Concerns about Procyclicality
This fi nancial crisis highlighted the issue of procyclicality caused by Basel II. In other words, fi nancial institutions and some authorities feared that Basel II would further increase the swing of credit cycle. This mechanism is as follows. First, in a crisis, an increase in losses and in estimated risk amounts decreases fi nancial institutions ’ capital adequacy ratio. Then this decline pressures fi nancial institutions to recover the ratio by raising capital in the market or reducing the size of risk assets. Third, if fi nancial institu- tions heavily depend on the latter, this would reduce their lending and thus deepen the recession. Risk-sensitive Basel II surely has these types of char- acteristics. However, it has also already introduced various measures that contain this side - effect (see table 5.5 ).
There are, however, some limits to the effects of the above measures. For example, if we set the degree of stresses at the level of once every 25 years for assessing the required capital, then required capital might be too high to help the macroeconomy grow in a smooth way through lending. Besides, if we require fi nancial institutions to use very prescriptive scenarios for the stress testing purpose, this would surely discourage them from thinking of measures that would be really helpful to their risk management. Meanwhile, if the authorities react to the crisis only after it comes, they tend to be too late and fall off the cliff without doing anything under political pressure.
152 POST-CRISIS RISK MANAGEMENT Table 5.5 Various measures of Basel II to contain its procyclicality
P illar 1 P illar 2
IRB requires fi nancial institutions not to stop at point in time (PIT, or assigning ratings based on the different phase of business cycle of the time) but also to assume the element of through - the - cycle (TTC, or assigning ratings based on the bottom conditions over the business cycle).
Loss given default (LGD) estimates should be at the bottom of the cycle (so - called, downturn LGD).
Weighted average by probability of default may satisfy the above.
Op risk amounts in BIA and TSA are associated with gross profi ts, so are expected to be anticyclical (although this is not explicitly mentioned in Basel II.)
Risk parameters used for IRB (such as probability of default) should be long - term average.
Stress testing assuming two consecutive quarters with zero growth rates should be done.
Outcome of stress testing including Pillar 1 should be considered in fi nancial institutions ’ assessment of capital adequacy.
Stress testing assuming some structural changes in macroeconomy is expected to be done by fi nancial institutions.
The Need for a Credit - Cycle - Smoothing Macroprudential Policy
To overcome these problems, there is an idea of introducing the policy mea- sures that would impose uniform regulation on all fi nancial institutions, and that is fl exible and proactive enough to react to changes in the credit cycle in a proactive and pre - emptive way while leaving room for fi nancial institu- tions to improve their own risk management. This is a policy measure that assumes a feature of Pillar 2 (leaving more room for individual institutions ’ discretion) in terms of horizontal frequency but also assumes a feature of Pillar 1 (imposing uniform restrictions on all institutions) in terms of histori- cal frequency.
More specifi cally speaking, this should be a policy measure that would affect all institutions evenly to contain bubble factors in a pre - emptive way based on some targeting indicators that represent bubble factors. This should look quite similar to the conduct of monetary policy. Some might wonder whether the purpose of monetary policy already includes containing fi nancial bubbles, and whether this should be enough. However, the current fi nancial crisis was not necessarily judged as a result of the failure of mon- etary policy alone. In the US, there seems to be quite a few who argue that
this is the case. Of course, many experts actually argue that the long - contin- ued easy monetary policy in the era of Chairman Greenspan surely laid the ground for creating the fi nancial bubble. Still, generally speaking, it is well understood that the central bank, which is expected to stabilize the general price level, cannot easily manage asset prices. For example, it is hard for the central bank to tighten monetary policy only because of an increase in asset prices or a decline in risk premiums.
This is exactly the same dilemma that was faced by the BoJ during the bubble era. In other words, the central bank usually has only one policy tool, money - market operations, and even if we provide more policy objec- tives than policy tools, it is quite hard to attain them.
Policy Targets
We have recently heard many arguments for a so - called credit - cycle - smoothing, macroprudential policy from the authorities ’ side. However, it is not necessarily clear what kind of cycle the discussed policy is trying to manage. In this context, some authorities have argued for the introduction of “ dynamic provisioning, ” which has already been introduced in Spain as a prototype of a credit - cycle-smoothing, macroprudential policy. In this pro- visioning system, the level of provision against loans is decided considering the situation at the bottom of the cycle. So when we approach the peak of the cycle, the system requires fi nancial institutions to put aside more provisions for the coming bottom.
This dynamic provisioning, however, has a similar feature to the mon- etary policy because both polices try to manage the same ordinal business cycle. And the current fi nancial crisis, for example, seems to follow a quite different cycle from the ordinal business cycle represented by, say, real GDP growth rate. Besides, if you look at the macroeconomic situation of Spain, which introduced this system a few years ago, you will soon fi nd that its economy among the major European countries, was the most badly plagued by residential mortgage bubbles after the UK. So looking at the case of Spain, dynamic provisioning does not necessarily seem to be effective in containing bubble factors leading to the current crisis in a pre - emptive way.
This indicates that we should fi rst discuss clearly the type of cycle to be smoothed by the new macroprudential policy. As already discussed in chapter 4 , the basic purpose of macroprudential policy is to prevent shocks that menace the stability of the fi nancial system. History indicates that these shocks will not occur as frequently as an ordinal recession. Experts in fi nancial crises often mention a frequency of roughly once every 10 years.
In other words, previous crises to the current one are the events from the middle to the end of the 1990s, the Japanese banking crisis, the Asian and
154 POST-CRISIS RISK MANAGEMENT Russian crises, and the turmoil caused by the failure of LTCM. Before these crises was Black Monday during the end of 1980s, and before that was the Latin American debt problem during the end of 1970s and the US S & L cri- sis at the beginning of 1980s.
Strictly speaking, there were many other shocks, and also the degrees of shocks varied signifi cantly. So they do not necessarily indicate the length of the credit cycle precisely. They, however, surely provided some images (or a kind of common factor) of shocks to be managed by credit - cycle - smoothing, macroprudential policies. They are the creation of a fi nancial bubble and an excessive decline in risk premiums in the asset market, which were often followed by long - continued high profi tability of fi nancial institu- tions and benign macroeconomic conditions. Based on the experiences of the Asian crisis, for example, some Southeast Asian countries and the IMF have been developing early warning indicators that could indicate the signs of a currency crisis. We might adopt a similar approach to shocks that menace the stability of the fi nancial system.
If we seek simplicity, as in the case of the Taylor rule for monetary policy, we have to select a small number of representative indicators. The simpler rule is often expected to lead to a surer attainment of the policy, thanks to the transparency of the policy - deciding process that it entails.
For these indicators, for example, we might use the following indicators of risk appetite and credit, market, or liquidity risks, which have long been shown in the GFSR published by the IMF (see fi gures 5.9 , 5.10 , and 5.11 ).
It should be noted that these indicators have actually been used by interna- tional organizations and the authorities as major indicators of vulnerability of the fi nancial system.
Policy Tools
Because the central bank uses mainly money market operations to set mar- ket interest rates at the targeted level, the macroprudential authority needs policy tools to smooth the credit cycle. For this purpose, we might use the scaling factor that is currently used for the calculation of capital adequacy ratio by fi nancial institutions.
Basel II does not use the direct output from the IRB formula for capital calculation. It actually uses the output of the IRB formula multiplied by “ 1.06. ” This 1.06 is called a “ scaling factor, ” and was not necessarily set to calculate the risk amounts more precisely. The reason for this scaling factor lies in the authorities ’ wish to keep continuity of fi nancial institu- tions ’ capital amounts between Basel I and II. In other words, when the authorities tentatively calculated the Basel II number for fi nancial institu- tions, they found that the required capital under Basel II could decrease by
Figure 5.9 Developments of risk allowance
Source: IMF (2008a) based on source material from Emerging Portfolio Fund Research, Inc. Goldman Sachs, Merrill Lynch, State Street
⫺50
⫺40
⫺30
⫺20
⫺10 0 10 20
60 70 80 90 100 110 120
⫺0.6
⫺0.4
⫺0.2 0 0.2 0.4 0.6 0.8 1.0
0 2 4 6 8 10 12 Merrill Lynch Fund Manager Survey
(Net percent of investors reporting higher risk-taking than benchmark)
State Street Investor Confidence Index1
2001 03 05 07 1999 2001 03 05 07
2001 03 05 07 1990 93 96 99 2002 05
Total net inflows into emerging market bond and equity funds (In percent of assets
under management, 13-week moving average)
Goldman Sachs Risk Aversion Index
Note: Dashed lines are period averages. Vertical lines represent data as of the April 2008 GFSR.
1The estimated changes in relative risk tolerance of institutional investors from Froot and O’Connell (2003) are integrated to a level, scaled, and rebased so that 100 corresponds to the average level of the index in the year 2000.
Increased risk taking Increased
risk taking
Greater risk aversion
156 POST-CRISIS RISK MANAGEMENT
Figure 5.10 Developments of credit risk indicators
Source: IMF(2008a) based on source material from Merrill Lynch, Moodys, Bloomberg, L.P., MBA, Federal Reserve
50 150 250 350 450
10 15 20 25 30
0 2 4 6 8 10 12
3 4 5 6 7
2.5 3.5 4.5 5.5 6.5
Forecast default rate
Actual default rate 1998 2000
1998
1991 94 97 2000 03 06
2000 02 04 06 08 2004 05 06 07 08
02 04 06 08 1999 2001 03 05 07
Merrill Lynch Global Corporate Bond Index spread
(in basis points)
Moody’s speculative grade default rates: actual and 12-month forecast (in percent)
Share of CCC or lower-rated corporate securities in Merrill Lynch Global High-Yield Index (in percent)
Expected number of bank defaults given at least one bank default (among 15 selected banks)
Delinquency rate on consumer and mortgage loans2 (in percent)
Note: Dashed lines are period averages. Vertical lines represent data as of April 2008 GFSR.
1Measuring the largest probability of default among the sampled 15 banks each day.
230-, 60-, and 90-day delinquencies for residential and commercial mortgages, and credit card loans in the United States.
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
6 10 14 18 22
0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70
1 2 3 4 5 6 7 8
0 50 100 150 200 250 300 350 400
⫺2.0
⫺1.5
⫺1.0
⫺0.5 0 0.5 1.0 1.5 2.0 2.5 3.0
199496 982000 02 04 06 08
1999
Note: Dashed lines are period averages. Vertical lines represent data as of the April 2008 GFSR.
136-month rolling regressions of hedge fund performance versus real asset returns.
2Data represent the absolute value of the net position taken by noncommercial traders in 17 selected U.S. futures markets. High values are indicative of heavy speculative positioning across markets, either net-long or net-short.
3Represents an average z-score of the implied volatility derived from options from stock market indices, interest, and exchange rates. A value of 0 indicates the average implied volatility across asset classes is in line with the period average (from 12/31/98 where data are available). Values of
⫹/⫺1 indicate average implied volatility is one standard deviation above or below the period average.
4Based on the spread between yields on government securities and interbank rates, spread between term and overnight interbank rates, currency bid-ask spreads, and daily return-to-volume ratios of equity markets. A higher value indicates tighter market liquidity conditions.
1996 98 2000 02 04 06 2001 03 05 07
1993 95 97 99200103 05 07 1997 99 2001 03 05 07 1997992001 03 05 07 Hedge fund estimated
leverage1 (sum of betas across asset classes)
Absolute value of net noncommercial positions in US futures markets2 (in percent of open-interest across select futures markets, 30-day moving average)
Estimated common component in asset class returns (share of the variation in returns, 90-day moving average)
Composite Volatility Index3 (in standard deviations from the period average)
Funding and Market Liquidity Index4 (January 1996 ⫽ 100) World implied equity risk premia (in percent)
Figure 5.11 Developments of market and liquidity risk indicators
Source: IMF (2008a) based on source material from Credit Swiss, Tremont Index LLC, Bloomberg, L.P., JP Chase Morgan, IBES, Morgan Stanley Capital International
158 POST-CRISIS RISK MANAGEMENT 6 percent compared to that under Basel I, on average. Therefore, authorities that feared a sudden decline in capital argued for this measure to be taken. If the authorities were to have strong confi dence in the outcome of the Basel II formula, they might not care about the drop. Otherwise, however, authori- ties ’ fears of a signifi cant drop in required capital as a result of the change in regulation was understandable. And this concern was more than suffi ciently borne out by the current crisis.
So the scaling factor was invented to fi ll the gap between the numbers calculated by the IRB formula and the expectation of authorities, which have not yet been confi dent enough in the new formula. And I think we should use this factor as a candidate for the tool of a credit - cycle - smoothing, macroprudential policy (see fi gure 5.12 ). The most appealing merit of this tool is its simplicity. What we have to do is increase or decrease this scal- ing factor from the current 1.06 according to the level of selected reference indicators. Meanwhile, the most serious demerit is that this tool looks too simple for one to be sure whether it is really effective in smoothing the credit cycle.
For example, a large part of the fi nancial losses that occurred in the cur- rent crisis were not necessarily well covered by Pillar 1 of Basel II. Given the
Credit cycle Business cycle
106 13 12.3%
Scaling factor 1.06 Original risk assets
100 Risk assets
after adj. Capital Capital ad.ratio
137.8 10 7.3%
Scaling factor 1.06 Original risk assets
130 Risk assets
after adj. Capital Capital ad. ratio
Capital adequacy ratio Capital adequacy ratio
- Bubble collapse Increase in risk or risk asset ⫹ decrease in capital decline in capital adequacy ratio - Scaling factor is kept constant
Risk assets after adjustment ⫽ original risk assets ⫻ scaling factor
160 140
106
12.3%
15.0%
10.0%
5.0%
0.0%
120 100 80 60 40 20 0
13
160
140 137.8
7.3%
15.0%
10.0%
5.0%
0.0%
120 100 80 60 40 20 0
10
Figure 5.12(a) Macro impacts of capital adequacy ratio—the current situation
fi nancial institutions ’ diverse positions that are not covered well by Basel II, requesting them to put aside larger capital based on a uniform formula might cause some ineffi ciency in attaining the policy target. Besides, we might hear the complaints of globally active fi nancial institutions that “ the use of differ- ent scaling factors among different countries could impede this global - base business, ” or “ business in one country could be better than in another. ” This type of complaint, however, could be fl atly dismissed, because even now in the area of monetary policy, we see different levels of interest rates, about which no one has complained.
The Agency to Conduct the Policy
Finally, I would like to discuss the issue of the agency that would conduct this credit - cycle - smoothing macroprudential policy. This should naturally be the bank regulatory agency, and in Japan, for example, it is the Japanese FSA. In this case, however, these agencies would need to be equipped with more macroeconomic research capability. Because central banks are in general very good at doing this type of research, it may be better that the regulatory agencies seek more cooperation with the central banks in decid- ing macroprudential policy. Moreover, the regulatory agency that decides
150 18 12.0%
Scaling factor Original risk assets
100 Risk assets
after adj. Capital Capital ad.ratio
137.8 15 10.9%
Scaling factor Original risk assets
130 Risk assets
after adj. Capital Capital ad. ratio
Capital adequacy ratio Capital adequacy ratio
Bubble collapse Increase in risk or risk assets ⫹ decrease in capital Adjustment of scaling factor slight decline in capital adequacy ratio
Risk assets after adjustment ⫽ original risk assets ⫻ scaling factor
160 140
150 10.0%
5.0%
120 0.0%
100 80 60 40 20 0
18
160
140 137.8
10.9%
10.0%
5.0%
120 0.0%
100 80 60 40 20 0
15
1.5 1.06
Credit cycle Business cycle
Figure 5.12(b) Macro impacts of capital adequacy ratio—after the introduction of credit-cycle-smoothing, macroprudential policy
160 POST-CRISIS RISK MANAGEMENT macroprudential policy should be politically independent as the central banks are required to be. To implement this policy framework successfully, we need to establish a framework that can facilitate the effective conduct of prudential policy. Table 5.6 compares a credit - cycle - smoothing macropru- dential policy with monetary policy in some important aspects.
Table 5.6 Comparison of credit - cycle - smoothing macro prudential policy and monetary policy
M acroprudential p olicy M onetary p olicy
Policy target Prevention of a fi nancial system crisis
Price stability (smoothing the business cycle) Forward - looking
policy indicators
Risk premium information as variously observed in markets
Various leading indicators of business conditions (e.g. money supply, yield curve, and industrial production)
Policy tools Adjustments of the scaling factor that is currently used for the capital adequacy ratio
Money - market operations to set money - market rates at the targeted level
The agency to conduct the policy
Bank regulatory agency with political independence
Central bank
161
CHAPTER 6
Strategic Reaction to the Financial Crisis: The Japanese and Asian Perspective