It looks like Sirius-XM is now contemplating bankruptcy (HT: Danny Sokol). There were quite a few critics of the Bush administration’s decision not to challenge the merger. Various antitrust commentators and critics (as well as rivals like the NAB) lined up on the side of enforcement, arguing that the merger would lead to a monopoly in the satellite radio market and hammering the administration for its failure to challenge. The DOJ defended its decision in a press release with distributors (automakers) also coming to its defense.
One obvious testable implication of the theory that the deal would create a merger to monopoly (in whatever relevant antitrust market you’d like) is that the post-merger firm would earn monopoly profits. Of course, one of the challenges of retrospectives is that conditions change in the post-merger world. That’s obviously the case here. However, that does not mean that one can’t analyze whether the rate of return earned by the post-merger firm is consistent with the anticompetitive theory. I wonder if the critics of the DOJ decision not to challenge this merger would view a post-merger bankruptcy as inconsistent with their theory?
For those interested in some antitrust analysis of the merger, you might be interested in checking out the conference on Merger Analysis in High-Technology Markets that my colleague Thomas Hazlett and I put on at George Mason University on behalf of the Information Economy Project. While the entire conference was filled with some really nice papers, forthcoming in the Journal of Competition Law & Economics, the third panel features competing analyses of the Sirius-XM merger by Tom Hazlett (link to paper here) and J. Greg Sidak. You can access the papers at the same link or listen to the audio on iTunes here.
Related posts:
haw haw disappointed that disappointed didn’t keep his promise to leave
Thanks for the response to the question, Entity.
I note that I thought it universally accepted that a testable implication of an economic theory is not the same as a proposed legal test. For example, an implication of the “cartel” theory of RPM is that prices go up. But this would make a horrible legal test for RPM since the efficiency based promotional services theories also predict a price increase. A post-RPM realization of higher prices wouldn’t tell us much about which theory explained the practice. Yes, I realize some argue this is precisely the right test. But thats wrong for obvious reasons. Here, where all else constant, the anticompetitive theory predicts better post-merger performance ceteris paribus (a condition CLEARLY violated) and we observe worse financial performance, I thought it interesting to talk about how this would impact the priors of those who believe the anticompetitive story is more likely.
Josh’s original post said “One obvious testable implication of the theory that the deal would create a merger to monopoly (in whatever relevant antitrust market you’d like) is that the post-merger firm would earn monopoly profits.” Disappointed, David and I read this language as proposing a test. Josh later clarified that this was never his intention, but it is still interesting to discuss the proposed test on the merits (even if no one actually proposed it). My post 7 attempts to do so, and to address Josh’s question from his post 4: “But how should be [we] update our Bayesian priors? Does the observation make the merger less likely to have been anticompetitive, more likely, or no change?” I pointed out that the evidence on post-merger financial performance could in principle affect our prior as to whether the merger created efficiencies as well as our prior as to whether it created market power. Josh now asks how much our priors ought to be modified with respect to XM Sirius. My guess is very little (with the caveat that I have done no actual analysis to see if the following conjecture is correct): the collapse of automobile sales (the major market for satellite radio) combined with the credit crisis (which would likely push a leveraged firm toward bankruptcy even if the firm’s assets are most profitably employed in their current business) would likely dominate the effects of the merger on current financial performance, so the fact of an impending bankruptcy would provide little information about the explanation for merger.
Glad to have you back disappointed.
Against my better judgment, let me repeat a few points for clarity and I’ll be happy to let you have the last word:
1. It was not a test. That’s clear from a plain reading of the language in the post and subsequent comments. Not much else to say there.
2. Your language about necessary and sufficient conditions pretends that it was a test. It wasn’t. The argument was that it was a relevant fact to evaluating the ex ante theories. I don’t think that proposition is even within the scope of reasonable debate. You’re either arguing post-merger performance is entirely irrelevant — in which case you are wrong, or arguing that by my suggesting it is relevant that I’ve proposed it as a sufficient condition — in which case you’re mistaken.
3. I completely agree with you that other factors would be necessary to tease out whether the post-merger performance truly said something about the market power or efficiency stories — OR was attributable to other factors. Obviously many other factors exist. You’ve named some. There are others too.
This is a very simple point. Does the change in post-merger performance, understanding that market conditions have changed in various ways, adjust priors. In fact, my post explicitly mentions changes in market conditions. I don’t list them. Why should I? That’s not the question I want to ask in the post. Its whether the fact of post-merger bankruptcy influences the views of its critics.
How this is “ignoring facts” is beyond me. And will continue to be beyond me … being simple minded I suppose, is a significant tax on my comprehension from time to time.
One might imagine an answer in which a critic says “the post-merger performance gives me less confidence in my priors despite the fact that I think that these other factors are the primary causes of the bankruptcy” OR “it doesn’t change my priors at all.” I wondered which view is more prevalent among critics. This seems like an interesting question to me.
4. Anonymity is fine. I get it. But you started with comments about “knee jerk” ideology above and beyond commenting on the merits (or lack thereof) in the post. If you are willing to go beyond the merits of the post to attack me in this way, it is certainly fair game to point out that personal shots from an anonymous commenter are well …. easy to come by on the internet.
I have a personal email account. Its linked above. I’m happy to respond to the substantive stuff on the merits here on the blog, but not the personal stuff. If you’d like a response to that stuff, you can feel free to contact me offline via email privately.
I’ve promised you the last word if you want it.
Oh, one last response that I thought was inappropriate to include the previous response. Re: the crack about my slinging mud about your ideology anonymously. I didn’t, and don’t now, question your character or your intent. But the things you say, do reveal something about where you’re coming from.
That was my explicit point earlier.
Which I further clarified recently by pointing out that the proposed casually empirical enquiry was so low powered as to be irrelevant, and some would say misleading, in the specific instance in which it was posed (i.e. the Sirius XM merger).
So I’m not questioning your character. I am making what some would call a reasonable inference as to what it means that you’d propose the inquiry given the weakness of the empirical foundation of that inquiry (i.e. all the facts that contra-indicate the probativeness of the inquiry).
I did point out that there is another, perfectly reasonable, hypothesis to explain the OP. It is simply that you posted before thinking it through. That happens to all of us, and it doesn’t make you a bad person. It makes you human. While I have offered that hypothesis several times, it has not been embraced. Instead, you’ve dug in and defended the OP. While continuing to ignore the facts that I presented.
Finally, as to the anonymous point. I know that you understand that not everyone is an academic in the antitrust world, so that anonymity is sometimes required.
OK, against my better judgement, I’ll try once again.
While I (obviously) fully understand (in fact, I advocate) wanting to walk away from it, the query in the OP not only proposed a test, it posed it as a rhetorical question. To at least this reader (and I think others), the implication in the OP was that the question had answered itself — the prospective bankruptcy of Sirius XM revealed that the merger was competitively benign. You disavow that. In the interest of moving on, I’ll accept that.
So, I am very glad to see that you at least agree that the test (or if you prefer the semantics, the casually empirical inquiry) is, at best, low powered.
I think we’re all agreed that finding out that a firm went bankrupt post-merger is neither necessary nor sufficient to establish that the merger was competitively benign. It is not neccesary for obvious reasons — the vast bulk of competitively benign mergers don’t lead inexorably to bankruptcy. Nor is it sufficient, for the reasons that all the posts have discussed: there are many other factors, orthogonal to the the competitive impact of the merger, that affect whether a firm goes bankrupt.
Antitrust entity further pointed out that the converse also provides at best, a low powered test. That is to say, observing increased profitablity is neither neccesary nor sufficient to conclude that a merger was anticompetitive. Not neccesary, because all kinds of factors orthognal to the competitive impact of the merger could have impacted the profitability of the firm. Not sufficient, because, as AE pointed out, a competitively benign merger that leads to firm-specific efficiencies will lead to an observed increase in profitability.
So, the test (or if you prefer, casually empirical inquiry) that the OP posited is, in the most generous description, a very low-powered test.
The problem here is one of just ignoring the empirics and specifics. The OP was not an abstract question about some hypothetical merger. It referred specifically to the Sirius XM merger.
As I’ve mentioned, in this specific merger, there are many facts available in the public record that make it clear that the financial results of the firm are substantially (possibly determinatively) affected by decisions that the CEO made that have nothing to do with the competitive impact of the transaction.
The OP simply ignored all these facts that strongly suggested that the test (or casually empirical inquiry), applied in the specific instance in which it was raised, is so low powered that it is better described as meaningless. Some might even say, misleading.
All the subsequent posts from the host authors here have reveled in abstract discussions, fierce defenses of the OP, and in “mudslinging”. But none have responded to any of the facts.
How do you do that? How do you propose a test/casually empirical inquiry about a specific transaction, ignore the facts at the start of the inquiry, and then continue to ignore the facts in subsequent posts?
If nothing else, what about the price increases? Do you seriously believe that the recently announced price increases would have happened if two independent satellite radio firms were competing in an environment where demand from their primary customers (the auto companies) had gone further south than Antarctica? Am I saying that any observed price increase implies that the merger was anticompetitive? No. But I am saying that this particular observed price increase, observed in circumstances when demand had just suffered severe negative shocks, is strongly suggestive (not conclusive, but strongly suggestive) that the merger did reduce competition. I am also saying that this particular casually empirical inquiry is much more probative than the one posited in the OP. This one is not confabulated by the CEO’s financial/strategic decision. And the observed price change is in the opposite direction of what one would expect given the very large negative demand shock that they have suffered.
Did you know about the financing problems at Sirius XM when you made the OP? Did you know about the poison pill strategy to fend off a hostile takeover by Ergen/Echostar? Did you know that Sirius XM recently anounced price increases?
Obviously, the last set of questions are somewhat rhetorical. An answer of “yes” implies that you knew that the proposed test/casually empirical inquiry was essentially irrelevant. An answer of “no” is a better answer, but still not a very good one. A couple of minutes of googling would have yielded those facts, with the benefit of which, the question wouldn’t have been posed (the question being specific to this merger).
A last point on Mr. Manne’s final post. Gosh, I sure hope we don’t end up in a world where we require only (ex-post) efficient transactions. Leaving aside the ex-ante / ex-post problem, I sure hope that despite what we see happening around us, we don’t end so utterly repudiating the property rights that form the foundation of our society.
Hey–give that man/woman a cigar! Let me repeat: it’s not a “test,” not even a “proposed test.” It’s a statement of fact. I really don’t get this. Is everyone so scared of what retroactive analysis might uncover that they see in it an evil Chicago School plot? Yes, the fact that the combined firm went bankrupt should indeed bear on our confidence that the merger was efficient. Frankly, the fact that firms go bankrupt should bear on our confidence that we know anything about how the economy will work.
Now, there is indeed a subtle anti-enforcement point in this: Antitrust laws prohibit market power, they don’t require moves toward efficiency. So there’s an asymmetry: If the bankruptcy tells us that we are less confident that the merger was anti-competitive, that’s an argument that enforcement would have been inappropriate. To the extent that we learn that the merger was not necessarily efficient, however, we have not learned that we should have prohibited it–we’ve only learned that it is more likely that it was a poor decision from the firms’ point of view. (Maybe we learn in some cases that an efficiency defense wasn’t as strong as we thought). So your point is well-taken, but I still think Josh’s framing of the question (do we learn anything about the correctness of our enforcement theory from the subsequent bankruptcy?) is the more relevant one.
Entity: this is late enough in the thread that you should not be using the term “proposed test.” The question was whether the post-merger bankruptcy is relevant to evaluating the anticompetitive theory. I don’t think anybody seriously contends that post-merger performance is not relevant (including you). Its trivially obvious to know that one actually doing the analysis would want to be careful to control for all sorts of things. You raise the interesting additional point that the post-merger bankruptcy might also reflect how we think about ex ante efficiency justifications. I totally agree — though I suspect with the right data one might be able to tease out market power effects from efficiency effects.
Disappointed (just in case you’re still reading):
No. The question “I wonder if the critics of the DOJ decision not to challenge this merger would view a post-merger bankruptcy as inconsistent with their theory?” is simply not a test under any reasonable reading. For instance, one might answer: “yes, the data allows rejection of the theory” or “the observation weakens my confidence in the theory but I still think its the best explanation of the merger” or “no, the data doesn’t update my priors at all.” Its a simple question. Understanding the various changes in conditions and other factors that might contribute to post-merger bankrupcty, and without access to a comprehensive regression analysis to tease these factors out (which would be a publishable paper not a blog post!), what do the critics think about whether the post-merger observations say about the competing theories about the merger ex ante? Its a perfectly reasonable question to ask. And if anybody, including you, has misread me as meaning something else I hope the question is now as clear as can be.
But lets be clear, nowhere, and I mean nowhere, in the post do I take the position anywhere close to what you attribute to me, i.e. a legal “test” that says post-merger bankruptcy means there could not have been creation of market power ex ante. I’m baffled of this reading and do not think it is made in good faith.
I refuse to engage in discussions about how “knowing you” would reveal I was off base about your ideology given that you’ve chosen to take the courageous approach of slinging mud about my own simple mindedness and ideology anonymously on the internet.
Lastly, you’ve got your interpretation of the post in the comment thread and I’ve got mine. I’m perfectly happy to let the readers who chose to continue coming to the blog be the judge of the substantive merits of what I write or whether I continue to be worth reading.
If the proposed test involves making an inference as to the motive for merger from the fact of (a presumed impending) bankruptcy, it has another problem: the test doesn’t discriminate between the market power explanation and the efficiency explanation. That is, both explanations should raise merged firm profits relative to what the firms would have earned absent the merger. If the fact that a firm goes bankrupt “bear[s] retroactively on our confidence” that the merger created market power, it should also bear retroactively on our confidence that the merger generated cost savings or other efficiencies.
Josh can speak for himself, but, Disappointed, the statement you identify as a test in Josh’s initial post is no such thing. It is exactly as he points out in his comment: The bankruptcy is a fact, perhaps inconsistent, perhaps consistent, with the anticompetitive theory, permitting Bayesian updating of our confidence in the theory. This is precisely what Josh is asking when he asks whether the fact is inconsistent with the prior. I think you assume that he means to imply that if bankruptcy is inconsistent, the theory must be thrown out. Quite the contrary, Josh is far too careful to make that claim in the face of changed circumstances, and, as he clarifies in his comment, asks for marginal updating, not wholesale dismissal (or acceptance). You go on to say that Josh said that “if the firm goes into bankruptcy, it can’t have market power.” I’m still looking for any words to that effect in Josh’s post, but I’m pretty sure he never said any such thing. Put it this way: I’m confident that Josh agrees that firms with market power can go bankrupt, for a host of possible reasons–but the fact that a firm goes bankrupt may bear retroactively on our confidence that the firm had market power to begin with.
Now I’m even more disappointed, because you’re twisting and misrepresenting my post and your post.
I never said that looking at post-merger performance is simple minded or involves ideology over facts. What I said is that taking a simple minded look at post-merger financial performance — without considering the impact of strategic decisions that the CEO made regarding financing, corporate control, etc. that would also have affected the financial performance regardless of the merger — amounts to mispractice at best, and, arguably — given the fact that you are obviously neither stupid nor a bad economist — a case of ideology over empiricism.
At the same time, you claimed that I did not read your post carefully, because “I (i.e. Josh) didn’t propose a model or test”. Josh, your original post queried: “I wonder if the critics of the DOJ decision not to challenge this merger would view a post-merger bankruptcy as inconsistent with their theory?” That *is* a test that you proposed. How do you claim otherwise?
Look, here’s the bottom line. You proposed a very simple test: if the firm goes into bankruptcy it can’t have market power. I pointed out this doesn’t control for many of the determinants of bankruptcy that were in play in this case. I also pointed out that you can look directly at prices, which increased here.
You chose to simply ignore both those points, and instead fling around rhetoric.
Your prerogative to do so. While it is not a particularly effective way to convince, it is a fine response if you just want to preach to the choir. As such, it reveals something about why you write.
Just saying.
Look, I’m guessing you posted the OP in haste, and hadn’t quite thought through what you were proposing. It is undeniably simple minded (it doesn’t control for anything!), and doesn’t control for all kinds of other factors that we know to have been at play wrt Sirius XM that would impact the observed rate of return.
Since you’re a good economist, you’re perfectly aware of the importance of doing these tests correctly by controlling for these other drivers. What’s left as an explanatory hypothesis? Someone whose ideological reflex triumphed over careful thought and empiricism. Happens, but when it does, it behooves us to admit it and correct.
Instead you’re choosing to dig in. Not a good decision. Never is. Nor is launching “counter allegations” about the ideology of the person who points out that your post is flawed, or saying that I didn’t engage in empirical analysis or thought.
Let’s do the ideology first. You don’t know me, and you don’t know my ideology. I’ll say this — you’re way off base in what you’re saying. Your last comment in this regard is way way off base, and reveals lack of judgement (on many levels).
Now let’s turn to empiricism — I gave you lots of things to look at — payments to on-air talent, financing choices, poison pills in a takeover battle. None of these are abstract — they are actual factors affecting Sirius XM’s financial performance. How do you then characterize my comments as devoid of empirical analysis and thought.
I am, now, very disappointed.
Enough so that you’ve lost a reader. Learned something about you that I didn’t know earlier.
Entity, no I do not believe accounting returns are good for this and never said so — and also said that the change in conditions made this extremely difficult to do. I think we basically agree.
What I did write, consistent with Entity and David, was that the interesting question was whether the economic performance of the post-merger firm was consistent with the anticompetitive theories of the merger. In fact, that is an interesting question. They provide a plausible story (and answer to the question) that would allow the theory to be consistent with the observed evidence. One can imagine a plausible story that runs the other direction. This is what made it interesting to ask in a blog post. I never argued that the poor performance rendered it IMPOSSIBLE for the theory to stay intact! But how should be update our Bayesian priors? Does the observation make the merger less likely to have been anticompetitive, more likely, or no change?
I really don’t get the criticism from Disappointed. I didn’t propose a model or a test — much less a simple minded one. Careful reading makes this obvious. And of course looking at changes in prices and all sorts of other things (natural experiments where possible, demand models, simulations) would be very useful! All of the things you propose (controlling for various factors to predict post-merger pricing) are a proxy for asking if the post-merger firm gained market power. So is looking at actual returns. There is an entire event study literature on this. But surely you are not suggesting looking at post-merger performance is simple minded or involves ideology or theory over facts? That doesn’t make any sense at all.
Anyway, if asking the question whether the the post-merger performance is consistent with the anticompetitive theory elicits such a reaction about things that are “misleading” and “ideological” rather than empirical analysis and thought … I wonder what wouldn’t? Disappointed’s interpretation of the short blog post asking how critics of the merger are interpreting the real world post-merger performance is quite revealing about ideology in its own right.
This is really disappointing. Are you really proposing as simple minded a test as this? Antitrust entity and David already posted good points. Let me add some more. Have you done any reading on the financing of the deal and the interaction between those issues and the takeover battle between Karmazin and Ergen? The financial straits of Sirius-XM have a lot to do with the strategic decisions made by Karmazin — paying talent, *wild* bets on financing choices where the bets came bust once the credit freeze hit, scorched earth/poison pills to fight off the Ergen/Echostar takeover, etc. Nor have you controlled for the purchase price (not that I know that it was a factor here, but the purchase price has been a critical factor in similar “analyses” in other matters). All of these things must be factored in to the scenario where you look for Sirius’ profits without the anticompetitive rents. Yet you did none of the above. Let’s look at this another way though. Have you also thought about perhaps looking directly at price evidence? Sirius-XM announced price increases recently. What do you suppose the likelihood of that would have been if they were independent firms in this market environment (i.e. look at the shape of their automobile company customers)? Look, this is really a disappointing post by you. It is so simple minded that it is misleading and reflects knee-jerk ideology, rather than empirical analysis and thought. Think about whether you want to continue putting it out there. It sticks out like a sore thumb compared to your other work, and thus does not serve you well.
Following up on the perceptive questions posed by antitrust entity, I’d say it’s quite likely that the transaction was financed with a lot of debt. This would have left the combined entity highly leveraged, and prone to failure in response to a strong inward shift of the demand curve during the recent economic contraction.
But if the economic conditions existing at the time of the merger continued to the present, or even improved, it’s quite possible that the combined, highly leveraged entity would have enjoyed something akin to “monopoly profits.”
“Of course, one of the challenges of retrospectives is that conditions change in the post-merger world. That’s obviously the case here. However, that does not mean that one can’t analyze whether the rate of return earned by the post-merger firm is consistent with the anticompetitive theory.” Two questions.
1. Do you believe that accounting returns are likely to be good proxies for economic returns? (Subsidiary question: if so, should the FTC start to collect line of business data again in order to identify poorly-performing markets using accounting profits data?)
2. Even if you trust the accounting data, why don’t changing conditions make this analysis problematic? Don’t you have to construct the rate of return in the but-for world in order to evaluate the merger to monopoly theory? And isn’t that analysis more difficult when conditions change?