World Development Vol. 32, No. 9, pp.

1525–1544, 2004
Ó 2004 Elsevier Ltd. All rights reserved
Printed in Great Britain
www.elsevier.com/locate/worlddev 0305-750X/$ - see front matter
doi:10.1016/j.worlddev.2004.05.004

Foreign Direct Investment, Vertical Integration,

and Local Suppliers: Evidence from the Polish
Dairy Sector
LIESBETH DRIES
Katholieke Universiteit Leuven, Belgium

and

JOHAN F. M. SWINNEN *
The World Bank, Washington, DC, USA
Katholieke Universiteit Leuven, Belgium
Available online
Summary. — Studies argue that foreign investment has negative implications for small local
suppliers, as they cannot comply with higher standards or they are laid off to reduce transaction
costs. We analyze the impact of FDI in the Polish dairy sector, a sector dominated by small
suppliers and of crucial importance for poor rural households. The analysis shows that FDI does
not cause a rapid consolidation of the supply base. Instead, foreign companies introduce farm
assistance programs to overcome market imperfections. Through vertical and horizontal spillover
effects, this leads to improved access to finance, increased investments, product quality
improvements, and growth of small local suppliers.
Ó 2004 Elsevier Ltd. All rights reserved.

Key words — Central Eastern Europe, Poland, dairy sector, vertical coordination, FDI

1. INTRODUCTION yet other studies conclude that the impact on

The public debate on globalization has
renewed interest in the effects of foreign direct
investment (FDI). Some see foreign investment
as a beneficial factor that can be an important
source of much needed capital, technology,
knowledge etc. for poorer countries. Others
point at the dangers of multinational compa-
nies crowding out local companies as well as
introducing imperfect competition.
There is a growing empirical literature on the
impact of FDI, which can be separated into
two strands. A first group of studies focuses on
horizontal spillover effects of foreign invest-
ment on domestic firms. The conclusions from
these studies are mixed: some studies find
positive effects (Hu & Jefferson, 2002; Liu,
2002), and others, no significant effect (Kokko,
Tansini, & Zejan, 1996; Konings, 2001), and
1525
local firms is negative (Aitken & Harrison,
1999). The difference in the findings comes from
two opposing effects of FDI. On the one hand,
FDI can introduce new products and technol-
ogies, and domestic companies can benefit from
this through personnel turnover, demonstration
* The authors thank Hamish Gow for discussions on the
issues addressed here and for collaboration in part of the
data collection. Furthermore we are grateful to Ewa
Maciag, Mr. Klekotka, Dr. Uminski and students from
the University of Gdansk for their help with the survey.
This research was supported by the European Com-
mission under its Phare/ACE Research Program (P98-
1007-R). The opinions expressed in this paper are those
of the authors only, and do not necessarily reflect those
of the institutions with which they are affiliated. Final
revision accepted: 12 May 2004.
1526 WORLD DEVELOPMENT
effects and knowledge spillovers. These hori-
zontal spillovers are however, only important if
the technology gap between the foreign and
domestic firms is not too large (Kokko, 1994).
A negative FDI effect can come from FDI
cutting into the local companies’ market share.
Hence, the different findings of the studies
reflect the relative importance of these two
factors in the various countries and sectors.
A second group of studies focuses on vertical
spillover effects. Studies find that foreign firms
facilitate the adoption of new technologies and
can solve contract enforcement problems (Gow
& Swinnen, 1998; Key & Runsten, 1999). Yet
most studies conclude that the impact on local
suppliers is mostly negative, in particular for
small suppliers in developing countries (Dolan
& Humphrey, 2000; Weatherspoon & Reardon,
2003). The latter often cannot comply with the
higher standards and grading requirements for
the supplied products (Farina & Reardon,
2000; Henson, Loader, & Brouder, 2000;
Reardon, Codron, Busch, Bingen, & Harris,
1999). Moreover, foreign investors prefer to
deal with a few large suppliers to minimize
transaction costs, forcing consolidation of the
supplier base and hence separating many small
suppliers from their traditional outlets (Hollo-
way, Nicholson, Delgado, Staal, & Ehui, 2000;
Runsten & Key, 1996; Winters, 2000). Reardon
and Berdegue (2002) show, in the case of retail
investors in Latin America, how this process
can lead to the rapid exclusion of thousands of
small suppliers.
The first objective of this paper is to study the
impact of foreign direct investment on suppli-
ers, and in particular small suppliers, in tran-
sition countries. Many companies in transition
countries are (or were) in severe need of
restructuring and upgrading of capital, tech-
nology, and management. This holds across the
transition world, but is especially pronounced
in those countries which are now most open to
external competition, either because trade
restrictions were liberalized in the transition
process, or, for several Central and Eastern
European countries, because they will soon be
integrated in a single EU market in which they
will have to compete with other EU companies.
At the same time, transition countries, and in
particular those closer to the EU, have received
a large inflow of FDI over the past years. For
these reasons, studying the impact of FDI in
transition countries can provide very useful
insights.
The second objective of this paper is to
analyze and understand the mechanism and
process of how FDI affects the local economy,
and in particular (small) suppliers. A major
problem in transition countries is the break-
down of exchange systems and contract
enforcement mechanisms (Blanchard, 1999;
Konings & Walsh, 1999). Private institutional
innovations have solved these problems in
some countries (Johnson, McMillan, &
Woodruff, 1999; Mcmillan & Woodruff, 1999).
Case studies suggest that foreign investors have
played an important role in this process
through vertical integration (Foster, 1999;
Gow, Streeter, & Swinnen, 2000). At least in
some cases such FDI-induced vertical integra-
tion has contributed to improved access to
finance and inputs, and productivity growth of
suppliers (Gow & Swinnen, 2001). But, so far
there is no representative statistical evidence on
these effects. This paper will be the first to
provide representative evidence on these effects,
and to estimate econometrically the effect of
these institutional innovations on the survival
and growth of small suppliers.
Our empirical analysis uses data from Poland
and specifically from the Polish dairy sector. To
identify both the effects of FDI and the process
through which these effects occur, we collected
data at several levels. More specifically, we
collected data through a series of in-depth
interviews with domestic and foreign owned
companies at the level where the foreign
investment took place (dairy processing and
marketing) as well as through a random survey
of (potential) local suppliers (dairy farms) to
these companies. We also interviewed some
dairy equipment suppliers. In combination the
collected information constitutes a unique
dataset on the impact of FDI on (small) sup-
pliers.
We selected the Polish dairy sector for several
reasons. First, Poland is the largest of the EU
accession countries, yet a small economy in the
world market. Poland produced around 12
million tons of milk in 2000, which represents
2.5% of total production in the world. The
accession of Poland alone would increase total
milk output in the EU with 10% (FAO, 2003).
Yet milk production and the dairy sector have
been severely affected by the economic and
institutional reforms over the past ten years.
Milk production and the number of dairy cows
fell by almost 30% during 1989–96. Productiv-
ity also declined initially but has turned around
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS
since 1992 and since 1997 yields are above their
pre-reform level. 1
Second, agriculture is a very important sector
in the Polish economy and characterized by
unfavorable structures and low incomes.
Almost 20% of the population is employed in
agriculture, mostly on small farms. Poland is
unique among the transition countries in that it
had a mixed institutional structure in agricul-
ture under the Communist regime. Small pri-
vate family farms survived the Communist
collectivization and occupied 76% of total
agricultural land. 2 The remaining land was
used by large-scale state farms. 3 Hence, in
contrast to other Communist countries where
small farms resulted from the fragmentation
and decollectivization of the former collective
farms, both small farms and large farms have a
strong historical and institutional basis in
Poland.
Third, dairy plays an important role in Polish
rural areas since many of the small farms have
at least some milk production. Out of approx-
imately 1.3 million dairy farms, 89% had only
1–4 cows in 1996 (see Table 1). Farms with
fewer than 10 cows produced 75% of Poland’s
milk. Less than 60% of total milk production
was delivered to dairies; the rest was used for
self-consumption or directly sold on the local
market. By 2000, 85% of Polish dairy farms still
had fewer than five cows (GUS, 2001).
Fourth, the dairy sector––both the process-
ing companies and the farms––were (and still
are) in need of substantial restructuring in
order to be competitive on the international
market. In the early 1990s Polish milk pro-
duction was generally characterized by low
productivity and low quality. While the situa-
tion has improved importantly since the mid
1990s, even in 1999 only 20% of the 450,000
producers delivering milk to dairies delivered
exclusively milk of the highest quality (Swedish
Board of Agriculture, 2001). The small scale of
the family farms creates specific investment
problems for upgrading milk quality, as well as
problems for investors in the dairy processing
companies, because of transaction costs of milk
collection.
Fifth, Poland has attracted significant FDI
in the dairy sector, 4 yet at the same time local
companies continue to have a large share of
the market. The liberalization of the Polish
trade system and the privatization of the
processing industry in the 1990s opened the
Polish dairy sector to increased competition
from abroad, allowed Polish exporters to
1527
search for new markets, and allowed foreign
companies to invest in the Polish dairy sec-
tor. 5 By 1999 there had been a total inflow of
US$4.6 billion of foreign investments into the
Polish agri-food sector, 5% of which has gone
to dairy processing and dairy equipment
companies.
The combined impact of privatization and
FDI on the structure of the dairy sector has
been modest (see Table 2): the total number of
dairy processing companies with more than 50
employees decreased by 22% during 1993–99.
The decrease was mainly in the number of
cooperatives, as the number of (noncoopera-
tively-owned) private companies doubled. Yet,
cooperatives still controlled 70% of the dairy
market by 1999. Twenty of the 50 privately-
owned dairies had majority foreign investor
ownership.
In combination, these characteristics make
the Polish dairy sector a very interesting case
and potentially a rich source of insights for the
study of FDI impacts, in particular regarding
vertical spillover effects and the impact on small
suppliers. Moreover, continued FDI in the
Polish dairy sector could have very significant
repercussions for the sector, for the many small
supplying farms, and obviously for rural
welfare and development more generally.
Therefore understanding the impacts is useful
both for the study of Poland and for more
general lessons.
The paper is organized as follows. In Section
2, we discuss the data that were collected both
at the dairy-processing level and through a
unique survey of local milk producers. Next, we
discuss qualitative evidence on the impact of
foreign direct investments and vertical integra-
tion in the dairy sector on supplier restructur-
ing. In Section 4, we present an econometric
analysis of the impact of FDI on survival and
growth of local milk producers. Finally, Sec-
tion 5 draws conclusions.
2. DATA
To identify both the effects of FDI and the
process through which these effects occur,
we collected data through a series of in-
depth interviews with domestic and foreign-
owned dairy-processing companies and
through a random survey of local dairy farms
which are potential suppliers to these compa-
nies.
 1528
Table 1. Structure of private farms in the Polish dairy sector, 1996–2000
Total Herd size (number of cows per farm)
1 2 3–4 5–9 10–19 20–29 30–49 50–99 100–199 >200

WORLD DEVELOPMENT
Number of 1,307,320 547,122 363,004 249,714 126,579 19,281 975 246 218 141 40
farms (1996)
As a percentage of total number of farms
1996 100 41.9 27.8 19.1 9.7 1.5 0.1 0 0 0 0
1998 100 40.6 26.9 19.3 10.6 2.5 0.1 0 0 0 0
1999 100 42.8 26.1 17.8 10.0 2.9 0.3 0.1 0 0 0
2000 100 44.6 24.6 16.9 10.0 3.4 0.4 0.1 0 0 0
% Change +2.7 )3.2 )2.2 +0.3 +1.9 +0.3 +0.1
1996–2000
Source: GUS (2001).
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS
Table 2. Number of dairy companies with more than 50 employees in Poland, 1993–1999
1993 1994
Total 410 332
Cooperatives 352 309
Public companies 30 12
Commercial law companies 28 11
Source: Majewski and Dalton (2000).
(a) Small suppliers
The farm-level data collection focused on
small suppliers and the data were collected in a
2001 survey of 290 dairy-producing rural
households in the Warminsko-Mazurskie
region in the North-East of Poland. Warmin-
sko-Mazurskie is an interesting region for this
analysis because it is an important dairy region
in Poland and because it has a mixture of large-
scale and small-scale farms––unlike some other
regions in Poland. At the start of transition
large-scale state farms (cooperatives were
almost nonexistent in Poland) farmed between
30% and 50% of agricultural land in the
region. 6
We interviewed 290 rural households who
had at least had some dairy production in the
past six years. Specifically, only households
were selected which produced and delivered
milk to a dairy processor in 1995. The survey
therefore also covers households that have
stopped producing and/or delivering milk to a
processor since 1995. By using this methodol-
ogy we have tried to minimize sample selection
bias due to exits.
This survey was performed in the fall of 2001
and included retrospective questions on chan-
ges that had occurred over the previous six
years––more or less the period after the arrival
of foreign investors in dairy companies in the
region. The households were selected randomly
in certain municipalities. As in the rest of
Poland, domestic dairies still far outnumber
foreign-owned dairies. To ensure that the
sample included a considerable number of
farmers that had been in contact with foreign-
owned dairy companies and their policies, we
overrepresented municipalities in the vicinity to
the three foreign-owned dairies in the region
(ICC––Paslek; Warmia Dairy; Kraft/Bel––
Chorzele). 7
Table 3 shows how the 290 households in the
survey deliver (or delivered) milk to 24 different
dairy companies, which greatly vary in size.
1529
1995 1996 1999 Change 93–99 (%)
318 321 320 )22
284 280 270 )24
0 0 0 )100
34 41 50 +79
Around 45% of the households in the survey
supplied milk to foreign-owned dairies, 55% to
domestically-owned.
Most of the so-called farms listed in the
official Polish statistics as dairy farms are
merely households producing for home con-
sumption. They account for the vast majority
of the one- and two-cow ‘‘farms’’ which make
up 70% of the total number of dairy farms in
Poland (Table 1), and 36% of dairy farms in
Warminsko-Mazurski (Table 4). Because of the
focus of our analysis, i.e., to understand how
the changes in the processing sector introduced
by FDI affected the suppliers, our survey con-
centrated on those households which delivered
at least some milk to dairies at the start of the
period covered by the survey (1995). As a
consequence, households with 1–2 cows repre-
sent a smaller group in our survey sample: 3%
in 1995 and 10% in 2000.
Even with this selection focus however, the
vast majority of the farms in the sample are
very small by (West or East) European stan-
dards. The majority of farms in the sample
(57%) had less than 10 cows and 96% of the
farms had less than 20 cows in 1995 (Table 4).
The average size of dairy farms in the sample
was 8.8 cows in 1995 and 10.5 cows in 2000.
(b) Dairy companies
We selected six dairy companies for in-depth
interviews with the management. Key charac-
teristics of the companies are summarized in
Table 5. The selection of the dairy companies
was based on three criteria: FDI, ownership
structure, and size. In terms of foreign invest-
ment, two of the selected companies are
majority foreign-owned, two have important
links to foreign companies, and two are purely
domestic. Four are medium-size companies
(50–70 million liters of milk), one large (420
million liters) and one small (2.5 million liters).
Three are cooperatives, two private, and one a
1530 WORLD DEVELOPMENT

Table 3. Dairy companies in the survey

Dairy Actual # farmers # Farmers in the sample # Farmers in the sample
delivering to this delivering to this dairy in delivering to this dairy in
dairy in 2001 1995 2001
OSM Elblag n.a. 30 20
Nowy Dwor n.a. 0 6
ICC Paslek 461 42 25
Olsztyn 232 4 3
Warmia Dairy 2,600 58 56
Lukta 130 13 7
Ostroda n.a. 4 0
Morag n.a. 3 10
Mostkowo n.a. 1 0
Marowieckie n.a. 1 2
Milejewoa n.a. 1 0
Lecza n.a. 1 0
Mazowsze, Chorzelea 2,500 26 25
Szczytno 500 9 2
Gizycko 2,200 19 26
Nidzicki 540 0 1
Mragowob (2,300) 34 0
Olecko 1,500 28 25
Mlekpol, Grajewo 14,000 1 26
Ostrolecka n.a. 0 5
Kurpie, Baranowo 3,450 3 7
Jezioranya n.a. 2 0
Gdansk n.a. 1 1
Other n.a. 2 1
# % # %
FDI processors total 126 45 106 43
No-FDI processors total 157 55 143 57
Total (%) 283 100 249 100
a
Kraft/bel––Chorzele.
b
Bankrupt in 2001. Mragowo dairy merged with Mlekpol in Grajewo.

Table 4. Share of farms in our survey by size classes and processor

Number of cows per farm
1 2 3–4 5–9 10–19 20 Total
Sample 1995 1.7 1.4 12.8 40.7 39.3 4.1 100
Sample 2000 5.1 5.9 10.3 26.9 35.9 12.4 100
W-Ma 2000 22 13.8 19.1 29.1 13.1 2.9 100
No-FDIb 1995 1.3 1.3 12.1 40.1 42 3.2 100
No-FDI 2000 6.4 3.8 10.2 29.9 31.8 12.7 100
FDI 1995 0.8 0.8 12.7 42.1 38.1 5.6 100
FDI 2000 3.2 5.6 10.3 23.8 42.9 12.7 100
a
Warminsko-Mazurskie region.
b
No-FDI is the group of farmers that were delivering to a domestic dairy company in 1995; FDI includes farmers
delivering to a foreign owned dairy in 1995.

joint venture of a cooperative and a private cies vis-

a-vis their suppliers based on the
company. information collected in both the interviews
In the next section we discuss how the foreign and the surveys. Later on we develop an
investments have affected dairy company poli- econometric model to assess how these changes
 Table 5. Fact sheet on selected Polish dairy companies
Mlekpol Mleczarnia Kurpie Mazowsze ICC Paslek Warmia Dairy
Location Grajewo Paslek Baranowo Chorzele Paslek Lidzbark-Warminski

FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS

Legal structure Cooperative Private company Cooperative Cooperative Joint venture Private company
private and coop
Main products Drinking milk, Yoghurt Cheese, butter Drinking milk, Cheese, butter, Skimmed milk pow-
cream, butter, butter, cheese, drinking milk, der (85% of output)
milk powder, cream, milk for yoghurt powder,
cheese, yoghurt further processing whey powder
Foreign owner No No No No Yes Yes
Since when? 1994 1995
Foreign share (%) 70 100
Home country USA NL
Supply to foreign dairy Hochland (2000) Bel/Kraft (1995)
Number of employees 900 10 200 240 250 310
Annual milk supply (ltr.) 420 mio 2.5 mio 65 mio 55 mio 52.5 mio 70 mio
Does company offer the
following programs?
Credit program Yes No Yes Yes Yes Yes
Input supply program Yes Yes Yes Yes Yes Yes
Agricultural extension Yes No Yes Yes Yes Yes
Veterinary service No No No No No Yes
Bank loan guarantee Yes Yes Yes No Yes Yes
Since when? 1994 1992 1991 1992 1995 1995
Since when do you apply 1994 1999 1999 1995 1995 1999
EU standard classification
system?

1531
1532 WORLD DEVELOPMENT
have affected the survival and growth of small
suppliers to the dairy companies.
3. FOREIGN INVESTMENT, VERTICAL
INTEGRATION, AND SUPPLIER
RESTRUCTURING
(a) Foreign investment and vertical integration
Foreign companies played an important role
in demonstrating the importance of supplier
assistance programs and quality improvement
strategies. For example, when Land O’ Lakes
invested in ICC Paslek in 1994, milk quality of
its supplying farms––as everywhere in the
region––was poor. From the start, the foreign
investor set out a clear strategy to increase the
quality of delivered milk. The strategy included
basic changes and investments. One of the first
changes was to invest in cooling tanks in col-
lection points where small suppliers delivered
their milk. 8 Furthermore, they invested in
agricultural extension to raise small farmers’
awareness of the importance of milk quality
and to teach farmers about basic hygienic rules
in handling the milk. From the beginning, the
foreign investor also required germ count and
cell count tests (in accordance with EU stan-
dard tests for milk quality classification).
Farmers were also allowed to have their milk
tested for antibiotic residues free of charge in
the dairy’s laboratory. This was especially
helpful for farmers who had had a cow disease
at their farm and who needed to make sure that
no antibiotics residue was left in the milk.
Faced with small suppliers unable to make
basic investments and restricted in their access
to basic inputs (such as feed, working capital,
etc.) due to a variety of market imperfections,
foreign investors also introduced other farm
assistance programs, such as input (feed) sup-
ply programs, trade credit, investment assis-
tance programs, etc. (see below). Payments for
this supplier assistance and the loans is done
typically through deductions of the ‘‘milk
check’’ at the time of payment for the milk.
Enforcement of the payments is done by
effectively interlinking output and input mar-
kets: assistance programs are restricted to
farms supplying to the company and part of an
(implicit) contract between the farm and the
company. 9 In other words, if successful, 10 this
process of FDI-induced vertical integration
between supplier and company enhances the
farms’ access to basic inputs, credit, and output
markets, while it ensures timely delivery and
high-quality supplies for the company.
(b) Spillover effects
These initiatives by foreign investors not only
had important direct vertical spillover effects on
their own suppliers, but also much wider spill-
over effects. Local dairy companies quickly
learned about the change in company policies
implemented by foreign owners and they star-
ted to copy the successful programs and adjust
their own company policies. Our interviews
show that by 2001 all the interviewed dairies
have programs that assist their supplying
farms:
––All companies have an input supply pro-
gram: the companies provide suppliers ac-
cess to feed and to inputs for on-farm feed
production, such as seed and fertilizer.
Farmers purchase the inputs through com-
pany shops and the inputs are paid from
the milk checks.
––Five out of six companies assist farms in
investing through credit and investment assis-
tance programs. Investment assistance takes
the form of leasing of milk equipment and
leasing of cows, as well as loans for buying
new or second hand cooling and milking
equipment. Payments are deducted from fu-
ture payments for milk deliveries.
––Five of the dairies provide bank loan guar-
antees for bank loans to farmers. Almost all
bank loans for farm investments have prefer-
ential interest rates (subsidized interest rates
around 5% compared to commercial loans
with interest rates often above 20%). In or-
der to obtain such a loan, the farmer needs
collateral. In many cases, however land or
buildings are not accepted as a bank guaran-
tee. Therefore, most interviewed dairies pro-
vide an additional service to their suppliers
by co-signing the bank loan. In this way
the dairy provides the bank loan guarantee
and facilitates its farmers’ access to bank
credits.
––The companies also provide extension ser-
vices to their suppliers. The only dairy that
did not provide credit assistance programs
or agricultural extension services to its sup-
pliers was the small dairy, Mleczarnia, prob-
ably because it did not have sufficient means
(size).
The conclusion that horizontal spillover
effects have caused a rapid replication of these
programs by domestic companies is confirmed
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS 1533

Table 6. Foreign ownership and financial assistance programs (% of farms delivering)

Total Foreign owned Domestic
Credit program on-farm investments 71.5 71.6 71.4
Credit program cows 72.1 73.9 70.7
Input supply program 78.1 78.9 77.5
Loan guarantee program 37.2 46.2 29.8
Average 71.5 71.6 71.4

by data from our farm survey, as summarized
in Table 6. By 2001, more than 70% of the
farms deliver to companies with input supply
and credit programs, and there is no significant
difference between foreign and local companies.
The only difference is in access to bank loan
guarantees, which is significantly higher for
those delivering to foreign companies (46%)
than for those delivering to local dairies (30%).
(c) Impact on farm investments
The assistance programs had an important
impact on dairy farm investments in the region.
More than three-quarters (76%) of all farms in
the survey made investments in the past 10
years, and 92% of all farms with more than 10
cows (see Table 7). Of those that invested, 58%
used loans, almost half of which came from the
dairies .
These numbers however, underestimate the
impact of the dairy company programs on farm
investment for three reasons. First, dairy loans
are more important for dairy-related invest-
ments than bank loans which are more gener-
ally used: 86% of all loans from dairy
companies are used for investments in enlarg-
ing and upgrading the livestock herd (30%) and
cooling tanks (56%). In contrast, only 29% of
all bank loans are used for these types of
investments.
Second, loans from dairies are only a partial
indicator of the financial assistance offered by
dairies. As explained above, assistance also
includes guarantees for bank loans. Hence, part
of the loans given by the banks is indirectly due
to these loan guarantee programs of dairies.
The importance of the loan guarantee pro-
grams should not be underestimated. Almost
half (45%) of the households who could not
obtain preferential bank loans identified lack of
sufficient collateral as the main reason.
Third, the programs which assist farms in
accessing inputs (mainly feed) enhance invest-
ment indirectly by lowering input costs, or
reducing transaction costs in accessing inputs,
and consequently, through improved profit-
ability.
(d) Impact on quality
The combined impact of these programs is
also reflected in the quality of the milk deliv-
ered. Figure 1 illustrates how the quality of the
milk delivered to the six interviewed dairy
companies improved dramatically over 1996–
2001. While in 1996 only around 30% of the
milk delivered to the dairies was, on average, of
the highest (extra) class of milk quality (the
highest quality by EU standards), this share has
increased to more than 80% by 2001.
(e) Dynamics of spillover effects
The evidence suggests that foreign invest-
ment has played a more important role early on

Table 7. Investments and loans of farm households

Size Invests Uses loan to Uses dairy Uses bank Uses dairy loan Uses bank loan
(# of (% of invest (% of A) loan (% of B) loan (% of B) (% of A) (% of A)
cows) total)
A B C D E F
1–5 52 54 41 50 21 26
6–10 78 51 43 70 22 36
>10 92 74 43 75 31 54
All 76 58 43 69 25 40
1534 WORLD DEVELOPMENT

Share of Extra Class Milk in Total (%)

100
90
80
70 Mlekpol
60
Mleczarnia
50
Kurpie
40
Mazowsze
30
ICC Paslek
20
10 Warmia Dairy
0
1996 1998 2001

Figure 1. Change in share of highest quality milk (EU standard) by dairy company.

in the transition as an initiator of change and companies we interviewed had on-farm cooling
institutional innovation. We found no signifi- tanks: the share of suppliers with on-farm
cant difference in 2001 between assistance pro- cooling tanks was 3% on average for the four
grams provided by foreign-owned companies domestic companies and only 9% on average
and domestic dairies, except for the loan even for the two foreign-owned companies. By
guarantee programs, which were more exten- 2001, the share had grown to 51% on average
sively provided by the foreign dairies. for the foreign-owned companies, while only to
The survey data confirm that important 22% on average for the domestic companies.
convergence has occurred in milk quality
between domestic and foreign-owned dairies (f) Impact on survival and growth of small
and their suppliers as spillover effects in com- suppliers
pany programs have multiplied. In 1995 the
supply of extra class milk was significantly How did the opening of the dairy sector to
larger among farmers delivering to foreign- foreign investment, competition and increased
owned dairies (58%) than among farmers quality requirements affect the survival and
delivering to domestic dairies (38%). By 2000 growth of dairy farms? Were small dairy
however this gap had almost disappeared: 83% farmers forced out of the market or did they
of foreign versus 79% of domestic dairy sup- manage to survive and upgrade their farms
pliers (see Figure 2). with the assistance of the programs discussed
Adjustments and convergence are slower above?
when they require important investments. In Our survey data show that of the 283
1996 very few of the suppliers to the six dairy households delivering milk to dairy companies

90
Share of farms with Extra class milk

80 FDI
70 no-FDI
60
50
40
30
20
10
0
1995 2000

Figure 2. Change in share of highest quality milk (EU standard) in the farm survey.
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS
35
30
25
Frequency
20
15
10
0
1 3
Figure 3. Size distribution of dairy farms in total survey sample. Each data point gives the average frequency of farms in
three consecutive size classes of which the data point itself is the size class in the middle. For example, the data point for
herd size three signifies the average number of farms with herd sizes 2, 3 and 4.
in 1995, 36 (13%) stopped delivering milk dur-
ing 1995–2000. Ten of them (4%) stopped
producing altogether while the rest (9%) kept
some cows for home consumption. Hence, 87%
continued delivering to dairies despite radical
restructuring of the dairies and tightened
quality demands. Moreover, most of those who
stopped delivering may have stopped anyhow:
the average age of those who stopped is 56
years, compared to 45 years for the entire
sample. In fact, the average annual decline in
delivering farms in our sample ()2.6%) is less
than the average annual decline in agricultural
employment in Poland over the same period
()3.3%).
The size distribution changed importantly
(see Figure 3). Three-quarters of the house-
holds (211) had between four and 12 cows in
1995. The share of farms in the 4–12 cow cat-
egory has reduced significantly with about the
same amount upgrading to a larger size and
falling back to smaller, presumably subsistence,
farms producing for home consumption. More
specifically, of the 211 household farms, 135
(65%) had still between four and 12 cows in
2000; 35 (17%) had less than four cows in 2000,
while 41 (19%) had more than 12 cows in 2000.
Farmers with growing farms were significantly
younger (42 years on average) than those whose
farm size declined (51 years on average).
In summary, these data suggest that most
small dairy farms have survived this restruc-
turing process well. They have not been cut out,
made important investments, upgraded their
farm business, and improved quality. Those
who fell out were mostly older farmers that
may have gone out of business anyhow.
1535
1995
2000
5 7 9 11 13 15 17 19 21 23 25
Herd size
In the next section we further analyze this
restructuring process. We econometrically esti-
mate the impact of various factors, among
which the ownership structure of the dairy
company and assistance programs offered to
suppliers, on the survival and growth of local
suppliers.
4. ECONOMETRIC EVIDENCE
(a) Model and variables
To complement our qualitative insights and
econometrically to quantify the direct effect of
foreign investment and the assistance programs
on small suppliers, we estimate a model based
on the firm growth literature. This literature
starts from the ‘‘law of proportionate effects’’
or Gibrat’s law, stating that firm growth rates
are independent of initial firm size. Following
Evans (1987), Hart and Oulton (1996) and Hall
(1997) the supplier growth relationship is
specified as follows.
d
Si;t ¼ ½F ð:; Si;t0 Þ
ðSi;t0 Þð1 þ ei;t Þ ð1Þ
where Si;t and Si;t0 denote the size of supplier i
respectively at time t, the current period and t0 ,
the initial period. ½F ð:; Si;t0 Þ
is a function of size
and some other variables and d is the time
interval over which growth is measured or in
other words, d ¼ t  t0 . Finally, et is the pro-
portionate rate of growth between t0 and t. If d
is small and we express Eqn. (1) in logarithms,
we obtain the following general growth func-
tion.
1536 WORLD DEVELOPMENT
lnðSi;t Þ  lnðSi;t0 Þ
¼ ln½F ð:; ðSi;t0 ÞÞ
þ ui;t
d
Using (2), we estimate the impact of foreign
investment and supplier assistance programs on
supplier growth with the following model:
lnðSi;t Þ  lnðSi;t0 Þ Xk Xl
¼ a0 þ aa Ai;a;t0 þ
d 1995, and takes the value of zero otherwise.
a¼1 b¼1
Xm
þ vc lnðSi;c;t0 Þ þ ni;t
c¼1
where A is a vector of k variables related to the
dairy company, including firm ownership and
assistance programs; B is a vector of l variables
controlling for differences on the level of the
farm household and farm operator; S is a vec-
tor of m variables that measure differences in
initial farm size, and finally ni;t is the errorterm.
Estimating Eqn. (3) using ordinary least
squares (OLS) regression techniques will only
take into account the growth of farms that still
exist at the end of the period that we consider.
Such analysis based on a sample of surviving
farms only, may be biased due to sample
attrition. Instead, we use a two-step Heckman
model of firm survival and growth (Weiss,
1999). First, a selection equation is estimated
by maximum likelihood as an independent
probit model to determine which farms have
survived using information from the whole
sample of suppliers. A vector of inverse Mills
ratios (estimated expected error) can be gener-
ated from the parameter estimates. Supplier
growth is observed only when the selection
equation equals 1. Then, for the surviving
suppliers, growth is regressed on the explana-
tory variables and the vector of inverse Mills
ratios from the selection equation by least
squares. Therefore, the second stage reruns the
regression with the estimated expected error
included as an extra explanatory variable,
removing the part of the error term correlated
with the explanatory variables and avoiding the
bias. 11
The dependent variables in the two estima-
tions are SURVIVAL and GROWTH, respec-
tively. SURVIVAL is a dichotomous variable
that takes the value of one if the household is
still delivering milk to a dairy in 2000 and it
takes the value of zero if the household stopped
delivering milk. GROWTH is defined as the
difference between the natural logarithm of the
number of cows owned in 2000 and the natural
logarithm of the number of cows owned in
1995, divided by the number of years between
ð2Þ the two points of observation, i.e., five.
The first set of explanatory variables mea-
sures the impact of dairy company character-
istics, including its ownership structure.
FDI is a dichotomous variable, which takes
the value of one if the household delivered its
milk to a foreign-owned dairy company in
bb Bi;b;t
The expected impact of FDI on survival and
ð3Þ growth of local suppliers is uncertain ex ante.
Most studies in the literature suggest that the
impact is negative, especially for small suppli-
ers. But our evidence, as discussed in the pre-
vious section, indicates that FDI has
contributed to access to finance, inputs, and
technology of small suppliers and as such may
have contributed to the survival and growth of
the farms.
FDISIZE is an interaction term of FDI and
SIZE95, an indicator of initial herd size. This
interaction term is included to test the
hypothesis of foreign investors’ bias toward
large suppliers. If foreign investors prefer to
deal with large suppliers, this may have a neg-
ative survival, and possibly also growth, effect
for small producers that initially supplied to the
foreign-owned dairy company. On the other
hand, this bias can lead to an additional posi-
tive growth effect for large suppliers that deliver
to foreign-owned companies. We therefore
expect FDISIZE to have a positive effect on
survival and growth.
PROGRAM measures the number of dairy
assistance programs that are offered by the
dairy company. Assistance programs include
extension, veterinary assistance, credit pro-
grams either for buying cows or for making
other on-farm investments, input supply pro-
gram, bank loan guarantees, and on-farm milk
collection. PROGRAM takes a value between
zero and seven with seven indicating that the
dairy offers all seven possible assistance pro-
grams.
From the information gathered in our inter-
views it was clear that foreign-owned dairy
companies had initiated and were the first to
implement such assistance programs. Hence,
the variables FDI and PROGRAM may not be
independent. But the correlation between both
variables in the sample is very small: the cor-
relation coefficient is 0.08. Still, to test for the
impact of possible multicollinearity problems in
the estimation, we also ran restricted models
(Models B and C) where FDI and PROGRAM
are excluded, respectively.
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS
A final variable related to the dairy company
to which the household delivers is COOP.
COOP is a dichotomous variable, which takes
the value of one if the household head is
delivering to a cooperative and takes the value
zero otherwise. We expect COOP to have a
positive effect on survival as it is more compli-
cated for a cooperative dairy company to
exclude suppliers, which are typically members
of the cooperative, than for a noncooperative
private company.
The next group of explanatory variables in
the analysis includes characteristics of the
household head. AGE measures the age of the
household head. AGE may capture two differ-
ent effects. As a proxy for experience one
should expect a positive effect on growth.
Younger people are however typically more
dynamic and entrepreneurial and age may
therefore have a negative impact, especially
since previous management experience may be
less relevant in a transition environment.
EDU is the education level of the household
head and takes a value between one and seven
where one stands for the lowest level of edu-
cation and seven the highest. Better educated
farmers are also more likely to survive and
grow, hence, EDU is expected to have a positive
coefficient.
PENSION is defined as a dichotomous var-
iable that takes the value of one if the house-
hold receives income from pensions and zero if
not. PENSION is assumed to have a negative
impact on survival and growth because it
reflects lack of incentives to invest.
COMMUNITY is an indicator of the activity
in the community of the household head, and
thus his social capital, and is measured as the
number of times per year that the household
head takes part in meetings concerning the
village. Social capital is expected to have a
positive impact on the likelihood of farm sur-
vival and growth.
WAGE is defined as a dichotomous variable
that takes the value of one if the household
receives income from off-farm employment and
zero in the other case. WAGE may capture two
opposing effects. Employment in other sectors
can offer extra sources of finance for the
household to invest in the farm operation. With
important rural credit constraints, the effect
would be positive on farm survival and growth.
On the other hand, with rural labor market
constraints, access to off-farm employment can
act as a stepping-stone to leave the agricultural
sector.
1537
HHMEMBER is the number of household
members and is expected to have a positive
effect on survival.
Finally, farm size is proxied by the natural
logarithm of the number of cows owned by the
household in 1995 (SIZE95). To capture pos-
sible nonlinear effects of this relationship we
include a variable SIZE95SQ, the squared
value of SIZE95 and SIZE95CU, the cubic
form of SIZE95.
(b) Regression results
Table 8 shows the estimation results for 283
households that delivered milk to a dairy in
1995 and for which the relevant data were
available. 12
The estimated coefficients for FDI and FDI-
SIZE are insignificant for both the survival and
the growth models. These results imply, first,
that suppliers delivering to foreign companies
are equally likely to survive than those deliv-
ering to domestic companies––and hence that
we find no evidence that foreign companies are
more likely to cut down their supplier base.
Moreover, our results indicate also that foreign
companies are not more likely to cut off small
suppliers from their supplier base than domes-
tic companies. This result is contrary to most
arguments in the literature that suggest that
foreign investment causes companies to cut
small suppliers from their supply base to min-
imize transaction costs and because small
farmers cannot make the necessary investments
to reach enhanced standards.
One criticism on this interpretation is that the
variable FDI could be endogenous because
foreign firms may decide their investment
location based on the structure of the local
supplier base. In other words, foreign firms
may prefer to invest in dairy companies with
larger suppliers. In our sample however, this is
not the case. The initial size distributions of
farms delivering to domestic dairy companies
and farms delivering to foreign companies are
not significantly different (see Table 9; Pear-
son’s v2 ¼ 4:72).
Assistance programs that dairy companies
provide for their supplying farms have a highly
significant and positive impact both on the
likelihood of farm survival and on growth.
Dairies that provide more programs to their
suppliers have fewer suppliers dropping out,
and their suppliers grow more. We know from
our interviews that foreign investors started
first with these programs, beginning in 1993–95.
 1538
Table 8. Regression results
Model A Model B Model C
Survival Growth Survival Growth Survival Growth
Coefficient jz-valuej Coefficient jz-valuej Coefficient jz-valuej Coefficient jz-valuej Coefficient jz-valuej Coefficient jz-valuej
Dairy company
FDI )0.497 (0.48) 0.047 (0.99) 0.271 (0.35) 0.042 (0.87) – – – –
FDISIZE 0.456 (0.80) )0.017 (0.79) )0.098 (0.23) )0.015 (0.68) – – – –
PROGRAM 0.541 (6.03) 0.006 (1.83) – – – – 0.520 (6.06) 0.006 (1.78)
COOP 0.933 (2.39) )0.001 (0.05) 0.666 (2.34) )0.011 (0.86) 0.912 (2.34) )0.001 (0.10)

WORLD DEVELOPMENT
Farm characteristics
SIZE95 0.788 (2.75) )0.336 (3.61) 1.131 (4.79) )0.406 (4.92) 0.943 (3.85) )0.351 (3.79)
SIZE95SQ – – 0.134 (2.95) – – 0.149 (3.53) – – 0.136 (2.99)
SIZE95CU – – )0.017 (2.32) – – )0.018 (2.54) – – )0.017 (2.36)

Household characteristics
AGE )0.020 (1.60) )0.001 (2.33) )0.007 (0.71) )0.001 (1.97) )0.020 (1.55) )0.001 (2.26)
EDU )0.032 (0.27) 0.009 (2.11) 0.116 (1.21) 0.008 (1.63) )0.013 (0.11) 0.010 (2.19)
WAGE )0.929 (2.84) )0.038 (2.36) )0.689 (2.83) )0.023 (1.34) )0.860 (2.72) )0.039 (2.40)
PENSION 0.245 (0.76) )0.019 (1.48) 0.057 (0.24) )0.020 (1.50) 0.250 (0.79) )0.019 (1.51)
COMMUNITY 0.073 (0.87) 0.003 (1.65) 0.031 (0.53) 0.003 (1.59) 0.062 (0.77) 0.003 (1.66)
HHMEMBER )0.006 (0.07) 0.005 (1.47) 0.101 (1.63) 0.004 (1.18) )0.010 (0.13) 0.005 (1.52)

Constant )0.695 (0.73) 0.262 (3.19) )1.577 (1.92) 0.390 (5.42) )0.894 (0.98) 0.287 (3.64)
Log likelihood 193.3 163.0 191.8
Rho 0.176 (0.595) )0.687 (4.10) 0.107 (0.35)
Observations 283 283 283 283 283
*
Significant at 10%.
**
Significant at 5%.
***
Significant at 1%.
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS 1539

Table 9. Initial size distribution of farms delivering to domestic and foreign-owned dairy companies (1995)
Number of cows Farms delivering to:
per farm
Foreign owned dairy companies Domestic dairy companies
# % # %
<2 1 0.8 2 1.3
2–3 6 4.8 11 7.0
4–5 27 21.4 25 15.9
6–7 18 14.3 26 16.6
8–9 19 15.1 22 14.0
10–11 24 19.1 27 17.2
12–13 15 11.9 19 12.1
14–15 6 4.8 15 9.6
>15 10 7.9 10 6.4
Total 126 100 157 100
2
Pearson’s v ¼ 4:72 ðp ¼ 0:787Þ

But over the time period covered by our survey least part of the exits from the sector were
(1995–2001) we found no statistical interaction involuntary.
between the programs and foreign ownership of While we found no size effect of suppliers
the companies. The correlation coefficient concerning foreign investment, the significant
between FDI and PROGRAM is 0.08, and coefficient of SIZE1995 indicates that in gen-
leaving out the respective variables, did not eral larger farms were more likely to survive
significantly affect the results. This suggests that 1995–2000 and the associated restructuring
domestic companies have rapidly replicated the process.
successful programs and have been equally The size-growth relationship is complex with
successful in assisting and upgrading their all three SIZE variables significant. A first
suppliers. implication is that initial farm size does have a
COOP has a significant impact, but only on significant effect on farm growth, which is
the survival of suppliers. This confirms our inconsistent with ‘‘Gibrat’s law,’’ which
hypothesis that suppliers delivering to cooper- states that firm growth is independent of the
atives are more likely to survive because it is initial size. Second, the significant coefficients
more difficult for cooperatives to refuse to take for SIZE95SQ and SIZE95CU indicate
milk from their suppliers, which are typically that the size-growth relationship has two turn-
cooperative members. This suggests that at ing points. On the basis of the estimated

150
Change in number of cows

-150
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Herd size

Figure 4. Change in number of cows by farm size, 1995–2000. Each data point gives the average change in number of
cows in three consecutive size classes of which the data point itself is the size class in the middle. For example, the data
point for herd size three signifies the average change in number of cows in all farms with herd sizes 2, 3 and 4.
1540 WORLD DEVELOPMENT
coefficients, one can calculate that the turning
points occur at a herd size of eight cows (min-
imum) and 24 cows (maximum). Figures 3 and
4 illustrate the developments behind these esti-
mation results. Eight cows was approximately
the median herd size in 1995 (see Figure 3) and
more than 97% of our observations had a herd
size smaller than 24 cows. The group of
smallest farms (1–4 cows) has grown and also
the group of larger farms (more than 16 cows).
The medium group (5–15 cows) has reduced in
size. Part of the farms in this group has reduced
their size and others have increased their size.
Several of these farms have increased their herd
size to more than 15 cows. The regression
results indicate that this is mostly the case for
farms already having more than eight cows in
1995. Some very small suppliers (1–4 cows)
have grown (since growth was relatively larger
in the 1–4 group than in the 5–8 group). The
group of very small suppliers has grown mostly
because of suppliers of the medium group
(especially those with 5–8 cows) have cut down
their herd. Figure 4 illustrates how this trans-
lates itself in the herd distribution. While in
numbers of farms the smallest farms have sig-
nificant growth rates, in terms of livestock the
important growth occurred in the group of
farms with more than 13 cows.
It is unclear whether these developments
over 1995–2000 reflect a ‘‘normalization’’ of a
farm distribution which was distorted by the
Communist legacy––the 2000 farm distribu-
tion appears less distorted than the 1995 dis-
tribution, as illustrated in Figure 3––or
whether it is only one stage in a further
development toward a bimodal farm size dis-
tribution in the dairy sector. There is mixed
evidence whether such bimodal distribution,
with both small part-time farmers on the one
hand, and large full-time commercial farmers
on the other hand, has emerged in Western
market economies. Some studies confirm such
development of a ‘‘disappearing middle’’
(Edwards, Smith, & Peterson, 1985; Garcia,
Offutt, & Sonka, 1987; Weiss, 1999). A recent
study by Wolf and Sumner (2001) finds,
however, that farm size distributions for US
dairy farms are not bimodal.
Some other characteristics of the farms also
affect their growth and/or survival. The indi-
cators of human and social capital character-
istics do not have an impact on the survival of
farms, but only on the growth of the farms. Age
has a significantly negative effect on farm
growth. Hence, while age has not significantly
affected the likelihood of the farms continuing
to deliver milk to companies despite increased
demands, farms run by younger households are
more likely to increase their herd size and to
grow. Similarly, while education does not affect
the likelihood of survival, better educated farm
managers are associated with more growth. The
result is the same for the indicator of social
capital. Playing an active role in the community
does not seem to have a relationship with
farmers’ survival over the past five years, but it
is positively correlated with farm growth. The
results are robust for different model specifica-
tions.
The WAGE variable has a significantly neg-
ative impact both on farm survival and growth.
Households with household members that earn
income from wages in other sectors are more
likely to quit their dairy operation. Households
with off-farm employment also have slower
growing farms. These results suggest that
households with off-farm income sources are
more likely to leave the sector, while house-
holds that depend solely on the agricultural
sector as a source of income tend to stay in the
sector. Weiss (1999) also finds a significantly
negative relationship between off-farm
employment of Austrian farming households
and farm survival. These findings confirm the
view that part-time farming is a stepping-stone
out of the agricultural sector and that part-time
farming promotes the restructuring of the farm
sector (Pfeffer, 1989; Roe, 1995).
This estimation result also suggests that
access to credit for the farms in the sample may
be less constraining than is often argued (e.g.,
European Commission, 1998; World Bank,
2001). Other studies find that in regions where
rural credit market imperfections are a very
important constraint for farms, households’
access to off-farm income sources has an
important positive impact on farm develop-
ment (e.g. Rizov, Gavrilescu, Gow, Mathijs, &
Swinnen, 2001). Hence, our result confirms that
the vertical integration process and the farm
assistance programs by the dairy farms, in
combination with the general progress in the
Polish rural finance system, have a very signif-
icant positive impact on access to credit for the
farms, and on mitigating their credit con-
straints.
This conclusion is consistent with the insig-
nificant result of PENSION. The negative
coefficient of PENSION in the growth equa-
tions, suggests that, if anything, access to pen-
sions reduces farm growth. Finally, the size of
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS
the household does not affect the survival or
growth of the farms.
5. CONCLUSIONS
Previous studies argue that foreign invest-
ment leads to a rapid consolidation of the local
supplier base with negative implications for
those suppliers who cannot comply with higher
standards and grading requirements, or who
are cut out by the company in order to reduce
transaction costs. Studies argue that this effect
can be especially dramatic for small suppliers in
developing countries.
The conclusions of our analysis are different.
We do not find that foreign investment leads to
either a rapid consolidation of the local sup-
plier base or to small suppliers being cut out. In
other words, we do not find the negative direct
effects for local suppliers that are often attrib-
uted to foreign direct investments in the liter-
ature. On the contrary, our evidence shows that
the indirect effects of foreign investment––or its
spillover effects––lead to improved access to
finance, increased investments, and (dramatic)
quality improvements by small local suppliers.
The mechanism through which this happens
consists of two steps. First, after foreign
investment, processing companies start a pro-
cess of vertical coordination through contract-
ing with local suppliers in which input and
output markets are interlinked. The contracting
is associated with enhanced standards require-
ments of supplies while at the same time the
companies provide assistance programs to
improve supplier management, and to enhance
access to technology, credit and other inputs. In
combination, the contracts and assistance pro-
grams are designed to overcome market
imperfections. The contracts are enforced by
interlinking the various markets. This process
leads to important positive vertical spillovers
NOTES
1. Important causes of the output decline were declin-
ing relative prices and subsidy cuts following liberaliza-
tion (terms of trade for milk producers declined by
almost 90% during 1989–92). Both output and produc-
tivity were negatively affected by disruptions from
privatization and restructuring, as well as contracting
problems in the absence of enforcement mechanisms and
1541
for the suppliers. Our estimations show that the
assistance programs had a significant effect
both on the survival and the growth of suppli-
ers.
The second step is that of horizontal spill-
overs. When domestic companies observe these
successful vertical integration strategies, they
start copying the strategies. Our analysis shows
that these horizontal spillover effects are strong
and rapid. For several of the effects, after five
years there is no longer a significant difference
between foreign owned and domestic compa-
nies, and their suppliers. Only in some aspects,
such as medium term investments, convergence
had not yet occurred.
In combination, these effects have caused
significant improvements in small suppliers’
investments, productivity and product quality
over the five-year period studied here. More
than 85% of all suppliers continued supplying
despite restructuring of the dairy companies
and strong tightening of quality requirements.
The reduction in suppliers over the period is
less than the average reduction in agricultural
employment in Poland, and most of those who
stopped supplying were older farmers who
might have stopped delivering milk anyhow.
Younger and better educated managers were
associated with stronger supplier growth. The
average size of the farms increased, but rela-
tively little, and the vast majority of suppliers
remain small.
In summary, our study finds that foreign
investment through this process of vertical
integration did not result in the negative direct
effects for local suppliers that are often attrib-
uted to foreign direct investments in the liter-
ature. On the contrary, our evidence shows that
there are significant positive effects on small
suppliers, but that the most important effect
was, over the period analyzed here, indi-
rect through vertical and horizontal spillover
effects.
the disruption of traditional exchange relationships
between farms and up- and downstream companies
(Dries & Swinnen, 2002; Macours & Swinnen, 2000).
2. In the European former Communist countries, only
in former Yugoslavia private farming also survived to a
large extent.
1542 WORLD DEVELOPMENT
3. Poland had very few collective farms.
4. FDI has resulted from several company strategies:
to serve the local market when trade constraints limit
imports, to use the Polish economy advantages for
exporting to the home market or to third markets, etc.
Poland performed well in FDI inflows because of its
stable political and institutional system, advanced
reform strategy, and cheap but relatively well educated
labor force. The EU accession process has further
stimulated FDI because it reinforced the institutional
and economic stability, the prospect of a large single
market, growth in Polish incomes and food demand,
and––in some cases––expectations of EU subsidies
(Walkenhorst, 2001).
5. Poland is a net exporter of dairy products. The main
buyers of Polish dairy products in 1998 were the former
Soviet Union countries (FSU). Imports of dairy products
in Poland mainly originate from the EU. Although the
total value of imports was halved during 1992–98, the
share of imports from the EU has increased. Poland
exports mostly lower value products to the EU, and
imports mostly higher value cheese and other products. In
1999, 59% of all imports were yogurts, almost all
imported from Germany. Only licensed dairies are
allowed to export their products to the EU. In 2000, only
19 out of the approximately 400 dairy processing compa-
nies have obtained such an export license. Although in
numbers these firms make up only 5% of the sector, in
terms of total milk procurement they account for 25%
(Kaspersson, Rabinowicz, & Schwaag Serger, 2002).
6. Estimate on the basis of data on old voivodship
classifications (Wies I Rolnictwo, 1999).
7. Using a list of supplying farmers from the foreign-
owned dairy companies would create a selection bias
since a list of current suppliers will exclude any farmers
that have stopped supplying over the past years.
REFERENCES
Aitken, B. J., & Harrison, A. E. (1999). Do domestic
firms benefit from foreign direct investment? Evi-
dence from Venezuela. American Economic Review,
89(3), 605–618.
Bardhan, P., & Udry, C. (1999). Development microeco-
nomics. Oxford: Oxford University Press.
Blanchard, O. (1999). The economics of post-communist
transition. Oxford: Clarendon.
Dolan, C., & Humphrey, J. (2000). Governance and
trade in fresh vegetables: the impact of UK super-
markets on the African horticulture industry. Journal
of Development Studies, 37(2), 147–176.
8. Formally, the foreign investor was in a joint venture
with a local cooperative and leased the collection
stations from the cooperative. It was the cooperative
which, under strong pressure from the foreign investor,
made the investments in the cooling equipment at its
collection stations.
9. The process of interlinking markets is traditionally
studied extensively in the development economics liter-
ature, typically focusing on landlord-tenant or trader-
farmer relationships in developing countries (see e.g.,
Bardhan & Udry, 1999).
10. Success of these programs and contract enforce-
ment is not obvious. Gow and Swinnen (2001) discuss
several cases of failure.
11. As the regression results in the next section will
show, Rho (the estimated correlation coefficient
between the residuals in the survival and growth
equations) is insignificant in our model. This indicates
that there are actually no unobserved characteristics
that make surviving farms grow faster than farms that
have quit. This means that the Heckman model that we
have estimated would give the same results as estimat-
ing a probit model of farm survival and a least-squares
model for growth separately. Since both estimation
methods will give the same results, we have opted to
keep the Heckman procedure as it is the standard
methodology used in the literature on firm survival and
growth.
12. On top of models A, B and C in Table 8, we have
also run restricted models to test the potential impact
of correlation between AGE and PENSION, because
older household heads are more likely to receive
pensions. We find that the results are robust across
these different model specifications and for simplicity
we have not included the results of these models in
Table 8.
Dries, L., & Swinnen, J. F. M. (2002). Institutional
reform and labor reallocation during transition:
evidence from Polish agriculture. World Develop-
ment, 30(3), 457–474.
Edwards, C., Smith, M. G., & Peterson, R. N. (1985).
The changing distribution of farms by size: a Markov
analysis. Agricultural Economics Research, 37(4), 1–
16.
European Commission (1998). Agricultural situation
and prospects in the Central and Eastern European
countries. Poland, Brussels: European Commission,
Directorate General Agriculture.
 FDI VERTICAL INTEGRATION AND LOCAL SUPPLIERS
Evans, D. (1987). Tests of alternative theories of firm
growth. Journal of Political Economy, 95, 657–674.
FAO (2003). FAOSTAT Agriculture Data [online].
FAO. Available at http://www.FAO.org.
Farina, E. M. M. Q., & Reardon, T. (2000). Agrifood
grades and standards in the extended Mercosur: their
role in the changing agrifood system. American
Journal of Agricultural Economics, 82(5), 1170–1176.
Foster, C. (1999). The impact of FDI in the upstream
and downstream sectors on investment in agriculture
in the NIS. In Agricultural Finance and Credit
Infrastructure in Transition Economies. Paris: OECD.
Garcia, P., Offutt, S. E., & Sonka, S. T. (1987). Size
distribution and growth in a sample of Illinois cash
grain farms. American Journal of Agricultural Eco-
nomics, 69(2), 471–476.
Gow, H., & Swinnen, J. (1998). Agribusiness restruc-
turing, foreign direct investment, and hold-up prob-
lems in agricultural transition. European Review of
Agricultural Economics, 25(4).
Gow, H., & Swinnen, J. (2001). Private enforcement
capital and contract enforcement in transition coun-
tries. American Journal of Agricultural Economics,
83(3), 686–690.
Gow, H., Streeter, D., & Swinnen, J. (2000). How
private contract enforcement mechanisms can suc-
ceed where public institutions fail: the case of
Juhosucor A. S. Agricultural Economics, 23(3), 253–
265.
GUS (2001). Statistical Yearbook. Warsaw: Central
Statistical Office.
Hall, B. H. (1997). The relationship between firm size
and firm growth in the US manufacturing sector.
Journal of Industrial Economics, 35(4), 583–606.
Hart, P. E., & Oulton, N. (1996). Growth and size of
firms. Economic Journal, 106(438), 1242–1252.
Henson, S. J., Loader, R. J., & Brouder, A. (2000). Food
safety standards and exports of perishable products
from developing countries: fish exports from East
Africa to the European Union. American Journal of
Agricultural Economics, 82(5), 1159–1169.
Holloway, G., Nicholson, C., Delgado, C., Staal, S., &
Ehui, S. (2000). Agroindustrialisation through insti-
tutional innovation. Transaction costs, cooperatives
and milk-market development in the East-African
Highlands. Agricultural Economics, 23, 279–288.
Hu, A. G. Z., & Jefferson, G. H. (2002). FDI impact and
spillover: evidence from China’s electronic and
textile industries. The World Economy, 25(8), 1063–
1076.
Johnson, S., McMillan, J. & Woodruff, C. (1999).
Contract Enforcement in Transition. EBRD Work-
ing Paper No. 45. London: European Bank for
Reconstruction and Development.
Key, N., & Runsten, D. (1999). Contract farming,
smallholders, and rural development in Latin Amer-
ica: the organization of agroprocessing firms and the
scale of outgrower production. World Development,
27(2), 381–401.
Kaspersson, E., Rabinowicz, E., & Schwaag Serger, S.
(2002). EU milk policy after enlargement––competi-
tiveness and politics in four candidate countries. Lund:
1543
Swedish Institute for Food and Agricultural Eco-
nomics.
Kokko, A. (1994). Technology, market characteristics
and spillovers. Journal of Development Economics,
43(2), 279–293.
Kokko, A., Tansini, R., & Zejan, M. C. (1996). Local
technological capability and productivity spillovers
from FDI in the Uruguayan manufacturing sector.
Journal of Development Studies, 32(4), 602–611.
Konings, J. (2001). The effects of foreign direct invest-
ment on domestic firms. Evidence from firm-level
panel data in emerging economies. Economics of
Transition, 9(3), 619–633.
Konings, J., & Walsh, P. P. (1999). Disorganization in
the process of transition. Firm-level evidence from
Ukraine. Economics of Transition, 7(1), 29–46.
Liu, Z. (2002). Foreign direct investment and technology
spillover: evidence from China. Journal of Compar-
ative Economics, 30(3), 579–602.
Macours, K., & Swinnen, J. (2000). Causes of output
decline during transition: the case of Central and
Eastern European agriculture. Journal of Compara-
tive Economics, 28(1), 172–206.
Majewski, E., & Dalton, G. (2000). The strategic options
for the Polish agro-food sector in the light of economic
analyses. Warsaw: FAPA.
Mcmillan, J., & Woodruff, C. (1999). Interfirm relation-
ships and informal credit in Vietnam. Quarterly
Journal of Economics (November), 1285–1320.
Pfeffer, M. J. (1989). Part-time farming and the stability
of family farms in the Federal Republic of Germany.
European Review of Agricultural Economics, 16(4),
425–444.
Reardon, T., & Berdegue, J. A. (2002). The rapid rise of
supermarkets in Latin America: challenges and
opportunities for development. Development Policy
Review, 20(4), 317–334.
Reardon, T., Codron, J.-M., Busch, L., Bingen, J., &
Harris, C. (1999). Global change in agrifood grades
and standards: agribusiness strategic responses in
developing countries. International Food and Agri-
business Management Review, 2(3), 421–435.
Rizov, M., Gavrilescu, D., Gow, H., Mathijs, E., &
Swinnen, J. (2001). Transition and enterprise restruc-
turing: the development of individual farming in
Romania. World Development, 29(7), 1257–1274.
Roe, B. (1995). A study of US farms exits with evidence
from the panel study of income dynamics: 1968–
1989. Paper presented at the AAEA annual meeting,
Indianapolis, August 6–9, 1995.
Runsten, D. & Key, N. (1996). Contract farming in
developing countries: theoretical issues and analysis
of some Mexican Cases. Report LC/L.989. Santiago,
Chile: UN-ECLAC.
Swedish Board of Agriculture (2001). A study of the
milk sector in Poland, Hungary, the Czech Republic
and Estonia. Market Regulations Division.
Walkenhorst, P. (2001). The geography of foreign direct
investment in Poland’s food industry. Journal of
Agricultural Economics, 52(3), 71–86.
Weatherspoon, D. D., & Reardon, T. (2003). The rise of
supermarkets in Africa: implications for agrifood
1544 WORLD DEVELOPMENT
systems and the rural poor. Development Policy
Review, 21(3), 333–356.
Weiss, C. (1999). Farm growth and survival: econometric
evidence for individual farms in Upper Austria. Am-
erican Journal of Agricultural Economics, 81, 103–116.
Wies I Rolnictwo (1999). Warsaw.
Winters, A. L. (2000). Trade liberalisation and poverty.
Available: http://www.worldbank.org/poverty/wdr-
poverty/winters.htm.
Wolf, C. A., & Sumner, D. A. (2001). Are farm size
distributions bimodal? Evidence from kernel density
estimates of dairy farm size distributions. American
Journal of Agricultural Economics, 83(1), 77–
88.
World Bank (2001). Poland: the functioning of the
labor, land and financial markets: opportunities and
constraints for farming sector restructuring. Wash-
ington, DC: The World Bank.