thus allowing for over-dispersion and providing a good fit to many types of
data.
For data like ours, with 86% of zero observations, zero-inflated (ZI) models
- which assume an extra proportion of zeros additionally to the zero obser-
vations arising from the count data distribution - should fit the data even
better. Zeros are allowed to occur as an outcome of two different regimes. In
one regime the outcome will always be zero and in the other one the standard
count process is at work resulting in either zero or positive values. The model
combines a binary variable describing the probability of extra zeros with a
standard count variable. The probability function is given by
ωi + (1 -ωi)g(yi = 0) ify = 0 ,
(1 - ωi)g(yi | yi > 0) ify ≥ 1,
where ωi can be a logit and g(yi) a Poisson or an NB density. All such
models (including Poisson, NB, and ZI versions thereof) can be estimated
by the maximum likelihood method. In our application, the binary process
reflects the economic suitability of a municipality for hosting a multinational
headquarters at all, and the conditional mean of the count process describes
the number of MNE headquarters that are actually attracted given a munic-
ipality’s general suitability for headquarters location.
Eventually, one might be worried about business tax rates being endoge-
nous. For instance, this would be the case if the average municipality al-
tered business tax rates to attract foreign headquarters in particular.18 Mul-
lahy (1997) derives moment conditions for generalized method of moments
(GMM) estimation of count data models with endogenous regressors and
valid instruments. Consider the Poisson regression model with a multiplica-
tive error term
yi = exp(x0iβ)ηi.
The following moment restriction follows from the first-order condition of the
18Notice that this is more than to say that municipalities used business tax rates to
attract firms in general. In our application, the average municipality is not able to attract
any foreign headquarters. This may be seen as an indication that the attraction of foreign
headquarters is not the most important policy objective of the average German munici-
pality. Hence, we expect the endogeneity issue as a subordinate one, here. However, we
mention and apply suitable methods for completeness and as a robustness check to the
conventional count data models.
12
More intriguing information
1. Multiple Arrhythmogenic Substrate for Tachycardia in a2. Examining the Regional Aspect of Foreign Direct Investment to Developing Countries
3. Campanile Orchestra
4. The name is absent
5. GROWTH, UNEMPLOYMENT AND THE WAGE SETTING PROCESS.
6. The name is absent
7. The Environmental Kuznets Curve Under a New framework: Role of Social Capital in Water Pollution
8. The purpose of this paper is to report on the 2008 inaugural Equal Opportunities Conference held at the University of East Anglia, Norwich
9. Popular Conceptions of Nationhood in Old and New European
10. The role of statin drugs in combating cardiovascular diseases