Securitisation versus Banking
The ever elusive CMU dream
There is(/was) renewed interest in EU-land over deepening a “capital markets union”, aka “CMU”. It is among the initiatives being pursued by the Commission in order to help accelerate growth in the European Union.
The initiative encompasses many elements, both around equity (shares) and debt markets. One important pillar of the CMU aims to re-launch some version of an EU securitisation market. This segment was never really defined in a EU-wide basis. In the pre-crisis European financial landscape there were wide disparities in the degree of adoption, legal frameworks, preferred structures etc. among the different countries comprising the EU. In any case, since the financial crisis there has been a steady decline in securitisation volumes, which amongst others, hinders certain types of “exceptional” central bank measures (i.e., purchasing securitised assets).
In the post-Brexit world the task of furthering CMU and in particular the securitisation market is further complicated by the fact that a large fraction of that activity was performed by London based firms.
The idea of a re-invigorated securitisation market seemed to have support in various high places but it also potentially treads on the toes of established business interests, namely the commercial banking industry which is very important in Europe. The same industry is currently also facing the specter of “fintech”, namely new intermediaries that can provide some or all banking services via new technology platforms and business models (e.g peer-to-peer lending).
The role of securitisation
Securitisation is an alternative channel to arrange finance for wide range of credit assets. Even though from afar it’s just another tool of the broader “banking industry”, from close up it is clear that it is an alternative mechanism and business model. While “sourcing” credit assets may involve commercial banks, core players in this channel are investment banking units, rating agencies and institutional end investors (which may also include… banks). The growth of securitisation is invariably at the expense of competing banking business lines. Hence it is to be expected that various arguments for and against securitisation are being heard, as the various interested parties jockey for advantage.
Fintech business models may also involve securitisation. Indeed one of the first major scandals in the fintech world involved the securitisation of P2P loans. Since securitisation involves the pricing and transfer or risk, it goes to the very core of financial services provision and it is to expected that its future will be entangled with the evolution of the fintech sector.
This post is a side-by-side comparison of some aspects of securitisation with the corresponding features of bank lending, to try to elucidate some points we which think are relevant in the debate. We will avoid technicalities and try to be clear about the core points. Two important caveats:
- This is not an exhaustive comparison of these two areas (each being a huge financial practice in its own right and with countless variations of legal and structural form across jurisdictions. Especially after the financial crisis there have been excellent reviews of some of the conflicts of interest etc. that can plague securitisation. Instead we focus on select topics touching upon complexity, risk management and transparency.
- Securitisation is in principle a very flexible financial construct. Reflecting this fact, the public debate is about the desirable form of a new type of “good” securitisation. What exactly this new securitisation will look like is open. It can range from: cosmetic changes to past practices to more a aggressive reshaping that creates substantially different incentives, transparency and efficiency versus previous incarnations.
To be sure, this post is not about proposing a concrete organizational model, at best it would help point out opportune “thought angles” to explore further.
Describing securitisation as “simple” may ring odd to many readers. Isn’t securitisation the poster child of complex toxic products? Well, yes and no, but mostly no. There is huge potential for confusion as to what exactly “complex” means in this context, but clarifying this is essential for a rational discussion of the pros and cons of securitisation versus bank lending (and potentially other channels).
There is one key question to help reframe this debate properly: “complex” versus which standard of simplicity? Is bank-intermediated finance actually simpler than securitisation in any concrete sense? We will see that analyzing this question is very useful in understanding also what kind of securitisation might be most beneficial to re-introduce.
To be precise, we want to compare the complexity of a securitisation bonds (liabilities) versus analogous bank liabilities, from the perspective of an external investor. The first angle to explore is to simply Read The Full Manual or RTFM in short. The role of the manual in the securitisation case is served by the prospectus. A securitisation prospectus is a very long document indeed (can be 200-500 pages!). It is supposed to describe in great detail all material aspects of the underlying loan pool and the issued bonds. What to compare this monstrous complexity with?
The proper comparison would be against an equivalently detailed “banking prospectus” for a bank’s liabilities and the assets backing those liabilities. Alas, this is not possible as such detailed descriptions do not exist. What bank investors get is summary information in the form of annual reports and other forms of corporate communications. If one attempted to describe the inner workings of even the smallest bank detail equivalent to a securitisation prospectus one would likely need many thousands of pages (we’ll get back to this).
This is not to say that there were no securitisation structures with higher than necessary complexity. But it is very instructive to realize that design elements that increase securitisation complexity are features that aim to make securitisation behave more like banking finance. Examples of such features are i) maturity mismatch, ii) managed portfolios, and iii) embedded optionality. Lets examine them in turn:
This design feature is a massive complexity driver for banking structures. Banks can finance and refinance credit assets with a range of liabilities, from sight deposits to long-term loans and bonds which are in general not maturity matched to the assets. The flexibility and risks involved are huge (cf. Savings and Loans crisis in US) and have little to do with the credit risk profile of the underlying assets. Managing these risks necessitates significant effort on behalf of the bank around liquidity / ALM management. For this reason, ABCP programmes and other securitisation structures that move away from the matched funding paradigm increase securitisation complexity in the direction of banking and make securitisation structures more fragile.
A bank has large freedom as to how it manages its credit portfolio. It can accept new clients, change product mix, vary risk appetite across many different dimensions etc. This offers many advantages to the bank but also makes the assessment of bank debt risk more difficult. Similarly, managed CLO’s (Collateralized Loan Obligations) and other securitisation structures that move away from the static pool paradigm increase securitisation complexity in the direction of banking.
There is no such thing as a vanilla loan. Bank balance sheets are choke full of lending products with large number of clauses that confer options to both bank and client. Most prominent are the prepayment and drawdown options. Including such products in securitisation increases its complexity and brings it closer to banking complexity.
Another way to evidence the huge disparity in complexity between securitisation and banking: Consider the rating methodologies used for rating securitisation bonds versus bank liabilities. The former are relatively complex / detailed bottom-up methodologies using the underlying portfolio data, bond subordination etc. Bank rating models are essentially simple high level top-down scorecards that abandon any pretense that you can actually “model” in detail the inner workings of a bank. Instead they focus on such elusive factors as “quality of management”. Indeed in the end, a key assurance that a bank can meet the many dynamic risks it faces due to its complex design, is the ability of its management to respond to the unexpected.
Yet another way to help confirm the complexity point: To obtain assurance that a bank has met a certain minimal solvency threshold, regulators have tens of thousands of pages of regulation (with associated internal risk models, aggregation and capital standards). This complex and largely undisclosed internal model based methodology is the closest we have to a bottom-up view of a bank. The complexity and amount of information involved in constructing this framework is staggering.
Amusingly, one aspect where the bank structure might indeed be simpler than an old style securitisation is the number of debt tranches with different subordination one has to contend with. Existing “old style” securitisations have a ridiculously high number of different rated “tranches” spanning all letters of the alphabet. From a risk perspective these can only be treated as different if one has never heard of the term model risk. A simple securitisation would indeed only distinguish a very limited and robust set of liabilities along the lines of: equity, mezzanine and senior debt.
End result: Significant Advantage: Securitisation
By performance we mean a historical track record of “trouble-free” financing where the channel has managed to withstand a severe crisis with proper distribution of risks.
This is a tricky one, where in our view neither of the two camps can objectively gain advantage. The problem is – and this is the hallmark of any poor security prospectus – we can always pick up selectively a historical period that favors one’s argument. Financial history is not that replete with observation windows so that we can do controlled comparisons and derive solid conclusions.
But there are some conclusions around performance that can indeed be drawn: It is both logically obvious and has been shown during the crisis that neither “generic banking” nor “securitisation” perform well just because of their design. Despite mountains of regulation, banks were perfectly able to lower their origination standards and eventually bite the dust, at huge expense to taxpayers globally. The devil is in – occasionally very simple – quantitative details of the credit pipeline, such as the credit acceptance criteria, the amount of subordination etc.
Furthermore, past performance is no indicator of future performance. In principle, with adequate risk management and capital buffers versus the risks being financed either design could be made as safe as desired. And vice-versa.
End result: Advantage: Neither
Fees & Costs
The key metric to watch: The percentage of investor funds that reach the borrower.
This is an important issue when one decides on how to set up the plumbing of the financial system. When you design a railroad network you don’t make the routes longer or the connections more obscure so that passengers have to travel and pay more than they need to. Everything else being equal, the task is to match investors with borrowers at the lowest possible transaction costs.
There is no doubt that “old securitisation” was a gravy train for all sorts of agents, which was a strong contributing factor for its explosive pre-crisis growth. But this inefficiency is not intrinsic and could be removed by design (standardized structures, low barriers to entry, heavy use of modern IT etc.) On the other hand the bundled nature of banking services makes the identification and optimisation of banking costs more difficult. Unbundling and allocating revenue and costs within a bank is notoriously controversial.
We will not make here a detailed analysis of the various means for reducing transaction costs. We only want to point out an important factor that will play a huge role in reducing financial transaction costs overall in the future, which is the dramatic advancement of information technologies that are usable in financial context. Not only would the transmission and processing of data, documentation, models etc that are required for the securitisation channel be available at a dramatically reduced cost, one can also easily imagine here-to-fore unavailable services added at very low additional cost. In principle IT benefits would apply in banking as well, but legacy IT issues and intrinsic complexity will be harder to overcome in banking.
End result: Advantage: Securitisation
Transparency, in our context, means the ready availability of information about the financing instruments to a wide range of constituencies. Information consists for example:
- volumes of lending, prices (credit spreads)
- credit performance, down to the possible individual exposures
- underwriting standards (borrower characteristics)
- precise linkage of liability performance to the performance of underlying assets (aka the modellable prospectus)
In the banking channel the above information is largely private to the banks. In a capital markets channel one must normally disclose to investors much more of the above information.
With information disclosure, investors can (in principle) analyse securitisation credit risk positions much deeper than if they were to invest indirectly via bank liabilities.
End result: Advantage: Securitisation
In some fundamental sense this is the most important added-value of any “financing technology”. Risk assessment / analysis is the means by which the, much vaunted, “efficient allocation of capital” is supposed to be performed in a market economy. Alas risk analysis is both difficult and does not lead to immediate profits. Hence in practice risk management, more often than not, is just a cost item.
In “old securitisation” risk management was essentially outsourced to third parties underscoring its status as a cost item. In this respect, whether “new” securitisation will offer any added-value to the economy rests to a large degree on the question of who, how and with what incentives will perform risk analysis. As before, we are not going to go into all possible modes and solutions.
To complicate matters, the banking channel starts with an important advantage here: Loans and relationships are long-term and hence bankers will both have access to better information and (in principle) face the consequences of their actions if they get it wrong. But the incentive for bankers to do proper risk management is reduced by the infamous too-big-to-fail status and ineffectual contractual arrangements of bank managers. And as we stated, investors don’t invest directly in bank selected assets, but rather in once removed bank liabilities.
In the new securitisation scheme, whoever does the risk analysis of the underlying credits must have more than “skin in the game”. This should be their game. Their bonus/malus should be directly linked to their ability to select credit worthy projects over a long performance period. Assuming that a serious improvement in securitisation risk management is in place, the simplicity argument assures that this should also be more effective.
End result: Advantage Securitisation
If you formed a “tag cloud” of typical investor language, you would see primarily two high frequency words: “risk” and “return”.
We will not attempt here to enter into the complicated considerations around assessing the risk-and-return profiles of securitisation bonds versus bank liabilities for various investors. One of the key complexities is the diversity of possible investors who have very different portfolio constraints and need to comply with different regulations. Peculiar and “uneconomic” constraints can create all sorts of unhealthy “arbitrage”.
Fortunately there are two important facts following from the previous discussions:
- Obvious Fact No 1: Risks can be analyzed better for simpler investments.
- Obvious Fact No 2: Returns will be on expectation higher the lower the fees and costs.
Hence the simplicity of the securitisation structure, together with the elimination of extraneous costs should make securitisation relatively more attractive to investors (all else being equal).
End result: Advantage Securitisation
We left – on purpose – this element for the end. It is actually the most important element. Any discussion on revitalizing securitisation aims to make the case that securitisation will help Small and Medium Enterprises (SME), which is the backbone of any economy. Hence it is important to understand how “SME friendly” might be securitisation versus bank finance. The answer is, not much, at least not without losing the main features that make it an attractive alternative, namely its simplicity.
In principle borrowers want the lowest possible cost of funding with the maximum flexibility around terms of both disbursement and repayment of funds.
Borrower friendliness is one area where banking based finance is the undisputed king. It is likely to remain so, no matter how good the new securitisation channel becomes. This simply follows from the design of the bank structure which allows it to “absorb” the complexity of product features that are desirable from borrowers and also the fact that clients may behave unexpectedly. Lets look at those in some more detail:
In good times banks can offer clients enormous product flexibility and mixing of lending products with other non-lending services.
Bank lending products can have huge optionality in the terms they can offer to clients, for example in the form of the already mentioned embedded options: option to
prepay, option to draw more etc.
The issue of what happens when things go sour for the borrower is a particularly weak point for securitisation. The long-term banking relationship creates knowledge and trust which may help banking clients through a workout. A bank has more flexibility compared to an SPV when trying to resolve problems with borrowers that go into difficulties.
As discussed extensively under “complexity”, securitisation can improve this lack of flexibility but at the expense of complexity.
End result: Slam dunk Banking
This significant advantage of the banking model versus borrower friendliness immediately suggests one key design principle for a succesful “new securitisation” paradigm:
Securitisation is not a replacement model for banking. It can never match the flexibility of services offered by the later without losing its key advantages of simplicity, transparency and efficiency. Securitisation should instead aim to be the most transparent, low-cost, soundly risk managed channel for simple loan products. This alone would be a huge step forward.
With the thinking that backs each of the scores appended below, without further ado, this is the summary of our results:
|1||Complexity||++||–||A simple securitisation is much less complex than a simple bank|
|2||Track record||–||–||No matter who you are, past performance does not predict future performance|
|3||Fees & Costs||++||–||Trimming fat is a very disruptive idea|
|4||Transparency||+||–||Openness is the new standard|
|5||Risk Management||+||–||Central to the business, not something you can outsource|
|6||Investor friendliness||+||–||Increase your expected return, know your unexpected return|
|7||Borrower friendliness||–||+++||Your friendly banker is irreplaceable|