Existing Experiences

3.1 Spain

The introduction of dynamic provisioning in Spain should be seen in the context of the profound impact of the euro adoption in the Spanish economy. Traditionally this economy was characterized by a certain propensity to instability, which implied ample cyclical swings, difficulties in maintaining price stability, chronic balance of payments problems, and periodic currency crises to restore competitiveness levels.

The Exchange Rate Mechanism (ERM) crisis of 1992–93 was an example of this type of recurrent crisis. After the crisis (and the restoration of the external balance) economic policies were oriented in the mid-1990s towards fulfilling the convergence criteria for euro adoption. This strategy was based on two pillars: (i) reducing the inflation differential vis-à-vis Germany, and (ii) maintaining a sound fiscal policy. It was assumed by the authorities that the remaining convergence variables (long-term interest rates and exchange rates) would behave consistently with nominal stability and converge towards the reference values.

This strategy was successful and Spain joined the European Monetary Union (EMU) from the start, which implied the irrevocable fixing of the parities on 1 January 1999. As a result of this process, the Spanish economy benefited from a significant reduction of risk premia, in particular those related to inflation and currency risk. The real long-term interest rate (defined as the difference between nominal rates and contemporary inflation) moved from a level of 4–5% in the 1980s and first half of the 1990s to around zero in the aftermath of monetary union.

The expansionary impact of the reduction in real interest rates on the Spanish economy was very significant (see Figure 2 [ PDF 24.7KB | 1 page ]). Domestic credit growth, which ranged between 5–10% in the mid-1990s, accelerated to rates above 15% in 1998–2000. House prices increased at an annual rate of around 10% in the same period. Inflation accelerated from 1.9% in 1997 to 2.2% in 1999 (3.5% in 2000). The differential in domestic demand growth between Spain and Germany in the early years of monetary union was around 3.6 percentage points, mostly related to the gains from price stability and policy credibility for Spain (and in general peripheral countries), whereas Germany, where credibility was already high, did not experience a similar effect.

The European Central Bank kept interest rates in the late 1990s around 4.0%, a level which was consistent with average conditions in the eurozone, but which implied very lax monetary conditions for the Spanish economy. This expansionary impact was compounded by the depreciation of the euro vis-à-vis the United States dollar in these years. It is interesting to note that the situation of Spain (and other peripheral countries) in the first years of EMU presented some similarity with what was anticipated a few years earlier by Allan Walters, economic advisor of the British Prime Minister Margaret Thatcher in the late 1980s. According to the “Walters' Critique”, ERM membership (and by the same token EMU) would imply too lax [tight] monetary policies for countries with above-average [below-average] inflation rates, which would tend to perpetuate (or even amplify) inflation differentials⁵. It is true that convergence criteria for euro adoption limited the extent of these initial discrepancies. But a different mechanism—the asymmetric shock of the reduction of risk premia for peripheral countries—had a similar effect⁶.

In the early years of the 2000s, therefore, the Spanish authorities saw with increasing anxiety the combination of high credit growth, inflation differentials with the Eurozone average, loss of competitiveness, and widening current account deficits. Monetary policy and the nominal exchange rate were no longer available as policy instruments. In this context, dynamic provisions (or statistical provisions, according to the denomination they received at the time) were seen as an instrument with a double objective: (i) to contain credit growth, by increasing the cost (in terms of provisioning effort) of the granting of new credit, and (ii) to protect Spanish banking institutions from future losses as a consequence of the relaxation of lending standards typical of the boom phase. While the first objective was probably more important at the time of adoption of this system, the results—as we will see below—were much more satisfactory in terms of the second objective.

Dynamic or statistical provisioning was therefore a truly macroprudential tool, in the sense that a prudential instrument (provisions) was used to achieve a systemic or macroeconomic goal (limiting credit growth). As concerns the second objective, it was mostly addressed at ensuring an adequate protection to individual institutions (and therefore could be seen as a microprudential tool), but to the extent that excessive risk assumption was partly a result of herd behavior and collective myopia by credit institutions, it had also a certain macroprudential aim.

3.1.1 How was the system expected to work?

As can be seen in Figure 3 [ PDF 15.3KB | 1 page ], under a normal provisioning system provisions are a function of contemporary nonperforming loans (NPLs), although this may be smoothed by the possibility of using “generic” provisions based on the credit stock. In the upturn, when gross domestic product (GDP) grows above potential, credit growth also accelerates. Since business conditions are favorable, collateral prices are increasing and optimism is pervasive, debtors have in general no problem in servicing the debt. The low provisioning effort fuels low risk aversion and credit growth, thus feeding back economic growth. In the downturn the opposite spiral operates: the difficult economic environment is accompanied by high NPLs, which require a bigger provisioning effort. This in turn decreases risk appetite and feeds credit contraction. Hence the pro-cyclical pattern of normal provisions.

The objective of dynamic provisions is to smooth the provisioning effort along the cycle, as shown in Figure 4 [ PDF 15.3KB | 1 page ]. How much? This is an open question. While the idea is to avoid the pro-cyclical effect of the normal system, a regulator would hardly aim at an opposite pattern of provisions (i.e. increase in the good times and decrease in the bad times), since risk is cyclical and this reality should be reflected in provisions. A reference would be to try to obtain a flat provisioning effort along the cycle in terms of the ratio of provisions to credit. The chart above—which should be taken only as a reference—depicts provisions with a smoothed pro-cyclical pattern, which was more or less what was aimed at in Spain.

3.1.2 How did the system really work?

As can be seen in Figure 5 [ PDF 16.8KB | 1 page ], credit growth stabilized at around 15% annually after the introduction of dynamic provisioning in 2000, and decreased slightly between 2001 and 2004. It is difficult to assess however to what extent this was related to the new provisioning system. Most probably the impact of the burst of the dotcom bubble was more relevant in this period. After 2004, however—coinciding with a reform of the provisioning system—credit accelerated sharply and reached rates of growth above 25% in 2006. The impact of the global financial crisis since mid-2007 implied a sharp contraction of both credit and GDP.

To understand these patterns it is useful to recall how the system was designed and how it was reformed in 2004.

Initially the system reform of 2000 was based on three types of provisions: specific, generic (both already existing), and statistical (introduced in 2000). Specific provisions depended on current bad loans, generic provisions were 1% of the credit stock, and statistical provisions were designed to offset specific provisions and depended on credit growth.

This mechanism was criticized on several grounds: First, by international accounting bodies, which argued that it implied profit smoothing along the cycle which masked the real situation of the banks. Second, Spanish financial institutions complained about being subject to higher provisioning requirements than their competitors, which was considered an important competitive disadvantage in the single European market for financial services.

By 2004 there was a sense that these provisions were excessive. By that time, they reached a level of more than 2.5% of credit (of which less than 0.5% was specific provisions, i.e., related to bad loans), as can be seen in Figure 6 [ PDF 17.2KB | 1 page ]. Furthermore, the coverage of provisions over bad loans reached nearly 500% (Figure 7 [ PDF 17.2KB | 1 page ]).

For this reason, and also to counteract the criticism by accountants, the system was reformed. The changes basically implied the integration of the generic and the statistical provisions and the establishment of limits to the accumulated fund. According to the formula:

Generic provisions = α Δ Credit + β Credit – Specific provisions

Where 0 ≤ α ≤ 2.5%

and 0 ≤ β ≤ 1.64%

Δ stands for change

The coefficients of the different types of assets were as shown in Table 1 [ PDF 40.7KB | 1 page ].

The limits of the Generic Fund, which was the result of accumulated provisions, were set between 0.33% and 1.25% of the alpha. Since a number of institutions were at that time at or very close to the limit, this implied the liberation of €14 billion from the Generic Fund. These “liberated” provisions were, however, not distributed, but consolidated as reserves.⁷ In the subsequent quarters, as more institutions reached the upper limit of the Generic Fund, and credit accelerated over 25% annually, the ratio of provisions to credit went down, from 2.5% in 2004 to 2.2% in 2007.

To a certain extent, the 2004 reform can be assessed in retrospect as a “lack of faith” in the system. It was innovative, with no precedent and no similar system in any other country, contested by the banks and by the international accounting bodies. The Spanish authorities started wondering whether the system could be unsustainable and whether there would be limits in the accumulation process. Had the authorities knew the magnitude of the shock that was incubating—and that would erupt in 2007—they would probably not have changed it, or at least not set the limits so close to the then prevailing levels.

The events since 2007 show a dramatic turn. GDP and credit dropped rapidly, NPLs started rising swiftly, and specific provisions grew fivefold from the summer of 2007 to the spring of 2009. Generic provisions also decreased very quickly, but not sufficiently to compensate for the increase in specific provisions, so that total provisions to credit in early 2009 exceeded the maximum reached in 2004, also due to the rapidly decreasing credit growth as the global crisis hit Spain. This limited use of generic provisions in the downturn can be explained by the prudence of financial institutions (which were aware that the worst was yet to come) and the authorities´ guidelines (aimed at limiting profit distribution when the impact of the shock was starting).⁸

Some preliminary lessons emerge from the Spanish case. First, dynamic provisions helped creating a cushion in the good times, but hardly discouraged credit growth or rises in house prices. When the size of the boom is big enough, the impact of an additional provision on credit supply is marginal. Second, the Spanish system—although being rule-based—allows for some discretion. Despite the fact that Spain has probably one of the most complete and reliable data set of credit and NPLs based on a long standing Credit Registry, the initial difficulty in calibrating the cycle “ex ante” led to doubts about the reliability of the estimates. This explains why the rules were changed in the middle of the game. Third, the treatment of off-balance sheet entities (OBSEs) also played an important role in the system. The issuance of covered bonds and securitization did not “save” capital for the institutions. The joint effect of dynamic provisions and treatment of OBSEs explains why the Spanish banking system confronted the crisis in a better initial situation than others in Europe.

3.2 Colombia

In 2007 Colombia adopted a model of dynamic provision for commercial and consumer loans, which represent about 90% of the total outstanding loan portfolio. The banking regulator implemented reference models for commercial and consumption credit risk. Although each bank could use its own credit risk model, which must be approved by the regulator, at present all banks are using the reference model.

The reference model established three types of provisions which are tax deducible: Individual, countercyclical, and generic provisions. Individual provisions reflect the characteristic risk of every borrower and every type of loan, and can only be used if the loan becomes nonperforming. Countercyclical provisions seek to cover changes in borrower's credit risk due to changes in the economic cycle and have the same characteristics as individual provisions. With the present regulation it is not easy to distinguish between individual and countercyclical provisions as both go to the same balance account. Finally, generic provisions are at least 1% of the total loan portfolio and this type of provisions can be used to meet countercyclical provision regulation requirements.

As one can see in Figure 8 [ PDF 18.8KB | 1 page ] once the model of countercyclical provisions was implemented there was a dramatic fall in generic provisions. In fact, the system was criticized since the rise in the increase in the individual provisions, through countercyclical, was compensated in part by the reduction in generic provisions.

3.2.1 How is the System Designed?

The regulator, using historical data, calculates two risk scenarios, A and B (where B is a riskier scenario). The outputs of this calculation are two default probability matrixes which contain default probabilities for every type of credit and borrower. Provisions are the result of:

P = OVL*DP*LOD
Where:
OVL = Outstanding Value of the Loan
DP = Default Probability
LOD = Lost Once Defaulted

Every year the regulator decides which matrix will be used to compute individual provisions. During years of high credit and economic growth, matrix A is used to determine the accumulation of individual provisions and matrix B will be used to calculate the riskier scenario provisions, so that countercyclical provisions will be the difference between the riskier scenario provisions and the individual provisions. During years of low growth matrix A will be used to calculate individual provisions and there will be no accumulation of countercyclical provisions.

The regulator can also exercise discretion in determining when banks can use countercyclical provisions to compensate the increase in individual provisions during an economic downturn. Once the regulator declares the change of state all banks can use countercyclical provisions, regardless of the financial health of individual institutions.

Such a discretionary model—with no rules determining the change of state (and thus of provisioning)—created a great uncertainty, which has led the Colombian regulator to announce a revision of the system in a direction that would make it more rules-based and more similar to the Spanish system.⁹

Although these changes have not yet been detailed, the new system will be based on the following principles. First, rules will be used instead of regulator's discretion in declaring the change of state. Second, the change of state will not be announced for the system as a whole but will be determined individually for each institution according to rules to be established. Third, clearer rules on the accumulation and drawing down of countercyclical provisions will be adopted. Fourth, dynamic provisioning will be used as generic ones—and not individual—in the downturn. Fifth, there will be differentiation between institutions for the building-up of the countercyclical provisions, so that banks with higher credit growth rates will accumulate higher countercyclical provisions.

3.3 Peru

After the emerging markets crisis of the late 1990s, which led to a credit crunch in Peru until 2003, the Peruvian economy began a period of fast economic expansion. Although initially fueled by exports, this boom was later related to private investment and consumption fueled by a credit boom, as shown in Figure 9 [ PDF 16.7KB | 1 page ].

Credit to all types of clients showed significant growth rates in this period, in particular that to higher-risk agents such as micro-firms and consumers (over 30% year on year [yoy] as shown in Figure 10 [ PDF 24.6KB | 1 page ]). In this context, and even though credit over GDP was still relatively low (compared to other countries in the region); concerns grew on whether these rates could be unsustainable or could partly be related to a less rigorous banks risk assessment. This is when the idea of introducing business cycle-adjusted provisions as a tool both to moderate credit expansion and to generate buffer provisions should the cycle turn down became attractive to policy makers. In 2008, GDP grew 9.8% and credit 36%, changes in generic provisions were introduced. This change partly turned voluntary provisions banks had accumulated in the last two years into permanent provisions. Figure 11 [ PDF 24.6KB | 1 page ] and Figure 12 [ PDF 23.1KB | 1 page ] show the evolution of credit, total, and voluntary provisions.

Before going into how these cyclical provisions were implemented, it is useful to describe first how the provisioning system worked before this. In Peru, loans are classified according to the type of debtor, which can be commercial, micro-firms, consumers, or mortgage.

Since December 2008, the generic rate depends on the type of debtor and is not homogeneous anymore: 0.7% in the case of all commercial and mortgage “normal” loans, and 1% in the case of all micro-firms and consumers “normal” loans. With this change, generic rates now penalize more those (riskier) loans that have historically shown a higher non-performance. Secondly, cyclical provisioning was introduced, primarily aiming at moderating credit growth rates and reducing the probability of eventual consumer over-indebtedness.

3.3.1 How is the system designed?

The Peruvian financial supervisor/regulator (Superintendencia de Banca, Seguros y AFP [SBS]) has set a rule based on GDP growth. In this way, cyclical provisioning is activated when the rate of growth of GDP exceeds a certain threshold (in boom periods), which is related to an estimation of potential output growth. Figure 13 [ PDF 35.7KB | 3 pages ], as well as the three graphs, illustrates the rule.

These cyclical provisions are part of generic provisions. When cyclical provisioning is activated, generic provision charges increase (although this depends on the type of debtor). Table 2 [ PDF 20.1KB | 1 page ] shows how these charges change.

Rates on additional generic provisions were based on data from the last episode of financial crisis in the late 90s crisis. They were therefore calibrated for a stress situation. In times of economic slowdown, on the other hand, the rule is deactivated and generic rates are reduced. Diagram 3 and the two graphs summarize the functioning in stress situations.

Figure 14: Cyclical Provisioning Deactivation [ PDF 26.4KB | 2 pages ]

It should be noted, however, that although additional accumulated generic provisions cannot be directly allocated to profits, the possibility of using them to cover other required provisions reduces the provisioning effort banks need to make during the cycle's downturn. Thus, they indirectly benefit banks profits in bad times, smoothing them over the cycle.

Why is the rule based on GDP? Why not credit (a banking system variable)? According to the SBS, it is assumed that GDP precedes credit. In this sense, credit growth would not be a good variable to anticipate future bank losses and thus reduces the desirability to relate provisions to credit growth.

Another issue to consider is that a GDP based-rule is systemic. This means that its activation does not depend on a bank's behavior, but on the economy's (system) as a whole.

For this reason, the effect could be asymmetric on banks: it could be the case that a more prudent bank would have to increase generic provisions.¹⁰

Regulations state that since January 2010 instead of classifying loans into four groups (by debtor type), financial institutions will have to classify them into eight groups. This should increase the homogeneity of loans in each credit type, which favors the accuracy of the assessment that can be made and therefore enhances risk management. Provisioning charges will then be as shown on Table 3 [ PDF 15.5KB | 1 page ].

Cyclical provisioning¹¹ was activated in December 2008, at the very same time it was implemented. However, given the fast deceleration the Peruvian economy has experienced since the fourth quarter of 2008, it is expected to be deactivated by rule B2 in the coming one or two months as can be seen in Figure 15 [ PDF 17.8KB | 1 page ] and Figure 16 [ PDF 17.8KB | 1 page ].

Download this Paper [ PDF 259.4KB| 32 pages ].

Comment(s)

There are [0] comment(s) for this entry. Post a comment.