Archives For

Here’s another reason—as if you needed it—as to why getting a reasonable bailout of the mortgage-backed securities market through Congress was so difficult. Below I reproduce an analysis from Gregg Easterbrook of the Brookings Institute:

Supposing we assume the bailout is required, here is what bothers me about the plan so far: Taxpayers don’t get stock, what they get is warrants that can be exchanged for stock, and nonvoting stock to boot….  Even if the warrants are called [sic, Easterbrook means exercised], taxpayers get no voting positions…. A week ago, Warren Buffett rescued Goldman Sachs by injecting $5 billion in capital. Did Buffett bargain for warrants that can be exchanged at an unknown later date for nonvoting shares? No: He is not a fool. Buffett gave Goldman Sachs $5 billion in return for senior preferred stock, the kind that votes and also is more valuable than ordinary shares…. The United States Congress and the White House should use the public’s $700 billion to buy … senior preferred shares. Why are Congress and George W. Bush not simply following the road map laid out on this problem by the smartest investor of our era?

The problem is that this is factually mistaken from beginning to end. Easterbrook says that Buffett did not get warrants, but in fact he did—indeed, warrants to purchase $5 billion of common stock with a strike price of $115 per share and a five year term. Easterbrook says that the preferred shares Buffett received are “the kind that votes,” but they’re not. Like the six classes of preferred stock Goldman already has outstanding, Buffett’s shares do not vote in the election of directors. Like virtually all preferred shares, they have very limited voting rights that do not allow the holder any ability to control the affairs of the company. This is all set out in the press release Goldman Sachs issued to announce the transaction and in Item 303 of the Interim Report on Form 8-K it subsequently filed with the Securities and Exchange Commission.

More generally, it’s apparent from his blog that Easterbrook simply doesn’t understand what preferred stock is and how it relates to common stock. As most readers of this blog know, preferred stock is “preferred” not because it’s “more valuable than ordinary shares” but because it ranks ahead of common shares in order of payment. That makes it less risky than common shares but not generally more valuable than common stick. Indeed, since return is proportionate to risk, preferred shares generally have a lower expected return than common shares, and since control follows risk, preferred shares generally have less voting rights than common shares. The preferred shares Buffett purchased from Goldman are thus perfectly typical in these respects.

My point here is not that Easterbrook doesn’t understand what he’s talking about (though he clearly doesn’t). My point is that an intelligent person in the opinion-making class trying his honest best to inform the American public about the merits of the bailout can utterly fail to understand the issues at stake. That’s not because people like Easterbrook are fools (they certainly are not) but because the matters are so complex and technical that even informed, intelligent non-specialists can get them very badly wrong.

This is a genuine challenge for democracy. In The Federalist No. 10, Madison argued that representational republicanism is superior to direct democracy because it can “refine and enlarge the public views by passing them through the medium of a chosen body of citizens, whose wisdom may best discern the true interest of their country and whose patriotism and love of justice will be least likely to sacrifice it to temporary or partial considerations.” In other words, democratically elected representatives are on the whole likely to make better decisions than the people themselves, in part because they are better informed than the average voter. History has by and large proved Madison correct in this. Part of the problem in passing the bailout program last week, however, was that our elected representatives were dealing with issues that were often too complex and difficult even for intelligent, informed people who were trying hard to understand them. I do not for a moment suggest that anyone else should get to decide such matters except the elected representatives of the people, but I think we should realize that the world has become so complex that issues like this may begin to appear with increasing frequency.

I want to respond to some of the comments on my blog regarding whether part of the ideological difference between liberals and conservatives can be explained by their differing estimations of the elasticity of various curves.

First, Thom reminds us of Posner’s idea that the difference between liberals and conservatives is that liberals think that people are selfless but stupid whereas conservatives think that people selfish but smart. This idea is closely related to the one I proposed. For, if people are selfish but smart, then when the price of something rises, they will take note, seek out substitutes, and adjust their behavior accordingly—and hence the demand curve will be elastic. On the other hand, if people are selfless but stupid, when the price of something rises, either they will not notice or else they will be insufficiently self-interested to substitute other products; either way, they pay the higher price, and so the demand curve will be inelastic. Posner’s account, if true, thus plausibly explains why liberals and conservatives would systematically disagree about elasticities.

Second, the liberal-conservative divide over whether a requirement to show identification will discourage many people from voting may well be independent of questions of the elasticity of demand for voting. For example, suppose that, as liberals seem to think, the demand curve here is highly elastic (e.g., say elasticity of 2.0), but that, as the conservatives seem to think, the increase in cost to voters will actually be quite small, say .01%. Then the law requiring voters to show identification would reduce voter turnout by only .02%, i.e., one voter in 5,000. A conservative could then say that this was a reasonable price for us as a society to pay in order to deter vote fraud. I have no idea what the actual numbers are, but it would seem to make sense to lose one legitimate vote in 5,000 in order to preempt, say, one fraudulent vote in 4,000. Even if the percentage of fraudulent votes was less than the percentage of votes lost through the identification system, the gain from people being more confident in the system might make checking identification worthwhile. The defining difference between liberals and conservatives here may thus relate not to elasticities but to the perceived increase in the cost of voting on a percent basis.

Lest anyone be scandalized by the idea that it could be worthwhile to let some legitimate votes be lost, let’s remember that any system at the polling place that limits who can vote, including the present vote, deters some people from voting. It’s only a question of how many lost votes we think acceptable in this context.

Third, I agree with the many readers who said that any explanation of the difference between liberals and conservatives in terms of disagreements about elasticities is partial at best. It explains nothing, for example, about differing attitudes concerning, say, abortion. Differences on life issues, I think, go to differences on fundamental meta-ethical premises. Here, conservatives tend to think that morality is, in some sense, based on a human nature that is shared by all members of the biological species homo sapiens. Since human fetuses or embryos are members of that species as much as anyone else is, they are generally entitled to the same moral protections as anyone else. Liberals, on the other hand, tend to think that morality is, in some sense, based not on human nature per se but on the human capacity for, say, higher mental functions. Hence, members of the human species that are incapable of such functions (e.g., the very young, the severely disabled, etc.) are not generally entitled to the same moral protections that healthy adult human beings are. If some account along these lines is right, then the philosophical differences between liberals and conservatives run very deep indeed.

One of my favorite intellectual puzzles is figuring out what deep conceptual presuppositions cause some people to be conservatives, others to be liberals. That is, on a range of issues that would seem largely unrelated—say, abortion, affirmative action, and gun control—it turns that people’s positions are highly correlated. For instance, people who are pro-life tend also to be against affirmative action and against gun control, whereas people who are pro-choice tend also to be in favor of affirmative action and in favor of gun control. Why is this?

I’m still working on a general solution, but one thing is pretty clear. Conservatives tend to think that demand curves are elastic, liberals that they’re inelastic. Economists talk about demand for a product or service as being elastic if a 1% increase in price produces more than 1% decrease in quantity sold, inelastic if a 1% increase in price produces less than a 1% decrease in quantity sold. Elasticity is a precisely defined concept, but the basic idea is easy enough to understand: roughly, demand is inelastic if, when you raise the price, people keep buying the product at the higher price, but elastic if, when you raise the price, people cut back on their purchases of the product and do something else with their money.

So, for example, conservatives think the demand for crime is elastic: if you raise the price of crime to the criminal by increasing prison sentences, you’ll get a lot less crime. Liberals, on the other hand, tend to think that increasing prison sentences will have little effect on crime rates: in other words, they think the demand for crime is inelastic relative to prison sentences. Similarly for taxes. Conservatives tend to think that if you raise income taxes, people will work a lot less, whereas liberals tend to think that if you can raise income taxes, people will generally work as much as they did before the tax increase.

A fascinating role-reversal is thus at work in the voting rights cases that the United States Supreme Court heard earlier this week. As this story in the Legal Times explains, the Court is considering a constitutional challenge to an Indiana statute that requires citizens who want to vote to show at the polling place a state-issued photo identification such as a drivers license. Conservatives generally favor the law, and liberals generally oppose it, perhaps because the law is generally perceived as helping Republicans and hurting Democrats.

Whatever may be the real motives on either side, the Indiana Democratic Party and the ACLU say that the law is unconstitutional because it will deter people—especially old people, the poor, and minorities—from voting. They are thus in effect saying that the demand for voting is very elastic: make it even a little more difficult for people to vote, and many people will stay away from the polls. The conservative supporters of the law, on the other hand, are saying just the opposite: raising the effective cost of voting will not affect how many people vote because the demand for voting is inelastic.

Where does the truth lie? As a political conservative, I usually think that demand curves are pretty elastic. Nevertheless, all my intuitions run in favor of the view that the Indiana statute would not deter many people from voting. If I ask myself why my intuitions run in this direction, however, and if I’m being completely honest, I would have to say that I don’t really know.

Much discussion of corporate governance in the last few years has centered on reforms advocated by ISS and CII and indices of good corporate governance practice created and maintained by such groups. A new study by Roberta Romano, Sanjai Baghat, and Brian J. Bolton, however, concludes that there is “no consistent relation between governance indices and measures of corporate performance.” The authors continue,

[T]here is no one “best” measure of corporate governance: the most effective governance institution appears to depend on context, and on firms’ specific circumstances. It would therefore be difficult for an index, or any one variable, to capture critical nuances for making informed decisions. As a consequence, we conclude that governance indices are highly imperfect instruments for determining how to vote corporate proxies, let alone for portfolio investment decisions, and that investors and policymakers should exercise caution in attempting to draw inferences regarding a firm’s quality or future stock market performance from its ranking on any particular corporate governance measure. Most important, the implication of our analysis is that corporate governance is an area where a regulatory regime of ample flexible variation across firms that eschews governance mandates is particularly desirable, because there is considerable variation in the relation between the indices and measures of corporate performance.

The paper is entitled The Promise and Peril of Corporate Governance Indices and the full text is available on SSRN.

According to a news story from Reuters, a recent Tufts University study (available here) says that “if nothing is done to combat global warming,” then by the year 2100, “two of Florida’s nuclear power plants, three of its prisons and 1,362 hotels, motels and inns will be under water” because of rising sea levels. This is a rather dubious proposition since all of those assets would, I assume, outlive their useful lives long before they get submerged, and no one would rebuild in an area soon to be underwater. Still, let’s leave aside such objections. The study tells us that rising sea levels could cost Florida $345 billion a year and goes on to state that “the sort of mitigation efforts needed to restrict sea level rises to 7 inches or less would cost a U.S. state like Florida between 1 percent and 2 percent of GDP.” Hence, “doing something may seem expensive, but doing nothing will be more expensive.”

Steven Landsburg has written that “the antidote to naïve environmentalism is economics,” and I think that dictum applies here. According to this report from the Commerce Department’s Bureau of Economic Analysis, the GDP of Florida in 2006 (in chained 2000 dollars) was about $610 billion. One percent of that figure is $6.1 billion. The authors of the study thus argue that it’s a bargain to spend $6.1 billion to avoid a cost of $345 billion—which it would be, if we were talking about spending $6.1 billion today to avoid spending $345 billion today. The question, however, is whether we should spend $6.1 billion today to avoid spending $345 billion ninety-three years from now in 2100.

The Tufts study purports to take account of inflation, so we need be concerned only with the pure time-value of money and a risk premium. For the former, let’s say 1.7%, which is a commonly used estimate. The latter is harder to calculate, but I venture to say that predictions of costs almost a century in the future are at least as risky as small-cap stocks, and Ibbotson Associates has calculated the historical risk premium on such stocks to be 13.4%. So what happens if we discount $345 billion in 2100 back to present value using a 15.1% discount rate? Answer: that $345 billion becomes a mere $720,799. In other words, we’re being told that it’s bargain to spend $6.1 billion today to avoid a cost with present value of well under a million dollars. I think not.

A story in the New York Times explains that in Germany booksellers are legally prohibited from discounting books below the price set by the publisher. It’s not clear from the story, but it thus seems that Germany has a legally-mandated system of minimum resale price maintenance. Not surprisingly, this favors small bookstores. “In the United States chain stores have largely run neighborhood bookshops out of business. Here in Germany, there are big and small bookstores seemingly on every block.” The Times story goes on to explain that a similar system in Switzerland has recently been abolished, and now Swiss booksellers are selling books into the German market, undercutting the German booksellers.

The Times story provides very few details, and so it’s difficult to draw conclusions about what’s really going on here. Does the publisher have complete freedom to set the resale price? If so, given that US booksellers haven’t found RPM to be in their interest, why would German booksellers be any different? Why not just set the RPM price so low that even discounters would be selling above the RPM price? Or does the German system somehow facilitate a publishing cartel?

One indication that this might be the case emerges in the Times article. “Last year 94,716 new titles were published in German. In the United States, with a population nearly four times bigger, there were 172,000 titles published in 2005.” Again, it’s hard to tell, but if the publishers are making monopoly profits on books, perhaps they’re competing some of those profits away by publishing more titles in an effort to capture a larger share of the market. Compare how the airlines, when still regulated, competed on quality.

By the way, one thing is clear: at least for popular books, prices are a lot higher in Germany than in the US. You can buy the English version of Harry Potter and the Deathly Hallows from Amazon in the U.S. for $19.24, but the German version from Amazon.de in Germany costs EUR 24.90, or about $35.81. Part of the disparity is due to the current weakness of the dollar, of course, but even if the dollar and euro were of equal value, the German version of the book would still cost about 20% more.

Back in April, private equity fund J.C. Flowers, along with JP Morgan Chase and Bank of America, agreed to acquire Sallie Mae, the largest provider of student loans in the United States. Between then and now, Congress passed the College Cost Reduction and Access Act of 2007 (CCRAA), which reduced in various ways the subsidies the federal government provides the education loan industry. J.C. Flowers and the other buyers have declared that the passage of the CCRAA (either taken alone or in conjunction with other events not here relevant) has caused a Material Adverse Effect on Sallie Mae (as defined in the merger agreement). If this is correct, then the buyers are entitled to walk away from the deal. If not, they may still walk away (the contract expressly excludes a specific performance remedy), but only by paying Sallie Mae a reverse breakup fee of $900 million. The parties are currently litigating the matter before Vice Chancellor Leo Strine of the Delaware Court of Chancery.

At the time the merger agreement was negotiated, other legislation, also adverse to Sallie Mae, was pending before Congress, and the agreement’s definition of “Material Adverse Effect” expressly excludes the effects of certain proposed legislation described in the company’s 10-K. The parties generally agree that the CCRAA is even more adverse to Sallie Mae than such proposed legislation. A central issue in dispute in the litigation is this: assuming that the CCRAA is more adverse to Sallie Mae than the proposed legislation described in the 10-K, in determining whether an MAE has occurred, is it only the incremental effect of the actual legislation over the proposed legislation that counts, or is it the entire effect of the actual legislation? Naturally, Sallie Mae says the former, the buyers the latter.

According to the buyers, they agreed in the contract to accept a certain level of risk, including the risk of the legislation described in the 10-K, but not more. Hence, if the actual legislation is more adverse than the legislation described in the 10-K, the question is whether such legislation in its entirety causes an MAE. Sallie Mae points out that, if this view is correct, then if the actual legislation is just one dollar more adverse than that contemplated at the time of the agreement, the existence of an MAE could turn on this one additional dollar of adverse impact, which seems ridiculous. According to Sallie Mae, the incremental impact in itself should have to be material, i.e., cause a MAE, if the buyers are to get out of the deal without paying the reverse breakup free.

The buyers are right here. Imagine that, at the time of the agreement, the buyers valued the company at V, but were willing to buy the company at the purchase price even if it were worth as little at rV, where r is some percentage reflecting a diminution in value of the company to the buyers from whatever causes (e.g., r = .80, meaning that the buyers would have been willing to buy the company even if it were worth only 80% of its value at the time of the agreement). The intuition here is that rV is lowest value of the company at which it still has not suffered an MAE; if the company is worth less than rV, then it has suffered an MAE.

Since the buyers were willing to bear the risk of the proposed legislation, we have to conclude that they thought that the proposed legislation, if enacted, would reduce the value of the company to, say, xV for some discount factor x such that xV > rV (e.g., perhaps x = .85, meaning that the proposed legislation would reduce the value of the company to .85V, which is still above the .80V threshold of an MAE).

Now, what happens when Congress passes legislation even more adverse to Sallie Mae than that which was being considered at the time of the agreement? Well, the value of the company is reduced to some value yV such that y < x. Has the legislation caused an MAE? It depends. All we know is that y < x. It could be that r < y < x, in which case rV < yV, and there has been no MAE. On the other hand, it could also be that y < r < x, in which case yV < rV, and there has been an MAE. Hence, the buyers’ view makes perfect sense: if the actual legislation is more adverse to Sallie Mae than the proposed legislation (if y < x), then the issue should be whether y < r, i.e., whether the legislation taken as a whole reduces the value of the company to the point that it has suffered an MAE.

Sallie Mae’s view, on the other hand, is incoherent. According to Sallie Mae, there is an MAE only if the incremental effect of the legislation (which is (x – y)V), taken alone, causes an MAE, i.e., only if V – (x – y)V < rV or, what amounts to the same thing, 1 – (x – y) < r. But this means that the buyers can be obligated to buy the company or pay the breakup fee in some cases when the company has suffered an MAE. For example, let r = .80, so that a 20% diminution in the value of the company causes an MAE, and let x = .85, so that the proposed legislation would have caused only a 15% diminution. According to Sallie Mae, if the actual legislation causes a diminution in the value of the company of more than 15% but less than 15% + 20% = 35%, e.g., say y = .70 for a 30% diminution, then the buyers have to buy the company or pay the breakup fee, even though the diminution exceeds 20% and causes an MAE. This, as I say, has to be wrong.

What about Sallie Mae’s argument that the buyers’ interpretation implies that legislation that is even one-dollar more adverse to the company than the proposed legislation could cause an MAE and so the existence of an MAE could turn on one dollar, which seems absurd? To this, I think, there are two responses.

First, all Sallie Mae is doing here is noting what philosophers call the Sorites Paradox—the idea that if you have a heap of sand and remove one grain, you still have a heap, and so by iterating this process you can prove that no matter how few grains of sand you have, even zero, you still have a heap. Analogously, if the proposed legislation doesn’t cause an MAE, then neither does legislation just one dollar more adverse. Iterating this argument, we can show that no legislation, no matter how adverse, causes an MAE. As to why sorites arguments fail, philosophers disagree (there’s a large literature on the philosophy of vagueness), but that such arguments are fallacious is clear enough.

Second and more important, Sallie Mae is right that, since the buyers were willing to accept the risk of the proposed legislation described in the 10-K, if the actual legislation is to cause an MAE, the actual legislation must be materially more adverse to Sallie Mae than the proposed legislation. But saying that the incremental impact of the actual legislation over the proposed legislation is material is not the same as saying that the incremental impact, taken alone, would have an MAE. To pursue the point from above, saying the incremental impact is material amounts to saying that (x – y) ≠ 0, or, more accurately, that (x – y) is not de minimis. But (x – y) can be significantly greater than zero even though (x – y) < r, i.e., even though the incremental impact, taken alone, does not cause an MAE. For example, if x = .85 and y = .70, the difference (x – y) = .15, or 15% of the value of the company, which is surely material, even if only a 20% diminution in value would cause an MAE on the company because r = .80. Put yet another way, the actual legislation could be materially different from the proposed legislation not because the incremental impact taken alone causes an MAE but because the actual legislation, in its entirety, causes an MAE whereas the proposed legislation did not.

Steven Davidoff responded to my blog here last week regarding Merrill Lynch & Co., Inc. v. Allegheny Energy, Inc. and made the excellent point that just how bad for Merrill the representation I quoted really was depends in part on the limitations on indemnification that were included in the purchase agreement. For example, if the representations survived for only a short period and were subject to a large deductible and low cap, Merrill’s liability under the representation would be sharply limited. Unfortunately, the second circuit opinion says nothing about the indemnification provisions, and so I don’t know how much Merrill’s lawyers were able to take back with the left hand what they had given away with the right.

Steven also mentioned a representation that as counsel on the buy-side he generally tried to get and that is in some ways similar to the one I so criticized Merrill’s lawyers for giving on the sell-side. The representation Steven mentioned provides as follows:Â

There does not now exist any event, condition, or other matter, or any series of events, conditions, or other matters, individually or in the aggregate, adversely affecting Seller’s assets, business, prospects, financial condition, or results of its operations, that has not been specifically disclosed to Buyer in writing by Seller on or prior to the date of this Agreement.Â

As seller’s counsel, I would be willing to give this representation only if a few key changes were made. First and most important, it would have to be qualified by some materiality threshold. Every day any number of things “adversely affect” the company—far too many to be listed on a disclosure schedule—and so this representation would be false when made unless qualified as to materiality. In my view, the correct qualification here would be qualification to a “MAC”—a “material adverse change” on the company (defined with all the usual carve-outs, etc.). If the buyer’s counsel argued for a lower level of materiality—say anything “materially adversely affecting” the company—I would push back very hard. I think the seller ought not be in the position of having to determine for the buyer what is “material” to the value of the business from the buyer’s point of view.

This brings me to the second change I would think necessary: the MAC in the qualification should be a MAC on the company as a whole (including its “business” and “financial condition”), not on the company’s “assets” and certainly not on its “prospects.” “Assets” is wrong because something can adversely affect the company’s assets and not make the company itself worse off (e.g., a fully-insured asset is destroyed in an act of God). “Prospects” is wrong because no rational seller guarantees the future results of the business.Â

Third, the representation should be qualified as to time, meaning that it should pertain only to the period beginning on the date of the most recent audited financial statements of the company. Anything prior to the date should be covered in those financial statements.

With these three changes, the representation would sayÂ

Since the date of the Seller Financial Statements, there has not occurred any Material Adverse Effect on the Seller, its business or financial condition, except as has been specifically disclosed to Buyer in writing.Â

Thus modified, the representation is, I believe, customary. A representation substantially identical to it appears, for example, in the KKR-First Data merger agreement (the merger proxy for the deal, with the agreement appended, is available here):Â

Section 4.7 Absence of Material Adverse Change. Since December 31, 2006, … there has not been any Material Adverse Change or any Effect that would, individually or in the aggregate, reasonably be expected to have a Material Adverse Effect on the Company.

Representations like this, I think, are significantly different from the representation Merrill gave in the Allegheny deal. For one thing, the representation above does not amount to representing that all statements in the diligence materials were true. To prove a breach of the representation above, the buyer would have to show (a) there was a fact that amounts to a MAC on the company, and (b) the seller didn’t disclose that fact in diligence. It would be a question of a MAC-level omission from diligence, not a question of a false statement in the diligence materials. From a litigation point of view, the issue would probably come down to the seller combing the diligence materials and arguing that some obscure statement in the materials really did disclose the fact in question. From an economic point of view, the representation above creates an incentive for the seller to disclose more information rather than less in diligence, and that sounds efficient to me.

Although the representation in Merrill-Allegheny makes Merrill liable for omissions from diligence and so creates a similar incentive, it also creates an exactly opposite incentive as well. For, under the Merrill-Allegheny representation, the buyer can comb the diligence materials for an arguably-false statement and then argue that the falsity of the statement breaches the representation. Hence, the representation creates an incentive for the seller to disclose less information rather than more. To this extent, it’s very likely inefficient.

There’s another difference too. Nothing in the representation above requires the seller to make normative judgments about whether the information delivered is “appropriate” to value the company. The only judgment that the seller has to make is whether a fact amounts to a MAC on the company. It would seem possible that the seller could disclose all negative information about the company while still not disclosing some information needed to “appropriately” value the company. Imagine, for example, that inventory turnover rates are very important in the business being sold, and so valuation methodologies for this kind of business usually make use of such information. Under the Merrill Lynch representation, I think the seller would have to disclose inventory turnover information, even if there was nothing “bad” in the information (e.g., even if the turnover rates were average for the industry) and even if the buyer never asked for it. Under the representation above, as long as the turnover rates were not unusually bad, I don’t think the failure to disclose information about them would amount to a breach.

So although I didn’t want to harsh all over Merrill’s lawyers, I still think they gave a representation that was unreasonably unfavorable for their client. Steven and I both recall seeing counsel, even from elite firms, make serious mistakes similar to this—missing double materiality qualifications and so on. Steven suggested that the explanation may lie in the negotiation process being private (hence most mistakes don’t lead to public embarrassment), in an unthinking adherence to the firm’s form agreements (hence once a mistake is incorporated into the form, it gets repeated), or in over-reliance on network effects (other lawyers know what they are doing and so the associates will learn from them).

I’m inclined to think the problem is simply one of agency costs. Clients don’t read business combination agreements, and they certainly don’t read the representations—much less do they understand what can reasonably be given in representations and what cannot. Hence, when lawyers make a serious mistake, it’s often the lawyers themselves who are advising the client as to how serious the mistake was and how it ought be handled. The clients are ill-equipped to make an independent judgment. The lawyers can thus cover up the mistake at the client’s expense.

Unless, of course, some tiresome law professors want to blog about them.Â

Last week the Second Circuit Court of Appeals decided Merrill Lynch & Co., Inc. v. Allegheny Energy, Inc. (available here), which involved fraudulent inducement and breach of warranty claims in connection with a business combination agreement. The case is a straightforward application of familiar principles of blackletter law. I mention it because of certain highly unusual provisions in the purchase agreement between the parties.

First, some background. Back in 2000, Allegheny was considering purchasing GEM, a subsidiary of Merrill involved in trading energy commodities. In connection with Allegheny’s due diligence review, Merrill provided Allegheny with a great deal of information about GEM. The opinion doesn’t say what happened in this case, but usually such information is delivered under a confidentiality agreement that provides not only that the potential buyer will not disclose the information but also that the potential seller makes no representations or warranties as to the completeness or accuracy of the information provided. Such a disclaimer makes perfect sense. The materials being produced are the business records of the seller, and they were generated in the ordinary course of running the company, not to be relied on by third parties with interests adverse to those of the seller. The efficient error rate for business records is presumably much higher than the efficient error rate for contractual representations, breaches of which will almost inevitably subject the person making them to liability. Hence, if the seller represented that materials produced in due diligence were, say, “true and complete in all material respects,” this representation would very likely be false. The usual thing, therefore, is for the seller to make no representations or warranties on such materials.

Merrill and Allegheny eventually entered into a definitive purchase agreement for the sale of GEM, and this agreement, like virtually all business combination agreements, included elaborate representations and warranties by the seller about the condition of the business. This too makes sense. Limiting representations to ones expressly made in the agreement reduces uncertainty and forces the parties to bargain over exactly which representations will be made.

After the deal closed, Allegheny discovered that some of the key financial information Merrill provided in due diligence was false. The facts get very complicated at this point, in part because the Merrill employee running the GEM business prior to the transaction had embezzled millions of dollars from Merrill (he’s now in jail) and in part because of disputes about accounting methodologies used in preparing the information. The parties disagree about exactly which statements in the information Merrill produced in due diligence were false, why they were false, and what various of Merrill employees knew or should have known about their falsity at the time the agreement was signed.

A disappointed buyer like Allegheny can make either or both of two claims. It can argue fraudulent inducement, and so have to prove that the seller knowingly (or recklessly) included false information in the due diligence materials and that the buyer justifiably relied on such information to its detriment. Or, the buyer can argue breach of warranty, and so have to prove that the seller made a false statement in the representations in the agreement and the buyer was harmed thereby. In the first case, the universe of relevant statements is much wider, but the buyer has to prove not only that one such statement was false but also that the seller made the false statement knowing it was false (or in reckless disregard of its truth). In the second, the universe of relevant statements is confined to those made in the agreement, but all the buyer has to prove is that one such statement was false; the seller’s knowledge of its falsity is irrelevant.

The agreement between Merrill and Allegheny, however, contained some highly unusual provisions. In particular, Merrill warranted that the information provided to Allegheny “in the aggregate, in [Merrill’s] reasonable judgment exercised in good faith, is appropriate for [Allegheny] to evaluate [GEM’s] trading positions and trading operations.” This should take the breadth away from any practicing M&A lawyer. In effect, Merrill is representing not only that it believed the information provided in due diligence was true but also that it reasonably believed such information was true. Hence, any false statement in the diligence materials becomes potentially actionable: the seller will argue that Merrill should reasonably have known such statement was false.

Even worse, Merrill is representing that the information it provided was “appropriate” for Allegheny’s evaluating the business. At the very least, this means that Merrill is warranting that it reasonably believed that it delivered all the information that Allegheny needed to value the business. Hence, omissions from due diligence will become actionable. If Merrill had any information it did not produce to Allegheny in due diligence, Allegheny will now argue that such information was reasonably necessary for it to value the business and so its non-delivery to Allegheny was a breach.

Moreover, representing about whether the materials produced were “appropriate” for valuing the business also requires Merrill to make judgments about what information a buyer reasonably needs in a valuation study and so involve making judgments about what were appropriate valuation methodologies for the business in question. That, quite simply, is not the seller’s job. How the buyer values the business is something only the buyer can really know. After deciding how to value the business, the buyer should ask for what information it needs, and, even better, bargain for representations in the agreement concerning that information.

Not surprisingly, the Second Circuit reversed the district court’s summary judgment in favor of Merrill and remanded for further proceedings. The lesson here—besides the obvious one that Merrill cut a very bad deal and should hire better lawyers—is that representations should be about matters of fact the seller can be fairly certain it knows to be true, not normative judgments about what’s “appropriate,” especially from the point of view of an adverse party. Voluntary exchanges are efficient because each of the parties prefers the result of the exchange to the status quo ante. A representation by one party that the other party has evaluated the change correctly—i.e., “really” prefers it—is beyond what that party could know. Giving such a representation almost amounts to insuring the transaction for the other party. Merrill, I suspect, did not realize that it was doing that, but the result follows from the contract language, and the Second Circuit quite properly is holding Merrill to the bargain it struck.