Abstract
We conduct a Meta-analysis of 54 papers that study the relationship between multinationality and firm performance. The aim is to understand if any systematic relationships exist between the characteristics of each study and the reported results of linear and curvilinear regressions to examine the multinationality-performance relationship.
Our main finding, robust to different specifications and to different weights for each observation, is that when analysis is based on non-US data, the reported return to multinationality is higher. However, this relationship for non-US firms is usually U-shaped rather than inverted U-shaped. This indicates that US firms face lower returns to internationalization than other firms but are less likely to incur losses in the early stages of internationalization.
The findings also highlight the differences that are reported when comparing regression and non-regression based techniques. Our results suggest that in this area regression based analysis is more reliable than say ANOVA or other related approaches.
Other characteristics that influence the estimated rate of return and its shape across different studies are: the measure of multinationality used; size distribution of the sample; and the use of market-based indicators to measure firm performance. Finally, we find no evidence of publication bias.
Our main finding, robust to different specifications and to different weights for each observation, is that when analysis is based on non-US data, the reported return to multinationality is higher. However, this relationship for non-US firms is usually U-shaped rather than inverted U-shaped. This indicates that US firms face lower returns to internationalization than other firms but are less likely to incur losses in the early stages of internationalization.
The findings also highlight the differences that are reported when comparing regression and non-regression based techniques. Our results suggest that in this area regression based analysis is more reliable than say ANOVA or other related approaches.
Other characteristics that influence the estimated rate of return and its shape across different studies are: the measure of multinationality used; size distribution of the sample; and the use of market-based indicators to measure firm performance. Finally, we find no evidence of publication bias.
| Original language | English |
|---|---|
| Pages (from-to) | 23–47 |
| Number of pages | 25 |
| Journal | Management International Review |
| Volume | 52 |
| Issue number | 1 |
| Early online date | 18 Oct 2011 |
| DOIs | |
| Publication status | Published - Feb 2012 |
Bibliographical note
The final publication is available at link.springer.comKeywords
- meta-analysis
- multinationality
- performance
- publication bias