Unconventional monetary policies after the financial crisis have been extensively discussed and analyzed, with the notable exception of the collateral policies at their core. The work by Kjell Nyborg is a rare exception. On the basis of an in-depth analysis of the collateral policies pursued by the European Central Bank in the wake of the financial crisis, he argues that those policies aggravated the sovereign debt crisis and put the survival of the euro at risk. Nyborg’s analysis of collateral policy is an important contribution to the literature, but his critique of the ECB’s crisis response is misguided. Moreover, his proposal to retailor haircuts on central bank liquidity in a manner that would deepen the ECB’s role in the fiscal disciplining of member states is dangerous: if adopted, it would be profoundly procyclical and destabilizing. Our new INET Working Paper analyses Kjell Nyborg’s work to identify a set of collateral policy principles that likely will ameliorate impending market liquidity crises as opposed to exacerbating them.
What is the role of collateral policy?
Central banking is widely seen as first and foremost a matter of using interest rates to achieve monetary policy goals. Central banks lend not merely at a cost, however, but always against securities. Borrowers pledge assets to access central bank funding. In this sense, the lending is secured. While secured lending exists in many forms, one feature is a constant: what is accepted as collateral varies significantly over time.
What the European Central Bank (ECB) accepted as collateral before and after the financial crisis were two altogether different things. When money and credit markets are liquid and well-functioning, central banks take a conservative approach, accepting only high-quality assets as collateral. In periods of market stress, on the other hand, they usually respond by accepting a wider range of assets as eligible collateral.
Overall, three core factors define the contours of central bank collateral policies. First, eligibility criteria set out what assets are eligible as collateral when banks seek access to central bank money; second, haircuts determine how much central bank money a bank will receive (as a percentage of the market value of the collateral) for different types of eligible collateral; and third, stipulations on counterparty access define what types of financial institutions the central bank is willing to provide lending to.
The haircut can be seen as the central banks’ insurance against liquidity risk. Should the borrower be unable to pay back the loan, the central bank can avoid a net loss, even if it has to sell the collateral at a price below the original market value. In this sense, haircuts are a risk management technique for central banks. Since the global financial crisis, it has become apparent, however, that haircuts have important implications for liquidity in the markets where collateral trades.
Since most collateral in the Eurozone is issued by Member States, the ECB’s collateral policy has significant impacts on liquidity and price in sovereign bond markets – that is, ECB’s collateral policy has significant, if deeply underappreciated, fiscal spillovers. In the early stages of the crisis, spreads between German bonds and ‘periphery’ Eurozone countries were amplified by the ECB’s collateral policy.
What is wrong with Nyborg’s critique of the ECB’s collateral policy?
The main message of Nyborg’s book is that the terms on which ECB supplied money for collateral during the crisis were “overly generous.” We argue that Nyborg’s characterization of the ECB’s collateral policy is highly misleading, however. In fact, haircuts were increased several times for assets with low credit ratings. Moreover, haircut differentials – the spread between haircuts on collateral assets with a high and a low credit rating – widened, hence producing a contractionary rather than an expansionary effect on collateral space.
Before October 2008, the ECB applied identical haircuts to all European government debt. There was no distinction between high- and low-quality collateral in this asset class. After the collapse of Lehman, the haircuts on highly-rated government debt remained at the same level, while all lower-rated government debt was assigned haircuts 5 percentage points higher.
Overall, three observations about changes made by the ECB to its haircut schedule stand out. First, haircuts for high-quality collateral were kept low throughout the crisis (and even declining for longer residual maturities). Second, for government bonds with a low credit rating, the opposite trend prevailed. Assets with a low rating faced a dramatic increase in haircuts, in the range of 550 to 850 basis points (depending on residual maturities), seen over the full period. Third, the haircut spread – between assets with a low (B to BBB-) and a high credit rating (A to AAA) – jumped 500 basis points in October 2008, was unaffected by the January 2011 revision, but increased again in October 2013, with 50 to 400 basis points (depending on residual maturities). For short residual maturities, the haircut spread jumped by 50 basis points (from 500 to 550), while for long residual maturities it increased by 400 basis points (from 500 to 900).
Over the full period, haircuts on government bonds with a low credit rating and residual maturity of less than one year were increased 12-fold, from 0.5% to 6%, whereas haircuts for the same class of government bonds with a residual maturity of 7-10 years nearly tripled, from 4.5 % to 13 %. These are hardly trivial increases.
We suggest that the haircut changes do not match their depiction by Nyborg as “overly generous”. On the contrary, it is difficult to imagine that haircut increases at this scale did not add to the already severe liquidity strains of troubled banks and governments in distressed countries.
What’s the way forward for collateral policy?
In Nyborg’s view, future instances of over-borrowing by banks and sovereigns ought to be prevented by using haircuts in a punitive manner. “The idea is simple,” he says, “if a debt-to-GDP ratio of no more than 60 percent is desired,” all you need to do is “increase haircuts progressively in the debt-to-GDP ratio beyond this.” The same mechanism can be established for fiscal deficits, such that haircuts are increased progressively as fiscal deficits exceed agreed thresholds. “My proposal works,” explains Nyborg, “by reducing the liquidity and value of a highly indebted country’s bonds.” By increasing borrowing costs, the “appetite” for borrowing in excess of agreed thresholds should recede.
We strongly oppose Nyborg’s proposal. If haircuts were proportional to fiscal deficits and public debt to GDP, collateral policies would exert a pro-cyclical and destabilizing influence not just on collateral markets, but on the financial systems they anchor. For a central bank to combat a market liquidity crisis effectively, it must decrease haircuts, not increase them – and more so for assets with low ratings, such that haircut differentials narrow rather than widen. This is essential to market liquidity. Incidentally, it is also by far the best risk management strategy, because the need for liquidity injections and asset purchases will be much more speedily satisfied with this policy mix.
Our Working paper argues that the ECB’s ambivalent strategy – of providing liquidity but raising haircuts on distressed assets – did not amount to “lending freely, against any and all collateral that is good in normal times”, as we believe Bagehot’s rule advises. By expanding collateral eligibility but raising haircuts and haircut differentials, the ECB was undermining the market liquidity it was trying to restore. To stop collateral valuation spirals, central banks must lower haircuts and haircut differentials – and suspend rather than follow the collateral valuation practices of financial markets.
Should we hold back for fears of moral hazard?
Countercyclical collateral policies are likely to be subjected to a standard criticism against measures that ease access to central bank liquidity. Such policies cause a moral hazard for both governments and banks, who would get access to funding on “subsidized” terms. Notably, lowering haircuts on central bank liquidity would amount to encouragement of “overborrowing” by banks that might in fact be insolvent and hence should not receive central bank funding.
Against such objections, we suggest that one must first acknowledge that liquidity provision and moral hazard are best dealt with separately. In much the same way as it would be “a terrible mistake”, in the words of Paul de Grauwe, if a central bank were to “abandon its role of lender of last resort in the banking sector because there is a risk of moral hazard”, we suggest that compromising collateral expansion by raising haircuts is a highly unfortunate conflation of strategies. The point is not that moral hazard problems should be ignored; only that they should be addressed differently. The solvency of individual banks is a task for micro-prudential regulation and supervision, not a concern that should be held against collateral policies designed to ease a market liquidity crisis.
Research has established that the moral hazard effects of liquidity provision are less detrimental than those resulting from direct recapitalizations of banks, the dominant response to bank insolvencies. The upshot is, we argue, that even if the main objective is to reduce moral hazard issues in banks to the largest possible extent, lowering haircuts likely will in fact contribute positively. By helping abate the liquidity crisis, incidences of banks becoming insolvent are reduced, and hence moral hazard in its severest form is minimized.