Comparison between Spain, Peru, and Colombia

The first important difference between the three systems is how are they activated or deactivated (see Table 4 [ PDF 16.7KB | 1 page ] for a full comparison). The Spanish and the Peruvian system are both rules-based, whereas the Colombian system is discretionary, in the sense that the “state” of accumulation mechanism is decided by the banking supervisor. The Colombian authorities have announced however that they will reform the system in a rules-based direction¹².

Another important difference is the variable chosen to calculate the amount of provisioning required. The Spanish system is based in credit whereas the Peruvian system is based on GDP. In Colombia (given its discretionary nature) so far there is no explicit variable used, although the authorities have announced that credit will be taken into account. These differences have important implications. First, since credit is a banking variable and institution specific, provisions under the Spanish system are based on the performance of each institution, whereas in the Peruvian system the activation or deactivation of the mechanism is common to the whole system.

Choosing a path which is common for all banks may have different implications for institutions depending on their strategy, their geographical or client specialisation, or their efficiency and profitability. Some may be gaining market share and others may be shrinking their presence in the system, but a system that is not institution-specific will tend to treat banks similarly (although the size, variation, or riskiness of their portfolio will imply of course differences in provisions even in system-based mechanisms). Second, the choice of GDP as an aggregate variable, as opposed to domestic demand, also poses questions. In fact, with a current account deficit, a domestic demand objective would ceteris paribus introduce more pressure than a GDP objective and would facilitate an automatic correction mechanism for the deficit. On the other hand, with a current account surplus, a GDP objective would be more demanding than a domestic demand objective, and the automatic correction mechanism would operate symmetrically. The differences however would depend crucially in the calibration and the choice of parameters in each case.

Finally, even if one opted for an aggregate variable, credit would seem more naturally linked to banking activity than GDP and it is directly linked to banks' behavior (whereas GDP is not a variable to which the banking system has any direct impact). On the other hand, in countries in the process of financial deepening (like Peru or Colombia) a high credit growth could not be a signal of excess in the financial sector, but a result of a healthy financial inclusion process. From this point of view the Peruvian system could be more tailored to the needs of emerging market economies (EMEs) (whereas in the case of Spain it is not an issue and high credit growth can be considered more as a prima facie indicator of financial excess than of financial development).

Another important difference lies on the sources of the data. Credit is a banking statistic and, therefore, much easier to use by the central bank and/or supervisor, whereas GDP is an estimate normally calculated by the statistics agency. Interestingly, Peru's choice of GDP coincides with its exceptional division of labor in terms of statistics. More specifically, GDP is calculated by the central bank and this is done monthly (which is also exceptional, and raises some reliability issues.)

One more relevant distinction is whether the provisioning mechanism is system-wide or individual. The Spanish system is individual, whereas the Peruvian and the Colombian provisioning mechanisms are systemic, although the latter has been announced to change to institution-specific in the next reform. Under the Spanish system, some banks may be increasing generic provisions while others are reducing them (for instance because the former are gaining and the latter are losing market share, or because there is an asymmetric negative [positive] shock in the latter [former] geographical area). The Peruvian system is activated for the system as a whole, although its impact on each institution depends on the riskiness of its portfolio. This implies that an institution losing market share, or with a more prudent lending policy, or which is experiencing a negative shock in its area of activity will be forced to provision above the normal level, simply because GDP is growing above a certain threshold.

The implications of the above are interesting form a competition point of view. On the one hand, one possible criticism of the Spanish system is that it could penalize institutions that are gaining market share because they are more efficient. On the other hand, the Peruvian system can be criticised for penalizing institutions that are more prudent. It also treats differently small and big institutions. The bigger (more systemic) a firm is, and the more diversified geographically, the less likely it is that you it face a rate of expansion very different from the average. In this regard, the Peruvian system could have a certain bias against smaller institutions.

Another difference lies in the fact that specific and generic provisions can net off. In the Spanish case this compensation is in principle automatic (although, as we are seeing now there is a certain room for discretion, both for the institution and for the supervisor, in the use of the generic provisioning in the downturn). The benchmark is to try to reach a constant total provisioning effort along the cycle¹³. Constant overall provisions along the cycle are arbitrary, but any other objective would probably be even more arbitrary. In the Peruvian case there is no benchmark. Banks are only required to provision more in the boom phase, without any real reference.

Finally, on the important issue of compatibility with international accounting standards (IAS), the Peruvian model seems even less compatible with IAS than the Spanish one. In fact, it is explicitly based on the expected loss model, which in IAS is only used for Off balance sheet items. The Spanish model, after the 2004 reform, tried to achieve a higher degree of compatibility with IAS. This is admittedly a secondary discussion at this stage, since IAS is expected to adopt (or at least admit) the expected loss model. . Constant overall provisions along the cycle are arbitrary, but any other objective would probably be even more arbitrary. In the Peruvian case there is no benchmark. Banks are only required to provision more in the boom phase, without any real reference.

One interesting proposal is that of Restoy and Roldan (2009), formulated similarly also in Turner (2009), whereby, a transparent distinction between regular profits and distributable profits in public financial statements should be made, which would imply that, without undermining the discipline of accounting standards, the regulator could use an anti-cyclical tool along the cycle to smooth reserves. But these proposals belong to the family of capital rather than provisions (see the discussion in Section 2.2 above).

Accounting principles—including the chosen provisioning model—would govern, as at present, how the regular profit and loss account is prepared. Regulators would however set clear rules establishing which portion of income could actually be paid out as dividends. The difference between those two concepts of profit would therefore be a set of publicly-reported compulsory reserves that would not interfere with the determination of the regular profit and loss account. That set could include a (through-the-cycle) reserve that would be earmarked against future losses and crafted along the lines of the Spanish dynamic provision.

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

Comment(s)

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