Lecture notes:
James R. Markusen
Department of Economics
University of Colorado
Boulder CO 80309-0256
james.markusen@colorado.edu
Multinational Firms
1.
Review of empirical evidence
2.
Dunning's OLI, joint inputs, firm versus plant-level scale
economies
3.
A model with endogenous multinationals
4.
Pattern of trade in goods and services
5.
Motives for internalization
6.
A model of internalization
MULTINATIONAL FIRMS AND THE NEW TRADE THEORY
Firm and Industry Characteristics
(1) Multinationals are associated with high ratios of R&D relative
to sales.
(2) Multinationals employ large numbers of scientific, technical,
and other "white collar" workers as a percentages of their work
forces.
(3) Multinationals tend to have a high value of "intangible assets";
roughly, market value minus the value of tangible assets such
as plant and equipment.
(4) Multinationals are associated with new and/or technically
complex products.
(5) Evidence suggests that multinationality is negatively
associated with plant-level scale economies.
(6) Multinationals are associated with product-differentiation
variables, such as advertising to sales ratios.
(7) A minimum or "threshold" level of firm size seems to be
important for a firm to be a multinational, but above that level
firm size is of minimal importance.
(8) Multinationals tend to be older, more established firms.
Country Characteristics
(1) The high-income developed countries are not only the major
source of direct investment, they are also the major recipients.
Most direct investment seems to be horizontal.
(2) There has been a major boom of direct investment into the
developing countries in the 1990s, but most of it has gone to
the more advanced LCDs and to China. Little goes to the least
developed countries.
(3) Direct investment stocks have grown significantly faster than
trade flows over the last two decades, even though trade
barriers have fallen dramatically.
(4) High volumes of direct investment are associated with
similarities among countries in terms of relative factor
endowments and per capita incomes, not differences.
(5) A high volume of outward direct investment is positively
related to a country's endowment of skilled labor and
insignificantly or negatively related to its physical capital
endowment.
(6) There is little evidence that direct investment is primarily
motivated by tariff avoidance or measurable transport costs,
(7) There is mixed evidence that tax avoidance and/or risk
diversification are important motives for direct investment.
Some evidence does suggest that political risk discourages
inward investment.
(8) Infrastructure, skill levels, and a minimum threshold level of
per capita income seem to be very important determinants of
direct investment.
(9) There is evidence that agglomeration effects are important in
direct investment. But it is admittedly difficult to distinguish
agglomeration effects from firms being drawn to the same
(unobserved) site-specific resources.
Table 1
Annual growth rate (%), all countries
1986-1990
1991-1996
FDI inflows
FDI stocks
Sales of foreign affiliates
Royalties and fee receipts
24.7
19.8
17.4
21.8
14.4
11.4
6.6
12.9
GDP at factor cost
Exports of goods and non-factor
services
10.8
14.3
6.4
7.4
Table 2
FDI inflows and outflow, share in total
Year
Developed
In out
Developing
in out
CEE
in out
1983-1987
1988-1992
1993
1994
1995
76
78
62
59
65
24
21
35
39
32
0
1
3
3
4
95
93
85
83
85
5
7
15
17
15
0
0
0
0
0
Share of inward FDI to/from developing countries going to:
China
Least developed
inward
(48 countries)
inward
LDC share of Hong
Kong
1984-89
1990-92
1993
1994
1995
10
15
38
39
38
2
2
2
1
1
world GDP
inward
4
5
4
4
4
outward
24
28
54
56
51
UNCTAD world investment report, 1996. Zhang and Markusen,
1996.
Correlation between Foreign Ownership Shares and Characteristics of
Industries (Andrew Dick, JIE 1993) (1982 and 1987 data)
Variable
Pearson Correlation Coefficient
Net Income divided by Assets
(proxy for intangible assets)
0.357
Advertising Intensity
0.283
% of Total Employees in R&D
0.228
Median Plant Shipments/
Industry Average Shipments
(economies of scale?)
0.610
Capital to Value Added Ratio
0.175
Patent Effectiveness
process R&D
product R&D
0.814
0.574
Research Lead Time
process R&D
product R&D
0.785
0.615
Point of Departure for Theory: Firms incur significant costs of doing
business abroad relative to domestic firms in those countries.
Therefore, for a firm to become a multinational, it must have offsetting
advantages.
Dunning (OLI): There are three necessary conditions for firms to be
willing to undertake investments abroad
Ownership Advantage: the firm must have a product or a
production process such that the firm enjoys some market power
advantage in foreign markets.
Location Advantage: the firm must have a reason to want to
locate production abroad rather than concentrate it in the home
country, especially if there are scale economies at the plant level.
Internalization Advantage: the firm must have a reason to want
to exploit its ownership advantage internally, rather than license
or sell its product/process to a foreign firm.
Ownership Advantages, Firm-Specific Assets, and Knowledge Capital
Multinationality related to R&D, marketing, scientific and
technical workers, product newness and complexity, product
differentiation.
MNEs intensive in knowledge capital, knowledge-based assets
(1) services of knowledge capital easily transported to
distant plants
(2) joint input or "public goods" nature of knowledge
capital.
Physical capital intensity by itself should not give rise to
multinationality.
What is being traded? Multinationals are exports of the services
of knowledge-based assets: managerial and engineering services,
financial services, reputations and trademarks.
Location advantages.
Horizontal multinationals producing the same goods and services
in each location: Large markets and high trade costs.
Vertical multinationals geographically fragmenting the
production process by stages: factor-price differences across
countries are linked to the factor intensities of different stages,
low trade costs.
Internalization advantages.
The same joint-input, public-goods property of knowledge that
makes it easily transferred to foreign locations makes it easily
dissipated. Firms transfer knowledge internally in order to
maintain the value of assets and prevent dissipation.
Elements of a simple model.
Firm's technology.
F - firm-specific fixed costs
G - plant-specific fixed costs
c - constant marginal cost per unit
t - shipping or tariff or other cost unit
Multi-plant economies of scale.
F is a joint-input across plants.
Fixed costs of a two-plant firm is less than the fixed costs
of a one-plant national firm.
(F + 2G) < (F +G) + (F + G) = 2F + 2G.
But, this does not immediately imply that all firms will be
multinationals. That will depend on the size of G, the fixed costs
of a second plant, versus t, the unit trade cost.
A vertical dimension is added by allowing F, G, and c to have
different factor intensities; e.g., skilled-labor intensity is higher
for F than for G which is in turn higher than for c.
Markusen and Venables, "Multinational Firms and the New Trade
Theory", Journal of International Economics, 1998
1.
Two Goods, X and Y
2.
Y - competitive, constant returns to scale
3
Two Factors, L and R. R specific to Y sector, L mobile between
sectors
4.
X is homogeneous, produced from labor only. Technology in X
F
G
c
ô
- firm-specific fixed cost in units of labor
- plant-specific fixed cost in units of labor
- constant marginal cost in units of labor
- unit shipping cost in units of labor
5.
X is imperfectly competitive, firms are Cournot competitors.
Markets are segmented.
6.
There are two countries, country h and country f. Factors L and
R are immobile between countries.
7.
There are four firm types, and there is free entry and exit into and
out of firm types.
Type mh - multinational firms which maintain plants in both
countries, firms-specific capital is located in country
h.
Type mf - multinational firms which maintain plants in both
countries, firms-specific capital is located in country
f.
Type nh - national firms that maintain a single plant in country
h. Type h firms may or may not export to country f.
Type nf - national firms that maintain a single plant in country
f. Type f firms may or may not export to country h.
We will model this as a complementarity problem; that is, at set of
inequalities with associated non-negative variables.
But it is analogous to a game where there is a continuum of players.
A firm decides whether or not to enter, and if entry, what type. After
entry firms play a Cournot output game.
An equilibrium is
charaterize by no active or inactive firm wishing to change is strategy.
Production function for Y
á
1&á
Yi ' Liy R i
i ' h, f.
,
(1)
Factor prices
wi ' á ( Liy /R i )á & 1 ,
ri ' (1 & á) ( Liy /Ri )á
i ' h, f.
(2)
Labor demand in country i by type-ni firms
n
n
cXii % (c % ô)Xij % G % F,
i, j ' h, f,
i … j.
(3)
Labor demand in country i by type-mi firms
m
c X ii % G % F,
i ' h, f.
(4)
Labor demand in country i by type-mj firms
m
c X ij % G,
i, j ' h, f.
(5)
Labor market adding up condition
n
n
L̄i ' Liy % ni ( cXii % ( c % ô) Xij % G % F )
%
m
mi ( c Xii
% G % F) %
m
mj ( cXji
(6)
% G)
Income
Mi ' wi L i % ri Ri
i ' h, f.
(7)
Utility
â
1& â
Ui ' Xic Yic ,
n
n
m
m
Xic ' ni Xii % nj Xji % m i X ii % m j X ji
(8)
Y ic ' (1&â)Mi.
(9)
Demand
Xic ' â M i /pi,
Price equals marginal cost conditions
n
n
pi ( 1 & eii ) # wi c
(10)
( Xii )
n
n
pj ( 1 & eij ) # wi (c % ô)
m
pi ( 1 & eii ) # wi c
m
pj ( 1 & eij ) # wj c
( Xij )
(11)
( Xii )
m
(12)
m
(13)
( Xij )
Cournot markup formulas
k
k
eij '
X ij
k
'
Xjc
pj X ij
k ' n, m,
âMj
i, j ' h, f.
(14)
"Supply" functions
n
Xii $ â M i
n
Xij $ â M j
m
Xii
m
$ â Mi
Xij $ â M j
p i & wi c
2
pi
pj & wi (c % ô)
2
pj
p i & w ic
2
pi
p j & w jc
2
pj
(15)
.
,
(16)
(17)
(18)
Free entry conditions
n
n
n
n
( n h)
(19)
n
n
n
n
( nf )
(20)
m
m
m
m
m
m
m
m
ph ehh Xhh % pf ehf Xhf # wh ( G % F )
pf eff Xff % ph efh Xfh # wf ( G % F )
ph ehh X hh % pf ehf X hf # wh ( G % F ) % wf G
(21)
( mh )
pf eff X ff % ph efh X fh # wf ( G % F ) % wh G
(22)
( mf )
If outputs are positive, then using (14)-(18), these free entry conditions can be
expressed as:
â Mh
â Mh
â Mh
â Mh
p h & wh c
2
% Mf
ph
ph & wf (c % ô)
ph
p h & wh c
2
% Mf
ph
p h & wh c
ph
2
% Mf
pf & wh (c % ô)
2
pf
2
% Mf
pf &wf c
p f & wf c
pf
pf
( nh )
(23)
2
# wf ( G % F ) ,
( nf )
(24)
2
pf
p f & wf c
# wh ( G % F ),
# w h ( G % F ) % wf G ,
( mh )
(25)
# wf ( G % F ) % wh G ,
( mf )
(26)
2
These can be expressed more simply as (a > b):
n
(27)
n
(28)
m
(29)
m
(30)
Ðh ' a h M h % b f M f & d h # 0
Ðf ' b h M h % a f M f & d f # 0
Ðh ' a h M h % a f M f & d h & e f # 0
Ðf ' a h M h % a f M f & d f & e h # 0
Change in Total Income: dMh = dMf > 0.
m
m
n
n
d Ðh ' d Ðf > d Ðh ' d Ðf $ 0
Change in the Distribution of Income: dMh = - dMf > 0
n
m
m
n
d Ðh > d Ðh ' d Ðf ' 0 > d Ðf
Change in w: dwf = - dwh> 0
n
m
m
n
d Ðh > d Ðh > 0 > d Ðf > d Ðf
Change in Firm versus Plant Cost Ratio: dF = - dG > 0
m
m
n
n
d Ðh ' d Ðf > 0 ' d Ðh ' d Ðf
Change in Transport Costs: dô > 0
m
m
n
n
d Ðh ' d Ðf ' 0 > d Ðh ' d Ðf
Now let us summarize these results.
Multinational firms will have an advantage relative to type-nh and/or
nf firms when:
(1) The overall market is large.
(2) The markets are of similar size.
(3) Labor costs are similar.
(4) Firm-level scale economies are large relative to plant-level
scale economies. (The added fixed costs of becoming a
multinational firm are low.)
(5) Transport costs are high.
Figures 1, 2, 3, 4
Extension of the model to six firm types. This permits “vertical” firms
which have a headquarters and single plant in different countries.
Characteristics of US transnational Corporations in Manufacturing,
1989
R&D expenditures as a percentage of Sales
Parents
Affiliates
Developed countries
Developing countries
3.33
1.12
1.27
0.30
R&D employment as a percentage of total employment
Parents
5.46
Affiliates
2.42
Skill level (compensation per employee $000)
Parents
Affiliates
Developed countries
Developing countries
38.9
25.2
33.3
9.5
Physical capital intensity ($000 per employee)
Parents
47
Affiliates
by industry of parent
31
by industry of affiliate
31
Assets per employee
Parents
Affiliates
Developed countries
Developing countries
186
114
147
52
Ratio of ratios (Slaughter)
(Parents non-production workers/
parents production workers)/
(affiliates non-production workers/
affiliates production workers)
1.626
Table 1: Selected statistics, unweighted averages
Skilled worker
% of labor force
Seven Developed
Countries
26.0
Ratio of skilled
to unskilled
1.81
All Developed
Fifteen Developing
Countries
11.0
All Developing
Sources:
3.55
Inward FDI
% of GDP
Outward FDI
% of GDP
13.4
16.5
9.1
11.5
19.6
5.2
15.4
4.5
UNCTAD world investment report, 1996 and 1997. GTAP data set, 1995.
Developed countries are: USA, Canada, Japan, Denmark, Germany, Great Britain,
Sweden. Developing countries are: Mexico, Korea, Singapore, Phillipines, Malaysia,
Thailand, Indonesia, China, Countries, Brazil, Chile, Turkey, Venezuela, Columbia,
South Africa, Sri Lanka
Are MNEs predominantly horizontal or vertical?
Characteristics of US Transnational Corporations in Manufacturing,
1989
Trade as a % of Sales
Parents
Total Exports
11.6
Exports to affiliates
5.3
Imports from affiliates
4.2
Affiliates
Total Exports
37.8
Sales to US parents in the US
Industry of affiliate
Industry of parent
11.2
9.0
Imports from parents
Industry of affiliate
Industry of parent
13.2
12.0
Source: World Investment Report 1993, UNCTAD, page 64.
Vertical six-firm-type model
1.
Two Homogeneous Goods, X and Y;
Two Countries, h and f.
Two Factors, unskilled labor: L skilled labor: S.
2.
Y - competitive, constant returns to scale, L intensive
3.
X - imperfectly competitive, increasing returns to scale, S
intensive overall.
"headquarters and "plant" may be geographically separated.
a firms may have plants in one or both countries
4.
There are six firm types, with free entry and exit into and out of
firm types. Regime denotes a set of firm types active in
equilibrium.
Type mh - horizontal multinationals which maintain plants in both
countries, headquarters is located in country h.
Type mf - horizontal multinationals which maintain plants in both
countries, headquarters is located in country f.
Type nh - national firms that maintain a single plant and headquarters
in country h. Type h firms may or may not export to
country f.
Type nf - national firms that maintain a single plant and headquarters
in country f. Type f firms may or may not export to country
h.
Type vh - vertical multinationals that maintain a single plant in
country f, headquarters in country h. Type vh firms may or
may not export to country h.
Type vf - vertical multinationals that maintain a single plant in
country h, headquarters in country h. Type vf firms may or
may not export to country f.
Factor-intensity assumptions: ranked from most skilled-labor-intensive
to least skilled-labor-intensive
Activities
[headquarters only] > [integrated X] > [plant only] > [Y]
Firm Types
[type-m firms] > [type-v and type-n firms]
Operations of a given firm within a Country
[local type-v firm] > [local type-m firm] > [local type-n firm] >
[plant of foreign type-m or type-v firm] > [Y]
What is Traded: X, Y, and headquarters' services, the latter produced
with skilled labor. Factors are not directly tradable.
What is Technology Transfer: Headquarters' services, "blueprints".
They may be complements or substitutes for host-country skilled labor.
This depends on whether branch plants draw skilled labor from the
host-country Y sector or an existing X sector.
Why is Investment Liberalization different from Trade Liberalization:
Imports of X embody technology, but they can only be substitutes for
host-country X production, and thus substitutes for substitutes for hostcountry skilled labor.
Determininants of the equilibrium regime
Total Market Size: Type-m firms have an advantage when the total
market size is large.
Trade costs: Type-m firms have an advantage when ô is large.
Firm versus plant level scale economies: Type-m firms have an
advantage when firm-level costs are large relative to plant-level costs.
Differences in Market Size: Type-m firms have a disadvantage when
the markets are very different in size. A type-n firm headquartered in
the large country or a type-v firm in the other country has the
advantage.
Differences in factor prices: Type-m firms and type-n firms have a
disadvantage when factor prices are very different across countries.
Type-v firms headquartered in the skilled-labor-abundant country with
their plant in the unskilled-labor-abundant country have the advantage.
Figures 5, 6, 7
Specific Case: country f small, unskilled-labor abundant.
focus on country f
Motivated by: Mexico liberalizing with the US
Eastern Europe liberalizing with the EU
Parameterization: Country h three times as large
(S/L)h > 2.5*(S/L)f
HP
IL
TL
FF
-
high pportection: high trade costs, no FDI allowed
investment liberalization: high trade costs, yes FDI
trade liberalization: no trade costs, no FDI
free, free: investment and trade liberalization
Production Regime
HP
IL
TL
Types nh, nf
Types mh, nh Type nh
FF
Types vh, nh
Trade Regime of country f
HP
IL
TL
FF*
export Y
import X
export Y
import X
import S
export Y
import X
import Y
export X
import S
Results for a Specific Case:
Variable, country f
HP
IL
TL
FF
Welfare
0.90
0.94
0.99
1.00
X production
0.16
0.49
0.00
1.00
z
0.87
1.79
0.65
1.00
w
0.90
0.92
1.00
1.00
z/w
0.97
1.95
0.64
1.00
zf/zh
0.91
1.75
0.65
1.00
wf/wh
0.91
0.92
1.00
1.00
Variable, country h
HP
IL
TL
FF
Welfare
1.00
0.99
1.00
1.00
X production
1.26
1.14
1.39
1.00
z
0.96
0.99
0.97
1.00
w
1.00
0.99
1.00
1.00
z/w
0.96
1.00
0.97
1.00
Country f, Policy Criterion
Welfare
FF
>
TL >
IL
>
HP
X production
FF
>
IL
>
HP >
TL
z
IL
>
FF
>
HP >
TL
w
FF
>
TL >
IL
>
HP
z/w
IL
>
FF
HP >
TL
zf/zh
FF
=
TL >
IL
>
HP
wf/wh
IL
>
FF
HP >
TL
>
>
General comment
Trade liberalization and investment liberalization are not substitutes.
1.
Trade liberalization and investment liberalization have opposite
effects on some variables such as factor prices.
2.
Neither trade liberalization nor investment liberalization along
exhausts gains from trade (in the sense of achieving FPE).
3.
For a small, skilled-labor-scarce country, trade and investment
liberalization are "complements", in a welfare sense and often in
a volume-of-trade sense.
Empirical Results (Carr, Markusen, Maskus, AER 2000)
RSALES
Production by affiliates of i parents in country
SUMGPD
Sum if country i's plus country j's GDP
GPDDIFF
Difference between country i and country j's GPD.
GDPDIFSQ
Sq diff between country i and country j's GDP.
SKDIFF
Skilled labor abundance of country i relative to j.
SKDIFSQ
SKDIFF squared
(GDPDIFF*SKDIFF)
INVCJ
Cost of investing in country j
TCJ
Cost of exporting to country j
(TCJ*SKDIFSQ)
TCI
Cost of exporting to country i
RSALES =
Other Variables
B1(SUMGDP)
(+)
B2(GDPDIFSQ)
(-)
B3(SKDIFF)
(+)
B4(GDPDIFF*SKDIFF)
(-)
B5(INVJ)
(-)
B6(TCJ)
(+)
B7(TCJ)(SKDIFSQ)
(-)
B8(TCI)
(-)
DIST
Distance from i to j
Table 1. Results for Real Sales Volume of Affiliates: WLS and Tobit estmination
Variable
WLS
Estimate
(t stat)
Sign as
Predicted?
(marginal
significance)
Tobit
Estimate
(Chi-sq)
Sign as
Predicted?
(marginal
significance)
SUMGDP
13.92
(9.80)
Yes
(0.0001)
15.04
(105.5)
Yes
(0.0001)
GDPDIFSQ
-0.0014
(-8.90)
Yes
(0.0001)
-0.0010
(34.67)
Yes
(0.0001)
SKDIFF
31044
(4.01)
Yes
(0.0001)
61700
(52.98)
Yes
(0.0001)
GDPDIFF
*SKDIFF
-4.27
(-2.12)
Yes
(0.035)
-10.20
(18.81)
Yes
(0.0001)
INVCJ
-455.6
(-3.92)
Yes
(0.0001)
-378.6
(7.98)
Yes
(0.005)
TCJ
190.6
(2.20)
Yes
(0.029)
156.2
(2.28)
Yes
(0.131)
TCJ*SKDIFSQ
-569.9
(-0.41)
Yes
(0.683)
-1264 Yes
(0.57)
(0.573)
TCI
-93.3
(-1.14)
Yes
(0.256)
-122.0
(2.13)
Yes
(0.144)
DIST
-1.34
(-6.63)
?
(0.0001)
-1.48
(41.92)
?
(0.0001)
INTERCEPT
-5381
(-0.42)
(0.676)
-23282
(2.59)
(0.108)
Observations
Adjusted R2
Log Likelihood
509
628
0.60
-5755
Result 1: An increase in the host-country's trade costs will raise
production by affiliates of parent country firms.
Result 2: A bilateral-increase in parent and host-country trade costs:
(A) Weighted least squares: decreases affiliate production, so
trade and investment are "complements".
(B) Tobit: generally decreases affiliate production when the
non-US country is a developing country ("complements")
but increases affiliate production when the non-US
country is another high-income country ("substitutes").
Result 3: A convergence in income (GDP) between the United
States and country j (holding the sum their incomes
constant) increases affiliate sales in both directions.
Result 4:
(A) When the US is parent, an increase in country j's skilledlabor abundance increases US affiliate production in
country j (production by US affiliates is attracted to
skilled-labor abundant host countries).
(B) When the US is host, an increase in country j's skilledlabor abundance increases country j's affiliate production
in the US.
Result 5: Affiliate production is income elastic: a bilateral increase
in parent and host country incomes increases affiliate
production by a greater proportion.
Markusen and Maskus 2000b
RSALESL
RSALESE
Local sales of affiliates of i parents in country j
Export sales of affiliates of i parents in country j
RSALESUS
RSALESF
Export sales back to the US (US outward sample only)
Export sales to third countries (US outward sample only)
SUMGPD
GDPDIFF
GDPDIFSQ
Sum if country i's plus country j's GDP
Difference between country i and country j's GPD.
Squared difference between country i and country j's
GDP.
Skilled labor abundance of country i relative to j.
SKDIFF
(GDPDIFF*SKDIFF)
INVCJ
TCJ
TCI
Cost of investing in country j
Cost of exporting to country j
Cost of exporting to country i
RSALES =
B1(SUMGDP)
(+) (> for local)
B2(GDPDIFSQ)
(-)
B3(SKDIFF)
(+)
B4(GDPDIFF*SKDIFF)
(-)
B5(INVCJ)
(-)
(> for export)
B6(TCJ)
(+)
(- for export)
B7(TCI)
(-)
(> for export)
Other Variables
DIST
Distance from i to j
Table 2
Results
1.
The theoretical model receives support in the empirical
estimation. Both horizontal and vertical motives for foreign
production seem to be present in the data.
2.
Empirical results confirm the hypothesis that the local market
size is more important for production for local sales than for
export sales.
3.
Empirical results confirm the hypothesis that production shifts
relatively in favor of production for local sales when the host is
more skilled-labor abundant and relatively in favor of exports
when the host is skilled-labor scarce.
4.
Empirical results confirm the hypothesis that the ratio of
production for export to production for local sales is negatively
related to host-country investment and trade costs.
5.
There is a quantitative difference between the two-way and USoutward-only samples with respect to the host-country skilledlabor share.
For the two-way sample, production for export is attracted by
host-country unskilled-labor abundance.
A hypothesis that US affiliate production for export is attracted
by host-country unskilled-labor abundance is not supported.
Markusen and Maskus 1999
Explicitly runs a horse face between three models: An unrestricted
model and two restricted versions.
KK model
(KK1)
There are firm-level as well as plant-level scale economies
(KK2)
A single-plant firm may geographically separate
headquarters and plant
(KK3)
Firm-level fixed costs are skilled-labor intensive relative
to plant-level fixed costs and the marginal costs of
production.
VER model
(VER1)
There are no firm-level scale economies
(KK2)
A single-plant firm may geographically separate
headquarters and plant
(KK3)
Firm-level fixed costs are skilled-labor intensive relative
to plant-level fixed costs and the marginal costs of
production.
HOR model
(KK1)
There are firm-level as well as plant-level scale economies
(HOR2) A single-plant firm may not geographically separate
headquarters and plant
(HOR3) Firm-level fixed costs, plant-level fixed costs and the
marginal costs of production all use factors in the
same proportion.
KK model
(1) Both type-H and type-V multinationals can exist.
(2) Multinational are important when countries are similar in size
and in relative endowments, and trade costs are moderate to
high (type-H multinationals).
(3) Multinationals are important when countries differ in relative
endowments, particularly if the skilled-labor abundant country
is small.
VER model
(1) Only type-V multinationals can exist.
(2) Multinationals are important when countries differ in relative
endowments.
(3) Multinationals do not arise between identical countries.
HOR model
(1) Only type-H multinationals can exist.
(2) Multinational are important when countries are similar in size
and in relative endowments, and trade costs are moderate to
high.
Figures 8, 9
Tables 3, 4, 5
Results
The econometric results give strong support to the HOR model, and
strongly reject the VER model.
The coefficient estimates in the HOR model have the right signs and
high statistical significance. An F-test cannot reject the HOR
model's restrictions.
These results tell us what researchers have long believed from casual
empiricism: direct investment is most important between countries
that are similar both in size and in relative endowments.
It is the "hill" of Figures 1c and 2c rather than the "valley" of Figures
1b and 2b that best describes the world.
The VER model clearly should not be taken seriously as a general
description of the world.
Internalization
General Idea: some of the same properties of knowledge capital that
create ownership advantages create internalization advantages.
These arise from the jointness property of knowledge along with
moral hazard, asymmetric information, and the infeasibility of
complete and/or enforceable contracts.
Some internalization models involving the stylized facts on
knowledge capital, product newness and complexity.
(1) A firm is reluctant to reveal its product or process to a
licensee, who may reject the proposal, but now has the knowledge.
But the potential licensee is not going to sign an agreement without
knowing what it is buying.
(2) The licensee knows that the firm may not have an incentive
to truthfully reveal the product's quality.
(3) The newness of the product may create an informational
asymmetry in the opposite direction: the potential licensee may have
a much better idea of how the product will sell in its local market,
while the MNE does not. The licensee extracts rent to reveal the
information.
(4) Bi-lateral uncertainty over stat-up problems, worker
productivity and learning rates.
(5) Knowledge is easily learned by new employees. The
licensee may be able to defect, starting a new firm in competition
with the MNE.
(6) Product quality is an intangible asset. A licensee may have
an incentive to reduce quality, capturing a short-run gain at the
expense of losing the contract.
(7) Difficulties in choosing between costly monitoring and
suffering the costs of moral hazard when employing licensees.
(8) Parties must make relation-specific investments (implies
investments are sunk and cannot be used for other uses).
(9) Differences in objectives and goals between the firm and
the licensee.
Elements of the Model (Markusen, JIE 2000)
(1) The MNE introduces (or attempts to introduce) a new
product every second time periods. Two periods are referred to as a
"product cycle". A product is economically obsolete at the end of
the second period (end of the product cycle).
(2) The probability of the MNE successfully developing a
new product in the next cycle is 1/(1+r) if there is a product in the
current cycle, zero otherwise (i.e., once the firm fails to develop a
new product, it is out of the game). The probability of having a
product in the third cycle is 1/(1+r)2 etc. Ignore discounting.1
(3) The MNE can serve a foreign market by exporting, or by
creating a subsidiary to produce in the foreign market.
(4) Because of the costs of exporting, producing in the
foreign country generates the most potential rents.
(5) But any local manager learns the technology in the first
period of a cycle and can quit (defect) to start a rival firm in the
second period. Similarly, the MNE can defect, dismissing the
manager and hiring a new one in the second period. The (defecting)
manager can only imitate, not innovate and compete in the next
product cycle.
1
Alternatively, r is the discount rate between product
cycles, with discounting within between periods of a cycle
ignored (not elegant I know, but it avoid very messy present
value expressions).
(6) Initially, no binding contracts can be written to prevent
either partner from undertaking such a defection.
(7) Initially, I will assume that the MNE either offers a selfenforcing contract or exports. The possibility that defection occurs
as an equilibrium is allowed later in the paper.
(8) Notation is as follows.
R-
Total per period licensing rents from the foreign country.
E-
Total per period exporting rents (E < R).
F-
Fixed cost of transferring the technology to a foreign partner.
These include physical capital costs, training of the local
manager, etc.
T-
Training costs of a new manager that the MNE incurs if it
dismisses the first one (i.e., if the MNE defects).
G-
Fixed cost that the manager must incur if he/she defects. This
could include costs of physical capital, etc.
Li- Licensing or royalty fee charged to the subsidiary in period i (i
= 1,2).
V-
Rents earned by the manager in one product cycle: V = (R-L1)
+ (R-L2).
V/r- Present value of rents to the manager of maintaining the
relationship.
The manager ("a" for agent) has an "individual rationality"
constraint (IR): the manager must earn non-negative rents. The
manager also has an incentive-compatibility constraint: the manager
must not want to defect in the second period.
(1) (R - L1) + (R - L2) $ 0
IRa
(2) (R- L2) + V/r $ (R - G)
ICa
where V = (R- L1) + (R - L2)
V/r is the present value to the manager of the future rents, if there
are any. (R - G) is the payoff to unilaterally defecting.
The MNE similarly has an "individual rationality" constraint (IR):
the MNE must earn non-negative rents. The MNE also has an
"incentive-compatibility constraint: the MNE must not want to
defect (fire the manager) in the second period.
(3) L1 + L2 - F $ 2E
IRm
(4) L2 $ R - T
Icm
Combine the IC constraints.
(5) R - T # L2 # G + V/r
Firm's objective is to minimize V subject to this incentive
compatibility. Solving this problem yields:
(6) 2R - L1 - L2 = V = r(R - T - G) > 0 (rent share to the
manager)
Result 1:
If R # G + T, the MNE captures all rents in a product cycle,
henceforth referred to as a rent-capture (RC) contract. This
situation occurs when
(1) The market is relatively small.
(2) Defection costs for the MNE (T) are high.
(3) Defection costs for the manager (G) are high.
If R > T + G, there is no single-product fee schedule that will not
cause one party to defect.
Now consider the case where the manager's IR constraint does not
hold; that is, the MNE shares rents with the manager.
Combine (2) and (3).
(7) R - T # L2 # G + V/r
It is clear that the multinational should minimize V. V is therefore
chosen such that (7) holds with equalities, L2 = R - T. Write (7) as
an equality and substitute for V.
(8) R - T - G = V/r = (2R - L1 - L2)/r = (2R - L1 - (R - T))/r
(9) r(R - T - G) = 2R - R + T - L1
(10) L1 = R + T - r(R - T - G)
Adding together L1 and L2 = R - T and subtracting F, we get the
earnings of the MNE
(11) L1 + L2 - F = 2R - F - r(R - T - G) > 0
The MNE has to share some rents with the subsidiary. The
subsidiary earns
(12) 2R - L1 - L2 = V = r(R - T - G) > 0.
The subsidiary earns positive rents, and no potential rents are
dissipated.
Result 2:
If R > G + T, the MNE can credibly offer a long-term commitment,
but must share rents with the subsidiary. This is henceforth referred
to as a rent-sharing (RS) contract. The one-period rents earned by
the subsidiary are smaller as
(1)
(2)
(3)
(4)
r is small (future rents are more valuable)
G is large (the incentive to defect is smaller)
T is larger (the MNEs incentive to defect is smaller).
R is larger (the subsidiary's share increase faster than R).
The one-period rents earned by the MNE are larger when (1), (2),
(3) and (4) hold.
Increases in F reduce profits (but do not dissipate the rents 2R) and
make the MNE worse off and leave the subsidiary unaffected.
Increases in G shift rents toward the MNE (G is never incurred in
equilibrium).
The boundary between the area in which a rent-capture contract is
possible and the area in which a rent-sharing contract is needed is
given by SC in Figure 1. RC occurs to the right, RS to the left of
SC.
MNE chooses exporting over the RC contract if 2R - F < 2E.
Indifference between exporting and the RC subsidiary is given by
the horizontal line in Figure 1, F = 2R - 2E. We label this line EC.
At point A in Figure 1, exporting and a RS subsidiary also yield the
same profits (by transitivity). Beginning at this point, differentiate
the right-hand side of (16) holding it equal to zero, in order to derive
the locus of indifference between exporting and a subsidiary.
(13) dF - rdG = 0
dF/dG = r
The locus of points giving the same profits from a subsidiary as
point A in Figure 1 is given by SE in Figure 1. This line intersects
the vertical axis (G=0) at F = 2R - 2E - r(R-T).
Result 3
Figures 10, 11
(1) Above SE and EC in Figure 1, the MNE chooses
exporting
(2) Below EC and right of SC, the MNE chooses a subsidiary
and captures all rents.
(3) Left of SC and below SE, the MNE chooses a subsidiary
and shares rents.
The exporting region of Figure 1 below EC and above (left of) SE is
a region in which the MNE prefers to dissipate some rents in
exporting rather than share rents with a subsidiary.
Extension: Property rights and contract enforcement can be modeled
by adding a defection penalty. Results shown in Figure 2.
A few summary comments.
The theory of foreign direct investment and multinational location
decisions should be viewed as largely distinct from the theory of
portfolio capital flows.
The theory of foreign direct investment is based on the existence of
firm-specific assets which give firms ownership advantages
necessary to overcome the costs of doing business abroad.
First-specific advantages are often located in knowledge-based
assets that create firm-level scale economies (multi-plant economies
of scale) that encourage horizontal expansion into foreign markets.
Vertical geographic fragmentation of activities is encouraged by
differences in factor intensities among stages of production,
combined with differences in factor prices across countries.2
The same properties of knowledge-based assets that create
ownership advantages, often create internalization advantage,
making it profitable to maintain firm ownership across
geographically dispersed plants rather than use arm’s-length
markets.
Note some confusion in terminology. The activities are vertically
integrated within the firm by definition of being a MNE, but geographically
disintegrated.
2
Figure 1: Equilibrium Regimes, τ = .15
Of
.95
.90
.85
.80
Endowment of Resources
.75
.70
.65
.60
.55
.50
.45
.40
.35
.30
.25
.20
.15
.10
.05
Oh .05 .10 .15 .20 .25 .30 .35 .40 .45 .50 .55 .60 .65 .70 .75 .80 .85 .90 .95
Endowment of Labor
Multinational firms only
Mixed regimes of national and multinational firms
National firms only
Figure 2: Compatative Statics
Of
.95
.90
.80
Countries Smaller: 40%
Endowment of Resources
.85
.75
.70
.65
.60
.55
.50
.45
.40
.35
.30
.25
.20
.15
.10
.05
Oh
.05
.10
.15
.20
.25
.30
.35
.40
.45
.50
.55
.60
.65
.70
.75
.80
.85
.90
.95
Endowment of Labor
Of
.95
.90
Trade Costs Smaller: 10%
Endowment of Resources
.85
.80
.75
.70
.65
.60
.55
.50
.45
.40
.35
.30
.25
.20
.15
.10
.05
Oh
.05
.10
.15
.20
.25
.30
.35
.40
.45
.50
.55
.60
.65
.70
.75
.80
.85
.90
.95
Endowment of Labor
Of
.95
MNE fixed costs higher: 1.75
Endowment of Resources
.90
.85
.80
.75
.70
.65
.60
.55
.50
.45
.40
.35
.30
.25
.20
.15
.10
.05
Oh
.05
.10
.15
.20
.25
.30
.35
.40
.45
.50
.55
.60
.65
Endowment of Labor
Multinational firms only
Mixed regimes of national and multinational firms
National firms only
.70
.75
.80
.85
.90
.95
0.25
0.25
0.20
0.20
0.15
0.15
0.10
0.10
0.05
0.05
0.00
0.00
-0.05
-0.05
-0.10
-0.10
-0.15
-0.15
-0.20
-0.20
-0.25
Production minus labor share in
country h without MNEs (diamonds)
Production minus labor share in
country h with MNEs (squares)
Figure 3: Effect of Multinational Entry on the Location of Production,
Welfare, and the Volume of Trade: NW-SE Diagonal
-0.25
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17
NW-SE Diagonal (countries differ in relative endowments)
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17
NW-SE Diagonal (countries differ in relative endowments)
18 19
Effect of multinational entry on the
welfare of country h
1.20%
1.00%
0.80%
0.60%
0.40%
0.20%
0.00%
18
19
Volume of trade and affiliate sales
450
400
350
300
250
200
150
100
50
0
1
2
3
4
5
6
7
8
9
10 11
12 13
14 15
16 17
18 19
NW-SE Diagonal (countries differ in relative endowments)
Volume of trade - NE model
Volume of trade - MNE model
Affiliate sales - MNE model
Figure 4: Effect of Multinational Entry on the Location of Production,
Welfare and the Volume of Trade: SW-NE Diagonal
0.08
0.08
0.06
0.06
0.04
0.04
0.02
0.02
0.00
0.00
-0.02
-0.02
-0.04
-0.04
-0.06
-0.06
-0.08
-0.08
Effect of multinational entry on the welfare of
country h
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
3.00%
2.50%
2.00%
1.50%
1.00%
0.50%
0.00%
-0.50%
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
200
450
180
400
160
350
140
300
120
250
100
200
80
150
60
40
100
20
50
0
Volume of affiliate sales
Volume of trade in X
SW-NE Diagonal (countries differ in size)
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
SW-NE Diagonal (countries differ in size)
Voume of trade - NE model
Volume of trade - MNE model
Affiliate sales - MNE model
Figure 5: Volume of Affliate Production: 25% trade costs
Affiliate Production
500
450
400
350
300
250
200
150
7
0
3
Skilled Labor
50
11
19
15
100
1
2
3
4
5
6
7
8
9
15 16 17 18 19
10 11 12 13 14
Unskilled Labor
Origin for
Country h
Figure 6: Affiliate Production by h-owned plants in country f
Affiliate Production
500
450
400
350
300
250
200
150
19
100
15
50
11
0
7
3
Skilled Labor
1
Origin for
Country h
2
3
4
5
6
7
8
9
18 19
15 16 17
14
13
10 11 12
Unskilled Labor
Figure 7: Change in Affiliate Production, h plants in f:
25% trade costs minus 5% trade costs in both directions
Change in
Affiliate Production
200
150
100
50
0
-50
-100
-200
Skilled Labor
3
7
11
15
19
-150
1
2
Oriigin for
Country h
3
4
5
6
7
8
9
15 16 17 18 19
10 11 12 13 14
Unskilled Labor
Table 2: Effects of Host-Country Size and Skilled-Labor Abundance on Foreign Affiliate
Production for Local Sale and Export (derviatives evaluated at the mean of independent variables)
Effect on:
$1 billon
increase in
country j's GDP
One percentage
point increase in
SKLJ*
One point
increase in
INVCJ
One point
increase in
TCJ
Local sales of country i
affiiates in country j
(inward and outward data)
$16.7 million
ε = 1.558 **
- $12.1 million
ε = -0.017
- $633.4 mill
$366.6 mill
Export sales of country i
affiliates in j to all countries
(inward and outward data)
$4.7 million
ε = 1.118
- $189.0 million
ε = -0.681
- $277.2 mill
$52.1 mill
Local sales of US
affiliates in country j
(US outward data only)
$22.7 million
ε = 1.044
$280.5 million
ε = 0.599
- $517.8 mill
$314.8 mill
Export sales of US
affliliates in j to the US
(US outward data only)
$1.3 million
ε = 0.213
$8.3 million
ε = 0.064
- $346.6 mill
$207.6 mill
- 211.3 mill
- $8.6 mill
Export sales of US
affliliates in j to other
(US outward data only)
εw =
0.681
$9.0 million
ε = 0.989
εw =
0.062
$12.0 million
ε = 0.061
* By a one-percentage-point increase in SKLJ we mean, for example, an increase from
15% to 16%, not an increase from 15% to 15.15%
** ε denotes elasticity
Figure 8a: Affliate Production, 25% trade costs, KK model
500
400
300
200
19
16
100
origin for country h
19
17
15
13
11
9
7
5
3
1
1
4
7
10
13
0
Skilled
Labor
Unskilled Labor
Figure 8b: Affiliate Production, 25% Trade Costs, VER model
500
400
300
200
19
17
15
13
11
9
7
5
3
1
1
4
7
10
Skilled
Labor
0
13
19
16
100
Unskilled Labor
Figure 8c: Affilliate Production, 25% Trade Costs, HOR model
100
80
60
40
Unskilled Labor
19
17
15
13
11
9
7
5
3
1
1
4
7
10
Skilled
Labor
0
13
19
16
20
Figure 9a: Affiliate production by h-owned plants in f: KK model
1
0.8
0.6
0.4
7
Unskilled Labor
1
4
10
Skilled
Labor
0
13
19
16
0.2
origin for country h
Figure 9b: Affiliate production by h-owned plants in f: VER model
500
400
300
200
19
17
15
13
11
9
7
5
3
1
1
4
7
Skilled
Labor
0
10
19
16
13
100
Unskilled Labor
Figure 9c: Affiliate production of h-owned plants in f: HOR model
70
60
50
40
30
20
Unskilled Labor
19
17
15
13
11
9
7
5
3
1
1
4
7
Skilled
Labor
0
10
19
16
13
10
Table 3. Theoretical Predictions for Three Models ( i = parent, j = host)
Variable
KK
HOR
VER
SUMGDP
+
+
0
GDPDIFSQ
-
-
0
D2*SKDGDPD
-
0
0
D2*SKDSUMG
+
-
+
D1*SKDSUMG
0
-
0
DISTANCE
?
?
?
INVCJ
-
-
-
TCJ
+
+
+
TCI
-
-
-
SUMGDP
= GPDi + GDPj
GDPDIFF
= (GDPi - GDPj)
SKDGDPD
= SKDIFF*GDPDIFF = (SKi - SKj)*GDPDIFF
SKDSUMG
= SKDIFF*SUMGDP = (SKi - SKj)*SUMGDP
D1
= -1 if SKDIFF = SKi - SKj < 0
=0
D2
GDPDIFSQ = (GDPi - GDPj)2
if SKDIFF = SKi - SKj > 0
= 1 if SKDIFF = SKi - SKj > 0
= 0 if SKDIFF = SKi - SKj < 0
D2 is non-zero if the parent country is skilled-labor abundant,
D1 is non-zero if the host country is skilled-labor abundant
Table 4. WLS Estimation Including Investment and Trade Costs (509 obs)
KK
HOR
SUMGDP
15.04
(0.0001)
Yes
15.01
(0.0001)
Yes
GDPDIFSQ
-0.001
(0.0001)
Yes
-0.001
(0.0001)
Yes
VER
D2SKDGDPD
5.39
(0.12)
No
D2SKDSUMG
-6.85
(0.04)
?
-2.94
(0.21)
Yes
D1SKDSUMG
-12.83
(0.0001)
Yes
-12.53
(0.0001)
Yes
DISTANCE
-1.31
(0.0001)
?
-1.23
(0.0001)
?
-2.21
(0.0001)
?
-436.6
(0.005)
Yes
-367.9
(0.002)
Yes
-639.3
(0.0001)
Yes
TCJ
173.9
(0.03)
Yes
149.9
(0.05)
Yes
321.6
(0.001)
Yes
TCI
-90.69
(0.26)
Yes
-85.7
(0.29)
Yes
-85.2
(0.40)
Yes
INTERCEPT
-7597
(0.55)
-10592
(0.40)
43337
(0.0001)
0.61
0.61
6.60
6.65
0.37
87.41
3.34
INVCJ
Adjusted R2
F-Test
Critical F 99%
3.79
(0.16)
Yes
Marginal significance levels are in parentheses:
"Yes" and "No" indicate conformity to predictions.
Variable
Table 5. Tobit Estimation Including Investment and Trade Costs (628 obs)
KK
HOR
SUMGDP
17.4
(0.0001)
Yes
17.23
(0.0001)
Yes
GDPDIFSQ
-0.001
(0.0001)
Yes
-0.001
(0.0001)
Yes
VER
D2SKDGDPD
11.83
(0.003)
No
D2SKDSUMG
-15.05
(0.0001)
?
-6.93
(0.001)
Yes
D1SKDSUMG
-24.51
(0.0001)
Yes
-23.97
(0.0001)
Yes
DISTANCE
-1.48
(0.0001)
?
-1.36
(0.0001)
?
-2.63
(0.0001)
?
-386.9
(0.008)
Yes
-227.2
(0.09)
Yes
-616.3
(0.0004)
Yes
TCJ
134.1
(0.15)
Yes
76.3
(0.40)
Yes
261.2
(0.03)
Yes
TCI
-136.7
(0.10)
Yes
-125.1
(0.13)
Yes
-243.4
(0.02)
Yes
INTERCEPT
-26271
(0.07)
-32791
(0.02)
46477
(0.0001)
-5747
-5751
8.0
7.88
-5933
372.0
14.86
INVCJ
Log-Likelihood
LR Test
Critical Chisq
13.17
(0.0001)
Yes
Marginal significance levels are in parentheses:
"Yes" and "No" indicate conformity to predictions.
Variable
Figure 10: Values of F and G supporting alternative Modes
SC
F
SE
E
A
2R - 2E
EC
RC
2R - 2E - r(R-T-2P)
RS
R - T - 2P
G
Figure 11: Elimination of Contract Enforcement
SC
F
SC'
SE'
E
SE
2R - 2E
EC
RC to E
RS to E
2R - 2E - r(R-T)
RC
RC to RS
RS
R - T - 2P
R-T
G