,
15 tweets,
5 min read
Read on Twitter
Our @OECDagriculture study on concentration in seed markets contradicts several of the points made in this @CivilEats article. I'd like to pick on one argument in particular - that mergers lead to less innovation. This has not been shown. 1/
It's true that a 2004 study (sciencedirect.com/science/articl…) concluded that "Increases in seed industry concentration have reduced biotech research intensity in the United States in the 1990s"; this study is the one referred to by the @CivilEats article (and many others) 2/
But that study has a flawed methodology. First, because the theoretical model assumes identical firms, they use (1/Number of firms) as concentration measure - a case of confusing theoretical simplification with empirical reality... 3/
Second, I'm very, very doubtful that you can do two-stage least-squares with 9 or 10 observations, as the authors do. In fact, even a simple OLS regression would be suspect with such a small sample size... 4/
But perhaps more importantly, they measure "market concentration" not by using actual market data but with the same dataset of GM field trials - i.e. they are not measuring market concentration but rather *the number of firms engaged in field trials*. 5/
The empirical strategy boils down to comparing the total number of field trials per crop divided by USD sales of that crop, with the total number of firms doing trials. They pick up a positive correlation, i.e. more firms = more trials 6/
But here is something odd. Their own data (Figure 1 in the paper) shows an *increase* over time in the number of firms registering field trials! And indeed, over this period the number of field trials went up as well... 7/ 
It's not obvious how this supports the argument that *fewer* firms leads to *less* innovation. Over time, their data has *more* firms doing trials, and total nr of trials going up. You can't simply put a minus sign in front of it to conclude something about mergers! 8/
Some other studies have tried to use the same data (on GM field trials), but I also have concerns about those, as discussed in our OECD report - see p. 101-104 (oecd.org/publications/c…) 9/
Here is how the raw data of the APHIS GM field trial database looks like. If you calculate concentration measures using this data (as some papers do), you'll conclude that *higher concentration* is associated with *more* field trials... 10/
In reality, what happened is that Monsanto emerged, started doing massive amounts of GM field trials, then scaled down again. Since Monsanto became such a large part of the total, it influences concentration measures calculated using these field trial data 11/
Long story short, when analysing the APHIS field trial database, don't use the same database for your measures of market concentration... 12/
To add a more constructive element in the debate: we have also done our own empirical analysis on innovation and market concentration in seed markets using EU data on new entries to the National List, and actual data on market concentration 13/
At least based on these results, there doesn't appear to be much of a link between market concentration and innovation rates in EU seed markets. This analysis has plenty of limitations, but it's the only one of its kind as far as I know. (p. 171 in the report) 14/
Long story short, I don't think there is sufficient evidence so far to support the claim that increasing concentration in seed markets reduces innovation. I'm not saying it is impossible; just that the evidence so far does not show it...
