FDI of German Companies During Globalization and Deglobalization

Abstract

Based on micro-level data of German companies from 1873 to 1927, we identified horizontal and vertical FDI applying a Knowledge-Capital model and analyzed individual FDI decisions. Our KC model revealed that market-driven FDI predominated; however, wage gaps and differences in human capital stimulated cost-driven FDI flows, which accounted for up to 10% of total FDI. On an individual level, large companies with high profitability conducted more FDI. Higher tariffs after WWI enhanced FDI, as companies could circumvent trade barriers—but declining openness reduced FDI. In spite of disintegration after WWI, the propensity to invest increased due to higher market concentration and firm specific investment patterns—albeit industry agglomeration effects were of minor importance.

Appendix: Variable definitions and data sources

Endogenous variables:
log (invest jt )	Natural logarithm of aggregated FDI flows between Germany and the respective host country (indexed j) at time t. We convert FDI flows into German currency and deflate using Hoffmann’s price index; thus, FDI is expressed in real terms. Data source: Amount of investment was deflated, using Hoffmann’s (1965, p.601, col.15) price index.
fdi it	We measured the binary choice of individual FDI; hence, the dummy variable takes the value one if firm (indexed i) conducted FDI in year t.
Explanatory variables (for the KC model):
Sum_gdp	We calculated the sum of Germany’s and the host country’s GDP: ln(GDPf)+ln(GDPHome). Data source: Maddison (1995, 2003).
Dispersion	To indicate the degree of similarity in terms of market size, we defined the dispersion index as follows: (ln(GDPf)−ln(GDPHome))2. Data source: Maddison (1995, 2003).
Distance	Distance between Berlin and the host country’s capital in km. Data source: http://www.macalester.edu/research/economics/PAGE/HAVEMAN/Trade.Resources/Data/Gravity/dist.txt
Language	The dummy variable takes the value of one if German is the official language of the host country and zero otherwise.
Patents	The variable is defined as the number of foreign patents, which have been extended (so-called high-value patents), of companies of the host country in Germany. This indicator should highlight the degree of technological progress of the host county. Data source: Kaiserliches Patentamt (1875–1927).
Enrollment	We used primary school enrollment rates in percent. To measure the skill difference, we calculated the difference of the host country’s enrollment rate and the top 25% of all host countries (enrollment); thus, a negative value indicates that the respective host country has lower enrollment rates compared to the leading 25% of host countries. Data source: Benavot and Riddle (1988) and Lindert (2004)
Wagediff	We determined the difference in real wages of the host country compared to Germany; ln(wageHome)-ln(wagef). Hence, a positive value indicates that the host country has lower real wages than Germany. Data source: Williamson (1995).
Protect	We measured protection by the average tariff rate, which is defined as the ratio of custom revenues to the total value of imports. Data source: Accominotti and Flandreau (2006), Clemens and Williamson (2004) and Rubio (2006).
Openness	We determined the ratio of the value of merchandise exports relative to GDP. Standard practice is to add imports and exports and divide by GDP—but we do not have reliable import data for 1873–1927. We use Maddison’s (2003) data for 1870, 1913 and 1929 on merchandise exports to construct the openness indicator (exports/GDP). Data source: Maddison (2003).
Additional explanatory variables (for logit/probit models):
Size	We used the natural logarithm of total equity as measure for firm size
ROE	Return on equity as a measure of firm’s profitability was defined as net income (distributable to common shareholders) divided by total common equity (excluding preferred equity).
Foundation	Year of establishment when joint stock company entered share register.
Past_fdi	Indicator variable which takes the value one if the firm invested abroad during the last 10 years and zero otherwise.
HI	The standardised Herfindahl Index shows the level of market concentration in the respective industry. We determined the sum of squared market shares (SSMS) based on total equity (as industry revenue figures are not available) and standardized as follows: (SSMS-min(SSMS))/(max(SSMS)-min(SSMS))*100.
Chemical/electric/machinery/metal	Used as industry dummy variables, except in Table 7 as we determined industry agglomeration effects defined as the number of FDI conducted by the industry within the last decade in the respective host country.
	The individual industries listed in the source on which have companies were:
	Asphalt
	Baths
	Banks
	Real estate
	Breweries
	Cement
	Chemicals
	Iron and coal
	Electrotechnical
	Dyes
	Other finance
	Gas
	Grain mills
	Glass
	Other raw material
	Rubber
	Colonial goods
	Other food
	Machinery/engineering
	Metal processing
	Musical instruments
	Food processing
	Oil mills
	Optical instruments
	Paper and printing
	Petrol
	Porcelain
	Gun powder
	Chocolate
	Other
	Toys
	Starch
	Textiles
	Railways and urban commuting
	Clocks
	Other transports
	Insurance

Main data source (except mentioned in the table above):

Handbuch der deutschen Aktiengesellschaften (1897–1901, 1912, 1927).