FDI Entry Mode and the Activity of Affiliates of Multinational Firms∗ Natalia Ramondo UC-San Diego and NBER October 31, 2016

Abstract This short paper documents a novel pattern in the data for Asian countries: The entry mode for Foreign Direct Investment (FDI), in the form of Mergers & Acquisitions (M&A) or Greenfield projects, into a host market is linked to the activity performed by the affiliates in that market. In particular, relatively more FDI entry under greenfield projects is associated with more affiliates engaged in international trade activities (global-value-chain FDI). This is observed at the country-sector and bilateral country-sector level.

∗

This short paper was prepared for the Asian Development Bank under contract No. 125047-S87377.

1

1

Introduction

Multinational Enterprises (MNEs) can choose to enter a foreign market by either buying an existing local firm or by building a new facility. While the former mode of Foreign Direct Investment (FDI) entry is called "Merger and Acquisition FDI" (henceforth, M&A), the latter is commonly called "greenfield FDI" (henceforth, GF). Empirical and theoretical studies dedicated to various aspects of the behavior of MNEs is large, but the literature on FDI entry modes is small. A notably exception is Nocke & Yeaple (2007) which develop a model of FDI entry in which firms choose if entering a market through buying an existing firms or building a completely new firm. They assume that firms are composed of two corporate assets: one embedded in the headquarter and the other embodied in the production division. M&A FDI occurs when a firm buys a production division abroad, while GF FDI occurs when a firm relocate (at a cost) a production division abroad. M&As are understood as trade in corporate assets in order to exploit complementarities given by various sources of heterogeneity: country, industry, and firm level heterogeneity. Their model provides some guidance on which correlations one should look at in the data. In particular, we should observe more GF FDI, relative to M&A FDI, among more productive firms, going to poorer and less distant markets. They confirm these correlations postulated by the model using data on the activity of U.S. multinational firms abroad. More generally, several paper have documented that the dominant mode of MNE expansion is through M&As in the developed world and GF in the developing world, even though M&A FDI is becoming more common also to penetrate developing countries.1 This paper links the mode of FDI entry with the activity that the MNE predominantly carry in a given host market. concretely, using novel data from D&B on the activity of foreign affiliates in Asia and data on GF projects and M&As from the FT FDI Intelligence unit, we document the relationship between FDI entry mode and vertical and horizontal FDI. We ask: Is the choice of the entry mode into a market (and industry) linked to the subsequent function that the MNE gives to its affiliates? We find that the presence of affiliates that are exposed to international trade (i.e., export-platform and vertical FDI, or GVC-FDI) is associated with more GF entry, relative to M&A entry, in a given market and industry. Even though we do not propose a theory about why this may be the case, one can speculate. In particular, one can think of MNEs wanting to acquire domestic firms when their goal is to penetrate a domestic market; the 1

See Nocke & Yeaple (2007), Head & Ries (2008), and UNCTAD (2000).

2

domestic firms would provide strategic assets in the form of local knowledge, in terms of institutions, suppliers, customer base, labor force, and perhaps, local capital markets. On the contrary, if the MNE wants to use that particular market as an export platform, the value of a local partnership decreases, and GF entry becomes a more attractive option.2

2

Results

Our two variables of interest are the following. We first construct the ratio of the number of GF FDI transactions to the number of M&A FDI transactions. Second, we compute the fraction of foreign affiliates that export and import (GVC-FDI), as the fraction of total foreign affiliates. These two variables are computed at the bilateral (host-source) country-sector level; that is, we know the number of transaction types (affiliates’ types) from country s in country h and sector s. The average ratio of GF to M&A FDI across host countries is 3.11, with a median value of 2.36; Bangladesh has the highest ratio of GF entry relative to M&As entry, while the richest countries in the region (Korea, Japan, Australia, and New Zealand) have the lowest (non-zero) ratios. The average ratio of GVC-FDI affiliates is of around 0.30 (and a median of 0.29), reaching a (non-zero) minimum in Australia (0.16) and a maximum in China (0.79), followed by Taiwan and Viet Nam. Table 1 presents both ratios, by country, as well as the position of each country in the ranking of each of the variables. Figure 1 shows the correlation in the raw data of the ratio of GF to M&A FDI and the share of affiliates engaged in GVC-FDI, by receiving country, overall and by sector. Clearly, the relationship is positive: A higher number of GF FDI entry is associated with a higher share of trade-oriented affiliates. Quantitatively, a ten-percent higher ratio of GF to M&A entry ratio is associated with 3.8 percent higher share of GVC-FDI affiliates, which means an increase of almost one-percentage points from the mean GVC-FDI ratio of 30 percent. The positive relation between GF FDI entry and GVC-FDI is strongest in mining. Figure 2 gains observations by showing the relation between the M&A to GF FDI Ratio and the share of GVC-FDI at the bilateral-country level, all and by sector. The positive relationship between GF FDI entry mode and GVC-FDI not only survives, but also gains significance (one percent). As for the relation at the host country level, at the bilateral country level, the mining sector presents the strongest correlation. 2 This speculation resonates with the findings in Guadalupe et al. (2012): in explaining the pattern of foreign M&As in Spain, they find that MNE provides access to global markets to the local partner, which induces these local firms not only to absorb the (better) technology brought by the MNE, but also to innovate further.

3

Figures 3 and 4 show the relation between entry mode and the activities of foreign affiliates for selected source and host countries, respectively. The relation becomes cloudier, but the positive correlation observed in previous figures is preserved, in particular for affiliates from the United States and Japan. We next explore the relationship between FDI entry mode and the activities of foreign affiliates as a function of some characteristics of the host country. Table 2 presents some description of the data, while table 3 presents a more formal regression analysis. In particular, table 2 shows the average ratio of GF to M&A entry and the average share of GVC-FDI, across host countries belonging to two different groups: the group with a value for a particular country characteristic above the median and the group with a value below the median. The table shows that, for instance, on average, MNEs have more affiliates dedicated to international trade and chose more GF entry in poorer countries, smaller countries, and labor-abundant countries. MNEs in Asia have, on average, more horizontal affiliates in countries with a more educated population. These markets are also the ones in which M&A entry is larger vis-a-vis GF entry. Finally, we turn to the regression analysis. In table 3, our dependent variable is the one representing GVC-FDI while our control variable of interest is the ratio of GF to M&A FDI (counts), both in logs. Clearly, the positive relationship between the two variables survives the addition of other country controls and sector fixed effects. Moreover, the relation is significant for all sectors pooled together and for manufacturing; for mining, significance is lost because the number of observations drops drastically. Column 1 says that doubling the number of MNEs choosing GF entry, relative to M&A entry, at the bilateral country-sector (i.e. an increase equivalent to moving from the 90 to the 95 percentile) increases the share of affiliates exposed to trade by almost 10 percent. Similarly, as shown in table 4, we can use as explanatory variable the share of GVCFDI and as dependent variable the share of GF FDI entry, relative to M&A FDI entry. Results entail a similar positive correlation, but quantitatively, the relation is much larger: Doubling the share of GVC-FDI increases the ratio of GF to M&A FDI entry by almost 40 percent.

3

Policy Implications

In previous work that analyzed the factors of attraction of GVC FDI, we concluded that the major factors of attraction for trade-oriented affiliates were labor abundance, low trade barriers (in the form of expedited procedures related to international trade as well 4

as low costs of exporting and importing), as well as an already existing network of domestic firms linked by input-output relations. Poorer countries were more likely to host this type of trade-oriented affiliates, due not only to low labor costs, but also to the presence of special economic zones (SEZ). The existence of these special zones, that may be acting as "shields" from the domestic environment for MNEs, may be the reason why we also found that MNEs locate their trade-oriented affiliates in markets with poorer rule of law and poorer governance environment in general. Investment treaties and trade treaties in general, once trade costs are taken into account, do not seem to particularly affect the location choice of MNEs with respect to the activities of their affiliates. However, countries that are already specialized in more downstream sectors (i.e. closer to the final consumer), regarding their exports, are the ones favored by trade-oriented MNEs affiliates. In this paper we document that GVC FDI is linked to a particular mode of entry by the MNE: greenfield projects. Poorer, smaller, and (unskilled) labor-abundant countries are the ones in which GF FDI, relative to M&A entry, seems dominant.3 That is, horizontal FDI is linked to M&A entry, while GVC FDI is linked to GF entry. Hence, besides the considerations mentioned above to effectively attract GVC FDI, facilitation of the establishment of wholly-owned affiliates seems crucial. This may be particularly important for less developed countries that lack a strong network of domestic firms with who the MNE may be interested in forming partnerships; in other words, firms in these poorer countries may not have much to offer to the MNE to make them attractive for an M&A. But facilitating GF-GVC FDI may be a vehicle to build a denser network of local firms which, through interactions with the MNE can climb the technology ladder and acquire knowledge of how to operate in the global market. Particularly illuminating is the case of Wal-Mart in China, as documented by Head et al (2014). Even though Wal-Mart, at the end, decided not to serve the Chinese market (i.e. horizontal FDI), they kept their "global procurement centers" (GPC) in China, buying local products to export to their other locations around the world. Local Chinese firms got in this way access to the international market: Chinese suppliers whose products were exported through Wal-Mart, eventually, by having their brand known in the rest of the world, started exporting themselves. A final note of caution is in place about our finding in this paper. One cannot discard, however, that the positive correlation found between the presence of GVC FDI and GF entry is a result itself of restrictive policies toward FDI. For instance, take the case of China in the 2000s’. Its regulations regarding entry of MNEs to serve the Chinese market (i.e. horizontal FDI) were high; in particular, foreign firms were required to form a jointventure with a local partner. Since this requirement was not in place if the foreign firm 3

We also find some evidence that the presence of a double-taxation treaty favors GF entry.

5

was export-oriented, most foreign firms entered China through GF FDI, with whollyowned affiliates, and located primarily in SEZs which also had very beneficial tax and tariff incentives.

References [1] Guadalupe, Maria, Olga Kuzmina, and Catherine Thomas. 2012. “Innovation and Foreign Ownership.” American Economic Review. [2] Head, Keith, Ran Jing, and Deborah Swenson. 2014. “From Beijing to Bentonville: Do Multinational Retailers Link Markets?” Journal of Development Economics. [3] Head, Keith, and John Ries. 2008. “FDI as an Outcome of the Market for Corporate Control: Theory and Evidence” Journal of International Economics. [4] Nocke, V., and S.R. Yeaple. 2007. “Cross-Border Mergers and Acquisitions versus Greenfield Foreign Direct Investment: The Role of Firm Heterogeneity.” Journal of International Economics. [5] UNCTAD. 2000. World Investment Report.

6

Figure 1: GVC-FDI and FDI entry mode, country level, by sector.

CHN

1

IND IDN

0

GEO

MYS

KOR

-1

AUSJPN

-2.5

-2

-1.5

-1

-.5

0

2

-2.5

-2

-1.5

-1

-.5

0

MYS

0

IDN

KOR JPN

-.5

0

(log) number of M&A to greenfield, host country level

LKA

THA PHL VNM

2

KHM TWN

1

SGP HKG

CHN

IND

MYS IDN

0

IND

BGD

AUSJPN

KOR

-1

1

KOR

AUSJPN

-2

CHN

(log) frac of aff that export and import, host country level

2

TWN

-1

(log) frac of aff that export and import, host country level

VNM

SGP

-1

IDN

(log) number of M&A to greenfield, host country level

THA

-1.5

MYS

(d) Biz services

PHL

-2

CHN

IND GEO

(c) Mining

AUS

TWN

SGP HKG

(log) number of M&A to greenfield, host country level

HKG

THA PHL VNM KHM

1

TWN

SGP HKG

LKA

0

KHM

BGD

-1

THA PHL VNM

-2

LKA

(log) frac of aff that export and import, host country level

2

BGD

(b) Manufacturing

-2

(log) frac of aff that export and import, host country level

(a) All sectors

-2.5

-2

-1.5

-1

-.5

0

(log) number of M&A to greenfield, host country level

Notes: OLS coefficients for a fitted line are: Upper panels: left = 0.385 (s.e. 0.36); right = 0.385 (s.e. 0.364); lower panel: left = 1.19 (s.e. 0.307); and right panel = 0.381 (s.e. 0.37).

7

-4

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

0 -.5 -1 -1.5 -2 -2.5

(log) frac of aff that export and import, bilateral country-sector level

-2

-1

0

(c) Mining

-4

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

(d) Biz services 0

4

-1

2

-2

0

-3

-2

(log) number of M&A to greenfield, bilateral country-sector level

(b) Manufacturing

-4

-4

(log) frac of aff that export and import, bilateral country-sector level

-4

-3

-2

-1

0

(a) All sectors

-3

(log) frac of aff that export and import, bilateral country-sector level

(log) frac of aff that export and import, bilateral country-sector level

Figure 2: GVC-FDI and FDI entry mode, bilateral country level, by sector.

-4

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

Notes: OLS coefficients for a fitted line are: Upper panel: left = 0.159 (s.e. 0.027); right = 0.163 (s.e. 0.021); lower panel: left = 0.243 (s.e. 0.069); and right = 0.17 (s.e. 0.063).

8

THA

-2

TWN IDN PHL HKG

VNM

KOR

AUS

-3

AUS

-2

-1

0

1

2

3

(log) number of M&A to greenfield, bilateral country-sector level

(c) China 0

TWN

THA

-1

-.5

VNM

HKG

SGP

-1.5

HKG

SGP

MYS

-2

AUS

HKG

IND

JPN

-4

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

0 -1

KOR

IDN TWN

AUS JPN

HKG SGP

PHL

CHN IND

AUS

-2

SGP CHN IND AUS

KOR

-3

-1

MYS

(log) frac of aff that export and import, bilateral country-sector level

VNM

IDN HKGIDN TWN IND KOR

KOR AUS JPN

-4

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

(d) India PHL

0

THA

CHN PHL

LKA SGP PHL CHNTWN MYS THA PHL SGP VNM HKG IND

CHN

THA

CHN JPN IDN LKA VNM SGP MYS

-1

THA CHN TWN VNM PHL SGP

PHL KOR

AUS

SGP

AUS

CHN

-2

MYS

(b) United States

AUS

-3

0

SGPKOR

(log) frac of aff that export and import, bilateral country-sector level

(a) Japan

-2.5

(log) frac of aff that export and import, bilateral country-sector level

(log) frac of aff that export and import, bilateral country-sector level

Figure 3: GVC-FDI and FDI entry mode, selected source countries.

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

Notes: OLS coefficients for a fitted line are: left upper panel = 0.469 (s.e. 0.141); right upper panel = 0.219 (s.e. 0.120); left lower panel = 0.201 (s.e. 0.163); and right lower panel = 0.124 (s.e. 0.162).

9

HKG SGP

CHE AUS

KOR SWE

JPN USA

-2

NLD IRL HKG

IND

GER GBR FRA

-2

0

2

4

(log) number of M&A to greenfield, bilateral country-sector level

0

THA ISR

DNK SWE AUT

BEL NLD

-.5

TUR

FRA

GER SGP SGP AUS

-1

-1

TWN ISR DNK

ESP

NLD

CHE JPN

ESP

IRL

CAN FRA GBR GER KOR USA LUX

-1

0

1

CHN

2

(log) number of M&A to greenfield, bilateral country-sector level

Notes: OLS coefficients for a fitted line are: left panel = 0.011 (s.e. 0.086); right panel = 0.002 (s.e. 0.106).

10

KOR

FIN CHE AUS MEX BRA GER NOR ITA JPN USA GBR IRL FRA DNK

-1.5

GBR KOR ZAF ZAF IDN CAN CAN NORFIN DNK KOR SWE LUX GER JPN ITA AUS NLD CAN IND TWN CHE GER THA USAGBR HKG BEL SGP FRA USA IRL FIN BRA JPN MYS FRA

(b) India

-2

0

CHE

(log) frac of aff that export and import, bilateral country-sector level

(a) China

-3

(log) frac of aff that export and import, bilateral country-sector level

Figure 4: GVC-FDI and FDI entry mode, selected host countries, all sectors.

3

Table 1: GVC-FDI and FDI entry, by country. Host country

rank GVC-FDI

rank GF-M&A ratio

GF-M&A ratio

GVC- FDI

China Taiwan Viet Nam Thailand Malaysia Philippines Indonesia Singapore Cambodia Georgia Korea India Brunei Sri Lanka Hong Kong Japan Australia Bangladesh Armenia Afganisthan Uzbekistan Pakistan New Zealand Nepal Azerbaijan Kyrgyzstan

12 9 7 5 16 6 17 10 8 19 20 14 26 3 13 21 22 2 1 24 18 4 23 11 25 15

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26

2.36 3.17 4.54 5.83 1.98 5.00 1.44 2.94 3.77 1.31 0.58 2.13 n/a 6.58 2.28 0.52 0.48 7.00 7.00 n/a 1.37 6.45 0.27 2.67 n/a 2.00

0.79 0.74 0.70 0.66 0.66 0.59 0.43 0.40 0.37 0.36 0.36 0.35 0.33 0.29 0.24 0.18 0.16 0.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Notes: GF-M&A ratio refers to the ratio of the number of greenfield projects to the number of M&A in a country. GVC-FDI refers to the share of foreign affiliates in a country that both export and import. The rank variables just rank the country with respect to each variable.

11

Table 2: GVC-FDI, FDI entry, and country characteristics, selected variables. GVC-FDI share

GF-M&A FDI ratio

rgdpl low high rgdp low high K-L ratio low high Years of schooling low high Exports upstreamness low high

0.246 0.348

3.943 2.039

0.197 0.396

3.586 2.684

0.218 0.376

4.163 1.973

0.345 0.255

3.914 2.502

0.310 0.283

3.977 1.995

Notes: GF-M&A ratio refers to the ratio of the number of greenfield projects to the number of M&A in a country. GVCFDI refers to the share of foreign affiliates in a country that both export and import. High/low indicates whether the variable referring to a host-country characteristic is above/below the median across all countries. The number showed in each cell corresponds to the average GF-M&A ratio and GVC-FDI share, respectively, among countries in each group.

12

Table 3: Determinants of GVC-FDI. OLS. Dep var (in logs)

log GF to M&A (counts) log dist D(sharing language) D(sharing colonial past) D(RTA) D(DTT) D(BIT) log rgdpl log KL log rgdp rule of law Observations R-squared sample

log of GVC-FDI, bilateral country-sector level (1)

(2)

(3)

0.095** (0.026) -0.067 (0.044) -0.086 (0.077) -0.240* (0.118) -0.118 (0.079) 0.087 (0.103) -0.050 (0.058) 0.921** (0.215) -0.537** (0.206) -0.069** (0.026) -0.604** (0.109)

0.076** (0.025) -0.028 (0.039) -0.061 (0.073) -0.128 (0.081) -0.078 (0.068) 0.032 (0.077) -0.073 (0.054) 0.734** (0.197) -0.284 (0.198) -0.063** (0.022) -0.705** (0.102)

0.049 (0.076) 0.034 (0.137) -0.158 (0.257) 0.018 (0.451) -0.444 (0.295) 0.589* (0.281) -0.095 (0.186) 2.014+ (1.044) -1.387 (1.100) -0.249 (0.150) -0.958* (0.442)

416 0.548 all

266 0.387 manufacturing

38 0.609 mining

Notes: The dependent variable is the number of affiliates with export and import activities, as a share of total affiliates, at the bilateral-country sector level, in logs. The control variable of interest is the number of greenfield (GF) to the number of M&A FDI, in logs, at the bilateral-country sector level. Specification in column a with sector fixed effects. Standard errors, clustered at the host-source country level, in parentheses. Levels of significance are denoted by ** p<0.01, * p<0.05, + p<0.1.

13

Table 4: Determinants of FDI entry mode. OLS. Dep var (in logs)

log gvc fdi log dist D(sharing language) D(sharing colonial past) D(RTA) D(DTT) D(BIT) log rgdpl log KL log rgdp rule of law Observations R-squared sample

Ratio of GF FDI to M&A (counts), bilateral country-sector level (1)

(2)

(3)

0.386** (0.110) 0.397** (0.095) -0.450** (0.152) 0.554** (0.167) -0.247 (0.162) 0.486* (0.192) 0.087 (0.123) 1.486** (0.477) -1.271** (0.465) -0.234** (0.055) -0.684** (0.201)

0.496** (0.181) 0.223* (0.103) -0.681** (0.178) 0.440+ (0.228) -0.312 (0.198) 0.485* (0.224) 0.107 (0.141) 1.530* (0.616) -1.465* (0.593) -0.137* (0.061) -0.534* (0.233)

0.281 (0.478) 0.126 (0.281) -0.513 (0.559) 0.467 (0.635) -1.065 (0.740) 0.264 (0.723) -0.906 (0.754) 4.374* (2.056) -3.644+ (1.851) -0.758* (0.304) -1.577* (0.753)

416 0.384 all

266 0.407 manufacturing

38 0.476 mining

Notes: The dependent variable is the ratio of the number of greenfield (GF) to the number of M&A FDI, in logs, at the bilateral-country sector level. Specifications in columns 1 and 2 with sector fixed effects. Standard errors, clustered at the host-source country level, in parentheses. Levels of significance are denoted by ** p<0.01, * p<0.05, + p<0.1.

14