September 29, 2014 | By Justin Sydnor | Back to blog

Multinational companies’ increasing importance to the business world requires that we understand the role that location plays in firm performance. In particular, how well can multinational companies transfer their practices from their home country to their operations abroad? When investigating multinationals’ ability to transfer their best practices to different locations, some studies have documented convergence in organizational practices and productivity levels across plants in different countries, whereas others find that firms have difficulty replicating the practices of the plants in the home country abroad.

However, it is often challenging to empirically identify the effect of location-specific factors on firm outcomes, because it is difficult to separate their role from other aspects of the firm. Moreover, even if firm and industry-specific factors are controlled for, companies might be producing different products (and for different markets) in different places, thus making the comparison of outcomes across locations hard to interpret.

My colleague Nicola Lacetera of the University of Toronto and the National Bureau of Economics Research (NBER) and I wanted to know whether location-specific factors, such as the education and attitude of the local workforce, supplier networks, institutional infrastructure, and local “culture” are important for understanding persistent heterogeneities among firms. We decided to address this question in the context of the automobile industry.

Using a unique data set of over 565,000 used-car transactions at wholesale auctions, we tested whether the long-run value and quality of otherwise identical cars depends on the country of assembly.

We exploited the natural experiment provided by the establishment of assembly plants in the U.S. by Japanese auto manufacturers and the fact that some of the most popular Japanese car models are assembled in both Japan and the U.S.

Because a number of Japanese car models, including some very popular, high-volume automobiles, are produced in plants in both Japan and the U.S. (e.g., Honda Accord, Toyota Camry), we were able to assess whether the short- and long-run quality of cars is affected by the country of production.

Our data contain detailed information on each car that allowed us to identify essentially identical cars at the auction; for instance, the sample of 2002 Honda Accord SE four-door sedans with four-cylinder engines sold in 2005 at the auction site in Chicago. We were also able to use the Vehicle Identification Number (VIN) that lists the country of manufacture for each vehicle to investigate whether there were long-run quality differences for otherwise identical Japanese-make cars, depending on whether they were assembled in the U.S. or Japan.

Our primary measure of quality was simply the sale price. The bidders at wholesale auctions are used-car dealers who intend to resell cars to final customers. Given the competitive nature of these auctions, any observable quality differences should be incorporated into the sale price of used cars. Thus, for our main analysis, we estimated whether the sale price of cars was affected by whether they were assembled in Japan or the U.S., after controlling for other important characteristics (e.g., mileage, auction location).

We found evidence that the Japanese-assembled cars sell for more, on average, than those built in the U.S., but the estimated difference is only $62. The average differences are driven almost entirely by older-model Toyotas, for which we found a more meaningful difference between the Japanese- and U.S.-built cars. Japanese-built Hondas and more recent Toyota models are no more valuable than those built in the U.S.

We found the same basic pattern when we looked at other direct measures of quality, including an estimate of the costs to recondition the car, whether it had known defects, and whether it sold at the auction. In each case, we found that there were only slight differences in favor of the Japanese-assembled cars and that these were usually only significant for older-model Toyota cars. Overall, our results suggest that Japanese automakers have been successful, though perhaps with some lag in the case of Toyota, at transferring their high-quality practices to their U.S. transplants. Our findings also suggest that there is not an inherent limitation to the U.S. manufacturing environment that prevents the production of high quality cars in America.

The case of Toyota appears to be especially interesting. Our data suggest that prior to around 2002, Toyota cars assembled in Japan were of modestly higher quality as they aged than equivalent cars assembled in the U.S. However, around 2002, Toyota launched a re-design of their flagship Camry, and news reports suggest that at that same time, Toyota put effort into redesigning some of their assembly processes and communications in the US plants. After that redesign effort, the quality differences we measured for Toyota cars assembled in the U.S. disappeared and might even have reversed a little. This suggests that adapting best practices to foreign plants is possible but may involve a very conscious design process, rather than happening as gradual improvements.