«CONCEALED INTEREST INCOME OF HOUSEHOLDS IN THE NETHERLANDS: 1977. 1979 AND 1981 Brugt Kazemier The views expressed in this paper are those of the ...»
CENTRAL BUREAU OF STATISTICS
The Ne therlands
National Accounts Research Division
CONCEALED INTEREST INCOME OF HOUSEHOLDS IN THE
NETHERLANDS: 1977. 1979 AND 1981
The views expressed in this paper
are those of the author and do
not necessarily reflect the views
of the Netherlands Centra1 Bureau
CONCEALED INTEREST INCOME OF HOUSEHOLDS IN THE NETHERLANDS;1977, 1979 AND 1981 The major problem in estimating the size of hidden income is that total income, reported plus unreported, is unknown. However, this is not the case with total interest income on giro, bank and savings accounts, (mortgage) bonds, savings certificates etcetera of households in the Netherlands. This makes it possible to estimate at least the order of magnitude of this part of hidden income. In this paper it wil1 be shown that in 1977, 1979 and 1981 almost 50% of total interest received by households was concealed.
2. Some background information
3. Adjustments to the Nationai Accounts
4. Adjustments on the Income Distribution Statistics
5. Concealed interst income of households
6. Micro discrepancy analysis
7. Conclusions Appendix References Introduction
In genera1 for taxpayers there are three methods to reduce taxation:
underreporting income, overreporting tax reliefs and taking advantage of loopholes in tax laws. Most of these activities are hidden. Nevertheless, many researchers have tried to quantify them and a vast variety of sometimes ingenious methods have been used. Yet, there is still no consensus on their magnitude.
The major problem in estimating the size of hidden income is that total income, reported plus unreported, is unknown. This, however, is not the case with total interest income on giro, bank and savings accounts, (mortgage) bonds, savingscertificates etcetera of households in the Netherlands. This fact makes it possible to estimate at least the order of magnitude of this part of hidden income.
The basic methodology underlying these estimates is a detailed analysis of the discrepancies between the National Accounts (NA) estimates of total interest income of households and the estimates in the Income Distribution Statistics (IDS). The estimates are based on different data sources: the first on data obtained from the giro, trade and savingsbanks, the Centra1 Bank, the government, the stock exchange, institutional investors, pension funds, large companies etcetera. The IDS are mainly based on data derived from taxforms. Therefore the discrepancies between both sets of statistics can give an indication of the order of magnitude of concealed interest income.
Agenera1 belief arose that the total amount of evaded taxes due to concealed interest income was very large. To be certain of this, Parliament decided on a sample study in which the interest receipts reported in the tax returns of 3000 income tax payers, were compared with the corresponding information in the accounts of the giro and al1 trade and savingsbanks in the Netherlands. For this purpose a special law was passed in which matters of privacy were settled and the co-operation of the banks was enforced.
The results of this micro discrepancy analysis became public and can be compared with the results of the macro discrepancy analysis described in this paper. The results substantiate each other.
This paper is built up as follows. Section 2 gives a brief background of the Dutch National Accounts and the Dutch Income Distribution Statistics.
The Same section als0 includes a brief description of the Dutch income tax system as far as it is of interest with respect to the methodology of the macro discrepancy analysis. Sections 3 and 4 describe the adjustment made to the NA and IDS estimates. These adjustments were necessary because of differences in definitions end imperfections in both sets of statisticc. In section 5, conclusions on concealed interest income are drawn. In section 6, the results of the above mentioned micro discrepancy analysis are summarized and compared with the conclusions in section 5. This paper concludes with section 7, a summary of the main results.
2. Some backnround information In the Dutch income tax system, interest income is treated in a very special way. Whereas wages, salaries, social benefits etcetera are liable to a personal income tax, interest income is not. The latter is considered to belong to al1 members in a household. Therefore, the interest income of al1 members is counted together and must be added to the personal income of the main wage earner, that is the person with the highest income. There are few exceptions. The most important of these is that children of age and parents who live with their children are not considered to be members of the household. Their interest income is added to their o m personal incomes.
In 1977 al1 interest income was subject to income tax. In 1979 and 1981 it was not. In 1979 there was an exemption for the first 200 guilders per household. In 1981 this exemption was 700 guilders, but applied to the balance of interest income and interest payments (e.g. on mortgages).
Notwithstanding these exemptions, one was obliged to report total interest income, total interest payments and the resulting exemption separately. The reason for this obligation was the existence of an inverse relationship between the cum of al1 positive income elements and some tax deductions.
For example, tax relief on charity gifts is higher for lower income groups.
Consequently, for everybody who completed a tax return, any not reported interest income must be considered as concealed, even if it would not have influenced the assessment.
Not everybody has to complete a tax return, especially when wages, salaries or social security benefits do not exceed a specified maximum. In these cases the pay-as-you-earn tax (P.A.Y.E.) is considered to equal income tax. This applies to about 45% of al1 income tax payers. It does not apply to those
- who receive a tax form in spite of this. When the inland revenue thinks for some reason that the P.A.Y.E. does not equal the income tax it sends out a tax form which must be completed and returned.
- whose taxable income is 600 guilders higher than the income on which P.A.Y.E. was paid, because of e.g. a large amount of interest income. If the inland revenue does not send a tax form, one must ask for it oneself.
Besides these two exceptions, people who think that P.A,Y.E. exceeds their income tax are advised to ask for a tax form to obtain a refund. Obviously interest income of those people who are legally justified in not completing a tax return must not be considered as concealed.
Both the accounts of the income tax and those of the P.A.Y.E. are used as a data source for the IDS (CBS, l982b, 1983b). The first accounts are more detailed than the latter. Among other things they contain information on interest income, in contrast to the P.A.Y.E. accounts which do not. The IDS do not include interest income of people who pay only P.A.Y.E.
The NA are completely different (CBS, 1983a). They are not based on one or two single sources, but are the result of a detailed comparison of a large number of data in an input-output framework. The data are obtained from many different sources and sets of statistics. Fiscal sources hardly play a role at all. As already mentioned in section 1, the cum of al1 interest flows is routinely calculated from data obtained from the giro, trade and savingsbanks, the Centra1 Bank, the government, the stock exchange, institutional investors, pension funds, large companies etcetera.
In addition, an independent estimate of the interest flows per sector is made, except for the household sector, which in the NA includes nonprofit organizations like trade unions, charities, sports clubs etcetera. The interest payments and receipts of this household sector are calculated as a residual.
As can be seen from the last two paragraphs, the IDS and NA cannot be compared without some adjustment. Firstly, the definition of a household differs. Secondly, the IDS contain interest flows between households, while the NA do not. On the other hand, the NA contain interest income of non tax payers and those tax payers who pay only P.A.Y.E., whereas the IDS do not.
Further, the NA are transaction based and the IDS are cash based. Finally, IDS the distinguish between three different interest categories, as discussed below.
3. Adjustments to the National Accounts As section 2 described, the interest income of households is deduced as the sum of al1 interest flows which cannot be assigned to other sectors.
However, interest received on giro, bank and savingsaccounts and on al1 types of savingscertificates can also be directly deduced from other published statistics. These cover precisely two of the three interest categories distinguished in the IDS. Interest received on bonds can then be calculated as the difference between total interest receipts and the two direct estimates. This enables a more detailed discrepancy analysis.
The interest on giro, bank and savingsaccounts can be deduced from the Statistics of Savings (CBS, 1980, 1981a, 1982a, 1983~). The definition of households used in these statistics corresponds with the one used in the NA. The Statistics of Savings do not report interest payments, but they report the total end-of-yearbalances of savings of many different types of accounts and their corresponding interest rates. Together, these statistics provide enough information to estimate total interest received on these accounts. From end-of-year data, average interest rates and average balances were calculated for each type of account. ühere no interest rates were given, information from similar accounts was used. Next, the available monthly data on giro accounts was used to adjust the average balances for seasonal fluctuations. Finally the averaged interest rates and seasonally adjusted average balances where multiplied to estimate the total interest due.
In 1978 the definition of savings used in the Statistics of Savings was changed slightly. From then on, deposits of less than 500,000 guilders were also counted as savings. It was possible to calculate the effects of this change in definition and to adjust the 1977 estimates accordingly. This new definition almost coincides with that of the IDS, &xcept for the maximum of 500,000 guilders. This limitation, however, is not of great importance, because in most cases, deposits with a higher balance are of a business nature and are rightly excluded from the analysis. If they are not businesslike, they affect only the distribution of concealed interest income over the three categories. Concealed interest on giro, bank and savingsaccounts are then slightly underestimated, while concealed interest on bonds are overestimated by the Same amount.
The estimation of interest on al1 kinds of savingscertificates falls into two parts. The balances and interest rates of one type of savingscertificate can be found in the above mentioned Statistics of Savings. The interest due on these savings is calculated in the Same manner as that on giro, bank and savingsaccounts. Further data on the balances of savingscertificates can be found in DNB (1983), but no data on interest rates is provided. It is known, however, that these rates are heavily related to the interest rates of longtenn loans. As an approximation, the interest rates of public loans were used. Due to heavy fluctuations in this rate in the late 1970s, the total interest received in a given year on savingscertificates can be calculated only when the composition of year of issue and maturity dates is known. The following distribution of maturities was assumed: 2 years 10%, 3 years 10%, 4 years 20%, 5 years 20%, 20% 6 years, 7 years 10%, 8 years 5% and 9 years 5%. Using this distribution and starting from the first year for which data on savingscertificates were published, and assuming that this was their first year of issue, total interest due was calculated. Also several other distributions of maturities were simulated. They al1 gave approximately the Same result. Al1 savingscertificates were assumed to be possessed by households only.
As already mentioned, interest income on bonds is calculated as the difference between total interest income according to the NA and the above described independent estimates of the other two categories. An objection to this procedure is that al1 errors in the first estimates affect this last one. An alternative independent estimate is impossible because of lack of data. An attempt was made, however, to estimate the size of this third interest category independently. The data used were based on the bonds issued by national and local goverment, universa1 banks and other money creating institutions, assurance companies (DNB, 1983), the 'Bank Nederlandsche Gemeenten' (Dutch Municipality Bank) and the "Nederlandsche Waterschapsbank' (see e.g. BNG, 1982; NWB, 1982), religious communities, large limited companies or corporations listed at the share market and data on mortgage bonds (see e.g. CBS, 1979a, l982c, 1982d, l982e, l983d). The interest households received on these bands was calculated to be at least 1700 million guilders in 1977, 2000 million guilders in 1979 and 2900 million guilders in 1981. Although these figures do not account for bonds issued by smal1 limited companies or corporations listed on the stock exchange, unlisted companies, hospitals, savingsbanks, associations and institutions or foundations with businesslike goals, foreign bonds etcetera, they account for some 75% of the residually estimated total interest received on bonds. Therefore, these estimates are considered to be fairly reliable. The first column of table 1 presents the calculated distribution of total interest received by households.
Table 1. Interest receipts of households according to the National Accounts 1977, 1979, 1981.
x 1000 million guilders Interest on giro, bank and savingsaccounts, and between households Interest on bonds Interest on al1 types of savingscertificate Total interest The definition of the household sector in the NA is more comprehensive than that in the IDS. The first includes some nonprofit organizations whereas the latter does not. For this disparity an adjustment is made.