# «CONCEALED INTEREST INCOME OF HOUSEHOLDS IN THE NETHERLANDS: 1977. 1979 AND 1981 Brugt Kazemier The views expressed in this paper are those of the ...»

These adjustments were calculated from data retrieved from a large variety of sources, such as annual reports. Total interest received by the nonnatura1 persons in the NA household sector was estimated at about l00 million guilders in 1977 and the same in 1979. In 1981 it was just over 200 million guilders. The distribution of these amounts over the three interest categories is deduced from data of national associations of e.g. sports and dancing, nonprofit broadcasting organizations, employers' organizations, employees' organizations and religious communities (see e.g. CBS 1979b, 1981b, 1982d). These corrections are probably underestimates, because of interest on bonds, received by relief funds and strike funds of trade unions. No data could be found on these. The NA estimate of total interest income of households, however, is most probably also underestimated for the same reason and by approximately the same amount. Therefore this underestimation does not affect the estimate of concealed interest income.

The NA includes only interest flows between sectors. Ergo the interest flows within sectors are excluded. However, the latter are subject to income tax and they are therefore included in the IDS. There, they are titled 'overige vorderingen' (other claims), which in this paper are put together with the interest income on giro, bank and savingsaccounts.

Separate data on these claims are not available, but can be partly deduced from total interest payments of households. A comparison of these payments according to the IDS and NA shows that the former exceeds the latter. It is assumed that this difference is due to interest payments within the household sector. Most probably, the mutual interest payments are much higher, because the do not account for payments not reported.

IDS Consequently, the estimates of concealed interest in this paper are somewhat conservative. In 1977 the difference between NA and IDS estimates was 300 million guilders, in 1979 it was 100 million guilders and in 1981 it was almost zero. As this decline is not very likely, the estimates for 1977 were also used for the other two years. As this component constitutes only a very minor part of total interest income, this discrepancy wil1 not alter our main findings.

Interest on savingscertificates is normally paid at the end of the contract period. At that moment it is subject to income tax and not in the years before. Therefore in the IS this interest is registered on cash basis. In the NA, however, it is registered on acrual basis. The difference between the two can be calculated. In 1977 the differente was about minus 300 million guilders, in 1979 it was almost minus 400 million guilders and in 1981 again minus 300 million guilders. For bonds it amounted to minus 100 million guilders per year. For giro, bank and savingsaccounts it was almost zero each year, For selfemployed persons, it is often difficult to keep private and business interest receipts separated. Errors are easily made. It is assumed that these errors are not very important and have not influenced the IDS systematically. Consequently, no adjustment for this is considered necessary. Table 1 shows the NA data on interest receipts of households and the sum of the necessary adjustments.

4. Adjustments on the Income Distribution Statistics The IDS includes only the interest receipts of those people who returned a tax form. It does not cover the receipts of those who pay only P.A.Y.E. As argued in section 2, not al1 this interest is concealed. Only if a person has so much extra income that taxable income is 600 guilders more than the income subjected to P.A.Y.E., it must be considered as concealed, if not declared. Interest minus exemption falls within this extra income. In 1977 there was no exemption on interest income at all, in 1979 this exemption was 200 guilders on interest receipts and in 1981 it was a maximum of 700 guilders on the balance of interest receipts and payments. Thus i£ interest income was the only extra income, no interest was paid and there were no other deductions, P.A.Y.E. payers who did not complete a tax return but did receive more than 600 guilders interest in 1977, 800 guilders interest in 1979 or more than 1300 guilders interest in 1981 were tax evaders.

For tax payers who paid only P.A.Y.E., no data are available, except their number. Therefore, to estimate the size of correctly nonreported interest receipts some assumptions were necessary. Firstly, interest is the only extra income. No interest was paid, nor were other tax deductions allowed. Secondly, for each income bracket (32 in total) per socio-economic category (5 in total), the average realized exemption on interest income of people who paid P.A.Y.E. but did not complete a tax return equals that of those people who did complete one. Thirdly, the Same assumption holds for the interest income distribution. The method used to estimate the correctly nonreported interest income is described in the Appendix. The results can be found in table 2.

A second but minor adjustment should be made for households who did neither completed a tax return, nor paid P.A.Y.E., e.g. students without earned income. Lack of data makes this impossible. This adustment, however, would not be very large and would hardly influence the conclusions on concealed interest income. Further, no full account is taken of income from properties of under-aged children whose parents or guardians are using those properties. This income must be reported as 'overig inkomen' (other income), and consists of e.g. income from real estate, dividends etcetera and for a minor part of interest income too. As the total 'overig inkomen' in the IDS is less than 100 million guilders, the interest included is negligible.

** Table 2. Reported interest receipts of households according to the Income Distribution Statistics 1977, 1979, 1981.**

Table 3 shows the estimates of interest tncome after adjuctment according to the NA and the IDS. The difference between them is so large that statistical reasons alone cannot explain it a l l. It is more realistic to assume that the discrepancies are caused by fraud, resulting in a systematic downward bias in the IDS estimates. Consequently, the discrepancies between the adjusted NA an IDS estimates can be considered as indicators of the order of magnitude of concealed interest income of households.

** Table 3. Discrepancy between the receipts of households according to the adjusted National Accounts and the Income Distribution Statistics 1977, 1979, 1981.**

Interest on giro, bank and savings accounts, and between households Interest on bonds Interest on al1 types of savingscertificate Total interest

1) Calculated frm unrounded results Al1 three years under consideration show roughly the Same percentage of concealed interest. This also applies to each interest category: 30-35% for giro, bank and savingsaccounts, around 70% for bonds and over 80% for savingscertificates. There is a downward tendency for savingsaccounts and an upward tendency for bonds. These tendencies, however, do not justify conclusions, as they may wel1 be insignificant and fa11 within the confidence intervals.

Although the amount of concealed interest did not grow in relation to total interest receipts, it doubled in absolute terms from about 3300 million guilders in 1977 to 6500 million guilders in 1981, and in magnitude is possibly more significant than hidden extra earnings (Van Eck and Kazemier, 1988 and 1989). In relation to national income it grew from 1.2% of GNP in 1977 to 1.9% of GNP in 1981.

6. Micro discre~ancvanalvsls In 1983 the Dutch Parliament instituted an enquiry in which the interest receipts, reported in the tax returns of a sample of 3000 income tax payers were compared with the corresponding information in the accounts of the and savingsbanks in the Netherlands. The enquiry giro and al1 trade details and matters of privacy were governed by the 'Wet Rentesteekproefonderzoek 1984/1985' (Law Interest Sample Research 1984/1985) of 1985 may 21at.

The sample consisted of 1369 units, i.e. a single person, a married couple without children or a family. Children of age were handled as a separate unit. The 1369 units correspond with almost 3000 persons. The Dutch Centra1 Bank collected the 1981 data on interest received, as far as they could be found in the accounts of the giro or the banks. The Department of Finance collected the corresponding reported data on interest, as far as tax returns had been completed. Both data files were put in the Same order, made anonymous and cent to the Centra1 Bureau of Statistics for further analysis. The details and results of this micro discrepancy analysis were reported in TWEEDE KAMER (1986).

A comparison of the results of the micro analysis with those of the macro analysis shows remarkable similarities. In the micro analysis, the adjusted but still underestimated interest receipts on giro, bank and savingsaccounts were 8100 million guilders in 1981. In the macro approach these were estimated at 8500 million guilders (NA estimate minus other claims ('overige vorderingen')). The difference is less than 5%. Further, the interest reported in the micro approach is circa 5700 million guilders, whereas in the macro approach interest reported was 5500 million guilders (IDS estimate minus 'overige vorderingen'), als0 a difference of less then 5%. Finally, the correctly nonreported interest of tax payers who paid only P.A.Y.E. is approximately the Same. These striking similarities, arrived at by completely different methods from completely different data sources, strengthens the confidence in the results of both analyses.

7. Conclusions In 1977, J979 and 1981 around half of total interest receipts of households in the Netherlands was illicitly concealed from the tax authorities.

Because of the rapid growth of total interest, the concealed interest income grew from 3000-3500 million guilders in 1977 to 6000-7000 million guilders in 1981. Also, for the three interest categories distinguished, the percentages of interest concealed were approximately the came in each 80-90% on year considered. In descending order: al1 types of 30-35% on giro, bank and savingscertificates, 65-80% on bonds and savingsaccounts. The results of this last category are confirmed by those of a micro discrepancy analysis.

Appendix For 1979 there were IDS data available on the average interest income and average interest exemption per socio-economic category per income bracket.

This made it possible to estimate a two parameter interest income distribution function per category per bracket. The most flexible specification often used for income distribution purposes is the Gamma distribution. An alternative is the Pareto distribution function but it is not defined for incomes of less than a certain pre-defined level. A second alternative is the Lognormal distribution but this can never be strictly decreasing (KLOEK and VAN DIJK, 1978). Therefore, the Gamma distribution was chosen. Its general forrn is

** r r-l -Xx fx - X x () e /r(r)**

For values of rl, this function first increases to its maximum and then decreases uniformly. If rsl this function is strictly decreasing. For each socio-economic category i and each income bracket j the average interest income is Zij,the average interest exemption is mijand the number of people who paid only P.A.Y.E. is U i j. The maximum exemption x,,, in 1979 v,, was 200 guilders. Using the Gamma distribution and the known averages the values of Ai j and rij can be calculated.

**and**

The first equation gives the ratio between r i j and Aij. Using the second equation, the absolute value of both parameters can be calculated. The results of the calculations can be found in Table A. For some income brackets, the number of observations was rather small. Therefore some brackets were taken together. The results of the categories "self employed" and "managing directors" were omitted, as they hardly contribute to the total adjustment. The number of self employed or managing directors who pay only P.A.Y.E.is very small.

Table A. Estimates of the parameters X and r of the Gamma distribution to describe the 1979 interest income distribution, per socio-economic category and income bracket

The calculated values of X and r show a satisfying regularity. Both decrease when total income increases. The obtained interest income distribution is used to calculate the adjustments for correctly nonreported interest income. This adjustment equals

BNG, 1982, N.V. Bank voor Nederlandsche Gemeenten, verslag over het boekjaar 1981 (BNG, 'c-Gravenhage).

CBS, 1979a Hypotheekbanken voor onroerende goederen 1977. Maandstatistiek financiewezen 1979, 4, pp 253-261 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1979b, Maandstatistiek van het financiewezen 1979, 8 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1980, Statistiek van de spaargelden 1976-1978 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1981a, Statistiek van de spaargelden 1977-1979 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1981b, Maandstatistiek van het financiewezen 1981, 8 and 10 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1982a, Statistiek van de spaargelden 1979-1980 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1982b, De personele inkomensverdeling 1977, kerncijfers (CBS, Voorburg).

CBS, 1982c, Bouwfondsen en hypotheekbanken voor onroerende goederen 1978-1979. Maandstatistiek finianciewezen 1982, 1 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1982d, Financiële gegevens kerkgenootschappen 1978 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1982e, Statistiek van balans en resultatenrekening beurs N.v.'s 1980 (Staatsuitgeverij, 's-gravenhage).

CBS, 1983a, Nationale rekeningen 1982 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1983b, De personele inkomensverdeling 1979, kerncijfers, herziene uitgave (CBS, Voorburg).

CBS, 1983c, Statistiek van de spaargelden 1981 (Staatsuitgeverij, 's-Gravenhage).

CBS, 1983d, Financiële gegevens kerkgenootschappen 1980 (Staatsuitgeverij, 's-Gravenhage).

DNB, 1983, Jaarverslag van de Nederlandsche Bank N.V. (Kluwer, Deventer).

Kloek, T. and H.K. Van Dijk, 1978, Efficient Estimation of Income Distribution Parameters. Journal of Econometrics, Vol 8, pp 61-74.

NWB, 1982, Jaarverslag van de Nederlandsche Waterschaps Bank 1981 (NWB, %-Gravenhage).