Speech Financial Stability and the Banking Sector
Head of Financial Stability Department
Sydney Banking and Financial Stability Conference
I'd like to thank the University of Sydney and the organisers for the opportunity to address this conference, which is on the subject of banking and financial stability.
Over the years I've had the privilege of being able to speak publicly about many aspects of financial stability and its drivers. But perhaps nothing is more important to financial stability than a sound banking system. To understand why, we first need to understand what makes banks different from other players in the financial system. After discussing that question, I will explain why the unique features of banks make the banking system so important for financial stability, before turning to the implications of this for policy across a range of domains.
What Is a Bank and Why Are Banks Different?
When I was a more junior person in the Reserve Bank, I had the task of helping edit and produce the conference volume of the Bank's 1996 conference, on the topic of ‘The Future of the Financial System’ (Edey 1996). In fact the event was almost exactly 20 years ago today. Coming back to those papers 20 years later, I was struck by how contemporary some of them seem. But the issues these papers canvassed, which seem so up-to-date, are the ones that didn't play out as the authors anticipated. The supposition of some observers at the time was that banks were in decline (Llewellyn 1996). They would be squeezed out by niche competitors and, in particular, institutional investors (Davis 1996). That financial markets would grow in importance was also one of the underlying assumptions of the Wallis Inquiry into Australia's financial system, which reported the following year (Financial System Inquiry 1997).
These are all arguments we hear today, as well. The challenge to banks from new entrants enabled by new technologies, the growing role of fund managers – these were issues then as now. The sequel might not turn out the same way, but we have already seen earlier episodes of this particular movie franchise.
Banks didn't fade away two decades ago. They evolved, and if anything have become even more important than before. We can see in this chart that the share of banks in the overall financial system started rising noticeably in the 1990s. Funds management, in the form of Australia's (compulsory) superannuation system, has also become more important. As a share of total financial sector assets, though, banks have become even more central to Australia's financial system in the past 20 years (Graph 1). Back in the early 1980s, non-bank financial intermediaries were almost as important as banks. In contrast, nowadays the share of financial intermediation outside the prudentially regulated sectors is very small – the orange bars compared with the sum of the two blue bars.
In some respects, banks have grown in importance because they are no longer being restrained by the tight restrictions of prior decades. The supposition of the debate in the 1990s had been that deregulation would make it harder for banks to compete with new entrants. The reality was that banks were being constrained by those regulations more than they were being shielded by them. Certainly that's the message of the non-banks during this period, many of which voted with their feet and became banks, just as a number of credit unions have done in recent years. The next graph, which shows bank lending (excluding bills) as a share of total credit intermediation rather than total assets, shows this turnaround even more starkly than the previous one (Graph 2).
How did banks not only retain their role in the financial system, but expand it, against those earlier expectations? To understand that, we need to consider what it is about a bank that makes it different from other kinds of financial players. In doing so, I need to make it clear that when I refer to ‘banks’ in a contemporary Australian context, I really mean ‘Authorised Deposit-taking Institutions’ (ADIs). As well as banks, these include building societies, credit unions and a few other small entities, all of which are regulated by the Australian Prudential Regulation Authority (APRA) and all of which are functionally banks for the purposes of this discussion; the function matters, not the label.
That phrase ‘Authorised Deposit-taking Institutions’ is a bit of a mouthful, but it gets to the heart of what is special about banks. Uniquely among financial services businesses, banks can take deposits. A deposit is a claim on another entity that can be converted at par and on demand into cash or used to make payments. Its nominal value can be relied upon, and so can its liquidity. This is a valuable service: in their seminal paper, Diamond and Dybvig (1983) showed that deposit contracts improve on a competitive market's capacity to share risk between people who cannot perfectly predict when they will need to make payments.
The issue here is that no private sector entity can actually promise to provide value at par on demand at all times, good or bad, without having access to central bank liquidity. Many of the problems revealed during the Global Financial Crisis were the results of entities and markets that were trying to get around this fact. From asset-backed commercial paper to money market funds, from repo to auction rate securities, the financial system tried to create products that offered deposit-like liquidity. None of them could make good on that promise when risk appetite turned. Fake liquidity is fake. It can work for a while, but it won't work if people start to panic.
There is more to this than maturity transformation. Yes, banks are in the business of borrowing short and lending long. But they aren't unique in doing so. Two things make banks different in this area. First, banks don't just borrow short: in the case of most deposit products, they are promising to return funds on demand. Many investment funds allow investors to redeem funds faster than some of the underlying assets can be sold, but in most countries there are limits on the speed and scope of redemption. Most investment funds don't – and in Australia are not allowed to – offer full redemption on demand. Banks don't just transform maturity; they create liquidity.
Second, when banks borrow by way of accepting a deposit, their creditors don't want to and shouldn't need to care about the bank's creditworthiness. The promise that banks are making is that deposits are essentially risk-free. For the purposes of making payments, a deposit is as good as cash, or at least it should be.
There are times, though, when depositors panic and want cash, not deposits. This is the story behind most historical financial panics, and the Global Financial Crisis was no exception to that. As has been detailed by Gary Gorton, among others, depositors will sometimes stop treating deposits as risk-free, because the assets backing them, by definition, cannot be risk-free (Gorton 2012). (The same issue applies to other short-term debt instruments or investment funds that people usually treat as risk-free or nearly so.) The only way to make those deposits truly risk-free in all states of the world is if the public sector ensures that convertibility to cash is guaranteed. This usually involves government insurance of deposits and/or the central bank acting as lender of last resort (Diamond and Dybvig 1983). That is why other products fail to act like deposits when it really counts: without public sector backing of their liquidity, it doesn't work.
We know this is true, because the alternative has been tried and shown not to work. ‘Free banking’ was permitted at times in past centuries, for example in parts of the United States. In that regime, anyone can set up a ‘narrow bank’ that ostensibly offers risk-free deposits by only investing those deposits in safe assets such as government bonds. But as has been documented elsewhere, free banking didn't do away with financial panics, and completely failed to provide an asset that traded at par, the way a deposit should. Most banknotes in the free banking era traded at large discounts. This imposed considerable costs and inconvenience on the broader economy (Gorton 2012).
If it's so hard to provide a deposit-like asset without central bank support, one might reasonably ask why society allows private sector banks to do this at all. Why don't we just insist that if the central bank didn't issue it, it isn't redeemable at par? Why don't we just let everyone have a deposit account with the central bank and make everything else a mutual fund with varying redemption values?
The answer to this is connected to the other function of banks that makes them different, and gives them a special role in the functioning of the financial system. The amount of safe, deposit-like assets that society demands is large relative to the size of the economy (Graph 3). If all of that were directly a liability of the central bank, it would need an equivalent amount of assets on the other side of its balance sheet. (That's how balance sheets work.) If only central banks could offer deposits, they would end up as the credit providers of first resort in an economy. Decisions about who would and would not get finance would more often than not be made by an agency of government. You don't need to be an advocate of laissez faire market forces to be uncomfortable with that outcome; it also implies a lot of monopoly power accruing to the central bank as credit provider. While in this post-crisis era, some central banks' balance sheets have grown very large, ordinarily they are much smaller than total deposits in the financial system, let alone total credit.
This brings us to the second special role of the banking sector, that of credit provider. We can see in the graph below that in most countries, banks and other deposit-taking entities account for the largest share of total credit and (non-intermediated) debt funding of the private non-financial sectors (Graph 4).
Australia is one of the more bank-oriented financial systems when it comes to providing credit, but it is hardly alone. And some of the countries at the lower end of the range, such as the United States and Canada, are there partly because their governments support the securitisation market in various ways. These interventions allow banks to take some exposures, particularly mortgage exposures, off their balance sheets. In some cases they also allow some non-bank loan originators to operate at larger scale than might otherwise be possible.
There are two reasons for banks' special role in credit provision. First, there are plenty of potential borrowers, specifically small businesses and households, who do not have direct access to finance from wholesale markets. Second, credit risk assessment is hard and costly. So except for the large borrowers with access to wholesale funding, there will be strong incentives to do that assessment only once.
Let's look at this first reason in more detail. Many small businesses, particularly startups, don't have a payoff profile that is compatible with debt finance. If most startups fail, but there's a small chance they're Google, the payoff structure is negative in most states of the world but extremely positive with some small probability. That's an equity-return profile, not a debt-return profile. Debt funding has no upside beyond the agreed interest rate.
But there are some small businesses, especially those with some collateral, that make good candidates for debt funding. More established businesses with collateral and at least moderately stable revenues might be riskier than some other borrowers, but they offer payoff streams with the right structure for a debt contract.
It's the same with households. Most households have a more or less predictable stream of labour income that fits the payoff structure of a debt contract rather well. And as I've pointed out in an earlier speech, households can't issue equity in themselves, so debt funding is their only option for private finance (Ellis 2016).
Although these small business and household borrowers are good candidates for debt funding, they can't access wholesale markets because they are simply too small to do so. There are fixed costs to getting a bond issue away; no household or small business has the scale to pay these. And in any case, the scale of these borrowers relative to the portfolio size of the investors is small. There is no benefit to splitting the exposure across multiple wholesale lenders. It is easier to take on a single small borrower's debt in a single portfolio, diversified with many other small borrowers.
The reason why those small borrowers' loans tend to end up with a single lender stems from the second reason banks have a special role in credit supply. Credit risk assessment is hard, and it takes real resources. If a single lender can fund a single borrower, they will do so. Splitting the borrowing across multiple creditors just duplicates the resource costs involved in deciding if somebody should get credit at all. Distributing loans across many small creditors will have its limits, too. Setting up the capacity to assess credit risk involves some fixed costs, and thus some economies of scale. And the diversification benefits are less when you take on a few whole loans instead of a slice of a larger portfolio. The incentive to delegate to an intermediary is therefore quite strong. In other words, financial intermediation exists precisely because it uses resources more efficiently than disintermediated market-based finance if the borrower and the end-investor are small.
Market-based finance has evolved structures intended to reduce this cost disadvantage. Some of these include using credit rating agencies instead of replicating credit risk assessments, or pooling smaller loans into securitisations. But as the experience of the Global Financial Crisis showed, they are not a perfect substitute for traditional financial intermediation. The incentives to do good credit risk assessment are more powerful when it is your own balance sheet at stake. And while agency problems also occur in traditional intermediation performed by banks, they are far more complex for the chain of relationships in market-based disintermediated finance.
The difficulties involved in credit risk assessment also mean that it can be harder for new entrants to break into this area than some observers might think. While I have no doubt that modern data analytics can assist the process of credit risk assessment and make it more rigorous, setting up the right algorithms is a hard problem. It will require data and human skills that are in short supply.
You could reasonably ask: why don't credit funds account for a greater share of credit provision? Why does a credit-heavy asset side of the balance sheet have to go together with a deposit-heavy, or debt-heavy, liabilities side? Why not fund those debt investments entirely with equity-type claims? In other words, why is the banking system so leveraged?
The main reason that this pattern arises is that equity-type claims expect equity-style returns. But the design of a loan contract is inconsistent with that. The nature of the payoff structure simply doesn't match an equity claim. There is no upside to a debt investor if the borrower does particularly well. There is only the downside if the borrower defaults. The payoff is by design more stable and therefore lower risk. Debt is also lower risk because in most legal systems, including Australia's, creditors get paid out before equity investors. And because debt is lower risk, on average debt involves a lower return than equity. Lower-return, lower-risk investments will tend to attract investors that can fund themselves at the lowest cost, and that isn't equity funding.
Low-leverage or unleveraged credit funds will always play some role in the system, of course, but they are unlikely to be the main provider of credit to the rest of the economy. They are more likely to fill in some niches that the mainstream banking system doesn't service. In particular, they will tend to be at the riskier end of the spectrum, often lending to borrowers that the banks rejected. There is nothing inherently wrong with that business model, as long as the end-investors understand that they are funding higher-risk business with no upside payoffs, and are compensated accordingly in the returns they are offered.
If credit provision is primarily debt-funded – whether in the form of deposits or something else – it raises other issues, including for financial stability. Let's turn to those issues.
Why Are Banks So Central to Financial Stability?
Another reason that banking systems are special is that their performance is central to achieving financial stability. We know that banking crises are all but synonymous with economic crises. We know that economic downturns accompanied by banking crises are usually more painful than downturns where the banking system remains sound (Reinhardt and Rogoff 2009). Recoveries are usually weaker than normal or even non-existent when a banking crisis occurs (Cerra and Saxena 2008). Even the best-handled resolutions of a failing banking sector have involved large declines in economic performance. And as we have seen in recent decades, first in Japan and then in Europe, if the banking system remains weak for many years, so does the economy.
Why are banking crises so damaging, and so critical to generating financial crises? We could seek to understand the banking system's special role through the lens of the Reserve Bank's simple four-point framework for identifying parts of the system that pose systemic risk. As set out in the Bank's submission to the 2014 Financial System Inquiry, systemic risk can be seen as coming from an entity or sector's size, its interconnectedness, its correlation or its procyclicality – or some combination of these drivers (RBA 2014).
It is evident that the banking system is big; it is also evident that it is interconnected with every other sector both as a borrower of deposits and wholesale funding, and as a provider of credit. The services provided by the banking system are essential to the efficient functioning of the broader economy. Perhaps less obvious is their correlation with each other. As was described in that submission, banks tend to have similar exposures, so if one of them becomes distressed, others probably will, too. Or you could focus on the procyclicality of banks inherent in their leverage.
But there is a simpler way to think about the special role of banks in financial stability. It stems from work in the 1980s, specifically the seminal work of Diamond and Dybvig on bank runs and their consequences (Diamond and Dybvig 1983). And it can be summarised in a line from another 1980s hit, Kate Bush's 1985 song ‘Cloudbusting’: ‘what made it special, made it dangerous’. It is the banking system's special roles in the payment system and the provision of credit that mean that dysfunction in the banking system can be damaging to financial stability and the economy.
To see why this is the case, consider again the world of free banking, when people could not be sure how much their banknotes would be worth at the moment they came to spend them. Consider the cost and disruption that would be imposed on the rest of the economy if the payment system worked this way. If a business wasn't sure that the funds to make this week's payroll would be worth what they expected, what would that do to its decisions to employ or invest?
Consider also the costs imposed on the real economy if businesses could not get credit to fund their daily operations. Imagine if households could not use short-term credit to smooth fluctuations in their expenses or partly finance long-lived items like homes and cars with credit. Think how long it would take someone to achieve home ownership if they had to pay for their home all in cash, up front, with no mortgage.
So it is not hard to see that if the banking system is not playing its role effectively, the cost to the economy can be large.
The special fragility of banks comes back to the special nature of their liabilities. Banks offer value at par and on demand. As I described earlier, they create liquidity as well as transform maturity. Yet the assets backing those deposits cannot all be realised immediately. Those that can be liquidated will probably require accepting a significant ‘fire-sale’ price discount. This remains true even when we are talking about maturing short-term wholesale funding instead of redeeming deposits.
Because much of the bank's asset base is loans, liquidating some assets quickly might not be feasible at any price. If any of you have a mortgage, consider whether you could pay it back by the end of the day, in cash, if the bank called you right now. I am guessing that you could not do so. That is why banks can't pay out all their depositors at once, even if every borrower could make their previously agreed repayments. Knowing that if a run starts, the banks' assets are unlikely to cover its liabilities, at least not at fire-sale prices, it is in all depositors' interests to run.
One of the challenges is that a run can occur even when the bank is sound. Certainly a bank is more likely to fail if there are doubts about the quality of its assets, but that isn't even necessary. All that is required is for depositors to believe that other depositors will run on the bank; then they better be the ones to get their money out first. Rumour and panic can bring down a bank. Or as Kate Bush might put it, ‘just saying it could even make it happen’.
But if the bank were to actually try to meet redemptions by realising assets, by calling in loans, think how damaging that would be. Asset values would fall and damage the ability of unrelated parties to finance their activities. The act of calling in loans would also be directly disruptive to the affected borrowers.
So we've established that distress in the banking system can harm the rest of the economy, and that even sound banks can be subject to runs and fail. The special role of the banking system goes beyond these issues, though, because banks are all too prone to get it wrong on asset quality, if left to their own devices. In other words, banks are also prone to runs that happen for good reason.
Getting it wrong on asset quality is generally a story of making loans that go bad. Sometimes this is understandable. As I mentioned earlier, credit risk assessment is hard. The economic cycle generates correlations in loan defaults in ways that are hard to predict. Loans that seemed sensible in good times, or even in moderately bad times, might still default in particularly bad times. Sometimes those defaulting loans can end up concentrated in one or a few institutions, whether through bad luck or poor management.
Sometimes, though, poor asset quality in a bank isn't the result of bad luck in the face of the best will in the world. Sometimes people forget that credit risk is correlated over the cycle, or stop thinking beyond the short term. The private incentives to do good credit risk management are stronger when the entity's own balance sheet is at stake than it is in an originate-to-distribute business model. But even on-balance-sheet lending might still result in more risk and worse asset quality than is socially optimal. Banks are, by the nature of their liabilities, highly leveraged entities. And even when they have lots of equity, they are still limited liability companies. It is still other people's money at stake. So like any leveraged entity, they will face the temptation to take more risk in order to boost returns. Those returns might be to the shareholder or to the internal decision maker, depending on remuneration structures. Like any organisation, a bank contains a great number of agents, making decisions on behalf of principals who cannot observe them perfectly. The interests of agents might not always align with those of the principals. This is why remuneration structures and governance are so important.
Also relevant, but in my view less important, is the issue of moral hazard. It is commonly assumed that banks take more risk because of their special role, because they believe that they are too big or important to be allowed to fail. In principle, this can be an issue. In my experience, though, no financial institution ever made a decision on the basis that they would be bailed out. Nobody expects a bail-out, because nobody expects to fail. The institutions that failed during the Global Financial Crisis were often driven by a belief that they were too smart to fail.
What Are the Implications for Policy?
In a world where banks are central to financial stability, and society wants financial stability, what does this mean for policy?
The roles of the regulators
The first implication is that banks will always need to be regulated and supervised. The quid pro quo of the social licence to offer deposits is a requirement to submit to public intervention. Strong, effective and holistic prudential supervision of deposit-takers is essential to financial stability. The Australian financial system has managed to weather the external shocks of the past two decades reasonably well. Strong prudential supervision has helped achieve that positive outcome.
Supervision goes far beyond ensuring that banks have enough capital, though that's important. Leverage matters, but how an entity behaves, given their leverage, matters more. As my APRA colleague Charles Littrell has pointed out at a recent Centre for International Finance Regulation panel, history shows that banks can have much higher shares of capital in their liabilities than we see nowadays, and still find ways to fail, if they take enough risk on the asset side of their balance sheet. We should remember that the policy measures that safeguard the liquidity of bank deposit liabilities, such as deposit insurance and liquidity provision by the central bank, can create incentives for banks to take those risks.
If the ultimate goal of financial stability policy is the real economy, it is not enough to require banks to hold enough capital to absorb losses, while disregarding the scale of those losses. The losses themselves can represent distress in the economy. The holders of capital are often part of the same economy, so absorbing the losses does not make them go away. Absorbing the losses and thus avoiding a collapse of the banking system prevents the knock-on effects on other parts of the economy, which is better than nothing. But it would be irresponsible to disregard the risk profile of the banking system's assets, as long as banks have enough capital to cover those risks.
Another implication for policy is that if society has decided that a private entity can offer deposits, which requires central bank liquidity to work, then the central bank must indeed provide that liquidity. Central banks are not in the business of lending to insolvent institutions (in the sense that they have no capital) or otherwise unviable institutions. But we can't be too precious about this point, either. Of course we need to consider moral hazard, but moral hazard isn't the only consideration. Given the costs of a crisis, it is surely a worse mistake to refuse to provide liquidity to a bank that is probably solvent but might not be, than to provide liquidity to a bank that looks solvent but turns out not to be.
I don't want to minimise the difficulty of making that assessment at the point it is needed. Valuing illiquid assets, such as portfolios of loans, is hard, especially in the midst of a crisis. There will always be some chance that the actual, realised valuations are much lower than can be reasonably expected. There will always be some chance that some new shock comes along that wipes out the remaining capital of an already distressed bank. But a central bank that hesitates in the face of these risks is doing nobody any favours. In a liquidity crisis, a central bank's role is to lend freely (but at a high rate) against good collateral. Obviously that requires banks to have enough good collateral to lend against. Asset quality is key. But we don't need to be too picky about what kind of assets the central bank might lend against in times of stress, as long as they are of sufficient quality.
This principle is nothing more than what Bagehot (1873) said all those years ago. But people sometimes forget that this means lending against collateral valued at pre-panic prices. Some people will always regard whatever happened before the panic as inflated and somehow illegitimate. While some asset prices can reach over-exuberant levels in the lead-up to a panic, fire-sale prices in a panic are no more based on fundamentals than mania-phase prices are.
In fact, if the central bank's commitment to convertibility is credible, it might not actually be necessary to provide liquidity at all. The historical Australian experience suggests that reassuring public statements can also be effective in assuaging the fears of nervous depositors (Fitz-Gibbon and Gizycki 2001).
That's the response in a systemic event. Banks still need to take care of their own liquidity needs in normal and moderately stressed times. That is why prudential supervisors around the world have long imposed various kinds of requirements on banks' liquidity management. In recent years, those requirements have been harmonised in the Basel III package of reforms. Specifically, the Liquidity Coverage Ratio has already been introduced in Basel Committee member jurisdictions, including Australia, and the Net Stable Funding Ratio requirement will be implemented in the near future.
The broader policy environment
A third implication for policy is that, even while banks remain at the core of financial stability issues, policymakers still need to look at the whole system. The objective of financial stability policy is to protect the real economy, not look after banks for their own sake. Regulatory and supervisory measures that might seem superfluous to protecting the solvency of individual institutions might still be necessary for the stability of the system as a whole and of the economy.
The policies adopted with that systemic state of mind will in fact take many forms. Some will be squarely in the prudential realm. The recent supervisory measures affecting mortgage lending are a good example of this. Nobody is suggesting that the risks in banks' mortgage books would, on their own, threaten banks' solvency. But a financial stability mandate (which APRA also has) is a mandate to think about the resilience of the economy more generally. That means thinking about the broader risk environment, including those faced by other sectors. In the end, though, a resilient customer base is also in the interests of the banking sector.
Another good example of this broader policy perspective is resolution policy. If the mandate were solely about safety and soundness of banks, yet one failed anyway, would it matter how badly it failed or who else got hurt? By contrast, if the mandate is the real economy, policy should seek to minimise the economic costs of failure. Effective resolution tools that do not presuppose particular solutions are essential. That is why the Australian authorities have been working on progressive improvements in our domestic and cross-border resolution capacity.
Some of the policy domains that might be relevant to the interaction of banking systems and financial stability might be less obviously connected to banks. But everything is connected. To do financial stability policy effectively, you need to be able to join the dots. In recent years, financial stability policymakers in Australia have had cause to take views on areas as diverse as tax policy, migration numbers, urban planning and pension finance. Because society has other goals as well as financial stability, the policy configuration that best supports financial stability might not be the one we end up with. And that's completely fine. Social policy choices are all about weighing up competing priorities. It's not for those with just one mandate out of many to say how that should be done.
The resilience of credit supply
Related to the need to look at the whole system is that, in their efforts to protect the real economy, policymakers need to ensure that credit is still being supplied to good borrowers even in bad times. A healthy and resilient banking sector can help achieve that; indeed, it would be difficult to manage it without one.
The ongoing need for credit, and thus the value of a well-functioning creditor sector, is sometimes underappreciated. Especially since the crisis, the dangers of too much credit have become all too apparent. Over-exuberant lending and borrowing can mean that some people are getting loans that they have little prospect of being able to repay even in good times.
Less well appreciated are the costs of having too little credit available. The point here is simply that in recognising that too much credit can be dangerous, we should not instead fall into the trap of thinking of all borrowing as illegitimate or somehow immoral. Less credit isn't always and everywhere better. The low levels of credit available in economies in the regulated era of past decades are not the benchmark we should be evaluating ourselves against now when we try to assess the risk in the system. Some activities can and should be financed with at least some debt, even in bad times – even though there are plenty of others that should not.
Let me be clear that Australia is not anywhere near having this problem; whatever the concerns about concentration and competition in the Australian financial system, there is plenty of finance readily available to lower-risk customers. But some recent examples overseas show the damage that can be done when there isn't enough credit available.
If I were to pick an analogy to credit, it would be food. We know that excess weight can in certain circumstances be dangerous to our health, and excess debt can be dangerous for our wealth. But the solution to obesity is not to stop eating completely. In fact, eating too little can be far more damaging than eating a bit too much. Likewise, financial stability is not best pursued by banishing debt altogether.
Neither is there, in the case of either eating or borrowing, an obvious bright line between what is safe and what is disastrous. There will, as a matter of arithmetic, be times when borrowing activity is stronger than average. Policymakers must of course be alert to the possibility that these periods will increase overall risks to financial stability, and respond if necessary. But in some cases the effects will be reasonably expected to be immaterial, or at least manageable. Policymakers should seek to control the risks posed by credit growth, not to prevent all episodes of strong credit growth. That would be like trying to ban date nights and Christmas dinners, when what you really want to do is reduce ill health.
The analogy can only go so far, though, because science has provided us with a good idea of how much to eat, given our activity levels. Economics has not, unfortunately, provided us with much of a clue about safe debt levels across a whole economy. And that is precisely because knowing what works for an individual doesn't help much at the national level. Knowing what is healthy eating for an individual doesn't tell us what the average weight of the whole country should be. Likewise, a good understanding of the credit risk posed by an individual borrower doesn't tell us the safe level of total debt for the whole country. It depends on where the debt is distributed.
The analogy also only goes so far because the economic and financial systems and the entities within them are so interconnected. One person's food choices might affect a few others in the same household, if they get to decide the menu. But the borrowing decisions of a much wider range of economic entities will affect the economic fortunes of each of us. If those decisions put those borrowers into distress, they will spend less. This will drag on economic performance in ways that can affect everyone. And if those borrowers are businesses that end up going bankrupt, whole supply networks can be disrupted in ways that are hard to restore.
This question of distribution is possibly also one of the reasons why some observers don't appreciate the need to maintain credit supply in bad times. If the original problem was too much debt, they reason, how can the solution be more debt? But remember that we are usually talking about different borrowers. One borrower might well have taken on too much debt, but the solution is not to deprive someone else of finance that they can reasonably handle.
It might be too hard to know what a healthy debt level might be in aggregate. Nonetheless policy can do what it can to encourage good borrowing and lending habits. If the banking system maintains sound underwriting standards at the level of the individual loan, we can have some confidence that the aggregate outcome is also sound. I find it extraordinary that in all the avalanche of reforms that have occurred since the crisis, there has been so little international attention to lending standards. There is still no internationally accepted common language about lending standards. The few international initiatives have been narrowly focused on specific portfolios. The conversation ends up being all about quantities, not quality.
What should we conclude from all this? First, we can conclude that a modern economy cannot do without banks. But by their nature, banking systems imply risks and challenges that we cannot eliminate. We will always have to accept a certain level of risk both to individuals and to the system, but there are ways to keep these risks tolerably low. In particular, policymakers need to supervise rigorously and provide liquidity when warranted.
We should also be prepared to conclude that financial stability policy, like life, is messy. Society has many priorities; financial stability is just one of these. Public policy is all about balancing competing considerations. Financial stability policy is also messy because of what we don't know. Mechanistic approaches to financial stability policy are unlikely to be particularly fruitful given the current state of knowledge; indeed they might never be. Asset quality usually matters more than raw quantities. But it is much harder to measure. And for that reason, good lending and borrowing habits are more important to instil, than a policy approach centred on a few numbers.
Thank you for your time.
I would like to thank Tim Atkin for assistance in preparing this talk. [*]
Some deposit products, such as term deposits, are not payable on demand, but they share with at-call deposit products the status of being treated as risk-free and not information sensitive, in the sense of Gorton (2012). Government insurance or guarantees of deposits generally also does not distinguish between term and at-call deposits, though there is usually a size limit to this protection, such as the $250,000 limit per customer per institution that applies in the Financial Claims Scheme in Australia. 
I am referring here to the provisions in Australia's Corporations Act 2001 that requires managed funds to freeze redemptions if the fund should become illiquid, as defined in the Act. 
Bordo (1990) records that the free banking era of 19th-century Scotland performed better, partly because the banks could diversify by setting up branches. However, this was not a test of a system without a public sector lender of last resort because Scottish banks still had access to central bank liquidity if needed. 
This would be true even if the assets had no appreciable credit risk, such as domestic-currency government bonds. 
In the song, it was ‘something good’ that was going to happen, which a bank run certainly is not. 
This is the channel of transmission emphasised by Fisher (1911) and Bernanke and Gertler (1989). 
A good example of this is set out in the UK Parliamentary Inquiry into the failure of HBOS, which pointed to problems stemming from ‘a belief that the growing market share was due to a special set of skills which HBOS possessed and which its competitors lacked.’ (UK Parliamentary Commission on Banking Standards 2013, p 8). 
See Goodhart (1985) for an earlier statement of this point. Lowe (2013) made the more general observation that there are times that the public sector's balance sheet can play an important stabilising role in a crisis. 
This is not a new point. See Debelle (2011). 
Some of the places these views were expressed include RBA (2014) and RBA (2015). 
See, for example, the Financial Stability Board's Principles for Sound Residential Mortgage Underwriting Practices, available at <http://www.fsb.org/2012/04/cos_120401/>. 
Bagehot W (1873), ‘Lombard Street: A Description of the Money Market’, HS King, London.
Bernanke B and M Gertler (1989), ‘Agency Costs, Net Worth, and Business Fluctuations’, ‘American Economic Review’, 79(1), pp 14–31.
Bordo MD (1990), ‘The Lender of Last Resort: Alternative Views and Historical Experience’, ‘Federal Reserve Bank of Richmond Economic Review’, pp 18–29.
Cerra V and SC Saxena (2008), ‘Growth Dynamics: The Myth of Economic Recovery’, ‘American Economic Review’, 98(1), pp 439–457.
Davis EP (1996), ‘The Role of Institutional Investors in the Evolution of Financial Structure and Behaviour’, in M Edey (ed), ‘The Future of the Financial System’, Proceedings of a Conference, Reserve Bank of Australia, Sydney, pp 49–99.
Debelle G (2011), ‘Collateral, Funding and Liquidity’, Conference on Systemic Risk, Basel III, Financial Stability and Regulation, Sydney, 28 June.
Diamond DW and PH Dybvig (1983), ‘Bank Runs, Deposit Insurance, and Liquidity’, ‘Journal of Political Economy’, 91(3), pp 401–419.
Edey M (ed) (1996), ‘The Future of the Financial System’, Proceedings of a Conference, Reserve Bank of Australia, Sydney.
Ellis L (2016), ‘Booms, Busts, Cycles and Risk Appetite’, Macquarie University Financial Risk Day 2016 Conference, Sydney, 18 March.
Financial System Inquiry (1997), ‘Financial Inquiry Final Report’, Australian Government Publishing Service, Canberra.
Fisher I (1911), ‘The Purchasing Power of Money: Its Determination and Relation to Credit, Interest and Money’, Macmillan, New York.
Fitz-Gibbon B and M Gizycki (2001), ‘A History of Last-Resort Lending and Other Support for Troubled Financial Institutions in Australia’, RBA Research Discussion Paper No 2001-07.
Goodhart CAE (1985), ‘The Evolution of Central Banks’, London School of Economics and Political Science, London.
Gorton GB (2012), ‘Misunderstanding Financial Crises: Why We Don’t See Them Coming’, Oxford University Press.
Llewellyn DT (1996), ‘Banking in the 21st Century: The Transformation of an Industry’, in M Edey (ed), ‘The Future of the Financial System’, Proceedings of a Conference, Reserve Bank of Australia, Sydney, pp 141–179.
Lowe P (2013), ‘Some Tensions in Financial Regulation’, Address to the Institute of Global Finance Second Conference on Global Financial Stability and Prosperity, Sydney, 4 July.
RBA (2014), ‘Submission to the Financial System Inquiry’, Financial System Inquiry, March.
RBA (2015), ‘Submission to the Inquiry into Home Ownership’, House of Representatives Standing Committee on Economics Inquiry into Home Ownership, June.
Reinhardt CM and K Rogoff (2009), ‘This Time is Different: Eight Centuries of Financial Folly’, Princeton University Press, Princeton.
UK Parliamentary Commission on Banking Standards (2013), ‘“An Accident Waiting to Happen”: The Failure of HBOS’, Fourth Report of Session 2012–13, Volume 1, 7 March. Available at <http://www.publications.parliament.uk/pa/jt201213/jtselect/jtpcbs/144/144.pdf>.