12 March 2010

Draft Submission on NBDT liquidity consultation

I have drafted my submission on the RBNZ NBDT liquidity consultation document, and would welcome any comments or corrections. It is due for submission on Monday 15th March 2010.

Submission Summary
I submit that the RBNZ should recommend 'Maintaining approximately the status quo' because:
  1. The public policy case for regulation is very weak. The two arguments raised for regulating liquidity are flawed and I will address them in my submission.
  2. The diversity of business models in the NBDT sector make a diverse range of liquidity management policies appropriate, including definitions, indicators and levels of liquidity risk. This makes the option of 'Prescribing a measurement framework' inappropriate.
  3. Imposing quantitative liquidity requirements would be ineffective, or competitively non-neutral, and/or overly complex, and this option should be decisively rejected.
  4. The particular quantitative liquidity requirements and indicators presented in the consultation document as an illustration from the regulations for registered banks, are flawed and inappropriate for the NBDT sector. I will criticise these indicators in my submission.
Preliminary Assessment
The consultation document asks: 'Do you agree with this preliminary assessment of liquidity risk in the NBDT sector?'
I agree with the following assessment, and reproduce it here to highlight the different strategies and challenges in managing liquidity risk:
During this period [of stress], finance companies within the consumer financing sector have shown some success in their ability to meet deposit outflows with matching maturities of their loan book. However, finance companies within the property financing sector have been more vulnerable to liquidity shortfalls as their loan book assets have proven to be highly illiquid in a stressed market. Saving institutions have generally held high levels of depositor confidence and accordingly liquidity has been managed by the reliance on high levels of reinvestment.
This assessment correctly shows that:
  1. Where the institution's solvency and asset quality are not in question, there is no need for high levels of liquid assets or long term funding, because the institution can retain the confidence of its depositors, experiences no runs, and is able to roll over its liabilities in the normal course of business.
  2. Where the institution's solvency or asset quality are in question, even high levels of liquid assets or long term funding are ineffective in preventing runs and the demise of the institution, they just slow them down a little.
How would a policy of requiring high levels of liquid assets or long term funding, which would have the principal effect of slowing down but not preventing the failure of a financial institution of questionable solvency, serve the purposes of:
  • '(a) promoting the maintenance of a sound and efficient financial system; or

  • (b) avoiding significant damage to the financial system that could result from the failure of a deposit taker'?

There is a missing link here, since the expected number of institutional failures is not likely to be reduced, the damage from failures would be about the same as without such regulations.
'The main argument here is that the failure of a deposit taker in benign times due to a sudden large liquidity shock could significantly affect confidence in similar entities. A deposit taker’s management would not typically factor this public cost into their decision on how much to spend on liquidity protection, and as a result prescription of liquidity requirements could provide reasonable comfort that the degree of liquidity risk would be lower than it would be without intervention.'
This is a remarkably weak 'main argument'. For several reasons:
Firstly, why should a sudden large liquidity shock occur during benign times? I do not believe that this is an appropriate to just assume that large changes will happen without a reason. For a large liquidity shock to occur a large number of the institution's creditors must decide to act at the same time to withdraw their funds or not roll them over. Later in the consultation document there is a reference to funding 'sources that are less likely to withdraw their money at any sign of problems' which provides a good theoretical basis for creditors to cause a liquidity shock to an institution: i.e. the creditors are concerned about the solvency of the institution and their actions are to avoid a high risk of loss. I submit that this assumption has no theoretical or analytical basis and should therefore be dismissed. This scenario is like asking what if all drivers on the roads became very sleepy at the same time? Serious road safety policy does not concern itself with such implausible scenarios, and neither should serious financial policy concern itself with large numbers of people doing unusual things with their funds at the same time without cause.
Secondly, the scenario has no historical precedent. There is no historical example of a banking system in benign times, where one financial institution has had a liquidity problem (while still being solvent), and the problem has spread. In fact even very wobbly and unsound banking systems avoid liquidity problems during benign times. Banking system problems only happen during stress events, and only spread to institutions of questionable solvency. For example, Dr Lawrence H. White, the leading expert on free banking and banking history, and a Professor at George Mason University, in Banking Without Regulation wrote that under free banking:
But bank failures were almost never contagious, or prone to spread to sound banks, for several reasons. Each bank tried to maintain an identity distinct from its rivals, and was able to do so when it was not compelled by any regulation to hold a similar asset portfolio. Depositors then had no reason to infer from troubles at one bank that the next bank was in trouble. Banks were generally well capitalized, so that fear of insolvency was remote.
Thirdly, if we substitute a more plausible scenario for the implausible one in the main argument, i.e. during stressed conditions and affecting similar institutions of questionable solvency, then it is a good thing for funding providers to discipline those other institutions too. (Recall the 'market discipline' is one of the three pillars of banking system regulation.) Is the goal of liquidity regulation to insulate financial institutions from market discipline? No!
The consultation document proceeds to consider stressed conditions:
We can contrast this conceptual situation with the recent sustained period of stress that finance companies experienced, where liquidity management for the most part proved adequate except for vulnerabilities in relation to property financing companies. The main difference here is that, in the recent period, the stress developed over time allowing deposit takers the time to respond and manage stresses as they developed. Another crucial difference is that the stress mainly affected the finance company sector where the long term nature of funding prevented large scale withdrawals of call funding. This argument would suggest that there is potential for NBDTs’ appetite for liquidity risk to be higher than socially optimal in normal times (unlike the current period), but it is not immediately obvious that it has been, or if it was, significantly so.
Firstly, the argument presented above is invalid. The argument boils down to:
  1. During the recent period of stress, most finance companies appear to have had adequate liquidity positions.
  2. During the recent period of stress, the stress developed slowly, allowing more opportunity for management responses.
  3. During the recent period of stress, mainly finance companies suffered.
  4. Finance companies had mostly long term funding.
  5. Therefore: there is potential for NBDTs’ appetite for liquidity risk to be higher than socially optimal in benign times (unlike the recent period of stress)
Point 5 does not logically follow from points 1-4. Points 1 and 4 are basically saying the same thing, that most finance companies had very low levels of liquidity risk because they had mostly long term retail funding, and had generally significantly higher levels of loans maturing than liabilities maturing in the short term (less than 12 months) during the recent period of stress. However, the finance companies' appetite for liquidity risk did not appear to be significantly higher during the benign times before the recent period of stress. Point 2 is at least somewhat debatable.
Secondly, even if point 5 was true, i.e. there was a potential problem, the consultation document itself puts doubt on whether the potential was realised in the benign conditions that preceded the recent period of stress, or whether it would be likely to be realised in a future benign period.
I have to therefore submit that:
  1. There is no theoretical or analytical reason to suspect that there could be a problem,
  2. There is no evidence that there has been a problem recently, and
  3. There is no evidence that any future potential problem is likely to be realised.
The consultation then introduces a further argument for liquidity regulation:
'Prescribed liquidity requirements can also be justified from the purpose of NBDT prudential regulation more generally, that is, to avoid significant damage to the financial system that could result from the failure of a deposit taker. While NBDTs are not systemic to the financial system, they are systemic to a local group of similar entities. Since similar entities have similar business models (particularly in terms of lending) the failure of an entity would have an impact on the availability of funding to particular segments of the economy.'
This argument boils down to:
  1. Quantitative liquidity requirements are prudential regulations
  2. Prudential regulations have the purpose of avoiding significant damage to the financial system resulting from financial institution failures
  3. NBDTs are not systemic to the financial system
  4. NBDTs may be systemic to a group of similar entities
  5. Similar NBDT entities have serve similar borrowing customers, who serve particular segments of the economy, could suffer liquidity problems at the same time, causing reduced access to funding for that segment.
  6. Therefore, quantitative liquidity requirements on NBDTs are justified.
Point 6 does not follow from points 1-5. There are a number of missing links here:
From 5. (segments of the economy) to 2. (significant damage to the financial system).
Firstly, for any segment of the economy, there is no reason to believe that access to funding relies on only a few similar NBDTs, even if they provide most of its funding now. Finance markets are the broadest of markets, and thus we should expect that sources of finance could change quickly in the event of trouble with a few NBDTs.
Secondly, the causation in the above argument is around the wrong way. Normally it is trouble within a segment of the economy that causes trouble with similar financiers, if their exposure to that segment is not well diversified. For example it was not liquidity problems with NBDTs serving the property development segment of the economy that caused a lack of funding to the property development segment in the recent period of stress. Instead, problems in the property development segment caused solvency problems for NBDT financiers, many of whom failed because of bad loans, and none of whom failed due to liquidity shortfalls notwithstanding the value of their assets being more than that of their liabilities (in all cases that I know of the failures of property financiers was caused by extremely large credit losses that typically left first ranking secured debenture holders with less than 50% of their funds).
Thirdly, even if 'the failure of an entity would have an impact on the availability of funding to particular segments of the economy' this is outside of the scope of the prudential regulation, which is concerned not with funding availability to particular segments of the economy but about 'significant damage to the financial system'.
From 1. and 2. (liquidity requirements as prudential regulations) to 6. (justification of liquidity requirements).
Quantitative liquidity requirements are not likely to have a significant impact on the chances of a NBDT or group of similar NBDTs getting into trouble, or the severity of their trouble. Most failures, past, present and future, will continue to be caused by credit risk (with market risk a distant second), and liquidity risk is unlikely to feature in anything more than isolated failures. In fact in the Basle Committee on Banking Supervision, Working Paper No. 13, Bank Failures in Mature Economies, published in 2004 (see http://www.bis.org/publ/bcbs_wp13.pdf?noframes=1 ), although the report identifies a wide range of bank failures, from large parts of the banking system to isolated failures of individual banks, not a single case of a bank failure caused by liquidity risk was found. The same paper examines four cases where only small banks failed -- potentially similar to New Zealand NBDTs -- and in all four cases credit risk was a cause of the failures. It follows that quantitative liquidity requirements cannot be reasonably expected to reduce number of institutional failures, and therefore the merits of quantitative liquidity regulations as prudential regulations are highly questionable.
Furthermore, we have good reasons for believing that mandating stronger liquidity positions will increase rather than decrease the costs and risks of institutional failures. Mandated stronger liquidity positions reduce the power of market discipline and the viability of substitutes such as transparency.
The argument for reduced market discipline goes as follows:
  1. Most credit institution failures are caused by credit risk (i.e. too many bad loans)
  2. Credit institutions with the highest credit losses suffer them primarily as a result of lower efficiency (e.g. less effective credit control processes and systems) and worse management (e.g. poor credit judgement)
  3. Information about lower efficiency, poor management and higher past and prospective credit losses causes depositors and other funding providers to withdraw or not roll over funding, causing liquidity problems for the institution.
  4. Higher required liquidity positions give the institution and its management more time to manage the institution, before they are replaced.
  5. Therefore, higher required liquidity positions has the effect of putting resources in less efficient institutions and under poor management for longer, worsening the social costs of such inefficiency and poor management.
Mandated high liquidity positions will also weaken transparency. This effect is modeled and documented by Dr. Lev Ratnovski, an IMF economist specialising in bank liquidity risk management in his paper Liquidity and Transparency Risk in Bank Risk Management (see http://sites.google.com/site/ratnovski2/LT_new.pdf ). Dr. Ratnovski's model accounts for liquidity risk as being due to lack of perfect transparency. If there was perfect transparency, a solvent financial institution would always be able to refinance its liabilities. Liquidity risk arises as a result of imperfect transparency causing a risk of adverse solvency signals to be received from solvent institutions that require re-financing. Dr. Ratnovski's research models the interaction between two ways of managing this risk:

To manage that risk, banks can accumulate liquid assets, or enhance transparency to facilitate refinancing. A liquidity buffer provides complete insurance against small liquidity needs, while transparency offers partial insurance against large ones as well. ...

Liquidity requirements can compromise banks endogenous transparency choices, leaving them exposed to large shocks. Liquidity risks can increase in response to seemingly more stringent regulation.

Dr. Ratnovski's conclusion holds because the expected benefits from transparency are reduced by higher liquidity ratios. Mandating higher liquidity ratios should therefore be expected to cause better capacity to handle small liquidity shocks, and reduced capacity to handle large liquidity shocks. Wouldn't it be better to address the cause of the risk, i.e. increase the incentive of the institution to invest in improving their capacity to credibly signal their solvency? Institutions could do this by, for example, having more transparent assets, more frequent release of financial accounts, having a more credible trustee and auditor, having a higher capital ratio or having lower risk assets -- the point is that there are complex trade offs between these different approaches, and regulators have no way to judge the merits of the different approaches, which require trial and error, market selection and evolution instead.

I therefore conclude and submit that the argument presented in the consultation document quoted above is so deeply flawed as to merit abandonment.
Next the consultation document states:
However, as stated above, the economic argument for precise quantitative liquidity requirements is not as strong as the case for capital or related party requirements, and issuing further non-binding guidelines could potentially address some of the current liquidity issues. In this case there are still arguments for prescriptive requirements to be set to achieve consistency across the sector and to prevent liquidity management presenting relative competitive advantages or disadvantages across the sector. Short of prescribing quantitative liquidity requirements, prescribing measurement frameworks could also achieve some amount of consistency across the sector and could fill gaps identified in the current framework.
What are the 'current liquidity issues'? No 'current liquidity issues' have been established, only a few invalid or weak argument as to why there might be a potential problem. The RBNZ should not accept that there are 'current liquidity issues' without first establishing and documenting them in a robust manner. There is a danger here of imposing a regulatory solution in search of a problem.
Consistency across a diverse sector such as NBDTs is not appropriate nor desirable. Furthermore, this paragraph concedes that the arguments for regulation in this area are weak ('not as strong').
Finally, if liquidity requirements are imposed, they should be imposed at the appropriate group level to match with the trustee supervision and financial reporting.
'There is also the risk that the presence of the DGS poses a moral hazard complication, but there is reason to believe that industry and trustees will continue to be conservative.'
What reason would that be? If there is reason to believe that, why doesn't the document give the reason? I would say, to the contrary, that there is a reason to believe that the moral hazard complication will reach well past the expiry of the scheme into the next credit cycle, as the RBNZ and Treasury advised the government on 10 Oct 2008, RBNZ and Treasury that the scheme would 'enduringly change expectations about government responses to financial stresses and institutional failures in the future.'
Do you agree that consistency of measurement frameworks for key liquidity constructs across the NBDT sector is advantageous?
No, I disagree that it would be advantageous. This is because of the diversity of the NBDT sector makes different liquidity strategies and risk levels appropriate in different institutions, and this would make potentially different liquidity risk and position indicators appropriate for measurement. There is also a danger that investors, institutions and trustees would take false comfort from inappropriate indicators.
Are there any other matters the Reserve Bank should take into consideration when considering prescribing a liquidity risk measurement or management framework?
Yes, the RBNZ should take into consideration
  1. the inherent complexity of managing liquidity risk
  2. the subjectivity inherent in measuring future potential cash flows
  3. the difficulty of determining the relevant future scenarios to plan for
  4. the inherent limitation of any liquidity indicator or indicators, and
  5. the futility of good liquidity management in addressing institutional insolvency.

To support my 5 points above consider what liquidity risk is and how it can be measured. Liquidity risk is the risk that the institution will run out of cash in the future. What happens in the future depends on future conditions, future perceptions, and future actions taken by the institution's counterparties. The future is unknown, so the best we can do is make a load of assumptions that institutions, managers, supervisors and others can debate and discuss or perhaps come to some kind of compromise that will inevitably turn out to be flawed. This makes regulation of liquidity risk by way of cash flow modeling very troublesome. Regulators can therefore be tempted to avoid the use of cash flow modeling and use liquidity indicators instead. Liquidity indicators are basically accounting ratios that can be more easily measured, because they are measuring the present rather than the future, but provide a limited and flawed measure of liquidity because they ignore the future cash flows that determine whether or not the institution actually runs out of cash. The most important limitation of liquidity ratios is that they are totally ineffective at curing the majority cause of institutional failure: insolvency.

As a result of considering these factors the RBNZ should appreciate that imposing a measurement framework or suite of indicators is will:
  1. Not be an effective method of promoting the purposes of the Reserve Bank Act (i.e. an efficient financial system and avoiding significant damage to the financial system)
  2. Be a risky, costly policy that will have very limited benefits. The benefits are likely to be small in relation to the costs.

'37 Our preliminary analysis suggests that many of the larger NBDTs would easily exceed the ratio requirements calibrated for locally incorporated banks. This result is unsurprising given NBDTs’ reliance on retail funding compared with banks. Other obvious differences between NBDTs and banks which would require different calibrations in the policy are the assumptions around reinvestment rates of retail term funding and the accessibility of committed lines.
38 To calibrate the bank liquidity policy for NBDTs the main issue, as outlined previously, is that of diversity of business models within the sector. Here NBDTs differ from the banking sector where diversity is mostly in terms of size. Theoretically, the bank liquidity policy could be recalibrated to the particular requirements of the different business models of finance companies and bank-like savings institutions. The complexity that arises here is the distortionary effects from varying requirements within a sector, and consequently the basis on which different calibrations should apply. Another option is for the bank policy to be calibrated at a minimum level across the sector. However, this could pose moral hazard where NBDTs could be less inclined to form their own view of how vulnerable they may be to liquidity risks, and take other necessary steps. The presence of trustees somewhat mitigates this risk.'
These paragraphs present the key disadvantages of imposing quantitative liquidity requirements:
  1. If imposed on the same basis as on registered banks, they would be non-binding and therefore achieve no liquidity increase effect, and therefore be bad public policy.
  2. If imposed at a higher level, as might be appropriate for many but not all NBDTs, it would be competitively non-neutral, and therefore bad public policy. It would also give a false comfort to investors, institutions and trustees, and make them less likely to independently assess their policies, and this also makes it bad public policy.
  3. If imposed at different levels within the different business models in the NBDT sector, it becomes very complex, and even more competitively non-neutral, and therefore bad public policy.
One must therefore conclude, and I submit that, quantitative liquidity requirements are bad public policy at any calibration, and must be rejected

The calibration of quantitative liquidity requirements, should the RBNZ be determined to impose them on the NBDT sector, should take into account:
  1. The possibility that a long term liability may become due early due to credit rating downgrade or breach of financial covenants such as interest coverage. For example, South Canterbury Finance US Private Placement Notes became repayable on demand when the company's credit rating was downgraded, which resulted in it running out of cash. I suggest that any long term funding measure be not subject to early repayment other than in the case of receivership or liquidation of the institution.
  2. The security ranking of funding. To qualify as long term funding, the funding should be unsecured or should rank equally with the principal class of debenture stock.
  3. Availability of unused allowances for prior charges. These can be utilised in times of stress, and many stressed and failed NBDTs have used such allowances to delay running out of cash despite failing due to insolvency.
  4. The liquidity that can be provided from supposedly illiquid assets such as loans and advances. These can provide liquidity by way contractual repayments, securitisation or use as security for borrowings, or by way of sale of senior tranches. I suggest that impaired and past due assets be excluded from any such measure.

No comments: