Drivers of Developing Asia’s Growth:
Past and Future
Donghyun Park, Asian Development Bank+ and
Jungsoo Park, Sogang University++
September 2010
Abstract
While developing Asia has recovered strongly from the global crisis, the region faces the
medium- and long-term challenge of sustaining growth beyond the crisis. The central
objective of this paper is to empirically investigate the sources of economic growth in 12
developing Asian countries during 1992-2007 via a 2-stage analysis. In the first stage, we
estimate total factor productivity growth (TFPG) and account for the relative importance of
labor, capital and TFPG in growth. In the second stage, we examine the effect of fundamental
determinants of growth such as human capital on both economic growth and TFPG. We find
that TFPG is becoming more important as a source of growth and a significant positive effect
of some fundamental determinants. Overall, our results strongly confirm the relevance of
supply-side factors for developing Asia’s medium- and long-term growth. The overarching
implication for policymakers is that supply-side policies which augment productive capacity
will be vital for sustaining developing Asia’s future growth in the post-crisis period.
Keywords
Growth, total factor productivity, factor accumulation, growth accounting, determinants of
growth, panel data, Asia
JEL codes
O40, O47
Acknowledgment
The authors gratefully acknowledge the valuable comments of participants at the Asian
Development Bank Workshop on Pushing Developing Asia’s Frontier Forward: Sustaining
Growth Beyond the Crisis, held at ADB Headquarters, Manila, Philippines, on 20 July 2010.
+
Principal Economist, Macroeconomics and Finance Research Division, Economics and Research Department,
Asian Development Bank, 6 ADB Avenue, Mandaluyong City, Metro Manila, PHILIPPINES 1550 [E-mail]
dpark@adb.org, [Tel] 63-2-6325825, [Fax] 63-2-6362342
++
Professor of Economics, School of Economics, Sogang University, Seoul, KOREA 121-742. [E-mail]
jspark@sogang.ac.kr, [Tel] 82-2-705-8697, [Fax] 82-2-704-8599
1 Introduction: Sustaining Developing Asia’s Growth Beyond the Global Crisis
During 2008-2009, developing Asia was fully preoccupied with overcoming the adverse
impact of the global financial and economic crisis. In sharp contrast to the Asian financial
crisis of 1997-1998, which first broke out in the region’s financial markets and then later
spread to its real economy, the region’s financial stability was largely unimpaired. Instead
trade was the primary transmission mechanism which spread the crisis spread from the
advanced economies to the region. More specifically, the collapse of global trade brought
about by the compression of demand in industrialized countries climaxed during the 4th
quarter of 2008 and 1st quarter of 2009, with predictably dire consequences for the region’s
exports and growth. During those two quarters there were widespread concerns that the
export-dependent region would suffer a deep and protracted recession. Fortunately, however,
the region has staged a stunning V-shaped recovery which has surpassed all expectations.
Developing Asia’s resilience during the global crisis was as remarkable as its resilience
during the Asian crisis. While the region’s GDP growth rate slipped from an average of 8.83%
during 2005-2007 to 6.6% in 2008 and further to 5.2% in 2009, it is projected to rebound
strongly to 7.5% in 2010 and 7.3% in 2011. The region is recovering faster than other parts of
the world and is in fact helping to lead the global economy out of recession.
Understandably and appropriately, the overriding priority of developing Asian governments
in 2008-2009 lay in mitigating the impact of the global crisis on domestic economic activity.
Even though the region’s financial systems did not suffer the credit crunch which paralyzed
their counterparts in the advanced economies, the region’s governments supported them with
various liquidity support measures to ensure their continued stability. Central banks around
the region decisively and repeatedly cut interest rates to support both their financial systems
and real economies. Above all, governments across the region quickly rolled out sizable fiscal
stimulus programs to support aggregate demand in the face of plunging exports and feeble
2
private consumption and investment. Developing Asia’s fiscal stimulus programs, the most
notable of which was the PRC’s 4 trillion yuan, about 13% of GDP, are widely believed to
have contributed to the region’s recovery. ADB (2010) finds some evidence that the region’s
fiscal stimulus, in particular higher public spending, had a significant, positive impact on the
region’s output during the global crisis. While the exact contribution of the stimulus to the
recovery remains uncertain, at a minimum, the fiscal and monetary stimulus is likely to have
boosted consumer and business confidence at a time of plummeting confidence.
Short-run output stabilization was at the front and center of developing Asia’ economic
policymakers’ agenda during the global crisis. The region’s unprecedented fiscal and
monetary stimulus was an unprecedented policy response to an unprecedented external shock.
In contrast to industrialized countries, developing Asia has only limited experience with using
macroeconomic policy for stabilizing short-run output fluctuations. For example, automatic
fiscal stabilizers such as unemployment benefits remain underdeveloped in developing Asia
relative to industrialized countries. For the most part, macroeconomic policy had been geared
toward safeguarding macroeconomic stability – e.g. low inflation and healthy public finances
– rather than smoothing the business cycle. The immediate lesson from the global crisis for
the region is that countercyclical fiscal and monetary policy can help cushion the impact of
severe adverse shocks. The broader lesson is that developing Asia’s tradition of sound and
responsible fiscal policy has an important but hitherto underappreciated benefit – the fiscal
space to unleash fiscal stimulus during extreme crisis. At a more fundamental level, the
global crisis brought to the fore the desirability and feasibility of using fiscal and monetary
policy for short-run output stabilization under certain circumstances.
As the crisis recedes and recovery gathers momentum, however, short-run output
stabilization will give way to long-run output growth as the top priority of developing Asia’s
policymakers. It is true that there is no clear-cut dichotomy between short-run stabilization
3
and long-run growth, and the former can pave the way for the latter. If developing Asia had
suffered a much more pronounced impact from the global crisis, the region would have been
in far worse position for medium-term growth in the post-crisis period. Nevertheless, the
distinction matters because short-run output fluctuations are more influenced by aggregate
demand whereas long-run growth depends more on supply-side factors which augment an
economy’s productive capacity. Therefore, policies which reduce short-term output volatility
are targeted toward aggregate demand whereas policies which foster long-term growth are
targeted toward boosting the supply of productive factors and their productivity. While
developing Asia has surpassed all expectations in weathering and recovering from a once-ina-lifetime external shock, the key question now becomes whether the region can sustain rapid
growth beyond the recovery. That is, what are the major obstacles to long-term growth in the
post-crisis world and what must the region do to successfully overcome them?
For developing Asia, sustaining growth in the medium- and long-run matters and matters
hugely for a number of reasons. Above all, for all its sustained rapid growth in the decades
prior to the global crisis and its remarkable resilience during the global crisis, developing
Asia remains by and large a poor region. The region remains home to two-thirds of the
world’s poor despite the massive reduction of poverty which has accompanied the region’s
rapid growth. One has to remember the very low initial base – i.e. very low per capita income
levels and correspondingly high poverty rates – from which the region began its economic
ascent. There is undoubtedly an element of truth in the countless news headlines proclaiming
the dawn of the Asian century and a seismic shift in the global balance of economic power
from the West to the East. However, such headlines should not detract from the fundamental
fact that the region still lags far behind the industrialized countries in terms of living
standards and hundreds of millions of its citizens still live below the poverty line.
Furthermore, there is no guarantee that the stellar growth of the pre-crisis period will
4
automatically carry over into the post-crisis period. There is also an unfortunate tendency in
the mass media to highlight GDP – e.g. “China set to become the world’s second biggest
economy” – and neglect GDP per capita which is much more relevant for living standards. In
short, sustaining long run growth in the post-crisis period is important both for lifting up
general living standards and making a further dent on still-widespread poverty.
Sustaining growth in the post-crisis world will be more challenging in many ways than in
the pre-crisis world because the external environment is likely to be less benign. In particular,
the global crisis could have far-reaching ramifications for the region’s export-led growth
paradigm. In striking contrast to most previous financial crises, the current global crisis
originated in industrialized countries. As a result, the global crisis has hit industrialized
countries much harder than developing countries and recovery has been noticeably quicker
and more robust in the latter group of countries. This asymmetry marks a continuation and
intensification of a gradual but secular rise in the relative importance of developing countries
in the world economy. Part of the burden of unwinding the global current account imbalances
which contributed to the global crisis will fall on the deficit countries, in particular the US.
The projected decline in US consumption and rise in US savings is desirable for global
stability but will adversely affect developing Asia’s export and growth prospects in the short
run. Likewise, the general weakening of European economies reeling from fiscal problems
does not bode well for their revival as dynamic markets for Asian exporters. The post-crisis
world is also likely to be more volatile since industrialized countries, bedrocks of stability
prior to the crisis, have become potential sources of volatility since the crisis. Instability
emanating from industrialized countries will have a proportionately bigger impact on global
stability as the 2008-2009 global crisis painfully illustrated.
As developing Asia’s recovery consolidates and gains a firmer footing, medium and long
run growth will reassert themselves as the region’s top priorities. The global crisis has
5
highlighted the feasibility and desirability of short-run output stabilization in the face of a
severe external shock to a region unaccustomed to using fiscal and monetary policy for
countercyclical purposes. However, over a longer time horizon, the still-low average per
capita incomes and still-large poor population of the region means that medium and long run
growth overshadow short run stability as the overall macroeconomic objective. Maximizing
long-run growth matters more than minimizing short-run output volatility in developing Asia,
and the positive experience with countercyclical stabilization during the global crisis should
not obscure this fundamental point. An important question which arises in connection with
maximizing the region’s medium and long run growth is whether there needs to be a
fundamental re-think of the Asian growth model. What has made this question all the more
relevant is the changes to the global economic environment which may call for adjustments to
the pre-crisis growth model. All in all, the global crisis provides an opportune time for taking
stock of the region’s sustained rapid growth in the pre-crisis period as well as thinking about
effective ways to cope with constraints to growth in the post-crisis period.
2 Toward a New Asian Growth Paradigm?
Many observers might find it puzzling as to why we are even posing the question of
whether developing Asia needs a new model in order to sustain growth beyond the crisis.
After all, the region has stood out for its sustained rapid growth and easily outperformed the
rest of the world for decades. As a result of its strong fundamentals, developing Asia is also
recovering more quickly and strongly than other regions and is leading the world out of
recession. Therefore, it may seem an odd time to re-examine the appropriateness of Asia’s
time-tested growth model. Indeed many elements which contributed to the region’s superior
long-run performance in the pre-crisis period will continue to serve it well in the post-crisis
period. For example, developing Asia’s prudent fiscal and monetary policy laid the
foundation for macroeconomic stability which enabled the region’s firms and households to
6
plan for the long term. Fiscal and monetary policies which foster macroeconomic stability
will remain highly relevant for long run growth in the post-crisis world. Another example of a
timeless ingredient of the Asian miracles is openness to foreign trade and technology. An
outward-looking growth strategy which reinforces the region’s vital links with the outside
world will continue to deliver huge benefits for the region’s growth and welfare in the postcrisis period. High savings and investment rates which enabled a rapid build-up of the
region’s physical capital stock in the pre-crisis period will do so beyond the crisis as well. In
short, many ingredients of the recipe for pre-crisis success remain valid for post-crisis success.
The one ingredient of the region’s pre-crisis growth paradigm that has been called into
question by the global crisis is the region’s export-oriented growth. However, developing
Asia’s experience during the global crisis does not invalidate developing Asia’s export-led
growth strategy. Exports enabled Asian producers to overcome the limitations of small
domestic markets and forced them to become more efficient in order to compete successfully
in highly competitive foreign markets. What the global crisis highlights is not so much the
risks of growing via integration into the world economy but rather the costs of neglecting the
potential contribution of domestic demand to growth. To the extent that various structural
distortions impede domestic demand and production for the domestic market, removing such
distortions can provide the economy with an additional source of growth and dynamism
[ADB (2009a)]. In fact, rebalancing Asia’s growth toward domestic sources has become a
policy priority in many Asian countries, most notably the PRC. Rebalancing involves not
only demand-side measures aimed strengthening domestic demand – e.g. social protection –
but also supply-side measures which boost industries and firms that cater to domestic demand
– e.g. services. A natural consequence of stronger domestic economies would be stronger
intra-regional trade which would enable regional countries to exploit hitherto under-exploited
gains from trade with their neighbors [ADB (2009b)]. Rebalancing and intra-regional trade
7
are thus two new and distinctive elements of the post-crisis Asian growth paradigm.
While a greater role for domestic demand and intra-Asian trade will be important elements
of developing Asia’s post-crisis economic landscape, in this paper we want to instead explore
the supply side factors which become more influential over a longer time horizon. In the long
run, an economy’s growth is determined primarily by supply side factors, in particular the
accumulation of factors of production – capital and labor – and their productivity. Those
factors are likely to overshadow the structure of demand as the determinants of developing
Asia’s long-run growth performance. There is a long-running debate on whether developing
Asia’s growth was driven by factor accumulation or productivity gains. The balance of
evidence suggests that both played a substantial part in driving the region’s sustained rapid
growth. Yet there are good reasons to believe that productivity gains will become relatively
more important for growth than factor accumulation in the post-crisis period. For one, the
high-savings, high-investment paradigm of the pre-crisis period will yield smaller benefits
when the physical capital stock has largely been built up, as is the case in many East and
Southeast Asian countries, precisely as a result of the high-saving, high-investment paradigm.
These economies are maturing and set to experience diminishing marginal returns to capital.
In the case of the PRC, the evidence indicates that the country is saving too much and
investing too much [ADB (2009c)]. The evidence also indicates that the post-Asian crisis
decline of investment rates in countries hardest hit by the Asian crisis represents a reversal of
over-investment rather than under-investment. All these considerations suggest that
productivity growth holds the key to sustained growth in the post-crisis period.
In addition to a less benign external environment arising from weaker demand and greater
volatility in industrialized countries, developing Asia also faces a number of homegrown
structural shifts which impinge upon the region’s long-run growth. Above all, the
demographic dividend which has been a major contributor to growth in the past is coming to
8
an end. While there is considerable heterogeneity across regional countries in terms of the
demographic transition, and some countries will reap demographic dividends for years to
come, the overall pattern for the region as a whole is one of rapid ageing. Favorable
demographic structures with high shares of working-age population are giving way to older
populations with a growing share of economically inactive retirees. The prospective decline
in labor force as a result of demographic transition means that labor inputs will become less
important as sources of economic growth. To some extent, encouraging women and retirees
to work more can compensate for the smaller working-age population but population graying
will have an impact. The demographic transition will also adversely affect national savings
rates since individuals tend to save when they are young and working and draw down their
savings when they are old and retired. The reduction in savings may encourage a shift away
from capital-intensive growth which was partly a result of abundant savings. Developing
Asia’s population aging is an important additional reason why its growth is likely to depend
more on productivity growth and less on more labor and capital in the post-crisis period.
The global crisis emphatically shattered the notion that sustained rapid growth was an
automatic feature of developing Asian economies. Although the region has rebounded from
the crisis with admirable resilience, the crisis served as a timely and sobering reminder that
growth can be derailed by external and internal shocks. It is certainly true that developing
Asia has performed remarkably well in the pre-crisis period. The region’s sustained growth,
which has lifted hundreds of millions of its citizens out of poverty into more humane and
productive lives, is one of the biggest success stories in economic history. It should be
emphasized that many of the elements of the Asian growth paradigm which served the region
well before the global crisis will continue to serve it well beyond the crisis. The region did so
well because it got many of the fundamentals “right” and there is no compelling reason to
move away from those fundamentals. At the same time, however, precisely because Asia
9
grew so rapidly the Asia of today is far different from the Asia of yesterday. This means that
some of the ingredients of Asia’s spectacularly successful recipe for growth are less relevant
for today’s industrialized, middle-income Asia than they were for yesterday’s agricultural,
low-income Asia. In particular, productivity growth will become a relatively more important
source of growth than re-allocation of surplus rural labor and rapid accumulation of physical
capital. The key to sustaining rapid growth beyond the global crisis thus lies in improving the
productivity of labor and capital on a sustained basis.
Mapping out the region’s future growth requires a basic understanding of the region’s past
growth. In this connection, the key question is “What have been the drivers of the region’s
growth in the past?” A related question is “Has the relative importance of the different growth
drivers changed over time?” This evolution in the sources of growth over time holds telling
clues about the likely sources of future growth. Above all, any shift in the relative importance
of growth drivers – e.g. some drivers are becoming more important, other drivers are
becoming less important – in the pre-crisis period can inform us about the likely structure and
direction of growth in the post-crisis period. This, in turn, can inform policymakers about the
major constraints to growth which have to be addressed and, more broadly, the kinds of
policies that they need to put into place in order to sustain growth beyond the crisis. The next
two sections describe the empirical analysis of developing Asia’s growth drivers during 19922007, and report and discuss the main findings from the analysis.
3 Recent Patterns of Growth in Developing Asia: Growth Accounting
In this section, we examine recent patterns of growth in developing Asia. More specifically,
we examine the relative importance of capital, labor and total factor productivity (TFP) in
explaining the region’s economic growth during 1992-2007. As explained earlier, the two
primary sources of growth are (1) the accumulation of factors – i.e. growth in the quantity of
capital and labor and (2) total factor productivity (TFP) growth. Although labor productivity
10
growth, or the increase in output produced by one unit of labor, is also a widely used
indicator of productivity, TFP growth is a more accurate indicator of productivity
improvements since it indicates improvement in the efficiency in production, controlling for
the contribution of all factors of production that are used. In addition, in the case of
developing Asia, the key question regarding growth has always been the relative importance
of capital accumulation versus productivity growth, and this can be better resolved by looking
at TFP rather than labor productivity.
3.1 Empirical Framework and Data
The twelve developing Asian countries in our sample are PRC, Hong Kong, India,
Indonesia, Korea, Malaysia, Pakistan, Philippines, Singapore, Taipei, Thailand and Viet Nam.
The twelve countries are divided into three groups – (1) the PRC, (2) four NIEs – Hong Kong,
Korea, Singapore, and Taipei, and (3) seven developing Asian economies – India, Indonesia,
Malaysia, Pakistan, Philippines, Thailand and Vietnam. The PRC is treated as a separate
group in light of its size, exceptional growth and unique structural characteristics. For
comparative purposes, we also include the G5, which we divide into Japan and non-Asian G5
– i.e. US, France, Germany and UK. We divide our sample period – 1992-2007 – into three
distinct sub-periods with different structural characteristics: 1992-1997 marks the pre-Asian
crisis period characterized by imbalances which brought about the crisis, 1997-2002 marks
the immediate post-Asian crisis period characterized by restructuring and reform, and 20022007 marks the most recent sub-period characterized by rapid pre-global crisis growth.
In calculating TFP growth, we assume two-input neoclassical production function with
constant returns to scale. TFP growth is calculated based on the following equation.
∆ 𝑙𝑛(𝑇𝐹𝑃) = ∆ 𝑙𝑛(𝑌) − (1 − 𝑎𝐿 )∆ 𝑙𝑛(𝐾) − 𝑎𝐿 ∆ 𝑙𝑛(𝐿)
(1)
where Y is GDP, K is the capital stock, and L is the labor.
This basic formula uses a labor input without quality adjustment for human capital
11
differences. The parameter 𝑎𝐿 is the output elasticity with respect to labor. In most growth
accounting literature, it is common to assume competitive labor market under which output
elasticity with respect to labor is equal to the labor shares of GDP. The data on labor
compensation is available in the National Accounts Statistics, UN. However, for the 12 Asian
economies in question, the data on labor compensation is available for only limited countries
and for only limited period. Therefore, we have chosen to produce TFP growth based on two
different methods. First, we have calculated labor shares for the Asian economies with labor
compensation data. As for the countries without the labor share data, we have borrowed and
applied those of the Asian economies with labor share data. Secondly, we followed the study
of Fischer (1993) and have assumed a common labor share to be 0.6.
Labor is usually considered to be augmented by enhancements in human capital. Since
formal education is a major source of human capital enhancement, we incorporate average
educational attainment years of the population (h) of each country to augment labor.
Here, we consider two types of labor quality adjustments. First, labor (L) is linearly
adjusted by human capital (h): hL. The TFP growth estimates is calculated based on the
following equation.
∆ 𝑙𝑛(𝑇𝐹𝑃) = ∆ 𝑙𝑛(𝑌) − (1 − 𝑎𝐿 )∆ 𝑙𝑛(𝐾) − 𝑎𝐿 [∆ 𝑙𝑛(𝐿) + ∆ 𝑙𝑛(ℎ)]
(2)
Second, labor is exponentially adjusted by human capital: exp(0.08*h)L. This adjustment
method is taken from Barro and Lee (2010). The TFP growth estimates is calculated based on
the following equation.
∆ 𝑙𝑛(𝑇𝐹𝑃) = ∆ 𝑙𝑛(𝑌) − (1 − 𝑎𝐿 )∆ 𝑙𝑛(𝐾) − 𝑎𝐿 [∆ 𝑙𝑛(𝐿) + 0.08∆ℎ]
(3)
3.2 Empirical Results
In this sub-section, we report and discuss the results of the empirical analysis described in
the preceding sub-section. Table 1 reports the TFP estimates calculated for 1992-1997 on the
basis of equation (1) without any adjustment for human capital. The first three rows of Table
12
1 report the average growth in output, capital and labor during 1992-1997 for each of our
country groups – four non-Asian G5 countries, Japan, 4 NIEs, China and 7 developing Asian
economies. The next three rows report the respective contribution of capital, labor and TFP to
output growth under the assumption that labor share is equal to actual share of labor in
national income. This assumption is denoted as C1. Contribution of capital is the percentage
point of the output growth that is explained by the growth in capital (= (1 − 𝑎𝐿 )∆ 𝑙𝑛(𝐾)).
Contribution of labor is the percentage point of the output growth that is explained by the
growth in labor (= 𝑎𝐿 ∆ 𝑙𝑛(𝐿) ). Contribution of TFP is the percentage point of the output
growth that is explained by the TFP growth (= ∆ 𝑙𝑛(𝑇𝐹𝑃)) The next row reports the relative
portion of output growth that is explained by the TFP growth ( = ∆ 𝑙𝑛(𝑇𝐹𝑃) /∆ 𝑙𝑛(y)) . For
example, for the non-Asian G5, since the contribution of TFP growth to output growth is 1.04%
and output growth is 2.35%, the relative contribution of TFP growth is 44% (=1.04/2.35).
The next four rows report the results of the identical growth accounting exercise under an
alternative assumption about the share of labor, which is assumed to be 0.6 as in Fischer
(1993). This alternative assumption is denoted as C2. Table 2 and 3 reports the results of the
identical empirical analysis for 1997-2002 and 2002-2007, respectively.
[Table 1]
[Table 2]
[Table 3]
The most striking result from Tables 1, 2 and 3 is that there has been a clear shift in the
sources of growth from physical capital accumulation to TFP in developing Asian countries.
Prior to 2002, the expansion of the capital stock was the main source of output growth in the
region but after 2002 TFP growth accounted for a much larger share of growth. Throughout
the entire sample period, the contribution of labor was minimal for all Asian economies. The
contribution of TFP growth for the Asian economies are lower when actual labor shares are
13
used since their actual labor shares are typically less than 0.6. As a result, higher weights are
applied to capital stock growth, which was very high. The relative contribution of TFP was
lower than for the non-Asian G5 until 2002. However, estimates and contributions of TFP
growth have increased significantly in the period of 2002–2007 for the 4 NIEs and 7 ADEs.
The TFP growth estimates for the 11 Asian economies for this sub-period are even higher
than those of the non-Asian G5. The estimates and contribution of China’s TFP growth are
strongly positive throughout the entire sample period, showing a very different pattern
compared to those of the Asian economies at a similar developmental stage.
Table 4 reports the TFP estimates calculated for 1992-1997 on the basis of equation (2) and
(3) to adjust labor for human capital. In addition to the average growth in output, capital and
labor during 1992-1997, we also report the average growth in the two alternative definitions
of human capital – human1 = h and human2 = exp(0.08*h). We report the contribution of
capital, labor, TFP and human capital to output growth. The contribution of human capital is
the percentage point of output growth that is explained by the growth in human1 (=
𝑎𝐿 ∆ 𝑙𝑛(ℎ𝑢𝑚𝑎𝑛1) ) and human2 ( = 𝑎𝐿 ∆ 𝑙𝑛(ℎ𝑢𝑚𝑎𝑛2) ). C3 denotes the results for linearly
adjusting labor for human capital – i.e. human1 – while C4 denotes the results for exponential
adjustment – i.e. human2. Table 5 and 6 report the results of the identical growth accounting
exercise for 1997-2002 and 2002-2007, respectively. Labor shares are assumed to be 0.60 in
all three tables.
[Table 4]
[Table 5]
[Table 6]
As was the case when we did not adjust labor quality for human capital, the most striking
result is the shift in the source of growth from physical capital accumulation to TFP after
2002. Although additions to the capital stock were the main driver of growth until 2002, the
14
relative importance of TFP rose noticeably after 2002. For the 4 NIEs, the relative
contribution of TFP growth was sizeable in 1992–1997 but dropped during 1997–2002.
However, the absolute size and relative contribution of TFP growth became dominant after
2002. For the 7 DAEs, TFP growth was either negative or marginal until 2002, but became
the dominant driver of growth after 2002. The estimate of TFP growth and its contribution to
output growth increased significantly in the period of 2002–2007 for both NIEs and DAEs.
The TFP growth estimates for the 11 Asian economies for this sub-period are even higher
than those of the non-Asian G5. The contribution of TFP growth is lower for Asian
economies when labor is adjusted linearly for human capital. Throughout whole sample
period, the contribution of labor was minimal for all Asian economies. Growth in human
capital for 4 NIEs was lower than the G5 until 2002 but turned higher afterwards. For the 7
DAEs, the growth in human capital was higher than all other groups except China for all
periods. The estimate of China’s TFP growth and its contribution to China’s output growth
remain strongly positive throughout the entire sample period.
Most other growth accounting studies for Asian economies look at the pre-1990 period.
Collins and Bosworth (1997) performed growth accounting for 1960–1994 period, but this
still does not overlap much with our sample period. Their sample included 3 NIEs and 4
ASEAN countries. Other studies include Young (1994) and World Bank (1993), both of
which look at 4 NIEs and 3 ASEAN countries. Interestingly and significantly, previous
growth accounting studies based on the pre-1990 period have found that capital accumulation
or more broadly input-based growth was the main source of growth in Asian economies. Our
study finds that this growth pattern continued up to 2002 but TFP growth emerged as a
relatively more important growth driver since then. Taken together, the evidence from our
study and existing studies suggests that factor accumulation, in particular capital
accumulation, drove the region’s growth for an extended period of time but the growth
15
paradigm is recently shifting toward one in which productivity plays a bigger role.
The consistently large size of China’s TFP estimate and its large influence on output growth
is something of a puzzle in light of the widespread perception that high savings and high
investment underlie the country’s exceptionally rapid growth. There are a number of possible
explanations for our finding of an oversized role of TFP growth in China’s growth. First,
capital stock may be underestimated, in which case TFP growth will be overestimated.
Second, China’s huge TFP growth may reflect to extensive, economy-wide re-allocation of
resources from low-productivity areas to high-productivity sectors – i.e. from rural to urban,
from agriculture to manufacturing. However, this type of re-allocation of resources across
sectors also happened in Korea and other Asian countries in the 1970s and 80s, but these
countries did not experience such spectacular TFP gains. A third explanation has to do with
China’s transition from a centrally planned economy to a market-oriented economy and the
consequent removal of pricing and other distortions. According to this view, the jump in TFP
is a consequence of getting prices right, establishing private property rights, opening up to
foreign trade, and other efficiency-promoting structural changes.
To sum up the results of our growth accounting exercise, which sought to assess the relative
importance of capital, labor and total factor productivity (TFP) in developing Asia’s
economic growth during 1992-2007, the source of growth seems to be shifting from capital
accumulation to TFP growth since 2002. Prior to 2002, capital accumulation was the
dominant source of growth prior to 2002, and this result is consistent with the evidence from
the existing literature. While the existing literature looks largely at data prior to 1990, our
evidence indicates that the capital-led pattern of growth persisted until 2002. Our central
finding – i.e. the emergence of TFP growth as an important source of growth since 2002 –
holds for both NIEs and developing countries. Furthermore, the finding is robust and
consistent across different specifications, including different adjustments for human capital.
16
Overall our evidence suggests that fostering TFP growth holds the key to sustaining the
region’s growth although capital accumulation will continue to contribute sustantially.
4 Explaining Output Growth and TFP Growth: Estimation of Growth Equations
In this section, we attempt to explain output growth and TFP growth through a number of
explanatory variables which are widely used in the standard empirical literature on growth.
The growth accounting exercise of the previous section attributed economic growth to two
broad sources – growth in the supply of productive factors and TFP growth. In this section,
we incorporate a much larger number of explanatory variables in order to identify more
specific sources of growth. Following most of the literature, we define output growth as the
growth rate of GDP per worker. In light of our key finding of the growing importance of TFP
growth in output growth in developing Asia since 2002, we also seek to explain TFP growth,
estimated in the previous section, with the same set of explanatory variables. Therefore, we
look at two dependent variables – output growth and TFP growth. The explanatory variables
include growth of capital stock per worker, per capita GDP relative to the US to incorporate
the catch-up effect, life expectancy, human capital, population size, percentage of tropical
area, openness and inflation. Four additional explanatory variables are related to various
aspects of governance – government effectiveness, rule of law, control of corruption and
regulatory quality.
4.1 Empirical Framework and Data
The production function we use is basically the Cobb-Douglas specification. Human
capital is assumed to improve the quality of labor exponentially: exp(0.08*h)L. The level of
technology (A) depends on catch-up effect, human capital, and other country-characteristics.
Human capital therefore affects output through two channels. It is a factor of production and
also a contributor to the technological level.
𝑌 = 𝐴𝐹(𝐾, 𝐻𝐿)
17
𝑌 = 𝐴𝐾 1−𝑎𝐿 (𝐻𝐿)𝑎𝐿
Ȧ⁄A = 𝐹(𝑐𝑎𝑡𝑐ℎ − 𝑢𝑝 𝑒𝑓𝑓𝑒𝑐𝑡, ℎ𝑢𝑚𝑎𝑛 𝑐𝑎𝑝𝑖𝑡𝑎𝑙, 𝑜𝑡ℎ𝑒𝑟 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑎𝑛𝑡𝑠)
In order to identify determinants of the per worker GDP and TFP growth, we adopt the
following empirical models from Bosworth and Collins (2003) which is based on conditional
convergence theory (or catch-up effect): a country with a low initial income per capita
relative to its own long-run (or steady-state) potential level of income per capita will grow
faster than a country that is already closer to its long-run potential level of output per capita.
In their study, catch-up effect, openness, geographical factors, and institutional quality are
shown to be influential in GDP and TFP growth.
Studies such as Benhabib and Spiegel (1994), Dinopoulos and Thompson (2000), Bils and
Klenow (2000) and Sachs, Radelet, and Lee (2001) have adopted models where ‘level of
human capital’ influences productivity growth. Thus, we have augmented the model to
additionally reflect the role of human capital level in determining the per worker GDP growth
and TFP growth. We will use a comprehensive international country-level panel data to
estimate the main determinants. The specification is as follows
∆ 𝑙𝑛(𝑌/𝐿)𝑖𝑡 = 𝛽0 + 𝛽1 ∆ 𝑙𝑛(𝐾/𝐿)𝑖𝑡 + 𝛽2 𝑙𝑛 (
∆ 𝑙𝑛(𝑇𝐹𝑃)𝑖𝑡 = 𝛽0 + 𝛽1 𝑙𝑛 (
𝑌𝑖0
) +𝛽3 ℎ𝑢𝑚𝑎𝑛 + 𝛾 ′ 𝑍 + 𝑑𝑢𝑚_𝑦𝑟𝑡 + 𝜀𝑖𝑡
𝑌𝑢𝑠,0
𝑌𝑖0
) +𝛽2 ℎ𝑢𝑚𝑎𝑛 + 𝛾 ′ 𝑍 + 𝑑𝑢𝑚_𝑦𝑟𝑡 + 𝜀𝑖𝑡
𝑌𝑢𝑠,0
Where Y = output, L = labor, K = capital sock, Yi0 = country i’s initial per capita income, YUS0
= the US’s initial per capita income, human = human capital, Z = vector of control variables
and dumyr is dummy variable for time periods.
The base model equation includes initial conditions such as initial income per capita(-)
𝑌
relative to the U.S. level (𝑌 𝑖0 ), educational attainment level (human) as the level of human
𝑢𝑠,0
capital and other potential determinants which includes the following variables: (i) initial life
18
expectancy relative to the U.S. (initial health condition), initial population; (ii) trade
instrument such as openness variable from Penn-World Tables (PWT); (iii) geographical
factor such as composite average of the number of frost days and tropical area; (iv) policy
variables such as inflation rates, current account balance relative to GDP; (v) institutional
factor such as rule of law, government effectiveness, control of corruption, or regulatory
quality. We also include time period dummies for the first two 5-year periods - 1992-1997
and 1997-2002. The only difference between the equation for output per worker growth and
TFP growth is that the former includes capital stock per worker as an explanatory variable.
We use an unbalanced international country-level panel data set from 1992 to 2007.1 The
data set includes 125 developing and developed countries.2 Appendix 1 lists the countries in
our sample. Appendix 2 shows the definitions of all dependent and independent variables,
along with their data sources. Since the annual variation of TFP growth is usually governed
by noise, we construct a ‘five-year interval data set’ which consists of average values or
initial values of variables from each five-year non-overlapping intervals within the full
sample.
3
Initial values of each respective interval are considered for the variables
representing initial conditions such as initial income per capita relative to the U.S. level,
initial life expectancy relative to the U.S., and initial population. To control for the omitted
time effect of each five-year intervals, panel regression with period-fixed effect is performed
on the five-year interval panel data set.
In the following three sub-sections, we report and discuss the results of the empirical
analysis described in the preceding sub-section. Section 4.2 reports and discusses the results
of estimating output per worker growth equations while Section 4.3 does the same for TFP
1
The ending year is set at 2007 due to the limited data availability of PWT.
We have excluded extreme data points where average annual TFP growth, per worker GDP growth, or per
worker capital stock growth is greater than 20% or less than – 20%.
2
3
More specifically, the five-year intervals considered are 1992-1997, 1997-2002, and 2002-2007.
19
growth equations. Section 4.4 empirically assesses the relative importance of the different
determinants of growth based on the results of Sections 4.2 and 4.3.
4.2 Empirical Results: Output per Worker Growth Equation Estimation
Table 7 reports the results of regressing the 5-year average growth rate of output per worker
on the explanatory variables. The following explanatory variables seemed to be significant –
growth in capital stock per worker (mdkkl), initial conditions: log of the per capita GDP
relative to that of the U.S. in the initial year of each respective 5-year interval (catch-up effect,
lny_us), initial population size (lnpop), human capital: 5-year averages of educational
attainment level (mhuman), geographical factor: percentage tropical area (mtropic), openness:
log of openness index from PWT (mopenc) and government effectiveness (from World Bank,
World Governance Indicator, mgoveff). Growth in capital stock per worker, population size,
human capital, openness, government effectiveness positively contributed to the growth in
GDP per worker. A lower initial per capita GDP relative to the US and relatively small
tropical area also had a positive effect. Variables that were not significant were life
expectancy, inflation rate and current account balance relative to GDP.
[Table 7]
Table 8 augments the analysis of Table 7 by incorporating 4 different indicators of
governance indicators - rule of law, government effectiveness, control of corruption,
regulatory quality - into the growth regression. Among the 4 governance indicators,
government effectiveness and control of corruption were shown to be significant in the
regression. Model (6) includes two interaction terms: mgoveff_a (= mgoveff* dummy_asia12)
to gauge the differential impact of government effectiveness for developing Asian economies
and mgoveff_o (= mgoveff * dummy_oecd) to do the same for OECD economies. In model
(6), the coefficient for the mgoveff rises and the coefficient for interaction term with OECD
dummy is significantly negative. This implies that the government effectiveness is more
20
important in GDP growth per worker for the non-OECD economies – i.e. developing
economies.
[Table 8]
To sum up our empirical evidence for GDP per worker growth regressions, the following
results were robust. The level of human capital was found to contribute to growth.
Furthermore, growth is higher when the country is more open in terms of trade. Government
effectiveness also contributes to growth. Non-tropical countries and countries with larger
population grow faster. There is some evidence of convergence, or the catch-up effect.
Among 4 different measures of governance indicators (rule of law, government effectiveness,
control of corruption, regulatory quality) government effectiveness and control of corruption
were shown to be significant in the regression. Finally, the role of government effectiveness
was greater for the non-OECD countries
4.3 Empirical Results: TFP Growth Equation Estimation
Output per worker can be decomposed into growth in capital stock per worker and TFP
growth. TFP growth is of particular interest to us since the growth accounting analysis of
Section 3 indicated that its relative importance as a source of growth was rising. Table 9
reports the results of regressing the 5-year average growth rate of output per worker on the
explanatory variables. The following explanatory variables seemed to be significant – initial
conditions : log of the per capita GDP relative to that of the U.S. in the initial year of each
respective 5-year interval (catch-up effect, lny_us), human capital : 5-year averages of
educational attainment level (mhuman), geographical factor : percentage tropical area
(mtropic), openness: log of openness index from PWT (mopenc), government effectiveness :
government effectiveness (from World Bank, World Governance Indicator, mgoveff). Human
capital, openness and government effectiveness positively contributed to TFP growth. A
lower the initial per capita GDP relative to the US and a relatively smaller tropical area also
21
had a positive effect. Variables that were not significant were life expectancy, population size,
inflation rate, and current account balance relative to GDP.
[Table 9]
In Table 10, we investigate whether determinants that have shown significance in Table 9
have differential impact in three different groups of countries – OECD, 12 Asian countries
and the rest of the world. We will consider interaction terms for three variables: human
capital (mhuman), openness (mopenc), and government effectiveness (mgoveff). Model (2)
includes the following interaction terms : mhuman_a (= mhuman*dummy_asia12) and
mhuman_o (= mhuman*dummy_oecd). In model (2), the coefficient for mhuman_a is
positive and significant, but mhuman_o is not significant. These interaction terms are additive
to mhuman. This implies that the role of human capital is greater in the 12 Asian economies
than in other countries. Model (3) includes the following interaction terms : mopenc_a (=
mopenc * dummy_asia12) and mopenc_o (= mopenc*dummy_oecd). In model (3), the
coefficient for mopenc_a is positive and significant, but mopenc_o is not significant. These
interaction terms are additive to mopenc. This implies that the role of openness is greater in
the 12 Asian economies than other countries. Moodel (4) includes the following interaction
terms : mgoveff_a (= mgoveff*dummy_asia12) and mgoveff_o (= mgoveff*dummy_oecd).
In model (4), the coefficient for mgoveff_a is not significant, but mgoveff_o is not negatively
significant. These interaction terms are additive to mgoveff. This implies that the role of
government effectiveness is greater for the non-OECD economies compared to the OECD
economies. The model (5) includes all interaction terms considered in models (2) to (4). In
model (5), the differential effects that we saw in the models (2) to (4) all disappear. This may
be due to multicolinearity problem due to inclusions of so many interaction terms.
[Table 10]
Table 11 shows the results of TFP growth regressions when we include the R&D variable
22
mdrk_l2. The following explanatory variables seemed to be significant – initial conditions :
log of the per capita GDP relative to that of the U.S. in the initial year of each respective 5year interval (catch-up effect, lny_us), human capital: 5-year averages of educational
attainment level (mhuman), geographical factor: percentage tropical area (mtropic), openness:
log of openness index from PWT (mopenc), current account deficit relative to the GDP
(mca_gdp), government effectiveness: government effectiveness (from World Bank, World
Governance Indicator, mgoveff), and growth in R&D stock per worker. Human capital,
openness, current account deficit relative to the GDP, government effectiveness, and growth
in R&D stock per worker positively contributed to TFP growth. Lower initial per capita GDP
relative to the US and relatively small tropical area also had a positive impact. Variables that
were not significant were life expectancy, population size and inflation rate.
[Table 11]
To sum up our empirical evidence for TFP growth regressions, human capital and openness
seem to have a positive significant effect on TFP growth. Government effectiveness also
seems to benefit TFP growth. Taken together, these results indicate that countries which
invest more in human capital, have higher levels of integration into the world economy and
enjoy stronger governance and institutions will enjoy more rapid TFP growth. We also find
that geography – i.e. smaller tropical areas – and R&D has a positive effect on TFP growth.
Our results lend support to convergence or the catch-up effect. Interestingly, we find that
human capital and openness play a bigger role in the TFP growth in the 12 Asian countries
than elsewhere. The role of government effectiveness was larger for non-OECD countries,
which implies that governance matters more for TFP growth in developing countries than in
developed countries.
4.4 Empirical Results: Relative Importance of Determinants of Growth
The previous two sections identified the statistically significant determinants of output per
23
worker growth and TFP growth. In this section, we attempt to measure the relative
importance of these determinants in output per worker growth and TFP growth. For output
per worker growth, we use coefficient estimates from model (3) of Table 8 to calculate the
relative contribution of the determinants to growth. More precisely, we calculate the relative
contribution as follows. We first calculate the predicted growth in GDP per worker for each
country. We then take the difference between the predicted growth in GDP per worker for
each country and global average of the predicted growth in GDP per worker - (predicted
dln(Y/L) of country j – global average of predicted dln(Y/L)). We take the difference
between each regressor for each country and global average of the respective regressor i - (Xi
of country j – global average of Xi ). We multiply the differenced values – i.e. the gap in
value from the global average – by the corresponding coefficient estimates of model (3) of
Table 8. The resulting values are presented in the table in bold figures. We divide the 12
Asian countries into 4 groups – (1) 4 NIEs – Hong Kong, Korea, Singapore and Taipei, (2)
ASEAN-4 – Indonesia, Malaysia, Philippines and Thailand, (3) 3 developing Asian
economies – India, Pakistan and Viet Nam, and (4) China.
The results of this exercise are reported in Table 12, which we can interpret as follows.
Predicted growth in GDP per worker refers to the predicted growth in GDP per worker based
on the estimates of model (3) of Table 8. Predicted growth gap in GDP per worker - gap from
the global average - refers to how much the predicted value of dln(Y/L) differs from the
global average of dln(Y/L). For example, for OECD during 1992–1997, the predicted growth
of GDP per worker was 0.65 percentage point higher than the global average. Of this gap in
growth, catch-up effect contributed minus 2.16% – i.e. the initial per capita income level is
higher than the global average, human capital contributed 1.02% – i.e. human capital is
higher than the global average, and so forth.
[Table 12]
24
We now look at and compare the results across groups of countries. For the period of 1992–
1997 and 2002– 2007, 12 Asian economies grew much faster than the OECD and other
developing economies. Growth in 2002 – 2007 are significantly higher relative to those of
the 1992–1997 period for China, 4 ASEAN, 3 DAEs, but lower for 4 NIEs and OECD. For
the 1997–2002 period, the Asian economies slowed down in growth while OECD and other
developing economies were relatively unaffected. The inequalities in Table 13 compare the
contribution of the determinants of output per worker growth across different groups of
countries in 1992–1997 and 2002-2007. For example, human capital contributed 1.02% to the
growth gap for the OECD countries, which is higher than 0.82% of the NIEs, and so forth.
Changes in relative rankings across different groups of countries are colored red but these are
relatively few.
We now look at the results over time – between 1992-1997 and 2002-2007 – for each of the
4 groups of Asian countries. For the 4 NIEs, catch-up effect became significantly more
negative as the relative income has risen compared to the global average (- 1.83 to – 2.02).
Contribution of human capital increased slightly (0.82 to 0.88), whereas contribution of
openness has risen significantly (0.56 to 0.66). Effectiveness of government relative to the
global average has reduced (0.86 to 0.78). For ASEAN-4, catch-up effect became more
negative as the relative income has risen compared to the global average (-0.06 to -0.14).
Contribution of human capital (-0.12 to -0.05) and openness (0.15 to 0.23) have risen
moderately. Effectiveness of government relative to the global average has fallen (0.18 to
0.09). For the 3 DAEs, catch-up effect has fallen as the relative income has risen compared to
the global average (1.16 to 0.98). Contribution of human capital (-1.13 to -1.00) and openness
(-0.52 to -0.28) have risen significantly. Effectiveness of government relative to the global
average has not changed. For China, catch-up effect has fallen as the relative income has
risen compared to the global average (1.08 to 0.25). Contribution of human capital (-0.23 to 25
0.05) and openness (-0.40 to -0.23) have risen significantly. Effectiveness of government
relative to the global average has fallen mildly (0.07 to 0.01).
[Table 13]
We now repeat the above exercise for TFP growth in order to identify the relative
importance of the different significant determinants of TFP growth. The results of this
exercise are reported in Table 14. Predicted TFP growth refers to the predicted TFP growth
based on the estimates of model (2) of Table 9. Predicted TFP growth gap - gap from the
global average - refers to how much the predicted value of dln(TFP) differs from the global
average of dln(TFP). For example, for OECD during 1992–1997, the predicted TFP was 0.75
percentage point higher than the global average. Of this gap in growth, catch-up effect
contributed minus 2.27% – i.e. the initial per capita income level is higher than the global
average, human capital contributed 1.06% – i.e. human capital is higher than the global
average, and so forth.
[Table 14]
We now look at and compare the results across groups of countries. For the period of 1992
–1997 and 2002–2007, 12 Asian economies grew much faster than the OECD and other
developing economies. TFP growths for all 12 Asian economies in 2002–2007 are
significantly higher than those of the 1992 – 1997 period. As for the 1997 – 2002 period, the
Asian economies slowed down in TFP growth while OECD and other developing economies
were relatively unaffected. The inequalities which compared the contribution of determinants
of TFP growth across different groups of countries are identical to the inequalities which
compared output per worker growth in Table 4.4.2 above.
We now look at the results over time – between 1992-1997 and 2002-2007 – for each of the
4 groups of Asian countries. For 4 NIEs, catch-up effect became significantly more negative
as the relative income has risen significantly compared to the global average ( - 1.92 to –
26
2.12 ). Contribution of human capital (0.85 to 0.91) and of openness has risen moderately
(0.37 to 0.44). Effectiveness of government relative to the global average has reduced (0.94
to 0.86). For ASEAN-4, catch-up effect became more negative as the relative income has
risen compared to the global average (-0.06 to -0.15). Contribution of human capital (-0.12 to
-0.05) and openness (0.10 to 0.15) have risen moderately. Effectiveness of government
relative to the global average has fallen (0.20 to 0.10). For the 3 DAEs, catch-up effect has
fallen significantly as the relative income has risen compared to the global average (1.22 to
1.03). Contribution of human capital (-1.17 to -1.04) and openness (-0.35 to -0.18) have risen
significantly. Effectiveness of government relative to the global average has not changed. For
China, catch-up effect has fallen significantly as the relative income has risen compared to
the global average (1.13 to 0.26). Contribution of human capital (-0.24 to -0.05) and openness
(-0.27 to -0.15) have risen significantly. Effectiveness of government relative to the global
average has fallen mildly (0.08 to 0.01)
5 Concluding Observations and Policy Implications
The single most interesting and significant finding which emerges from our empirical
analysis is that the primary source of developing Asia’s economic growth is shifting from
accumulation of physical capital to total factor productivity growth. For a prolonged period of
time, the region’s growth was driven by high investment which led to a rapid build-up of the
region’s capital stock and productive capacity. This should not be surprising since developing
Asia was a largely rural low-income region with a limited manufacturing base prior to the
rapid industrialization which transformed it into the most dynamic component of the world
economy. Given the speed and scale of developing Asia’s transformation, it is easy to forget
that the region was a typical part of the developing world just a generation ago. High
marginal returns to capital and hence high investment rates are to be expected in very poor
countries with very limited capital stocks. In the case of developing Asia, the region’s high
27
savings rates and consequently ample pool of savings further accentuated the bias toward
capital-intensive pattern of growth. Given the severe scarcity of capital, which implies a
plethora of highly profitable investment opportunities, the efficiency of investment is likely
to have been secondary to the quantity of investment as the driver of economic growth. Our
finding is largely consistent with the empirical literature on developing Asia’s growth which
finds that capital accumulation drove the region’s growth prior to 1990. According to our
evidence, this pattern of growth continued until around 2002.
Our finding, which is robust and consistent across different modes of adjusting labor
quality for human capital, suggests that there has been a major structural shift in the pattern
of developing Asia’s economic growth around 2002. More specifically, we find that total
factor productivity begins to play a much larger role in the region’s growth. One of the key
stylized facts of the modern economic growth is that as countries grow richer, TFP growth
emerges to the forefront of the growth process and factor accumulation recedes to the
background. This has been the general experience of the industrialized countries and there is
no reason why developing Asia should be an exception. The basic intuition here is that as
countries grow richer – i.e. accumulate an increasingly larger capital stock – beyond a certain
point, diminishing marginal returns to capital sets in so that additional investment becomes
less and less productive. Although most of developing Asia is nowhere near as rich as the
industrialized countries, these forces can help to explain the region’s transition from an inputbased growth to productivity-based growth. Furthermore, the transition may have started a
little earlier in the region in light of the region’s heavily capital-intensive growth pattern. The
specific rationale for the growing relative importance of TFP growth will differ from country
to country. For example, NIEs have reached income levels which are close to those of the
industrialized countries. In the case of the ASEAN countries as well as Korea, higher
productivity may reflect the restructuring and reform in the financial and corporate sectors
28
since the Asian crisis. In the case of India, the key driver of productivity growth is likely to
be deeper integration into the world economy. Finally, China has enjoyed robust TFP growth
throughout our entire sample period and this may be the consequence of on-going reforms
toward a market economy. Further research is needed to identify the country-specific sources
of TFP growth but the size of TFP growth and its relative contribution to growth seems to be
rising for the region as a whole.
The primary policy implication which arises from our main finding – i.e. the growing
relative importance of TFP growth in developing Asia’s economic growth – is that
governments around the region should more forcefully pursue policies which foster higher
productivity. Given that center of gravity of the region’s economic growth is shifting from
factor accumulation to productivity growth in recent years and this trend is likely to gain
momentum in the future in light of the region’s fast-rising income and development level,
policies which promote the productivity of all productive factors will hold the key to
sustaining growth beyond the global crisis. Total factor productivity growth consists of two
components – technological progress (TP) and technical efficiency change (TEC). TEC refers
to narrowing the gap between potential and actual output or, equivalently, moving from inside
the production frontier toward the frontier. For example, at a micro level, better management
enables a firm to be more productive with the same level of inputs and technology. An
economy-wide example is more flexible labor markets which result in a more efficient
allocation of labor across firms and industries. On the other hand, TP refers to shifting out of
the production frontier due to technological innovation. For the more advanced countries,
technological progress will involve investment in R&D and knowledge-creating activities.
For the less advanced economies, technological progress will largely involve the adoption of
new and existing technologies created by countries closer to the world technology frontier.
There are a number of specific areas in which developing Asia’s governments can foster
29
higher productivity. For example, better transport, communication, energy and other
infrastructure improves the productivity of all firms and industries. Although some parts of
the region’s infrastructure are among the best in the world, the region’s very success is
creating new demands for more and better infrastructure. It is estimated that between 2010
and 2020, developing Asia needs to invest a staggering total of US$8 trillion in infrastructure.
Another area where effective policies are required to foster productivity growth is human
capital. While developing Asia has traditionally invested heavily in education, the region has
to do a better job of producing workers with the “right” skills needed by employers. Shortage
of skills in key areas can become a bottleneck to growth. For example, for all the success in
its world-class export-oriented IT services sector and its large and growing army of university
graduates, one of the key constraints to further growth of the sector is the shortage of workers
with the required skills. A third area relevant for speeding up developing Asia’s transition to
a productivity-led growth is financial development. While the region’s financial systems have
become noticeably stronger and more efficient since the Asian crisis, they still lag behind the
region’s dynamic, world-class manufacturing sector. Yet sustaining growth in the post-crisis
will depend more on the efficiency of investment and less on the quantity of investment,
which means that financial systems will have to do a better job of allocating capital to its
most productive uses. A fourth area for promoting productivity is international trade and,
more generally, openness. Developing Asia’s remarkable economic success in the past
closely paralleled its growing integration into the world economy. In addition to static
welfare gains based on comparative advantage, trade delivers substantial dynamic efficiency
gains. In particular, globalization forces firms and industries to raise their game to survive
competition in both domestic and foreign markets.
The shift from growth based on inputs and capital accumulation to growth based on
productivity does not diminish the importance of investment in medium- and long-run growth.
30
For policymakers, this means that the priority on fostering productivity growth does not
obscure the need to create a conducive climate for private investment. Indeed dynamic
productivity gains are bound to be limited in a business environment in which firms and
entrepreneurs actively cannot seek out and invest in profitable investment opportunities.
However, given that developing Asia is no longer a poor, capital-scarce region with
intrinsically high marginal returns to capital and an abundance of profitable investment
opportunities, the priority for policymakers must be to improve the efficiency of investment
rather than to augment the quantity of investment. Indeed policies which seek to boost the
quantity of investment when investment rates are more or less at the “right” levels or above
“right” levels are not only ineffective for promoting growth but downright harmful. Overinvestment can have severe negative consequences, as the Asian crisis painfully illustrated.
Policies must concentrate instead on enabling productive investment by the most efficient
firms and industries, including new firms and new industries. Broader, deeper, more liquid
and more sophisticated financial markets will be the key in this connection. Improving the
efficiency of investment also reduces the amount of consumption that must be foregone – i.e.
savings – to finance a given level of investment and hence growth. In addition to enabling
private investment, the government should play a more active role in particular types of
investments. In particular, public-private partnerships can help to finance and meet the huge
demand for infrastructure in the coming years. The government should also play the role of
catalyst in green investments which promote environmentally sustainable growth in light of
underdeveloped markets for environmental goods.
At a broader level, our empirical evidence strongly re-confirms the relevance of supply-side
factors for developing Asia’s medium- and long-term growth. The explanatory supply-side
variables drawn from the standard empirical literature do a reasonably good job of explaining
growth in a panel of 125 countries during 1992-2007. Much of the results of regressing the
31
growth of output per worker and total factor productivity growth are sensible and consistent
with economic intuition. For example, the growth of capital stock per worker had a positive
effect on the growth of output per worker. We also find evidence of conditional convergence
– i.e. poorer countries tend to grow faster than richer countries if we control for the other
variables which help explain growth – for both output and TFP growth. Interestingly and
significantly, we find that human capital and economic openness, both of which were
significant, played a bigger role in the growth of developing Asian countries than in other
countries. This makes sense in light of the region’s traditional high priority on investment and
outward-looking export-oriented growth strategy. The overarching implication for
policymakers is that supply-side policies which augment productive capacity will be vital for
sustaining developing Asia’s future growth in the post-crisis period. While aggregate demand
policies – i.e. fiscal and monetary stimulus packages – contributed to the region’s faster- and
stronger-than-expected rebound from the global crisis and thus laid the foundation for growth,
it is time that policymakers accord a higher priority to the supply side.
The region-wide analysis of developing Asia’s past growth drivers that we provide in this
chapter is useful in that it gives us a general sense of the future pattern for the region as a
whole as well as the kinds of policies that will promote sustained growth for the region as a
whole. Nevertheless, developing Asia is a heterogeneous region encompassing countries with
a wide range of income levels and diverse structural characteristics. For example, with
respect to investment, the growing primacy of efficiency over quantity is more relevant for
the higher income countries. Lower-income countries such as Viet Nam need to boost their
investment and build up their capital stock. A big difference between the two Asian giants is
that China has a relatively well-developed infrastructure whereas a large infrastructure deficit
is a major bottleneck to India’s achieving even higher growth. In terms of technological
progress, R&D and knowledge creation is more applicable to more developed countries with
32
larger pools of scientists and engineers. Predictably, the critical constraints to growth differ
from country to country. According to ADB’s growth diagnostic study series, the critical
constraints to Indonesia’s growth are inadequate and inefficient infrastructure, weaknesses in
governance and institutions, and unequal access to and poor quality of education. In the case
of Nepal, the same series identifies the critical constraints as weak governance and slow
recovery from civil conflict, inadequate infrastructure, poor industrial relations and labor
market rigidities, and inability to address market failures. For Philippines, the most binding
constraints to growth are tight fiscal situation due to weak revenue generation, inadequate
infrastructure, weak investor confidence due to governance concerns, and small and narrow
industrial base due to inability to address market failures.
Our evidence indicates that good governance and institutions matter for both economic
growth and TFP growth. In particular, government effectiveness and control of corruption has
a significant positive impact. Our evidence also suggests that governance has a bigger effect
on developing countries than industrialized countries. The government tends to play a larger
role in the allocation of resources in developing countries since they tend to have weaker
institutions and less developed markets. Furthermore, governments in industrialized countries
have stronger institutional capacity for revenue generation and public spending. Relatively
competent and honest governments which efficiently deliver basic public services such as
administration, education and health care which raise the productivity of all firms and
industries in the economy. Such governments are also more conducive for political stability
and a more benign overall investment climate. Furthermore, governments are often directly
involved in investment, especially infrastructure such as transport, communication and energy.
Corruption is a symptom of weak and unstable governments, and acts as an unpredictable and
high tax on investment and economic activity. Increasingly, an important dimension of strong
governance and institutions will be the capacity to deliver inclusive growth which spreads the
33
fruits of growth to as much of the population as possible. As the success of conditional cash
transfers shows, well-targeted inclusiveness-promoting programs need not be fiscally costly
and can make a big dent on poverty. By promoting social and political stability, inclusive
growth can foster a more conducive social and political environment for growth.
Regional cooperation and integration has been progressing in developing Asia, especially
among East and Southeast Asian countries, but the global crisis has added a sense of urgency
to the process. The post-crisis world is likely to witness a further shift of global economic
power from industrialized countries to developing countries, in particular fast-growing
developing Asia. The rising income levels and purchasing power of the region, combined
with the relative decline of traditional export markets in the industrialized countries, point to
the growing relative importance of intra-regional trade in the post-crisis period. Rebalancing
toward domestic demand holds the key to addressing the negative impact of the prospective
weakening of demand from the industrialized countries. An important complement and
consequence of stronger domestic demand is stronger intra-regional trade. ADB (2009) points
out that while tariffs and non-tariff barriers to trade have come down sharply in the region as
a result of extensive liberalization, a host of border and behind-the-border obstacles to trade
still prevent regional countries from exploiting the potentially large gains from intra-regional
trade. In addition, regional integration which moves the region or sub-regions toward a truly
single market will augment the dynamic efficiency gains from increased competition.
Regional integration can not only benefit the region’s goods markets but also its financial
markets. In particular, greater integration of the region’s relatively underdeveloped bond
markets can help to create bigger, deeper and broader bond markets which will serve as a
secure and stable source of long-term capital. Infrastructure development will also benefit
from regional cooperation. More specifically, Asia needs to invest about US$290 billion on
regional infrastructure projects which would enhance pan-Asian connectivity in transport,
34
communications and energy networks.
Developing Asia has recovered from the global crisis with remarkable speed and vigor. In
contrast to the Asian crisis, when developing Asia relied on exports to markets outside the
region to regain its footing, this time around the region is leading the world out of the
recession. In light of its robust recovery as well as its track record of superior growth
performance on a sustained basis prior to the crisis, it might be tempting for Asian
policymakers to believe that the growth model which served the region well before the crisis
will continue to do so after the crisis. Certainly, Asia got many of the fundamentals “right”
for an extended period of time – e.g. macroeconomic stability, high degree of integration into
the world economy, and so forth – and the crisis has done nothing to invalidate these timeless
ingredients of the recipe for growth. However, now is a good time to take stock of Asia’s
medium- and long-term growth and policies for growth for a couple of reasons. First, the
post-crisis world is likely to present a less benign global environment and, at the same time,
throw up a number of structural long-term challenges, in particular aging. Second, and much
more fundamentally, some of the ingredients which were appropriate for yesterday’s Asia
will no longer be appropriate for today’s Asia, because the region’s very success has
transformed it from a stagnant low-income region to a dynamic middle-income region. In
particular, for the region as a whole, the source of growth will shift further from factor
accumulation to TFP growth, a trend that has already begun according to our empirical
analysis. Therefore, supply-side policies which augment productive capacity, in particular
polices targeted at areas which promote productivity growth - e.g. infrastructure, human
capital, financial development, trade, hold the key to sustaining the region’s medium- and
long-term growth and allow the region to make further progress on reducing poverty and
spreading the benefits of growth to wider swathes of population.
35
References
Barro, R. and J.-W. Lee. 2010. “A New Data Set of Educational Attainment in the World,
1950–2010.” NBER Working Paper No. 15902. National Bureau of Economic Research.
Massachusetts.
Benhabib, J. and M. Spiegel (1994) “The role of human capital in economic development:
Evidence from aggregate cross-country data,” Journal of Monetary Economics 34, 143173.
Bils, M. and P. Klenow (2000). “Does schooling cause growth?” American Economic
Review 90(5), 1160-1183.
Bosworth, B. and S. Collins. 2003. “The Empirics of Growth: An Update.” Brookings Papers
on Economic Activity (34)2: 113–206.
Collins, S. and Bosworth, B. P. (1997). “Economic growth in East Asia: Accumulation versus
assimilation,” NBER Macroeconomic Annual, 135-203.
Dinopoulos, E and P. Thompson (2000). “Enodgenous growth in a cross-section of countries,
Journal of International Economics 51(2), 335-362.
Fischer, S. (1993). “The Role of Macroeconomic Factors in Growth,” Journal of Monetary
Economics 32, 485-512.
Gallup, John L. and Jeffrey D. Sachs (1998) “The Economic Burden of Malaria,”
Unpublished paper, Center for International Development, Harvard University, October.
Heston, A., R. Summers, and B. Aten. 2009. Penn World Table Version 6.3. Center for
International Comparisons of Production, Income, and Prices at the University of
Pennsylvania, August.
Lee, Jong-Wha (2010), “Computing Capital Stock Estimates,” unpublished manuscript.
Masters, William A., and Margaret S. McMillan (2001). “Climate and Scale in Economic
Growth.” Journal of Economic Growth 6(3), 167-186.
Sachs, J., S. Radelet, and J.-W. Lee. 2001. “Determinants and Prospects of Economic Growth
in Asia.” International Economic Journal 15(3): 1-29.
World Bank (1993), The East Asian Miracle: Economic Growth and Public Policy. Oxford:
Oxford University Press.
Young, A. (1994). “Lessons from the East Asian NICs: A contrarian view,” European
Economic Review 38, 964-973.
Young, A. (1995). “The tyranny of numbers: confronting the statistical realities of the East
Asian growth experience,” Quarterly Journal of Economics 110, 641-680.
36
Table 1
Contribution of Capital, Labor and TFP to Output Growth, 1992-1997
Labor Quality Not Adjusted for Human Capital
non-Asian
G5
Japan
4 NIEs
China
7 DAEs
Growth in
Output
2.35%
1.26%
6.99%
9.79%
5.64%
Capital
2.50%
3.29%
8.72%
11.45%
8.04%
Labor
0.50%
0.61%
2.14%
1.17%
2.33%
Capital
1.03%
1.63%
4.42%
5.39%
5.57%
Labor
0.29%
0.31%
1.03%
0.62%
0.70%
TFP
1.04%
-0.68%
1.55%
3.78%
-0.63%
44.00%
-53.52%
22.18%
38.63%
-11.24%
Capital
1.00%
1.32%
3.49%
4.58%
3.21%
Labor
0.30%
0.37%
1.28%
0.70%
1.40%
TFP
1.06%
-0.42%
2.22%
4.51%
1.03%
44.90%
-33.20%
31.77%
46.10%
18.26%
lsh1992
60.00%
60.00%
60.00%
60.00%
60.00%
labsh1992
60.06%
49.12%
49.71%
52.33%
30.06%
C1. labor share = actual
Contribution of
(Relative contribution
of TFP)
C2. labor share=0.6
Contribution of:
(Relative contribution
of TFP)
Source: Author Estimates
37
Table 2
Contribution of Capital, Labor and TFP to Output Growth, 1997-2002
Labor Quality Not Adjusted for Human Capital
non-Asian
G5
Japan
4 NIEs
China
7 DAEs
Growth in
Output
2.58%
-0.19%
2.57%
7.69%
3.16%
Capital
3.23%
1.59%
4.95%
8.74%
3.92%
Labor
0.66%
-0.30%
1.49%
0.96%
2.46%
Capital
1.33%
0.78%
2.43%
4.10%
2.75%
Labor
0.38%
-0.15%
0.75%
0.51%
0.76%
TFP
0.86%
-0.81%
-0.61%
3.08%
-0.34%
33.38%
431.85%
-23.75%
40.10%
-10.71%
Capital
1.29%
0.64%
1.98%
3.50%
1.57%
Labor
0.39%
-0.18%
0.89%
0.58%
1.47%
TFP
0.90%
-0.65%
-0.30%
3.62%
0.12%
34.73%
344.17%
-11.74%
47.07%
3.77%
lsh1992
60.00%
60.00%
60.00%
60.00%
60.00%
labsh1992
57.89%
50.77%
50.20%
53.11%
30.68%
C1. labor share = actual
Contribution of
(Relative contribution
of TFP)
C2. labor share=0.6
Contribution of:
(Relative contribution
of TFP)
Source: Author Estimates
38
Table 3
Contribution of Capital, Labor and TFP to Output Growth, 2002-2007
Labor Not Adjusted for Human Capital
non-Asian
G5
Japan
4 NIEs
China
7 DAEs
Growth in
Output
2.32%
1.73%
5.48%
12.20%
6.58%
Capital
2.78%
1.10%
3.74%
10.63%
4.92%
Labor
0.77%
-0.07%
1.46%
0.85%
2.25%
Capital
1.16%
0.55%
1.84%
4.98%
3.44%
Labor
0.44%
-0.04%
0.72%
0.45%
0.69%
TFP
0.72%
1.22%
2.91%
6.76%
2.45%
31.06%
70.46%
53.11%
55.44%
37.25%
Capital
1.11%
0.44%
1.49%
4.25%
1.97%
Labor
0.46%
-0.04%
0.87%
0.51%
1.35%
TFP
0.74%
1.33%
3.11%
7.44%
3.26%
32.10%
77.12%
56.74%
60.96%
49.53%
lsh1992
60.00%
60.00%
60.00%
60.00%
60.00%
labsh1992
59.05%
50.97%
51.46%
53.11%
30.55%
C1. labor share = actual
Contribution of
(Relative contribution
of TFP)
C2. labor share=0.6
Contribution of:
(Relative contribution
of TFP)
Source: Author Estimates
39
Table 4
Contribution of Capital, Labor and TFP to Output Growth, 1992-1997
Labor Quality Adjusted for Human Capital
non-Asian
G5
Japan
4 NIEs
China
7 DAEs
Growth in
Output
2.35%
1.26%
6.99%
9.79%
5.64%
Capital
2.50%
3.29%
8.72%
11.45%
8.04%
Labor
0.50%
0.61%
2.14%
1.17%
2.33%
Human1
1.43%
0.96%
0.86%
2.36%
1.81%
Human2
0.88%
0.68%
0.49%
1.00%
0.64%
Capital
1.00%
1.32%
3.49%
4.58%
3.21%
Labor
0.30%
0.37%
1.28%
0.70%
1.40%
Human1
0.85%
0.48%
0.40%
1.25%
0.56%
TFP
0.20%
-0.99%
1.71%
3.10%
-0.05%
8.37%
-78.65%
24.39%
31.66%
-0.97%
Capital
1.00%
1.32%
3.49%
4.58%
3.21%
Labor
0.30%
0.37%
1.28%
0.70%
1.40%
Human2
0.53%
0.41%
0.29%
0.60%
0.38%
TFP
0.53%
-0.83%
1.93%
3.91%
0.65%
22.38%
-65.37%
27.60%
39.96%
11.46%
lsh1992
60.00%
60.00%
60.00%
60.00%
60.00%
labsh1992
60.06%
49.12%
49.71%
52.33%
30.06%
C3. linear labor-quality
adjustment
Contribution of:
(Relative contribution
of TFP)
C4. exponential laborquality adjustment
Contribution of:
(Relative contribution
of TFP)
Source: Author Estimates
40
Table 5
Contribution of Capital, Labor and TFP to Output Growth, 1997-2002
Labor Quality Adjusted for Human Capital
non-Asian
G5
Japan
4 NIEs
China
7 DAEs
Growth in
Output
2.58%
-0.19%
2.57%
7.69%
3.16%
Capital
3.23%
1.59%
4.95%
8.74%
3.92%
Labor
0.66%
-0.30%
1.49%
0.96%
2.46%
Human1
1.22%
0.65%
1.06%
1.87%
2.00%
Human2
0.80%
0.48%
0.67%
0.88%
0.71%
Capital
1.29%
0.64%
1.98%
3.50%
1.57%
Labor
0.39%
-0.18%
0.89%
0.58%
1.47%
Human1
0.71%
0.33%
0.54%
0.99%
0.61%
TFP
0.17%
-1.04%
-0.94%
2.50%
-1.08%
6.46%
550.92%
-36.54%
32.50%
-34.10%
Capital
1.29%
0.64%
1.98%
3.50%
1.57%
Labor
0.39%
-0.18%
0.89%
0.58%
1.47%
Human2
0.48%
0.29%
0.40%
0.53%
0.43%
TFP
0.41%
-0.94%
-0.71%
3.09%
-0.31%
16.06%
496.47%
-27.41%
40.19%
-9.78%
lsh1992
60.00%
60.00%
60.00%
60.00%
60.00%
labsh1992
57.89%
50.77%
50.20%
53.11%
30.68%
C3. linear labor-quality
adjustment
Contribution of:
(Relative contribution
of TFP)
C4. exponential laborquality adjustment
Contribution of:
(Relative contribution
of TFP)
Source: Author Estimates
41
Table 6
Contribution of Capital, Labor and TFP to Output Growth, 2002-2007
Labor Quality Adjusted for Human Capital
non-Asian
G5
Japan
4 NIEs
China
7 DAEs
Growth in
Output
2.32%
1.73%
5.48%
12.20%
6.58%
Capital
2.78%
1.10%
3.74%
10.63%
4.92%
Labor
0.77%
-0.07%
1.46%
0.85%
2.25%
Human1
0.69%
0.58%
1.22%
1.39%
2.13%
Human2
0.47%
0.45%
0.82%
0.72%
0.84%
Capital
1.11%
0.44%
1.49%
4.25%
1.97%
Labor
0.46%
-0.04%
0.87%
0.51%
1.35%
Human1
0.40%
0.29%
0.62%
0.74%
0.64%
TFP
0.31%
0.98%
2.35%
6.60%
1.93%
13.49%
56.63%
42.88%
54.11%
29.34%
Capital
1.11%
0.44%
1.49%
4.25%
1.97%
Labor
0.46%
-0.04%
0.87%
0.51%
1.35%
Human2
0.28%
0.27%
0.49%
0.43%
0.51%
TFP
0.45%
1.06%
2.60%
7.01%
2.74%
19.53%
61.55%
47.47%
57.45%
41.62%
lsh1992
60.00%
60.00%
60.00%
60.00%
60.00%
labsh1992
59.05%
50.97%
51.46%
53.11%
30.55%
C3. linear labor-quality
adjustment
Contribution of:
(Relative contribution
of TFP)
C4. exponential laborquality adjustment
Contribution of:
(Relative contribution
of TFP)
Source: Author Estimates
42
Table 7
Output per Worker Growth Equation
Dependent var = dln(Y/L)
Model
VARIABLES
mdkkl
lny_us
(1)
a1
(2)
a2
(3)
a3
(4)
a4
(5)
a5
(6)
a6
0.448***
(12.23)
0.010***
(-5.186)
0.428***
(11.26)
0.010***
(-5.396)
0.428***
(11.23)
0.429***
(11.08)
0.010***
(-4.231)
-0.001
(0.0634)
0.004***
(4.811)
0.002*
(1.788)
0.009***
(-2.850)
0.009***
(2.834)
0.000
(1.007)
-0.000
0.403***
(10.06)
0.404***
(10.14)
0.013***
(-5.040)
lnlifes
mhuman
0.004***
(5.030)
lnpop
mtropic
mopenc
0.010***
(-2.910)
0.005**
(2.134)
0.004***
(5.158)
0.002*
(1.855)
0.009***
(-2.836)
0.009***
(2.817)
-0.010***
(-4.749)
-0.000
(0.00995)
0.004***
(5.028)
0.002*
(1.813)
-0.009***
(-2.808)
0.009***
(2.754)
minflat_cpi
mca_gdp
Constant
Observations
Adjusted R-squared
t-statistics in parentheses
*** p<0.01, ** p<0.05, * p<0.1
-0.008**
(-2.449)
0.008**
(2.533)
-0.010***
(-3.294)
-0.081***
(-3.451)
-0.082***
(-3.464)
315
0.453
315
0.451
315
0.459
315
0.459
-0.004
(-1.107)
0.010***
(-3.228)
0.047***
(-3.521)
-0.003
(-0.843)
0.009***
(-3.124)
0.081***
(-3.575)
-0.009***
(-3.107)
315
0.450
315
0.455
-0.003
(-0.837)
Source: Author Estimates
43
-0.008**
(-2.385)
0.008**
(2.453)
0.000
(1.228)
-0.000
(0.00117)
0.006**
(2.271)
-0.004
(-1.259)
0.004***
(4.764)
0.002*
(1.787)
-0.003
(-0.949)
0.010***
(-3.096)
0.083***
(-3.493)
(-0.158)
_Iyear_1997
(0.440)
0.004***
(4.498)
0.002
(1.590)
-0.000
(0.0113)
0.005**
(2.109)
-0.004
(-1.077)
0.010***
(-3.246)
0.083***
(-3.655)
mgoveff
_Iyear_1992
-0.014***
(-4.748)
0.005
Table 8
Output per Worker Growth Equation,
Governance Indicators Included
Dependent var = dln(Y/L)
VARIABLES
mdkkl
lny_us
mhuman
lnpop
mtropic
mopenc
mlaw
(1)
a1
(2)
a2
(3)
a3
(4)
a4
(5)
a5
(6)
a6
0.419***
(10.73)
0.011***
(-4.927)
0.004***
(5.058)
0.002*
(1.896)
0.417***
(10.64)
0.011***
(-5.255)
0.004***
(4.588)
0.002*
(1.838)
0.010***
(-2.930)
0.008***
(2.699)
0.404***
(10.25)
0.013***
(-5.559)
0.004***
(4.935)
0.002*
(1.798)
0.410***
(10.56)
0.013***
(-5.470)
0.004***
(5.015)
0.002*
(1.896)
0.400***
(10.00)
0.013***
(-5.496)
0.004***
(4.748)
0.002
(1.549)
-0.008**
(-2.456)
0.008**
(2.573)
-0.008**
(-2.249)
0.007**
(2.464)
-0.008**
(-2.285)
0.007**
(2.248)
0.386***
(9.594)
0.014***
(-5.925)
0.004***
(5.087)
0.002
(1.557)
0.009***
(-2.740)
0.005
(1.496)
0.004*
(1.912)
0.005
(0.912)
0.001
(0.151)
-0.008**
(-2.484)
0.008***
(2.715)
0.002
(0.962)
mregq
0.003
(1.129)
mgoveff
0.005**
(2.118)
mcontrolcorr
mgoveff_a
-0.003
(-0.947)
0.010***
(-3.194)
0.083***
(-3.634)
-0.003
(-0.958)
0.010***
(-3.235)
0.080***
(-3.515)
-0.004
(-1.102)
0.010***
(-3.286)
0.083***
(-3.661)
-0.003
(-0.887)
0.010***
(-3.321)
0.082***
(-3.546)
-0.003
(-0.979)
0.010***
(-3.367)
0.077***
(-3.253)
0.004
(0.720)
-0.007*
(-1.896)
-0.004
(-1.344)
0.010***
(-3.485)
0.068***
(-2.902)
315
0.455
315
0.455
315
0.461
309
0.446
309
0.446
315
0.469
mgoveff_o
_Iyear_1992
_Iyear_1997
Constant
Observations
Adjusted R-squared
t-statistics in parentheses
*** p<0.01, ** p<0.05, * p<0.1
t-statistics in parentheses
*** p<0.01, ** p<0.05, * p<0.1
0.010***
(2.605)
Source: Author Estimates
44
Table 9
TFP Growth Equation
Dependent var = dlnTFP
VARIABLES
lny_us
(1)
a1
(2)
a2
(3)
a3
(4)
A4
(5)
a5
0.010***
(-5.536)
0.014***
(-5.895)
0.004***
(5.291)
0.004***
(5.045)
0.011***
(-3.325)
0.006**
(2.346)
0.009***
(-2.849)
0.005**
(2.048)
0.015***
(-5.373)
0.005
(0.429)
0.004***
(4.860)
0.001
(1.163)
0.009***
(-2.676)
0.007**
(2.248)
0.012***
(-4.282)
-0.009
(-0.693)
0.004***
(4.975)
0.002
(1.603)
0.009***
(-2.662)
0.008**
(2.537)
0.000
(0.829)
-0.000
(-0.237)
0.006**
(2.414)
0.006**
(2.288)
0.014***
(-4.882)
0.005
(0.396)
0.004***
(4.645)
0.001
(1.142)
0.009***
(-2.703)
0.007**
(2.338)
0.000
(1.011)
-0.000
(-0.207)
0.006**
(2.342)
lnlifes
mhuman
lnpop
mtropic
mopenc
minflat_cpi
mca_gdp
mgoveff
mcontrolcorr
_Iyear_1992
_Iyear_1997
Constant
Observations
Adjusted R-squared
t-statistics in parentheses
*** p<0.01, ** p<0.05, * p<0.1
-0.004
(-1.180)
0.010***
(-3.207)
0.055***
(-4.078)
-0.004
(-1.354)
0.010***
(-3.272)
0.059***
(-4.335)
-0.004
(-1.127)
0.009***
(-3.136)
0.080***
(-3.392)
-0.004
(-1.258)
0.010***
(-3.152)
0.081***
(-3.430)
0.004**
(1.994)
-0.003
(-0.905)
0.009***
(-3.115)
0.087***
(-3.696)
315
0.186
315
0.199
315
0.198
315
0.196
309
0.183
Source: Author Estimates
45
Table 10
TFP Growth Equation,
Differential Impact on OECD and 12 Asian Countries
Dependent var = dlnTFP
VARIABLES
lny_us
mhuman
(1)
a1
(2)
A2
(3)
a3
(4)
a4
(5)
a5
0.014***
(-5.895)
0.004***
(5.045)
0.014***
(-6.071)
0.004***
(5.267)
0.015***
(-6.198)
0.004***
(5.219)
0.009***
(-2.849)
0.005**
(2.048)
0.014***
(-6.036)
0.004***
(5.141)
0.001*
(1.822)
-0.000
(-0.592)
0.011***
(-3.235)
0.004
(1.400)
0.011***
(-3.120)
0.003
(0.979)
0.006**
(2.414)
0.006**
(2.249)
0.011***
(-3.281)
0.004
(1.444)
0.002**
(2.299)
-0.000
(-0.152)
0.005**
(1.986)
-0.004
(-1.354)
0.010***
(-3.272)
0.059***
(-4.335)
-0.004
(-1.399)
0.010***
(-3.325)
0.053***
(-3.824)
-0.004
(-1.370)
0.010***
(-3.323)
0.054***
(-3.951)
0.009***
(2.716)
0.004
(0.816)
-0.006*
(-1.817)
-0.005
(-1.473)
0.010***
(-3.366)
0.048***
(-3.401)
0.015***
(-6.432)
0.005***
(5.217)
-0.003
(-1.205)
-0.001
(-0.743)
0.011***
(-2.992)
0.001
(0.295)
0.006
(1.640)
0.004
(1.193)
0.008**
(2.284)
0.005
(0.833)
-0.007
(-1.415)
-0.005
(-1.446)
0.010***
(-3.413)
0.047***
(-3.300)
315
0.199
315
0.206
315
0.209
315
0.210
315
0.215
mhuman_a
mhuman_o
mtropic
mopenc
mopenc_a
mopenc_o
mgoveff
mgoveff_a
mgoveff_o
_Iyear_1992
_Iyear_1997
Constant
Observations
Adjusted R-squared
t-statistics in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Source: Author Estimates
46
Table 11
TFP Growth Equation,
R&D Variable Included
Dependent var = dlnTFP
VARIABLES
lny_us
mhuman
(1)
a1
(2)
a2
(3)
a3
(4)
a4
(5)
a5
0.009***
(-4.289)
0.003***
(3.608)
0.012***
(-4.844)
0.004***
(4.152)
0.013***
(-3.903)
0.004***
(4.087)
0.001
(0.826)
0.013***
(-4.090)
0.004***
(3.937)
0.001
(0.584)
-0.009**
(-2.481)
0.008***
(2.966)
0.010***
(-2.820)
0.007**
(2.557)
0.011***
(-4.541)
0.004***
(4.216)
0.001
(0.863)
0.009***
(-2.620)
0.008**
(2.531)
0.000
(1.088)
0.001*
(1.839)
-0.008**
(-2.217)
0.008**
(2.381)
0.000
(1.274)
0.001*
(1.683)
0.003
(0.839)
-0.008**
(-2.293)
0.008**
(2.365)
0.000
(1.136)
0.001*
(1.772)
lnpop
mtropic
mopenc
minflat_cpi
mca_gdp
0.001**
(2.173)
mgoveff
mcontrolcorr
mdrk_l2
0.057*
(1.930)
_Iyear_1992
-0.003
(-0.710)
0.010***
(-3.265)
0.058***
(-3.935)
_Iyear_1997
Constant
Observations
Adjusted R-squared
t-statistics in parentheses
*** p<0.01, ** p<0.05, * p<0.1
176
0.239
0.061**
(2.077)
-0.001
(-0.280)
0.009***
(-2.978)
0.062**
(2.088)
0.059**
(1.997)
0.002
(0.673)
0.048
(1.609)
0.062***
(-4.245)
-0.001
(-0.352)
0.010***
(-3.035)
0.080***
(-3.082)
-0.002
(-0.529)
0.010***
(-3.084)
0.082***
(-3.129)
-0.001
(-0.334)
0.010***
(-3.166)
0.075***
(-2.805)
176
0.255
176
0.255
176
0.253
175
0.239
Source: Author Estimates
47
Figure 12
Relative Contribution of Explanatory Variables to Growth in GDP Per Worker
country
OECD
4 NIEs
China
4
ASEAN
3 DAEs
Other
Developing
Economies
19921997
growth in GDP per worker
predicted growth in GDP per
worker
predicted growth gap in GDP
per worker (gap from the
global average)
1.86%
4.85%
8.63%
3.56%
2.99%
0.21%
2.15%
4.05%
7.02%
3.52%
3.16%
0.92%
0.65%
2.55%
5.52%
2.02%
1.66%
-0.58%
catch-up effect
-2.16%
-1.83%
1.08%
-0.06%
1.16%
0.41%
log of population
0.58%
0.18%
1.08%
0.50%
0.73%
-0.08%
human capital effect
1.02%
0.82%
-0.23%
-0.12%
-1.13%
-0.28%
geographical effect
0.39%
-0.08%
0.37%
-0.41%
-0.01%
-0.08%
openness effect
-0.46%
0.56%
-0.40%
0.15%
-0.52%
0.02%
government effectiveness
1.06%
0.86%
0.07%
0.18%
-0.16%
-0.20%
growth in GDP per worker
predicted growth in GDP per
worker
predicted growth gap in GDP
per worker (gap from the
global average)
1.92%
1.08%
6.73%
-0.33%
2.09%
1.14%
2.05%
1.99%
5.02%
0.46%
2.10%
0.60%
1.02%
0.96%
3.99%
-0.57%
1.07%
-0.43%
catch-up effect
-2.17%
-2.06%
0.57%
-0.25%
1.03%
0.40%
log of population
0.57%
0.18%
1.08%
0.50%
0.73%
-0.08%
human capital effect
1.13%
0.81%
-0.10%
-0.11%
-1.10%
-0.29%
geographical effect
0.39%
-0.08%
0.37%
-0.41%
-0.01%
-0.08%
openness effect
-0.40%
0.57%
-0.51%
0.31%
-0.46%
0.00%
government effectiveness
0.92%
0.60%
-0.08%
-0.01%
-0.23%
-0.17%
growth in GDP per worker
1.55%
4.02%
11.35%
3.90%
4.90%
2.47%
3.01%
3.16%
7.20%
2.41%
4.15%
2.61%
0.19%
0.33%
4.37%
-0.41%
1.32%
-0.21%
catch-up effect
-2.20%
-2.02%
0.25%
-0.14%
0.98%
0.41%
log of population
0.56%
0.17%
1.07%
0.51%
0.74%
-0.08%
19972002
20022007
predicted growth in GDP per
worker
predicted growth gap in GDP
per worker (gap from the
global average)
48
human capital effect
1.14%
0.88%
-0.05%
-0.05%
-1.00%
-0.29%
geographical effect
0.39%
-0.08%
0.37%
-0.41%
-0.01%
-0.08%
openness effect
-0.42%
0.66%
-0.23%
0.23%
-0.28%
-0.01%
government effectiveness
0.88%
0.78%
0.01%
0.09%
-0.17%
-0.18%
Source: Author Estimates
49
Table 13
Comparison of the Relative Contribution of Explanatory Variables to Growth in GDP
Per Capita Across Different Groups of Countries
1992-1997:
Relative income level
OECD < 4 NIEs < 4 ASEAN < Other developing economies < China < 3 ADEs.
Human capital
OECD> 4 NIEs > 4 ASEAN > China > Other developing economies > 3 ADEs.
Government effectiveness
OECD> 4 NIEs > 4 ASEAN > China > 3 ADEs > Other developing economies.
Openness
4 NIEs > 4 ASEAN > Other developing economies > China > OECD > 3 ADEs
2002-2007:
Relative income level
OECD < 4 NIEs < 4 ASEAN < China < Other developing economies < 3 ADEs.
Human capital
OECD> 4 NIEs > 4 ASEAN > China > Other developing economies > 3 ADEs.
Government effectiveness
OECD> 4 NIEs > 4 ASEAN > China > 3 ADEs > Other developing economies.
Openness
4 NIEs > 4 ASEAN > Other developing economies > China > 3 ADEs > OECD.
Note: Change in relative rankings from the 1992-1997 period are colored red.
Source: Table 12
50
Table 14
Relative Contribution of Explanatory Variables to TFP Growth
country
OECD
4 NIEs
China
4 ASEAN
3 ADEs
Other
Developing
Economies
19921997
growth in TFP
0.43%
1.88%
3.80%
0.71%
0.39%
-0.81%
predicted growth in TFP
0.75%
0.77%
1.77%
0.26%
0.14%
0.30%
predicted growth gap in
TFP (gap from the global
average)
0.25%
0.27%
1.27%
-0.25%
-0.36%
-0.20%
catch-up effect
-2.27%
-1.92%
1.13%
-0.06%
1.22%
0.43%
human capital effect
1.06%
0.85%
-0.24%
-0.12%
-1.17%
-0.29%
geographical effect
0.46%
-0.10%
0.43%
-0.49%
-0.01%
-0.09%
openness effect
-0.31%
0.37%
-0.27%
0.10%
-0.35%
0.01%
government effectiveness
1.17%
0.94%
0.08%
0.20%
-0.18%
-0.22%
growth in TFP
0.33%
-0.78%
2.99%
-0.65%
-0.04%
0.15%
predicted growth in TFP
predicted growth gap in
TFP (gap from the global
average)
0.46%
-0.02%
0.85%
-0.30%
-0.27%
0.22%
0.15%
-0.32%
0.55%
-0.60%
-0.57%
-0.08%
catch-up effect
-2.28%
-2.17%
0.60%
-0.26%
1.09%
0.42%
human capital effect
1.17%
0.84%
-0.11%
-0.11%
-1.14%
-0.30%
geographical effect
0.46%
-0.10%
0.43%
-0.49%
-0.01%
-0.09%
openness effect
-0.27%
0.38%
-0.34%
0.21%
-0.31%
0.00%
government effectiveness
1.01%
0.67%
-0.09%
-0.01%
-0.25%
-0.19%
growth in TFP
0.40%
2.51%
6.93%
2.96%
2.22%
0.53%
predicted growth in TFP
predicted growth gap in
TFP (gap from the global
average)
1.62%
1.59%
2.09%
1.16%
1.20%
1.46%
0.07%
0.04%
0.54%
-0.39%
-0.35%
-0.09%
catch-up effect
-2.31%
-2.12%
0.26%
-0.15%
1.03%
0.43%
human capital effect
1.18%
0.91%
-0.05%
-0.05%
-1.04%
-0.30%
19972002
20022007
51
geographical effect
0.46%
-0.10%
0.44%
-0.48%
-0.01%
-0.09%
openness effect
-0.28%
0.44%
-0.15%
0.15%
-0.18%
0.00%
government effectiveness
0.97%
0.86%
0.01%
0.10%
-0.19%
-0.20%
Source: Author Estimates
52
Appendix 1
List of Countries
Albania
Argentina
Armenia
Australia
Austria
Bahrain
Bangladesh Barbados
Belgium
Belize Benin Bolivia
Botswana
Brazil
Bulgaria
Burundi
Cambodia
Cameroon
Canada
Central African Republic
Chile China Colombia
Congo, Republic of Costa
RicaCote d`Ivoire
Cyprus Czech Republic
Denmark
Ecuador
Egypt El
Salvador
Fiji
Finland
France Gabon Gambia
Germany
Ghana
Greece
Guatemala
Guyana
Haiti Honduras
Hong Kong Hungary
Iceland
India Indonesia
Iran Ireland
Israel Italy Jamaica
Japan Jordan Kenya Korea
Kuwait
Kyrgyzstan Laos Latvia Lesotho
Libya Lithuania
Luxembourg Macao Malawi
Malaysia
Maldives
Mali Malta Mauritania
Mauritius
Mexico
Mongolia
Morocco
Mozambique Namibia
Nepal
Netherlands New Zealand Nicaragua
Niger Norway
Pakistan
Panama
Papua New Guinea Paraguay
Peru Philippines Poland Portugal
Romania
Rwanda
Saudi Arabia Senegal
Sierra Leone Singapore
Slovak Republic
Slovenia
South Africa Spain Sri Lanka
Sudan Swaziland
Sweden
Switzerland Syria Tanzania
Thailand
Togo Tonga Trinidad &Tobago
Tunisia
Turkey Uganda
United Kingdom
United States Uruguay
Venezuela
Vietnam
Yemen Zambia
Zimbabwe
53
Appendix 2
Definitions of Variables and their Data Sources
Dependent Variables
dlnTFP
dln(Y/L)
Independent Variables
dln(K/L)
lny_us
lnlifes
mhuman
lnpop
mtropic
mopenc
minflat_cpi
mca_gdp
mdrk_l2
mlaw
Definitions
Five-year average growth rate of TFP.
TFP growth are derived from the above
equation (3)
Five-year average growth rate of GDP per
worker (Y/L)
Definitions
Five-year average growth rate of capital
stock per worker(K/L). Capital stock (K)
series are Lee(2010)’s estimates based
on the PWT investment series.
Log of the per capita GDP relative to that
of the U.S. in the initial year of each
respective 5-year interval,
Log of initial life expectancy relative to
that of the U.S. in the initial year of each
respective 5-year interval,
Sources
5-year averages of educational attainment
level (human) from Barro-Lee data set
Log of population in the initial year of each
respective 5-year interval,
Percentage tropical area
Barro-Lee
Log of openness index from PWT
Five-year average inflation rate (CPI)
Five-year average of current account
deficit relative to GDP
Five-year average growth rate of R&D
stock per worker. Real R&D stocks were
constructed on the basis of R&D to GDP
ratios.
_lyear_1992
Rule of law (from World Bank, World
Governance Indicator)
Government effectiveness (from World
Bank, World Governance Indicator)
Control of corruption (from World Bank,
World Governance Indicator)
Regulatory quality (from World Bank,
World Governance Indicator)
Dummy variable for the 1992-1997 period
_lyear_1997
Dummy variable for the 1997-2002 period
mgoveff
mcontrolcorr
mregq
PWT
Sources
PWT
PWT
WDI
PWT
Gallup
and
Sachs (1998)
PWT
WDI
WDI
WDI
and
OECD MSTI
(Main
Science and
Technology
Indicators)
WGI
WGI
WGI
WGI
Note: PWT is Penn World Tables version 6.3, WDI is World Development Indicators, and WGI is World
Governance Indicators from World Bank.
54
Appendix 2 (continued)
Summary Statistics of Variables
Variable
Obs
Mean
Std. Dev.
Min
Max
mdyyl
350
0.017
0.028
-0.125
0.122
mdtfp2
350
0.007
0.024
-0.113
0.098
mdkkl
350
0.015
0.033
-0.139
0.152
lny_us
350
-1.737
1.126
-4.008
0.466
lnlifes
350
-0.143
0.177
-1.163
0.055
mhuman
350
7.260
2.768
0.629
12.796
lnpop
350
9.127
1.696
4.523
14.066
mtropic
315
0.495
0.476
0.000
1.000
mopenc
350
84.079
52.350
16.936
418.998
minflat_cpi
350
13.067
56.667
-1.756
1007.457
mca_gdp
350
-1.729
6.954
-27.920
34.857
mdrk_l2
214
0.539
0.050
-0.108
0.208
mlaw
349
0.146
0.989
-1.850
2.058
mgoveff
349
0.225
1.007
-1.592
2.636
mcontrolcorr
339
0.177
1.049
-1.757
2.468
mregq
350
0.233
0.862
-2.104
1.992
55