The NYT reports that Lehman used accounting tricks to appear solvent while their bad mortgage holding ate away at their balance sheets. Specifically, they used repurchase agreements to temporarily move bad loans off their balance sheet just in time for quarterly reports. After the quarter was over and the numbers on which the report is based are set, Lehman would accept the bad loans back:
Repos, short for repurchase agreements, are a standard practice on Wall Street, representing short-term loans that provide sometimes crucial financing. In them, firms essentially lend assets to other firms in exchange for money for short periods of time, sometimes overnight.
But Lehman used aggressive accounting in its Repo 105 transactions: it appears to have structured transactions such that they sold securities at the end of the quarter, but planned to buy them back again days later. These assets were mostly illiquid real estate holdings, meaning that they were hard to sell in normal transactions.
The effect of the accounting was to artificially and temporarily lower the firm’s debt levels to hit certain targets, making the firm look healthier than it really was.
The revelation of Lehman’s Repo strategy now makes at least 3 firms in the last decade that have experienced sudden collapses following financial chicanery (Lehman, Enron, Tyco). Is this a coincidence? No. It is exactly what theory predicts should happen.
Theory predicts that managers can and will exploit information asymmetries to fool shareholders into allocating more capital to them than they deserve (See work by Oliver Williamson). But since there are limits to this information exploitation, there comes a point at which it is no longer sustainable. Mis-allocations are not sustainable in the long run, only the short run. By extension, if the short run price is biased high and the long run price is accurate, corrections will come in the form of steep drops (which are collectively bad for the market).
The underlying logic of this process is fairly simple. By manipulating their books, these firms allocate their “bad news” into the future. They take advantage of the fact that most news is somewhat indeterminate. Sometimes, news that looks bad is actually meaningless because other factors intervene. So when they get bad news, such as “our loans are performing poorly,” they invent concepts or categories that neutrtal-ize the news contingent on as yet unknown information.
For example, the loans may be under-performing because they are bad loans, or they may be under-performing because some random factor, such as unusual weather, has biased the defaults to come early. In other words, these are a regular set of loans, but because of the weather, the bad ones reported default early. If this is true, these bad ones won’t report default later, so in the end it will be a wash. So the firm can create a category called “March-June timing losses” and book the loans as performing regularly and book the under-performance as a “timing loss.” Then they can argue that the “timing losses” average zero over time because they are due to seasonal factors. The trick works because of the information asymmetry regarding what comprises “timing losses.” If shareholders knew that timing losses are simply regular loan losses, put in another category and assumed to average zero, the fraud would be obvious. But by giving the category a separate name, the firm suggests that “timing losses” have an independent history, an independent ontology as it were, and so statements about the performance of this category are subject to an independent debate. So it becomes plausible that “timing losses” average out over time even though it is implausible that bad loan losses average out over time.
Of course, this solves nothing, it only kicks the problem down the road. The implicit commitment of this strategy is that timing losses will average out (to zero) over time. But of course, if timing losses are simply allocations of bad loans, that is not going to happen. So after 12 months of timing loss accumulation, shareholders will again become concerned. But there is a way around this, too: keep changing the name — don’t call them timing losses every reporting period. If they’re timing losses in Q1, call them “reprobate contingency” in Q2 and zero out the timing losses. Claim that reprobate contingency is a general category that you only use rarely. Then, in Q3, put the same losses back in timing losses. Voila! Timing losses oscillate around zero, just like you said they would, and you churn them through a variety of meaningless, but rarely used, categories so they don’t accumulate in any one place.
The key to this logic is that solid inferences require the accumulation of compatible data. So the firm can obscure bad news by splintering and scattering bad news so that it is difficult for outsiders to re-assemble into a compatible set of information. It is because of this logic that this trick works in the short term, and it is also because of this logic that this trick is not sustainable in the long term.
In the short term, any given piece of bad news can be splintered and scattered. But in the long term, the splintered and scattered bad news accumulates. It accumulates in the places where it is easiest to hide it. Much like sweeping dust under the rug eventually leads to a rug with a weird, unsettling and obvious hump. The news, like the dust, has not gone away, it’s just accumulating in a place that is difficult to see. But every time news is put there, it becomes a bit easier to see. At some point, the task of splintering and scattering the news such that it cannot be observed with compatible information is greater than the firm can undertake — it involves calculations that are too complex or aligning incentives that are too disparate or costly. So the bad news accumulates and the shareholders notice. But when they notice it, they don’t just notice the most recent bad news, they notice all of it. It is the accumulation of all of it that makes any of it apparent.
I first observed such a process when I was a consultant to Campbell’s Soup in the mid/late nineties (mainly 1997). Campbell’s knew that what Wall St. cared the most about was top line revenue growth. Campbell’s was seen as a company with “flat growth” in a market that wasn’t expanding. To get their share price up, they needed to both promise and deliver on growing sales. So it was absolutely critical that Campbell’s met its quarterly sales volume targets. But, sadly, the demand for soup was not growing very much. So to make their quarter end sales numbers, Campbell’s used to call their customers (major supermarkets and big box stores) and ask them if they wouldn’t mind placing their orders for next month now, right before the end of this quarter. They called it the “quarter end load,” because what they were doing was “loading up” their customers.
Of course the problem with this is fairly obvious. If you “load” your customers at the end of this quarter, your sales will dip next quarter (in the first week or so), and so you’ll be playing catch up at the end of next quarter. But this is, of course, the short term/long term trade-off at the heart of the scheme. Today, your share price stays strong. Tomorrow, it is likely to be weaker. But tomorrow is a ways off (3 months in this case). Lots of things can happen. Maybe soup demand will finally take off! Maybe (the individual salesman thinks) “I won’t be in this job in 3 months.” Maybe (the CEO thinks) “I won’t be in this job in 3 months.”
And once they started thinking creatively about this, they didn’t need to be confined by a 3 month window. At some point, somebody had the brilliant idea of using trucks as carriers of phantom orders. Rather than try to convince Ralph’s to take extra soup before they wanted it, they placed orders for Ralph’s and put them on trucks. Then, when the next quarter started, they let Ralph’s cancel the orders. So the trucks just come back. You just pay a little extra in freight. Of course, freight costs go up, but that’s ok because freights costs are influenced by all kinds of factors, including energy prices and various complexities having to do with order sizes and what not. In other words, even if freight costs consistently go up because of quarter end loading, it is going to take a shareholder a really long time to figure this out.
But such charades cannot continue forever. Eventually, the phantom growth in soup sales is too great for all the trucks and all the stores that might discretely carry excess inventory. So what to do? I wasn’t at Campbell’s long enough to see how they handled it, but I know the general corporate strategy. Find an event that permits you to “credibly” take losses and then write-down all of your bullcrap in one fell swoop. If all of your dust is accumulating under the rug, you wait until the plumber has to come fix something, then you kick out all the dust and say “that plumber sure brought in a lot of dirt!” For example, any shock to the market that legitimately reduced soup demand could be used to absorb the accumulated losses. Rather than demand being down 5% they could claim it was down 15% and write-off all of the accumulated cancellations in one stroke.
This is just one example. And, as it should be apparent, if these kind of tricks can be played with a product as simple and concrete as soup, one can imagine how it could be done with “financial services” products.
0 Responses to “It’s ok, I’m sure Lehman Brothers is the exception”