology
adoption & the turning point indicates existence of U-shaped relationship
12. V.4 Introduction of Secondary Education:
• Could externalities originate from secondary-educated workforce?
↔ medium-skilled/secondary human capital, its interaction with
proximity & proximity-squared are added in regression:
○ Effect of tertiary human capita is positive and significant
○ Effect of secondary human capita is positive and insignificant
○ Interactions are jointly significant (not individually always)
• Plot of marginal effect reveals larger effects of tertiary human
capital than secondary & there is a U-shaped relationship between
tertiary human capital and productivity growth, which does not hold
for secondary education
• Marginal effect of secondary human capital is universally
decreasing, whereas that of tertiary reaches a minimum then starts
increasing
Overall, high-skilled human capital has increasing impacts on
productivity growth for countries close to frontier.
13. VI. Contributions/Conclusions of the Paper:
VI.1 Studies the relationship between high-skilled (tertiary) human
capital & productivity growth for both developed & developing
world
Main Conclusions:
• Tertiary human capital positively affects productivity growth in all
countries
• Tertiary education involve U-shaped impact on productivity growth:
Large impact for countries far from frontier, decreases as countries
move closer & then increases again.
• Confirms Vandenbussche [2006] hypothesis on complementarities
of high-skilled human capital & innovation in countries close to
world technology frontier
14. •Tertiary human capital in “innovating” country (close to
frontier) contributes to productivity growth, because high-
skilled people are aware & have better understanding of the
processes
•Tertiary human capital in an “imitating” country (far from
frontier) prompts new technologies being adopted & operated
efficiently through diffusion of knowledge
15. VI. 2 Allows for the possibility that externalities also originate
from ‘medium-skilled’ (secondary-educated workforce)
Main Conclusions:
• Medium skilled’ human capital has positive (smaller) effect
on productivity growth
• Secondary education involves decreasing impact on
productivity growth ↔ lacks the complementarities to
innovation
16. VI.3 Uses sophisticated productivity measure incorporating
inter-country differences in work-hours & educational
composition of workforce
○ Crude Productivity Measure: Vandenbussche et al [2006]
○ Sophisticated Productivity Measure: Inklaar et al [2008]
○ Inklaar et al [2008] found differing result: No evidence of
high-skilled human capital externalities with close proximity
to frontier.