CBO Cost Estimate of the Tax Deal

On December 10th, the Congressional Budget Office released a one-page table which shows the estimated cost for the so-called "tax deal". More specifically, it shows the estimated change in revenues and direct spending for S.A. 4753, an amendment to H.R. 4853, the Tax Relief, Unemployment Insurance Reauthorization, and Job Creation Act of 2010. The following graph shows the resulting revenues, outlays, and deficits when this change is added to the current baseline:



The graph also shows the revenues, outlays, and deficits for the baseline and the alternative under which EGTRRA and JGTRRA (the 2001 and 2003 Bush tax cuts) are permanently extended and AMT (Alternative Minimum Tax) is indexed for inflation. The actual numbers and sources for this and the following graph can be found at this link.

As can be seen, none of the alternatives have much effect on outlays. Regarding revenues, the tax deal causes a two-year slowdown in their projected recovery and the permanent extension and AMT fix causes a continuing slowdown with revenues remaining between 17 and 18 percent of GDP through 2020. As a result, a reverse relationship occurs with the deficit. The tax deal causes a two-year slowdown in the improvement of the deficit and the permanent extension and AMT fix causes a continuing slowdown with the deficit actually rising from 4.7% of GDP in 2014 to 6.2% of GDP in 2020.

The following graph shows the resulting gross federal debt and debt held by the public for the baseline, tax deal, and permanent extension and AMT fix:



The gross federal debt equals the debt held by the public plus the monies owed to the trust funds, chiefly Social Security. As can be seen, the tax deal causes an increase in both debts but one that does not grow over time. In fact the increase appears to lessen slightly, reaching 3.7% of GDP for both debts in 2020. On the contrary, the permanent extension and AMT fix causes both debts to continue a steady growth through 2020. The debt held by the public and the gross federal debt would be projected to climb to 90.3% and 120.8 percent of GDP by 2020, respectively. The 120.8% of GDP figure would be just short of the 121.7% of GDP peak that it reached at the end of World War II.

The above graphs suggest that a key concern is that the current tax deal not become a permanent extension. Unfortunately, the current situation shows how difficult it is to allow a long-term tax cut to expire, especially in a difficult economic environment. This would suggest that temporary long-term tax cuts are a very bad policy. If projections suggest that a long-term tax cut is not sustainable, the tax cut should be scaled back to a level where it is sustainable and to where no built-in expiration is required.

Will Higher Taxes Reduce the Deficit?

On November 21st, the Wall Street Journal ran an editorial titled "Higher Taxes Won't Reduce the Deficit ". The authors of the editorial are Stephen Moore and Richard Vedder. Moore is listed as a senior economics writer for The Wall Street Journal editorial page and Vedder is listed as a professor of economics at Ohio University and an adjunct scholar at the American Enterprise Institute. The editorial begins by mentioning that the draft recommendations of the president's commission on deficit reduction and a plan put forward by Alice Rivlin have proposed taxes as a part of reducing the deficit. It continues:

The claim here, echoed by endless purveyors of conventional wisdom in Washington, is that these added revenues—potentially a half-trillion dollars a year—will be used to reduce the $8 trillion to $10 trillion deficits in the coming decade. If history is any guide, however, that won't happen. Instead, Congress will simply spend the money.

In the late 1980s, one of us, Richard Vedder, and Lowell Gallaway of Ohio University co-authored a often-cited research paper for the congressional Joint Economic Committee (known as the $1.58 study) that found that every new dollar of new taxes led to more than one dollar of new spending by Congress. Subsequent revisions of the study over the next decade found similar results.

An initial problem that I have with the editorial is that it does not provide a link to the study or any of its subsequent revisions. I am very leery of any article that does not make it as easy as possible to check its sources. I make a point of giving all of my sources on my website and my blog. At most, such an article might motivate me go out and search for the data myself.

I was able to eventually find what appears to be a 1991 version of the study at this link, subtitled "The $1.59 Study". Note that this is one penny different from "the $1.58 study" cited in the editorial. This appears to have been an honest mistake because the paper contains the following endnote:

8. In 1987, we reported a $1.58 coefficient on the tax variable. Incorporating some data revisions to some variables, we now obtain a coefficient of $1.50 using the 1947-86 period, but the reported $1.59 using the 1947-90 period, suggesting the additional years strengthened the positive tax-spend deficit relationship.

Still, this mistake did have that effect of making it difficult to find the paper since googling "$1.59 Study" found the paper immediately but googling the "$1.58 Study" cited by the editorial did not. In any event, I took a quick look at the paper and noticed a couple of points. Graph 1 shows the estimated spending associated with each $1.00 increase in taxes for the four time spans. The following table summarizes that data:

time span  Estimated Spending per $1.00 Increase in Taxes
---------  ----------------------------------------------
1791-1825  -$0.03
1826-1860  +$0.20
1867-1913  +$0.72
1947-1990  +$1.59

As can be seen, the $1.59 figure is associated with just the last period of 1947 to 1990. The final sentences of the paper's conclusion refer to fact, stating the following:

Historically, there was a time when tax increases meant deficit reduction, but that time passed in the early part of this century. State and local governments still are able to constrain spending increases to levels equal to or less than the taxes raised. Why? We would tentatively suggest that the answer may lie in different institutional constraints, such as balanced budget amendments, spending limitation amendments, line-item vetoes, etc., measures that lower the marginal political benefits of new spending to political decisionmakers. In any case, the Federal fiscal problem is not likely to be solved without significant behavioral change on the part of those decisionmakers, and those changes are not likely given the current system of political rewards and costs.

The following graph shows the growth of spending from 1950 to 1990.



The actual numbers and sources for this graph can be found at this link. The graph and actual numbers show that, as a percentage of GDP, spending was generally increasing for 1947 through 1983, nearly all of the period in question. However, spending as a percentage of GDP dropped steadily from 1991 through 2000. I was unable to find one of the cited revisions to this study which might address this period. However, the editorial itself states the following:

Our research indicates this is a sucker play. After the 1990 and 1993 tax increases, federal spending continued to rise. The 1990 tax increase deal was enacted specifically to avoid automatic spending sequestrations that would have been required under the then-prevailing Gramm-Rudman budget rules.

The statement that "federal spending continued to rise" is highly misleading. As shown in the graph above, federal spending as a percent of GDP (the red line) sank sharply from 1991 through 2000. The first table at the following link shows that spending did rise measured in current dollars. However, any good economist would at least correct those numbers for inflation and likely population growth, if not GDP. The following table shows federal outlays in current and inflation-adjusted (FY 2005) dollars from 1990 through 2009.

       FEDERAL OUTLAYS (dollar amounts in billions)

            Actual Amount             Percent Change
      -------------------------  -------------------------
      Current  FY 2005  Percent  Current  FY 2005  Percent
Year  Dollars  Dollars   of GDP  Dollars  Dollars   of GDP
----  -------  -------  -------  -------  -------  -------
1990    1,253    1,832     21.9      9.6      6.3      3.3
1991    1,324    1,849     22.3      5.7      0.9      1.8
1992    1,382    1,858     22.1      4.3      0.5     -0.9
1993    1,409    1,846     21.4      2.0     -0.7     -3.2
1994    1,462    1,879     21.0      3.7      1.8     -1.9
1995    1,516    1,897     20.6      3.7      0.9     -1.9
1996    1,561    1,907     20.2      2.9      0.5     -1.9
1997    1,601    1,916     19.5      2.6      0.5     -3.5
1998    1,653    1,959     19.1      3.2      2.2     -2.1
1999    1,702    1,990     18.5      3.0      1.6     -3.1
2000    1,789    2,041     18.2      5.1      2.6     -1.6
2001    1,863    2,073     18.2      4.1      1.6      0.0
2002    2,011    2,201     19.1      7.9      6.2      4.9
2003    2,160    2,304     19.7      7.4      4.7      3.1
2004    2,293    2,378     19.6      6.2      3.2     -0.5
2005    2,472    2,472     19.9      7.8      4.0      1.5
2006    2,655    2,564     20.1      7.4      3.7      1.0
2007    2,729    2,565     19.6      2.8      0.1     -2.5
2008    2,983    2,704     20.7      9.3      5.4      5.6
2009    3,518    3,186     24.7     17.9     17.8     19.3

Source: Budget of the United States Government, FY 2011: Historical Table 1.3

As can be seen, outlays in FY 2005 dollars went up very slowly during this period. In fact, they went down from 1992 to 1993, reaching a level below their 1991 level. If they had also been corrected for population growth, they would have likely gone down over a much longer period. In any case, the editorial continues:

The only era in modern times that the budget has been in balance was in the late 1990s, when Republicans were in control of Congress. Taxes were not raised, and the capital gains tax rate was cut in 1997. The growth rate of federal spending was dramatically reduced from 1995-99, and the economy roared.

Of course, a dramatic reduction beginning in 1995 could not be due to a capital gains tax cut in 1997. The most recent tax change before 1995 was the 1993 tax hike. In addition, the above table shows the growth rate of federal spending over this period. As can be seen, the growth rate reached its minimum in 1993 and stayed low until 2000. Spending then began to rise sharply, even as a percentage of GDP, through 2009. Most of this was following the Bush tax cuts of 2001 through 2003. Hence, tax cuts don't appear to restrain growth in spending. This would suggest that growth in spending is largely independent of tax hikes and tax cuts.

For the above reasons, I find the Wall Street Journal editorial to be less than credible. At best, it suggests that we need to make a conscious effort to restrain spending, just as we did during the nineties.

Do Capital Gains Tax Cuts Increase Revenue?

At the end of my November 8th post, I asked for anyone who knows of an economic study that purports to show that any income tax cut has paid for itself to please post a link to it. On November 18th, I received the following reply:

I'm not an economist of any flavor, but I was wondering what you thought of the article by Stephen J Entin from the Institute for Research on the Economics of Taxation that I've linked below. He appears to argue that increased capital gains tax should decrease net federal revenues.

Any thoughts appreciated.

http://iret.org/pub/CapitalGains-1.pdf

The rest of this post will attempt to answer this reply:

My request at the end of my November 8th post did refer to income tax cuts, not capital gains tax cuts and those are the primary focus of my analysis which I referenced there. The only involved analysis of capital gains tax cuts that I have done are in my prior posts, here, here, and here. Following is my chief conclusion from that relatively short analysis:

The above discussion points to how difficult it is to study the relationship between capital gains tax rates and revenues. Hence, it is not intended to show a simple relationship but rather to show the problems with claims of simple relationships derived from looking at a few changes in the capital gains tax rate. That said, I did notice one interesting relationship in the data. From 1954 to 1982, there appeared to have been something of a positive correlation between the average capital gains tax rate and capital gains revenue. That is, they both tended in increase or decrease at the same time. After 1982, there appeared to be much more volatility in both the average tax rate and revenue and any obvious positive correlation disappeared. This would suggest that a more stable average tax rate might be desirable. This would lessen the need for investors to concern themselves with the timing of their stock transactions and allow them to concern themselves only with the long-term value of the investments themselves. If at any point that it is decided that the tax rate needs to be changed, it would likely be wise to phase in the change slowly.

Following is a graph of the data that this paragraph is referring to:



I did take a quick look through the paper you referenced and do have a few thoughts. First of all, I had some comments on the second paragraph which reads as follows:

However, taxpayers react to higher tax rates by earning and reporting less income. Higher taxes on capital retard capital formation and reduce wages across the board. The particular tax increases that the Congress and the Administration are most likely to adopt would damage the economy and reduce the tax base. In fact, they are likely to result in lower federal revenues, and larger budget deficits.

Except for the word "likely", these statements seems a bit absolute. In any event, this proposed effect is possibly countered by other effects. Consider the following excerpt from an article titled "Policy Points: Experts Agree That Capital Gains Tax Cuts Lose Revenue":

To raise revenue over the long run, capital gains tax cuts would need to have extraordinary huge, positive effects on saving, investment, and economic growth that virtually no respected expert or institution believes they have. In fact, experts are not even sure that the long-term economic effects of these capital gains tax cuts are positive rather then negative.

One reason is that preferential tax rates for capital gains encourage tax sheltering, by creating incentives for taxpayers to take often-convoluted steps to reclassify ordinary income as capital gains. This is economically unproductive and wastes resources. The Urban-Brookings Tax Policy Center’s director Leonard Burman, one of the nation’s leading tax experts, has explained, “shelter investments are invariably lousy, unproductive ventures that would never exist but for tax benefits.” Burman has concluded that, “capital gains tax cuts are as likely to depress the economy as to stimulate it.”

Secondly, I did notice that the paper seems to concede that it is in disagreement with the Treasury and Joint Tax Committee of the Congress. On page 3, it states:

The Office of Tax Analysis at the Treasury and the staff of the Joint Tax Committee of the Congress estimate revenues for the Federal budget. They also estimate changes in revenue associated with proposed changes in tax policy. They base their revenue forecasts on an assumed underlying state of the economy (the baseline economic forecast). The economic baseline specifies the amounts of total output, employment, capital, and the various types of income associated with them (the macro-economic conditions).

The revenue estimates of a proposed tax change are conducted on the assumption that the macro-economic baseline, with its aggregate levels of income and output, labor and capital, is fundamentally unchanged. This is called static revenue estimation. By contrast, estimation that allows for tax changes or other budget changes to affect macro-economic conditions, such as the size of the economy and the level of income, is called dynamic revenue estimation.

This disagreement is also mentioned by the CBPP article which also mentions the non-partisan Congressional Budget Office (CBO) and the Administration’s Office of Management and Budget as estimating that cutting capital gains rates reduces revenues over the long run. The paper does suggest in the excerpt above that this disagreement is because it is using a presumedly better dynamic revenue estimation versus static revenue estimation being used by the other studies. However, the CBO did do at least one study using various supply-side models in 2004 to analyze President Bush's proposals which included extending the tax cuts. As can be seen in the summary of that study at this link, those models estimated a significant cost just as did the conventional model. Now those proposals did include more than a capital gains tax cut. Still, it does show a case where the use of dynamic models by the CBO has had no significant effect on the estimated costs.

Thirdly, the conclusions of the paper seem to be based strictly on a theoretical model, the Cobb Douglas production function described starting on page 36. Although this model was doubtlessly analyzed against real data, the paper does not appear to do any analysis of recent data. Hence, I could only search for critiques of this model. This link states that the "Cobb-Douglas production function was largely abandoned after 1961" and there are some criticisms listed in the
Wikipedia entry for the model. It's not an easy thing to judge these criticisms but neither is it easy to judge the Cobb-Douglas production function itself.

In any event, the paper does seem to agree that there is a significant difference between the effect of income tax cuts and capital gains tax cuts. The following excerpt is from page 22:

The supply of labor is rather inelastic. Many primary workers (the main breadwinners in the households) are employed by others, and have limited ability to vary their hours worked (set by their employers) or the degree to which they participate in the work force (each family needs at least one breadwinner). Their hours worked do not vary a lot as wages rise or fall (so the supply curve in Chart 1 is steeply vertical.) Such workers are assumed to bear most of any taxes imposed on labor, including the income tax and the payroll tax, both the employee and employer shares. Secondary workers in the family, the self-employed, teenagers, and wealthier individuals have somewhat more flexibility in deciding whether or not to work, and how many hours to offer, but even they bear most of the tax on their labor income.

The effect of taxes on capital is quite different. The quantity of capital is far more sensitive to taxes than is the quantity of labor. (Its supply curve is more nearly horizontal.) It is easy and enjoyable to consume instead of save, and quite possible to invest abroad instead of in one’s own country. If the after-tax rate of return on saving and investment in the United States is driven down by an increase in the tax rate, U.S. capital formation may be curtailed, or shifted to other countries. The same phenomenon occurs within the country. When there is a tax increase on capital in any one jurisdiction, the amount of plant, equipment, and buildings in that region shrinks. As the capital becomes scarce, the rate of return on the remaining capital rises, until it is again earning a normal rate of return, after tax.

So this paper does not seem to be in any way suggesting that income tax cuts pay for themselves. Hence, I am still yet to find a single economic study that purports to show evidence that any income tax cut has ever paid for itself and would be interested in any that someone could post a link to. However, I am also interested in any additional studies that make that claim about capital gains tax cuts. Anyhow, I hope these thoughts were helpful to the commenter who asked for them. Feel free to leave any additional comments or questions.

Do Tax Cuts Increase Revenue?

On November 7th, David Stockman and Mike Pence appeared on This Week with Christiane Amanpour to debate tax cuts. David Stockman was a budget director for President Ronald Reagan and Mike Pence is a Congressman and has been the Chairman of the House Republican Conference for the past two years. A video of the tax cut debate can be found on the This Week website and a transcript can be found at this link. Following is an excerpt from that transcript:

PENCE: Before we get -- look, David Stockman is, you know, he's a really famous guy and a thoughtful guy. I just disagree with him vehemently and I, frankly, have for about 30 years. David believes that every tax increase equals a revenue increase, but that's not true. Anybody who is familiar with the historical data from the IRS knows that raising income tax rates will likely actually reduce federal revenues.

AMANPOUR: Let me ask David...

PENCE: So if we raise taxes, the American people are very likely going to -- the top 1 percent are going to send less money to Washington, D.C., and that will never get us out of this...

(CROSSTALK)

STOCKMAN: I just have to respectfully disagree. You will have some loss of revenue because some activity or transactions won't happen, but if you raise taxes on paper by $100 billion, maybe you'll get $90 billion or $85 billion. But it's just common sense fact that, when you raise the rates, you get more revenue. Normally, it's a bad thing to do. But we are in such dire shape that we have no choice but to accept the negative trade-off of some harm to the economy to start paying our bills. Otherwise, we're dependent on the Chinese, we're dependent on OPEC, we're dependent on a bunch of hedge fund guys to buy our debt, and this game is about over.

Pence states that "raising income tax rates will likely actually reduce federal revenues". This is in reference to the proposal to let all or part of the Bush tax cuts expire at the end of this year. The logical companion to this argument is the old supply-side claim that tax cuts increase revenues above what they would have been otherwise. I have looked at this argument and posted an analysis of it at this link. I could find nothing in the actual numbers to support this claim. In fact, I have asked supply-siders for years to provide the link to one credible economic study that purports to give evidence that supports this claim. As of yet, I have received none.

So far, every economic study that I have found that addresses the issue refutes the claim that tax cuts increase revenue. For example, there is a study titled "Dynamic Scoring: A Back-of-the-Envelope Guide" that was done by N. Gregory Mankiw and Matthew Weinzier. Mankiw was the chairman of President Bush's Council of Economic Advisors from 2003 to 2005. You can find a summary of the study at this link. It gives the following key quote at the top:

"In the long run, about 17 percent of a cut in labor taxes is recouped through higher economic growth. The comparable figure for a cut in capital taxes is about 50 percent."

This pretty much agrees with Stockman's statement that 10 or 15 percent of the revenue expected from a tax increase might be lost due to a decrease in activity or transactions. Both Stockman's statement and the quote above agree with the usual concept that a cut in the tax rate will lose that same percentage of revenue but that some of that loss will be recouped by higher economic growth. However, there are additional effects of tax cuts, negative as well as positive. Following is an excerpt from page 115 of the book "The Coming Generational Storm", co-written by economist Laurence Kotlikoff:

...For tax cuts to raise revenues, pretax labor earnings have to rise by a larger percentage than the tax rate falls.

There are two competing forces at play in determining whether pretax earning rise, stay the same, or fall. On the one hand, workers may say to themselves, "Boy, now that taxes are lower, I can work less and still receive the same after tax pay. I'm going to cut back my workweek." On the other hand, they may say, "Boy, now's a good time to work more and earn more because taxes are lower on every extra dollar I earn". Economists call the first of these reactions the income effect. They call the second reaction, the substitution or incentive effect.

Some of the best labor economists in the country have spent their lifetimes measuring the income and substitution effects. The broad consensus of these experts is that the two effects are roughly offsetting. This means that if wage tax rates are cut by, say 15 percent, tax revenues will fall by 15 percent.

Similarly, there are studies that suggest that certain tax increases can actually promote an increase in economic growth. An example is a study titled "The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks" by Christina and David Romer. Christina Romer recently resigned as the chairwoman of President Obama’s Council of Economic Advisers but will become a member of the President's Economic Recovery Advisory Board. The study found that "deficit-driven tax increases" have had a positive effect on economic growth. The following excerpt from page 14 describes these increases:

The figure makes clear that while deficit-driven tax increases have occurred throughout the postwar era, they were most prevalent in the early and mid-1980s. Many of these deficit-driven tax actions were related to Social Security: of the 26 actions in the category, 15 were designed to deal with the long-run solvency of the Social Security system. The Social Security Amendments of 1977 and 1983, in particular, were major tax actions that raised taxes in a number of steps and did not simultaneously increase benefits. The largest deficit-driven tax increases not related to Social Security were those contained in the Tax Equity and Fiscal Responsibility Tax Act of 1982, and the Omnibus Budget Reconciliation Acts of 1987, 1990, and 1993. The first two of these were Reagan-era measures; the third was the Bush tax increase that defied his famous "Read my lips: no new taxes" campaign speech; and the fourth was the Rubinomics-defining tax increase of the early Clinton administration.

The following graph from page 59 shows the estimated impact of a tax increase of 1% of GDP on GDP using long-run and deficit-driven tax changes:



The following excerpt from page 23 refers to this figure:

Panel (b) of Figure 6 shows the results for the two types of exogenous changes. For long-run changes, which make up most of this category, the results are quite similar to those for all exogenous changes. For tax increases to deal with an inherited budget deficit, the results are more interesting. The point estimates imply that output does not fall at all following deficit-driven tax increases. The estimated effect peaks at 1.4 percent after two quarters, and then fluctuates around 1 percent. However, there are too few tax changes of this type for the effects to be estimated very precisely. The maximum t-statistic, for example, is just 1.2. Nonetheless, the estimates are suggestive that tax increases to reduce an inherited deficit may be less costly than other tax increases, and they provide no evidence that they have substantial output costs.

The following excerpt from page 38 summarizes these effects:

In Section III, we found that the responses of real GDP to the two subcategories of exogenous tax changes appear to be quite different. The response to a long-run tax increase is negative, large, and highly statistically significant. In contrast, the response to a deficit-driven tax increase is positive, though not significant.

Finally, the following excerpt from the conclusions on page 41 expands on this positive effect:

Finally, we find suggestive evidence that tax increases to reduce an inherited budget deficit do not have the large output costs associated with other exogenous tax increases. This is consistent with the idea that deficit-driven tax increases may have important expansionary effects through expectations and long-term interest rates, or through confidence.

As I mentioned, I am yet to find a single economic study that purports to show evidence that any income tax cut has ever paid for itself. If anyone who reads this should know of one, please leave a comment with a link to that study. Thanks.

Long-term Unemployment (updated)

On July 2nd, the Bureau of Labor Statistics released its Employment Situation Summary for June of 2010. Following is the first paragraph:

Total nonfarm payroll employment declined by 125,000 in June, and the unemployment rate edged down to 9.5 percent, the U.S. Bureau of Labor Statistics reported today. The decline in payroll employment reflected a decrease (-225,000) in the number of temporary employees working on Census 2010. Private-sector payroll employment edged up by 83,000.

The drop of 125,000 in nonfarm payroll employment put the number of nonfarm payroll jobs at 130.47 million, just about the same level that existed in December of 1999. Hence, we have gone 10 and a half years with no job growth.

Perhaps more disturbing, however, has been the growth in long-term unemployment. On this topic, the December report stated the following:

In June, the number of long-term unemployed (those jobless for 27 weeks and over) was unchanged at 6.8 million. These individuals made up 45.5 percent of unemployed persons. (See table A-12.)

The following graph shows the average and median duration of unemployment since 1948:



Links to the actual data used to create this graph can be found at this link. As can be seen, the average and median duration of unemployment have risen to what is by far their highest level since 1948 and 1967, respectively. The average has reached 35.2 weeks and the median has reached 25.5 weeks weeks.

The following graph shows unemployment divided up into four durations:



As before, links to the actual data used to create this graph can be found at this link. As can be seen, the number of workers unemployed for 27 weeks and over reached 6.8 million in June. This is over double the prior high of 2.9 million reached in June of 1983. The number of workers unemployed for 15 to 26 weeks was 2.2 million in June, down from the recent high of 3.2 million reached in October of 2009. The number of workers unemployed for 5 to 14 weeks was 3.1 million in June, down from the recent high of 4.3 reached in May of 2009. Finally, the number of worker unemployed for less than 5 weeks was 2.8 million in December, down from the recent high of 3.6 million reached in January of 2009. Hence, all of the durations are starting to trend down except for the longest, those unemployed for 27 weeks and over.

It should be noted that neither of the above graphs include the Great Depression. However, they do show that long-term unemployment has by far reached its highest level since the end of World War II. This would suggest that long-term unemployment may require special attention in the effort to put the unemployed back to work.

Who is econdataus and why should we believe him?

Yesterday, I received an interesting question from Michael Fremer who has a blog at this link. Following is his question:

I sent my right wing friends some of your analyses and they gulp, but then respond "Who is econdataus and why should we believe him?"

And i can't answer that...

Can you?

My reply (with a few edits for readability) was as follows:

In regards to "who is econdataus", you can tell them that I am nobody they would know. I am a math major and software engineer. As I mention in my profile, I became interested in U.S. budget and economic matters back in 1992, the first time that I remember the debt becoming a major issue in a presidential election. Around that time, I became aware of more and more cases where public political discussion seemed to be based on bad or incomplete data. Hence, my first goal was to build a site with carefully compiled and sourced data in the areas in which I was interested. This gives me (and hopefully others) a ready source to point to when engaged in political discussions online. Giving the data and sources allows others to verify the data and judge the veracity of the sources.

The main analysis that I initially did was just to graph the data and list some fairly straightforward points (such as here). However, I later began to write an occasional analysis on an area in which I was interested (such as here) and at someone's suggestion, I started this blog. I did my best to keep it non-partisan. At this link, for example, I critiqued my own initial analysis stating "the difference in job growth between the two parties does greatly lessen, if not disappear, if one looks at job growth over entire business cycles". In any case, I truly believe that the country needs at least two strong parties to function at its best. And I do not believe that one party is inherently better than the other.

Regarding "why should we believe him?", I would say they should base their belief strictly on the data and their trust in the sources from which that data came. Regarding my analysis of the data, they should believe that only to the extent that it makes sense to them. As I mentioned, I am a math major and software engineer, not a professional economist. I have, to my knowledge, no public reputation, positive or negative. If they desire that, they will have to look elsewhere. The problem, of course, is that many of those reputable experts disagree. Hence, I would suggest that they demand sources for any data with which they are presented and verify it. I can't count the number of times I've come across data that is cherry-picked or just plain wrong. The only way to determine this is to ask for the source and verify the data.

R. Davis

To this, I would just add that they should question any analysis that doesn't totally make sense. In fact, it's probably best to question even those that do! As such, I am more than happy to answer any questions on my analyses or data.

The Risks of U.S. Sovereign Debt

From February 1st through the 4th, the Buttonwood's notebook blog of the Economist magazine had some interesting posts on the risks of sovereign debt. They were titled The debt trap: ranking the suspects, More debt rankings, The new gold standard, and Adding in the deficit. The first post addresses the effect of a debt trap, where the bond yield is higher than the economy's nominal growth rate. The second post adds in the effect of the debt-to-GDP ratio. Finally, the last post adds in the effect of the primary surplus/deficit. Also called the primary balance, this is equal to revenues minus spending, excluding interest costs. If the primary balance is positive, it's a primary surplus. If negative, it's a primary deficit.

The effect of these four factors (bond yield, GDP growth, debt-to-GDP ratio, and primary balance) are also addressed in a Reuter's column titled Sovereign debt maths show risk of vicious circle. The column mentions all four factors in the following excerpt:

To see how these play out, consider two countries. One has a moderate debt load, 50 percent of GDP, which carries a 4 percent average interest rate. If the budget is in primary balance, the government will still run a deficit of 2 percent of GDP, which is 4 percent (the interest rate) of 50 percent (the debt). As long as nominal GDP grows by 4 percent, the ratio of debt to GDP stays the same.

The validity of this final statement (that the ratio to GDP stays the same) can be seen via the following formula:

(new_debt) / (new_GDP) = (i * D - P) / (r * G) = (1.04 * D - 0.0) / (1.04 * G) = D / G

where i = interest rate
      D = debt (prior)
      P = primary surplus or deficit(-)
      r = rate of GDP growth
      G = GDP (prior)

The value of 1.04 in the above formula is of course the correct factor to calculate a 4 percent increase and is equal to (4/100) + 1. In any case, the column goes on to describe how this same formula applies to the country with the high debt load of 100 percent of GDP. It then continues:

But the equilibrium is fragile. It can be disturbed in three ways: nominal GDP growth can decline, interest rates can go up or the country can start running a primary deficit. The pain is much worse for highly indebted countries like Greece, which has managed all three at once.

Start with growth. Imagine nominal GDP growth drops to zero. If nothing is done, the debt/GDP ratio will rise by 2 percentage points in the moderately indebted country, but by 4 percentage points in the highly indebted one.

This situation can be represented by the following formula:

(new_debt) / (new_GDP) = (i * D - P) / (r * G) = (1.04 * D - 0.0) / (1.0 * G) = 1.04 * (D / G)

For the moderately indebted country, this will equal 1.04 * 50% of GDP which equals 52% of GDP, an increase of 2% of GDP. For the highly indebted country, this will equal 1.04 * 100% of GDP which equals 104% of GDP, an increase of 4 percent of GDP.

The column continues:

Countries can keep that key ratio from increasing, by running compensating primary surpluses. That means moving from balance to a surplus of 2 percent of GDP for the moderately indebted and from zero to 4 percent for the heavily indebted. The higher the debt level, the more the government’s belt will have to be tightened.

To verify this, one can first calculate the formula for the primary surplus in the case where the debt to GDP ratio does not increase as follows:

(D / G) = (i * D - P) / (r * G)  (old debt to GDP ratio equals new debt to GDP ratio)
(D / G) * (r * G) = (i * D - P)  (after multiplying both sides by (r * G))
(r * D) + P = (i * D)            (after simplifying and adding P to both sides)
P = (i * D) - (r * D)            (after subtracting (r * D) from both sides) 
P = (i - r) * D                  (after simplifying)

For the moderately indebted country, the necessary primary surplus will be

P = (1.04 - 1.0) * D = 0.04 * (G/2) = 0.02 * G = 2% of GDP

and for the highly indebted country, the necessary primary surplus will be

P = (1.04 - 1.0) * D = 0.04 * (G) = 0.04 * G = 4% of GDP

Hence, the above formulas show that countries with different levels of debt to GDP can maintain those levels as long as GDP growth keeps pace with the average interest rate and the countries run primary balances of zero. However, they also show that, if one or more of these factors fall out of balance, the country with the higher debt load will see greater negative effects.

The following graph shows the net interest on U.S. Treasury securities as a percentage of outlays, debt, and GDP, along with the nominal growth in GDP:



The actual numbers and sources for this graph can be found at this link. The net interest as a percentage of debt (the blue line) should give the average interest rate being paid on U.S. debt held by the public. As can be seen, the GDP growth rate was generally above the average interest rate from about 1951 to 1981 but was generally below the average interest rate from about 1982 to 2002. From 1951 to 1981, the average interest rate was about 4.2%, nearly half of the average GDP growth rate of about 8.2%. From 1982 to 2002, however, the average interest rate was about 7.3%, over a full percentage point more than the average GDP growth rate of about 6.1%. Hence, it is possible for the average interest rate to surpass the average GDP growth rate for an extended period of time.

The following graph shows the primary and unified surplus or deficit(-) of the U.S. budget:



The actual numbers and sources for this graph can be found at this link. As can be seen, the primary surplus appears to have been, on average, close to balance since 1950, at least until recently. In fact, from 1950 through 2007, the average primary surplus was about 0.14 percent of GDP. However, the graph shows that the primary surplus became a huge deficit in 2009 and is projected to just recover to about a one percent of GDP deficit by 2015. In addition, the tables at this link show that the primary deficit is projected to rise sharply over the coming decades. These numbers come from the most recent U.S. Budget and project that the primary deficit will reach a massive 24.3 percent of GDP by 2085.

Hence, the large primary deficits projected over the long-run would seem the most obvious risk to U.S. sovereign (government) debt. As explained in the above articles, however, investors could start demanding higher interest rates if they believe that these deficits increase the risk of holding government bonds.

On the topic of interest rates, Donald B. Marron, a visiting professor at the Georgetown Public Policy Institute, blogs:

According to the IMF, the average maturity of US debt is only 4.4 years. Portugal, Italy, Ireland, and Spain have maturities that are about 50% greater (from 6.2 to 7.4 years), and the UK is almost three times as long at 12.8 years.

The blog concludes:

The short maturity of US debt is a blessing in the short run, since we can benefit from lower interest rates. But it is also poses two risks in the long-run: greater exposure to interest rate increases (if and when they materialize) and a relentless need to ask capital markets to rollover existing debts. Both good reasons why Treasury should continue to gradually extend the maturity of federal borrowing.

The Long-Run Budget Outlook (2011 Budget)

As mentioned in my prior post, the U.S. Budget for fiscal year 2011 was released on February 1, 2010. As in prior years, it included the Analytical Perspectives which contains a section on the long-run budget outlook. The following graph shows the outlook for the federal receipts, outlays, and debt held by the public as projected by this section.



The actual numbers and sources for this and the following graph can be found at this link. As can be seen, receipts are projected to rise and spending is projected to drop over the next 10 years, causing the deficit to narrow to 4.2 percent of GDP by 2020. However, outlays are then projected to begin a steady rise, causing the deficit to reach a record 62.3 percent of GDP by 2085. This is projected to cause the debt held by the public to rise to 829.7 percent of GDP. This is over seven times the prior high of 108.7 percent of GDP reached in 1946, at the end of World War II. In addition, it is over double the level projected in the prior budget. The projections for this year and last year are the purple and green lines in the above graph.

Regarding the projected increase in the deficit and debt, Chapter 5 of the Analytical Perspectives says the following:

The key drivers of the long-range deficit are the Government’s major health and retirement programs: Medicare, Medicaid and Social Security.

Further on, it narrows this a bit, stating:

The most important single factor driving the long-run budget outlook is the growth of health care expenditures.

This growth in health care expenditures can be seen in the following graph which shows the projected growth in outlays according to the 2011 budget.



As can be seen, Medicare and Medicaid costs are projected to grow rapidly over the next 75 year. In fact, the first two tables at this link show that the cost of Medicare is projected to be greater than ALL federal receipts in 2085. This is over twice the cost that was projected in the prior budget. Chapter 5 explains the reason for the rapid rise in health care costs, stating:

In the base case, a continuation of the historical trend would see the per beneficiary cost of health care spending for Medicare, Medicaid, and private health care rising 2 percent per year faster than GDP per capita.

It goes on to list three major reforms being considered to slow this growth:

There are three broad reforms in the legislation under consideration in Congress that experts believe will produce significant savings relative to the historical trend: an excise tax on the highest-cost insurance plans will encourage substitution of more efficient plans with lower costs, while raising take-home pay; an independent payment advisory board will be empowered to suggest changes in Medicare and the health care system to improve the quality and value of its services; and an array of other delivery system reforms will gradually reduce costs.

These projections do not include any of these possible saving, however. As chapter 5 explains:

...but the baseline does not include these savings because the final form of the legislation was not resolved in time for the Administration to produce detailed estimates of its long-run effects.

Chapter 5 does go on to explore the projected effects of a number of alternative scenarios. They are summed up in the following table:

Table 5–2. FISCAL GAP UNDER ALTERNATIVE BUDGET SCENARIOS (Percent of GDP)

Baseline....................................................................  8.0
Health:
  Excess cost growth averages 1 percent.....................................  4.5
  Excess cost growth averages 1/2 percent...................................  2.8
Discretionary Outlays:
  Grow with inflation plus population.......................................  6.2
  Defense grows with inflation; nondefense grows with inflation + population  5.9
Revenues:
  Revenues exceed baseline by 2 percent of GDP..............................  6.4
  Revenues fall short of baseline by 2 percent of GDP.......................  9.6
Productivity:
  Productivity grows by 0.5 percent per year faster than the baseline.......  6.6
  Productivity grows by 0.5 percent per year slower than the baseline.......  9.6
Population:
  Fertility:
    2.3 births per woman....................................................  7.1
    1.7 births per woman....................................................  8.8
Immigration:
  1.3 million immigrants per year...........................................  7.5
  0.7 million immigrants per year...........................................  8.4
Mortality:
  Female life expectancy 82.7 years; male life expectancy 79.1 years in 2085  7.2
  Female life expectancy 89.9 years; male life expectancy 87.2 years in 2085  8.8

The accompanying text describes the fiscal gap as follows:

The fiscal gap is one measure of the size of the adjustment needed to preserve fiscal sustainability in the long run. It is defined as the increase in taxes or reduction in non-interest expenditures required to keep the long-run ratio of government debt to GDP at its current level if implemented immediately. The gap is usually measured as a percentage of GDP. The fiscal gap is calculated over a finite time period, and therefore it may understate the adjustment needed to achieve longer-run sustainability.

Table 5-2 shows fiscal gap calculations for the base case calculated over a 75-year horizon and for the various alternative scenarios described above. The fiscal gap in the base case is 8.0 percent of GDP, and it ranges in the alternative scenarios from 2.8 percent of GDP to 9.6 percent of GDP. In all cases, significant fiscal adjustments would be needed to achieve long-run sustainability.

As can be seen, the largest projected savings of 5.2 percent of GDP would come from cutting the average annual growth of health care costs from 2 percent to 0.5 percent above GDP per capita growth. The next largest projected savings would be just 2.1 percent of GDP if defense grows with inflation and nondefense grows with inflation plus population. Both are otherwise assumed to grow with GDP. In any event, both projected savings taken together would add up to 7.3 percent of GDP, still a bit short of the 8.0 percent of GDP baseline fiscal gap.

Budget of the United States Government for Fiscal Year 2011

On February 1, the U.S. Budget for fiscal year 2011 was released. Regarding the deficit, a Reuters article from that day stated the following:

The budget forecast a $1.56 trillion deficit in 2010, or 10.6 percent of gross domestic product (GDP), up from a 9.9 percent share of GDP in 2009.

But the shortfall was forecast to shrink to 8.3 percent of GDP in 2011. By the time his term ends in January 2013, it would have halved from the level Obama inherited on taking office last year, keeping a key pledge. That supposes Obama gets Congress to agree to spending cuts and that the economy rebounds strongly enough to sharply lift tax revenues.

The following graph shows selected measures of the deficit since 1970:



The actual numbers and sources can be found at this link. As can be seen, the numbers in the Reuters article are for the "unified deficit", shown in purple in the graph. The unified deficit is usually very close to the change in the debt held by the public (labelled "public deficit" in the graph above). As can be seen, however, they differ a fair amount in 2009 and 2010 with the "public deficit" reaching about 12 percent of GDP in those years. The reason for this difference is described on page 59 of the Analytical Perspectives from the 2011 Budget as follows:

In 2009 the deficit was $1,413 billion while these other factors—primarily the net disbursements of credit financing accounts—increased the need to borrow by $329 billion. As a result, the Government borrowed $1,742 billion from the public. The other factors are estimated to increase borrowing by $197 billion in 2010 and reduce borrowing by $66 billion in 2011. In 2012–2020, these other factors are expected to increase borrowing by annual amounts ranging from $59 billion to $145 billion.

In any event, the unified deficit equals federal receipts minus federal outlays. The following graph shows federal receipts and outlays since 1950:



As before, the actual numbers and sources can be found at this link. One thing that is very noticeable in the above graph is that outlays are projected to increase from about 20 percent of GDP from 2003 through 2007 to about 23 percent of GDP from 2012 through 2015. The following table shows the projected change in outlays by function over the eight years from 2007 to 2015. For comparison, it also shows the actual change in outlays by function over the prior eight-year span from 1999 to 2007.

     CHANGE IN OUTLAYS AS PERCENTAGE OF GDP

Superfunction and Function   1999-2007  2007-2015
---------------------------  ---------  ---------
National defense...........     1.0       -0.4
Human resources............     1.2        2.0
  Education, training, etc.     0.1        0.0
  Health...................     0.4        0.3
  Medicare.................     0.6        0.7
  Income security..........     0.0        0.2
  Social security..........     0.0        0.5
  Veterans benefits........     0.1        0.2
Physical resources.........     0.1       -0.3
Net interest...............    -0.8        1.3
Other functions............    -0.1        0.6
  International affairs....     0.0        0.2
  General science..........     0.0        0.0
  Agriculture..............    -0.1        0.0
  Administration of justice     0.0        0.0
  General government.......     0.0        0.0
  Allowances...............     0.0        0.4
Undistributed receipts.....    -0.2        0.1
Total, Federal outlays.....     1.2        3.2

Source: U.S. Budget, FY 2011, Historical Tables,
        tables 3.1 and 10.1

As can be seen, the functions with the largest projected increases in outlays are Net interest, Medicare, Social security, Allowances, and Health with increases of 1.3, 0.7, 0.5, 0.4, and 0.3 percent of GDP, respectively. In fact, these five functions could be said to account for the full 3.2 percent of GDP increase in all federal outlays with the other functions pretty much canceling each other out.

In any case, the projected 1.3 percent of GDP increase in Net interest follows a 0.8 percent of GDP decrease in the prior eight years. Hence, the increase in Net interest is likely due in part to a return to more normal interest rates and in part to an increase in the debt.

Medicare and Health (which consists chiefly of Medicaid) are projected to see a combined increase of about 1.0 percent of GDP from 2007 through 2015. This follows a combined increase of about 1.0 percent of GDP in the prior eight years. Social Security, on the other hand, is projected to increase by about 0.5 percent of GDP from 2007 through 2015. This follows no increase in the prior eight years. Table 5-1 on page 47 of the Analytical Perspectives from the 2011 Budget shows that Medicare and Medicaid are projected to continue growing as a percentage of GDP through 2085. Social Security, however, is projected to stabilize at about 5 percent of GDP after about 2010.

Finally, Allowances is projected to increase 0.4 percent of GDP from 2007 through 2015 when there was no increase in the prior eight years. Table 3.2 in the Historical Tables shows that the great majority of this increase in allowances is for Health Reform, chiefly in 2015.

Hence, the above table shows that the great majority of the increase in outlays that is projected through 2015 falls in two areas. The first area is entitlements, driven chiefly by health care costs but, to a lesser extent, by Social Security. The second area is Net interest. Since interest is pretty much an unavoidable cost, this suggests that the chief focus needs to be on the rise in the cost of health care and other entitlements. One-time increases in outlays, such as the 0.5 percent of GDP projected for Social Security, can likely be addressed by some combination of limiting the increase in outlays and increasing the revenues to pay for those outlays. On the other hand, continuing increases in outlays, such as those projected for health care costs, likely need to be addressed chiefly by limiting those continuing increases. This is especially the case for Medicare which is projected to increase from 3.1 percent of GDP in 2010 to 22 percent of GDP in 2085. Depending on ever-increasing revenues and/or cuts in other outlays to pay for such a large increase does not seem like a viable option.

Long-term Unemployment

On January 8th, the Bureau of Labor Statistics released its Employment Situation Summary for December of 2009. It stated that nonfarm payroll employment dropped by 85,000 in December. That put the number of nonfarm payroll jobs at 130.91 million, just about the same level that existed in February of 2000. Hence, we have gone nearly a decade with virtually no job growth.

Perhaps more disturbing, however, has been the growth in long-term unemployment. On this topic, the December report stated the following:

Among the unemployed, the number of long-term unemployed (those jobless for 27 weeks and over) continued to trend up, reaching 6.1 million. In December, 4 in 10 unemployed workers were jobless for 27 weeks or longer.

The following graph shows the average and median duration of unemployment since 1948:



Links to the actual data used to create this graph can be found at this link. As can be seen, the average and median duration of unemployment have risen to what is by far their highest level since 1948 and 1967, respectively. The former has reached 29.1 weeks and the later has reached 20.5 weeks weeks.

The following graph shows unemployment divided up into four durations:



As before, links to the actual data used to create this graph can be found at this link. As can be seen, the number of workers unemployed for 27 weeks and over reached 6.1 million in December. This is over double the prior high of 2.9 million reached in June of 1983. The number of workers unemployed for 15 to 26 weeks was 2.8 million in December, down from the recent high of 3.2 million reached in October of 2009. The number of workers unemployed for 5 to 14 weeks was 3.5 million in December, down from the recent high of 4.3 reached in May of 2009. Finally, the number of worker unemployed for less than 5 weeks was 2.9 million in December, down from the recent high of 3.6 million reached in January of 2009. Hence, all of the durations are starting to stabilize or trend down except for the longest, those unemployed for 27 weeks and over.

It should be noted that neither of the above graphs include the Great Depression. However, they do show that long-term unemployment has by far reached its highest level since the end of World War II. This would suggest that long-term unemployment may require special attention in the effort to put the unemployed back to work.

Interest Cost of U.S. Treasury Debt

Following is an excerpt from a Wall Street Journal article from December 31, 2009:

The Treasury sold more than $2.1 trillion in notes and bonds this year, more than in the previous two years combined, to fund a widening budget shortfall and finance programs to rescue the banking system and support the economy.

Yet, despite the supply onslaught, buyers—from foreign central banks to U.S. households and domestic commercial banks—flocked to the sales. As a result, the government's borrowing costs fell to historic lows in 2009.

Regarding these borrowing costs, the Treasury Department posts the average interest rates on U.S. Treasury Securities at this link. This shows the monthly averages from 2001 through 2009 and was apparently used to create the graph at this link. The table for December 2009 shows that the average interest rates for Treasury Bills (which have a maturity of one year or less) fell to a mere 0.237 percent at the end of 2009, compared to 1.082 percent at the end of 2008. The average interest rates for all interest-bearing debt fell to 3.29 percent at the end of 2009, compared to 3.866 percent at end of 2008. This second pair of numbers is a good amount higher than the first pair since it includes longer-term debt paying at even higher rates.

The following graph gives a longer-term view of the average interest rates on U.S. treasury securities:



The actual numbers used to create this graph can be found at this link. The red line shows the average interest rate paid on the gross federal debt and the blue line shows the average interest rate paid on the debt held by the public (see the note and table at the bottom of this page for the difference between these two debts). As noted at the bottom of this page, the gross interest rates were calculated by dividing the gross interest by the average gross federal debt for each year. Similarly, the net interest rates were calculated by dividing the net interest by the average debt held by the public for each year.

The yellow and green lines are calculated the same as the red and blue line except that that the denominator used is the gross domestic product (GDP). Hence, the yellow line is the gross interest divided by the GDP for each year and the green line is the net interest divided by the GDP for each year.

The graph shows that, since 1962, the average interest rate on the gross debt has been slightly more than it has been on the debt held by the public. This difference reached a maximum of about one percent in the mid-eighties and is projected to be at about that level from 2009 to at least 2014. The two rates generally rose from 1962 to a high of 11 and 10 percent in 1982 and then declined. They are projected to reach a low in 2010 and then start increasing again.

The graph also shows the annual change in the GDP and the Composite Outlay Deflator as the purple and light blue lines, respectively. As can be seen, GDP grew a good bit faster than the average interest rates from 1962 to 1981 but grew somewhat slower from 1982 to 2002. On the other hand, inflation (as measured by the composite outlay deflator) grew at about as fast as the average interest rates from 1962 to 1981 but much slower after that. In fact, the following table shows the averages for all of these measurements from 1962 through 2008:

Measurements from 1962 through 2008              Average
------------------------------------------------ -------
Annual increase in Gross Domestic Product (GDP).  7.251
Average interest rate on gross federal debt.....  6.482
Average interest rate on debt held by the public  6.007
Average increase in composite outlay deflator...  4.230

Hence, it is the case that, since 1962, the the average interest rate on the gross federal debt has been slightly above that for the debt held by the public. Also, both rates have been a bit below the growth in the GDP but well above the rate of inflation.

Finally, the graph shows that by 2014, the interest costs as a percentage of GDP are projected to rise to levels not seen since the late 1990s. However, the levels in the late 1990s were based on interest rates of over 6 percent whereas the levels in 2014 are based on lower rates of 4.5 and 2.5 percent respectively. This would suggest that higher rates of 6 percent or so would bring interest costs as a percentage of GDP to their highest level since at least 1962.