Employment protection and FDI revisited: New evidence from micro data*

• Author: Jing‐Lin Duanmu,Per Skedinger,Pehr‐Johan Norbäck
• DOI: http://doi.org/10.1111/twec.13087
• Published date: 01 March 2021
• Date: 01 March 2021

World Econ. 2021;44:645–670. wileyonlinelibrary.com/journal/twec

|

645

© 2020 John Wiley & Sons Ltd

Received: 5 September 2017

|

Revised: 28 February 2019

|

Accepted: 6 March 2019

DOI: 10.1111/twec.13087

LETTER TO THE EDITOR

Employment protection and FDI revisited: New

evidence from micro data*

KEYWORDS

employment protection, FDI, labour market institutions, micro data

1

|

INTRODUCTION

Many countries attempt to attract foreign direct investments (FDI), since potential host countries rec-

ognise that FDI can improve the productive efficiency of the business sector. Not only do multina-

tional firms (MNEs) pay higher wages than indigenous firms, they are also more productive and do

more R&D. Increasing globalisation and volatility of employment may, however, also increase the

demand for job security. A conflict may then arise between the host country government's ambition to

protect employment and its ambition to attract FDI. In this paper, we re- examine how FDI is affected

by the stringency of employment protection legislation (EPL) in the host country using affiliate micro

data on outward investments by Swedish multinational firms.

Investigating the relationship between EPL and FDI is challenging for at least three reasons: First,

it seems intuitive that more stringent EPL increases labour costs, for example, in the form of severance

pay, since it becomes costlier to dismiss workers. This effect is reinforced if stronger EPL improves

the bargaining position of workers and hence increases their wage demands. However, there are also

mechanisms through which stronger EPL can reduce wages. For instance, Lazear (1990) has sug-

gested that more job security may contribute to lower wages, and possibly leave total labour costs

unchanged, if employers can shift increased dismissal costs onto workers. Thus, the relationship be-

tween more stringent EPL and labour costs— and hence also the relationship between EPL and FDI— is

ambiguous a priori.1 Second, the term FDI refers to multiple decisions made by a multinational firm.

These may involve investing into a new affiliate or deciding on employment and sales in an existing

affiliate, where some sales may be destined to markets outside the host country market as exports. The

impact of more stringent EPL on these decisions may not be uniform. Finally, FDI may trigger re-

forms of EPL in host countries and these policies are not decided upon in isolation of other labour

market policies, or economic policy in general.

To identify the effect of more stringent EPL on the activities of MNEs, we use a micro dataset on affil-

iates of Swedish multinational firms for the period 1965– 1998. We estimate regression equations for the

investment decision of establishing new affiliates (extensive margin), affiliate employment, local sales

and affiliate export sales (intensive margin). To examine the impact of the stringency of EPL in the host

1If employers incur firing costs in excess of benefits accruing to workers, in the form of red tape and legal costs, these

additional costs cannot be undone in wage bargaining and thus contribute to lowering employment (Burda,1992).

*We thank Shon Ferguson for valuable comments and Fredrik Andersson, Joakim Jansson, Louise Johannesson, Dina

Neiman, Hedda Nielsen, Charlotta Olofsson and Nina Öhrn for expert research assistance. Financial support from Jan

Wallanders och Tom Hedelius stiftelse and the Marianne and Marcus Wallenberg Foundation (first and second author) and

from the Johan and Jakob Söderberg Foundation (second author) is gratefully acknowledged.

646

|

LETTER TO THE EDITOR

country, we use an index developed by Allard (2005) which is available for the 20 OECD countries we

consider for our period of study, but also try an alternative measure (to be explained in more detail later).

An advantage with using micro data is that the impact of EPL can be explored on both margins of

FDI. Previous studies have either explored the impact of EPL on the intensive margin, using aggregate

data which precludes an analysis on the extensive margin, or they have focused on the extensive mar-

gin when using firm- level data. The main advantage with our micro data is that the analysis can also

be performed at a more detailed level.

On the intensive margin, we distinguish between the impact of more stringent EPL on those affili-

ates that export and those that only sell to the local market. Earlier research (e.g. Olney, 2013) finds an

asymmetry where MNEs’ exports decline in EPL while their local sales are unaffected. While we find

that exports and employment in exporting affiliates decline when EPL becomes stronger, it remarkably

turns out that sales and employment in non- exporting affiliates increase when EPL becomes stricter.

On the extensive margin, we compare new affiliates which only sell to the local market and those

that also export. Previous research (e.g. Javorcik & Spatareanu,2005) has found that more stringent

EPL deters MNEs from investing into new affiliates. We also find that more stringent EPL induces

Swedish MNEs to invest in fewer new affiliates, but a closer inspection reveals that the negative im-

pact stems from establishing fewer affiliates that export.

In the working paper version of this paper (Norbäck et al., 2012), we show how the above asymme-

try can be consistent with a heterogeneous Cournot model of FDI, where more stringent EPL increases

actual or expected wage costs. Briefly, if an affiliate is more productive than the average firm active

in the local market— a condition which is likely to be fulfilled, given the well- known stylised fact that

MNEs on average are more productive than indigenous firms— more stringent EPL, which increases

general wage costs, can improve the affiliate's relative competitiveness in the host country market.

This strategic advantage vis- à- vis indigenous firms can then explain why employment and local sales

can increase when the stringency of EPL increases. However, in its export markets the MNE faces

overseas rivals not affected by changes in EPL in the host country. Intuitively, if more stringent EPL

raises wage costs affiliate exports will then decline.

We are thus able to reproduce the findings in the previous literature, but our study also makes

significant contribution by shedding new light on the complex micro activities of MNEs, and con-

sequently, their differential response to the host country's EPL. All our regressions include country

fixed effects, which means that we identify the effect of more stringent EPL on our different affiliate-

level measures of FDI using the within- country variation in EPL. We control for other labour market

policies and in unreported robustness checks, we use numerous controls for globalisation, culture,

institutions, geography and taxes without changes in results. Moreover, endogeneity of EPL should

not be a problem in our study, since we focus on outward FDI from a small country.2

Still, one might suspect that a change in labour market policies (such as more or less stringent EPL)

will not take place in isolation of other economic policies. Sweden in the early 1990s is a good exam-

ple, where— as a response to a deep economic crisis— fundamental reforms were made in the tax

system and the regulation of FDI. Sweden also achieved central bank independence and a flexible

exchange rate system, as well as new budget rules for the government. This occurred at the same time

as the rules concerning temporary employment were made less stringent. The Swedish example sug-

gests that we should control for such omitted policy bias.3 We therefore add regressions with host

2Sweden accounted for just 2.8% of all outward FDI flows during the period 1970– 1998, according to UNCTAD. We also

perform robustness tests by excluding countries that may be more dependent on FDI from the sample.

3Note that in our data Sweden is the home country for the MNEs. To control for omitted home country- specific variables, we

use year- specific effects in all regressions.