The Problem of Estimating the Proportion of Income Identified
and Issues for the Future
- Too early to tell if the gap has narrowed significantly since the 1990s -
October 07, 2005
The need to address the "ku-ro-yon problem", namely, unfairness in Japan's tax system due to the fact that the ability of the tax authorities to identify income is high as regards employment income but relatively low as regards business income or farming income, is more urgent today than ever before. One reason is that concrete proposals for tax reforms, particularly a review of deductions from employment income, have been made from within the government. Deductions from employment income not only constitute the deduction of amounts corresponding to expenses from the employment income of salaried workers, but also have the character of compensation for the fact that it is easier to identify employment income than business income. For this reason, any review is conditional on the resolution of the "ku-ro-yon problem". Given that income is the basis for the payment of benefits such as child allowance, and for the calculation of insurance contributions, it is also important if the fairness of the social security system is to be assured.
In Japan, however, the government itself does not assess or publish figures on the actual proportion of income identified, and estimates by expert researchers have traditionally played an important role. The estimates by Hiromitsu Ishi (1981), widely considered to be pioneering examples of this type of work, indicate that, despite some slight variation, the ratios of 90%, 60% and 40% that are the origin of the name "ku-ro-yon" (ku-wari, roku-wari, yon-wari; nine-tenths, six-tenths, four-tenths) can largely be confirmed in quantitative terms, as far as the 1970s are concerned. The most recent estimates by Ohta, Tsubouchi and Tsuji (2003) of the Cabinet Office Economic and Social Research Institute, which are based on the same methodology as the Ishi estimates, suggest that by 1997, the ratios of income identified were approximately 100%, 90% and 80%, and that the ku-ro-yon problem had been alleviated to a substantial degree.
However, it should be remembered that the interpretation of such estimates, especially those covering the 1990s, should be approached with even greater caution than is advised by those making the estimates. For a number of reasons, it is important that future policy debates should not be based solely on these estimates, even if only for a short period of time.
In the first place, it is likely that the assumptions on which most of these estimates are based are starting to collapse. Estimating the proportion of income that the tax authorities are able to identify requires, in addition to the tax income statistics gathered and compiled by the tax authorities, access to statistics that express the "true" amount of income that is being earned in Japan and, although many estimates of the proportion of income identified are based on the assumption that the figures they use they can be replaced by the SNA's National Income Statistics, it is probable that, in recent years, this assumption has become shaky. In fact, it is probable that the figures in the National Income Statistics since the second half of the 1980s have underestimated employment income to a degree that cannot be overlooked.
Second, the effect of problems inherent in the methodology of the estimates (of which the estimators are not aware) on the results of the estimates is likely to have increased, particularly in recent years. For instance, even within income derived from real estate, such as car parking space rental fees, the classification of that income under the tax system will differ according to whether or not the taxpayer receives the income as the fruits of a business activity, so that the income may be treated either as "business income" or as "real estate income". Thus, the estimate methodology does not make proper allowance for the fact that "income" is a many-sided concept and this is likely to have imparted a significant bias to the results of the estimates.
Third, the estimators themselves recognize that when undertaking an estimate, statistical limitations make it is necessary to set arbitrary assumptions and that these assumptions have a considerable influence on the outcome of the estimate.
Needless to say, if future policy debate is to be constructive, it is essential that more reliable, quantitative information on the proportion of income identified be made available. Besides the improvement and diversification of estimate methodology, another important requirement is that the government should learn from the examples of the United States, Sweden and other countries and itself investigate and publish its findings on what proportion of income is being identified.
For more information on the content of this report, please contact: Kazuhiko Nishizawa the Japan Research Institute, Limited.