Yves here. It’s a welcome surprise to see economists devise a model that delivers generally sensible results. Here, three economists looked at how financial innovation leads to an bloated financial sector as well as greatly increasing the risk of meltdown.
One quibble is that they characterize financial innovation as a benign process at the outset that is all too quickly abused, particularly when the authors assume information asymmetries (as in the investors don’t fully understand what they are buying). One of their examples, collateralized debt obligations for subprime securitizations, disproves the thesis.
The writers miss that an earlier version of this type of asset-backed CDO also existed in the 1990s, and as in its 2000 reincarnation, was critical to keeping subprime mortgage securitizations going. As we discussed long form in ECONNED, the ability to sell these deals was constrained by the difficulty of placing the least attractive tranche, the BBB slice. Those (along with more attractive financial sausage constituent parts) were rolled into CDOs, which were tranched just as the original mortgage backed securities has been.
Now astute students of finance know that pretty much any product defect can be solved by price. That means those deemed-to-be-drecky BBB tranches could have sold if they had had higher interest rates. Of course, that would have meant some combination of higher interest rates on the underlying mortgages (as in less product to sell, since fewer people can afford payments on more costly mortgages) and/or lower fees to the deal participants. So these CDOs, which proved to be a Ponzi scheme in both their 1990s and 2000s versions, were a “financial innovation” created to get around the need to price RMBS more realistically. And the result in both cases was too much subprime origination and a CDO crash.
By Bruno Biais and Jean-Charles Rochet, professors at the Toulouse School of Economics, and Paul Woolley, Senior Fellow at the London School of Economics. Originally published at VoxEU
The Global Crisis has intensified debates over the merits of financial innovation and the optimal size of the financial sector. This column presents a model in which the growth of finance is driven by the development of a financial innovation. The model can help explain the securitised mortgage debacle that triggered the latest crisis, the tech bubble in the late 1990s, and junk bonds in the 1980s. A striking implication of the model is that regulation should be toughest when finance seems most robust and when innovations are waxing strongly.
One of the curiosities of the modern economy is why the finance sector is so large. Economists have only recently sought to document and ponder this phenomenon. Empirically, Greenwood and Scharfstein (2013) find that, in the US, financial services, which accounted for 2.8% of GDP in 1950, made up 8.3% of GDP in 2006.
In Biais et al. (2014), we offer a formal analysis suggesting why finance has become so large, asking whether it is in society’s interest and, if not, what might be done about it. In our model, the growth of the financial sector is driven by the development of a financial innovation. The analysis is applicable to a wide range of innovations – from specific instruments such as collateralised debt obligations (CDOs) or credit default swaps (CDSs), to asset classes such as junk bonds or exchange-traded funds (ETFs) – and management styles.
Our model considers two types of market participants: investors who own the assets, and managers who operate the innovation. The analysis is conducted first under the assumption that investors can evaluate the ability and actions of managers – i.e. there is ‘symmetric information’. This provides a benchmark for the next step of recognising the reality that investors are uncertain of managers’ competence or diligence, and therefore information is asymmetric.
The innovation can be fragile and prone to shocks. Alternatively, it can be quite strong. Based on observed performance, market participants learn about the strength of the innovation. They rationally interpret good performance and the absence of shocks as a sign of robustness. Thus, when no shock occurs, confidence increases and the innovation grows, attracting funds and managers. As the innovation flourishes, managers’ earnings increase and exceed the norm of the other sectors, consistent with the empirical findings of Philippon and Resheff (2008).
Investments in the innovative sector are robust to shocks only if managers carefully analyse risks and valuations. Early on, when confidence in the innovation is still limited, managers understand that effort is essential for success. Only the most efficient managers, for whom the cost of effort is relatively low, enter the innovative sector, and all exert effort. After success, however, managers with inefficient risk-control and evaluation skills also find it worthwhile to enter the innovative sector. Because for these managers the cost of effort is large, they dispense with risk-prevention and prudent valuation.
In this context, if the industry is eventually hit by a negative shock, the sheer size of the industry, combined with the large fraction of managers who shirk, triggers large and widespread losses – a crisis occurs, and only investors whose managers maintained vigilance throughout are spared its worst effects.
In this version of the model the innovation has remained at a size that is socially optimal. Investors know what they are getting and are paying a competitive price for it. Now make the plausible assumption that opacity and complexity of the innovative sector prevent investors from perfectly monitoring the risk-control systems and efforts of managers. With this change, the predictions of the model take a dramatic turn. Under information asymmetry, investors are unable to tell managers with efficient risk-control systems (who exert risk-prevention effort) apart from those with inefficient systems (who dispense with risk-prevention.) Both types receive the same compensation. Hence, negligent managers earn excess profits (‘informational rents’). The influx of relatively inefficient managers attracted by rich rewards for little effort spurs faster growth than under symmetric information. This takes the scale of the innovation beyond its social optimum and also increases the vulnerability of the sector. Consequently, information asymmetry implies more severe crises.
While this theoretical model could also be applied to nonfinancial innovations, it is particularly appropriate for finance. Three of the most important features of financial innovations play a key role in the analysis:
• First, risk-control and management are key to the success of financial innovations, and it is precisely these activities which the managers of the model are in charge of.
• Second, the complexity and nonphysical nature of financial innovations make it difficult for outside investors to observe finance sector managers’ actions, which generates moral hazard, as in the model.
• Third, when financial innovations prove to be weak, this generates severe losses for a large cross-section of institutions, again as in the model.
Relating the Model to Events
The finance industry, especially in the last three decades, has been plagued by innovations that end in tears. The more egregious examples have been the securitised mortgage debacle that triggered the latest crisis, the tech bubble in the late 1990s, and junk bonds in the 1980s. Their dynamics are in line with the implications of our model.
The initial growth of the CDO innovation began in the mid-2000s, with investment banks offering securitised mortgages with a yield advantage over standard investments. Consistent with our model, initial success attracted managers and investors, but confidence led to negligence. Many managers ceased to examine for themselves the contents of the loan bundles, relying instead on agency ratings. The originators of the CDOs also felt confident enough to lower the credit quality of the component loans. Both developments constitute shirking in response to strong growth, in line with the implications of the moral hazard model. Also in line with the model, there were differences across managers. Only those who remained vigilant saved their investors from the worst losses.
Active fund management boils down to two basic strategies: fundamental investing based on the hard work of estimating future cash flows, and momentum investing. At the end of the 1990s there was a surge of innovation in the telecom, media, and technology industries, associated with an increase in financial investments in those sectors. It was, however, complex and costly to assess the future revenues to be generated by those investments. Again in line with the model, sustained outperformance raised confidence, and an increasing fraction of managers switched from the difficult task of fundamental valuation to the easier strategy of momentum-riding, which, in the language of the model, can be interpreted as shirking. When the crash came, investors whose managers exclusively rode momentum paid the price (Daniel and Moskowitz 2012 and Daniel et al. 2012 document the vulnerability of momentum strategies to shocks). In contrast, those who continued to bear the cost of careful asset valuation were spared the worst.
Our theoretical analysis also sheds light on the leveraged-buyout (LBO) and junk-bond wave of the 1980s. Empirical research at the time had shown sub-investment-grade bonds had a much lower default rate than implied by their yield spread. This spurred the issuance and distribution of junk bonds. Consistent with the model, initial success increased confidence, which led to reduced risk-prevention and valuation efforts. This made the innovation vulnerable and, when hit by a negative shock in 1989, it collapsed. Also in line with the model, the innovation was not expunged, and, after several years without crisis, has lived to fight another day.
Asset-owning principals should be more aware of the agency problems, write better contracts, exercise greater vigilance, and insist upon fuller disclosure by agents. They should lengthen the period of time for intermediaries’ performance review and reward. Shirking is more difficult to sustain in the longer term because individual failure or universal crisis is likely to intervene. The accumulated rewards promised to managers can be cancelled in case of losses, increasing the managers’ ‘skin in the game’ and aligning their interests with those of the principals (see, for example, Biais et al. 2007).
In practice, however, final asset owners are often small and dispersed. They typically delegate investing to trustees of pension funds, mutual funds etc. This creates a chain of agency problems, which the final principals are not well equipped to deal with. Policymakers should give guidance to trustees in the form of a code of best practice showing how to minimise the problems arising in delegation. This can be underlined by extending the interpretation of fiduciary duty to include compliance with the code.
If gentle nudges are insufficient, regulatory intervention may be necessary. Regulators could rule on the acceptability of products, bearing in mind that opacity worsens agency problems. They could enforce reporting requirements, for example, on gross returns, costs, portfolio choices, margins, etc. One of the most striking implications of the model is that regulation should be toughest when finance seems most robust and when innovations are waxing strongly. Beware of complacency and notions of ‘Great Moderation’, because that is when shirking is most prevalent.
Please see original post