Asymptosis

I’m going to start this post with a proleptic response:

No, nobody is suggesting a Maoist cultural revolution, forcing executives and professors to muck shit on collective farms.

But I feel compelled to share the results from my latest, comparing income inequality to prosperity and prosperity growth in the fifty states.

Contrary to the supply-sider narrative, income inequality appears to have almost no correlation with prosperity growth (.05). What correlation it does show is the opposite of what the supply-siders would have you believe.

This suggests — also contrary to all-too-commonly received wisdom — that there is not in fact an inevitable trade-off between equity and efficiency. That it’s a false choice. That we can in fact have more of both.

Update: I erroneously attributed the following post to Tyler Cowen instead of Alex Tabarrok. Fixed in the title.

Marginal Revolution: The Dark Magic of Structured Finance.

I’ll let you read it yourself, but here’s the takeaway. If the chance of underlying mortgages’ defaulting goes from 5% to 6% (a 20% increase),

the probability of default in the 10 tranche[s] jumps from p=.0282 to p=.0775, a 175% increase.  Moreover, the probability of default of the CDO jumps from p=.0005 to p=.247, a 45,000% increase!

These securities don’t just leverage the risk; they massively leverage the risk of risk.

It never ceases to amaze me how tiny technical details can have massive implications. Here’s perhaps the biggest example I’ve ever seen.

In a recent post I cited the 1987 Long-Term Capital Management hedge-fund fiasco, which occurred because they were assuming a normal distribution bell curve for financial events. “A one-in-five-hundred-year event on a normal bell curve was a one-in-five-year event in the actual distribution of market movements.”

Now the New York Times (from Reuters) reports that even 20 years after the 1987 LTCM meltdown, Wall Street risk mangers were still using that la-la land distribution. August, 2007:

“We were seeing things that were 25 standard deviation moves, several days in a row.”
–David Viniar, chief financial officer, Goldman Sachs

25 standard deviations means about one chance “in the lifetime of a billion universes.” Think it might be time to take another look at that model?

And even today, after the 08/09 catastrophe — at least according to this article — many still use the “normal” or “Gaussian” distribution — what you get if you throw a bunch of pennies on the ground, measure all their distances from the center, and plot those distances on a histogram.

Why do they assume that the penny distribution is the same as the distribution of financial events encompassing hundreds of trillions of dollars and millions of (presumably, sort of) intelligent agents? No reason that I can discern. And I can’t find anyone who has attempted to plot the actual distribution of market events. (Pointers to the contrary from my gentle readers much appreciated.)

The closest, cited in the article, was the mathematician Benoit Mandelbrot (of fractal fame, among others). In 1962 he looked at a hundred years of cotton prices, and plotted a bell-curve for that actual emprical data.

So here’s the tiny detail: the shape of a bell curve is determined by a single simple number — the exponential rate of decay. The normal, gaussian distribution has an ERD of 2. Flatter bell curves with fatter tails — more unusual events — have lower numbers. Mandelbrot came up with 1.7 for the distribution of cotton-market events. According to Reuters, with some exceptions the normal, gaussian distribution is still “the one that is actually used.”

I recommend giving it a read for more details. It’s quite short.

Commenters on my cross-post over at Angry Bear have asserted among other things that real GDP/capita growth has been steady for more than a century.

See? This proves, they assert, that the policy differences between Republicans and Democrats don’t really affect anything.

My response was to point out how hard it is to eyeball important changes in this graph — it looks like a straight line with blips for the Depression and WWII. It’s very hard to see what in fact constitute important or even profound changes in our well-being.

In particular, look at growth from 1820 to 1929, and consider how the trend compares to post-war (i.e. post-New Deal) performance:

Growth moved above trend starting the 60s, and has continued doing so since. (You could move the trend line down a little without departing the facts — passing it through the center of the pre-1929 movements rather than the ’29 top — making post-war growth look even better by comparison.)

We can say at least this with complete certainty: the disastrous long-term results predicted for (and contrafactually attributed to) New Deal policies did not occur.

Cross-posted at Angry Bear.

Thanks to yeoman’s work by Larry Bartels, Mike Kimel, and a host of others, we’ve seen that over many decades, the American economy has performed far better, by almost any measure, under Democratic presidents.

Larry Bartels’ key graph mapping income growth by quintile, 1948–2005 (from page 33 of Unequal Democracy) is perhaps the best demonstration of that.  Even the rich get richer, faster, under Democratic presidents. The poor and the middle class get far richer:

All those findings came as a surprise to me when I first saw the data. I’d pretty much accepted the Republican “party of growth” party line, after hearing it repeated thousands of times over several decades. It turns out that at least for American presidents, it just isn’t true. Not even close.

Even though the data’s been sliced, diced, and analyzed every which way from Sunday, with consistent results, critics still try to second-guess it. Many or most of these objections are spurious — statistically illiterate, logically flawed, or just plain self-contradictory special pleading.

But at least one repeated question does bear examination: What about Congress? Don’t they supposedly control the purse-strings?

Here are some questions — some of which may seem to have “obvious” answers, though we really can’t know until we look systematically:

1. Do Democrats and Republicans in the different branches deliver systematically different results in spending, revenues, and deficits (and more tenuously, second-order effects like real GDP per capita)?

2. What are those party differences? Are they large differences?

3. Do those difference themselves differ depending on whether the parties are in control of the Senate, the House, and/or the Presidency? IOW, do Dems/Pubs act differently in the House than they do in the Senate or the presidency?

4. Which branches have the greatest effects on different economic measures?

5. In which branches do Democratic and Republican results differ the most? Does that vary depending on the economic measure you’re looking at?

6. Do certain party combinations in the Senate/House/Presidency show systematically different results in spending, revenues, etc.? Is there any sign of an optimal or dystopic combination?

7. Do there seem to be systematic differences between mixed and monolithic control of the branches?

8. Directly addressing the objections mentioned above: does an analysis of congressional results undercut, disprove, or otherwise alter the conclusions from the research on presidential outcomes linked above?

To look into these questions I built a spreadsheet for the years 1961–2009, based on one built by my friend Steve, who tagged all the years for which party controlled each branch. (Thanks, Steve!) The spreadsheet (with data sources cited) is here.*

I chose to start in ’61 not only because the period basically spans my sentient life (self-serving bias?), but because 1. it’s a reasonably long period (49 years), and 2. by then the wild economic swings following the Depression and World War II had settled down. If anyone wants to extend the period backward, it’s an easy matter to add the data from the sources cited in the spreadsheet.

You can jump down to the resulting graphs, or even the summary, but some discussion is in order first.

Details

The central issue here: there are many different ways to look at this data. I’ve chosen ones that I think will give the most inclusive, comprehensive, and comprehensible insights into the questions above, revealing both the big picture and the sometimes messy, contradictory, and ambiguous results. Given the nature of the data, it simply can’t answer some of the questions  above.

I detail my choices here. I encourage others to use the spreadsheet to create different views — hopefully widely representative ones, eschewing intentional (or even unconscious) cherry-picking. My choices were driven by just plain curiosity about the “facts on the ground.” I had some notions of what I might find (Dems tax and spend? Pubs borrow and spend?), but given how wrong I was on presidents and economic growth, I wanted to see the numbers. I encourage others to operate from similar principles.

For each year, I entered the change in spending, revenues, debt, and real GDP per capita. Then I built “lags,” so we can look at results in ensuing years. (This assuming that it takes some time to implement then see the effects of policy changes; it doesn’t happen instantly.)

Here’s the upper left corner of the spreadsheet. The “change” percentages are from a simple calc: (year2-year1)/year1.

							CHANGES IN:
							SPENDING AS % OF GDP
year	senate	house	president	dominant	congress	Mix	spend0	spend1	spend2	spend3
1961	D	D	D	D	DD	DDD	1.51%	-1.31%	-0.23%	-0.21%
1962	D	D	D	D	DD	DDD	-1.31%	-0.23%	-0.21%	-1.54%
1963	D	D	D	D	DD	DDD	-0.23%	-0.21%	-1.54%	0.49%
1964	D	D	D	D	DD	DDD	-0.21%	-1.54%	0.49%	2.35%
1965	D	D	D	D	DD	DDD	-1.54%	0.49%	2.35%	0.67%
1966	D	D	D	D	DD	DDD	0.49%	2.35%	0.67%	-0.39%
1967	D	D	D	D	DD	DDD	2.35%	0.67%	-0.39%	0.92%
1968	D	D	D	D	DD	DDD	0.67%	-0.39%	0.92%	0.49%
1969	D	D	R	D	DD	DDR	-0.39%	0.92%	0.49%	-0.13%
1970	D	D	R	D	DD	DDR	0.92%	0.49%	-0.13%	-1.58%
1971	D	D	R	D	DD	DDR	0.49%	-0.13%	-1.58%	0.45%

As an example of lagging, the “spend1” data is simply shifted up one year, meaning that the parties controlling the Senate, House, and Presidency get “credit” (or blame) for changes that happen a year later. Ditto for the two-, three-, and four-year lags.

Note that lagging by an additional year means there’s one less year (at the end) to evaluate. We can’t look at a three-year lag for 2008, for instance, because 2011 hasn’t happened yet. Since we’re averaging over a 49-year period, it’s to be hoped that loss of those later years will not corrupt the results excessively. (This gets more problematic when you start slicing up the data; more on that below.)

Since I don’t use any of the zero-year-lag results (I think for obvious reasons, mentioned above, which necessitate the use of lags in the first place), the graphs below don’t tell us anything about Obama’s results in 2009. Previous years’ regimes are analyzed based on the economic changes in 2009, but we have no lagging data on ensuing economic changes with which to analyze 2009. (If anyone wants to add projected data for 2010 and beyond, feel free. Just be sure to tell your readers.)

Yes, this means that the Pub presidents get blamed for 2009, but so does the Democratic congress in ’07 and ’08. Carter gets blamed for some rocky years under Reagan in ’81 and ’82. And etc., throughout the table. This is why we need to look at multiple lags and many years; it serves to average out all the allocations that might seem “unfair” in the particular.

Another method would look at the total (summed or averaged) changes over one-, two-, three-, and four-year ensuing periods. As you’ll see in the graphs below, I chose to summarize similarly by averaging the multiple lag results. I encourage others to try the other methods. I can only display so many graphs here.

As I mentioned above, the analysis gets problematic quite quickly when you start slicing the data into smaller pieces. Our short, 49-year sample quickly rears its head.

Smaller, shorter slices, for instance, might only reflect where they happen to land in the business cycle. And slices concentrated at the beginning or end of the period might be distorted by long, secular trends (geopolitical, cultural, technological) that have nothing to do with party control — Matthew Arnold’s “ebb and flow of human misery.”

Here’s the number of years since ’61 for different regime combinations (Senate/House/President):

RRD	6
RDR	6
RRR	6
DDD	15
DDR	16

A comparison of DDD to DDR looks like the most promising in this set of slices because they’re both lengthy and scattered through the 49-year period. Since the only difference is in presidential control, that comparison may have special applicability to the presidency question that launched this investigation, #8.

The smaller slices will be much more subject to business cycles and secular trends. For instance, we’ve got three six-year, single-president slices. RDR is a single span from ’81–’86, when Volcker and Reagan, with a mixed Congress, were trying to pull us out of the biggest recession since World War II (and succeeding) and a 40-year cold war (eventually succeeding). It seems foolish to draw any grand, sweeping conclusions from that single and arguably anomalous slice.

Likewise: RRD is 1995 to 2000 (Clinton), and RRR is 2001 to 2006 (Bush II) — the first a massive boom (bubble?), the other a recession and tepid recovery. Hard to make useful comparisons.

DDD and DDR, by comparison, reflect sixteen- and fifteen-year slices scattered from the beginning to the end of the period.

Anyone can feel free to write narratives based on comparing all these slices. I’ve presented them in the graphs below. But with the exception of DDD and DDR comparisons, I would suggest that those spins will be statistically spurious. You’re just telling stories about single presidencies, which is exactly what we’re trying to overcome here.

Here are the Senate/House combos:

DD	31
RD	6
RR	12

In those 49 years we’ve never had a Democratic Senate and Republican House. And in only six years have we had mixed Congressional control (those six Reagan years). These two facts, especially combined, make it largely impossible to throw any real light on question #7 (mixed versus monolithic control) with this analysis.

DD is a large and scattered sample with both R and D presidents. RR is a single span late in the period — 1995–2006 — so it’s more subject to distortion by longer-term trends, but it’s half under Clinton, half under Bush II. RR and DD comparisons might yield insights.

Larry Bartels ran into these same difficulties when trying to add the effects of the legislature to his regressions. He uses different phrasing in his explanation, which I’ll let you peruse in a footnote.*

If you’re so inclined, you can also slice the data in the spreadsheet by the “dominant” party — the one controlling two out of three branches. I didn’t find much of interest there, but go to town. Again, I can only display so many graphs without it all becoming meaningless spaghetti.

The Graphs

One more bit of explanation — of the graphs themselves — is necessary here.

I’ve made the assumption that parties can have a somewhat immediate effect on taxes, spending, and debt. So for those measures I’ve given the average of only the one- and two-year lags. I’ve shown all four bars, though, so you can see the numbers and, if you think it’s useful, eyeball-average them.

For changes in real GDP per capita — which is a second-order effect and hence (in [my] theory at least) takes longer to appear — I’ve given the average of all four lags. You can produce any other calculations or presentations you want using the spreadsheet.

The highlighted averages are generally in the vertical position indicating their values, except where I’ve had to move them to avoid obscuring information.

That said, here we go.

Taxes

Revenues as a percent of GDP. This means there’s no need to correct for inflation, with the inherent uncertainties of that correction. Short story, everybody cuts taxes except Democratic presidents.

President

Democratic presidents raise taxes. Republican presidents lower them. Alert the media. (This graph shows nicely why I don’t pay much attention to three- and four-year lags for changes in revenues, spending, and debt.)

Senate

Surprise: Dem senators are tax cutters — just not as much as Pubs. For both parties, the effects are dwarfed by presidential differences.

House

Pretty much the same as the Senate, except that Dem House members cut taxes more than Republicans.

Congress

DDs and RRs are remarkably similar, with small magnitudes.

Combined

The DDD and DDR results are quite similar to the presidential results, above.

Spending

Again, as a percent of GDP so we needn’t correct for inflation. Republican presidents and Democratic senators are the big spenders, followed by Democratic house members. Democratic presidents are as frugal as Republican legislators.

President

Republican presidents raise spending significantly faster than Democrats.

Senate

Dem senators, on the other hand, increase spending something like eight times as fast as Republicans. The data reads darned consistently here, and unlike with taxation, the magnitudes are similar to presidential changes.

House

Almost the same as the Senate: Democrats increase spending faster than Republicans, and with magnitudes similar to the Senate and Presidents.

Congress

Again very similar to the Senate and House. Democrats in Congress increase spending much faster than Republicans.

Combined

DDD vs. DDR basically reflects the presidential differences.

Debt

As a percent of GDP. Somebody might want to add columns to the spreadsheet showing deficit changes, and look at it that way instead.

President

Profoundly large magnitudes and differences here, and remarkably consistent even into the three- and four-year lags.

Senate

Democratic senators, unlike presidents, increase the debt — but nowhere near as fast as Republicans. Smaller magnitudes here than with presidents.

House

Not so the House. Republican congresspeople are more frugal than their Democratic colleagues. Magnitude a bit less than the Senate, far less than presidents.

Congress

DDs and RRs are basically the same.

Combined

DDD versus DDR again mirrors the presidential results.

Real GDP per Capita

This measure is, necessarily, corrected for inflation. Changes in real personal income, a la Bartels, would arguably be a good alternative or additional measure.

President

Lags two and three show the parties in parity. Lags one and four create a Democratic advantage to the tune of .5% extra growth per year.

Senate

Very consistent, and very little difference between the parties. Magnitudes similar to presidents’.

House

Also fairly consistent (though the one-year lag shows party parity) and with similar magnitudes, with the Democrats associated with faster growth.

Congress

Democratic congresses are associated with higher GDP growth than Republicans.

Combined

DDD and DDR again mirror presidential results, though somewhat more pronounced.

Summary

Here’s a summary of all the average changes. (Remember: revenues, spending, and debt changes are averaged over the one- and two-year lags. Real GDP per capita changes are averaged over all four lags.)

Conclusions

Here’s my narrative. You can call it spin, but I hope you’ll at least agree that it’s not contradicted by the facts as they’re analyzed here. Feel free to copy and paste these graphs (credit please), or create your own, and write your own story.

1. On revenues and spending,  presidential effects appear to overwhelm congressional effects. This is not surprising when you consider the presidential drives behind Reagan and Bush II’s tax cuts and defense buildups, Johnson’s Great Society plus Vietnam, Clinton’s tax increases, etc.

2. Everyone increases spending (Republican legislators just barely) — Democrats significantly more than Republicans.

3. Only Democratic presidents have the political courage to raise taxes. (Given this, their ability to get elected at all says a lot for their popularity.)

4. Of Republicans, only House members hold the line on debt, due to tiny changes in both revenues and spending.

5. Democratic presidents are the only ones to reduce the debt — and they do so in a big way.  (See #3.)

6. Overall, Democratic control is associated with greater growth in real GDP per capita. (In the Senate, Pubs and Dems are essentially the same.) To do the arithmetic for you: if Republican presidents had managed equal increases during their 28 years in power, GDP/capita would be approximately 15% higher than it is today — $53K per person per year, as opposed to $46K.

And if Larry Bartels’ graph is any indication, far more of that (greater) GDP would be flowing through the hands of the middle class.

That’s my story and I’m sticking to it (unless you convince me otherwise).

————————

* You’ll need to know, or learn to use, Excel’s pivot tables — which you should do in any case cause they kick ass. I’ve given brief instructions in the spreadsheet. The graphs require some slightly sophisticated hand-tuning (data sources, label displays, axes, etc.) depending on what you’re trying to show.

** From Unequal Democracy, page 34 (screen grab so I don’t have to type it in):

Just following up on my recent post showing that progressivity in state taxes seems to have no significant relationship to prosperity:

These findings suggest to me that the supposed tradeoff between equity and economic efficiency is a false choice (by this measure, at least). More equitable states (far more equitable) are just as prosperous, at least in aggregate, as states with wildly draconian tax regimes.

So if we can have both equity and efficiency, why shouldn’t we?

File under: Painfully Strained Mixed Metaphors. (Unless: “with” means “using.” But still.)

The Norquististas want to make government small enough to drown it in a bathtub.

Bruce Bartlett was there when the Republicans ginned up this ideology. He participated. And he takes it down here.

In effect, STB became a substitute for spending restraint among Republicans. They talked themselves into believing that cutting taxes was the only thing necessary to control the size of government. Thus, rather than being a means to an end—the end being lower spending—tax cuts became an end in themselves, completely disconnected from any meaningful effort to reduce spending or deficits.

I would just disagree on one point: tax cutting has not been an end in itself for Republican politicians. It very quickly turned into an utterly intentional, Machiavellian political strategy to buy votes.

The Reaganomics Strategy.

It worked brilliantly, and unfortunately it still does.

I posted recently about how profoundly regressive state and local taxes are, with my home state of Washington being the very worst. A new initiative proposal by Bill Gates Sr. to institute a state income tax on high earners (while reducing business and property taxes) prompted me to revisit the issue.

My question: are states with more progressive taxes more or less prosperous? Do their economies grow faster or slower?

Short answer: no. In aggregate, there’s almost no difference between states with more-progressive taxes and those with less progressive regimes.

Even though both progressivity and economic growth vary wildly among states, there’s only a very small positive correlation between progressivity and growth in prosperity (change in Real GDP per capita):

And there’s a small negative correlation between progressivity and prosperity (2008 GDP per capita):

Both of these correlations (.06 and -.05) are too small to tell us anything at all. Either progressivity has no effect on these measures, or we can’t tell what the effect is.

For those who want the details or to play around with the numbers, the spreadsheet’s here.

Update May 7: I forgot to mention that I looked at this same subject before, comparing prosperous countries. More progressive countries seem to grow slightly faster. (Positive correlation: .13.)

I got curious about this. Here’s what WikiPedia says:

——————-

Oil spills of over 100,000 tonnes or 30 million US gallons, ordered by tonnes^[a]

Spill / Tanker	Location	Date	Tons of crude oil	Reference
Gulf War oil spill	Persian Gulf	January 21, 1991	1,360,000–1,500,000	[19][20]
Ixtoc I oil well	Gulf of Mexico	June 3, 1979–March 23, 1980	454,000–480,000	[21]
Atlantic Empress / Aegean Captain	Trinidad and Tobago	July 19, 1979	287,000	[22][23]
Fergana Valley	Uzbekistan	March 2, 1992	285,000	[20]
Nowruz oil field	Persian Gulf	February 1983	260,000	[24]
ABT Summer	700 nautical miles (1,300 km) off Angola	1991	260,000	[22]
Castillo de Bellver	Saldanha Bay, South Africa	August 6, 1983	252,000	[22]
Amoco Cadiz	Brittany, France	March 16, 1978	223,000	[20][22]
Amoco Haven tanker disaster	Mediterranean Sea near Genoa, Italy	1991	144,000	[22]
Odyssey	700 nautical miles (1,300 km) off Nova Scotia, Canada	1988	132,000	[22]
Sea Star	Gulf of Oman	December 19, 1972	115,000	[20][22]
Torrey Canyon	Scilly Isles, UK	March 18, 1967	80,000–119,000	[20][22]
Irenes Serenade	Navarino Bay, Greece	1980	100,000	[22]
Urquiola	A Coruña, Spain	May 12, 1976	100,000	[22]

^a One tonne of crude oil is roughly equal to 308 US gallons, or 7.33 barrels.

——————-

News reports say the current Gulf of Mexico well is leaking 200,000 gallons a day, which comes to 650 tons. Do that for a month and it’s about 20,000 tons.

And did I just miss it, or was there essentially no mainstream coverage of the Gulf War oil spill?

The Exxon Valdez spilled about 37,000 tons.

I’m wondering: how much was spilled during the Battle of the Atlantic in WWII?