Unused Roads

I don't remember how I first became aware of this, but several years ago while doing research on the recession, I came across figures for miles driven on US highways. What was fascinating about them was that the figures perfectly corresponded to the dates set as the beginning and end of the recession. Miles driven peaked in November of 2007, fell for the following 18 months, and started to recover in June of 2009. I figured that the number would make its way back up and that would be the end of the story. But it wasn't. I checked again some months later, and the number had actually declined. As it sits now, the number has been improving since November 2011, but it's still far below it's 2007 high.

As it stands, total miles driven in the US are currently the same as in 2004. A closer look at the numbers reveals that the decline may have begun earlier. Up until 2002, monthly changes in miles driven closely matched a linear trend line, increasing around 5 billion miles every month. However, since November 2002, every month has been below what a linear trend line would predict. Even after switching to a polynomial trend line, the predicted and the actual results begin to diverge after July 2004. If the linear trend from 2002 had been maintained, as it had been for 15 years, miles driven would be 3.4 trillion instead of 2.9 trillion.

There is one more thing to consider, the US population has increased by 11 million since the recession began. So whereas total miles driven has more or less flat-lined since 2009, per capita miles driven continues to decline.

While total miles driven is the same as 8 years ago, miles per person is the same as 14 years ago. Similar to total miles driven, miles per person used to follow a linear trend line, but it began to diverge in December 2000.

The big question is, what does this mean? The sharp decline after 2007 is easily explained as a result of the recession while the continued decline could reflect the weakness of the recovery. However, slowing trends could be seen as far back as 2000. The lack of recovery in highway might not reflect the recession so much as a change in driving behavior. There are studies showing that car ownership is declining especially among young people. A further question is, will the current trend continue or will a confluence of improving economic conditions and population increase drive the number back up?

Election Spending

Total campaign spending per vote in federal elections (in 1998 dollars).

2010: $30.15
2008: $30.69
2006: $28.76
2004: $29.03
2002: $24.71
2000: $27.50
1998: $22.14

Mexican Border

I recently read a review of Robert Kaplan's book "The Revenge of Geography" wherein he writes about how geography shapes society. One part of interest to me, but won't expand on in this post is the importance of Ukraine to Russia in whether or not Russia takes a more European or Asian orientation. What I want to talk about here is actually a quote from another book about the Mexican border. The reviewer critiques Kaplan on his views about the Mexican border and quotes another author, George Friedman who wrote "[Mexico's] real, social border will be hundreds of miles north of the legal border.” I thought this was an interesting concept. In terms of geography, the Mexican border is pretty poorly defined and was established at a time when the region was sparsely populated. So here lies a poorly defined border seperating two countries with a massive income gap. This of course leads to massive immigration. Because of the level of immigration and the proximity of Mexico, Mexicans are not being assimilated in the same way immigrants in the past were. They have such a large presence that they are about to maintain their culture and language and current technology allows them to communicate and visit with family in Mexico regularly, so they are never alienated from their history. This means that the Southwest is increasingly looking like its southern neighbors, so while the Mexican border may be south of Texas, Mexican society extends farther north.

In looking at this issue, I wanted to know just how far north Hispanic culture extends today. To do this, I looked at Census data on the Hispanic population just north of the Mexican border. I then used this data to map out a region of the US where the population is 50% Hispanic.

There are counties north of this region which are over 50% Hispanic, but I wanted to create as clean a border as possible without exclaves. Likewise, the region includes counties with very low Hispanic populations, but were surrounded by counties with high Hispanic populations. I left one American exclave though, San Diego and Orange County, because it has a large population, but a relative small Hispanic population.

Overall, this region represents 36 million people, 18 million of whom are Hispanic. New Mexico is clearly the most Hispanic state, although it is sparsely populated so the majority of this region lives in California. Over a quarter of the population is in Los Angeles County alone.

What is striking about this region is how uneven it is. It extends as far north as the Bay Area in California and parts of Colorado and Kansas, but only incorporates the border counties of Arizona and just barely reaches Austin and Houston. Arizona has a relatively low Hispanic population despite being flanked by large Hispanic populations in California and New Mexico. The Hispanic population isn't moving uniformly northward but is following its own patterns, possibly adhering to more sensible boundaries than the arbitrary ones currently in place.

Majority-Minority 2011

It's been awhile since I've looked at figures for the majority-minority states. In my last post on the subject, new data from the 2010 census revealed that minority populations had been under counted, and that the estimated year for when most states would become majority-minority states moved forward considerably. Well, I finally got around to including 2011 in my estimates. Not much has changed from 2010, the next ten states likely to become majority-minority states, remain the same, although the years change a little.

Nevada - 2014
Maryland - 2017
Georgia - 2019
Florida - 2022
New Jersey - 2025
Arizona - 2025
New York - 2029
Louisiana - 2030 (est. excludes 2005)
Mississippi - 2030
Delaware - 2031

Only three states, Hawaii, Massachusetts, and Rhode Island, showed a proportional increase in the white population between 2010 and 2011, but in Massachusetts and Rhode Island, it wasn't large enough to reverse the trend, and in Hawaii, the white population is already so small, there is no significant trend. DC is the only region that clearly shows a reversal as the city becomes more gentrified. According to current trends, DC will be majority white by 2029. However, I'm skeptical this trend will continue as the pace of change decreased greatly between 2010 and 2011.

Divergent Staaten

While on the subject of economic convergence and divergence, I thought it would be interesting to look at Germany. Since reunification, Germany has been marked by the stark economic differences between eastern and western Germany. During their time as separate states, West Germany greatly out paced East Germany in economic growth.

Since reunification though, the German government has poured tons of resources into the eastern Staaten, hoping to bring about a convergence. So far success has been limited. At first, they thought they could pay for reunification through the privatization of East German factories, but they turned out to be in such bad shape they were essentially given away. When replacing the East German currency, the government gave a highly favorable exchange rate, in an effort to increase wealth. However, this ended up inflating labor costs, which led to persistent unemployment. A massive investment in construction proved fruitless as the population continued to decline, leaving many disused buildings.

However, the government learned much from these early efforts, and with the recent growth of the Germany economy, one would hope that some convergence has occurred. Unfortunately, I can't find an comparison of incomes over a long enough period of years to determine if there has been a convergence in that area. An article from Spiegel says that in 20 years, household incomes in the east went from 35% to 53% of that in the west.
I have more complete data on unemployment though.
Instead of there being a more immediate convergence, there was a divergence in unemployment rates up until 2001, and the unemployment rate didn't peak until 2005. However, since then, the drop in unemployment in the east has outpaced the west.

Another chart I find interesting is comparing the number of unemployed people for each job vacancy.

After a massive spike in 2004, the number of job vacancies in eastern Germany greatly increased. So it appears, at least in terms of employment figures, eastern Germany didn't begin to converge with the west until around 2005, when Germany's economy started to improve.

While doing this research, I found one interesting article that now I can't find, wherein it concluded that attempts to compare the economies of the west and the east were flawed as the east is more rural. While the largest city in Germany, by far, is Berlin, all other cities in the top 10 are in western Germany. When compared to just rural areas in the west, the level of convergence is much higher. It's just that recent economic growth in Germany has mostly taken place in the cities, which the east is lacking.

Divergent States

The different rates of economic growth is a rather common topic when comparing countries. There is always a question of whether growth leads to convergence, assuming growth has decreasing marginal returns, or to divergence, assuming growth acts as a feedback loop. Of course, given the number of variables involved, there are plenty of examples to support either claim. While many Asian countries are approaching the levels of wealth seen in Western countries, there are many African countries that have been left behind.

Focusing on just divergence, one reason it occurs is because countries are sovereign. Just because one country has a growing economy, doesn't mean that another randomly selected country will have one. It takes a certain level of exchange between the two countries to forestall divergence. For example, since the introduction of NAFTA, economic growth in the US and Mexico has been almost the same every year. This is of course a situation where there is neither divergence nor convergence. Arguably though, there is a positive relationship between integration and economic convergence.

One reason to believe this is true is to look at regions within a country. It appears to be very rare that two regions within a country have divergent growth rates, at least not in the long-run. China is currently experiencing divergent internal growth rates between the coastal regions and the interior (per capita GDP in the wealthiest province is 5 times greater than in the poorest). China is actively trying to fix this problem, but other countries have mechanisms in place that also address this problem, although more passively, such as the free movement of capital and labor. For example, when a region becomes poorer than its neighbors, property and labor will be cheaper which entices businesses to move there.

What I am curious about though is how effective these mechanisms are. It's actually kind of a marvel that given how big the US is, that most people can expect the same standard of living in every state. There are pockets of wealth and poverty, but those are scattered throughout the US, not concentrated in certain areas.

I wanted a long term look at how GSP per capita in each state grew relative to the US GDP per capita. Unfortunately, there was a change how GSP is measured in 1997, so older data will not much up. Regardless, there have been some interesting movement since 1997.

The most basic way to look at divergence is to find the standard deviation of GSP per capita between states. In 1997, the average state had a GSP per capita of $35350 with an standard deviation of $6943. In 2011, the average state had a GSP per capita of $41446 with an standard deviation of $8109. While the standard deviation increased, so did the average, and relative to the average, nothing has changed as in both 1997 and 2011, one standard deviation was 20% of the average. However, in the years available, 1997 and 2011 are both high points. In 2003, one standard deviation was 17.5% of the average. So since 2003 there seems to have been a divergence, but a glacial one.

Looking at a more individual level, in 1997 the wealthiest state, Alaska, had a GSP 2.4 times greater than the poorest state, Idaho. In 2011 the wealthiest state, Delaware, had a GSP 2.2 times greater than the poorest state, Mississippi. So not only has the gap between the wealthiest and the poorest decreased, but the states holding the top and bottom positions has changed. This indicates that no state is soaring ahead or falling behind.

That said, there is a huge difference in growth rates from 1997 to 2011. During that time period, GDP per capita in the US increased by 14%. At the low end, Michigan's GSP per capita declined by -2%. At the high end, North Dakota's GSP per capita increased by 41%. However, despite such different growth rates, neither state have a GSP per capita much different from the rest of the country. All it really changed was their relative position. North Dakota has intriguing gone from being nearly one standard deviation below the average to nearly one standard deviation above the average.

Another interesting finding is that the current 3 wealthiest states all had below average growth. Which indicates there is some convergence happening at the top. However, there is some indication the poorest states are not converging with the average. Mississippi, the current poorest state, had only 9% growth, while South Carolina, the 3rd poorest state, had only 1% growth. While there's no indication that this trend will continue, it is still a worrying position to be in. It will be interesting to see when this divergence reverses itself.

New Top Computer

Top 500

Despite my prediction that K Computer would remind in the top spot for a long time, it has already been over taken by Sequoia, a computer that can calculate 16 PetaFLOPS. Not only that, after two rather lackluster updates, the Top 500 this time around saw a massive change. Overall, the total FLOPS calculated by the entire top 500 increased by 67%, the second highest increase ever, just behind June 2008's 68% increase.

Most of the change took place in the top 10, with 6 new computers. Compare that to November 2011's list which saw the exact same top 10 with just K Computer getting an upgrade. Overall, the top 10 increased it's total FLOPS by 113%. However movement at the bottom was rather low. Despite 170 new computers on the list, the total FLOPS of the bottom 10 only increased by 20%.