Stacking the Deck

How much does your vote count? Not much, right? Even in a historically close election, we are talking about hundreds of votes, possibly less than errors in counting or casting.

So what if we added in the influence you have by talking about your candidate and maybe the motivation you give like minded friends to actually cast their votes by saying you are coming in late to work to cast yours? Might we dream up an instance where you could tip the balance?

Let's say we get close to believing it possible. Is there a greater chance of your vote counting in a presidential election or a congressional race? We'll assume you are in a "purple" state (sorry, Oklahoma, by the time you are a swing state, the outcome of the election has likely been decided). Does gerrymandering affect your answer? (For those not familiar, gerrymandering is manipulating political lines to pack all of one group in a minimal number of districts/regions to increase the representation of another).

North Carolina's Congressional Districts as of 2013.

A court has ordered redrawing of Districts 1 and 12

(though interestingly, not 4).

 Image from US GIS in public domain, https://commons.wikimedia.org/w/index.php?curid=36507070

I've heard complaints of gerrymandering giving minorities representation they did not deserve (reverse discrimination) as well as being responsible for republican gains at the federal level in recent years. Being someone who values credibility and consistency over partisanship, I thought of this as one of those issues where agreeing with one side and not the other (if that ended up being my conclusion) would involve the kinds of subtleties that don't play well in political discussions. It turns out, that is not the case and there is a very simple approach to this issue that is consistent.

North Carolina Congressional Districts 1973-1982

  Image from US GIS in public domain, https://commons.wikimedia.org/w/index.php?curid=36507057

Let's imagine a state with 15 districts; we'll make this state a rectangle 300 miles long and 500 miles wide, so each district can be a square 100 miles to a side. Imagine then that these districts have ~ 500k residents and are fairly uniform, 67 percent one group and 33 percent another (I'll call them republicrats and democans, but you can plug in racial/ethnic groups, environmentalists and industrialists or any other opposing factions that might exemplify the greatest political differences in our fictitious state that I will call Eastconsin).

In a perfect world, we would expect 10 republicrats and 5 democans sent to the House of Representatives based on a 67/33% split. But what does our Congressional delegation actually look like? 

This:

 

Republicrats get 100% of the seats despite democans being 33% of the population. While this may seem far fetched, the idea of a state having some sizeable minority worthy of representation, but not concentrated enough to gain that in any reasonable geographic divisions is very real.

Brian Olson created this map by minimizing distances from any census block to the center of the district.

Optimizing compactness takes partisanship out of the equation and works well if groups are segregated enough.

See this map and other work here

Of course, this is not fair. Could it be fixed? Certainly. One way would be to try to segregate the two parties to the greatest extent possible, perhaps if the uniformity between the districts but the democans in one district lived along the borders and in the corners, clustering with democans in another. We might be able to obtain 10 districts of 500k republicrats and 5 with 500k democans. Not much diversity within districts and unlikely we will see something like this in real life, but it works out well:

Never one to leave well enough alone, let's imagine a democan takes office and is tasked with redrawing districts. This clever fellow realizes that packing all of his sympathizing partisans together isn't necessary to win a district. He can put 250 001 of his guys in a district with 249 999 republicrats and still win. He starts digging through the numbers and finds that the 33% figure we have been using for his party is rounded and there are actually 33.33467% or 2 500 010 democans in the state. This is enough to take 50% +1 in 10 districts.

 Here, the goal is to to have two different configurations of districts: first, 50% +1 democans per district until we run out of democans and then all the rest are pure republicrat. This is a bit of an exaggeration, as the person drawing the map wouldn't want to leave it for a small turnout differential to decide the election (in any of the salmon colored districts, two more democans than republicrats staying home on election day results in a tie; three, a loss for the democans), nor would they want to risk a moderate opponent pulling a handful of "their" voters. So, most likely, we would have 9 democan dominated districts with 55 000 + vote margins and 6 still packed with republicrats.

I'll end with that simple approach to what is a proper or improper use of gerrymandering. The answer has to do with diffusion of important groups. If a group with a need for political representation exists as a large enough proportion of a state to deserve representation in Congress does not have that representation or is under-represented by one or more seats, this is a valid tool to use until it somehow infringes on the representation of another group. If districts are gerrymandered in such a way as to over-represent a group, it is invalid. Such is the case of North Carolina, a "purple" state that went for Romney by a 3 percent margin in 2012 and Obama by a 0.3 percent margin in 2008, but has a 10-3 ratio of republican-democratic representation in Congress. Also, looking at the CPVI (a measure of partisan leaning) for each district, the democrats are +17, +17 and +23 in their favor in the three districts they hold, while republicans have no district more than +12 in their favor (this demonstrates real world vs my ideal examples).

Now, my previous post pointed out part of the reason for divisive partisanship and one step toward resolving that. I think this one goes further, as it shows how we could have a republican dominated House of Representatives for 15 out of the  last 21 years despite having a democratic president for 13 of those years. It is a large part of the reason why party discipline and the steady march to the right while the country slips left has cost them more in presidential elections than congressional. But mostly, it makes your vote count less no matter which side you are on if you are in one of the many states that allow partisan redistricting. 

Citations for images of 2013 and 1973-1982 maps above, as offered by Wikipedia.org :

By Authors: 1) All GIS data presented in this project was originally collected and published by the United States Department of the Interior. 2) US district GIS shapefile data created in association with the UCLA by Jeffrey B. Lewis, Brandon DeVine, Lincoln Pitcher, and Kenneth C. Martis. (2013) Digital Boundary Definitions of United States Congressional Districts, 1789-2012. 3) Data was rendered using ArcGIS® software by Esri. 4) File developed for use on Wikipedia and elsewhere by 7partparadigm. - US Department of the Interior, public data retrieved from http://cdmaps.polisci.ucla.edu on October 18, 2014., Public Domain,

Spoiler Alerts

It's election season again and this one is providing more headlines and social media attention than any I have seen. It also brings the mandatory gripes about the system being rigged, the media being biased and politicians being liars, cheats and slow drivers in the passing lane.
So what better time to talk about those elements that are truly failures of the system and what the alternatives are? I plan to address just a couple that I think get little attention or the wrong kind.
These next couple of posts will be quick and lack the editing, visuals and maybe some of the humor I try for, but I want to get them out and also am trying to get back into the fast writing needed to finish my second novel, so I hope the ideas carry enough weight to make them worth reading.

First, lets look at the two party system and independent candidates:

In 2000, we had the most contentious and drawn out result in recent history due to some combination of poor ballot design, voter error, Supreme Court decision, poor campaign decisions and... Ralph Nader. This last is usually only heard in democratic circles and even then gets less blame than the SCOTUS decision. Interestingly, numerous recounts by media yield different results depending on how many counties are included and what standard is used (the "voter intent" standard, which many believe to be the truest measure yields a victory for Gore if performed across the state, but nearly any other attempt at recount, including those most likely absent the ruling leave Bush as the winner), while third party ("spoiler") candidates clearly shift the victory to Gore both in Florida and New Hampshire (either of which change the outcome of the election) whether using Nader or another of the left wing independents.
In 1992, we had another spoiler and one who garnered more attention (and is more widely accepted as the reason for Clinton defeating GHW Bush (41), even if the numbers don't back up the claim). Ross Perot was the most viable third party in ages and for a time looked like he had a real shot at winning (before he suspended his campaign and lost credibility due, he claimed later, to threats from republicans to disrupt his daughter's wedding). Still, he managed enough support that pundits and voters outside of the republican party still credit/blame him for the outcome.
And now, with the 2016 primary under way (for readers in other countries, worlds or dimensions, look it up if you must, but don't judge our country on this convoluted process; if it made sense, I would not be writing this), we have another wealthy and somewhat popular player threatening to enter the election.

Now, the first obvious question is:
Since we have a primary system designed to choose the best from opposing parties, do we need more choices?
And the answer is: Yes.
And not just yes for those who think they are two sides to the same two-headed coin. Even in a perfect world, a system where each half of the spectrum picks the perfect candidate to represent them in the election gives us two candidates halfway between the middle and extreme positions (the Median Voter Theorem says that two candidates will place themselves just slightly closer to the middle than his opponent and multiple iterations of this will yield two indistinguishable moderates, but this assumes two candidates who did not have to reach out the ends of their own parties to get there). The fact that those who vote in primaries tend to be closer to the extremes of their respective parties only makes this worse. Looking at this mathematically, a uniform distribution of voters (where we are all evenly spaced along the political spectrum with no clustering in the middle or at the ends) all voting in the primaries (half on either side of the middle) choosing the perfect "middle" candidate for their group will end up producing a president either 1/4 of the way down the spectrum or 3/4. If we use a 0-100 scale and calculate the average distance between voters and president, we find a result of 31.25 (.75*37.5 + .25*12.5). Had the Median Voter theorem worked, we would have had 25. Given a more extreme example (say, one where the more partisan half participates in the primary), we get a little over 39.
This is nowhere close to ideal, but a moderate like the MVT gives us has little to no chance in the primaries and will sabotage the more moderate of the two party candidates along with himself if he runs independent (this may be true even if a large number of voters choose their best candidate without fear of "wasting their vote" if the other two manage to not be too extreme; take primary candidates from the first case at 25 and 75 and a moderate at 50 and ideal voting gives the moderate independent the range from 37.5-62.5, 1/4 of the vote, while the other two get the quarter from their position to the extreme and the eighth from their positions to half way to the independent for 3/8 of the vote).

 The second obvious question is:
Would this blog post have been written if there were not a better way?
And the answer is, "I don't know because there is one".
This answer is often called Instant Runoff or Alternate or Ranked Choice Voting. It allows voters to cast their votes to whomever they prefer, ignoring the possibility of that vote giving the election to someone they like the least.
The way it works is ranking candidates from best to worse. If one candidate gets at least 50 percent of first place votes, he wins. If not, the one with the least first place votes gets dropped and all those who put him first have their second choices thrown in. This gets repeated until someone has a majority. This candidate will, in all likelihood, fit the Median Voter well, but even more, the winner will likely have very few voters with a strong dislike for him. Some of the other positive aspects of this system is the reduction in negative campaigning (no one wants to alienate someone whose second vote might matter; this has proven out in the real world), a reduction in the influence of money on elections (you would need enough to stay visible, but not necessarily enough to ruin every television commercial break), cut down on the cost and time of elections (no need for primaries when you allow some reasonable number of viable candidates to all be on the ballot) and proving me right.

The last question is:
Can we make this happen?
And the answer is maybe. The parties, lobbyists, partisans and big money would all be against it. But, if we all hate the current system enough to join up and get this done, it will happen.
Personally, I think it will take some level of support already existing when the absurd finally happens (as someone suggested to me, a Cruz nomination with Trump and Bloomberg as independents and either of the democrats, which leads to the House deciding; with the unpopularity of Cruz and the blame for the mess falling on Trump, we get a negotiated deal ala Rutherford B Hayes for Bloomberg with the help of some strange bedfellows and everyone insists that this should not happen for another 192 years).

Rabbits, Roths and Returns: A Retirement Primer

Everyone today is worried about their retirement, but most of us do not know just how worried we should be or what exactly we should worry about (other than the general thought of our 401k shrinking), much less what to do about it.  Say what you will about the value of choices, we are so overwhelmed with them today that we don’t have the resources to weigh them.  Just in relation to retirement, there are a number of variables to consider:

• Company funded plan or individual?
• Traditional or Roth?
• How much can I afford?
• How much do I need?
• How much time do I have until retirement?
• Stocks?
• Bonds?
• REITs???
• Money Market?
• Full service broker or discount?
• Domestic or international?
  This list does not even come close to be inclusive.  Under the stocks heading, there are individual stocks and mutual funds; under mutual funds, there are managed funds and index funds, closed and open, tax managed, the list goes on.  So, for the most part, we either do nothing, put away a token amount in a random vehicle, or take the advice of someone who is probably just as lost as we are.
  If this seems a foolish way to handle something so important…. it is!
  Though few of us have the time to fully educate ourselves on every aspect of financial planning, there are a few concepts that are worth being informed on.
  Age Adjusted Allocation and Dollar Cost Averaging
  First is the volatility of the market.  The market decline that followed the housing bust and bank bailouts may not have been the worst ever, but for those watching their retirement dollars shrink, it seemed like it.  Many delayed retirement, while others contemplated pulling their savings out of the market in favor of something more stable.  Neither should have happened if a little more information had come along with the initial decision to invest.
  A quick look at a chart of Dow-Jones monthly opening prices shows that the market has consistently grown in the long term over the last 80 plus years.  The black lines indicate mathematical ‘fit’ lines that attempt to (and do a fair job of) approximate the average growth of the market over time.
  
  It is even possible to use these mathematical models to give an average monthly return on investment (slightly over one percent).
  But look at the yellow line (actual values) and it is obvious that this upward trend is not a smooth climb.  The trend involves numerous small up and down movements, along with a few more major bounces.  It is because of these ‘runs’ and ‘corrections’ that stock market investment should be done with the long term in mind.  Someone early in their career with several decades to weather the ups and downs of the market can afford to have a very large share of their retirement funds in the market.  But as retirement age approaches, the percentage subject to these unpredictable fluctuations should necessarily decrease, so that a large dip does not wipe out the majority of a retirement account just as it was supposed to start paying out.
  Beyond adjusting for risk based on age, the potential still exists for irrecoverable loss, even for the youngest workers.  The worst drop on this chart, in 1929, which appears proportionately smaller than it was due to the scale of the chart, did not recover to its peak levels for over 25 years.  That means that at any time before 1954, some portion of the money invested at the peak in 1929 was gone (a very large chunk for the decade after the crash).  Even after the market returned to pre-crash levels, a dollar in 1954 was worth less than it was in 1929, so there were still losses in real dollars; a certain amount of return on investment is necessary just to break even.
  
  Of course, few of us invest for retirement in one lump sum and this is fortunate.  A look at the value of 1000 dollars invested in a lump sum versus the same amount invested in smaller sums over time, starting just before the crash, demonstrates the advantage to investing over time (called dollar cost averaging).  The third column shows 1000 invested over time starting a year before the crash; more dollars invested at the peak carry a penalty, but this is still overcome by the number of years investing at lower market prices:
  Note that the 1000 invested in a lump sum does not recover its value, even after 25 years, where the 1000 invested over time recovers its value before 10 years is out and shows total returns in the 80-140 percent over 25 years.  Also remember that as the 1000 dollars is spread over time, the amount per month goes down and the percentage of earnings put into retirement shrink as the 3 dollars per month invested in 1954 makes up a smaller part of the average paycheck than the 3 dollars invested in 1929.
  Why is this?
  A quick manipulation of a few Dow averages shows how dollar cost averaging means more stock shares are purchased at market bottoms than at the peaks, so the average cost paid per share is much lower.  This chart compares a few historical values for the Dow with the number of ‘shares’ 1000 dollars would have purchased at that level.
  
  At the pre-crash level, 1000 dollars bought about two and two-thirds times the shares that made up the Dow.  Three years later, the same amount bought over 23 times the shares (or nearly 10 times as much stock as it did pre-crash).  By 1988, that 1000 only got about half the shares (or about 1/5 of what it did in 1929).  So, by investing over time, not only is the risk reduced, there is actually a benefit from market fluctuations, to the point that profit can be made even over time frames where the market’s overall trend is negative or flat.  The magnitude of these fluctuations, by themselves, are unimportant to this principle, as demonstrated by the fact that increasing the size of monthly variation by 10 percent only made a 10 dollar difference over 25 years; what is important is the amount of time the market stays below the level at which withdrawals are made and the average difference between the market level at withdrawal and the level while contributions were being made.  So, a 30 year old who sees their 401k drop by 20 percent should not be considering a change in investment allocation; if anything, they should be thinking about increasing the level of contributions.
  Thinking About Taxes
  Roth IRAs are all the rage right now because of the difference between how they and traditional IRAs are taxed.  Where traditional IRAs, like 401k or 403b plans, are funded with pre-tax dollars, reducing their impact on take home pay, a Roth is funded with after-tax (take home) dollars.  On the other end, a traditional IRA is taxed when withdrawals are made, while withdrawals from a Roth are tax exempt.  So, the argument goes, in an IRA, the contributions and earnings on investment are delayed until the money is accessed, but with a Roth the contributions are taxed upfront, but the earnings on investment are NEVER taxed.  Sounds like a big advantage, right?
  Maybe.
  All things being equal, it really doesn’t matter.  If a certain amount of net or gross income is set aside for retirement, the total contribution that comes out of that is higher for the traditional IRA than for the Roth, which means there is more money in place to grow over time.  For instance, if someone can afford 100 dollars per month out of their household budget for retirement, that hundred can go into a Roth directly as 100 dollars; if they decide to do a traditional IRA instead, assuming a 20 percent tax bracket, they can put 125 dollars in.  This is because the 100 dollar Roth contribution actually came from 125 dollars in gross pay that had 25 dollars (20 percent) taken out for taxes; the IRA, being pre-tax, allows 125 dollars to come out of gross pay, reducing take home by 100.
  Some Roth proponents will now argue that because earnings make up such a large part of long term investment, the tax exempt nature of earnings in a Roth make it preferable, while IRA proponents will say that the tax advantages of their plan allow more money to get in, which means there is more to earn returns on.  Rather than bog down in a mathematical argument on this one, a visual will more easily convey the truth:
  Imagine you have 5 rabbits and ask a friend to help build a cage.  In return, the friend will accept either 20 percent of the rabbits you now have or 20 percent of the rabbits you will have in a year.  For the sake of argument, assume that you have an abundance of rabbit food and that there is no extra cost or effort to take care of the fifth rabbit, so all that matters is how many rabbits you end up with in the end.  Finally, assume that all rabbits are pregnant females and will have two babies by the end of the year.
  To start, you have 5 rabbits (green background represents your rabbits):
  
  Now, if you keep all five and give 20 percent to your friend at the end of the year, the result is 15 rabbits – 3 for your friend (20 percent of 15) = 12 rabbits for you:
  
  Now, consider the other alternative:
  
  Here we have 4 rabbits for you and 1 for your friend.  It should be easy enough to see that the one in the red background will have 2 baby rabbits, while the four in the green background will have eight more, bringing the total to 12 and 3 again:
  
  The allegory is not hard to apply to money: Investing pretax dollars in a traditional IRA, the money grows (tripling like the rabbits, quintupling or just growing by 10 percent does not matter) and taxes are paid on the whole thing.  Investing after tax in a Roth, imagine the dollars invested in the green background, while the dollars paid in taxes sit in the red background, along with the earnings not accrued on them.
  So, all things being equal, how do you decide?  There are a few considerations that vary by circumstances.  An above average earning individual late in their career who has saved nothing to this point will likely be in a higher tax bracket now than in retirement, so getting the break from the IRA now makes more sense (those in the top brackets do not have the Roth option).  A lower earning individual early in their career, especially if they are a disciplined saver, will benefit from the tax exemption later in life, as their bracket will be higher then.  Beyond that, there are personal views on what tax rates may do over time, fear of changing tax laws regarding the status of Roth’s might also play into it.  But there is one case in which the choice is clear.
  If you are eligible for a Roth and will be making the maximum contribution, the Roth has the advantage of allowing a greater amount of gross income into retirement.  This is because 5000 dollars pre-tax into a traditional account is 5000 gross invested.  But 5000 invested after tax represents a greater amount, depending on the marginal bracket of the investor; using 20 percent, 5000 into a Roth is 6250 from gross earnings.  As it has been shown that the outcome for similar contributions from gross is the same for either, the ability to make a greater contribution from gross allows the Roth to yield a greater total value.
  Another feature of a Roth is that contributions (not earnings) may be withdrawn at any time without penalty.  For some, this may be a safety net, but others might find it a temptation- the judgment is up to the individual in this case.
  Finally, for those with 401k or 403b plans that offer an employer match, this is just free money.  Imagine with the rabbits that the store where they were purchased called to say they were retroactively offering a buy one-get one free deal and were bringing 5 more (pregnant) rabbits over.  In the end, you will be losing three more to your taxing friend, but will have 12 more, too.  Other than that and the possibly more limited investment choices (and management costs), these plans offer the same kind of tax treatment as a traditional IRA.
  Investing Early and Often
  There are a number of allegories about compounding returns, one famous one has someone placing a piece of rice in the corner of a checkerboard, two on the next space, four on the next, then eight, etc, covering every space.  Another has someone saving a penny one day, two the next, four, eight, sixteen, and so on for a month.  The questions, respectively, are how many pieces of rice are on the last space of the checkerboard and how much money is saved at the end of the month.
  Taking the second question first, start by summing up the days:

• Day 1- 1c
• Day 2- 1c + 2c = 3c
• Day 3- 3c + 4c = 7c
• Day 4- 7c + 8c = 15c
  The pattern here for the total is that each day’s total is one cent less than the next day’s contribution.  The pattern for the contribution is 2^(x-1), with x being the number of the day (the ‘^’ sign means to the power).  This problem is not too difficult to take to completion with a calculator, but plugging in the formula, the last day’s total is one cent less than the next day’s contribution.  Assuming a 30 day month, the 31^st day would have 2^30 c, so the total for the 30^th day is one cent less than that. This equals 10, 737, 418. 24 – over 10 million dollars!
  The checkerboard problem is simpler, as it only asks for the last entry, which is 2^63, which is over 9 quintillion (a quintillion is a 1 with 18 zeroes after it).
  This relates to investing in that when you double something twice, you actually multiply by four.  So if your investment doubles every 10 years, after 20 you have four times as much and after 30, eight times as much.  By the same principle, it does not take 10 years at 10 percent to double, since each year earnings accrue on the original amount AND the previous year(s)’ earnings.
  For example, 1000 dollars at 10 percent annual yields 1100 dollars after one year.  But the second year, that 10 percent is applied to the 1100 (which is 1000 investment and 100 return), yielding 1210.  The third year yields 1331; the fourth, 1464.10 and the investment doubles after seven and a half years.
  Taking all this together, an investment at 10 percent (which is slightly less than the market average historically) over 30 years is not 10 times 30 =300 percent (quadrupling the original investment by returning triple in returns), but involves doubling 4 times (30/7.5=4), which means 2^4, or 16, times the original investment.  A person putting 1000 in at age 25 will have 16000 at age 55.  Cut that time in half and an investment of 1000 at age 40 will only yield 4000; or, conversely someone who wanted 16000 at age 55 could invest 1000 at age 25 or 4000 at age 40.  Time is a friend to the investor and the more there is, the better.
  Summary
  Investing for retirement can be as complex as desired, but it can only be simplified to a limited extent without losing valuable information.  The basic concepts of compound returns, dollar cost averaging and pre-tax versus after-tax investing are three that every one should understand if they want to get the most out of their retirement contributions.  Beyond that, there is a wealth of information on types of investments, strategies and personal finance, much of it free online, in podcasts or at the library; retirement years are becoming longer each generation- it is worth spending a little bit of time now preparing for what will hopefully be a lot of time in financial security then.