Please Do Not Cite Without Permission - Paper in Progress. Presented at the 3rd Political Networks Conference, Duke University, May 20, 2010.


An extensive literature in international and comparative political economy has focused on the how the mobility of capital affects the ability of governments to tax and regulate firms. The conventional wisdom holds that governments are in competition with each other to attract foreign direct investment (FDI). Nation-states observe the fiscal and regulatory decisions of competitor governments, and are forced to either respond with policy changes or risk losing foreign direct investment, along with the politically salient jobs that come with these investments. The political economy of FDI suggests a network of investments with complicated dependencies.

We propose an empirical strategy for modeling investment patterns in 24 advanced industrialized countries from 1985-2000. Using bilateral FDI flow and stock data, we examine the nature of the networks in relation to a set of covariates - in particular differences in tax rates between pairs of nations. Our statistical model is based on the methodology developed by Hoff (2005), Westveld (2007), Westveld and Ho ff (2009b). The model allows the temporal examination of each nation's activity level in investing and attractiveness to investors. Additionally, the model considers the temporal examination of reciprocity between pairs of nations, as well as the notion of clusterability. For both the flow and stock data, there exist a data set based on reports from senders (out-reported-data) and a data set based on reports from receivers (in-reported-data). We extend the model by treating these two data sets as independent replicates (for the flow and stock data separately), conditional on a mean parameter representing an underlying value of FDI, along with random effects within the variance portion of the distribution of the response that allows for discrepancy between the two data points (in and out data). Using a fully Bayesian approach, we also impute the missing data within a MCMC algorithm used to fit the model.