GROWTH POUCY 55

Getting interventions right:

how South Korea and Taiwan
grew rich

Dani Rodrik
Columbia University

1. INTRODUCTION

To any economist interested in growth, the East Asian experience since the early
1960s poses enduring challenges. In 1960, South Korea was poorer than many
sub-Saharan African countries, and Taiwan not all that much richer (Table 1).
Since then, these two countries have experienced average increases in per-capita
income of 6.8% and 6.2% respectively, with the result that they have left far
behind not only these African countries, but also others like Mexico and Argentina
which had been much richer. How these two countries managed to transform
themselves from economic basket cases into economic powerhouses remains
something of an enigma.
The standard story to which most orthodox economists subscribe is one of
export-led growth (see, for example, Tsiang, 1984; Kreuger, 1985; World Bank,
1993; little, 1994). During the 1950s, the story goes, both of these countries
engaged in traditional import substitution policies, with multiple exchange rates,
high levels of trade protection, and repressed financial markets. By the late 1950s,
each country had exhausted the 'easy stage' of import substitution. This, together
with the impending reduction in US aid - which had been the main source of

I am grateful tojagdish Bhagwati, Paul de Grauwe, Ann Harriion, Aivind Panagariya, A n d i ^ Rodriguez-Clare,
Robert Wade, Adrian Wood, Ahvyn Young, Panel memben and especially Gene Grouman and Charles Wyplosz
for helpful comments, Eytsung Kim for excellent research assistance, and the CEPR MIRAGE project for
financial assistance.
56 DANIRODWK

Table 1. CoIl^>arative growth experience

Country Per-capita GDP, Per-capita GDP, Per-capita GDP,

I960 1989 growth.
(1985 doUars) (1985 dollars) 1960-89 (%)

South Korea 883 6206 6.82

Taiwan 1359 8207 6.17

Ghana 873 815 -0.54

Senegal 1017 1082 0.16
Mozambique 1128 756 -2.29
Brazil 1745 4138 3.58
Mexico 2798 5163 2.36
Argentina 3294 3608 0.63

Source: Penn World Table 5.5.

foreign exchange for both economies - led policy-makers in the two countries to
alter their economic strategy and adopt export-oriented policies. These policies
included the unification of exchange rates accompanied by devaluations, various
other measures to stimulate exports (including most significantly duty-firee access
for exporters to imported inputs), higher interest rates, and some liberalization of
the import regime. As a consequence of these measures, as well as a broadly
supportive policy environment (encompassing macroeconomic stability and public
investment in infrastructure and in human capital), exports took off in the mid-
1960s. Export orientation led both economies to specialize according to
comparative advantage, resulting in rising incomes, investment, savings and
productivity.
This orthodox account has been criticized for downplaying the active role of
governments in Taiwan and South Korea in shaping the allocation of resources.
Observers like Amsden (1989) and Wade (1990) have argued that the reforms of
the 1960s went considerably beyond giving markets and comparative advantage
free rein. According to these authors, governments in both countries had dear
industrial priorities and they did not hesitate to intervene (through subsidies, trade
restrictions, administrative guidance, public enterprises or credit aUocation) to
reshape comparative advantage in the desired direction. Interestin^y, however,
the orthodox and revisionist accounts converge on the importance of the e:qx)rt-
oriented strategy in having disciplined firms and enhanced productivity growth.
The World Bank's detailed recent study. The East Asian Miracle (1993), has
attempted to incorporate some of the revisionist objections (particularly on the role
of directed credit) into the standard account.
I will argue in this paper that the standard story, as sketched above, is
incomplete and quite misleading on the importance it attaches to the role of ejqwrt
orientation in the growth performjuice. It also has backward the causal
GROWTH POUCY 57

relationship between exports, on the one hand, and investment and growth on the
other. As I will show, the increase in the relative profitability of exports iiround the
mid-1960s was modest in both countries, and can account fully neither for the
initial jump in the export-GDP ratio at that time nor for the subsequent steady
increase in this ratio.
A much more plausible explanation for the economic take-off is the sharp
increase in investment demand that took place in the early 1960s. The reason for
this investment boom is the key issue addressed in this paper. I will argue that in
the early 1960s and thereafter the Korean and Taiwanese governments managed
to engineer a significant increase in the private return to capital. They did so not
only by removing a number of impediments to investment and establishing a
sound investment climate, but more importantly by alleviating a coordination
failure which had blocked economic take-off. The latter required a range of
strategic interventions - including investment subsidies, administrative guidance
and the use of public enterprise - which went considerably beyond those discussed
in the standard account. That government intervention could play such a
productive role was conditioned in turn by a set of advantageous initial conditions:
namely, a favourable human capital endowment and relatively equal distribution
of income and wealth.
I wall elaborate on these arguments below. It is useful to set the stage first by
reviewing some of the key elements of the Taiwanese and Korean miracles (section
2). Next, I discuss the shortcomings of the export-based explanations of these
miracles (section 3). I then turn to some of the distinctive initial conditions -
relative abundance of human capital and equitable income and wealth distribution
- which appear to have played a role in both countries' economic performance
(section 4). Section 5 lays out the paper's central arguments on coordination failure
and the governments' role in removing it. Section 6 discusses the investment-
stimulating policies followed by the two governments in light of the preceding
analytical framework. Section 7 asks how it became possible for detailed
interventions to be carried out efficiently and with little rent seeking. In section
8,1 discuss a number of objections to the arguments. Section 9 closes the paper by
offering some concluding remarks. The formal model providing the foimdation for
the centrail argument is presented in the appendix.

2. THE CONTOURS OF A MIRACLE

We begin by reviewing some of the key facts about the two countries' economic
performance over the last three decades. Figure 1 shows their spectacular growth
performance since the early 1960s. We note that economic growth has fluctuated
widely around a high mean. Both economies were particularly hard hit by the two
oil shocks of the 1970s, but in each case output recovered remarkably quickly.
58 DANI RODRK

-Korea- •Taiwan

-2 I I I I I I I I I I I I I I I I I I- I I I I I I I I I I I I I I I I I I I
1954 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90

Figure 1. Per^capita growth rates (three-year moving averages), 1954-90

Source: Penn World Table 5.5.

Figure 2 is the chief exhibit for the export-led growth hypothesis. The
export-GDP ratio rose from virtually zero in Korea to more than 30% by the
early 1980s, and from around 10% in Taiwan to over 40%. In both countries, the
increase in export orientation was particularly rapid in the decadefi-omthe mid-
1960s to the mid-1970s, and hais abated somewhat since then.

• Korea • Taiwan |
I:
60

50

40

%30

20

10

n I'I •I I I I I I I I I I I I I I I I I I I I I I I I
1952 5 4 5 6 5 8 6 0 62 6 4 6 6 6 8 70 72 74 76 7 8 8 0 8 2 8 4 8 6 8 8 9 0
Figure 2. Export/GDP ratio*, 1952-90
Sources: Council for Economic Planning and Development, Taiiuan Slatislical Data Book, 1982 and
1992; Economic Planning Board, Mi^ StadsUcs of Urn Konan Eeoiumy, various issues; IMF, JnUnttUioiiat
Fbumdal SlaHstics.
GROWTH POUCY 59

-Korea •Taiwan I

40

35 A /

30

25

20
•S/
15
• • -
/
10

0 • I I I I I I Mil M i l l 1 1 1 1 1 1 1M i l l 1 M 1 M 1 1 1 1
1951 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89
Figure 3. Investment/GDP ratios, 1951-90
Source: Perm World Table 5.5.

Less discussed but certainly more important is the spectacular increase in the
investment effort, shown in Figure 3. Investment rose from around 10% of GDP in
the late 1950s in both countries to 30% in 1980. Since 1980, investment has
continued its upward trend in Korea, but has declined somewhat in Tziiwan. This
investment effort has been matched by a roughly equivalent increase in savings.

Table 2. Comparative productivity growth statistics

Country Period Total factor productivity growth (% per year)

Economy Nfanufacturing
South Korea 196&-90 1.2* 2.7
Taiwan 1966-90 1.8 1.4

Argentina 1940-80 1.0 n.a.

Brazil 1950-80 2.0 n.a.
1960-80 n.a. 1.0
Chile 1940-60 1.2 n.a.
Colombia 1940-80 0.9 n.a.
Mexico 1940-80 1.7 n.a.
1940-70 n.a. 1.3
Venezuela 1950-70 n.a. 2.6

* Non-agricultural economy.
Source: Young (1994). Latin American statistics are originally from Elias (1990).
60 DANI RODRK

Consequently, the net resource transfer from abroad has been either small
(Taiwan) or moderate but manageable (Korea).
Finally, to round out our discussion of the main outlines of the Korean and
Taiwanese experience, productivity performance has been respectable in both
countries, but hardly spectacular. Table 2 shows the results of Young's (1994)
careful calculations of changes in total factor productivity (TFP) for Korea and
Taiwan, along with similar numbers for Latin American countries. The East Asian
TFP figures do not stand out in comparison with those for Latin American
countries. A paper by Kim and Lau (1992), based on an econometric estimation of
a 'meta-production function' across countries, presents an even more dramatic
finding: the rate of technical progress in South Korea and Taiwan has been
essentially nil.' Surprising as these results may seem, they reflect the simple fact
that once the phenomenal rate of factor accumulation (primarily in capital) is
taken into account, there is very litde growth 'residual' left over to explain. The
inescapable conclusion is that the proximate determinant of the East Asian miracle
is capital accumulation rather than an increase in industrial factor productivity.^

3. WHAT IS WRONG WITH THE EXPORT-LED GROWTH HYPOTHESIS?

As pointed out in the introduction, the standard account gives priority to the role
of export orientation in explaining the economic performance summarized in the
previous section. A particularly clear statement comes from Ian Litde (1994):
the outstanding success of Korea 2uid Taiwan from the early 1960s to the mid-
1970s was based on a phenomenal growth of labour-intensive manufactures.
This branch of manufacturing took off because exports were highly profitable
once the bias against manufacturing for export was removed. The high
profitability also depended on a relatively well-educated hard working docile
labour force which was, apart from the natural rate of increase, fed by a large
movement out of agriculture High profits and increased earnings for
recruits to the industrial labour force led to a very rapid rise in savings. There
was thus a virtuous circle.
Upon a closer look, however, this account is not quite convincing for a number of
reasons discussed below.

' The World Bank (1993) study mentioned above reports high TFP growth in these countries, but its analysis has
been seriously challenged by Young (personal communication) and little (1994).
' This sutement does not contradict the fact that both countries have managed to increase greatly the
sophistication of the manufactured goods they produce, from toys and appard to consumer electronics and
semiconductors. What it suggests is that this transformation has been fiiUy paid for by investments in phyacal and
human capital.
GROWTH POUCY

3.1. The switch In relative Incentives towards exports in the early 1960s was not
significant enough to account for the export boom
Countries that have experienced sustained export growth outside of East Asia have
almost always done so as a consequence of a sharp increase in the relative
profitability of exports. What is striking about the experience of South Korea and
Taiwan is how stable the relative price of their exportables was around the time of
export take-off. In both countries, most of the important export incentives had
already been in place for several years before the export boom started. Once the
boom got under way, it picked up speed, even though the measured profitability of
exports did not increase further. Moreover, exports continued their inexorable
rise, often in the context of deteriorating incentives for exporting activities. The
following paragraphs elaborate on these points in greater detail.

3.1.1. Korea. Under the Rhee government of the 1950s, Korean policy was
preoccupied by largely political considerations, and the government attached no
particular importance to either economic growth or exports (Jones and Sakong,
1980). There were multiple exchange rates and a haphazard, ineffective
programme of export subsidies (Frank et al., 1975). However, exporters could
retain a share of their export earnings to import certain items for home
consumption, a system which translated into a large export subsidy whenever the
free-market exchange diverged greatly from the official rate. After 1958, export
incentives were increased. Exporters were given tariff exemption on imports of raw
materi2ils and spare parts in 1959. Subsidized credit was made available to
exporters for up to 75% of their production costs, also in 1959. And a devaluation
of the currency in 1961 brought the official exchange rate close to the free-market
rate.
Later, after President Park took over in a military coup on 16 May 1961, the
scope of export subsidization was greatly enlarged. The subsidy on export credits
was increased and exporters were exempted from the commodity tax and the
business activity tax. The income tax on export earnings was reduced. There were
also direct cash grants on exports, but these were phased out by 1965 (Frank et al.,
1975). However, the incentive effects of the devaluations and the cash grants were
eroded by expansionary macroeconomic policies that led to rising inflation in
1962-3 and a renewed gap between official and parallel exchange rates in 1963. A
large devaluation in May 1964 served once again to unify the currency. After
1965, export subsidy programmes were expanded further. In that year, the
existing practice of giving priority to exporters in acquiring import licences was
formalized and expanded. Exporters were allowed automatic access to duty-free
imports of raw materials and intermediate inputs up to a limit. This limit was
determined administratively, on the basis of firms' and industries' input-output
coefficients plus a margin of 'wastage allowance' (Frank et al., 1975). Since the
imports acquired under the wastage allowance could be sold in the domestic
52 DANI RODRIK

market, this was a significant subsidy and was consciously used as such. Subsidized
credit to exporters became particularly important after 1966. Frank et al. estimate
that the wastage allowance alone provided an export subsidy of 4.6% in 1968 on
average, and up to 17-21% in certain fabrics and footwear. Bureaucrats had
virtually unrestricted discretion in setting wastage allowances, and their generosity
varied from time to time.
There is no doubt that these measures increased the relative profitability of
exporting compared to the situation that had prevailed during most of the 1950s.
However, it also seems clear that the greatest impact of the incentives was felt
around 1959-60, rather than in the mid-1960s when the export boom began. This
is largely due to two reasons: (1) the export subsidy implicit in the export-import
link system was particularly significant in 1959-60 when the gap between the
official and parallel exchange rates was large;^ and (2) inflation eroded many of the
export incentives between 1961 and 1964. The devaluation of 1964 and the
widening scope of export subsidies could offset the deterioration of incentives since
1960 only partially.
This can be seen in Figure 4, which plots the real effective exchange rate for
Korean exports. This is a measure of the real exchange rate which includes the
monetary equivalent of all the subsidies on exports (export premia through the
import link, cash grants, tax incentives, duty-free imports, expwrt credits and the
like), and which is therefore an appropriate index of the profitability of exporting
relative to other activities in the Korean economy.* As the denominator of this
index is the domestic price level, subsidies or protection of non-e:qx)rt activities is
captured to the extent that such policies raise the domestic price level relatine to
prices of export activities.
We note that, even with the devaluation of 1964, the level of export incentives in
1964-5 was no more than 10% higher than in the preceding couple of years, and
actually below the level attained in 1959-60. Even though exports rose very fast
from 1964 onward, they were not to regain their 1959-60 level of profitability
until the early 1970s, and then again only briefly. By the mid-1970s, the
export-GDP ratio was nearly ten times larger than in the early 1960s, yet the
relative profitability of exports was lower! The 2u-gument that e^qxirt-oriented
policies were responsible for the increase in exports is often made in a different
guise (see, for example. Page, 1994) by pointing out that trade and exchange rate
policies in the 1960s were not overdy discriminatory against exports (as they

' At first sight it may seem strange that a multiple exchange rate system, with an overvalued ofiidal e x c h a i ^ rate,
would act as an export subsidy. But the import-export link (i.e. the ability of exporters to retain some of their
dollar earnings to import for the home market) meant that exporters received some of the scarcity rents created by
the system.
* The subsidy equivalent of the export incentives are taken from Kim (1988). This, and the earlier Frank <( U.
(1975) study on which these estimates are based, are the moM authoritative and widely dted sources on the
quantitative aspects of Korea's trade regime.
GROWTH POUCY 63

30 120
Relative price of exports
25 100

20 -80

15- - 60 S

10 T -40
Exports/GDP

5 -20

n.n.n.n.n.n.n.ll.ll
55 57 59 61 63 65 67 69 71 73 75

Fignre 4. Relative price of ei^orts: Korea, 1955-76

Source: Frank et al. (1975), Kim (1988) and Bank of Korea.

commonly have been in other developing countries). The evidence for this comes
from taking the ratio of the effective exchange rate for exports to the effective
exchange rate for imports (both calculated by Frank et al., 1975), and noticing
that the resulting number is around 1 or somevifhat larger during the 1960s.
However, it turns out that the comparable number for the second half of the 1950s
is much larger, suggesting (if the numbers are to be believed) a much greater
export bias in the earlier period (see Frank et al., 1975, Tables 5-10, 8-lOD, 8-lOE,
8-lOC).
Hence the export spurt was not associated with a significant increase in the
relative profitability of exports. This has been noted by others. In their
authoritative study of Korean development. Mason et al. explicitly state that 'the
industrial policy changes that took place in thefirsthalf of the 1960s did not dearly
result in a significant increase in the measurable incentive to export'. Frank et al.
(1975) attempt to estimate the sensitivity of Korean exports to exchange rates and
e:qx)rt subsidies, and note that 'the main difficulty [in doing so is] that from 1955
to 1970 the effective exchange rate for exports remained remarkably steady'. The
same point is noted by Jones and Sakong (1980) as weU.
In resolving the apparent paradox, these authors resort to arguments that are
not entirely satisfactory. Mason et al. suggest that it was the stability of incentives
that was responsible for the export boom (see also Frank et al., 1975). But since the
incentive in question is the profitability of exports relative to other activities, there
is no clear reason why enhanced stability should have favoured exports over other
64 ^ DANI RODRIK

activities.^ Jones and Sakong (1980) resort to a range of explanations, incJuciing

greater stability, reduction in rent-seeking opportunities, simpler input acquisition,
clear political leadership, a more favourable 'business climate', £uid non-pecuniary
incentives such as presidential awards for successful exporters. Once again, the
more compelling among these factors are not specific to e^qxjrting activities, and
therefore cannot account for why exports increased faster than other activities.
With regard to the import regime, there was no significant import liberalization
until 1967, when the switch from a positive list to the negative list was
implemented. Under the new regime, the regulations began to specify only those
items that were prohibited, rather than those that could be imported subject to
restrictions (as in the past). In the period 1961-3, the number of items positively
listed as importables subject to government licensing, quotas, foreign currency
allocation and other regulations ranged between 1000 and 1600. Under pressure
from the balance of payments, the government actually reduced the number of
importable items to fewer than 500 in late 1964. The 1961-3 level was restored in
1965, and further increased later, until 1967 when the switch occurred (Hong,
1993).

3.1.2. Taiiivan. In Taiwan most of the export incentives were put in place in the mid-
to late-1950s, even earlier than in Korea, and the currency was unified during
1958-61. By 1954-5, the system of import duty and commodity tax rebates for
exportable production had already been implemented. In 1956, manufacrturers
were allowed to retain up to 80% of the foreign exchange they earned bam
exports £ind use it for their own import needs. (This ratio was raised to 100% of
export earnings for most items after the exchange rate reform of 1958.) In 1957, a
relatively generous export credit programme was started. Finally, the multiple
exchange rate system was unified during 1958-61 in several stages: (1) in ^ r i l
1958, the multiple buying rates were consolidated into two buying rates, in parallel
with two selling rates; (2) in November 1958, exports and im]X)rts under the lower
rate were brought up to the higher rate; and (3) further minor devaluations and
simplifications were undertaken during the following two years (Hong, 1993; Iin,

It it true that, in the presence of sunk costs associated with exporting, uncertainty in the pre-1964 period may
have prevented entrepreneurs from switching coitiiy production from the home market to world markets.
However, the export boom that took place was not a matter of switching production: it entailed the establishment
of new capacity specifically oriented towards foreign markets. With greater stability in relative incentives, the first-
order eiTect should have been to enhance the profitability of investing in new capacity for both foreign airf home
markets.
* In discussing the same issue, Frank it aL (1975) draw what is in my judgement the correct conclusion: *i( •
plausible to hypothesize that South Korean exports were constrained more by the capacity to produce goo<fa than
by the relative profiubility of producing for export instead of domestic markets'. T o extend this to its kigical
conclusion, we must therefore search for explanations for why it became profitable to invest and expand
productive capacity.
GROWTH POUCY 65

Real exchange rate

\

Exports/GDP

1960 62 64 66 68 70 72 74 76 78 80

Figure 5. Real exchange rates and exports: Taiwan, 1960-81

Sources: Kuo (1983, Table 14.4) and Council for Economic Planning and Development, Taiwan
Stattstical Data Book, 1982 and 1991.

1973). By July 1960, the difference between the official exchange rate and the
market price of foreign currency had become insignificant.
Unlike in Korea, we do not have a synthetic measure of an effective exchzmge
rate for exporters. So we have to content ourselves with a simple real exchange
rate index (not inclusive of export subsidies), which is plotted in Figure 5.
However, as discussed above, we know that ail the significant export subsidies had
already been deployed by the late 1950s. Therefore Figure 5 should give us a fairly
accurate idea of the trend in the relative profitability of exports since 1960. The
diagram shows that the initial export spurt (in 1963-4) was actually associated with
a decrease in export incentives, indicated by a real appreciation of around 10%
(the product of a fixed exchange rate). After 1964, the relative profitability of
exports increased steadily until 1973. But it was not until 1969 that the
export-GDP ratio resumed its climb. By the early 1980s, the relative profitability
of exports stood roughly at its level of 1961, yet the export-GDP ratio was more
than four times as large. It is a safe guess that no international economist,
presented with a real exchange rate chart such as the one in Figure 5, would have
predicted a fourfold increase in the exports-GDP ratio. (The real exchange rate in
question is the domestic price of tradables, more specificiLQy exportables, relative
to the price of non-tradables. This ratio can change considerably even in a 'small'
country with no market power in international trade.)
With regard to import liberalization, the Taiwanese pattern is again similar to
Korea's. There is a trend towards liberalization after 1964, but this is very much
the consequence of the increase in exports and the improvement of the balance of
66 DANI RODRIK

payments position. In any case, the opening up is hardly drastic. As in Korea, one
could not possibly ascribe the export boom to import liberalization.
Could the boom of the mid-1960s have been a delayed response to the shift in
incentives towards exports during the late 1950s in both countries? Comparative
evidence indicates that exports tend to react quite quickly to changes in incentives.
The examples of Turkey and Chile will be discussed briefly below. Moreover, this
comparative experience is instructive in another respect as well: in cases like
Turkey and Chile, sustained export booms generated by export-oriented policies
have been associated with real exchange rate depreciations that are much larger
than any experienced in Korea or Taiwan. The stability of relative prices in the
latter is particularly striking in comparative context, and does suggest that much
more than export incentives was involved in boosting exports.
Of course, export incentives (and in particular a relatively free-trade regime for
exporters) must have been a necessary condition for exports to take off in Korea
and Taiwan: it is hard to imagine the export performance of these countries taking
place in the presence of grossly overvalued currencies or high barriers to trade in
imported inputs used in exportables. Nonetheless, the delay suggests that the
export incentives were not sufficient in themselves.

3.2. Export orientation versus investment boom

It is not clear why export orientation (or the increase in exports) should have led to
an investment boom. The standard account rarely confronts the question of wiiy
either the export incentives or the export boom should have led to an investment
boom. We have many cases where even more sizeable increases in the profitability
of exports either have not led to increases in investment or have done so only after

1979 81

Figure 6. Turkey, 1979-90

Source: O E C D , Economic Surveys: Turkey, various issues.
GROWTH POUCY 67

1980 90

Figure 7. Chile,
Smira: Bosworth et al. (1994).

considerable lag. This is obviously important, in so far as exports (unlike

investment) do not directly lead to economic growth.
Consider two important examples, Turkey and ChDe. In Turkey, there was a
massive increase in the profitability of exports in the early 1980s (of more than
50%), with considerable import liberalization as well. These reforms were
accompanied by an impressive increase in the export-GDP ratio (Figure 6). Yet, as
Figure 6 shows, private investment actually fell in this period, and did not recover
until the second half of the 1980s. In Chile, a fall in the relative profitability of
exports in the late 1970s was accompanied by an investment boom. Investment
collapsed in 1982-3 as a result of a major financial crisis. Export incentives
increased significantly after 1982, but private investment responded sluggishly
until 1989 (Figure 7).
We should not be surprised to find that export incentives and investment can co-
vary in different ways. In theory, there is no reason to suppose that export
orientation should be associated with an increase in investment demzuad. Export
orientation makes some sectors more profitable and others (import-competing
activities and non-tradables) less so. The same is true of import liberalization.
The net effect on investment demand is indeterminate. If anything, the logic of
relative factor endowments and the Stolper-Samuelson theorem suggests an
opposing presumption: in capital-poor countries like Taiwan and South Korea
during the 1950s and 1960s, an increase in the relative price of exports should
have been associated with a dKlim in the return to capital, and hence reduced
investment!
68 DANI RODRIK

3.3. The contribution of exports

Since the export base was so small early on (especially in Korea), the contribution
of exports to GDP growth could not have been very high until the mid-1970s at
least. Elxports were less than 5% of GDP in Korea around 1960, and barely over
10% in Taiwan. In a purely accounting sense, exports could have been responsible
for only a small fraction of the initial growth spurt in both countries, in view of the
small base from which they sprang. This comes out clearly in demand-side
decompositions of output growth (which break down growth into increases in
domestic demand, import substitution, exports and changes in input-output
coefficients). For example. Hong (1976) calculates that export expansion could not
have accounted for more than 10% of Korea's growth in 1966-70. Kuo (1983)
reports a higher contribution of exports to Taiwan's growth (but still below that of
domestic demand until the 1970s), in large part because the Taiwanese economy
was more open at the beginning of the transition.
Such demand-side decompositions of the 'sources of growth' are in themselves
unsatisfactory because the methodology is vague as to the mechanism, if any, that
relates exports growth to GDP growth. In a fully employed, small open economy
with marginal products of productive factors eiualized among different activities,
an increase in exports cannot raise output. Of course, an increase in exports
caused by an extem2d terms of trade improvement could be associated with an
increase in real income. However, terms of trade movements cannot account for
any significant part of Korea's and Taiwan's success. During much of the 1960s,
Korea and Taiwan had some unemployment as well as a productivity gap between
the modern and traditional sectors of their economies. Under such conditions, an
increase in exports can be expansionary, but the question is how much.
A useful first approximation is provided by a computable general equilibrium
(CGE) model constructed by Trela and Whalley (1992) for Korea. These authors
treat the agricultural sector differently from the other sectors of the economy, in
that they assume labour is paid its average (rather than marginal) product in the
agricultural sector. Consequently, as labour is drawn into the mcxlem sectors of
the economy (including exportables), total output rises. Even so, they find that
export-oriented policies can explain in their model only a very small firaction of
Korean growth during 1962-82. In their words, 'the results seem to imply that
outward-oriented policies in Korea have little significance in driving growth'
(1992). This is a consequence of the facts already discussed: exjxjrtables were a
smedl part of the economy and there was only a relatively small change in
measured incentives towards exportables.

3.4. Productivity spillovers from exports?

It is not clear that export growth was, or should have been, associated with
cumulative productivity spillovers to the rest of the economy. The previous
GROWTH POUCY 69

arguments can be countered by claiming that the contribution of exports to growth

came not from the demand side or through investment, but from widespread
technological ^illovers and cumulative productivity benefits deriving from export
performance. This is indeed a common presumption. The World Bank's (1993)
East Asian report gives it top billing in its exposition of the 'dynamic' benefits of
outward orientation. See also de Melo and Robinson (1992). These authors use a
CGE model with export externalities to account for those stylized features of East
Asian development that standard CGEs are unable to explain'.
The trouble is there is virtually no evidence that exports or outward orientation
were associated with technological externalities. The World Bank's report does not
provide any. It bases its arguments on a number of a priori reasons, which are in
themselves quite problematic (see RodHk, 1994, for a critique of the report, and
Rodrik, 1992, for a general discussion of the ambiguous theoretical case in favour
of productivity spillovers from exports). It is common in this literature to point to
correlations between export growth and TFP performance across firms or
industries as evidence for technological spillovers. Such correlations obviously say
nothing about the direction of causality, if any. The most plausible way to interpret
such findings is that firms and industries which are successful in improving their
productivity are better able to compete in world markets. Perhaps most telling on
this score is that, as we have seen, overall productivity growth in industry has not
been spectacular in either country, and can explain only a small part of total
growth. As Young (1994) puts it, 'it is not particularly difficult to find either
developed or less developed economies whose productivity performance, over time
periods spanning two decades or more, has matched or rivjilled that of the NICs
[including Korea and Taiwan]' (see also Young, 1993). Hence, we should be
suspicious of any argument that gives pride of place to the (assumed) presence of
technological spillovers from exports.
Moreover, the argument about technological spillovers from exports is usually
made in the context of manufactured, rather than aggregate, exports (World Bank,
1993; Page, 1994). That constitutes another problem, in view of the meagre base
from which the externalities in question are presumed to have come. In both
countries, manufactured exports accounted for a quarter or less of total exports
around 1960, and consequently they were an even smaller share of national
income. It is not easy to see how the spillovers from such a puny source could be so
strong as to set a process of aggregate economic growth into motion.

3.5. The rising share of exports In GDP Is consistent with investment-ied growth
As mentioned previously, the apparent clincher for the export-led growth
hypothesis is the steady increase in the exports-GDP ratio that both countries have
experienced. If, as argued above, exports are unlikely to have played much of a
causal role in growth, why did this ratio increase so much? The answer is provided
70 DANI RODRIK

Box 1. Endogenous response of exports to an investment boom

The initial levels of consumption and production are represented in Figure 8

by points Co and Qp respectively. We represent the increase in investment
demand as a shift in preferences, which biases home demand towards the more
capital-intensive importables. The relevant indifference surface now becomes
U\, with the associated consumption and production levels of C\ and Qj
respectively. (Note that the production mix remains the same, as the external
terms of trade are taken to befixed.)The economy's imports and exports both
expand. The impact effect of an increase :' investment is to render the
economy more open to trade. The longer-run effect is shown on the right
panel. As the economy's capital stock expands, the transformation frontier
moves outward in a fashion that is bijised towards the capital-intensive
importable. This has the effect of reducing trade over time, but, as drawn, the
trade triangle remains larger than in the initial equilibrium.
The figure also clarifies how, following an increase in investment (and
import) demand, an increase in exports can develop without any change in the
relative price of the exportable. The mechanism that enables this is the switch
in domestic expenditures away from the exportable, allowing in turn an
increase in exports. In the absence of non-tradables {zs in the diagram), the
increase in exports exactly matches the increase in imports, with no change in
relative prices. In the presence of non-tradables, some of the increased demand
for the importables would come at the expense of non-tradables, calling for a
reduction in the price of non-tradables (relative to tradables) to equilibrate the
market for non-tradables. The magnitude of the equilibrium real exchange
rate depreciation would then depend on the parameters of the model. ^

' The general case with non-tradables is analysed formally in Rodrik (1995) in the context of an explicit
intertemporal model. In such a model, there are two offsetting effects on the price of exportaUesrelativeto
non-tradables. The substitution effect (as expenditure switches towards the importable) tends to depress the
price of non-tradables and raise the relative price of exportables. The income effect (from the increase in the
profitability of investment) goes in the opposite direction.

by thinking in general equilibrium terms, with a formal iUustration provided in

Box 1.
Consider an (exogenous) increase in the profitability of investment, followcd by
an increase in the share of investment in GDP. Suppose that the country in
question (like Korea and Taiwan during the 1960s) has a comparative
disadvantage in producing capital goods. The investment boom will require a
GROWTH POUCY 71

.1
.9
72 DANI RODRIK

0.
o
O

1960 62 64 66 68 70 72 74 76 78 80 82 84 86 88

Figure 9. Imports and investment: Korea, 1960-48

Sources: Penn World Table 5.5 and Economic Planning Board, Myor Statistics of the Korean Eanumiy,
various issues.

commensurate increase in imports (also as a share of GDP). And if international

borrowing is not unlimited, exports must rise (as a share of GDP) to pay for the
imports. Hence, we will observe an increaising trade orientation alongside the
boom in investment. This story reverses the causality between growth and exports.
Export orientation enables growth (by allowing imports to increase), but it is not its

0.
a
(5

1952 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90

Figure 10. Iiiq>orts u u l inveatmenti Tai%van, 1952-90

Sources: Penn World Table 5.5 and Natmal Incom* in Takum Ana of Urn Rt/mblk ofOumi (1993).
GROWTH POUCY 73

14
A
12
/'^Machinery and transp. equip. /
10

• \Crude materials (exc.fuels)

1960 62 64 66 68 70 72 74 76 78 80 82 84 86 88

Figure 11. Imparts as a share of GDP: Korea, 1960-88

Source: Economic Planning Board, Major Statistics of the Korean Ecorumy, 1976, 1989.

ultimate determinant. Ultimately, the reasons for growth must be traced back to
reasons why it became profitable to invest.
This story is quite consistent with the Korean and Taiwanese experiences. First,
a casual look at the data shows that in both cases investment and imports are
closely related. As Figures 9 and 10 make cleeir, the behaviour of imports tracks
quite closely the behaviour of investment. In Korea, investment and imports both
rise (as a share of GDP) until around 1980, and then stabilize somewhat. In
Taiwan, investment and imports rise in tandem until the late 1970s, and then both
decline somewhat. Figures 11 and 12, showing the composition of imports, make

16

A
14
' \ Machinery and transp. equip. \
12 ," \
/ \ / \
10
» \ /
% 8 . Aqric. products
/.

4 \ / ^
^' •.. / Basic metals" ^ . _ S : /

Textiles
- t 1 1 I 1 t 1 1I 1 1 ) 1 1 1 1 1 I 1 1 I M 1 M 1 I 1 '1 1 ) I 1 ( 1 t t
1952 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90

Hgure 12. Imports as a share of GDPi Taiwan, 1952-90

Source: Council for Economic Planning and Development, Taiwan Statistkai Data Book, 1982, 1991.
74 DANI RODRIK

dear the connection. Since the mid-1960s, the share of capital goods (machinery
and transport equipment) in both countries' imports has risen sharply. In fact, it is
mainly the increase in capital goods that accounts for the rise in the imports-GDP
ratio. Hence, the increasing export orientation of the economy is quite consistent
with investment-led growth, with causality running from investment to imports
and from imports to exports. '

3.6. The role of savings

In both countries, savings rose alongside investment, and therefore enabled growth
to proceed without hitting balance of payments constraints. This is important as it
constitutes another part of the explanation of how a large export surplus could
develop without substantial contemporaneous changes in the relative price of
exportables. Most accounts view the increase in savings in Korea and Taiwan as
having been the result of economic growth itself and of a comparatively early
demographic transition (Collins and Park, 1989; Collins, 1991; World Bank,
1993).
Government policy was helpful as well. Real interest rates were raised for
depositors to levels that were either positive or only mildly negative (Taiwan in the
1950s and Korea in the 1960s). In addition, an increase in public savings made an
important contribution to total savings in both countries. In Taiwan, the increase
in the savings rate in the early 1960s coincided with a sharp increase in
government saving after 1961 (Kuo, 1983). This wais enabled in tum by a
reduction in government consumption. And in Korea, 'it was the rapidly
increasing contribution of government savings zmd the steady inflow of foreign
savings which enabled Korea to achieve the very high rate of investment during
the 1962-73 period' (Hong, 1976).

3.7. Recapitulation
The proposition that Korea's and Taiwan's economic performance can be
ascribed to export orientation faces serious difficulties. The switch towards export-
oriented pohcies cannot account for the sustained export boom since the mid-
1960s, and even less for the equally impressive and sustained investment boom.
Export growth itself can explain only a limited part of the early growth in output.
The increasing share of exports in GDP is quite consistent with a story of

According to Box 1, the openness of the economy must necessarily decline once the investment ratio stabilizes.
However, there are a number of confounding features in the Korean and Taiwanese experiences. Foremost
among these is the large-scale import liberalization that has taken place in both countries during the 1980s.
' As Robert Wade has reminded me, the argument about investment-led ejqwrts is not new. However, I have had
difficulty locating sources in the published literature which place priority on investment demand <W at the same
time explain the rising export-GDP ratio. Bradford (1990), for example, seems to suggest a causal role for
investment, but it is unclear as to why the investment was allocated di^roportionately in e^qxirt^oriented secton.
GROWTH POUCY

investment- rather than export-led growth. These problems should lead us to

search for more direct explanations for the apparent increase in the profitability of
private investment around the mid-1960s in both countries. The rest of this paper
jirgues that the explanation has to do with government policies that go
considerably beyond export incentives, and that Korea and Taiwan shared some
rather special initial conditions that made such government policies possible.

4. IMPORTANCE OF INITIAL CONDITIONS

While South Korea and Taiwan were both quite poor around 1960, their social
indicators placed them among the ranks of countries at several times their income
levels. Table 3 shows Adelman and Morris's (1967) index of socioeconomic
development for a range of countries, as measured around the late 1950s and early
1960s. This index is derived from factor analysis and is based on a large number of
indicators meant to capture characteristics of social structure and social
organization. (The indicators include the extent of dualism, urbanization,
importance of an indigenous middle class, social mobility, literacy, mass
communications, cultural and ethnic homogeneity, fertility, national integration
and sense of national unity, and modernization of outlook.) Adelman and Morris
place Taiwan and Korea in their most advanced group, even though their per-
capita incomes are considerably below average.
If we focus specifically on indicators of educationed attainment, we see the same
discrepancy with the level of per-capita income. Table 4 displays data on three
educational indicators which are commonly employed as explanatory variables in
cross-country growth regressions. The table shows the actual school enrolment and
literacy rates in Korea and Taiwan in 1960, as well as the corresponding rates that
would have been expected on the basis of these countries' per-capita income levels
alone. The latter are derived from cross-section regressions of educational
indicators on per-capita income and its square. We find that both countries had

Table 3. Socioeconomic development and income

Index of socioeconomic Per-capita GNP, 1961 ($)

development, c. 1960
South Korea 0.85 73
Taiwan 1.05 145

Brazil 0.79 186

Cambodia -0.55 101
Ivory Coast -0.98 184
Morocco -0.57 150
Cyprus 1.08 416
Jamaica 1.06 436

Source: Adelman and Morris (1967, Table IV-5).

76 DANI RODRK

Table 4. Educadomal indicatora

Primary enrolment ratio Secondary enrolment ratio literacy rate

Predicted Actual Predicted Actual Predicted Actual

Korea 0.57 0.94 0.10 0.27 0.31 0.71

Taiwan 0.62 0.96 0.12 0.28 0.36 0.54

Source: See text.

virtually universal primary school enrolment, while the norm for countries at their
income levels stood at around 60% only. Korea had more than double the literacy
rate compared to the norm, and Taiwan's literacy rate was one-and-a-half times as
high. It is clear that both countries had a labour force that was considerably better
educated than would be predicted from their income levels.
The other respect in which Korea and Taiwan stood out by 1960 was their
exceptionally equal distribution of income and wealth. This was due in part to
long-standing historical reasons, and in part to the serious lsmd reforms undertaken
in both countries during the 1950s. Figure 13 plots Gini coefficients for income
and land distribution for 41 countries for which both measures are available for a
yeeu" around 1960. Korea and Taiwan are the two countries closest to the origin:
that is, with the lowest overall inequality.
These initial conditions can account, in a statistical sense, for a large part of the
two countries' economic performance since 1960. Table 5 shows the results of
regressing growth and investment rates on initial primary enrolment and

070
•
0.65
« *
0.60
i
* • • *
t • •
r 0.50 Finjand
g 0.45
*
•: •• *
1 0.40 •• ••
c 0.35
Korea Taiwan
• *
0.30 •

0.25
0.20
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
GInl coefficient for land
Figure 13. MeaaurM of income and land diatribudoa, e. 1960
Source: Alesina and Rodrilc (1994).
GROWTH POUCY 77

Table 5. Growth regreuiona

Independent Dependent variables

Per-c^ita GDP growth, 1960-85 Investment/GDP, 1960-85

(1) (2) (3) (4)

Intercept 6.22* 3.71* 16.06* 18.06*
(4.69) (3.86) (2.64) (4.32)
Per-capita GDP, -0.38* -0.38* 0.94 0.49
1960 (-3.25) (-3.61) (1.76) (1.08)
Primary 2.66** 3.85* 11.01** 14.11*
enrolment, 1960 (2.66) (4.88) (2.40) (4.11)
Gini cocff. for -5.22* -5.50* -21.04* -16.59*
land (-4.38) (-5.24) (-3.85) (-3.64)
Gini coefF. for -3.47 14.44
income (-1.82) (1.66)
0.53 0.53 0.43 0.50
Sample size 41 49 41 49

Notes: Numbers in parentheses are /-statistics. Asterisks denote level of significance:

* Significant at the 1 % confidence level
** Significant at the 5% confidence level.

inequality indicators (as well as initial per-capita income) in the sample of countries
for which inequality data are available. The primary enrolment rate has a positive
and statistically significant coefficient (as expected). In addition, there is a strong
negative association between inequality (particularly in land distribution) and
subsequent growth. Despite the parsimonious specification (most notably, the
exclusion of investment as an explanatory variable from the growth equation),
these regressions do rather well, and explain around half or more of the cross-
national variation in growth and investment rates.
Table 6 shows that almost 90% of the two countries' growth experience since
1960 can be 'explained' by these initial conditions. When Korea and Taiwan are

Table 6. Frctpordon of growth explained by initial conditions

Actual growth Predicted growth Proportion explained

(1960-85) (%)
South Korea 5.95 5.24 88
Taiwan 5.68 4.96 87

Argentina 0.48 1.58 329

Brazil 3.52 1.96 56
India 1.37 3.46 253
Kenya 0.% 1.46 152

Notts: Predicted growth rates are from regresaon (1) in Table 5.

78 DANI RODRK

excluded from the original sample from which the predicted values are generated,
the percentage of actual growth predicted by the regression is a bit lower (82% for
Korea and 81% for Taiwan), but stiU striking. In a statistical sense, then, there is
nothing 'miraculous' about their experience. The real outliers are countries like
Argentina and India (whose actual growth is vastly overpredicted) or Brazil (whose
growth is vastly underpredicted). Of course, while these results may be interesting,
they do not amount to an explanation. We still need a theory on why these initial
conditions mattered as much.'" That is the task of the next section.

5. THE COORDINATION FAILURE INTERPRETATION

5.1. The argument

What we need is an explanation that can account for the main outlines of Korean
and Taiwanese economic performance, and which also has a role for the special
set of initial conditions discussed in the previous section. Such an explanation can
be constructed along the following lines.
First, by 1960 Taiwan and South Korea shared a set of advantageous initial
conditions relating to social infrastructure. In particular, both economies had a
skilled labour force, relative to their physical capital stock and income levels. These
initial conditions made both countries ready for economic take-off, in the sense
that the latent return to capital accumulation was high.
Second, for a number of reasons, the economic take-off could not take place
under decentrailized market conditions. Chief 2unong these reasons are the
imperfect tradability of key inputs (and technologies) associated with modem-
sector production, and some increasing returns to scale in these activities. These
conditions created a situation of coordination failure. In other words, while the
rate of return to coordinated investments was extremely high, the rate of return to
individual investments remained low.
Third, governments in both countries undertook a set of measures starting in the
late 1950s that not only removed some policy-induced distortions, but also served
to coordinate and subsidize private investment. These measures included: credit
subsidies, tax incentives, administrative guidance and public investment.
Fourth, this active government role helped remove the coordination failure that
had blocked industrial growth. As private entrepreneurs responded to these
measures, the resulting investments turned out to be profitable not only in financial
terms, but in social terms as well.

'" Another possibly important initial condition, emphasized by Adrian Wood in personal correspondence, is the
lack of a good natural resource base in Korea and Taiwan. This a in part related to the high educational
attainment ratios relative to income: as Wood points out, countries with the same level of education per worker
but more land would have had a higher GNP per capita and lie closer to the regression line. But in addition, the
meagre natural resources gave these countries a clear comparative advantage in manufactures, allowing them to
enjoy both rapid industrialization and rapid trade expansion.
GROWTH POUCY 79

Fifth, government intervention could be implemented in an effective manner

(without leading to rent-seeking behaviour) because initial conditions, once again,
had endowed the government in each country with an extraordinary degree of
insulation from pressure groups, and with leadership capability over them. Among
these initial conditions, a relatively equal distribution of income and wealth was
critical.
Sixth, as investment rose as a share of GDP, so did imports of capital goods, as
neither country had a comparative advantage in such goods. Thanks to
appropriate macroeconomic and exchange rate policies, export supply was
adequate to meet the increase in import demand, and rose alongside imports.
Seventh, as a consequence, the increase in exports played a critic£il role in
paying for the imports of capital goods. But it is more appropriate to view this
increase in exports as a consequence of the increase in investment demand, rather
than the other way around.

5.2. A framework of analysis

There are two critical claims in this story: (1) both countries were ready for
economic take-off by the early to mid-1960s, but economic growth was blocked by
a coordination failure; (2) governments in both countries were able to undertake
the measures needed to override this coordination failure. The evidence on the
presence of a coordination failure is necessarily circumstantial. I think the case is
reasonably compelling in view of the likelihood that all of the prerequisites for the
existence of a coordination failure were met in the two countries. The appendix
presents an aneilytical framework which makes explicit what these prerequisites
are. Here I rely on an intuitive exposition of the economic logic.
Imagine a smjill open economy, initially specializing in the production of
traditional goods. Alongside there exists a relatively capital-intensive modem
sector, which yields higher factor returns when it is viable. The modem sector
relies on specialized inputs (e.g. particular labour skills, technologies, intermediate
inputs or capital goods). These inputs share the following features: (1) they require
well-educated workers but at low cost; (2) they exhibit scale economies; and (3)
they caimot be perfectly traded in international markets. The viability of the
modem sector requires the local presence of these inputs, which in tum depends
(in part) on the existence of a sufficiendy well-educated workforce.
Such an economy is ready for take-off if there is enough skilled labour that the
modem sector would be viable if a large enough share of the economy's resources
were devoted to producing the specialized inputs. Yet there is no certainty that
labour and capital move from the traditional sector to the modem sector, leading
to specialization in the latter and to higher incomes. The reiison is that, because of
scale economies, only a large-scale movement of resources is guaranteed to be
profitable. From the perspective of an individual investor, it will not pay to invest
in the modem sector unless others are doing so as well. The profitability of the
80 DANI RODRIK

modem sector depends on the simultaneous presence of the specialized inputs; but
the profitability of producing these inputs in turn depends on the presence of
demand from a pre-existing modem sector. It is this interdependence of
production and investment decisions that creates the coordination problem.
Coordination failure is least likely to happen when the economy is well endowed
with both skilled labour and physical capital, for then production in the modem
sector is profitable even when entrepreneurs act in an uncoordinated manner. For
economies at the other end of the spectrum - lacking both skilled labour and
capital - the coordination issue is moot because the modem sector is not viable in
the first place. It is in the intermediate economies most reminiscent of Korea and
Taiwan in the early 1960s - weU endowed with skilled labour but poor in physical
capital - that the coordination problem is most severe.
Markets are known to handle resource allocation poorly in the presence of scale
economies and non-tradability: market prices reflect the profitability of diflFerent
activities only as they are currently undertaken; they do not provide any signals
about the profitability of activities that would require a large-scale reallocation of
resources within the economy (which, after all, is what economic development is all
about). These are, of course, old ideas that go back to Scitovsky's (1954) analysis of
pecuniary externalities and Rosenstein-Rodan's (1943) advocacy of big-push
policies. More recendy, the arguments have been formalized in papers by Faini
(1984), Pack and Westphal (1986), Murphy et al. (1989), Krugman (1991),
Matsuyama (1991), Ciccone and Matsuyama (1993), Rodriguez-Clare (1993) and
Rodrik (1993).
One problem with this literature has been that coordination failure is often
presented as a generic problem affecting all kinds of economies. The present
framework is more specific about the prerequisites. It highlights the following three
prerequisites for a coordination failure to become a serious issue: (1) some degree
of non-tradability in the technologies and/or goods associated with the modem
sector; (2) economies of scale; (3) a reasonably skilled labour force (but a low
endowment of physical capital). The last one dearly applies to the case of Korea
and Taiwan. Scale economies are also plausible in many of the modem-sector
activities. Hence, non-tradability is the feature that requires additional discussion.
Upon a moment's reflection, it should be dear that some degree of non-
tradability is necessarily associated with the types of goods produced by rich
countries. Otherwise poor countries would not remain poor for long: arbitrage
through trade would eliminate the disparities. In practice, the non-tradability of
modern-sector inputs is observed in a number of different ways. Labour services
are for the most part effectively non-traded, so that skilled and spedalized
workmanship must be locally available. The fixed costs often required to develop
these skills lead to scale economies. Intermediate and capital goods are in principle
tradable, but they sometimes require either geographic proximity to the final user
(as when they are manufactured to suppliers' specifications) or the use of
GROWTH POUCY

complementary local inputs before they can be put to use (as when skilled workers
are needed to operate sophisticated imported machinery). Often, the requisite
technologies also have a non-tradable element, in so far as much of the
technological capability is tacit and not explicidy codified in designs and
blueprints. As Pack and Westphal (1986) put it:
The tacitness of technology leads to problems in its communication over long
distances and across social differences, problems which can be overcome - if at
all - only at some cost . Moreover, knowledge that exists (somewhere in the
world) does not exist everywhere simultaneously because there are costs in
advertising its mere existence or in discovering its existence through search.
Only knowledge that is "close by' is known to exist Another significant
channel for inter-industry externalities is the exchange of technological elements
in transactions involving intermediate products and capital goods. Indeed many
such exchanges leading to better utilization of local resources and to
improvements in the design of capital goods have been observed. A salient
aspect of these exchanges is the dependence of their outcome on extensive
interaction between suppliers and users in iteratively changing both process and
product characteristics.

Some exzimples drawn from the East Asian experience may help bring these points
to life.

5.3. Case studies: Hyundai and Lucky-Goldstar

The importance of specialized labour skills and the complementarities they
generate across manufacturing activities is illustrated by the experience of
Hyundai, one of Korea's huge conglomerates {chaebol). Hyundai first entered
manufacturing in 1964 by building a cement plant. According to Amsden (1989):
Hyundai used its cement plant as a laboratory to train its managers with
background in construction, before assigning them to other manufacturing
aiFiliates. Trainees gained experience in inventory management, quality and
process control, capacity planning, and so on, thus spreading basic production
skills throughout the Hyundai organization. After Hyundai Cement, the next
manufacturing affiliate in the group was founded in 1967 and named Hyundai
Motors. Twenty years later it became the first independent automaker from a
late-industrializing country to export globally. The first president of Hyundai
Motors was a former president of Hyundai Cement.
Korean government policies were highly partial to conglomerates like Hyundai.
By giving them access to subsidized capital, the government allowed them to
internalize many of the labour market spillovers in the fashion described in the
quote.
82 DANI RODRIK

Hyundai's experience with shipbuilding provides a concrete instance of the

imperfect tradability of technology (and its interaction with scale economies). The
company started out by importing its basic design from a Scottish firm, but soon
found that this was not working out. The Scottish design relied on building the
ship in two halves because the original manufacturer had enough capadty to build
only half a ship at a time. When Hyundai followed the same course, it found out
that the two halves did not quite fit. Subsequent designs imported from European
consulting firms also had problems, in that the firms would not guarantee the rated
capacity, leading to costly delays. Engines were available from Japanese suppliers,
but apparently only at a price higher than that obtained by Japanese shipyards.
Moreover, ship buyers would often require design modifications, which Hyundai
would be unable to undertake in the absence of an in-house design capability.
Only with large enough capacity would it pay for Hyundai to integrate backwards
(into design and engine building). In a highly volatile business, scale in tum
depended on having access to a steady and reliable customer (a merchant marine).
The Korean government provided Hyundai with substantial assistance, as weD as
an implicit guarantee of markets. Hyundai eventually integrated both backwards
and forwards. The government's guarantee came in handy in 1975 when a
shipping slump led to the cancellation of foreign orders. President Park responded
by forcing Korean refineries to ship oil in Korean-owned tankers, creating a
captive demand for Hyundai Jones and Sakong, 1980).
The chairman of the Lucky-Goldstar group explains the success of his company
in this way:
My father and I started a cosmetic cream factory in the late 1940s. At the time,
no company could supply us with plastic caps of adequate quality for cream
jars, so we had to start a plastic business. Plastic caps alone were not suflBdent to
run the plastic-moulding plant, so we added combs, toothbrushes, and soap
boxes. The plastics business also led us to manufacture electrical and electronic
products and telecommunication equipment. The plastics business also took us
into oil refining which needed a tanker-shipping company. The oil-refining
company alone was paying an insurance premium amounting to more than half
the total revenue of the then largest insurance company in Korea. Thus, £in
insurance company was started. This natural step-by-step evolution through
related businesses resulted in the Lucky-Goldstar group as we see it today, (cited
in Amsden, 1989)

The quotation clearly illustrates the importance of local inputs and customers as
well as of scale economies in fuelling the growth o{ chadtoL While the chadtolcovlA
thus internalize some of the coordination issues, they were greatly assisted in doing
so by govemment policies which wiU be discussed in the next section.
In both Korea and Taiwan, the rate of retum to capital and profitability in key
manufacturing activities rose significantly from the late 1950s on. In Korea, Jones
GROWTH POUCY 83

and Sakong (1980) report (based on Hong, 1977) steadily rising real rates of retum
to capital in manufacturing: the range is 9-18% in mid- to late-1950s, 9-26% in
1962-6, 16-38% in 1967-72, and 17^0% after 1972. The rate of profit in
manufacturing steadily rose from 9% in 1951-3 to 16% in 1954-6, to 28% in
1957-62, and to 35% in 1963-70 (Hong, 1993, p. 347). Apparendy, investment
became more profitable as the investment rate rose." In Taiwan, profitability
rates rose in most of the private manufacturing industries after the late 1950s, with
the notable exception of textiles and wood products, two major exporting
industries (Lin, 1973). Interestingly, the greatest increase in profitability in the
post-1963 period (outside food, beverages and tobacco) was experienced by public-
sector manufacturing. As will be discussed in the next section, it was public
enterprises that supplied many of the key intermediate inputs in Taiwan. This is
how lin (1973) explains the increase in their profits:

The domestic consumption of the output of these non-food industries (which

produce petroleum products, chemical fertilizers, industrial chemicals, etc.)
increased tremendously during the 1960s, due to increased demand from
chemical-using industries (such as those making polyvinylchloride, monosodium
glutamate, and paper £ind pulp for both the export and domestic market), as
well as from the agricultural sector and the transportation industry.
In other words, intermediate industries became profitable thanks to expanding
linkages downstream.
We note finally that in both Korea and Taiwan the way policy-makers viewed
the economy and their role in it has parallels with the logic of the coordination
failure. As the discussion in the following section will make clear, the Korean
government has always perceived itself as a mediating agent and a facilitator for
bringing about industrial change, through arm-twisting, subsidies or public
enterprises as the circumstances may demand. In the words of Pack and Westphal
(1986):

In Taiwan, the basic philosophy underlying [the government strategy] is that an

economy will undergo certain stages of development, and at each stage there
are certain key industries (such as integrated steel mill, large shipyard, and
petrochemical plants) which through various linkages will bring about
development of the entire economy. This strategy also jissumes that government
officials know what those key industries are and what policy measures should be
adopted to develop these industries. (Hou, 1988, cited in Hong, 1993)

" little (1994) calculates that the annualized return to investment in Korea was 31.1% during the period
1963-73. However, his calculations also show a reduction in the rate of retum subsequently, to 18.3% during
1974-9. He attributes the decline to the HCI drive.
84 DANI RODRIK

Indeed, Taiwan's Fourdi Plan (1965-8) stated:

For further development, stress must be laid on basic heavy industries (such as
chemical wood pulp, petrochemical intermediates, and large-scale integrated
steel production) instead of end product manufacturing or processing. Industrial
development in the long mn must be centred on export products that have high
income elasticity and low transportation cost. And around these products there
should be development of both forward and backward industries, so that both
specialization and complementarity may be achieved in the interest of Taiwan's
economy, (quoted in Wade, 1990)
Hence, what these governments thought they were doing has much in common
with the ideas discussed here.

6. GOVERNMENT POLICIES TO SUBSIDIZE AND COORDINATE PRIVATE

INVESTMENT

Under the conditions discussed in the previous section, there exists a large role for
govemment intervention. Such intervention can take many different forms. Most
directly, policy-makers can coordinate private-sector production and investment
decisions through their control over credit allocation, the tax regime and trade
policy, as well as through 'administrative guidance'. Govemment policies to
subsidize investment in the modem sectors of the economy have a ljirge payofiF
because they get the private sector to internalize the coordination externalities.
The same outcome can also be obtained through investments by public enterprises
themselves. The Korean and Taiwanese governments used a combination of these
interventions, thereby raising the private retum to capital in the modem sectors to
the level of the social retum.

6.1. Improving the investment climate

As indicated above, both governments actively subsidized £md coordinated private
investment. However, one of the most important changes that took place in the
late 1950s in Taiwan and the early 1960s in Korea was a substantisd improvement
in the investment climate overall, brought about by a reorientation of govemment
priorities. We can view this as an important prerequisite to the effectiveness of the
interventions themselves.
During much of the 1950s, economic goals did not rank particularly high with
the Taiwanese leadership. The govemment was preoccupied instead with the
reconquest of the mainland. By the end of the decade, it became dear that the
communist regime was firmly entrenched: 'P^aiwan's] party elders came to see
that economic development could be a better guarantee of the party's survival'
(Wade, 1990). Thereafter, the govemment tumed its energies to eliminating nuuiy
investment-deterring distortions (such as multiple exchange rates and macro-
GROWTH POUCY 85

economic instability). An important turning point was the Nineteen-Point Reform

Programme instituted in 1960. This contained a wide range of subsidies for
investment, and will be discussed further below. What is more important at this
juncture is that this programme signalled a major shift in government attitudes
towards investment: Lin (1973) notes that '[with] the announcement of the
nineteen-point reform programme of 1960, the improvement of investment
climate became a catchword. The simplification of administradve procedures and
the liberalization of regulative measures with regard to economic matters became
an official goal.' And in 1965, many of the remaining administrative controls on
new plants or capacity expansion were removed altogether.
The story in South Korea is quite similar. President Park, who took power in a
military coup in 1961, could not have been more different from Syngman Rhee,
his predecessor (not counting the short-lived Chang Myon regime). Rhee's
attention had been focused on national consolidation and on political goals;
economic growth was never a priority. Park, on the other hand, gave precedence
to economics over politics, and to economic growth over other economic concerns.
These priorities were refiected in the amount of time he spent on economic
matters, and in his support of growth-oriented bureaucrats and businessmen (fones
and Sakong, 1980). Park made very clear early on in his rule that entrepreneurs
who undertook investments in line with his priorities would be richly rewarded,
while others were penalized. In fact, one of the first acts of Park was to arrest most
of the nation's leading businessmen, under a Special Law Dealing with Illicit
Wealth Accumulation, and to charge them with profiteering under the previous
regime. The businessmen were eventually set free, after Park had extracted a
commitment from them to undertake specified investments. While locking up
businessmen may seem an odd way of enhancing the investment climate, the
episode served to underline the expectation that entrepreneurs were to invest in
productive activities rather than rent seeking.

6.2. Investment subsidies

In Korea, the chief form of investment subsidy was the extension of credit to large
business groups at negative real interest rates. Korean banks were nationalized
after the military coup of 1961, providing the government with exclusive control
over the allocation of investible funds in the economy. 'Allocation of under-priced
credit [became] by far the most important single instrument of government
microeconomic control' Jones and Sakong, 1980). Korean firms were highly
dependent on external credit, as borrowing made up two-thirds of their cash flow
during 1963-74. According to Jones and Sakong, 'the general bank [lending] rate
has typically been half of the curb-market rate; and second, the real bank rate has
often been negative and generally below even the most conservative estimates of
the opportunity cost of capital'.
86 DANI RODRIK

With bank lending rates below market dearing rates, a rationing mechanism
was needed. Under the Rhee regime, political motives apparendy played a key
role in the allocation of scarce credit. Under Park, however, credit was allocated
on the basis of 'economic' criteria: namely, the priority given to different economic
activities. Deserving users were judged on the basis of their investment plans,
technology, domestic linkages and scale economies. Since credit was more likely to
be awarded to those with some track record, the loan allocations necessarily
favoured established firms, and the chaebol in particular. This explains why, unlike
in Taiwan, expansion of the manufacturing sector has come primarily through the
growth of existing firms, rather than the entry of new firms. Between 1966 and
1971, the value added share of firms with 200 or more employees rose from 58%
to 72% (Table 7). In the 1970s, Korean credit policy became even more partial to
chaebol as the govemment decided to use them as instruments for the government's
industrial diversification strategy (Pack and Westphal, 1986).
Another important manner in which investment was subsidized in Korea was
through the socialization of investment risk in selected sectors. The govemment -
most notably President Park himself- provided an implicit guarantee that the state
would bail out those entrepreneurs investing in 'desirable' activities if
circumstances later threatened the profitability of these investments. 'The Korean
govemment had extensively sociidized the investment risk for selected entrepre-
neurs, and such an arrangement invigorated the animal spirit of the big business
groups, inducing them to indulge in aggressive expansion through the faU-sqfe
government-sponsored investment activities' (Hong, 1993). The shipbuilding
industry is a good example. Without the personal involvement and encouragement
of President Park, Hyundai would not have embarked on or completed what
eventually became one of the world's best shipyards. The govemment guaranteed
the firm's external borrowing, provided extensive subsidies for the infrastructure,
and supplied financijil guarantees to get Hyundai its first order (Amsden, 1989). As
discussed above, when the world shipping industry collapsed in 1975, Hyimdai
could activate the government's contingent subsidy. The subsidy took the form of a

Table 7. Size distribudon of manufacturing firms: Korea

Number of 1966 1971

employees
Number of Percentage Percentage Number of Percentage Percentage
firms of workers of value firms of workers of value
added added
5-49 21013 39.4 24.9 21045 27.9 12.8
50-199 1326 20.8 17.6 1605 18.3 15.2
200 and
above 379 39.8 57.5 762 53.8 72.0

Source: Hasan (1976).

GROWTH POUCY 87

Tmble 8. Tax caredits and reimbursements as a share of

the relevant tax base: Tai%van

Year Percentage

1955 0.4
1956 0.7
1957 0.8
1958 1.5
1959 3.4
1960 5.6
1961 9.8
1962 8.5
1963 12.4
1964 12.4
1965 13.1
1966 14.9
1967 18.0
1968 17.2

Source: Iin (1973, Table 6-5).

regulation forcing Korean refineries to ship oil in Korean-ovmed tankers. Such

implicit investment subsidies were gready reduced during the 1980s, as a
consequence of the financial difficulties experienced by some of the investments
undertaken in the second half of the 1970s (the period of so-called heavy and
chemical industries).
In Taiwan, investment subsidies took different forms. Real lending rates were
generally positive and credit subsidies were much less important. However, public
enterprises did get credit at more favourable terms, and they also served to
socialize investment risk. The most important direct subsidies in Taiwan came in
the form of tax incentives. The Statute for Encouragement of Investment (enacted
in 1960 in conjunction with the Nineteen-Point Programme mentioned above)
represented a 'sweeping extension' (Lin, 1973) of the prevailing tax credit system
for investment. The maximum business income tax paid by enterprises was
reduced to 18% of annual income (from a previous maximum of 32.5%); the tax
holiday for new investments was extended from three to five years; tax exemption
was given to undistributed dividends for reinvestment, to 2% of foreign exchange
earnings, and to proceeds of export sales; 'productive' real estate was made either
exempt from stamp tax and deed tax, or taxable at reduced rates; and payments of
import duties on plant equipment were made deferrable, and payable in
instalments after starting operations. These incentives were further expanded in
1965, at which time the business income tax was reduced in all priority sectors
listed in the investment law, and specified manufacturing sectors (in basic metals,
electrical machinery and electronics, machinery, transporution equipment,
chemical fertilizers, petrochemicals and natural gas pipe) were given complete
88 DAM RODRIK

exemption from import duties on plant equipment. (Information in this paragraph

is taken from Iin, 1973; and Kuo, 1983.)
The quantitative significance of these tax incentives for investment can be
observed from Table 8, which shovv« the ratio of tax credits or rebates to the
relevant tJix base. We observe a sharp rise around 1960 and another one in 1965,
associated with the original statute and its revision, respectively. Not coindden-
tally, these increases are also associated with jumps in the investment-GDP ratio
around the same time.

6.3. Direct co-ordination of investment decisions

In addition to providing subsidies, the Korean and Taiwanese governments played
a much more direct, hands-on role by organizing private entrepreneurs into
investments that they may not otherwise have made. In Taiwan, it was the
government that took the inititJ steps in establishing such industries as plastics,
textiles, fibres, steel and electronics. In Korea, in the words of Amsden (1989),
'[t]he initiative to enter new manufacturing branches has come primarily from the
public sphere. Ignoring the 1950s . every major shift in industrial diversification
in the decades of the 1960s and 1970s was instigated by the state.'
Wade (1990) describes how Taiwan'sfirstplastics plant for PVC was built under
government supervision, and handed over to a private entrepreneur upon
completion in 1957. In 1966, three more private firms began producing PVC. All
four relied on an imported intermediate. Meanwhile, the state-owned Chinese
Petroleum Corporation (CPC) produced ethylene, from which an intermediate
suitable for processing into PVC could be derived at a cheaper price than the
imported intermediate. 'So the government forced the four private producers of
PVC to merge in a joint venture with the Chinese Petroleum Corporation and
another state-owned chemical company, in order to adopt a more eflScient
ethylene-using production method' (Wade, 1990). (While Wade is not e^qilicit on
this, there must have been some scale economies or complementarities that
prevented CPC from unilateridly moving into the production of the ethylene-
based intermediate, without waiting for a commitment from the downstream
producers.) The story illustrates nicely the coordinating role of the government.
A similar account is given regarding fibres:
The government. . decided to oversee the creation of a rayon-making plant as
part of a plan to diversify the textile industry away from cotton fibre. \ ^ t h
much help from US advisors it brought together an American synthetic fibre
company with several local textilers from both public and private firms, and
oversaw negotiations on the terms of the joint venture The resulting
corporation was the largest 'private' firm on the island at the time [1957]
In 1962, this same state-sponsored rayon company, together with a state
GROWTH POUCY 89

financing agency, created another company to make nylon. It started

production in 1964. (Wade, 1990)
Privatefirmssoon followed after this state-led entry into synthetic fibres.
Finally, the role of the Taiwanese state was crucial in the early stages of the
electronics industry. In 1974 the publicly owned Electronic Research and Service
Organization (ERSO) was formed to bring in foreign technology and disseminate
it to local firms. ERSO built the country's first model shop for wafer fabrication
and entered a technology transfer agreement with RCA. It trained engineers, who
later moved to private firms. The strategy led to many private-sector offshoots that
commercialized the technology developed by ERSO (Wade, 1990).
It is interesting to note that the Taiwanese authorities' approach to selecting
industries to nurture in this fashion was based on what Wade calls "engineering
concepts', such as take-off, linkages, gaps, substitutions and incremental
extensions - concepts which have litde place in conventional welfare economics.
Wade mentions that the justification for building a stainless steel plant in the early
1980s was to 'fill a gap in Taiwan's infrastructure'. Similarly, '[d]evelopments in
electronics are being promoted wath the aid of an input-output map which
highlights gaps in the production structure within Taiwan'. This concern with
linkages may sit awkwardly with neoclassical development theory, but it does
resonate with our emphasis on coordination failures.
In Korea, as we have seen, the presence of large conglomerates helped
internalize some of the industrial complementarities that Taiwanese policy-makers
had to nurture through more direct interventions. But the Korean goverrunent
was not hesitant to intervene in order to solve what it perceived to be leirger-scale
coordination problems:
The state masterminded the early import-substitution projects in cement,
fertilizers, oil refining, and synthetic fibres, the last gready improving the
profitability of the overextended textiles industry. The government also kept
alive some unprofitable factories inherited from the colonial period, factories
that eventuaUy provided key personnel to the modem general machinery and
shipbuilding industries, which the state also promoted. The transformation fi-om
light to heavy industry came at the state's behest, in the form of an integrated
iron and steel mill. [The government] was responsible for the Big Push into
heavy machinery and chemicals in the late 1970s. (Amsden, 1989)
The case of shipbuilding has already been discussed in some detail. As in T3uwan,
the government proceeded on the understanding that some industries and
products were more 'strategic' than others because they were the source of linkages
with the rest of the economy. A recent account about how Daewoo got into the
shipbuilding business provides yet another example: 'Mr Kim [the founder of
Daewoo] found himself in shipbuilding in 1978, when the government twbted his
90 DANI RODRIK

arm to take over a near-bankrupt project to build a giant shipyard at Okpo, on

Koje island near the southern port of Pusan. "I did not have a chance to say no,"
says Mr Kim. Indeed, the govemment simply announced the move when he was
out of the country' {The Economist, 26 November 1994, p. 81). The Okpo shipyard
is now "at the heart of [Korea's] achievement' in shipbuilding.

6.4. Use of public investment and public enterprise

Public enterprises played a very important role in enhancing the profitability of
private investment in both countries (perhaps more so in Taiwan than in Korea).
They did so by ensuring that key inputs were available locally for private
producers downstream. In Taiwan, as we have seen, it was common for the state to
establish new upstream industries and then either hand the factories over to
selected private entrepreneurs (as happened in the case of glass, plastics, steel and
cement) or mn them as public enterprises. In Korea, the govemment established
many new public enterprises in the 1960s and 1970s, particularly in basic
industries characterized by a high degree of linkages and scale economies. In both
countries, public enterprises were the recipient of favourable credit terms, as well
as direct allocations from the govemment budget.
Not only did public enterprises account for a large share of manufacturing
output and investment in each country, their importance actually increased during
the critical take-off years of the 1960s. This can be seen clearly in Table 9, where
data on three comparator countries are ailso listed. Public enterprises actually
accounted for a larger share of GDP in Taiwan than in such 'socialist' developing
countries as India and Tanzania.

Table 9. The importance of public enterprise in GDP and investment (%)

Public enterprise share of

Year
GDP Capital formation

South Korea 1963^ 6.7 31.2

1971-2 9.1 21.7
Taiwan 1954-7 11.7 34.3
1958-61 13.5 38.1
1962-5 14.1 27.7
1966-9 13.6 28.0
1970-3 13.3 30.5
1974-7 13.6 35.0

India 1966-9 6.5 29.6

Tanzania 1970-3 12.7 48.2
Argentina 1978-80 4.6 19.6

Sources: Wade (1990, Table 6.2), from original dau in Short (1983), except for public enterprise share
in GDP for Korea, which is from Jones and Sakong (1980, Table 24).
GROWTH POUCY

Jones and Sakong (1980) have analysed in detail the expansion of the public
enterprise sector in Korea. They find that the Korean government had a coherent
set of preferences with respect to where public enterprises should be set up. They
summarize their results thus: 'the industries chosen for the public-enterprise sector
[were] characterized by high forward linkages, high capital intensity, large size,
output-market concentration, and production of non-tradables or import
substitutes rather than exports'. These are exactly the characteristics associated
with a high potential for coordination failure.
The case of POSCO, Korea's state-owned integrated steel mill, is instructive (if
not entirely representative). In the early 1970s, the Korean government was turned
down by the World Bank when it applied for a loan to construct a steel plant. The
World Bjuik's argument was that Korea did not have a comparative advantage in
steel. The government was undeterred and went ahead nonetheless. The
government provided POSCO with capital assistance as well as infrastructure
subsidies (for the construction of water supply facilities, port facilities, an electricity
generating station, roads and a railway line). In addition, the government
supported downstream industries to ensure demand for POSCO's production.
POSCO eventually became, by the World Bank's reckoning, "arguably the world's
most efficient producer of steel' (cited in Wade, 1990), supplying Korean mini-
mills with steel at below world prices. Moreover, the presence of POSCO
stimulated in turn a wide range of upstream industries, ranging from capital goods
to spare parts. Between 1977 and 1984, the local content of POSCO's output rose
from 44 to 75%.

7. HOW COULD INTERVENTION BE IMPLEMENTED EFFECTIVELY?

To any economist with experience in the developing world, what is striking about
the policies discussed in the previous section is their similarity to those commonly
employed in many other, considerably less successful economies. Why have these
interventions, along with many others not specifically discussed (such as
quantitative trade barriers or local content requirements), been successful in
Taiwan and Korea and not elsewhere? One part of the answer to this question has
already been given: the imbalance between a well-educated labour force and a low
endowment of physical capital meant that the return to coordinated investments -
and therefore government policy aimed at coaxing these investments - was quite
high.
This is an important part of the story, but not the entire story. While the initial
human capital advantage may have been a necessary condition for intervention to
work, it was not sufficient. On top, what was required was a competent, honest and
efficient bureaucracy to administer the interventions, and a dear-sighted {lolitical
leadership that consistently placed high priority on economic performance. In
Korea and Taiwan, unlike in so many other developing countries, these additional
92 DANI RODRIK

requirements were present. Why? One important factor was clearly the availability
of relatively skilled labour, enabling the formation of a competent bureaucracy. In
addition, an exceptionally high degree of equality in income and wealth - one of
the other initial conditions mentioned earlier - was important as well.
How exacdy did the latter help? The absence of large inequities meant several
things. First, neither govemment had to contend with powerfiil industrial or
landed interest groups. Such powerful groups had been decimated by the Japanese
occupation (Korea), the settlement by the mainland Chinese (Taiwan), and land
reform (both countries). Therefore, policy making and implementation could be
insulated from pressure group politics. In both countries, the implementation of
grovk^-oriented policies required a number of institutional reforms, including the
centralization of functions previously distributed among multitudes of ministries
and agencies, and the creation of new bureaucracies (see Haggard, 1990). These
institutional reforms could be undertaken relatively autonomously, and with litde
pressure from the push and pull of daily politics. Economic laws £ind regulations
could be written by technocratic elites, with litde concern for their eflFect on
organized pressure groups.
Second, the absence of large-scale inequities meant that governments felt no
immediate need to undertake redistributive policies. The analytical literature on
the political economy of growth suggests that regimes which inherit large
inequalities are constandy under pressure to implement growth-retarding policies
(Alesina and Rodrik, 1994; Persson and Tabellini, 1994). An example is the
pursuit of populist fiscal and microeconomic policies (as in much of Latin America)
which engender high inflation, stop)—go cycles and low growth. The political
leadership in Taiwan and Korea could concentrate on expanding the pie instead.
Third, and related to the above, the fact that the top political leaders were free to
focus on economic goads meant that they could supervise the bureaucracy closely.
This is important because interventionist regimes are prone to two fatal problems
having their origin in the bureaucracy. The first is that interventions naturally
generate opportunities for rent seeking. A weak or poorly supervised bureaucracy
is incapable of reining in rent seeking (or becomes part of it). A strong
bureaucracy, on the other hand, can choke off entrepreneurial incentives by
sticking too closely to the letter of the law and imposing too many cumbersome
restrictions aimed at rooting out rent seeking. In both Taiwan and Korea, the top
political leaders closely monitored the bureaucracy to make sure that the
bureaucrats assisted rather than hindered private entrepreneurship. President
Park, in particular, was famous for his daily involvement in the impilementation of
his economic policies, and his willingness to override the bureaucracy at a
moment's notice when businessmen had legitimate complaints.
Hence, the initial advantage with respect to income and wealth distribution
played an important role in shaping the political landscape in both countries. This
is probably the single most important reason why extensive government
GROWTH POUCY 93

intervention could be carried out effectively, without giving rise to rampant rent
seeking.

8. SOME FREQUENTLY ASKED QUESTIONS

The airguments I have made in this paper are naturally controversial. I have
challenged one of the most widely held beliefs in development economics: namely,
the view that South Korea and Taiwan owe their growth to export-oriented
policies. I have also argued that both economies on balance greatly benefited from
extensive government intervention in markets, a proposition which neoclassical
economists are trained to regard with suspicion. The professionally correct
response to these arguments is, and perhaps should be, a high degree of scepticism.
All the more so since some of my arguments, regarding the existence of a
coordination failure in particular, lack direct empirical confirmation, even if they
are suggestive and broadly consistent with the evidence. Without more detailed
micro- and industry-oriented studies, we will not know with any degree of
confidence whether the hypothesis I have put forward has great relevance to these
two cases. My hope is that this paper will stimulate such research, as well as
providing some clues and a set of organizing principles to guide it.
This section addresses some of the more common reasons for scepticism.

8.1. Can we really distinguish between export and investment strategies?

One objection is that the distinction between a trade-oriented strategy and
investment policies is too sharp, that these governments were preoccupied with
both exports and investment. It is argued that the potential distortions of
investment incentives and other interventions were kept in check by the
requirement that enterprises eventually become competitive in world markets.
Indeed, some of the investment subsidies were in practice contingent on export
performance. Besides, since so much of investment went into exportables, exports
must indeed have been profitable.
Nothing I have said in this paper should be taken to imply that exports did not
play a facilitating role in Korea's and Taiwan's growth miracles. In fact, an
implication of my argument is that without the increase in exports the investment
boom would probably not have taken place. It is possible, as the above indicates,
that exports also made a contribution by acting as a disciplining device on subsidy-
receiving firms.'^ However, none of this is very helpfvd in understanding what
made these countries take off. An adequate analysis has to confront the questions:

However, the sinalytical basis for this argument has, to my knowledge, never been properly laid out. If fulfilling
an export target is the price to be paid for hefty jubsi<^, many firms will willingly pay the price. Thert is nothing
inherently efficient or deiirable about the reiulting exports.
94 DAM RODRK

why did investment and exports respond so vigorously to government policies, and
which of these served as the driving force behind economic growth? My view is
that the investment booms would not have been possible in the presence of gross
policy biases against exports, but that these booms were compatible with a wide
range of trade policy options.

8.2. Doesn't the experience of many other countries show that investment is not
enough for sustained growth?
Investment-led growth irresistibly invites comparison with the Soviet Union and
other Soviet-type economies. Doesn't this experience show that accumulation is far
from enough for sustained growth? It is not appropriate to compare economic
systems lacking markets and private property, and in which both the level and
allocation of investment are determined by central planners, to market systems
where investment decisions are made on the basis of profitability. For all the
government interventions, it is not possible to confuse South Korea and Taiwan
with the German Democratic Republic or the former Czechoslovakia.
Overlooking Soviet-type economies, the causal relationship between investment
and growth should be one of the least controversial propositions in economics. The
empirical correlation is clear enough, as can be seen from the scatterplot in Figure
14. Conceptually as well, all theories of growth are based at least in part on capital
accumulation. That is not to say that all countries which invest a lot have
experienced high growth: Figure 14 shows a number of outliers. But in view of the
0.40

0.35

0.30 * •

a. 0.25
Q
• • • • , • /orea
• • "•'
0.20 • • • Taiwan

0.15

0.10 >

*0.05

-0.04 -0.02 0.02 0.04 0.06 0.08

Growth
Figure 14. Reladonship between growth >ad inveatment, 1960-S5
GROWTH POUCY 95

overall correlation, it is these exceptions that require explanation, not the cases
(like South Korea and Taiwan) which have followed the general rule.

8.3. Isn't the evidence on coordination failures too weak in view of failed
government interventions elsewhere?
As already noted, the case for coordination failures is based on circumstantial,
rather than direct, evidence. This is not the kind of evidence that will move anyone
with strongly held priors on the inefficiency of government intervention.
Traditional arguments against it abound. The 'big push' ideas of the 1950s have
led nowhere. How is it possible for governments to 'pick the winners'? And what
about the exeimple of free-market, yet successful. Hong Kong?
Without disagreeing on the need for more evidence, I would point out a
number of things. First, I have tried to be explicit about the conditions under
which a coordination failure is most likely to exist, and have not treated it as a
generic issue, as in much of the early (and current) literature. There is nothing
implausible about the presence of these conditions in the South Korea and Taiwan
of the 1960s. Second, I have tried to bring a considerable amount of case study
evidence to bear on these issues. In fact, so voluminous is the case study literature
on government interventions (and their apparent success) in these countries that
one may as well regard the main problem as being one of finding an adequate
theory on which to hang this evidence, rather than locating the evidence to
support a particular theory. The framework I have proposed here helps us make
sense of the findings of this case study literature, something that is hard to do with
the convention^ approach.'^
This approach also clarifies why 'picking winners' was not so difficult in the
early years of the Korean and Taiwanese experience. Policy-makers in these
countries only had to look at Japan and more advanced countries to see their
future. Of course, once the catch-up is nearly complete, it becomes more difficult
to play the same game.
Finally, Hong Kong's experience is not as telling as it may seem at first glance -
and not only because Hong Kong is a smaii city state, with significant geographical
and historical advantages in foreign trade. Hong Kong was already a high-
investment country by 1960: its investment ratio stood above 20% (of GDP) in
1960, almost double the figure for Korea and Taiwan at the time. Consequently,
Hong Kong never faced the challenge of raising investment. And the absence of
government policy in this regard reveals itself in an investment ratio that has
remained virtually flat since 1960. Therefore, one might as well read the Hong

" A dear example is the need to square the Korean and Taiwanese governments' emphasis on inter-industry
linkages with the complete neglect of luch linkages in standard welfare economics.
96 DANI RODRIK

Kong evidence as suggesting that an increase in domestic investment requires

government action.

8.4. Doesn't the initial imbalance between human and physlcai capital suggest
an even simpler story of catch-up, with iittie roie for government policy?
Recent theories of growth show that an initial imbalance between human and
physical capital speeds up growth (Mulligan and Sala-i-Martin, 1993). That
growth is higher in economies where human capital is large relative to physical
capital holds even in the absence of increasing returns to scale or other possible
market failures. This applies well to South Korea and Taiwan, and perhaps we do
not need the role of government policy to explain fast growth in these two
countries.
Table 10 suggests, however, that such an imbalance does not necessarily
translate into high growth. I have listed in the table four countries which had
roughly the same human capital advantage (as measured by educational
indicators) in 1960 as did Korea and Taiwan: Dominican Republic, Philippines,
Paraguay and Sri Lanka. None of them experienced the kind of growth rates that
Korea and Taiwan did. It would seem plausible that the difference in outcomes
has much to do with differing government policies in the two sets of countries.

Table 10. Countries with human capital 'imbalance', c. 1960

Per- Primary enrolment ratio Secondary enrolment literacy rate

capita ratio
growth, Predicted Actual Predicted Actual Predicted Actual
1960-89
Dominican
Republic 2.48 0.64 0.98 0.13 0.07 0.39 0.65
Philippines 1.58 0.62 0.95 0.12 0.26 0.36 0.72
Paraguay 2.72 0.65 0.98 0.14 0.11 0.40 0.75
Sri Lanka 1.83* 0.65 0.95 0.14 0.27 0.39 0.75

•1960-85.
Source: Same as Table 4.

9. CONCLUDING REIMARKS

The role of government policy in Korea and Taiwan is open to diverse

interpretations. In principle, one could argue in favour of any of the following
propositions: * (1) government policy was a hindrance, but these countries

With thanks to Gene Grossman, for laying out these different poiitions. I am borrowing his wording here.
GROWTH POUCY 97

overcame it nonetheless; (2) it was irrelevant, not helping growth but not hindering
it either; (3) it was helpful, though not essential (that is, it speeded up a process
which would have happened anyway); (4) it was necessary for the growth
experience of these countries. Propositions (1), (2) and (3) are hard to sustain
because the experience of Taiwan and South Korea was far from exemplary
throughout the 1950s, prior to the concerted efforts made by governments in both
countries to make their economies grow. In fact, it was not uncommon for visiting
economists to despair that these economies would remain basket cases for the
foreseeable future. Hence, the real debate should be over which government
policies made the difference, not whether policy made a difference or how much.
I have argued in this paper that the South Korean and Taiwanese 'miracles' can
best be understood by taking seriously what the two governments thought and said
they were doing: namely, coordinating and encouraging private (and public)
investments with a high degree of linkages within the modem sector. Such policies
had a high payoff because they helped remove coordination failures in economies
where the latent return to investment was already high. A relatively skilled and
educated workforce was a necessary condition. So was a relatively equal
distribution of resources, which endowed the political leadership with insulation
and allowed it to focus on economic growth as the top priority. Export-oriented
policies (and chief among them exchange rate policies) were important in so far as
they enabled a steady rise in imported capital goods. But viewing export
orientation as the clue to the growth puzzle misses the mark by a wide margin.
This approach can explain why so many other developing countries have failed
miserably with government interventions that bear more than a passing
resemblance to those employed by the East Asian countries. South Korea and
Taiwan shared a number of special initial conditions - high levels of educational
attainment relative to income, and equsil distribution of income and wealth - that
these other countries lack. Consequently, the relevance of their experience with
government intervention to other developing economies may well be limited.

Discussion

Gene Grossman
Woodrow Wilson School, Princeton University
It is always a pleasure to read one of Dani Rodrik's papers. They are dear, well
organized and well argued. And they are always provocative.
This paper is no exception. Dani challenges the view that trade policy 'explains'
the success of South Korea and Taiwan. In fact, he takes the orthodoxy head on,
and tries to demolish it entirely. Export orientation could not deserve pride of
place in these stories, he argues, because: the timing of the export booms was not
98 DANI RODRIK

right to account for the growth experiences; the changes in relative prices were not
lzu-ge enough to have such a dramatic effect on growth performance; export
orientation itself has no obvious implications for the rate of investment, which
surely shot up in both countries and evidently played a major role in spurring
growth; and there is no real evidence of productivity spillovers from export
activities to the rest of the economy. For each of these arguments, Dani musters
empirical evidence from the raw data and by referring to the studies of others.
Having disposed of the traditional explanation, Dani sets out to build a new one
to replace it. He argues that Korea and Taiwan had (1) favourable initial
conditions thanks to the high levels of human capital in their labour forces and the
relative equality in their income distributions, and (2) good policy, which overcame
a coordination failure that otherwise would have impeded investment in the
modem industricJ sector. He lays out a theoretic£il structure (informally in the text,
more formally in the appendix) that explains how such a coordination failure could
arise, how the preconditions for its existence Eire consistent with the initial
conditions that prevailed in these countries in the late 1950s, and how government
policy could in principle be used to overcome the meirket failure. Finally, he
discusses the policies that were actually used, argues that their intended purpose in
the eyes of the policy-maikers was exactly the one suggested by the theory, and
shows that the evidence is at least consistent with the view that they were
successfully implemented.
I find the rejection of the export-led growth story to be rather convincing. How
could the trade sector have been the driving force? What was the exogenous
change? The relative price changes were surely too small to affect such a large
movement in resources. And even if productivity in the exportable sector
improved dramatically, this sector was much too small to begin with to generate a
large boost in the growth rate.
Barro and Lee (1993) provide additional corroborating evidence. They estimate
a relationship between growth in per-capita income and a number of country-
specific variables for the decades from 1965 to 1975 and 1975 to 1985. These
regressions, which use data for 85 countries in the first period and 95 in the
second, are similar to those reported by Rodrik in his Table 5, except that the list
of regressors is different. Barro and Lee use as explanatory variables: the beginning
of period log of real GDP per capita; measures of educational attainment by males
and females (entered separately); the log of life expectancy; the investment/GDP
ratio; the ratio of government consumption to real GDP less the ratio of nominal
spending on defence and non-capital expenditures on education to GDP; the log
of (one plus) the black market premium on foreign exchange; and the average
number of successful and unsuccessful revolutions per year. Recognizing the
possible endogeneity of many of these regressors, Barro and Lee estimate their
growth relationship using instrumental variable techniques. Note that the black
market premium here is intended as a proxy for the size of market distortions.
GROWTH POUCY 99

TaUc 11. Sources of growth im Korea and TaKvam

Korea Taiwan
l%5-75 1975-fl5 1965-75 1975-85

Net convergence effect 0.028 0.023 0.026 0.017

Investment/GDP ratio 0.002 0.008 0.003 0.008
Government consun^>tion 0.007 0.014 -0.008 -0.003
Black market premium 0.001 0.005 0.003 0.006
Number of revolutions -0.003 -0.003 0.002 0.002
Fitted value 0.081 0.061 0.060 0.04O
Actual per-capita growth 0.078 0.063 0.061 0.057

Note: GontKbutions to per-c^ita GDP growth measured as deviations from sample means.
Source: Barro and Lee (1993) and additional calculations by Robert Barro.

Using the estimated coefficients from the growth regression, Barro and Lee can
decompose a country's growth experience into its proximate 'sources'. Table 11
shows the results of this decomposition for Korea and Taiwjin.'^ The table shows
the contribution of each variable to the fitted growth rate of real per-capita GDP,
expressed relative to the sample mean for all countries. The 'net convergence
effect' is the combined impact from the initial values of log (GDP), male and
female educational attainment, and log (life expectancy); it corresponds roughly to
what Dani terms 'the initial conditions'.
The table shows that, with the possible exception of Taiwan in the period from
1975 to 1985, the short list of regressors 'explains' most of these two coimtries'
growth experience; perhaps there is no 'miracle' here. Moreover, the biggest
contribution to growth comes from the "net convergence effect', i.e. the fact that
these countries had low initial per-capita GDP levek, but high initial levels of
human capital per worker (and thus, by implication, low initial physical capital to
worker ratios). The relatively modest levels of government consumption ako
seemed to play a role. Apparently, the elimination of price distortions, to the
extent that this is captured in the change in the black market premium, is not a big
part of the story.
The argument that there existed coordination failures that inhibited investment
in the modem manufacturing sectors in Korea and Taiwan until the government
intervened to overcome them is intriguii^, but requires further elaboration and
empirical investigation. I am not sure how one would definitively establish the
existence of a coordination failure, let alone the potential for one in a
counterfactual world where government policy was different. I am naturally

" The Barro and lee paper does not report the decompodtion for individual countries. The table here is based
on cakulations perfonned by Robert Barro, to whom I am grateful.
100 DANIRODRIK

inclined to be suspicious of any argument that says, 'there are important non-
traded inputs that are critical to an entire sector's being whose production entails
such substantial scale economies that profitability requires a very large locd
market'. But this suspicion is nothing more than the usual suspicion economists
have when reference is made to unmeasured, and perhaps unmeasurable,
externalities.
Dani briefly addresses the possibility that the experiences of Korea and Taiwan
could be explained by a much simpler story, such as the one told by Mulligan and
Sala-i-Martin (1993). These authors investigate an endogenous growth model
where the production technology for final output is homogeneous of degree one in
physical and human capital, but there are no non-convexities or multiple
equilibria. Mulligan and Sala-i-Martln show that in the transition phase, the
growth rate rises with the magnitude of the gap between the ratio of physical to
human capital and its steady-state value. This 'imbalance effect' arises from the
fact that gross rates of investment in each type of capital must be non-negative.
Dani rejects this possible alternative explanation because it would have also
predicted 'miraculous' performances in countries like the Dominican Republic and
the Philippines. This is true, but on the other hand, the alternative may be
somewhat more consonant with the results reported in Table 1. The table shows
that, after taking account of the initial high levels of human capital accompanied
by the initial low levels of real GDP, the high ratios of investment to GDP in these
countries apparendy did not contribute much extra to growth. I would guess that
the 'imbalance effect', even if not the entire story, is an important part of it.
I would also draw attention to the recent argument put forward by Ventura
(1994), which seems relevant as well. Ventura notes that in an integrated world
economy, small countries like Korea and Taiwan can accumulate capital at very
high rates for long periods of dme without driving down the marginal product of
capital, because they can continuaUy reallocate resources among sectors with
different factor intensities. In this context, he shows, a small country that is a little
more patient than others can have dramadcadly faster capital accumulation and
superior growth performance for a prolonged period, until it eventually ceases to
be small.
In the concluding section of his paper, Dani lists the various roles that
government policy might have played in the Korean and Taiwanese ejqjerience.
My own personal assessment (which has a large standard error) is described by (3):
I suspect that the governments of Korea and Taiwan implemented wise policies
that speeded up a process that would have happened anyway. But Dani dismisses
this view with reference to the 1950s: if it would have happened, why did it not
start sooner? In one sense he is surely right. Governments can always find policies
that will interfere with growth, and Korea and Taiwan probably had such policies
in place in the early post-war period. So, by removing the impediments, the
change in government policy made growth possible. What I wonder is whether the
GROWTH POUCY

targeted industrial jX)licies, such as the selective investment subsidies, the direct
coordination of investment decisions, and the strategic use of public enterprises,
were necessary for the successful performance. I suspect not, but I really do not
know the answer. Dani's paper highlights this question; future researchers will
surely be inspired by his challenge to provide definitive answers.

Victor Norman
Norwegian School of Economics and Business Administration
There are two main arguments in this paper. One is that the success of South
Korea and Taiwan had little to do with export orientation. The other is that
government intervention, to overcome a coordination failure, had a lot to do with
their success. The arguments are linked, but it is useful to consider the two
separately.

Export orientation and growth

To a traditional trade theorist, the idea of export-led growth might be comforting

eind politically appeziling, but intellectually puzzling. If growth stems from a
growing resource base or better technology, there is no obvious link between trade
and growth, except in so far as a static gain from trade, by producing higher
income, could give higher savings. And even that link is weak, since the gain from
trade is a permanent one, with a correspondingly low marginal propensity to save.
In that perspective, it is not Rodrik's argument which is surprising. The
surprising assertion is the conventional one that Korea and Taiwan (and a number
of other countries) have taken off industrially through large-scale exports.
A puzzle remains, however. Why is the belief in exports as a source of growth so
persistent? And why is a strong correlation between export orientaton and growth
apparently observed in so many countries?
The answer is, I think, that there are good reasons to expect - at least for small
countries - a strong correlation between exports and growth, even in the absence
of a simple causality between the two. The correlation derives from perhaps the
oldest observations in economic theory: namely, Adam Smith's assertion that the
division of labour is limited by the extent of the market. If there are economies of
scale - internal or external to firms - efficient production is only possible if there is
access to markets of sufficient size. In countries such as South Korea and Taiwan,
that simply was not the case. To take one of Rodrik's own examples: the Korean
shipbuilding industry has a world market share of 25%, and much of its efficiency
derives from its scale. Three firms (Hyundai, Daewoo and Samsung) account for
93% of production; and Hyundai alone builds 10% of all new tonnage in the
world (Nagatsuka, 1994). Clearly, production units of this size could not have been
developed on the basis of the internal Korean market for transport equipment.
102 DANI RODRIK

In the Adam Smith view then, market size is a prerequisite for growth. It is not
the came of growth, however. Access to large markets is of no help unless the
internal conditions, such as the quality of the labour force, market infrastructure,
savings, etc., are right. The point is simply that internal conditions are not
sufficient, unless the market is large enough. That may explain why eiqxirts and
growth are strongly correlated for small countries, while countries such as
Germany and the USA could combine industrialization with protection.
Rodrik suggests an entirely different explanation for the correlation between
exports and growth - that growth generated import demand, which in turn
generated exports. That cannot be correct. As shown in Rodrik's Figures 4 and 11
(for Korea) and 5 and 12 (for Taiwan), the export boom preceded the import
boom. For Korea, for example, the ratio of exports to GDP trebled from the early
to the mid-1960s - from a level of around 2% to 6% in 1965, and roughly 7% in
1966-7. Imports of capital goods, on the other hand, did not take off until the
second half of the decade. They amounted to between 2 and 4% of GDP fi-om
1960 to 1966, then rose to around 7% in 1967 and more than 9% in 1968. It takes
exceptional! belief in the foresight of policy-makers if this is to be seen as consistent
with the view that 'thanks to appropriate macroeconomic and exchange rate
policies, export supply was adequate to meet the increase in import demand, and
rose alongside imports'.

Investment and coordination

If my interpretation of the role of exports is correct, it is easier to accept the rest of

Rodrik's story. There could well have been a coordination failure which prevented
industrial take-off and rapid export growth in the 1950s, and it mig^t be that
government intervention to overcome the failure was important. One should ask,
however, why he argues for a complex intervention story (direct intervention to
overcome a coordination failure) when a simple one — a high saving and
investment rate - suffices. As he demonstrates, in the 1960s the rate of investment
rose sharply in both South Korea and Taiwan. He also shows that saving by
publicly owned firms was a major source of higher investment. In Korea, for
example, public enterprises in 1963-4 accounted for 31.2% of capital formation,
while only contributing 6.7% of GDP. Furthermore, the governments in the two
countries stimulated private savings and investment through credit subsidies and
tax incentives. One can argue that these measures were sufficient to initiate a self-
reinforcing process of industrialization. In fact, in Rodrik's own model they would
be sufficient. Why then add a less convincing story about government coordination
of investment projects?
The complex story is particularly puzzling when contrasted with the examples
cited. Rodrik cites Jones and Sakong, who argue that the Korean government had
a 'coherent set of preferences' for public enterprises, favouring industries with
GROWTH POUCY 103

'high fonvzu-d linkages, high capital intensity, large size, output-market

concentration, and production of non-tradables or import substitutes rather than
exports'. He then takes the steel mill POSCO as an example, saying that it is
instructive, but not representative. The examples o{ privatefirmsthat he cites are
also curious relative to the stated government preferences. He makes a point of
government promotion of shipbuilding, which has no forward linkages, is labour
intensive, sells in an almost perfectly competitive market, and is export oriented.
The other Korean example, the Lucky-Goldstar group, was successively in
cosmetics, plastics, telecommunications, oil refining, tanker shipping and
insurance. How that fits in with the "dear government preferences' is unclear to
me. The story could be made consistent if the private firms were dependent on
inputs from public enterprises with the Jones-Sakong characteristics; but that is
equally unclear in the Hyundai and Lucky-Goldstar cases - Korean shipbuilders,
for example, import steel and main engines from Japan or Western Europe.
A central part of Rodrik's explanation is that successful government intervention
was possible in South Korea and Taiwan because the governments were insulated
from pressure groups. He then describes President Park, pointing out that he spent
a lot of time on economic matters and supported growth-oriented bureaucrats and
businessmen. To me, that does not sound like insulation from pressure groups. On
the contrary, policy-makers who are insulated from public opinion, and who
depend on a small group of 'prominent' associates, must be prime candidates for
regulatory capture. Did Edweird IV control the Woodvilles? Was the
military-industrizJ complex in the USA easier to control before it came under
public scrutiny?
The assertion of successful industrial policies in South Korea and Taiwan to me
remains unproven.

General discussion
Several Panel members wondered under exactly which conditions the results
claimed by Rodrik would obtain. Thus there were discussions about whether it was
a story about two sectors (one backward, one modem) or about capital
accumulation. The role of human capital was often explicitly mentioned.
In establishing a favourable climate for investment, George Alogoskoufis
pointed out, the political regime is importarit. It is essential to have a regime that
guarantees that there will not be expropriation. David Begg made a similjir point
regarding the role of the exchange regime, including restrictions to capital
movements, and the risk of partial expropriation through devJiluation and inflation
as well as limits to where to hold incomes from profit. Mathias Dewatripont
stressed the importance of the initial distribution of wealth, which is one of the
preconditions listed by Rodrik. Unequal distribution of wealth may create pressure
104 DANI RODRIK

for redistribution and thereby discourage investment. He also wondered whether

the more recently successful Asian countries had followed the same pattern of
growth and government intervention, or whether South Korea and Taiwan are
exceptions.
Hans-Werner Sinn thought that the paper dismissed too quickly the
Hecksher-Ohlin theory. He observed that this theory is compatible with the
stylized facts used by Rodrik. If the economy starts from a situation where only the
labour-intensive commodity is produced, then allowing capital movements
immediately triggers capital inflows. These inflows have nothing to do with
government intervention; they simply reflect natural scarcities.
Assaf Razin raised another issue. He questioned the treatment of relative prices
of exports as exogenous. On the contrary, he expected that a push for exports
would lead to an exchange rate appreciation. Dani Rodrik responded that it is
possible to treat that relative price as exogenous as long as it is acceptable to treat
Korea as a small country with no market power. Indeijit Singh wondered about
the role of the size of military spending. He noted that a country with less military
expenses should perform better. He also thought that the paper should have
looked at the importance of the homogeneity of society. Much as with equality, a
homogeneous society is often characterized by less social conflict. Alessandra
CaseUa questioned measures used to evaluate equality. She further asked whether
the policy which had been carried out had increased inequality.
Xavier Vives drew a parallel with Japan. In this case, he felt that the
coordination of investment had indeed been important, as stressed by Rodrik. Yet,
in his view, the role of public subsidies has been mostly symbolic as their size has
been quite modest. He was seconded by Michel Keen, who asked Rodrik to
provide more information on the size of investment subsidies.

APPENDIX

The model discussed here is taken from Rodrik (1993), to which the reader is referred for
further details (see also Rodriguez-Clare, 1993). We focus on a small, open economy that
can produce two tradable final goods. Both of these goods are produced under constant
returns to scale. The first of these is a labour-intensive good, requiring labour and capital,
which we associate with the 'traditional' sector. Its unit cost function is given by 9{w, r),
with w and r standing for the wage and rental rates prevailing in the economy. The other
good, produced in the 'modem' sector, uses capital and a range of intermediate goods
(producer services and specialized inputs) that are imperfect substitutes for each other. We
use the Dixit-Stiglitz-Ethier specification for the way that these intermediates enter the
production function of the modem good. In a symmetric equilibrium in which n
intermediate goods are available at price p, the unit cost function of the modem good is
given by (^(r,^"'/'""'*. where cr > 1 is the elasticity of substitution between any two of
the intermediate goods. Note that the productivity of the modem sector is linked to the
number of input varieties available: as n increases, unit costs in the modem sector decline.
GROWTH POUCY 105

As discussed in the text, a key assumption is that the intermediates are non-tradable.
They are produced using labour and under increasing returns to scale. Intermediate-good
production is assumed to be intensive in labour skills, so that the educational attainment of
the workforce is taken to be an important determinant of costs in this sector. The unit cost
function of the representative intermediate is expressed as wX(h)c(z), where A is an index of
the skill level of the workforce (so that k'(h) < 0), and ^ is the output level of the
representative intermediate. Due to increasing returns, (/{z) < 0. The presence of scale
economies implies that each intermediate will be produced by a single firm under
monopolistically competitive conditions. Therefore, when the intermediate sector is active,
p = wX{h)c{z). In addition, marginal costs equal (perceived) marginal revenue, so that
\p — MC]/p=l/cr. These two conditions, together with the separability of the
intermediates' cost function, imply that the ratio AC/MC depends only on z, and can
be written as AC/MC = n{z)- The result is that the equality between marginal cost and
marginal revenue can be written in a simple form:

(Al)

Thisfixesthe output level of each intermediate as a function of the elasticity of substitution

alone. Hence, any change in the scale of the intermediate sector will have to come from a
change in n. Let X and Y stand for the output levels of the modem and traditional goods,
and A" and L for the economy's (fixed) endowments of capital and labour. Let the world
prices of the modem and traditional goods be TT and 1, respectively. The remaining
equations of the model can be expressed in the following form;

0{w, r) > 1, r > 0, r[9{w, r) - 1] = 0 (A2)

<t>{r, prT^I^"-^^) > TT, Z > 0, X[<j>{r, ptT^I^"-^^) - TT] = 0 (A3)

ivX{h)c{z) >p, Z>0, z[wX{h)c{z) -p] = 0 (A4)

eu,{w, r)r+X(h)c{z)nz = L (A5)

</>r{u>, r)r + <t>r{r, / w - i / ( ' ' - ' ) ) Z = K (A6)

^^(r,/>«-'/(<'-'))Z = «^ (A7)
The first three equations are the appropriate complementary slackness conditions that
relate domestic costs to prices. The next two equations are the full-employment conditions,
where we have used the fact that the partial derivatives of the unit cost functions with
respect to factor prices yield unit factor demands. The last equation is the market-clearing
condition for the intermediate-goods sector. There are seven endogenous variables in this
system {w,r,p,n,z,X, T) to be determined by the seven equations above.
A key feature of the model is that the competitiveness of the modem sector depends on
both the skill level of the workforce (h) and the range of domestically produced
105 DANI RODRIK

intermediate varieties (n). For sufficiently low levels of A and K, the modem sector will not
be competitive even when the economy produces the maximum feasible number of
intermediates (which is reached when the entire labour force is employed in the
intermediate-goods sector). When h and K are sufficiently large, on the other hand, the
modem sector will be competitive even when a very small number of intermediate goods is
produced.
But when h is large and A" is low, the economy can have two equilbria: one in which the
economy specializes in the traditional sector and the modem sector remains
uncompetitive, and another one in which the modem sector is competitive and becomes
active. The possibility of multiple equilibria arises fi-om a coordination problem. If the
economy is initially specialized in the traditional sector, it will not pay for any sin^e firm to
enter the modem (or the intermediate-goods) sector at the prevailing factor prices, even
though a large-scale shift of resources in that direction can be both privately and socially
profitable. The reason, in turn, is that there will be demand for intermediates only if a
sufficiently large number of them is being produced. Hence the profitability of being in the
intermediate-goods sector depends on the number of other firms already there.
When there are two competing equilibria, specialization in the modem sector produces
at least as high a level of real income as specialization in the traditional sector (and
generally the inequality is strict). And the rate of return to capital (r) is necessarily higher in
the equilibrium with specialization in the modem sector. This is so for two reasons: (1) as
just explained, the modem sector produces higher incomes and can pay out higher factor
returns; and (2) the modem sector is capital intensive. Government jjolicies that push
resources into the modem sector will improve resource allocation as well as increasing the
return to capital. These results form the basis of the argument in the text.

REFERENCES
Adelman, I. and C. Taft Morris (1967). Society, Politics and Economic Dmlopmmt: A Quantittam Approach, The Johns
Hopkins Press, Baltimore, MD.
Alesina, A. and D. Rodrik (1994). 'Distributive Politics and Economic Growth', Qmrterfy Journal of Economics.
Amsden, A. H. (1989). Asia's Next Giant: South Korea and Late Industrialivtion, Oxford Univer^ty Press, New York.
Barro, R. J. andJ.-W. Lee (1993). 'Losers and Winners in Economic Growth", NBER Working Paper No. 4341.
Bosworth, B., R. Dombusch and R. Laban (1994). The Chilean Economy: PoUgi Lessons and Oullagis, Brookings
Institution, Washington, DC.
Bradford, C. I., Jr (1990). 'Policy Interventions and Markets: Development Strategy Typologies and Policy
Options', in G. Gerefli and D.L. Wyman (eds.), Mam^actimngMiracles: Paths ofIiubtstrialiialioB n latin America aM
East Asia, Princeton University Press, Princeton, NJ.
Ciccone, A. and K. Matsuyama (1993). 'Start-up Costs and Pecuniary Externalities as Barriers to Eoonomic
Development', NBER Working Paper No. 4363.
Collins, S. (1991). 'Saving Behavior in Ten Developing Countries', in B. D. Bemheim a n d J 3 . Shoven (eds.).
National Saving and Economic Performance, University of Chicago Press, Chicago, IL.
and W.-A. Park (1989). 'External Debt and Macroeconomic Performance in South Korea', i n j . Sachs and
S. Collins (eds.), Deixloping County Debt and Economic Pafmncma, vol. 3, University of Chicago Press,
Chicago, IL.
de Melo, J. and S. Robinson (1992). 'Productivity and Externalities: Models of Export-Led Growth', /irirnufiiiMJ
Trade and Economic DenelopmtnL
Elias, V. (1990). Sources of Growth: A Study ofSevm Latin American Ecommms, ICS Press, San Francisco, CA.
Faini, R. (1984). 'Increasing Returns, Non-Traded Inpub, and Regional Devdc^wient', Ecaumic JornmL
Frank, C. R., Jr, K. S. Kim and L. Westphal (1975). Fonip, Trade A « m i aid Ecommc Dmbpmatt: Sou* JCmo,
Columbia University Press, New York.
Haggard, S. (1990). Pathways fiom the Periphery: Vu Politics rfGnwtk in the Nmfy hdustnabcH^ Omtries, Corndl
University Press, Ithaca, NY.
Hasan, P. (1976). Korta: Problms and Issues in a Rapidly Gnwing Ecatomy, The Johns Hopkins University Press,
Baltimore, MD, and London.
GROWTH POUCY 107

Hong, W. (1976). Factm Suf^ and Factor Intensity of Trade in Korea, Korea Development Institute, Seoul.
(1977). 'Trade, Distortions and Employment in Korea', Korea Development Institute, Seoul.
(1993). 'Trade and Development: The Experience of Korea and Taiv«m', in G. Hasson (ed.), InttmatimuU
Trade and Development, Routledge, London.
Hou, C.-M. (1988). 'Strategy for Economic Development in Taiwan and Implications for Developing
Economies', paper presented at the Conference on Economic Development Experiences of Taiwan, Taipei,
&-IOJune.
Jones, Leroy and H Sakong (1980). Government, Business, and Entrepreneurship in Economic Development: The Koretai C
Harvard University Press, Cambridge, MA.
Kim,J.-I. and L. J. Lau (1992). 'The Source of Economic Growth of the Newly Industrialized Countries on the
Pacific Rim', unpublished manuscript, Stanford University, CA.
Kim, K. S. (1988). 'The Timing and Sequencing of Trade Liberalization Policy: The Case of Korea', unpublished
manuscrqjt. World Bank
Knieger, A. O. (1985). 'The Experience and Lessons of Asia's Super Exporters', in V. Corbo et al. (eds.), Export-
Oriented Development Strategies: The Success of Five Newly Industrializing Countries, Westview Press, Boulder, CO, an
London.
Krugm<ui, P. (1991). 'History versus Expectations', Quarterly Jowmal of Economics.
Kuo, S. (1983). The Taiwan Etmontf in Transition, Westview Press, Boulder, CO.
Lin, C.-Y. (1973). Industrialiation in Taiwan, 1946-72: Trade and Import-Suistitution Policies for Developing Countries,
Praeger, New York.
little, I. M. D. (1994). 'Trade and Industrialization Revisited', unpublished paper, Nuffield College, Oxford.
Mason, E. S., M. J. Kim, D. Perkins, K. S. Kim and D. Cole (1980). T?u Economic and Social Modernisation oft/ie
R/public of Korea, Harvard University Press, Cambridge, MA.
Matsuyama, K. (1991). 'Increasing Returns, Industrialization and Indeterminacy of Equilibrium', Quarterly Journal
of Economics.
Mulligan, C. and X. Sala-i-Martin (1993). 'Transitional Dynamics in Two-Sector Models of Endogenous
Growth', Quarterly Journal of Economics.
Murphy, K., A. Shleifer and R. Vishny (1989). 'Industrialization and the Big Push', Journal of Political Econonty.
Nagatsuka, S. (1994). 'Recent changes in the South Korean shipbuilding industry and its future prospects',
JAMRI Report No. 49, Japan Maritime Research Institute.
Pack, H. and L. E. Westphal (1986). 'Industrial Strategy and Technological Change: Theory versus Reality',
Jourruil of Development Economics.
Page,J. (1994). 'The East Asian Miracle: Four Lessons for Development Policy', in S. Fischer andj. Rotemberg
(eds.), MER Macroeconomics Annual 1994, Cambridge, MA.
Persson, T. and G. Tabellini (1994). 'Is Inequality Harmfvil to Growth?' Ameiican Economic Review.
Rodriguez-Clare, A. (1993). 'The Division of Labor and Economic Development', unpublished manuscript,
Stanford University, CA.
Rodrik, D. (1992). 'Closing the Technology Gap: Does Trade liberalization Really Help?', in G.K. HeUeiner
(ed.). Trade Policy, Industrialization and Development: New Perspectives, Oxford University Press, Oxford.
(1993). 'Coordination Failures and Government Policy in Intermediate Economies: A Model with
Applications to East Asia and Eastern Europe', unpublished manuscript, Columbia University, New York.
(1994). 'King Kong Meets Godzilla: The World Bank and The East Asian MiracW, in A. FUhlow, C. Gwin, S.
Haggau-d, D. Rodrik and R. Wade, Miracle or Desigti? Lessonsfiomthe East Asum Expentna, Overseas Development
Council, Washington, DC.
(1995). 'Trade Strategy, Exports, and Investment: Another Look at East Asia', Institute of Policy Reform
Discussion Paper, Washington, DC.
Rosenstein-Rodan, P. (1943). 'Problems of Industrialization of Eastern and South-Eastern Europe', Economic
Journal.
Scitovsky, T. (1954). 'Two Concepts of External Economies', Journal of Political Econonty.
Short, R. (1983). 'The Role of Public Enterprises: An International Statistical Comparison', International
Monetary Fund, Washington, DC.
Trela, I. andJ. Whalley (1992). 'The Role of Tax Policy in Korea's Economic Growth', in T. Ito and A. Krueger
(eds.). The PoUiiail Ecommey of Tax Reform, University of Chicago Press, Chicago, IL, and London.
Tsiang, S.C. (1984). 'Taiwan's Economic Miracle: Lessons in Economic Development', in A.C. Harberger (ed.).
World Economic Ormutli, ICS Press, San Francisco, CA.
Ventura, J. (1994) 'Growth and Interdependence', mimeo.. Harvard University, Cambridge, MA.
Wade, R. (1990). Governing the Mariiet: Economic Theory and the Role of Government m East Asian Industriatioitiai, Princeto
University Press, Princeton, I>JJ.
World Bank (1993). The East Asian Miracle: Economic Growth and Public Policy, Oxford Univeraty Press for the Worid
Bank, Washington, DC.
Young, A. (1993). 'Lessons from the East Asian NlCs: A Contrarian View', NBER Working Paper No. 4482.
(1994). 'The Tyranny of Numbers: Confronting the Statistical Realities of the East Asian Growth
Experience', unpublished paper.