Economics Letters 67(2000)179–186
/locate/econba
International relocation:firm and industry determinants
a,b ,a,b
*Enrico Pennings ,Leo Sleuwaegen a
Catholic University of Leuven ,Naamstraat 69,3000Leuven ,Belgium
b
Erasmus University Rotterdam ,Rotterdam ,The Netherlands
Received 16November 1998;received in revid form 12October 1999;accepted 21October 1999
Abstract
This article is the first to explore the determinants of international relocation of a firm.It is found that labour intensive firms in a highly industrialized and open economy such as Belgium tend to relocate more to other countries than their highly productive capital intensive counterparts.Access to a global network,firm size,and the rate of innovation have a positive effect on the probability of relocation.Uncertainty has a negative impact on the probability of relocation.The positive effect of firm size and profitability on the relocation decision is clearly distinct from its effect on the exit decision of a firm.©2000Elvier Science S.A.All rights rerved.
Keywords :Relocation;Multinationals;Investment;Sunk costs JEL classification :F21;F23
1.Introduction
Relocation of activities from one country to another is of immediate policy interest as business relocations are claimed to be job exporting with firms moving to low-cost labor-abundant locations (A
rthuis,1993;OECD,1995;Brainard and Riker,1997;European Parliament,1998).Especially small countries with an open economy are vulnerable to relocation.In Belgium,for example,the number of people employed in the diamond industry sharply decread from 19,010in 1961to 3831in 1992after massive relocation to —mainly —Thailand.
Thus far,most studies have looked at the determinants of investment abroad without making a distinction between expansion investments and relocation decisions that involve ‘the move of a manufacturing process from one place to another’(Mucchielli and Saucier,1997).Few studies have looked at the motives for relocation in such a strict n.From a sample of business relocation
*Corresponding author.Tel.:132-16-326-774;fax:132-16-326-732.E -mail address :enrico.pennings@econ.kuleuven.ac.be (E.Pennings)0165-1765/00/$–e front matter ©2000Elvier Science S.A.All rights rerved.
PII:S0165-1765(99)00269-4
180E.Pennings,L.Sleuwaegen/Economics Letters67(2000)179–186
announcements in the U.S.,Chan et al.(1995)find that main reasons for the business relocation are c
ost savings and business expansions,regardless of whether thefirm is a plant or a headquarter. Within a European context,Mucchielli and Saucier(1997)speculate that restructuring,andflexible respons to new market conditions for innovative products are equally,if not more,important motives.Belderbos and Sleuwaegen(1996)and Belderbos(1997)report an important number of Japane investments in Europe and the U.S.over the period1986–1989replacing Japane production as a respon to trade measures targeting electronics exports from Japan.
Using a novel data t,this paper ts out to systematically investigate relocation decisions offirms located in Belgium.The paper is structured as follows.In Section2we explain the data. Subquently,we build a logit model that explains relocation.Section4discuss the results and Section5summarizes thefindings.
2.Data
The sample offirms to which a questionnaire was nt out was obtained by merging two data ts. First,a sample of allfirms that reported a collective layoff in the period from1990to1996.Eachfirm in Belgium that:(i)has at least20employees;and(ii)lays off more than10%of its workforce needs to report such a dismissal at the government.This data t,maintained by the Federal Planning Bureau (FPB)i
n Belgium,consists of827firms.The cond sample was obtained by a draw fromfirms in a data t maintained by the National Institute of Statistics(NIS)in Belgium,which contains allfirms with a V AT-number.From this large sample2172firms with more than20employees were randomly lected.So the total number of questionnaires amounts to2999.
The respon rate to the questionnaire was15.5%resulting in a sample of466obrvations.Tables 1and2respectively show the ctors and thefirm size of the initial sample and the sample of respondingfirms.The tables illustrate that the distribution of the respondents well corresponds with the distribution offirms in the initial sample.Subquently,the data from the respondingfirms were linked to data from the balance sheet of the companies.This link resulted in afinal sample of372firms.
Within thefinal sample14.0%of thefirms responded that they relocated activities between1990 Table1
Distribution of ctors among initial sample offirms and responding sample
No.offirms(%)in No.offirms(%)
initial sample responding
1.Primary38(1.3)6(1.3)
2.Basic industry1487(49.6)247(5
3.0)
3.Construction177(5.9)28(6.0)
4.Trade681(22.7)112(24.0)
5.Services294(9.8)35(7.5)
6.Transport and communication204(6.8)23(4.9)
7.Financial corporations64(2.1)10(2.1)
8.Other rvices54(1.8)5(1.1)
Total2999(100)466
E.Pennings,L.Sleuwaegen/Economics Letters67(2000)179–186181 Table2
Distribution offirm size among initial sample offirms and responding sample
Firm size No.offirms No.offirms
(No.of employees)(%)in initial sample(%)responding
20–49897(29.9)112(24.0)
50–99745(24.8)111(23.8)
100–199628(20.9)103(22.1)
200–499463(15.4)85(18.2)
500–999149(5.0)23(4.9)
.1000117(3.9)32(6.9)
Total2999(100)466(100)
and1996.Table3shows the countries to which activities were relocated.The table illustrates that neighbouring countries are the most preferred destination regions for relocation.
3.The empirical model
The decision to relocate activities is modeled within a logit model relating the probability to
9 relocate to a t of explanatory variables x.In the logit model the probability of relocation is F x b
s d
i i where F.5exp./11exp.,and b is the vector of coefficients.
s d s d s d
f g
International relocation is defined as either the decision to move part of the production to another country,or to replace part of the production by a combination of an investment abroad and subcontracting during the period1990–1996.It thus excludes the cas of complete disinvestment with a relocation of all the activities to another country.Descriptive statistics and preci definitions of the explanatory variables are provided in Appendix A.
Among the explanatory variables,capital investment and the degree of sunkness of investment are proxied by twofirm-specific and two industry wide variables.Thefirm-specific variables are(i)C/L:
Table3
Host countries of relocation
Host country No.of Host country No.of
relocations relocations
France12Russia2
Netherlands7Austria1
UK5Canada1
Germany4China1
Tunisia3Czech Republic1
USA3Hungary1
Italy2Malaysia1
Luxembourg2South Africa1
Poland2Spain1
Portugal2Total52
182E.Pennings,L.Sleuwaegen/Economics Letters67(2000)179–186
1
ratio offixed capital to the number of employees,and(ii)S/T:the ratio of the sunk tangible asts
2
(plant,machinery and equipment)to the total tangible asts.Other sunk costs are given by industry wide variable IT1,an industry dummy variable that takes the value1if the company belongs to an R&D or advertising intensive industry.Industries displaying high R&D and/or advertising intensities are typically endogenous sunk cost industries(cf.Sutton,1991;Davies and Lyons,1996).Remaining m
anufacturing industries take a value equal to one for the dummy variable IT2.The rvices industries are taken as the reference ctor in the estimating model.Services ctors are primarily market oriented and need a clo connection with customers,and hence,will show a smaller probability to relocate.
Uncertainty is measured by the interaction term UNC,denoting the product of S/T and the variation coefficient of the de-trended sales.Profitability is measured by P/S,the profit to sales ratio, whilefirm size is measured by FS,the logarithm of the average turnover.The complexity of the operations and the degree of vertical integration of afirm is proxied by V/he value added to sales ratio.Innovation is measured by INN,a dummy variable that equals one if thefirm has successfully accomplished a combined product and process innovation.MUL is a dummy variable that equals1if the company belongs to a multinational group with more than one foreign subsidiary. The variable,CP,reprents a dummy variable that is1if the company faces incread competitive
3
pressure and0otherwi.
4.Results
Table4shows the results of the logit regressions.
The results can be explained by a blend of two streams within literature,(i)a cost minimizing literature which basically states thatfirms produce at locations where production is least costly,and (ii)a multinational investment literature claiming the importance of transferable technological advantages and the impact of operationalflexibility within a multinational framework especially in the prence of high uncertainty.Current theoretical rearch on relocation has mainly focud on thefirst stream of literature.Cordella and Grilo(1998)and Collie and Vandenbussche(1999)emphasize the importance of labour cost.Considering the cond stream,Motta and This(1994)show that sunk cost is an important impediment to investments in relocation.
Consistent with the cost-minimizing literature,the results of the logit model indicate that labour intensivefirms,displaying a low capital to labour ratio,are more likely to relocate activities from Belgium.From a location point of view,Belgian comparative advantage is strongly bad on the u of capital-intensive production technologies.Firms and industries using labour intensive technologies are at a relative disadvantage,unable to follow the high labour productivity gains made in the other ctors of the economy.High unit labour costs force them to either clo down or relocate activities to
1Fixed capital consists of formation expens,intangible asts,tangible asts,andfinancial asts.
2Apart from plant,machinery and equipment,the tangible asts include land and buildings,furniture and vehicles,leasing and other similar rights,asts under construction,and other tangible asts.
3INN,MUL and CP are directly obtained from the questionnaire.For CPfirms were asked if competitive pressure was declining,constant or increasing between1990and1996.
E.Pennings,L.Sleuwaegen/Economics Letters67(2000)179–186183 Table4
Results of the logistic regression:maximum likelihood estimation of international relocation
Variable Coefficient Coefficient Coefficient
(st.dev.)(st.dev.)(st.dev.)
Intercept25.452***25.471***25.290***
(1.947)(1.951)(1.963)
S/T0.6140.63820.670
(1.270)(1.317)(1.316)
C/L20.295**20.321**20.307**
(0.153)(0.155)(0.154)
CP20.05120.08120.083
(0.504)(0.511)(0.511)
MUL0.826***0.618*0.323
(0.351)(0.451)(0.585)
FS0.318***0.332***0.322***
(0.131)(0.133)(0.134)
INN0.908***0.864***0.912***
(0.355)(0.359)(0.364)
UNC214.609**214.408**218.874**
(7.372)(7.324)(9.642)
V/S21.15621.999**22.021**
(1.115)(1.195)(1.200)
P/S 5.188 5.398
(4.294)(4.355)
IT10.9130.9490.960
(0.921)(0.980)(0.973)
IT2 1.347* 1.458* 1.443*
(0.911)(0.976)(0.970)
MUL*P/S 4.309 3.616
(6.011)(6.034)
MUL*UNC8.955
(11.283)
2
R0.1540.1690.170
Loglikelihood2123.612120.592120.27
No.of obs.371371371
*p,0.1;**p,0.05;***p,0.01(one-tailed test).
low labour cost countries,especially when they are facing strong competitive pressure from imports. However,the competitive pressure variable is found to be not significantly different from zero. Largefirms not only have more plants that can be relocated,they can also profit more from relocating business.Especially when variable costs are low,largefirms will reap more benefits from relocation th
an smallfirms.The results indicate that it is indeed the largerfirm that relocates activities. Largefirms and more profitablefirms have also a better capacity tofinance and absorb the adjustment cost of the relocation investment(Caves,1996).
Considering the multinational investment literature,the dummy variable MUL suggests a positive impact on the relocation decision of a multinational network.When a company is part of a global network,production can easily be shifted within its network without incurring sunk cost when situation proves unfavourable in one of itsfirms.Kogut and Kulatilaka(1994)show that the value of
