2013 in review

December 30, 2013

The WordPress.com stats helper monkeys prepared a 2013 annual report for this blog.

Here’s an excerpt:

A San Francisco cable car holds 60 people. This blog was viewed about 1,600 times in 2013. If it were a cable car, it would take about 27 trips to carry that many people.

Click here to see the complete report.

Income Inequality: Cross sectional and Trends

October 5, 2013

Although there are many ways of measuring inequality of income and wealth, this time I’m only going to use one measure:  The Gini Index (or Gini Coefficient), to show income inequality.

The Gini index has a value from zero to one, but is usually shown as a percent from zero percent to 100 percent.  At zero, there is perfect equality and everyone in a population has the same income.  At 100 percent, only one person has all the income and nobody else has any income at all. Wikipedia has a very good explanation of it here

One has to be careful in interpreting the Gini Index, though, because it is a measure of dispersion not a measure of absolute well-being.  For example, there could be two countries, one very rich and one very poor.  Yet both of them may have a low Gini Coefficient.  Thus if everybody is equally poor, they will have a low Gini, and if everybody is equally rich, they, too, will have a low Gini.

If you want to look for the well-being of an entire population, you have to use other measures alongside the Gini, like, for example, the general health of the population, or the mortality rate, or food insecurity, or pandemics, or whether or not they are having civil wars, or the level of corruption in both government and business.

The Gini is also used to measure the dispersion of wealth, but I’m not going to delve into that here.

So, let’s begin.  The first two graphs show the twenty lowest Gini countries and the twenty highest Gini countries.  Data are from the World Bank and the U.N. Statistics Division.  The year being measured is 2011. (click the image to get a larger picture)

Twenty lowest Gini countries-2011

Twenty highest Gini countries-2011

The very lowest Gini Index was Iceland, and it is a relatively rich country.

The very highest Gini was Seychelles.  I know you may not have heard of this country, but it is a bunch of scattered islands off the east coast of Africa, not far from Madagascar.  If you want to know more about it, Here is the Wikipedia entry for it

Seychelles is somewhat of a mystery, since they have a high human development index, yet, according to Wikipedia,

It has the highest Human Development Index in Africa and the highest income inequality in the world, as measured by the Gini index. Seychelles is a member of the African Union.

Among those countries with a low Gini, there are the usual suspects: Sweden, Norway, Finland, and so on.  But looking at the top graph, you can see that Afghanistan is also in this group.  The others are relatively rich, but Afghanistan is very poor.

So you can see why there is a warning above about being careful in using the Gini as a measure of well-being.

Now, let’s turn to trends.  Since I am an American, my first impulse was to look at the U.S., as I assume that anybody in any country would look at theirs first.

But, really, the unemotional approach would be to look at the slopes of the curves and see which ones have an upward trend and which ones have a downward trend, and what the magnitude of those slopes were.

But I succumbed and here is what the long-run trend looks like in the U.S.

Gini index-US-1960 to 2011You can see that since 1960, the slope is upward, indicating an increasing rate of inequality.  But notice that the rate of upward change is slowing down, indicating some kind of limiting factor or factors.  I have no idea what those factors are, but if you do, please post them in the comments.

There are a number of ways to select trends in other countries.  One way is to bracket those within one or two standard deviations from the world mean (average).  Another way might be to include only OECD countries, and yet another could be to only highlight the extremes.  Another way might be to use the U.S. as a benchmark case and select those countries who are “nearest neighbors” and bracket anything that varies by a certain percentage from the U.S.

There is a technical problem with selecting trend lines based on the mean for all countries for all years in the dataset.  The reason is that before 1986 there were less than 30 countries in the sample universe.  A better starting year is 1997 with a population of 103 countries in the dataset, but the tradeoff is shorter trend lines.

So, given these limitations, here is the world average trend from 1997 to 2011.

(click on the graph to enlarge it)

World average Gini index-1997 to 2011You can see that worldwide, the distance between the rich and poor is declining.  Yet there are some countries where the slope is upward, while the bulk of them are downward.

I started with the idea of comparing all OECD countries, but that would make an unreadable graph, since there are 34 members.  So I decided to break the data into “chunks” to simplify things.

The most obvious candidates are the U.S. neighbors, Canada and Mexico, since they are members of NAFTA and are contiguous.  So here they are.

Gini index-Canada Mexico and US-1997 to 2011It’s difficult on this graph to see the slope in Mexico, but it is declining, even though it has the greatest gap between the rich and poor among the three.

Toward the end of the period, the U.S. also shows a small drop after 2009, but it is relatively insignificant.  Canada, on the other hand, is far below it’s neighbors, and the gap between the rich and poor is slowly declining.

Examining the extremes:

From the bar charts above, I’m going to pick the trends from four of the lowest Gini coefficients and four of the greatest, and show two trend charts.  Some of the data are short trends, so I’ll have to start with a target year and show them, even though some of the countries may have no data in the early years.  For the lowest index there is Iceland, Sweden, Norway and Finland, and for the greatest are Seychelles, South Africa, Comoros and Namibia.  The target beginning year is 1999.

Four lowest Gini index trends-1999 to 2011

Four greatest Gini indices

The first thing you notice is that the lowest Gini index countries are tracking together and are on average only three to four percentage points apart from lowest to highest.

The second thing is that the high Gini countries are converging, and are nearly the same at the end of the period.  From 2006 onward, the group is trending downward, indicating a trend toward less dispersion, even though the Gini is quite high compared to the world average, which has a beginning point of about 42 and an ending point of about 39.


As you can see, this subject could go on for quite some time, given a dataset of up to 191 countries.

This post has been simple descriptive statistics, and only a sketch at that, but it does give some insight into global trends.

The really interesting part will be when the correlates are introduced and analyzed.  But I am a long way from that stage, and it will take quite awhile and a lot of grunt work to put the pieces of this puzzle together.

Structural Unemployment in the U.S.

September 27, 2013

In the past, I have focused on unemployment rates and duration of unemployment.

This time I’m going to dig a little deeper and explore what appears to be an emerging structural unemployment based on three basic ideas:

1.  Employment population ratios over time.

2.  Productivity.

3.  The implications of automation.

First a picture showing the trend in employment population and productivity.

Here, I’m using the employment population ratio rather than the unemployment rates in all their permutations because it gives a more realistic picture of how many people are employed compared to the entire population.

Data are from the U.S. Bureau of Labor Statistics.

Employment population ratio vs output per hour-US-1948 to 2012The first thing you notice is that the crossover in the trends was around the year 2000.  The second thing is that these are long and persistent trends, so it appears that one cannot chalk the cause up to current political persuasions or cultural mood at any point in time.

Keep in mind this is U.S. data only, so it doesn’t show what is happening in any other country, and it doesn’t take into account any other sociopolitical structures or cultural mindsets of other developed nations.

Now, I’m going to switch gears and present a set of hypotheses.  These are not new ideas, but I think they are worth spelling out.

  • Political persuasions have little or no effect on the long-term trends.
  • Productivity is not strongly related to the number of people in the labor force as a proportion of the population.
  • Productivity is a function of increasing automation.
  • As the proportion of people in the labor force declines, purchasing power declines.
  • As automation increases, demand will decline (because there are fewer jobs)
  • As demand declines, the need for more output declines.
  • Neither neo-liberal nor Keynesian nor Marxian economic policy can alter the trends.
  • Unless there is some new way of distributing goods, income and wealth, the system will become self-limiting.
  • Neither the rich nor the poor will be immune to this trend.

Now, I want to introduce you to a remarkable book written by a young man (a Millennial) called Robots will Steal your Job, but that’s OK .  The book is online and he keeps adding material and you can get on his mailing list when he finishes each chapter.

Here is a snippet from the introduction:

You are about to become obsolete. You think you are special, unique, and that whatever it is that you are doing is impossible to replace. You are wrong. As we speak, millions of algorithms created by computer scientists are frantically running on servers all over the world, with one sole purpose: do whatever humans can do, but better. These algorithms are intelligent computer programs, permeating the substrate of our society. They make financial decisions, they predict the weather, they predict which countries will wage war next. Soon, there will be little left for us to do: machines will take over.

Start with this introduction, and the left sidebar will guide you through the contents chapter by chapter. The book is well-researched and well-written.

He also has a TEDx video if you are interested. Robots will Steal your Job

2012 in review

December 30, 2012

The WordPress.com stats helper monkeys prepared a 2012 annual report for this blog.

Here’s an excerpt:

600 people reached the top of Mt. Everest in 2012. This blog got about 4,100 views in 2012. If every person who reached the top of Mt. Everest viewed this blog, it would have taken 7 years to get that many views.

Click here to see the complete report.

Organic Farming Trends In The United States

September 4, 2011

Although organic agriculture is only a small fraction of U.S. agriculture, it is growing at a very rapid pace.  Below are some graphs that illustrate that growth rate.

NOTE:  All source data in this post are from the USDA web pages devoted to organic farming.

First, here is the amount of farmland certified by the U.S. Department of Agriculture as organic.  The graph shows the number of acres for range land and crop land.

First, notice that both types of use are growing exponentially.  Second, notice that crop land is growing at a faster pace than range land, and that the curve is much smoother than the range land use.  This pattern is consistent with the rapid growth of food co-ops and the philosophy of localism, which will be dealt with in a later post. (click the graph to get a larger picture)

Next is the average size of each organic operation.  These operations are typically small, ranging from just a few acres to not much larger than 500 acres.  In this case, the trend is linear rather than exponential, indicating that the production per operation is increasing faster than the number of operations. (click the graph to get a larger picture)

Next is livestock.  Typically, organic livestock operations have been small compared to vegetable food crops, but they are growing at a rapid rate.

The graph below is of cattle production, both for beef and for dairy.  Dairy production is by far the largest segment of this sector, and is growing at a much greater pace.  I wish that I could have found goat trends, but when I do, I will post them.

The next graph is of hogs and pigs, and you can see how this sector has also been growing at a rapid pace.  As with so many other sectors, the trend line fits an exponential curve.

Finally, in this section for livestock, is the growth in sheep and lamb.  As with the others, the best-fit curve is exponential.

Finally, is poultry production.  This is divided into broilers and layers, and broilers are the greatest and fastest growing of the two.  By looking at the growth curve, you can see that broilers are also exponential, but with my limited software, I couldn’t get the exponential regression line to show, so I just substituted a linear fit


Although this has been a short post with little narrative, this evolving trend will be expanded on in the future.

Individual and Corporate Tax Rates in the U.S.

August 15, 2011

On a discussion group I subscribe to, there has been much discussion of economic subjects, and that motivated me to look into historical patterns in the United States.

What follows are two sets of graphs:  One showing individual income tax rates in the lowest and highest brackets from 1913 to 2008, and the other showing corporate tax rates in the period from 1942 to 2009.  These were the longest series I could find at the moment, so this is the result.

As usual, clicking on the graph will bring up a larger,more readable image.

Individual Tax rates, 1913 to 2008

Corporate Tax Rates, 1942 to 2009

Significant Points:

For individual income tax rates, the lowest and highest in 1913 were at 1% and 7% respectively.  by 1944, the lowest bracket was 23% and the highest bracket was 94%.

After that, the top tax bracket remained at a range between 94% and 70% until 1980, after which it dropped steadily, reaching 35% in 2009.

For the low bracket, it was 1% in 1913, reaching a high of 22.2% in 1952 and thereafter dropping to 10% in 2008.

For Corporate taxes, the lowest bracket was 25% in 1942 while the highest bracket was 40%.  By 1952, the lowest bracket had reached 30% and thereafter steadily dropped to 15% by 2009.  During that same time period, the highest bracket began with a 40% rate in 1942, reaching a high of 53% in 1968, and thereafter began a drop until by 2009 it was 35%, with a low of 34% in 1988.

I am currently exploring correlations with these tax rates and other economic variables, and the only one I have looked at thus far is the unemployment rate, which has no significant correlation along the trend lines. (16% with the low income bracket and 0.00% with the high income bracket.

I haven’t done any correlations with the corporate rates yet, but suspect similar findings.

An interesting set of variables to explore would be with infrastructure spending and with defense spending, but those datasets are not complete yet.

More to come.

Duration Of Unemployment in the United States 1948 to 2011

December 28, 2010


The latest data for mean and median weeks unemployed has been published by the U.S. Bureau of Labor Statistics, which included the month of June, 2011.

This time, I decided to do the graph on a 3-month moving average rather than on an annual basis.  Even with the moving average, the graph is a little jagged looking, but it is more sensitive to individual time periods.

As usual, if you want a larger image, just click on the graph.

Weeks unemployed002


Although the most often cited unemployment statistic is the unemployment rate (in all it’s versions), the statistic that concerns me much more is the duration that someone is unemployed.

It’s possible to remain unemployed until unemployment insurance runs out, and there have been extensions to that time. My concern is that there is a limit to how many times these extensions can be made, and that the lower standard of living can result in the U.S. becoming a third world country. Production will decline, social insurance of all kinds will decline and general measures of social health, including physical health will decline.

The longer someone is unemployed, the lower the general morale will become. In an upcoming post, I hope to show the symptoms of this declining morale and the results may bring to the country.

So, consider this graph that runs from 1948 to 2010. In the past the only measure was the mean (average) number of weeks unemployed, but in 1967 BLS began recording the median number of weeks unemployed. The median is the number at which half the population is below that rate and half the population is above that rate.

Household Savings Rates, Longitudinal and Cross-Sectional

January 9, 2010

Recently, I asked a question on one of the discussion groups I subscribe to about any indicators in cultural changes they thought were significant.  The list owner suggested that  I look at changes in household savings rates, and how, in the U.S. it was increasing of late.

The implication was that with the current recession (which officially began in December, 2007), people were changing the way they thought of how they handled their money, and that was a change in cultural attitudes toward consumption, and materialism in general.  It seemed that this was a reasonable assumption, so I went off on a merry chase.

The first stop was the U.S. Federal Reserve, Saint Louis Regional Office. They collect and analyze data from many sources, but primarily from the U.S. Bureau of Economic Analysis, which, in turn is an aggregator of data from many other agencies.

But if this were an indicator of a change in American attitudes toward consumption and materialism, I wondered, what would a snapshot of other countries in the world look like?  Are some countries more prone to save their money than others?  And if so, what would those countries be?

Using just one variable doesn’t explain cultural habits toward savings, but it does give some clue about how, collectively, some countries behave differently than others.

So, to begin, here is a trend line of U.S. household savings rates.  There are two graphs; one by the Federal Reserve on a monthly basis from 1959 to the present, and one I annualized by averaging the months for each year to produce a trend on an annual basis.

The St. Louis graph below is a bit blurry, and the monthly trend data make it look rather jagged.  Still, you get the idea.  The vertical gray bars are recession periods.

For a clearer picture, Look Here

The next graph is the annualized trend from the raw monthly data, from 1959 to 2009.

You can clearly see that the savings rate is on the upswing, but it is still far below the overall average.

Looking at cross-sectional data by country:

The next step was to look at cross-sectional data.  In this case, I went to the OECD (Organization for Economic Cooperation and Development site and looked at several countries.  The list is not exhaustive, consisting of only 24 countries.

There was a trend line for each of these countries from 1992 to 2009 with projections to 2011.  In order to get a sense of the “character” of the savings habits in each of these countries, I averaged all the years and then sorted them on the average.

You can see the result below:

NOTE:  You can get a bigger, more readable picture by clicking on the graph.

Who saves the most?  Spain, followed by Belgium, France, Italy, Switzerland and Germany.

Who saves the least?  Denmark, followed by Australia, the United States, the Czech Republic,  Norway and the Slovak Republic.

Technical Notes:

I don’t think of this as a technical blog, but an informational one.  Sometimes, though, technical considerations cannot be avoided.  This is one of those times.

Here are some main considerations you should keep in mind while looking at these data:

  • Standards, quality control, and reliability of the data (more about that later)
  • The variability over time.  Some countries have very erratic trend lines and the average I used smooths all of those out.  Below are some illustrations of variability measures.

Standards, quality control and reliability of the data:

In general, the quality of  data since 1992 in the OECD has been very good, except for some former Soviet satellites.  The reason is because they adopted the uniform System of National Accounts since the breakup of the USSR.  Some of the former Soviet satellites are having difficulty reporting according to the SNA, mainly because of budget constraints.

Likewise, the data from north America has been improving rapidly.   This is because of the adoption of the National Income and Product Accounts system (NIPAs).  The result is that all product, income and expenditures are counted using the same criteria in Canada, Mexico and the United States.

In addition, there has been an ongoing effort to “harmonize” the SNA and NIPAs systems so that all items in the accounting systems mean the same thing on a worldwide basis, much the same way that the International Standards Organization (ISO) has standardized everything from Internet address, to the thread sizes of nut and bolts to the sizes of paper sheets.

Variability of the data:

Even though the graphs above show a ranking of countries in terms of their long-term average, the variability over the term of the trends can be considerable.  Generally, they can be categorized as follows:

  • The savings rate is high and the variability is erratic
  • The savings rate is high and the variability is small
  • The savings rate is low and the variability is erratic
  • The savings rate is low and the variability is small

Now, I’m afraid I have to get a little technical.  But not much…

Two of the most basic measures in statistics are the mean (average), and the standard deviation.  The average isn’t difficult; people use it all the time:  Add up the numbers and divide by the number of numbers.  So, the average of 5 and 3 and 7 is (3+5+7)/3  (there are three numbers) and the result is 5.  So far, so good.

The standard deviation is a little more tricky.   I launched into an explanation and then thought better of it.  Let’s just say it is an average of all the deviations from the mean.  If you want more detail, there is a very good explanation on Wikipedia.

Once the standard deviation is computed, then it can be taken as a percentage of the mean.  This is called the coefficient of variation, or C.V.  The advantage here is that it normalizes the variation regardless of the size of the numbers in the data.

Below is a graph of the same countries sorted on the coefficient of variation.  You can see which countries have been the most and least erratic over the trend.

If we plot the savings rate against the variability we can get a picture of those countries in each quadrant:  Low savings, erratic, high savings, erratic, low savings stable, and high savings stable.  The graph has the names of each of the countries at each data point, and the axes splitting them into the four quadrants.

If you can draw any coherent conclusions from this, I would sincerely welcome them.

Putting Unemployment in Perspective: 1929 to 2009

December 4, 2009

Most of the graphs you see showing the U.S. unemployment rate are short term trends.  A few of them go back as far as 1948, but I wanted to go back to the Great Depression and put these trends in perspective.

Below is a graph of the unemployment rate from 1929 to 2009.  I estimated the end point of 2009 to be 10%.  The most current Bureau of Labor Statistics report shows a drop from 10.2% to 10.0% from October, 2009 to November, 2009.  But it is a bit dicey whether it will finally end up less or increase yet again.

A note of caution:  Methods of computing the unemployment rate have changed over time.  But in my view, the adjustments in methods do not distort the magnitude of the Great Depression.

So, here is the long view.

Just to put this in terms of how real people were affected by this upheaval, here are some pictures of what it looked like then.  The one below is a bread line. The pictures are from the Library of Congress.

And here is a famous picture by Dorothea Lange of migrant farm workers.  The woman on the right is the one from the famous picture of a mother and child.  She cropped it from this picture and enlarged it.

Unemployment Update- October 2009

October 22, 2009

The following graphs update the employment picture I posted in July, 2009 (see earlier post below).

There are two graphs:

The duration of unemployment in median weeks (half the people were jobless  less than 18.1 weeks and half were jobless for more than 18.1 weeks as of the end of September, 2009.)

The mean (average) at the end of September, 2009 was 26.2 weeks.  Nearly always, the mean is greater than the median, especially when the distribution is weighted toward the larger numbers.  Without knowing the actual distribution of numbers, it is fairly safe to say that with such a large difference between the mean and median, the median is a fairly conservative number.

The unemployment rate is also a fairly conservative number, since “discouraged workers” don’t show up in this statistic.  The Bureau of Labor Statistics has a series of alternative measures of unemployment, which include part-time workers, those who left the labor force entirely after long periods of unemployment, or who ran out of unemployment insurance benefits.  It is also possible to get data on those who were out of work, couldn’t get a job, turned 62 years of age, and simply retired on Social Security.

So, given those caveats, here are the two graphs.  Each series runs from July, 1967 to September, 2009.

First, the unemployment rate trend, which was 9.8% at the end of September, 2009.

Unemployment Rate Jul1967 to Sep2009

And then the median duration of unemployment in weeks, which was 18.1 weeks at the end of September, 2009.

Median Weeks of Unemployment Jul1967 to Sep 2009

Since last June, the picture hasn’t changed very much, except that the numbers continued to rise.  Look back at the July post and you can see how the upward trend has continued, with no noticeable slowing of the curve.

Looking at the duration of unemployment in terms of the length of a year, it looks like this:

  • Median duration of unemployment (18.1 weeks) is 34.8% of a year
  • Mean duration of unemployment (26.2 weeks) is 50.4% of a year.

So, we are fairly safe in assuming that a large proportion of the workforce is unemployed between a third of a year and half a year.  The magnitude of these job losses will be covered in the next post.
Source:   Bureau of Labor Statistics