Goldman Sachs: Orders of Worth Colliding
April 28, 2010
Still with the on-going Goldman Sachs story: yesterday, during one of the hearings of the American Senate Governmental Affairs subcommittee we had one of these rare chances where worldviews collide ‘on air’. In yesterday’s hearing, Senator Carl Levin was questioning former Goldman Sachs Mortgages Department head Daniel Sparks about matters related to selling of structured mortgage-based financial products known as Timberwolf, during 2007. The full transcript is not available (you can see the video here), but a few lines can give us a gist of the dialogue that took place. When Levin asks Sparks why Goldman Sachs hid from the customers their opinion of the value of Timberwolf (a product that an internal GS memo described as a ‘shitty deal’), Sparks answers that ‘there are prices in the market that people want to invest in things’. On another occasion exchange, when asked what volume of the Timberwolf contract was sold, Sparks answered: ‘I don’t know, but the price would have reflected levels that they [buyers] would have wanted to invest at that time’.
This reveals the incompatibility in its naked form. While Levin focused on the discrepancy between the opinions among Goldman Sachs’ employees about the value of the product and between the prices paid for these financial contracts, Sparks placed ‘the market’ as the final arbiter about matters of value. That is, according to this order of worth it does not matter what one thinks or knows about the value of assets, it only matters what price is agreed on in the market. Both Levin and Sparks agree that not all information was available to all market actors. However, while this is a matter for moral concern according to Levin’s order of worth, it is merely a temporary inefficiency according to Sparks’ view.
Moreover, the fact that this dialogue took place in a highly-visible political arena, a televised Congressional hearing, entrenches the ‘ideal type’ roles that Levin and Sparks play. Sparks, no doubt at the advice of his lawyers, played the role of the reflexive Homo economicus, claiming, in effect, that markets are the only device of distributional justice to which he should refer. Levin, in contrast, played the role of the tribune of the people, calling for inter-personal norms and practices of decency. These two ideal type worldviews, as Boltanski and Thevenot show, cannot be reconciled. What we call ‘the economy’, then, is oftentimes the chronology of the struggle between these orders of worth
socializing finance 
April 28, 2010 at 5:56 pm
[…] Read full story […]
April 28, 2010 at 8:51 pm
You’re right in the Timberwolf context, but I’m not sure that this should be taken too far. I think, as you say, Goldman has probably been advised to play up the culture clash angle, but I think that’s smoke.
But the more Goldman gets to write this up as a culture clash, the more get away from the underlying issue: that – in at least the case of Paulson’s involvement in ABACUS – they deliberately concealed facts that they had should have disclosed. I’m not sure those sort of practices are quite as foreign to the professed norms of market professionals as all that.
Steve Randy Waldman explains this much better than me here:
http://www.interfluidity.com/v2/822.html
April 29, 2010 at 10:48 am
Hello Yuval, like Pete, I welcome your post because it addresses a white hot issue from the standpoint of familiar theories.
But I disagree with your characterization. Presenting Goldman and the SEC/ Senate as a controversy between orders of worth is a way of saying “both are right, in their own way.” That’s precisely what Goldman would have us believe.
To see the complexity of the case, one needs to understand the counterperformative process that was going on. A metadevice, the rating system, was created. It was working fine for corporate bonds and CDOs. It worked fine for mortgage bonds for a while.
But because it was so prevailing, some actors learnt how to game the system. Assemble CDOs that look good from a rating perspective, but that are worthless. The rating was more or less the same all along, it worked fine at the beginning, it stopped working at some point in mid 2000s. So one cannot blame the credit agencies only. One cannot just resort to the usual suspect — agency theory — to explain this. The problem was with the creators of these tainted CDOs.
For more on how this process takes place, see the following episode of this American Life, which my coauthor David Stark brought to my attention:
http://www.thisamericanlife.org/radio-archives/episode/405/inside-job
April 29, 2010 at 11:19 am
Daniel, as much as I would love this to be a performativity story, it is not. I’m not talking about setting up of the CDOs and the way the rating agencies operated and were gamed. I am talking about the latest showdown. I think that what we see in the SEC’s case and the Congressional hearings genuine expressions of worldviews. I am not sure that the Goldman people are cynical here and just want us to think one thing while believing something else themselves. After all, what would Goldman gain from portraying themselves as ‘defenders of the market’ in this public climate?
Pete, thanks for the link. I think that the discussion there is really fascinating (well, definitely for market geeks like me). I agree with you that Steve Randy Waldman explains things very well, but that he got it wrong. It is perfectly reasonable for a market maker not to disclose full details about the identity of the ‘other side’ to the trade or about the exact nature of the assets transacted, even if it comes at the expense of one of the parties to the interaction. In fact, market makers do it all the time. Of course, market makers may pay for such behaviour because markets also have norms, but, the sanctions for such behaviour, as I would guess a practitioner like Sparks sees it, would be performed ‘within the market’ and not in the political arena. That is, the party feeling hard done by would refrain from trading with that market maker, or Sparks suggests, would trade at less favourable prices.
April 30, 2010 at 4:42 am
I have some reservations about the usefulness of the concept of performativity. Its tempting to use it as shorthand to describe how calculations affect markets, but maybe the situation is more nuanced. Isn’t it just that social structures (even industries) form around some calculations that are deployed in the market, also spurring new forms of rationality that reference the calculation? The concept of performativity can occlude fine grained investigation of these social structures and forms of rationality.
April 30, 2010 at 7:07 am
Chris, this sounds interesting, but I am not sure what you mean by calculations that bring about new forms of rationality. Care to elaborate?
May 6, 2010 at 3:21 pm
I am very meyukai your blog because you are a good blog
May 8, 2010 at 4:17 pm
This is my first time posting on this blog, so let me first say that I always enjoy the thoughtful commentary posted here.
My thoughts are similar to Yuval’s. I think the key underlying controversy in these hearings is: “Where is the line drawn between a financial intermediary and a market maker?”
To my knowledge the traditional view in economics is that market makers facilitate the trade of exchange-traded assets, while intermediaries structure over-the-counter deals. This is at least how Merton (1995) frames the distinction. In this view, it’s the *type* of financial product that marks the distinction.
But Goldman (implicitly) offers an alternative definition: one based on the notion of calculative agency. Goldman claims it was acting as a market maker because its clients were large banks who possessed the calculative tools and skills necessary to value the assets themselves. So although they were structuring a customized OTC deal, they had no duty of care to their customers because revealing Paulson’s beliefs would have been redundant information (which could already be inferred through the prices of the CDSs in the synthetic CDO portfolio).
I personally don’t agree with Goldman that calculative agency was sufficiently distributed, but their framing of the market-maker / intermediary distinction is interesting.
From this perspective, “financial intermediaries” exist when calculative agency is not sufficiently distributed throughout the market for calculation to be decentralized among agents, whereas “market makers” can only exist where that agency is distributed. The interesting question is: at what point in the decentralization of calculative agency does the role (and its implied ethical responsibilities) shift?
May 8, 2010 at 5:14 pm
Taylor — that’s a great point. You’re bringing the literature to bear on this very controversial distinction between market maker and intermediary. My reading of your point: if you’re the one grinding the meat, you should not be able to bet against the hamburger. Or grind according to the instructions of those who bet against it.
However, I disagree with your point on decentralized agency. The problem with these CDOs was, according to MacKenzie, that agency was too decentralized. Everyone was relying on the rating agencies. And that allowed people like Paulson enlist Goldman to engage in what seems to be gaming the system by producing tainted hamburger that looked great from the rating standpoint. Counter performativity.
The issue then is, how do you prevent these counter performativity dynamics. I would argue that the guideline should be precisely the reverse that you suggest: be extra careful when the system is very distributed.
May 8, 2010 at 9:09 pm
I think we’re saying the same thing, but using the word “decentralized” in different ways. When you say “decentralized” you mean with respect to the agents making decisions; i.e. agents “out sourced” their valuation of ABS CDOs to AIG and the ratings agencies. What I meant is that you don’t want your calculative capabilities clustered in a single organization or two like AIG and Moody’s; you want it decentralized throughout the market, which I think is the same thing you said. In MacKenzie’s language, you want a rich calculative culture to blossom that is shared by many.
How one prevents counterperformativity is, of course, the million dollar question, and it’s difficult to speak in generalities. But I have observed this: central to many cases of counterperformativity is a combination of mutual ignorance and a break-down in reflexive “higher-order knowledge”, i.e. what market participants believe about each others’ beliefs, what they believe others believe about their beliefs, and so on. I think one of the keys to answering your question are to spend time thinking about this epistemic domain and how it relates to the material devices that constitute the market.
In the case of ABS CDOs, a breakdown of higher-order knowledge occurred at multiple levels: mortgage originators realized that those in the CDO business weren’t “reading the loan tape” and so they realized they could get away with issuing mortgages that had good FICO scores but were actually quite risky. Likewise, Goldman Sachs and others realized that the assets were flawed but knew they could sell them to ignorant investors (who used ratings scores) and have them insured by AIG (who also relied on ratings). Then there’s the case you mentioned where people like Paulson intentionally tried to build CDOs that looked good on paper but were specifically designed to fail. These beliefs were jarred back into alignment by events like the collapse of Bear Stearns and the credit crisis.
The combined epistemic/material dynamics of this are really quite fascinating — you have a material object — a rating score and a method for creating it — whose presence ruptures a whole set of fragile higher-order beliefs in the market that ultimately creates this huge disaster. And at each level is a mode of reflexive thinking: “I know that the ratings are flawed, but other people don’t, and I can use that to my advantage.”
The interesting thing is that most of the time these higher-order beliefs are aligned productively to “perform” the market. So traders may believe that other traders are mispricing an asset, but the material artifacts of the market are set-up to encourage him to signal that belief by buying/selling. When a trader sees a news item flash across his Bloomberg Terminal, he knows that hundreds of other people are seeing that at the same time on their Bloomberg Terminals, and he knows they possess conceptual equipment that will allow them to act on the news as well. Thus, he must act fast otherwise the arbitrage opportunity will disappear. Here again you have a fascinating epistemic/material dynamic, but this time it’s creating a desirable outcome: the trader knows everyone has the same terminal and tools as him, he knows they know, he knows that they know he knows, etc.
A financial innovation — whether it be a new formula or a new product — always threatens to disrupt these chains of beliefs simply because its new and its benefits and risks are uncertain. Moreover, the producers of that innovation are usually more informed about its true risks and benefits than its users. One of MacKenzie’s classic papers is instructive here — in his paper on the social construction of nuclear missile testing, there’s a diagram referring to what he calls “the uncertainty trough”: http://is.gd/c0vFB. The problem for finance is that the users of a financial product will tend to be less skeptical about its risks than the producers; but knowing this, the knowledge producers can exploit their ignorance.
To help prevent counterperformative instances, then, I think you need to look to both improving the material devices and people’s beliefs about those devices. That means working to build a broad, rich calculative culture where calculative agency isn’t tied up in a small set of actors and organizations. It also means making sure that people’s higher-order beliefs about those tools are well-aligned, and if not, figuring out ways to aligning them. I’m speculating here, but one possible way beliefs could be re-aligned is through credible public announcements, by Central Bank chairpersons or someone else of similar authority.
May 8, 2010 at 10:10 pm
Taylor, Daniel: a fascinating discussion! And thanks for expanding the theoretical and conceptual underpinning for the empirical snippet I brought in the original post. Taylor, I agree with the conclusion about Goldman’s duty of care and I would like to add that it alludes to a more basic issue: the fundamental inability, it would seems, to design and perform a ‘disinterested market maker’ – one that would be detached from the transacting parties, the asset traded and the organisational structure of the market and yet be able to perform as an effective intermediary. Goldman, in the case, were very much an interested market maker, of course, but frequently, so are exchange market makers who parcel out a large order for a favourable client and push it to many smaller clients. That said, I am not sure about the use of ‘calculative agency’. The way you build the argument it seems that you can simply replace ‘calculative agency’ with ‘information’ and would get the same result, which I believe you don’t intend to, but it’s not quite clear there.
Then, about the counterperformativity point: the ‘higher-order’ beliefs, as you call them Taylor, are always mediated/represented by the technological devices that are commonly used, or, more generally, by the material conditions in which the social action is situated. Hence (and here I agree with Daniel, but want to sharpen his point), the more commonly accepted a market device is, the more dangerous it would be if turns out that the device is faulty. So, paradoxically, the more accurate and valid the predications of a market device turn to be, the more popular it would be and thus more dangerous. Following this insight, if we try to limit the risk here by regulation then we put the regulator in an impossible situation whereby it would have to ask some market actors to use a ‘second best’ model/methodology for the sake of overall systemic safety. Boris Holzer and I made this point in a paper we wrote a few years ago. Similarly, distributing information to all market actors, as you imply, may bring about good results, but, I would guess, be extremely difficult to impose. By the way, this was tried, in a way, in the AZX ‘sunshine trading’ mechanism with mixed results.
May 10, 2010 at 12:56 am
Yuval, I agree, this is a great discussion about some fascinating issues! Although I’m not sure our “inability to design a disinterested market maker” is really fundamental to finance. Instead, this may be fundamental to the current design of exchanges. Before stock prices were decimalized in 2001, the market makers could profit from the spread; since then, those activities have become much less profitable, which is why we’ve increasingly seen liquidity provision handled through liq. rebates and high-frequency trading. Thus, this inability is really the consequence of a material fact of the market’s design.
The problem is that even with liquidity rebates, ECNs, and high-frequency trading, there seems to be a perceived need for old-fashioned market makers in the form of “NYSE Brokers” whose neutrality is legally guaranteed. I’m pretty fascinated by an issue that Daniel has mentioned from time-to-time on here: the growing controversy over whether HFT liquidity is “real.” Regardless of how that controversy plays out, the fact that people still trade on the NYSE and they’ve opted for a hybrid electronic/broker-dealer design suggests that people believe that “liquidity rebates” are insufficient for their purpose.
Regarding “calculative agency,” in the context of the Goldman discussion I can see how it looked as though I was defining the term merely as information. In my view, calculative agency essentially refers to the power to access, process, and act on information. This capability is a function of skills, technological artifacts, pricing formulae, etc.
Goldman had asymmetric information relative to its clients about the design of ABACUS. My point is that if the calculative agency for CDOs were as well-developed and widely shared as that for some kind of boring, well-known asset (stocks?), this information wouldn’t have mattered as much since the higher risk of the ABACUS deal would have been rendered more visible through the price of its underlying CDSs.
In the extreme, you can imagine the “neoclassical” ideal where markets are informationally efficient (because calculative agency is well-developed and widely shared), so Goldman’s “private” information would be rendered plainly visible to its customers. Once you creep away from that, Goldman’s intentions become more and more opaque and disclosure matters more and more. The question is: at what level of opaqueness does Goldman’s omission become “material,” thus constituting fraud?
I strongly agree that higher order beliefs are mediated and shaped by market devices, but my point is that you shouldn’t reduce these beliefs to the material factors/devices that shape them. To see this, consider how mortgage lending changed as the ABS CDO market grew. At some point, lenders began to realize that the CDO managers buying their mortgages were only looking at FICO scores, LTV ratios, etc. and not “soft” information (like information collected during interviews) that was contained on the “mortgage tape.”
Lenders’ decision to relax lending standards was motivated precisely upon (a) their (higher order) belief that banks wouldn’t notice the change, as the technology blinded them to the changing risk profile, and (b) that banks wouldn’t realize that other banks weren’t reading the tape either. Likewise, Paulson and Goldman Sachs believed that their customers and AIG were misinformed and only watched ratings, so they built high-risk ABS CDOs that had “good” ratings with the belief that their customers wouldn’t notice. The material artifact — the ratings agencies’ models — is central here, but ultimately it’s “acting at a distance” by misaligning people’s beliefs.
So, if in ANT-style, we wanted to “trace the associations” between the material device (rating models) and the change in lenders’ behavior, we’d have discontinuities if we only looked at the material world. For it to be continuous, you need to trace out how the material artifacts are changing agents’ beliefs.
I think this becomes explicitly clear when you think about economic models of moral hazard, like the Akerloff used car market. In these models, common knowledge is taken as a given — so the buyer and the seller each know what information is available to the other; specifically, the buyer knows that he’s misinformed relative to the seller. In these cases, when unresolvable moral hazard exists the market just breaks down.
This distinction matters for discussions of counterperformativity because counterperformative devices or strategies are usually not intrinsically bad or destructive. Instead, they become destructive because people’s beliefs about their effects are misaligned. LTCM is, I think, the best example of this. There was nothing intrinsically defective with their investment strategy. As MacKenzie points out, they were extremely careful at risk management. The problem arose because people copied the strategy but weren’t aware that others were copying it as well. VaR models fit this story as well. VaR is not, primarily, destructive because the model is intrinsically flawed; instead, it tends to exacerbate volatility when banks use VaR and aren’t aware that other banks also use VaR.
May 10, 2010 at 1:18 am
Sorry… I didn’t really finish my thought in the 2nd to last paragraph. Here’s the rest, below the original paragraph.
I think this becomes explicitly clear when you think about economic models of moral hazard, like the Akerloff used car market. In these models, common knowledge is taken as a given — so the buyer and the seller each know what information is available to the other; specifically, the buyer knows that he’s misinformed relative to the seller. In these cases, when unresolvable moral hazard exists the market just breaks down.
The seller does not have an opportunity to sell the buyer a bad car; he fully anticipates that the buyer will anticipate that he will sell him a bad car. In the real world, however, you not only have asymmetric information (as above) but a lack of common knowledge about the existence of asymmetric information. In the real world, higher-order beliefs are misaligned: there’s asymmetric information but the buyer may not realize this. Knowing this, the used car salesman successfully sells him a lemon.
In financial markets, new market devices create asymmetric information (as FICO ratings did) but agents (banks) fail to anticipate their effects; other agents (lenders) anticipate this lack of anticipation and act in a way that breaks down the market device.
May 25, 2010 at 6:25 am
[…] at the great academic site Socializing Finance: A Blog On the Social Studies of Finance in “Goldman Sachs: Orders of Worth Colliding“: When Levin asks Sparks why Goldman Sachs hid from the customers their opinion of the value […]