there are other ways in addition to partnerships in which individual taxpayers can also undertake tax-sheltered investments.

But neither the business entity used for the shelters nor the high-income individuals who participate in them are responsible for the unfairness in our tax system and the prominence of tax shelter activity. Rather, I think we totally agree with Mr. Dorgan's comment, the problem arises from the high marginal tax rates and the ability of taxpayers to reduce those rates by investing in activities that are given preferential treatment in our existing tax system. Thus, in many instances those tax provisions which are intended to encourage certain kinds of economic activity are also those that cause the perception that the system is unfair.

Now, my statement and my comments will be divided into three sections. The first reviews the high-income taxpayer analysis that we transmitted to you, to the subcommittee, in July. Second, I would like to review our more recent analysis of the growth in partnership activity over the past two decades; and then third, I will review with you our analysis of the most recent partnership tax return data and try to point out some of the highlights of those data.

In order to undertake the high-income taxpayer analysis, it was necessary for us to develop some concepts that are not commonly referred to and which are not part of normal tax return jargon. Therefore, I would like to spend at least a moment in explaining to you the concept of total positive income, which is very important to our analysis of high-income taxpayers.

In analyzing high-income taxpayer returns, we must be able to measure income before deductions or offsets to income reported on those returns. There are two categories of deductions or offsets. First the items that receive most attention are the so-called belowthe-line deductions in taxable incomes, the itemized deductions, and tax credits with which we are familiar and which show up very prominently on the tax return.

The second, but certainly as important a category of items, are the so-called above-the-line offsets. These include business and investment losses, exclusions, including the capital gains exclusion, and income adjustments. The above-the-line offsets do not always appear quite as clearly on the tax return. It is easy, as I indicated, to identify the below-the-line offsets because we can look at a tax return and identify those items of deduction or credit that are claimed so-called below the line. It is much more difficult, however, to identify the above-the-line offsets.

In order to be able to identify both, which is important in undertaking the analysis that we presented to you, we utilized a concept called total positive income. This concept uses tax return data, and includes primarily wages, salaries, dividends, interest, and income from profitable businesses and investments. Because total positive income excludes the current law capital gains exclusion, IRA contributions, Keogh contributions, business and investment losses, it permits a comparison of, if you will, the beginning point in terms of identifying income before both above-the-line and below-the-line offsets.

I want to point out, however, and I consider it an important point, that many losses reported on a tax return, whether from partnership activities or not, represent real economic losses. But many of them also reflect the use of tax preferences or tax incentives that generate losses for taxpayers and create opportunities for tax shelters. Above-the-line offsets represent the difference between all positive amounts of income and adjusted gross income.

In its annual reports on high-income individual returns, the Internal Revenue Service uses four income concepts ranging from the traditional and well-known concept of adjusted gross income to the concept of expanded income. Now, I am not going to go into the details of these various ways of analyzing returns except to say that for analyzing high-income returns, our concept of total positive income, or TPI, is a more useful measure of income than is adjusted gross income or any of the related concepts.

Like all measures, TPI is not perfect. There are no taxpayer data on the ideal measure of income economists frequently call economic income. In theory, economic income counts income from all sources, whether or not reported on the tax return, and accurately measures each form of income. Such an ideal measure of income would separate economic losses from losses claimed on returns. In practice, however, economic income is very difficult to measure and cannot be constructed using data only from tax returns.

For our purposes, we chose a high cutoff level, namely $250,000 of total positive income, for considering any return high income. We did that in part so that we could make sure to the extent that total positive income overstated economic income, that we were nevertheless talking about relatively high-income taxpayers.

After we identified the category of taxpayers we would consider high income, that is, taxpayers with over $250,000 of total positive income, we then performed a computer examination of all individual tax returns, where total positive income was in excess of $250,000. In 1983 over 260,000 returns fell in this category, representing about one-quarter of 1 percent of all returns filed for 1983.

A report that we submitted to the subcommittee on July 31 describes most of the results of that analysis. What I would like to do this morning is simply mention several of the findings and add a few points not reported in our July 31 transmittal to the subcommittee.

While much interest centers on high-income taxpayers who pay little tax, let me again emphasize that it is important to remember that most taxpayers filing the returns we examined paid a substantial share of their income in taxes in 1983. For example, nearly half owed tax equal to at least 20 percent of their total positive income. Nearly one-quarter, 23.6 percent, for example, owed tax equal to at least 30 percent of their total positive income, and 5 percent owed taxes equal to at least 40 percent of total positive income. If you want to examine the data more specifically, table 1 contains those data by income and tax-by total positive income and by tax rate or tax burden category.

On the other hand, that table indicates that a significant minority of taxpayers owed very low taxes in 1983 relative to their total positive income. For example, 11 percent, or almost 30,000, high TPI taxpayers paid less than 5 percent of TPI in tax. Of those, over 3,000 nad TPI in excess of $1 million. In addition, over 7,000 taxpayers, or 2.9 percent of those filing returns with TPI over $250,000, owed zero tax. Of those, 931 had TPI of at least $1 million.

For the high-TPI individuals who paid little or no tax, I think the question is: How did they do it? Table 2 seeks to show that the above-the-line offsets to TPI were much more important than the below-the-line offsets, such as itemized deductions. And let me say again it is for that reason, so that we could make that kind of analysis, that we used a TPI concept.

In the aggregate, if you are looking at table 2, for high-income taxpayers owing less than 5 percent of tax, that is the second line on that table, itemized deductions offset only 18 percent of TPI, whereas if you look at the left-hand portion of that table, the socalled above-the-line offsets, you will see that for these highincome, low-tax people losses and capital gains deductions actually exceeded TPI, offsetting 112 percent of TPI.

Also, among the above-the-line offsets, what we called current business losses were more important than the combination of capital gains deductions and other kinds of losses, such as losses carried over from previous years. Current business losses equal the partnership loss column, plus the column labeled "Other Current Business Losses." Current business losses for taxpayers with a tax liability under 5 percent of TPI offset 67 percent of TPI, compared with 46 percent for capital gains deduction and loss category.

Partnership losses dominated the category of business losses for these high-income, low-tax taxpayers. If you look at the data, you will see that of the 67 percent, 36.2 percent of those losses were partnership losses. It is this kind of data that suggests to us that it is appropriate to examine what it is about those partnership losses that contribute to the use of partnerships by high-income taxpay

ers.

High-income taxpayers as a group, including those paying significant taxes, also make substantial use of above-the-line offsets, but less than the high-income taxpayers with low total tax liability. You can determine that from looking at line one on table 2. Within the category of current business losses, partnership losses again, dominated, respectively, almost 11 percent of TPI or one-quarter of all above-the-line offsets for all high-income taxpayers. Thus, partnership losses clearly are a major, though not the only, source of reduction to taxable income and, therefore, in tax liabilities paid by high-income taxpayers.

High-income taxpayers also report a disproportionate share of the partnership losses that appear on individual tax returns. If you turn to table 3, you will see that approximately 166,000 returns, or 64 percent of all high-income taxpayers, registered some partnership loss in 1983. Only 2 percent of all taxpayers reported partnership losses. I think that statistic clearly indicates the major role partnership losses play with respect to the so-called high-income category. The aggregate amount of losses reported on the individual returns totaled $17.6 billion, or, 52 percent of the total of $32.2 billion in partnership losses reported on all high-income tax returns in 1983.

Just for comparison purposes, these high-income taxpayers generated less than 10 percent of all total positive income in 1983. Whereas partnership gains reported on all individual returns exceeded partnership losses by $900 million in 1983, among highincome returns, the losses exceeded gains by $6.5 billion.

Again, I want to emphasize, however, that the TPI measure does not distinguish between economic losses and losses that appear only for tax purposes. Some portion of these partnership losses, thus, represent real economic losses that properly reduced the taxable income shown on the returns examined. From our data, the real economic losses that result from bad luck, misjudgments, or other business errors cannot be distinguished from losses advertised by tax-shelter promoters. We know total positive income includes some real losses, but we also know that other frequently used measures, adjusted gross income, exclude all tax shelter losses.

Available information does permit us, however, to help understand better the source of these losses, and that is the focus that I want to turn to in the remaining portion of my testimony. Since our analysis that we transmitted to you, Mr. Chairman, in July has identified partnerships as the main source of business losses for high-income taxpayers and since there is much interest in partnership as a common vehicle for tax shelters, let us turn to a more specific focus on the partnership sector, and I would like to begin by focusing on some historical data on partnerships.

Historically, the partnership sector has been the source of substantial taxable income for individuals and for taxpayers as a whole. If you will turn to table 4, you will see that if you look at the partnership data in column 1, for example, in the mid to late 1960's-the bottom of that table-partnerships generated net income of roughly $10 billion a year. You can see that number was $9.7 billion in 1965 and rose to a high of $11.4 billion in 1968, but hovered in the $10 billion range throughout the mid to late 1960's.

During the 1970's, the gains and losses generated by partnerships grew substantially, although the net income for the sector as a whole and the net amounts reported on individual returns did not show similar growth until the latter part of the decade. But if you look at the 1970's, and I would suggest you look at the second and third columns for gain partnerships and for loss partnerships, you will see a rather substantial growth in both gain partnerships and loss partnerships.

The pattern of fairly stable or generally rising net income from partnerships changed abruptly in 1980. Between 1979 and 1980, net income for the sector dropped from $15.2 to $8.2 billion, and net losses, losses from loss partnerships, jumped from $24.8 to $36.8 billion. In 1981, partnerships in the aggregate reported a net loss for the first time in 25 years that annual statistics on partnership returns have been available. The net losses in the partnership sector have continued through 1983, most recent year for which statistics are available, with 1983 showing the greatest net loss of the 3 years. Losses have assumed a greater prominence in the last 3 years for which statistics are available than in any time during the three decades of statistics on partnerships. In 1981 and 1982, individual taxpayers overall reported net losses from partnerships on their individual income tax returns.

Generally over the past two decades, growth within the partnership sector has been concentrated in certain industries and types of partnerships. Our July 31 report to the subcommittee documented that. If you look at table 5, we have tried to identify broad industry categories. And what you will see in table 5 is that between 1965 and 1983-I am incidentally referring to the second category, the one that is headed "Number of Partnerships With and Without Net Income"-the number of partnerships in both real estate and finance more than tripled. Oil and gas partnerships more than quadrupled. Meanwhile, the total number of partnerships in other industries increased by only 15 percent.

Continuing to look at table 5, between 1965 and 1983, the number of partnerships reporting losses more than tripled, from 229,000 to 750,000 partnerships. Particularly rapid growth in the number of partnerships with losses appeared in the oil and gas and real estate industries. If you look at the third category, "Percentage of Partnerships Without Net Income," you can see by percentages the details in all of those industries and the growth in the industries that I mentioned to you.

Although limited partnerships represented only 1 in 12 of all partnerships in 1977, they accounted for one-third of the increase in the number of partnerships between 1977 and 1983. Only onequarter of all partners in 1977 belonged to limited partnerships, but 88 percent of the increase in partners over the next 6 years was attributable to limited partners. General partnerships experienced little growth during the same period.

Now let's focus on the third type of data we analyzed, that is, partnership losses during the period 1981 through 1983. Partnerships do not pay taxes themselves. Rather, a partnership is an entity that passes through income and expenses to its partners. Generally these items are taxed as if received by the partners, although special rules and limitations apply to some of these items. Consequently, what a partnership reports as partnership income is not what most people would generally consider income from a partnership.

Therefore, we made some adjustments to the measure of net income that partnerships report on form 1065-1065 is the number of the form that partnerships file with the Internal Revenue Service on which they show what are referred to as distributive share income and loss. We refer to the measure of income that we begin with, that is, what comes from the partnership return, as "1065 net income."

If you turn to table 6, and if you begin by looking at the fifth line on that table, you will see form 1065 income or loss. That is the beginning point in our analysis, and that is the number that we got out of our computer analysis of partnership returns filed during this 3-year period.

Now, as I indicated to you, we wanted to arrive at a concept that would reflect the contribution the partnership activities make to taxable income or losses reported by the separate taxpayers, by the partners, that is. Some of the adjustments we made involved items of income or expense which are subject to special treatment at the partnership level, such as investment interest expense or long-term capital gains. A few of the adjustments required estimates of items not fully reported on any of the partnership tax forms. Most of the mineral exploration costs fall into that category. Income after