locations (municipalities) considered here is much more homogeneous than
in cross-country data. For instance, in contrast to an international setting,
other tax parameters such as taxes on income are identical across German
municipalities since they are levied at the level of the federal union. More-
over, sub-national data allow one to control for the heterogeneity of locations
within countries which is more difficult with national data.
The location of regional headquarters is important because they are typi-
cally tied to the main production plants in the same municipality. Attracting
a regional headquarters of a foreign MNE to a municipality thus promises
jobs and local business tax income. There is anecdotal evidence of munici-
palities which lower their tax rates so as to explicitly attract foreign firms.
One well-known example is the small town of Holzkirchen, close to Munich
in Upper Bavaria, that lowered its business tax rate to appeal to Sandoz, a
big pharmaceutical firm.3 Another example constitutes the municipality of
Amering which managed to lure Kathrein, the world leader in satellite dishes,
to locate its headquarters there, by lowering its local business tax rate.4 We
would be able to provide further examples, but these are enough to illustrate
how local business tax rates may affect a firm’s location choice.
The majority of municipalities in Germany does not attract any foreign
headquarters, however. This could be to do with (high) business tax rates
as well as other factors. We will argue that (low) business tax rates alone
are not enough to attract foreign headquarters for the average municipality.
Econometrically, the fact that numerous municipalities are incapable of at-
tracting any foreign MNE headquarters suggests that the problem of interest
may not be approached appropriately by a linear regression model. More-
over, many of the municipalities which successfully attract foreign MNEs
are interested in how a given number of foreign investments to just one country (Germany)
is allocated within the country in response to tax rates. Hence, we disregard multilateral
considerations of investors - e.g., decisions related to whether invest in Germany at all.
Second, the huge number of host jurisdictions involves a relatively large fraction of munic-
ipalities where no investment is undertaken at all. The fraction of zeros would necessarily
rise more than proportionately if we distinguished the number of investors by their country
of origin. This would call into question the applicability of some of the methods applied
here. Hence, we have to relegate questions related to the origin of foreign investments
with this data-set to another paper.
3See http://www.sueddeutsche.de/wirtschaft/artikel/10/51958/ reporting that
Holzkirchen had lowered its local business tax rate by 30 percent, making it the
second-lowest in the state of Bavaria, in its (successful) attempt to attract Sandoz.
4See http://www.kathrein.de/de/presse/cont/texte2005/pi0553.htm.