Calculation and Allocation of FISIM

Calculation and allocation of FISIM

February 2005

Discussion on methodological

problems and solutions

This document is under constant changes, and aims in summarising the electronic discussion between MemberStates on calculation and allocation of FISIM

UPDATED on 12 May 2005

Table of content

General points

a.Calculation of internal reference rate

b.Calculation of externalreference rate

c.FISIM allocation among the domestic user sectors......

d.Allocation of FISIM exports......

e.Allocation of FISIM imports

f.Calculation of FISIM intermediate and final consumption by households

g.Calculation of FISIM at constant prices

General points

Belgium

Backdata on FISIM

The backdata issue appeared during the workshop last year. Although Eurostat was ready to allow us a delay to produce those backdata, internal requirements (quarterly accounts, accounts for general government) have put some pressure to produce those data more rapidly than expected.

The nature of FISIM margins applied on stocks of loans and deposits make it impossible to use a simple model to derive backdata (due to the high volatility of results). The best way to estimate FISIM production and allocation is to use the same methodology as described in the regulation448/98. That means to build two tables, one with the outstanding amounts of loans and deposits by counterpart sectors, one with the accrued interest paid and received by counterpart sectors. The reference rates (internal and external) must be estimated because it is not possible to calculate implicit rate in the past (due to change in the reporting scheme of banks). Work is underway, the tables of stocks and accrued interest are nearly completed for years 1985-1994. The backdata should be available during the second quarter of this year. The estimation will only cover FISIM produce by s122, but a estimation based on growth rate of s122 could be done for production of FISIM by s123 if required, the allocation will be based on constant keys.

Eurostat: We welcome the results of calculating backdata. We agree that the main focus should be on estimating FISIM produced by s122, and then use growth rate of s122 to extrapolate production of FISIM by s123. We agree because the main sector generating FISIM output is S.122 sector and the importance of sector S.123 is rather small.

Cyprus

Any specific methodological problems or questions concerning FISIM allocation will be communicated to you in the near future, when the calculation will reach at a more advanced level. Due to delays in data collection/processing and subsequently to calculations of reference rates, the whole work has been delayed. According to current indications, the work completion (at least for current prices) is expected to be finished by June-July 2005.

a.Calculation of internal reference rate

Netherlands

The calculation of the Internal Reference Rate (IRR): The regulation states that this IRR must be calculated as interest on loans between S.122 and S.123 divided by the outstanding loans between S.122 and S.123. Because the External Reference Rate (ERR) is calculated on deposits as well as loans, we also used the deposits in our calculations of the IRR. Furthermore we used the outstanding internal loans and deposits within and between S.122 and S.123. In our opinion the internal flows of S.122 give a better approach of an interbank or risk free rate. This method has been confirmed by the NAWG of November 2004.

Eurostat: There is a substantial difference on the quality and data availability for inter-bank relationships for the resident and non-resident units. There is good quality data available on inter-bank relationships of the resident financial units (S.122 and S.123). The matrices compiled for interests and stocks, which cover of all inter-bank relationships for all units within S.122, all units within S.123, and between all units S.122 and S.123, should be balanced. It means that the asset sides (loans) of both matrices must be equal liability sides (deposits). This is why the Regulation asks to calculate the IIR as a ratio of interests to stocks on loans. You may calculate the IIR based on sum loans and deposits, but the results will be the same as calculating based on loans only (assuming that the matrices are balanced - loans equal deposits).

The inter-bank relationship between resident FIs and non-resident FIs is somewhat different. There is no balance between loans and deposit due to variety of reasons, mainly due to coverage of information from non-residents FIs (asymmetries). Therefore, the Regulation requires the ERR to be calculated as a ratio based on both loans and deposits. It is not correct that the above issue has been confirmed at the NAWG in November 2004.

Netherlands (2 Mar 05)

The second remark concerns the calculation of the IRR. We calculate the IRR on interbank loans as well as on interbank deposits. In our opinion financial institutions do have both loans (AF.4) and deposits (AF.2) on the asset side of their balance sheet. On the liability side they also have both loans and deposits (S.122 only). We agree with you that all interbank relationships should be balanced. That means that the interbank relationships on loans (AF.4) should be balanced as well as interbank relationships on deposits (AF.2).

The interest rate on (interbank) deposits (a Euribor-rate for example) doesn’t equal the interest rate of other (long-term) interbank loans. The difference arises for example due to the maturity structure. That’s why we calculate de IRR on both interbank deposits and loans which results in a weighted average of the interbank interest rates on deposits (AF.2) and loans (AF.4). Calculation based on only interbank loans would in general over-estimate the reference rate and the IRR will in general be higher than the ERR, which is calculated on both loans and deposits.

France (7 Mar 05):Some numerical illustration can better illustrate the problem raised by out Dutch colleague. Let’s illustrate an example of inter-bank relationship on loans and deposits in the country’s economy that consists of only two banks. For clarity, it might be appropriate to avoid naming the financial instruments loans and deposits in the inter-banks transactions.

Say, Bank A makes a transfer of financial asset of 90 (stock value) to Bank B.

This transaction must be recorded in the accounts of both banks, i.e., in the account of Bank A this amount of 90 must be recorded on assets side, and in the accounts of Bank B on the liabilities side.

In addition, the Bank B makes a transfer of financial asset to Bank A of 10 (stock value). (See the illustration of recording the above mentioned inter-bank transactions below)

Concerning the NL point on the published rates, EURIBOR (mid-term) rate and long-term (probably yield of bank bond). It should be noted that in the inter-bank relationship the majority of transactions in loans are short term (see ESA95, para 5.75)

In theory, the weighted average of published rates (corresponding average weighted to the stocks of loans and deposits of FIBOR -short term, EURIBOR – mid term, and bank bonds yield - long term,) should be comparable with the calculated Internal Reference Rate.

For example, say that the asset of Banks A of 90 is a short-term loan (recorded also as liability in Bank B), and the asset of Banks B of 10 is a mid-term loan (recorded also as liability in Bank A). Therefore, the rate (that in theory would be) comparable to the Internal Reference Rate would be calculated as average of 90% weight of FIBOR rate, and 10% of EURIBOR rate.

Netherlands

At the moment the results of our first calculations are being analysed. This is because of the major revision of our National Accounts with significant adjustments in our financial accounts (relevant balance sheets) and the little experience we have with FISIM. In particular for the user sectors there can be a high variability of results. That is due to the fact that the IRR and ERR are highly dependent on a short term interest rate for deposits. When this rate changes quickly, also the difference between a more fixed rate on loans and the IRR or ERR can move quickly, and therefore lead to a high variability of outcomes.

Eurostat: It sounds possible if there is large proportion of instruments (deposits as mentioned) with the short maturity. Would you please confirm calculating the reference rates using average stocks of opening and closing period (i.e. average over one year).

Netherlands (2 Mar 05)

We confirm that we use average stocks for calculating the IRR, ERR and FISIM. Despite of this, the results during the period 2001-2003 show a high volatility because of the quickly decreasing reference rates. The results of Belgium also show this kind of decreasing reference rates.

b.Calculation of external reference rate

(see also point a above)

Malta

Result not satisfactory seems too high, I am in contact with the Central Bank to review the figures used.

Poland

The breakdown of the external reference rate (onto euros, non-euro and extra-EU) is still a problem.

Portugal

Non-residents Fis : On the FISIM Eurostat tables report, the Table 3 -"FISIM output of S.122 + S.123 : Exports" preview the allocation of FISIM to the Non-residents FIs. This treatment seems opposite to the Table 2 Eurostat report ( allocation of FISIM imports). In the other hand, we looked to a French document that consider the non-residents financials institutions only to the calculation of the external reference rate. It seems to us this is the correct treatment.

Eurostat: We do not support this split of external reference rate (ERR) into two ERRs:

• one to be used for compiling margins on FISIM exports, which would be calculated as ratio of interest received from non-resident FIs on assets of resident FIs to the corresponding stocks of assets
• one to be used for compiling margins on FISIM imports, which would be calculated as ratio of interest paid to non-resident FIs on liabilities of resident FIs to the corresponding stocks of liabilities

The Regulation is still obligatory;however, we would welcome countries comments on that subject, which has been presented at the FISIM workshop in June 2004. (see the minutes and the useful information on circa

documents “FISIM 07” and “FISIM 12”)

c.FISIM allocation among the domestic user sectors

c.1. Treatment of S.121 sector

Belgium

s121 is excluded from the calculation of FISIM, and its output is measured as the sum of costs (CR448/98). This production is supposed to be entirely allocated to the intermediate consumption of s122 and s123.

Does that mean that, compared with the present situation (FISIM not allocated), the intermediate consumption of s122 and s123 will be increased by an imputed consumption of monetary intermediation services (65A01) produced by s121? This consumption of monetary intermediation services produced by s121 as it is not explicitly priced to s122 and s123 should be added to the current intermediate consumption of s122 and s123 calculated from profit and loss account items, increasing the current P2 of s122 and s123.

Another problem is that the production of s121 is partly a market production, thus the production of monetary intermediation services (based on costs) must be corrected by the amount of market production before being imputed to s122 and s123. The market production being consumed (intermediate consumption) mostly by units of s11.

Eurostat:

Treatment of S.121 within FISIM

The long discussions we have had in the EU on measuring the FISIM output and identifying producer sectors (generating FISIM output), which can then be allocated, resulted in the Regulation 448/98 you mentioned above. This Regulation, Annex I, para: 3.63, makes it clear (by convention) that “The central bank must not be included in the calculation of FISIM: its output is measured as the sum of cost”. It also says in para: 3.70k that “by convention the central bank output should be entirely allocated to the intermediate consumption of other financial intermediaries (sub-sectors S122-S123)”.

In other words, you don’t have to think about any imputation method when calculating FISIM allocation at all.

General Treatment of S.121 (FISIM and non-FISIM) output

In the context of S.121 treatment outside FISIM, S.121 is involved in the NACE J. Although S.121 is a market unit, by convention (according to the Regulation) we measure its outputas a sum of costs (P1=D.1+K.1+P.2+D.29-D.39).

In other words, total S.121 (FISIM and non-FISIM) output should be included in NACE J (65.11) as mentioned.

Carmen Ortega of Spain provided a nice numerical example of S.121 treatment in the new recording system (FISIM allocated). On page 5 of the paper attached (S121fisimP1.pdf), the following changes are shown:

1. The total output of central bank S.121, which consists of FISIM and non-FISIM (fee + commission) charges, is calculated as a sum of costs (CR 448/98, annex I, para 3.63c). This amount is then recorded in the production account in output (P1), which in your example is

P1 = 120 (P.2) + 180 (D.1) + 1 (D.29 less D.39) + 15 (K.1)

= 316

2. In the old recording systems (FISIM not allocated) interest received (D.41) consisted of all interest received including FISIM charges, whereas interest paid did not included FISIM. In the new system (FISIM allocated), interest received must be recorded as free of FISIM. Therefore, in the allocation of primary income account, interest received is reduced by the amount of FISIM.

P1 (FISIM + non-FISIM) = 316

P1 (non-FISIM, = 20

measured on the bases of

fees and commissions)

Old D.41 received = 2100

NewD.41 received (free from FISIM) = 2100 – 296 (P.1 FISIM)

= 1804

Belgium

I've got a problem with that because if, as explained above, we increase the P2 of s122 and s123 with the amount of P1 produced by s121 (estimated based on costs) we will change the net borrowing/net lending of s122 and s123.This situation is annoying since nothing change in the financial assets and liabilities of those two sectors. Till now the allocation of FISIM was neutral on B9, but the treatment of P1 of s121 as described in the regulation 448/98 is not neutral. Generally, production estimated from costs are applied for non-market units (like s13, and s15) but those sectors are supposed to consume their own production. Here, with s121 the consumption is done by another sector...

Eurostat:

FISIM changes values from interest (D.41) to services (P.2) and therefore it must be neutral for net lending / borrowing (B.9) everywhere. Please see the numerical illustration above and in the file “S121fisimP1.pdf”.

c.2. Sectors S.122 and S.123 as FISIM consumers

Belgium

The second point concerns the consumption of FISIM by units of s123 which are not producers of FISIM. Following the workshop of last year, it appears that Belgium was the only member state for which a consumption of FISIM was recorded for s123, corresponding to consumption by units such as financial holdings, securities and derivatives dealers and erroneously by collective undertakings investments funds (which must be excluded of FISIM allocation). After consultation of Eurostat, it was agreed that s123 would not be considered as producer and consumer of FISIM at the same time, which I recognise, is wiser to make easier the understanding of the recording of allocation of FISIM for the user of ESA95. I am still confronted with two possibilities at that stage: to simply erase the previously estimated consumption of FISIM of s123 (option which I have chosen, maybe a bit rapidly) or report consumption of FISIM by s123 in another sector (for example s11). The intermediate consumption of FISIM by s123 is estimated between 79 and 220 millions Euro depending of the year. Do Eurostat have any preferences?

Eurostat:It seems that some units of S.123 sector that are not FISIM producersbut are FISIM consumershave rather small impact on total intermediate consumption or GDP (between 0.7% and 2% of S.12 intermediate consumption and much less than 0.001% of total intermediate consumption). However, looking from a different FISIM perspective, the value is quite important (between 1% and 3% of FISIM output and between 1.5% and 4.5% FISIM intermediate consumption).

In cases where the same unit (say belonging to S.123 sector) is FISIM consumer as well as FISIM producer, its consolidated (netted-off) value should be recorded as net FISIM producers or net FISIM intermediate consumerdepending on the sign of the result. For example, if the consolidated FISIM value of the same unit is positive, it should be recorded as net FISIM producer in the FISIM output.

However, form the above description provided, some units of S.123 sector are only FISIM intermediate consumers, and do not generate FISIM at all. This specific case is described below (see page 12).

Belgium

Another problem has arisen, due to differences in sectorisation for s124 in financial accounts and non-financial accounts, the consumption of FISIM by s124 is greater than interest recorded in the non-financial accounts. In order to circumvent this problem, may I record consumption of FISIM of s124 with consumption of FISIM of s11 (implying no record of FISIM in the account of s124) until the sectorisation for s124 has been solved? Consumption of FISIM by s124 is estimated between 13 and 69 millions Euro depending of the year (between 0,6 and 2,4% of the intermediate consumption for s11).

Eurostat:We do not have strong view on that. It seems to us that the shares you mentionedare percentages of the FISIM intermediate consumption for s11, and therefore consumption of FISIM by S.124 would be much less than 0.0003 of the intermediate consumption for s11.

Netherlands

The user sectors of FISIM: In regulation 448/98, and also in the questionnaire we had to respond for the seminar in June last year, only the sectors S.11, S.124, S.125, S.13, S.14, S.15 and S.2 (export) are mentioned as user sectors of FISIM. In our opinion there are more user sectors. First of all, we think that all sectors, the producers of FISIM included, can have loans and hold deposits. Therefore they make use of financial services, and must thus been seen as user sectors of FISIM. This certainly is the case for the part of S.123 that doesn’t produce FISIM (investment funds). With the assumption that all sectors consume FISIM we constructed a matrix of FISIM, just like the matrices of interest, loans and deposits. The matrix of FISIM consists of the producer sectors S.122, S.123 and S.2, and of all sectors except S.121 as user sectors.

Eurostat:You may record these amounts of units of S.123 sector as consumers (not producers), although we believe that consumption by S.123 sector is of negligible importance. Due to the fact that FISIM is net concept, this explains why only a general description has been given in the Regulation (CR 448/98, Annex III) that sectors S.122 and S.123 are not mentioned as net FISIM intermediate consumers (user sectors).

The value of the same unit of S.123 as FISIM consumer should be shown as netted-off (consolidated) with the same unit S.123 as FISIM producer. In other words, the production (output) of the same unit belonging toS.123 sector should be decreased by the value of intermediate consumption of the same unitbelonging toS.123 sector.