Friday, August 9, 2013

Homicide Statistics Bias by Gun Politics

A recent email I was forwarded, contained a list of Homicides/100,000 citizens for many countries around the world.  The email subject was "Eye Opener" and began with the title "World Murder Statistics".  What followed was a list of 109 Countries, with Honduras at the top of the list with 91.6 Homicides / 100,000.  Last on this list was the USA with 4.2 Homicides / 100,000 citizens.  The email ended with this statement:

 "ALL the countries (109) above America have 100% gun bansIt might be of interest to note that SWITZERLAND (not shown on this list)has NO MURDER OCCURRENCE!However, SWITZERLAND'S law requires that EVERYONE....

1. Own a Gun
2. Maintain Marksman qualifications....regularly
3. "Carry"........a Weapon."


As has been my habit when I see a list of numbers, I first went to the source of the data to confirm what I saw in the email.  Indeed, there is data supplied by the United Nations Office on Drugs and Crime (UNODC), which is different than the email's claimed source of the World Health Organization.  The following links will take you to these data summarized in an active table, but on these sites there are links to the complete data set from the UNODC which I downloaded and found to be the same as these links.

List of countries byIntentional Homicides / 100,000 Inhabitants

List of countries by firearm-related death rate per 100,000 inhabitants 

List of countries by gun ownership rate per 100 inhabitants

I first began to understand the "109 countries above America" and what this meant.  In order to find the Honduras rate of 91.6,  the email was referencing Intentional Homicide data.  All the data in the email were correct for all countries EXCEPT America!  The email stated that the United States rate was 4.2 / 100,000 but from the UNODC data set, the United States rate was 4.8.  In addition, there were 102 countries with rates higher than the US (not 109) and, not stated in the email, 104 countries with Homicide Rates LESS than the US.

In order to understand if all 102 (109) countries with rates worse that the US indeed had "gun bans", I utilized Gun Politics  for more information.  In summary, I could not find any country with a "100% ban" on guns.  However, many of these countries do indeed have stronger gun restrictions, but in these cases there are ways to obtain and possess a gun.  But to be clear, there are an equal number of countries with Homicide Rates LESS than the US that have more restrictive gun laws than the US.  So, it appears that restrictive gun laws do not seem to predict Homicide Rates.  But to test this I did download Gun Ownership data to correlate to Homicides which I will cover later.

Now, I wanted to investigate the comments about Switzerland! The statement that they have "no murder occurrance" is not accurate.  In fact, on this same list their Homicide Rate is 0.7 with 15 countries lower than Switzerland.  And finally, gun control in Switzerland is based on a militia concept as seen from this quote from Gun Politics.

Switzerland practices universal conscription, which requires that all able-bodied male citizens keep fully automatic firearms at home in case of a call-up. Every male between the ages of 20 and 34 is considered a candidate for conscription into the military, and following a brief period of active duty will commonly be enrolled in the militia until age or an inability to serve ends his service obligation.[76] During their enrollment in the armed forces, these men are required to keep their government-issued selective fire combat rifles and semi-automatic handguns in their homes.[77] They are not allowed to keep ammunition for these firearms in their homes, however, and ammunition is stored at government arsenals. Up until September 2007, soldiers received 50 rounds of government-issued ammunition in a sealed box for storage at home.[78] Swiss gun laws are considered to be restrictive.[79] 

So this law does not apply to everyone, but only to males.  They are required to keep a gun in the home for immediate call up to the militia (after serving in the armed forces) and does not mention anything about "carrying" a gun.  The marksmanship requirement I could not find either.  However, the most interesting fact was left out of the email.  Although required to keep the firearm at home, THEY HAVE NO AMMUNITION AT HOME!  All of it is stored in government run arsenals!  No wonder the death rate is so low!  Guns at home without ammo.

Now, moving on the actual data.  First I thought it interesting that this email used Homicide Rates by all methods.  There is a database of Homicide by Firearms by the UNODC which I found and began to look at relative to guns/firearms.  This database has fewer countries participating but there are still 70.

Applying statistics to all these lists, I was first interested in statistical differences between lower and higher rates.  I evaluated this using control charts with limits based on population sigma since the data were not time ordered, just alphabetical.  First we will look at Gun Ownership per 100 Residents.



The X chart at the top, clearly shows only one outlier country which is the US!  All other countries are within the normal range of per capita ownership.  This is probably not new news to most of you.

Next I looked at Homicide by Firearm Rates for Total, Homicide and Suicide.


There are two countries outside the upper limit signifying outside the "norm" for Total Firearm Homicides.  These two countries are El Salvador and Honduras.  If we take the 95% confidence (2 sigma), Columbia, Guatemala, and Sweden could also be considered outside the norm.  Switzerland is considered to have less restrictive gun laws, but Honduras more restrictive.


For Firearm Homicide Rates, the two outliers are Guatemala and Hungary,  At 95% confidence, add El Salvador, Honduras, Japan and Sweden.  Japan is considered to have more restrictive gun laws than the US.


Finally, the Firearm Suicide Rate shows Netherlands and Zimbabwe to be uniquely high.  Netherlands is considered to be more restrictive in their gun laws.

I could not find any correlations between gun ownership per 100 inhabitants and any of the firearm rates as seen below:


As you can see, between the two graphs, there is a small white box with a -2.8 which means the correlation is non existent and all others even weaker!  The number of guns don't correlate to any of the firearm death rates so I would conclude that other factors, including culture are more important.

My takeaway from this closer look at the email and the corresponding data is the culture of guns and gun politics has very little to do with firearm homicides and suicides.  It is time for the different groups battling over gun laws to take a new direction to make their respective cases!

Wednesday, February 20, 2013

Apple - Stock Price Drop Justified??

Following on my post of October 26, 2012, Corporate Quarterly Earnings Report's Negative Effect on Wall Street, I began to wonder if the recent $200 drop in Apple's stock price would correlate to its actual financial performance.  To note, the price began dropping in mid-September 2012, and might be now stabilizing as of this writing.

To begin my investigation, I collected, from SEC filings, Apple quarterly Revenue and Earnings figures back to March 1993.  As you have probably gathered in my other posts or from reading the information at my website (www.sustainthegain.com), I am not a fan of using Indices of these quarterly figures relative to a previous period!  Remember, a trend of one (recent quarter compared to one previous one) is not significant!  However, to reinforce this idea, I will produce a few examples of this analysis technique and make a few comments on them.  After these examples, I will return to the more meaningful analysis technique using the actual Revenue and Earnings data.

I have displayed these indices in a control chart, in order to gain some statistical reference!  I will start with Revenue, and in particular, Index versus Previous Quarter.


The first thing to notice is that at first glance, this chart appears to be stable at an average index of 1.065.  Compounded quarterly, this is a Compound Annual Growth Rate (CAGR) of 26.6%.  More importantly, the last two quarters ending Sept and Dec of 2012 are NOT uniquely different and, therefore, don't suggest any reason for a decrease in stock price.

There are some other things to take away from this chart.  The Upper and Lower Control Limits (UCL, LCL) are 1.78 and 0.35.  This means that any single quarter's index would need to be greater than 1.78 or less than 0.35 to be "out of the ordinary"!  As I have said, most companies are spending precious time explaining indices of 1.05 or .96 when a single, unique explanation is fruitless since only the common causes are acting on the results.  Whatever explanation is offered will now falsely become part of their institutional memory.   However, there is a distinct change in the pattern of these indices beginning at the middle of the chart, which is actually, March 2004, so lets take a closer look at this change.


 Although not obvious in the first graph, there has been a statistically relevant change 3/04 when the average index rose from 1.012 to 1.124.  This is the equivalent of increasing the CAGR from 4.9% to 59.6%.  Sounds great, but still nothing showing up for the 9/2012 stock drop!  My conclusion is that there was one sustainable positive change in 3/2004, and a positive "bump" in 12/1999 which could not be sustained.  My research on Apple SEC filings turned up a major accounting change in 2004 whereby the Revenue was reported differently!  It is pretty clear that the OND quarter is the highest index each and every year, since 2004!  This is one of the rare examples of indices, in control chart form, will indeed highlight a sustainable change.  The good news is that the actual quarterly results show this change as well, so still no need to use Index versus a Previous Quarter.

Below is the same Quarterly Revenue, but displayed as Index versus Year Ago (IYA).


It is more obvious that there is a change around 3/2004, but look how messy the individual indices are!  And the OND quarter pattern change is not indicated.  Also, the width of the UCL and LCL is quite large and the average 4 quarter index is 1.44 since 3/2004.  But, the last 3 quarters are all closer to the LCL than to the average, which is a signal of a possible change.  Could this explain the stock price drop??  I doubt that anyone on Wall Street is using control charts on indices!

Would Earnings as either Index Quarter Ago or Index Year Ago, show anything more??


 IQA does not show a average index shift in 3/2004 but the high OND quarter, each year, can be seen after this date.  The average index in Earnings is .962 which means that the earnings are shrinking!  This is likely an issue with the very low index at the beginning of the chart.  After 3/2004, the average index is 1.22.  Now look at IYA to see if there is anything more insights.


In the IYA case, the rise in the average index again shows near 3/2004.  But the most striking thing on this chart is the reduced variation in the earnings index after the accounting change of 3/2004.

In summary, the use of indices turned up only one sustainable change in performance since 1993, which was the accounting change in 2004.  We also did NOT see any significant changes in 9/2012 which would explain the stock price drop.  Sooooooo, we will move on to control charts using the actual quarterly results, starting again with Revenue.



Using the actual quarterly data for Revenue, you can find 4 timeframes of stable performance:  first from 3/1993 to 12/1995 when the CAGR was 18.3%;  then 1/1996 to 9/2004 when the Revenue dropped and CAGR was -0.6%;  next from 12/2004 to 6/2010 when the Revenue rose and the CAGR was 33.3%;  finally from 9/2010 to 12/2012 when Revenue jumped and the CAGR rose slightly to 35.5%.  The increasing variation (width of the blue UCL and the yellow LCL) in 9/2004 and 12/2010 is consistent with increasing quarterly values.  

In 1/1996, Windows 95 was introduced and most likely explains the drop in Revenue 1/1996!  In the following years, Jobs became CEO, Mac OS 9 ships, G4 Cube introduced, Apple Stores Open, i-Pod ships, Mac OS X ships and i-Tunes starts late 2003.  In 2004 we have the accounting change, 17" mac display and i-pod mini.  So what was the 2004 breakthrough......you pick, but my guess is accounting!  Had the pattern of Revenue (high OND quarters), I might have said this breakthrough was  i-Tunes.   The i-phone launches in 2007, 3G in 2008, but it is the i-Pad and i-Phone 4 that both launch in 2010 which creates the jump in 9/2010 and maybe only the i-Pad.  

However, trying to explain the stock price drop is more difficult!  Wall Street does not analyze using these techniques so they were not aware of the rise in CAGR!  The last two OND quarters were at or just above the UCL, but this should have been good news!  You can see that the Spring and Summer quarters had lower revenue but not outside the LCL!  It does appear that if you averaged all 4 quarters of 2012, you would get a number that falls right on the green trend line.  I decided to check this by obtaining the annual revenue numbers since 1993 which gave me the following graph.


Since 12/2004, the annual Revenue has had stable, predictable CAGR of 41% with 2012 landing right on the trend line!  So, it seems that Revenue should not have the caused of the stock to drop.  Could it have been Earnings??  Below is a graph of actual quarterly data.


There are only two sustainable breakthroughs in Earnings when there were 3 in Revenue (4 stable timeframes), but remember that one of the Revenue breakthroughs was the Accounting change that applied only to Revenue.  The Windows 95 intro in 1/96 did yeild a couple negative earnings quarters but not a shift in the Earnings.  In 12/2000 there was large loss in Earnings after 3 years of positive growth.  This lines up quite well with when Jobs became CEO.  He likely took a big write down after which the Earnings CAGR took off at 57.9%.  Then, simultaneous with Revenue, the Earnings jumped in 9/2010 but CAGR dropped to 40.1%.  Even if Wall Street had been tracking Earnings growth with control charts, the stock price should have dropped before 2012 since the 4 quarters of 2011 would have been sufficient to get a signal of this change.  But I'm sure they were not doing such an analysis!  An argument could be made that the OND 2011 and OND 2012 for Earnings were approximately the same value, when Wall Street would have expected at least a 20% year on year increase.  This was likely the stock downfall, but it is clear from the control chart that OND 2012 is just a random, non-significant result that fell between the control limits and should have been given no special consideration.  Had the stock problem been due to the i-Phone 5 intro and the "Apple Maps" problem, this should have shown up in Revenue.

An important note about Apple Quarterly Earnings News Releases:  In my 10/26/12 article referenced at the beginning of this post, I gave many examples of companies that spent significant time trying to explain every non-significant up or down in their results using Index Year Ago as the basis.  However, Apple does NOT report this way.  Every News Release follows the exact same format: the first paragraph reports actual results of this quarter and the same quarter year ago, but they do NOT use indices; the second paragraph gives sales figures but again avoids IYA; the third paragraph reports the dividends declared; the fourth and fifth paragraphs use the phrases "We are thrilled" and "We are excited" to describe the records they have set in Sales, Revenue and Earnings.  But they never attempt to tie a particular product event directly to a change from the quarter year ago!  Way to go Apple for not poisioning their institutional memory.   Might this be a contributor to their success??