Title: INT 97-23 NSF/Tokyo Report: Household Saving in Japan Date: 6/11/97 The National Science Foundation's offices in Tokyo and in Paris periodically report on developments abroad that are related to the Foundation's mission. These documents present facts for the use of NSF program managers and policy makers; they are not statements of NSF policy. Household Saving in Japan The following report was submitted by Dr. David W. Campbell, Associate Professor, Department of Economics, University of Essex. In January 1997, Dr. Campbell began a 12-month JSPS fellowship at the Institute for Posts and Telecommunications Policy (IPTP). His host is Dr. Masahiro Kawai, Institute of Social Science, University of Tokyo. IPTP made available to Dr. Campbell data from one of their recent surveys. Dr. Campbell has prepared this report with assistance from Mr. Wako Watanabe, a researcher at IPTP. Abstract Research over the past ten years or so has indicated that Japanese and American saving behavior may not be fundamentally different. In this paper using micro data from the 1994 Survey on the Financial Asset Choices of Households we tried to determine if aggregate saving in Japan as in the United States is driven by a small number of very wealthy or high income households. We found that about 12 percent of households (which we called influential households) account for 75 percent of total positive saving, 75 percent of total negative saving, and 75 percent of total net saving. The upper tails of the wealth and income distributions dominate net saving and influential households dominate the upper tails. These conclusions reinforce the hypothesis that the savings process in Japan is not distinctive and highlight the importance of research on heterogeneity of saving behavior. [Note to readers - The following report is modified from the original; it does not contain tabular material, appendices, and citations. Complete reports can be received without charge from: IPTP, Ministry of Posts and Telecommunications, 1-6-19, Azabudai, Minato-ku, Tokyo 106, Japan, tel: 81-3-3224-7310, fax: 81-3-3224-7314.] 1. Introduction Until very recently it was thought that from an international perspective Japan's saving rate was very high and that this was due to institutional and other factors peculiar to Japan. Examples of such factors are the bonus system, tax breaks for saving, high housing prices, and bequests and other transfers. Over the last few years, research indicates that the differential in the private saving rates between the US and Japan is largely due to differing accounting conventions. Further, with one exception, factors such as the bonus system have been shown to be insignificant vis-a-vis Japan's aggregate saving. The exception is bequests and other transfers; further work needs to be done to quantify the importance of these in wealth accumulation. However this last point is not surprising. The role of transfers in the saving process has been an issue of heated contention in most developed countries. In short then it is safe to conclude that, as far as saving goes, the Japan is different hypothesis is dead. This suggests that research agendas on saving adopted by researchers in other developed countries, particularly the United States, might be fruitfully applied to Japan. The most promising of these, in our opinion, is research on the heterogeneity of saving behavior. The abandonment of the stochastic life cycle model by the mid-1980's in the US and the more recent decisive rejection of the intergenerational altruism model in the US has spelled the end of the simple idea that one paradigm of saving was adequate to describe aggregate saving in the United States. Further, some recent work strongly suggests that a relative handful of very wealthy or high income households may dominate aggregate saving in the US. In Japan, too, the intergenerational altruism model has been decisively rejected. And two Japanese researchers (Takayama and Kitamura) have stressed the role of high income groups in wealth formation. In this paper we will try to determine for the first time if aggregate saving in Japan is also driven by a small number of households and if so what the identifying characteristics of these households might be. 2. Definition of saving and data source Our definition of saving is the change in nominal assets minus nominal capital gains/losses. Unadjusted sample size was 706 households; final adjusted sample size was 646.61. The data source was the 1994 Survey on the Financial Asset Choices of Households [hereafter the Financial Asset Choice Survey (FACS)], conducted by the Institute for Posts and Telecommunications Policy of the Ministry of Posts and Telecommunications. The Financial Asset Choice Survey has been conducted biannually since 1988; currently all households are eligible with the exception of those with heads under 20. Households are selected using the stratified multi-stage random sampling method. In 1994, survey forms were distributed to 6000 households; 3924 households returned forms for a response rate of 65.4 percent. Assets included in the FACS are financial assets, liabilities, and real assets. Financial assets consist of twenty-eight products in four categories: deposits, insurance, securities, and other financial assets. Liabilities comprise seven kinds of debts, and real assets are owner-occupied land, owner-occupied housing structure, other land and buildings, and other real assets. Cash, consumer durables, public pension wealth, company pension wealth, and assets of private unincorporated non-financial enterprises are excluded from the survey. 3. Results We were first interested in seeing, if in fact, a relatively small number of households dominate Japanese aggregate saving. We looked first at positive savers, who comprise two-thirds of all households. The top ten percent of these households account for 76.34 percent of total positive saving while the bottom fifty percent account for 5.58 percent. The situation is similar though not quite as skewed for negative savers with the top ten percent of the households accounting for 60.76 percent of total negative saving and the bottom fifty percent accounting for 6.64 percent. As one would expect from the analysis above, total net saving is dominated by households in the bottom and top deciles. One way of summarizing the above information is to note that the minimum percent of total households that account for 75 percent of positive saving, 75 percent of negative saving, and 75 percent of net saving is 12.17 percent. As in the US, in Japan a handful of households drive aggregate saving and the representative household contributes insignificantly to household wealth formation. We call these households influential households, influential households. Influential households save about 22 times as much, own about 3.4 times as much, and earn about 50 percent more than non-influential households. We can see this from another angle by investigating whether the upper tails of the wealth and income distributions dominate net saving. In particular, one is struck by the facts that the wealthiest twenty percent of households account for 86.23 percent of net saving while the poorest 60 percent of Japanese households account for less than 1 percent of net saving. On the income side the upper tail is not quite so domineering and the bottom half of the income distribution is responsible for 32.74 percent of net saving. It is worthwhile noting that the upper tails of both the wealth and income distributions are driven almost exclusively by influential households. In short the upper tails of the wealth and income distributions contribute most of total net saving, and nearly all of these contributions to net saving are made by influential households. [Note - For a comparison with the U.S. please refer to the complete report.] Other factors of course play significant roles in total net saving in Japan. The receipt of a bequest in the past appears to be a necessary prerequisite for substantial wealth accumulation for many households: 48.3 percent of influential households have received a bequest while only 15.4 percent of non-influential households have. Though over twenty percent of households are one person households their total contribution to net saving is less than one percent. The share of aggregate saving of households that received a bequest in 1994 is at 94.94 percent extraordinary high; in fact, given the literature in this field, we can say with confidence too high. Nevertheless this is not surprising, in our view, given the probability of receiving a bequest in a particular year and the size of the sample. We do not doubt that our main findings above will hold even if the share of 1994 bequest saving is very substantially reduced. Finally we examine here, just cursorily, mobility within the wealth distribution. 71.44 percent of all households remained in the same five percentile bracket during 1994. About 1 in 20 households (4.59 percent) in the top twenty percent of the 1993 wealth distribution dropped into the lower eighty percent of the 1994 distribution, accounting for -4.0157 percent of net saving. They were replaced by households responsible for 25.9372 percent of net saving. 4. Conclusion About 12 percent of households in Japan, those we have called influential households, account for 75 percent of total positive saving, 75 percent of total negative saving, and 75 percent of total net saving. The upper tails of the wealth and income distributions dominate aggregate net saving and influential households dominate the upper tails. It seems that as in the US, saving is done by a small number of very wealthy or high income households. This conclusion highlights the importance of research on heterogeneity of saving behavior. Further, it is highly consistent with the substantial body of work over the last ten years or so that has rejected the hypothesis that the savings process in Japan is distinctive.