1.olofsdotter ForeignDirectInvestment 1998
1.olofsdotter ForeignDirectInvestment 1998
1.olofsdotter ForeignDirectInvestment 1998
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Karin Olofsdotter
I. Introduction
Remark: I thank Magnus Blomström, Yves Bourdet, Hans Carlsson, Lena Dahl, Karo-
lina Ekholm, Fredrik Gallo, Fredrik Sjöholm, Peter Svensson, Johan Torstensson, Rasha
Torstensson, and an anonymous referee for valuable comments on this paper. Partici-
pants at the seminars at Lund University, University of Nottingham, and the workshop
on "Multinationals, Trade and Economic Geography" at IUI, 1996, have also contrib-
uted with helpful suggestions. Financial support from the Jacob Wallenberg Foundation
is gratefully acknowledged.
Most earlier works are micro-based studies that have concentrated on spillovers from
activities by MNEs, i.e. effects of FDI on host economies that are external to the MNEs.
See Kokko (1992) for an overview.
Yt=F(KnLt)An (1)
where Yt is total production,
labor, respectively, and A,
edge. Concentrating on At,
innovations are done abroa
is the available amount of
addition, we allow for a cat
logically backward country
country (see Edwards 199
5 The same argument is done by Edwards (1992). Furthermore, the role of populati
growth seems to be more important as a determinant of economic growth than
growth rate of the labor force (see, e.g., Levine and Renelt 1992).
III. Results
Base Estimations
these countries during the observed period, and the variable is expect-
ed to be negatively correlated with the growth rate.9 The results pre-
sented in Table 1, column (i), show that all these variables have the
expected signs and are statistically significant. In particular, an increase
in FDI seems to have a positive impact on the growth rate. Further, the
effects of these variables appear to be robust throughout all columns in
Table 1.
In the specifications (ii)-(v) we have added the joint effects of
FDI with OPEN, SCHOOL, PROP, and BUREFF, respectively. 10 Recall
that SCHOOL and BUREFF are expected to interact positively with
FDI, while the joint effects of FDI with OPEN and PROP are ex-
pected to be negative. In every specification, the separate effects of
the interactive terms are also included. As can be seen in Table 1, only
the interactive terms FDIxPROP and FDIxBUREFF turn out to be
significant at the 5 and 1 percent level, respectively. The joint effects
of the other capability factors OPEN and SCHOOL, however, have the
right signs but are insignificant. Finally, in column (vi), all the interac-
tive terms have been included in the regression. The interaction term
with BUREFF remains significant, while we do not find any statistical
significance of FDIxPROP in this specification. This could be
explained by a high correlation between the explanatory variables as
suggested by the Simple Correlation Matrix (Table A2 in the Appen-
dix).
It is interesting to notice that, in contrast to previous works, our
results reveal no positive growth effects from the stock of FDI in com-
bination with the degree of openness or the level of human capital,
which, at least, suggests that the results in earlier studies are sensitive
to the empirical method.11 Since these results could also be caused
by multicollinearity, various regressions with different combinations
of interactive terms and exclusions of one or more of the other explan-
atory variables have been tested. The results for FDIxOPEN and
FDIxSCHOOL, however, turned out to be very similar in every spec-
ification. The lack of significant effects for these variables is empiri-
cally supported by some earlier findings. Regarding the openness result,
9 This is also done by e.g. Levine and Renelt (1992) and Borensztein et al. (1995).
10 We have taken the natural logarithm of FDI since otherwise, there are very high sim-
ple correlations between the interactive terms.
1 1 The findings for OPEN and SCHOOL remain valid even after exclusion of the indu-
strialized countries in our data sample. Other measures of human capital and openness,
viz. initial secondary school enrollment and real exchange rate distortions, respective-
ly, have also been tested, with similar results. These results are available upon request.
Sensitivity Analysis
We will test the sensitivity of our results for the institutional vari-
ables by taking into account the problems of non-normally distributed
error terms, measurement errors, and endogeneity.
First, if the error terms are not normally distributed, OLS regression
will result in inefficient estimates. Following D'Agostino et al. (1990)
and Royston (1991), joint tests for skewness and kurtosis were per-
formed on the error terms in specification (iv) and (v) in Table 1, which
did not reject the hypothesis of normally distributed error terms at the
5 percent level.
A more severe problem might be measurement errors associated with
the use of proxy variables. Measurement errors will bias the estimated
coefficients, and may thus lead to incorrect inferences from the regres-
sion results based on OLS. The true maximum likelihood estimates can
be obtained by running reverse regressions (see Klepper and Learner
1984). We therefore regress those independent variables that we sus-
pect to be measured with errors on the variables in the original direct
regression, including the dependent variable. We then solve each regres-
sion equation for the originally dependent variable and obtain the
implied coefficients. These coefficients, in turn, put a lower and a high-
er bound on the true estimate. If the signs of the estimates from the
reverse regressions do not change, the estimates are robust to measure-
ment errors. On the other hand, if they do not have the same sign, the
estimates are unbounded and one should be careful when making infer-
ences from the OLS regression.
The results from the reversed regressions procedure are presented
in Table A3 and A4 in the Appendix. The coefficients in the specifica-
tion that contains the bureaucratic efficiency variable do not change
signs, while in the specification with the property-right variable the
dummy LATAFR does. This result, however, can be disregarded as the
variable does not involve any measurement problem. Thus, the results
seem not to be sensitive to measurement errors.
Finally, there might be an endogeneity problem between, in partic-
ular, the increase in the stock of FDI and the dependent variable since
it is possible that countries with high growth rates of GDP will attract
more FDI. In order to test that the estimated coefficients are not biased
due to misspecification, we perform a Hausman specification test for
contemporaneous correlation between the error term and GFDI on spec-
ifications (iv) and (v) in Table 1. In order to conduct the Hausman test
we need instrumental variables that are highly correlated with the var-
iables for which they are instruments but not with the error terms. Look-
ing at the Simple Correlation Matrix, SCHOOL seems to be a reason-
able candidate. According to the test results, though, we cannot reject
the null hypothesis of no contemporaneous correlation. In other words,
estimation with the OLS method seems appropriate.
Appendix
* Developing countries.
INV POPG IGDP GFDI OPEN SCHOOL PROP BUREFF FDI LATAFR
INV 1.00
POPG -0.55 1.00
¡GDP 0.63 -0.71 1.00
GFDI 0.32 -0.45 0.44 1.00
OPEN -0.66 0.63 -0.56 -0.25 1.00
SCHOOL 0.59 -0.75 0.84 0.45 -0.53 1.00
PROP -0.48 0.61 -0.58 -0.24 0.59 -0.62 1.00
BUREFF 0.49 -0.49 0.67 0.31 -0.46 0.69 -0.57 1.00
FDI 0.06 -0.05 0.22 -0.19 -0.09 0.07 -0.15 0.13 1.00
LATAFR -0.37 0.42 -0.32 -0.27 0.55 -0.30 0.17 -0.13 0.03 1.00
References
Abramovitz, M. (1986). Catching Up, Forging Ahead, and Falling Behind. The Journal
of Economic History 46(2): 385-406.
Balasubramanyam, V. N., M. Salisu, and D. Sapsford (1996). Foreign Direct Investment
and Growth in EP and IS Countries. Economic Journal 106 (January): 92-105.
Barro, R. J. (1991). Economic Growth in a Cross Section of Countries. Quarterly Jour-
nal of Economics 106 (2): 407-443.
Baumöl, W. J. (1986). Productivity Growth, Convergence, and Welfare: What the Long-
Run Data Show. American Economic Review 76(5): 1072-1085.
Penn World Tables (Mark 5.5) (1993). National Bureau of Economic Research, Harvard
University.
Pritchett, L. (1996). Where Has All the Education Gone? World Bank Policy Research
Working Paoer 1581. World Bank. Washington. D.C.
Royston, J. P. (1991). Comment on sg3.4 and an Improved D'Agostino Test. Stata Tech-
nical Bulletin 3: 23-24.
Sachs, J. D., and A. M. Warner (1995). Economic Convergence and Economic Policies.
Working Paper 5039. National Bureau of Economic Research, Cambridge, Mass.
Scully, G. W., and D. J. Slottje (1991). Ranking Economic Liberty Across Countries.
Public Choice 69(2): 121-152.
Stern, N. (1991). The Determinants of Growth. Economic Journal 101 (January):
122-133.
Summers, R., and A. Heston, A. (1991). The Penn World Table (Mark 5): An Ex-
panded Set of International Comparisons, 1950-1988. Quarterly Journal of Econom-
ics, 106 (2): 327-368.