Economists are often in the “it works in practice, but does it work in theory?” mode, but here we see a case where some are grappling with why some of their prized notions pre-crisis came a cropper.
A clever post at VoxEU discusses why financial innovation isn’t what it is cracked up to be, and typically leads to what economists call “rent seeking” and mere mortals call “ripoffs”. (As an aside, the Orwellianism of the branding of financial services tricks and traps as “innovation” is annoying. Readers are invited to come up with better nomenclauture and encourage its use).
Finding that “innovation” is mainly, if not solely, beneficial to the new product pushers seems so obvious in the light of the developments of the last twenty years of the financial services industry that it scarcely seems worthy of mention. However, the post by Bruno Biais, Jean-Charles Rochet, and Paul Woolley offers another insight:
We describe the evolution of a financial innovation and show how rents rise progressively to the point where the agents end up capturing the bulk of the return from the innovation. The key assumption of the model is the presence of information asymmetry; the agent has more information than the principal and the agent’s interest and objectives are not necessarily aligned with those of the principal.
Despite being based on a single innovation, our analysis can be used as a metaphor for the financial sector as a whole. The model also shows how innovations and rents carry the seeds of their own destruction to the point where principals are no longer receiving an adequate return and refuse to support the innovation, which then collapses. Perhaps in line the global financial crisis, the model suggests that high and rising rents of agents offer a lead indicator of crisis.
Yves here. This is a very powerful finding. If there is a way to further substantiate this observation, it proves what many believe: that an overly large (in terms of profit) financial sector is unstable. Its real economy cost become too high and users withdraw from the particularly abusive products. This would give arguments to rein in the financial services more heft.
The authors flesh out their ideas:
First consider the frictionless benchmark case in which principals and agents have access to the same information. The principals are a set of rational, competitive investors and the agents are a set of similarly imbued fund managers. A financial innovation is introduced but there is uncertainty about its viability.
As time goes by, investors and managers learn about this by observing the profits that come from adopting the new technique. If it generates a stream of high profits, this raises confidence that the innovation is robust. This leads to an increase in the scale of its adoption and therefore the size of the total compensation going to managers. Because of the symmetry of information, these gains are competitively determined at normal levels and the innovation flourishes.
Alternatively, profits may deteriorate, market participants come to learn of its fragility and the innovation withers on the vine. In both cases, while learning generates dynamics, with symmetric information there is no crisis….
In reality, innovative sectors are plagued by information asymmetry. It is hard for the outsider to understand everything the insiders are doing and difficult to monitor their actions.
We explore the implications of this lack of transparency and oversight using optimal contracting theory. Our model assumes that managers have a choice:
* They can exert effort to reduce the probability that the project fails even though such effort is costly.
* Alternatively they can cut corners and “shirk” – the term used by economists and familiar to every schoolboy meaning to avoid work. When agents shirk they fail to carefully evaluate and control the risks associated with the project.
Yves here. The authors mention CDOs as an area where this took place. While there were a lot of investors that were “shirkers”, the uglier reality is that CDOs were so complex as to be effectively unanalyzable in a granular fashion. A asset backed securities CDO contains 100 to 250 instruments. In 2006-2007, a “mezz” CDO would be 80% subprime bond tranches. Each subprime bond contains roughly 5000 mortgages. So the CDO contains exposures to 400,000 to one million mortgages. Oh, and each bond has a different waterfall, so you can’t throw them all together and analyze them tout ensemble. Oh, and another 10% of the CDO is tranches from OTHER CDOs, which means that each of those bits contains exposures to another 400,000 to 1,000,000 mortgages.
Now of course, people did miss the elephant in the room: that these pools were effectively a very junky subprime index. But how can you figure out how junky with so many moving parts? This was an inherently abusive product. Even if a buyer had the luxury of time and money, how can you make anything more than very approximate judgments about something this complicated?
Back to the post:
Our model also assumes that managers have limited liability. The inability to punish gives rise to the moral hazard that characterises finance at every level from individual traders to the banks that employ them (our simple moral hazard model is in line with that of Holmstrom and Tirole 1997).
This combination of opacity and moral hazard is the core of the agency problem. Investors have to pay highly to provide managers sufficient incentive to exert effort, and the greater the moral hazard, the larger are likely to be the rents. Our model shows the probability of shirking is higher when the innovation is strong than when it is weak. After a period of consistently high profits, managers become increasingly confident that the innovation is robust. They are tempted to shirk and it becomes correspondingly harder to induce them to exert continuing effort. As the need for incentives grow, the point is reached where agents are capturing most of the gains from the innovation.
Investors then become frustrated at the rents being earned by the agents and at their own poor return and eventually give up on incentives. The dynamics are such that when confidence in the innovation reaches a critical threshold, there is a shift from equilibrium effort to equilibrium shirking. Innovation collapses as managers cease to undertake the necessary risk assessment to maintain the viability of the innovation. In the end, an otherwise robust innovation is brought down by the weight of rents being captured.
The handling of portfolios of CDO’s in the run-up to the recent crisis illustrates this well. Fund managers had the option of diligently scrutinising the quality of the underlying paper or they could shirk by relying on a rating agency assessment and pass the unopened parcel on to the investor. While securitisation is a potentially valuable innovation, it also requires costly effort to implement properly….
Referring back to our model, this suggests that rent extraction was occurring at all operating levels within the institutions.
Our model’s second prediction is that innovations under asymmetric information are vulnerable to collapse. The current crisis seems to validate this prediction since structured credit, CDO’s and CDS’s were the immediate cause of the global financial crisis.
The article also refers to some empirical work by Thomas Philippon and Ariell Reshefalong similar lines:
Their study observes a burst of financial innovation in the first half of this decade and rapid growth in the size of the finance sector, accompanied by an increase in the pay of managers. They estimate that rents accounted for 30 to 50% of the wage differential between the finance sector and the rest of the economy. Philippon and Reshef point out that the last time this happened on a similar scale was in the late 1920’s bubble – also with calamitous consequences. It is significant that a high proportion of the net revenues of banks and other finance firms went to the staff rather than shareholders.
While correlation is not necessarily causation, the pattern is awfully persuasive.