All in, The Unbalanced Schedule Wasn’t That Unfair


The 2013 Major League season is (almost) over, with only a tiebreaker playoff between the Tampa Bay Rays and the Texas Rangers remaining.  But the complaints have been flying fast and furious that the Major League schedule for 2013 was unfair to teams in the American League East.  Specifically, contending teams from the West and Central divisions were able to fatten up on relatively weak opponents (Houston, Seattle, the White Sox, Minnesota) while the AL East was strong top-to-bottom.  Since the wild card is a competition among teams from all divisions, the claim is that in a balanced schedule the AL East teams would have been far more competitive for the two wild-card spots.  See, for example, for a representative complaint.

Interesting thesis: but is it true?  To find out, we calculated the Balanced Schedule Equivalent Record for major league teams.  The BSER calculates what team’s records would have been had the Major Leagues played a balanced schedule in 2013.

The logic behind the BSER is simple: each team played a total of 142 games against the 14 other teams in its own league, plus 20 interleague games.  This works out to an average of just a little over 10 games against each team in a balanced schedule, as opposed to 19 games against teams in its own division and an average of 6.6 games against teams in the other two divisions.  If we scale each head-to-head record by the appropriate factor (142/(19*14) for teams in one’s own division; 142/(6.6*14) for teams in the other divisions) we arrive at what the in-league record would have been for each team, had they played a balanced schedule.  Add in the interleague games, and, voila; we have the final BSER.

The results for the National League are shown in Table 1

Team BSER Wins Actual Wins
LA Dodgers 97.83413078 92
St. Louis 97.56504899 97
Atlanta 94.48849396 96
Cincinnati 92.28844839 90
Pittsburgh 91.04198916 94
Washington 83.27080108 86
Arizona 82.23747789 81
San Diego 77.39849624 76
Milwaukee 76.40145819 74
NY Mets 73.71747551 74
Colorado 73.56945359 74
Philadelphia 71.31897927 73
San Francisco 69.66860626 76
Chi Cubs 69.4157176 66
Miami 61.11854544 62

Though there are some differences here (LA finishes with the National League’s best record, instead of having the third seed, and Pittsburgh goes to Cincinnati for the wild card instead of the other way round), the playoff teams are unchanged.  In the American League this is (almost) true as well.

Team BSER Wins Actual Wins
Boston 97.52548651 97
Oakland 96.24581214 96
Cleveland 91.35019367 92
Tampa Bay 90.27956254 91
Detroit 89.38933271 93
NY Yankees 88.68694463 85
Baltimore 88.61631351 85
Kansas City 83.01655794 86
LA Angels 82.45823198 78
Texas 81.26986123 91
Toronto 77.72932331 74
Seattle 68.79630142 71
Minnesota 68.04830257 66
Chi White Sox 66.44679882 63
Houston 49.00820232 51

Boston keeps the league’s best record, and Oakland remains the #2 seed, a game behind. Interestingly, Cleveland actually wins the Central by two games over Detroit, and Tampa Bay is the other wildcard.  The complaints do have some validity: Texas did benefit enormously from the unbalanced schedule, winning 10 games more than they would have done in a balanced schedule.  However, the AL East teams wouldn’t have done a great deal better under a balanced schedule (Tampa Bay would actually have lost a game, and the Yankees and Baltimore would only have picked up three each).

The calculation here is quite simple; if Major League baseball is genuinely perturbed by this season’s outcome, it could shift to the BSER to calculate wild-card teams and home-field seedings for the playoffs, thus permitting MLB to retain the unbalanced schedule and the excitement of the division races while maintaing fairness in inter-division competitions.

Or not.  The flaws in that plan are easy to see from the above tables: the whole point of unbalanced schedules is to emphasize division games.  Suppose a team finished second in its division in the unbalanced schedule, but third in the wild-card race behind the third-place team in its own division (as you can see from the above, the BSER can invert standings in a division).  Who goes to the wildcard?  And should Cincinnati or Pittsburgh have the home-field this year?  Pittsburgh finished ahead in the division, but Cincy in the BSER race…

The World’s Most Exclusive Club

The Triple Gold Club[This is written jointly with Alvin AuYoung]

What’s the NBA’s Most Exclusive Club?  Well, let’s first consider this question in the context of other sports. The Club should honor specific, noteworthy achievements, and have very few members.  The Club of Golf Grand Slam Champions might qualify: a purist would argue that the only member is Bobby Jones, but with four straight Grand Slam Championships across two years, Tiger Woods might also be considered a member.

But we do know of one high-achievement club with only one member, and it’s very likely that it will contain only one until June of 2013: the Basketball Triple Gold Club.   Who’s the member?  Well, that’s the trivia question…but, first, we should describe the Basketball Triple Gold Club.

The Basketball Triple Gold Club consists of people who have won all three of basketball’s most prestigious tournaments: the NBA Championship, the Olympic Gold Medal, and the FIBA Basketball World Cup.  It’s modeled on the Hockey Triple Gold Club,  which features the winners of the Stanley Cup, the Olympic Gold Medal, and the IIHF World Championship.  There are 25 members of the Hockey Triple Gold Club: nine Canadians, eight Russians, six Swedes, and two Czechs, and  the hockey world does take it very seriously.

But the Hockey Triple Gold Club has a large, and unavoidable, element of bogusness to it: the IIHF World Championship resembles college basketball’s NIT more than anything else.  It’s held annually, and it is held at the same time as the Stanley Cup playoffs.  As a result, it features mostly players who couldn’t make the NHL, or whose teams missed the playoffs, and occasionally a few stars whose teams got knocked out in the first round.    Players whose teams perennially go deep in the NHL playoffs are unlikely to ever participate in the IIHF World Championship, so there’s a large element of luck involved in even playing in the tournament.

Triple Gold and Mystery Man

Who’s the Mystery Member?

The Basketball Triple Gold Club is not nearly as inherently suspect: the FIBA World Cup is held quadrennially, and doesn’t conflict with the NBA season.  The world’s best could, if they wanted to, show up every fourth year to play for their country.  And, typically, unless the players are Americans, they do.  However, in 1998 NBA players were barred from playing in the FIBA World Cup due to that season’s NBA lockout, and in 2002 a number of NBA stars skipped the FIBA World Championships.    The shocking sixth-place finish of the US in 2002 and its bronze medal in the 2004 games reinvigorated America’s commitment to international basketball, and it’s possible that the FIBA World Cup will attain the stature in basketball that the World Cup does in soccer.    Barring injury, no top soccer player would miss the FIFA World Cup, and no top European player would miss the UEFA European Championship.

OK, enough suspense: who’s the guy?  The sole member of  the Basketball Triple Gold Club is Shaquille O’Neal, and, yes, that noise you hear in the background is Kobe Bryant‘s teeth gnashing.  O’Neal won the 1994 FIBA World Championship, four NBA Championships, and an Olympic Gold Medal in 1996.  Kobe has five NBA championships and an Olympic Gold from 2008, but has never been on a FIBA World Cup champion.


The Sole Member of the Triple Gold Club

So who’s likely to join Shaq?  The only possibility at the Olympic Games this year is Pau Gasol.  Gasol has a FIBA World Championship (2006), and two NBA titles (2009, 2010, Lakers).  The only other NBA/FIBA champions are  Peja Stojakovic of Serbia (FIBA champions, 2002) and, recently, the Dallas Mavs (NBA champions, 2011), Lamar Odom  (World Cup, 2010; NBA Champion, 2009, 2010), and Chauncey Billups (World Cup, 2010; NBA Champion, 2004). But Peja has retired and Serbia didn’t qualify for the Olympics this year, Billups is unable to play until December with a torn Achilles and  Odom wasn’t named to the Olympic team.

Shaq might get company before too long, though.  The US are heavy favorites in London, and if they do bring home the  gold five members of the team will have both FIBA World Championships (from 2010) and Olympic Gold Medals — Kevin Love, Kevin Durant, Tyson Chandler, Russell Westbrook and Andre Iguodala; and the smart money is that an NBA title is in the future for at least some of these young and rising stars.  So Shaq might have company as early as June of 2013.

The next opportunity will be at the FIBA World Cup of 2014, and a number of the game’s greatest could have a shot then, if they want it: Kobe Bryant, LeBron James, Dwyane Wade, and Chris Bosh, (for the latter three, Gold Medal 2008, NBA Championship 2012), to name four.  Manu Ginobli has a Gold Medal (2004) and three NBA Championships (2003, 2005, 2007)  to his credit, and so long as he plays Argentina has a live chance.  He turn  37 in 2014, however, and has been a professional since 1995.

The National Popular Vote: A Good Idea that Won’t Happen, and A Bad Idea That Might

 The National Popular Vote System

In the previous three posts, we’ve introduced the Threshold Popular Vote, showed how it would work in practice, and compared it to other electoral college reforms.  In this post, we’ll look at the simplest of all reforms: dump the electoral college and elect the President directly.  It’s good enough for most countries with a Presidential system; it’s good enough for the 50 states when they elect governors; the electoral college is an 18th century artifact that was a sop to the slaveholding South (James Madison, in fact, wanted popular election, but Southern states feared the popular election of a Northern abolitionist); we’ve gotten rid of our previous forms of indirect election (the US Senate was elected by state legislatures until the ratification of the 16th Amendment); the US is a much more homogenous nation now than it was in 1789, so we don’t need to protect regional interests; and so on.

All correct.  But so what?  There are two variants of the National Popular Vote: a Constitutional Amendment, which is a reasonable idea that will never pass, and the National Popular Vote Compact, a terrible idea that just might pass.  We will consider both of these here.

The National Popular Vote Amendment

The only really reliable way to abolish the Electoral College is through a Constitutional Amendment, which requires the ratification of 3/4 of the states: in other words, 13 states can block an amendment.   And the current system favors the small states, and there are at least 13 small states who will never, ever ratify such an amendment.  Why should they?  Currently, 189,000 Wyomingans select one elector: 683,500 Californians do.  A Wyoming resident has roughly 3.5 times the voting power of a California resident, and simply won’t give it up.

In fact, it never gets that far: Congress has only once gotten a Presidential popular vote amendment out of committee, and it never passed the House.   In general, there is a minor flurry of activity after each close election, and then it dies out. Unless and until there is some reason to believe that the popular vote will overcome the political hurdles that have many times stayed it, this remains a dead letter.

There is another, more subtle and tenuous objection. The United States has two Federal institutions explicitly designed to ensure representation of minority regional interests: the Senate and the Electoral College. Further, despite its original design as a weak federation of strong states, the United States has evolved into a continentwide strong federation of weak states, with a relatively low level of regional tensions. Contrast this with Canada, designed as a strong federation of weak provinces, but without any real Federal institutions designed to protect regional interests (the Canadian Senate is a toothless sinecure for political appointees, and is strongly biased to Eastern and Central Canada: the six provinces east of the Lakehead have 78 Senators; the four to the west, 24). Canada has evolved into a weak Federation of strong provinces, with persistent centrifugal tensions. Of course there are other factors, such as the minor matter of the Civil War. That said, the lessons of Canada, and the European Union’s current difficulties, is that one cannot have a strong central government in a federation without central institutions designed for regional representation.

The logistical objection to the National Popular Vote is the probability of a national recount.  Elections in the United States are exclusively state and local elections; there is no national election, nor any means to administer one.  Even the presidential election, which we think of as a national election, is in fact 50 simultaneous state elections plus an election in the District of Columbia.    There are some common rules, by Constitutional law by statute: universal suffrage, national age of eligibility to vote, federal oversight over electoral districts, limits on campaign spending.  The Constitutional requirement to choose all electors “on the same day” probably requires all states to hold their election on a common date.    But places and times of voting, the actual ballot, ballot access, counting procedures, recount thresholds and procedures, are all under state control.  It was a state official, Florida Secretary of State Katherine Harris, who certified Governor Bush’s win over Vice-President Gore in Florida in 2000.   Gore v. Bush eventually wound up in the Supreme Court, but it first had to work its way through Florida’s state court system.

If we were to have a national popular vote system, this would mean overlying a purely Federal election over 51 state elections.  This isn’t impossible, but it does need to be thought through and carefully implemented.  Polling times, voting procedures, time and manner of ballot counting,  eligibility to vote, probably ballot design,  and so on should be standardized for a Federal election.  For example, in some states convicted felons who have served their sentences can vote, and in others they cannot; in a national election, a uniform standard should be adopted.  Probably the most important thing is to set standards for a recount: under what conditions should one be held,  who is the supervising body, should the recount be held simultaneously, and so on.

The reason recount procedures are important is that recounts are quite likely.  There  have been 16 elections since the end of World War II.    Of these, three (Kennedy-Nixon 1960, Nixon-Humphrey-Wallace 1968, Bush-Gore 2000) have come within the 0.5% threshold at which most states order automatic recounts.  Three more (Truman-Dewey-Thurmond 1948, Carter-Ford 1976, Bush-Kerry 2004) were within a couple of percent and might have been challenged.  Almost one Presidential  election in five results in an automatic recount, and another one in five in a possible recount.

Fair Vote prefers a national popular vote, and discounts the possibility of a national recount.  A Fair Vote conducted study shows that only one election in 332, almost exactly 3 in 1000, gives rise to a recount.  The Fair Vote study is correct but irrelevant.  Most elections in the United States are uncontested, or offer only token opposition to the certain winner.  This is partly due to the sheer number of elections we have in the United States: here, we elect many minor officeholders who are appointed in most countries.  Uncontested elections are partly due to gerrymandering; and partly due to dominant regional party strength.  Such effectively-uncontested elections form the great bulk of the elections in Fair Vote’s study.  However, all Presidential elections are hotly contested in an evenly-divided polity, and in these circumstances extremely close elections are frequent.

Finally, should we insist on a majority, or simply a plurality?  In most of our popular elections, it’s the latter, but there are exceptions.  A number of states (Louisiana, for example) require a runoff between the top two candidates in the event that no candidate wins a majority on election day.   This is also the case in a number of popular-vote countries, notably France.   Whether we should require a majority  is no idle question: in the 16 elections since World War II,  no candidate won a majority in five elections, and  in a sixth a candidate won a majority  by just eight-hundredths of a percent.  We’ve had a majority requirement in the electoral college since the dawn of the Republic; should we carry this forward to a majority requirement in the popular vote.

There’s actually an ingenious proposal to introduce a National Popular Vote without going through a Constitutional Amendment, and it’s getting some traction.  It’s brilliant, but in our opinion fatally flawed.  We’ll look at that one next.

The National Popular Vote Without An Amendment: A Great Idea that Just Won’t, and Shouldn’t, Fly.

But you just have to love the sheer brilliance of the idea.

The National Popular Vote Bill is the brainchild of Dr. John R. Koza, the founder of Scientific Games, Ltd.  The basic idea is to enlist state legislatures to pass bills which commits those states’ electoral votes to the winner of the national popular vote.  The commitments wouldn’t take effect until states with at least 270 votes passed similar legislation.

The sheer brilliance of the idea is also its most significant, and really only,  flaw:  Dr. Koza and his colleagues have devised a way to effectively amend the Constitution without going through an amendment process.  This leads to a host of problems, which we detail here:

  • The first problem is inherent: this is an attempt to end-run a Constitutional Amendment.  It’s a good question as to whether a subterfuge like this is Constitutional; there are two  procedures to amend the Constitution, and one has been used successfully 29 times, so the procedure is not too burdensome.    The second procedure is entirely in the hands of the states.  The National Popular Vote bill invokes neither; instead, any compact of states with 270 electoral votes between them could effectively, by statute, modify out system of government  Any Supreme Court might well hold that the Constitution cannot be amended by as few as 11 states acting on their own (California, Texas, New York, Florida, Pennsylvania, Illinois, Ohio, Michigan, Georgia, New Jersey, and North Carolina).  The National Popular Vote people, counter, correctly, that the Constitution permits each state to select electors by any manner it chooses, so long as all electors are chosen on the same day.  However, it is implicit in the Constitution’s wording that the electors be chosen by the people of the state or by their representatives; it is by no means clear that the legislature of a state can abdicate the choice to the voters across the nation.   Further, Article 1, Section 10, Clause 3 of the Constitution explicitly prohibits states from entering into compacts without the consent of Congress: “No State shall, without the Consent of Congress,…, enter into any Agreement or Compact with another State”.  At a minimum, the National Popular Vote bill is constitutionally dubious, which means it is certain to be challenged in court.  There is a good review of the issues involved here, with a pro-Popular Vote perspective:  The Constitutionality of the National Popular Vote: Refuting Challenges Based on Article II, Section One
  • The incentive to cheat in a close election is very large.  Our history has shown that partisans are more than ready to ditch high-minded good-government principle for party advantage on any slight excuse.  So consider the scenario where the 11 states mentioned above have joined the National Popular Vote compact, and no others have.  Now assume a Republican has narrowly won the popular vote, by a tenth of a percent or so, but nine of the 11 mentioned states (every one save Texas and Georgia)  have voted for the Democrat, some convincingly.  Finally assume that the Republican  has won less than 220 of the 269  electoral votes allocated to states which have not, in this scenario, passed the National Popular Vote bill.  The morning after the election the outraged citizens of California,  Pennsylvania, Michigan, New Jersey et al will demand that the legislature recant the National Popular Vote act and select electors in accordance with the wishes of the state’s voters.    The cheerleaders of the partisan media (, MSNBC, and so on) are sure to find good reasons for their renunciation.  [If this example seems offensive to the Democrats among my readers, simply reverse Democrat and Republican in the above example, choose different states, and insert Fox News and in the above; it would be no less credible].   The National Popular Vote Compact organization has, of course, considered this and put protections into their bill: a “blackout” period after July 20 in an election year,  prohibiting NPV Compact states from changing their electoral laws after that date.  FairVote also claims that Article I, Section 10, Clause 1 of the Constitution prevents states from refusing to honor interstate compacts.   The cited clause reads: “No State shall enter into any Treaty, Alliance, or Confederation; grant Letters of Marque and Reprisal; coin Money; emit Bills of Credit; make any Thing but gold and silver Coin a Tender in Payment of Debts; pass any Bill of Attainderex post facto Law, or Law impairing the Obligation of Contracts, or grant any Title of Nobility.”.  This doesn’t require states to keep their word to each other, and in any case Clause 3 says that they can’t enter into treaties with each other without getting Congress’ permission.
  • The final problem is that the states in the National Popular Vote Compact have no means of enforcing voting standards among the states that join the National Popular Vote Compact, much less the other states.  Under the National Popular Vote Compact, California will be casting its electoral votes for the candidate preferred by all 50 states and the District of Columbia.  But how does California know how those votes were cast?  More importantly, how does California know who won?  Consider, for example, a Kennedy-Nixon election of 1960, where the national popular vote winner (Kennedy) was defeated in California.  Kennedy won the national popular vote by only 16 hundredths of a percent: well within the automatic-recount margin for virtually every state.  Nixon won California by about 54-hundredths of a percent, just barely out of automatic-recount territory.  Question.  Under the National Popular Vote Compact, for whom should California cast its electoral votes?  Kennedy?  But under California law, Nixon could surely demand a recount.  But the recount would have to be national, and how could California coerce a non-compact member (say, Rhode Island) to conduct a recount?  Rhode Island  voted for Kennedy by over 60%, and let’s assume it isn’t a member of the compact; it’s casting its four electoral votes for Kennedy, inarguably, and has no internal reason to go through the expense and trouble of conducting a recount.  But there is no way that any state would accept a 16-hundredths of a percent margin without a recount; what is California to do?

For all of these reasons, whatever one thinks of the idea of a National Popular Vote, the National Popular Vote Compact is both a brilliant and a terrible idea.  In a close election — and the only time any of this really matters is in a close election — the National Popular Vote Compact is certain to cause chaos at a level that would make Bush-Gore 2000 look staid by comparison.  The result would certainly be a repeat of what happened following the tumult of 1800; the National Popular Vote Compact would collapse of its own internal contradictions and failings, much as the original election procedure laid out in Article II, Section 1 of the Constitution collapsed, and a Constitutional Amendment would fix things up, much as the 12th Amendment gave us our current election procedures.

The fact that Congress and the States will eventually be shocked into fixing our broken electoral college system by a disaster caused by the National Popular Vote Compact, though, is no reason to deliberately foster the disaster.

In the next post, we’ll look at adoption strategies for the Threshold Proportional Vote.

Comparisons to Other Reforms

In previous posts, we introduced the Threshold Proportional Vote and showed how it would work in practice.   In this post, we’ll look at other proposals for electoral college reform.

This isn’t the only proposal for electoral college reform.    There are three  proposals that have received a great deal of attention: a national Maine-Nebraska system,  the Pure Proportional Vote System, and the simple direct popular vote scheme.   The last of these three schemes is a sufficiently interesting and discussed topic that it deserves its own post, and we’ll do that in the next post.  In this post, we will look at the first two proposed reforms: the National Maine-Nebraska System and the Pure Proportional Vote system.

The National Maine-Nebraska System

The Constitution allows the states to pick their electors by any method they choose, and states have experimented with a number of methods in the past.  Currently, 48 states and the District of Columbia use the familiar winner-take-all method, but Maine and Nebraska each select two electors statewide and one for each Congressional District.  In 2008, in fact, Senator McCain won Nebraska and two of the CD’s, and President Obama won the district around Omaha: Nebraska split its electors, 4-1 for McCain.

So the National Maine-Nebraska plan is to elect one elector for each of the nation’s 435  CD’s, two from each state, and three from the District of Columbia.  The attractions of this plan are  that candidates would have to campaign not only in battleground states but also in battleground CD’s, and that the Battle of the Chads in Florida would have been only over the two statewide EV’s, not all 25 of Florida’s EV’s.

Attractive.  Unfortunately, an analysis of recent elections shows a couple of things (data from the Swing State Project):

  • [Bad] There aren’t all that many competitive CDs.  58 CD’s were decided by 5% or less in 2008; 45 in 2004; 49 in the squeaker of 2000.  Forcing candidates to fight for CD’s really isn’t expanding the electoral map much.
  • [Fatal].  The National Maine-Nebraska system shows an overwhelming bias for the GOP.  In 2000, under a National Maine-Nebraska System, Bush comfortably beats Gore, 298-240, despite losing the popular vote.  In 2004, Bush’s 2% popular vote squeaker over Kerry turns into a 90-vote 314-224 margin in the electoral college.  In 2008, Obama’s six-point landslide over McCain yields a 303-235 win.

The latter flaw is fatal: quite aside from the right and wrong of any system with a strong partisan bias, any meaningful reform will require the support of both parties.  In other words, the National Maine-Nebraska System is deader than a doornail.

 The Pure Proportional Vote

The pure proportional vote is a simple system: if a state has n electoral votes,  and a candidate gets p% of the vote, he gets p * n/100 electoral votes.  So, for example, a candidate who gets 43% of the vote in California gets .43 * 55 = 24 electoral votes.  There will obviously be steps due to rounding.  This seems quite similar to the Threshold Popular Vote, and is on the surface quite fair. endorses it as a reform, which may indicate that FairVote has a strong preference for systems which favor third-party candidates.  However, on the evidence, the voters don’t like it much.

The Pure Proportional Vote has been proposed once before: in Colorado, in 2004, as Amendment 36 to the Colorado Constitution.  It got drubbed in a referendum, 66%-34%.  A defeat this stunning leaves scars, but also teaches lessons, and the lessons are best taught by perusing the arguments of the opponents: Coloradans Against a Really Stupid Idea (CARSI).   Essentially, this group made two core arguments:

  1. Amendment 36 gave each candidate the exact proportion of electoral votes corresponding to their popular vote total.  Colorado had nine electoral votes, so this essentially amounted to awarding a candidate one electoral vote for each 11% of the popular vote he got.  The losing candidate in Colorado generally gets about 45% of the popular vote or more, so one electoral vote would be up for grabs in Colorado in every election.
  2. The Amendment would have applied to the 2004 election, when Bush was virtually certain to carry Colorado and Kerry to get 45% of the vote or so; since Colorado was the only state adopting this, this amounted to a net win for the Democrats of four electoral votes.  The Amendment was sponsored by Democrats from California, and the Republicans had no difficulty discerning that this was simply a blatant attempt to grab a few extra electoral votes in a tight election.

CARSI’s first objection is essentially that normal variation in popular vote won’t be enough to sway electoral votes; it takes too much change in the percentage of the vote to change any electoral votes.  To avoid this problem, we have advocated a threshold proportional vote system.  Under our system, in a state with nine electoral votes like Colorado, a candidate would get nothing for the first 40% of the vote, and then one electoral vote for each 2.222% of the vote thereafter; in contrast, Amendment 36 would have awarded each candidate one electoral vote for each 11.1111% of the popular vote.  Under the threshold proportional vote  system, every election would have several electoral votes up for grabs from Colorado.

The second objection is serious, and makes the point that the means of introduction of a new system is as important as the new system itself.  In enumerating the requirements for any new electoral system, we observed that any new system must not favor one party over another, nor should it leave any  state materially worse off than it currently is.  This is also true of its means of introduction.   If  all Democratic-leaning states adopted the Threshold Proportional Vote  system, and Republican-leaning states did not, this would lead to a strong bias in favor of the Republicans, and of course the converse is also true.  As a result,  the Threshold Proportional Vote  system must be introduced simultaneously, in a balanced way,  in Republican and Democratic leaning states.  Further,  it must be extremely difficult for states to return to the winner-take-all system, to avoid the problem of last-minute shanges to affect an election.  We’ll note this as a potential problem with the National Popular Vote Compact in the next post.

Oddly, Coloradans Against a Really Stupid Idea didn’t hit on the biggest objection to a pure proportional vote: if adopted nationally, it would send many elections to the House of Representatives.   In 1992, for example, Ross Perot would have gotten 103 Electoral Votes and George H. W. Bush 203; Bill Clinton would have been reduced to 231, far short of the 270-vote majority.    Bush/Gore 2000 would certainly have gone to the House, as would Kennedy/Nixon 1960, Nixon/Humphrey/Wallace 1968, and Truman/Dewey/Thurmond/Wallace 1948.   That’s five of the 16 elections since World War II,  and this would represent an enormous change in our political system.  We have had 52 Presidential elections since the adoption of the 12th Amendment.  Of these, only one (Adams/Jackson/Clay 1824) went to the House, and it resulted in backroom horse-trading in the Congress to elect John Quincy Adams President.  The Adams Presidency was crippled by the “corrupt bargain” that got him elected, and Adams was drubbed in his rematch with Jackson in 1828.   The consequences of the PPV would be far worse.  For example, in 1992 George H W. Bush lost the popular vote significantly, 43%-39%.  However, the Republicans controlled a majority of state Congressional delegations, so it’s likely George H. W. Bush would have been elected President.  But the Democrats controlled the Senate, and so it’s likely then-Sen. Gore would have been elected Vice-President.  Since the Democrats controlled the House by a wide margin, it’s even possible that the Democrats would have attempted to use the impeachment process to overturn an election they would certainly have seen as illegitimate.  The history of our Republic is that closely-fought elections leave procedural hurdles and scars that last throughout  an Administration: see  Adams/Jackson/Clay 1824, Hayes/Tilden 1876, Harrison/Cleveland 1888, Nixon/Humphrey/Wallace 1968, Bush/Gore 2000.  A Pure Proportional Vote system would greatly exacerbate that tendency.

In sum, the National Maine-Nebraska system would turn the country over to one party for the foreseeable future, and a Pure Proportional Vote system would wreak havoc and effectively turn the President into a vassal of the Congress.  Neither suits our democracy.

One idea that might is the National Popular Vote.  We’ll consider that, in two variants, in the next post.


The Threshold Proportional Vote System in Practice

In our previous post, we discussed the Threshold Proportional Vote System.  Here, we look at how it would work in practice.

We can go back into our election records to find out how the Threshold Proportional Vote System would work in practice.  We  looked at a number of prominent elections in our past, and simulated them using a simple  version of the TPV system.  For each state, we found the top two vote-getters, and then awarded each 5 * (percentage of  top-two vote – 40%)  of the state’s electoral votes, with rounding.

To see this, consider a state with 11 Electoral Votes, where candidate A gets 52% of the vote, candidate B gets 45% of the vote, and minor candidates get 3%.  A’s percentage is set to 53.6% (52/(52 + 45) = 52/97 = 53.6%) and B”s percentage is 46.4%.  A gets 5 * (.536 – .4) = 0.68; B’s is 0.32.  0.68 * 11, rounded, is 7, and 0.32 * 11, rounded, is 4.  So we give A 7 electoral votes, and B 4 electoral votes.

We looked at three sets of elections: our recent elections, prominent third-party elections, and notoriously close elections in our history.  All the data came from the US Election Atlas.

Recent Elections

Let’s take a look at how some of our recent elections would have turned out if we’d used the Proportional Electoral Vote system in every state.

Election Candidates Popular Vote Electoral Vote (Actual) Electoral Vote (TPV)
2008 McCain(R) 45.7% 173 198
Obama(D) 52.9% 365 340
2004 Bush(R) 50.7% 286 301
Kerry(D) 48.3% 251 237
2000 Bush(R) 47.9% 271 267
Gore(D) 48.4% 266 271

The results are different in all three elections, and in all three cases actually align more closely with the popular vote.  In the case of 2008, Obama’s landslide is slightly reduced, and in 2004 Bush’s modest victory in the popular vote is better reflected in the vote totals.  The 2000 cliffhanger had the smallest change in electoral votes, but the most significant change in outcome: Gore wins, 271-267, instead of losing 271-267 (Gore actually got 267 electoral votes, but a faithless elector in DC refused to cast her ballot).  And this is without awarding Gore Florida; under the Threshold Proportional Vote, Bush won Florida 13-12.  Had Gore won the Battle of the Hanging Chads, he would have won Florida 13-12.  Under the TPV system, only one electoral vote, not 25, was hanging in the balance.

Performance in Three-Party Elections

Another interesting question is what happens in three-party elections.  Third candidates are unusual in the American system, but they do happen: there were prominent third-party runs in 1992, 1968, 1948, and 1912, and (arguably) in 1960.  In these four elections the third-party candidate finished first or second in some state, enough to affect the outcome in the TPV system.  The third-party runs in 1980 (Anderson), 1996 (Perot, again) and 2000 (Nader) did garner significant support and in the case of Nader, arguably affected the outcome.  But we’ve already analyzed 2000 and 1980 and 1996 were landslides.   As for 1960, we’ll look at that in close elections, below.

Election Candidates Popular Vote Electoral Vote (Actual) Electoral Vote (TPV)
1912 Wilson(D) 41.84% 435 403
Roosevelt(P) 27.4% 88 97
Taft(R) 23.17% 8 31
1948 Truman(D) 49.55% 303 311
Dewey(R) 45.07% 189 182
Thurmond(I) 2.41% 39 38
1968 Nixon(R) 43.42% 301 278
Humphrey(D) 42.72% 191 208
Wallace(A) 13.53% 46 52
1992 Clinton(D) 43.01% 370 351
Bush(R) 37.45% 168 186
Perot(I) 18.91% 0 1

The most important note, of course, is that it didn’t affect the outcome in any case. A couple of points to note: in both 1912 and 1948, there were four major candidates: Eugene V. Debs and Henry Wallace, respectively. Wallace in fact nearly outpolled Thurmond in the popular vote, getting 2.37% to Thurmond’s 2.41%. But he didn’t threaten to join the top two in any state. Similarly, Perot’s impressive 1992 performance was national: he didn’t win any state, and finished second in only one (Maine).

Performance in Close Elections

Finally, we looked at close elections not already covered.  There have been several close elections in our history: the infamous 1876 election between Samuel Tilden and Rutherford B. Hayes, where Tilden won the popular vote by almost 3% but lost the electoral vote 185-184; the Kennedy/Nixon election of 1960, which arguably was a three-party election; and the 1976 Carter/Ford Election.   We look at these three here.

Election Candidates Popular Vote Electoral Vote (Actual) Electoral Vote (TPV)
1876 Tilden(D) 50.92% 184 204
Hayes(R) 47.92% 185 165
1960 Kennedy(D) 49.72% 303 268
Nixon(R) 49.55% 219 264
Byrd(I) 0.73% 15 5
1976 Carter(D) 50.08% 297 291
Ford(R) 48.02% 240 246

Notice that the results in 1876 would have been changed; instead of losing a squeaker Tilden actually wins a narrow election; but inarguably wins it (in fact, as an historical note, Tilden actually did win the elected electors 184-182. Colorado had just been admitted to the Union as a state, and hadn’t had time to organize a Presidential election in the state. Instead, the legislature selected Colorado’s electors, and they went 3-0 for Hayes. We assume they would have made a similar choice under the Threshold Proportional Vote system).

The election of 1960 is worth a special note, because under a straight calculation (which we show here) the election would have gone to the House of Representatives. 269 electoral votes were needed for a majority, and Kennedy only got 268. Further, in the actual election of 1960, “unpledged electors” were placed on the ballot by segregrationist Democrats in Mississippi and Alabama. In Mississippi they ran separately from Kennedy, winning the state, where Kennedy finished second. In Alabama, there was a single unified Democratic slate, but in the actual electoral college Alabama split 6-5 for Harry F. Byrd over Kennedy.

In this election the TPV system shows Mississippi as breaking 5-3 for unpledged over Kennedy, and we’ve shown Kennedy as taking the 9 votes Democrats got under the TPV system (Nixon gets two in Alabama, with 42.7% of the vote). If the 9 Alabama votes split 5-4 for unpledged, then Kennedy winds up with 263 to Nixon’s 264. Similarly, in the election of 1960 a faithless Oklahoma elector cast a vote for Byrd; if he did that as well under the TPV system and everything else went as it did, the actual results would have been 263-263-11, and we would have been headed for the House.

This is highly imponderable, though, and depends upon the actual implementation of Threshold Proportional Vote. For example, we show California, New Mexico, New Jersey, and South Carolina as tied, though Nixon won the first and Kennedy the latter three; it seems likely that any actual implementation would award the winner of a state the majority of electoral votes. Also, Kennedy got 42% of the vote in Vermont, but zero electoral votes, due to rounding. If an actual implementation gave one electoral vote to a candidate who got 40%, Kennedy picked up one, there.

All in, in the most  likely scenario under a TPV system Kennedy  winds up with 271 electoral votes, losing one in California but winning electoral votes in New Jersey, New Mexico, South Carolina, and Vermont, for a net pickup of three over our calculation; and, faced with a choice of throwing the election to the House, the unpledged electors of Alabama would likely have cast ballots for Kennedy.

Analysis of the Threshold Proportional Vote in Practice

The TPV system, in all cases, aligns the electoral vote better with the popular vote than does our current system.  In our history, it would have changed the results in only two elections — exactly those two where the winner of the popular vote didn’t win the electoral college.  In most cases, the TPV system narrows the electoral vote margin from the current system, and in one case possibly enough to throw the election to the House of Representatives.   In some cases (1876, 2004) it increased the margin of victory.

In the next entry, we’ll examine other proposed reforms to the electoral college, and evaluate them against the Threshold Proportional Vote.

America Needs a New Electoral System: And Here’s One That Would Work

We’re heading into another national election — well, actually not.  We’re heading into another 12-state election.  RealClearPolitics has 12 states with 147 electoral votes in the tossup column (New Hampshire, Virginia, North Carolina, Florida, Ohio, Michigan, Wisconsin, Iowa, Missouri, Colorado, Arizona, and Nevada).  At least a couple  of these — Wisconsin, Arizona — are really leaning to one side or the other.  And “leaning” is a polite way of saying “will almost certainly support”.   So this election will be fought in the 12 states mentioned above, or more likely 8 to 10 of them; the rest of us will watch from a distance.

This is bad for America, bad for our politics, and bad for our policies.  Our Cuba policy, to take one example, hurts us and almost certainly has held back democratization in Cuba.  But it’s still popular among the Cuban emigrants in Little Havana, and no Presidential candidate or President will kick away Florida’s 27 electoral votes: ergo, the policy stays.  We need a new electoral system, and this entry proposes one: the Threshold Proportional Vote System.

Under this system, each major candidate is awarded a portion of the electoral votes in a state proportional to the percentage of the vote he received in that state, so long as he achieves a minimum threshold of support.  The threshold is vital.   Majorities should be rewarded, and super-majorities should command overwhelming electoral vote totals.

The formula we advocate gives each of the top two  candidates in a state  electoral votes in proportion to the popular vote each got, so long as a candidate got a minimum of 40% of the votes cast for the top two candidates (that’s the threshold), with a linear scale between 40% and 60%: the formula is the 5 * (percentage of the popular vote – 40%).  A candidate who got 53% of the vote in a state would get 65% of its electoral votes; her opponent, with 47%, would get only 35% of the electoral votes.   At 60% of the vote, a candidate gets all the state’s electoral votes; at 40% or below, none.  This rewards candidates who win states, but offers some votes to strong minorities in a state.

To see how this works, let’s consider California, with 55 Electoral Votes, and see how they’d break under the current winner-take-all system and the Threshold Proportional Vote System.

Winner Vote % Loser Vote % Current System TPV System
Winner EV Loser EV Winner EV Loser EV
60 40 55 0 55 0
58 42 55 0 50 5
56 44 55 0 44 11
54 46 55 0 39 16
52 48 55 0 33 22
50.1 49.9 55 0 28 27

To see why this is a good idea, let’s look at some properties we’d like for any reformed electoral system:

  • It needs to force the candidates into a national campaign — otherwise, there is no point in doing it.
  • It has to be fair, and broadly follow the popular vote.
  • It has to  be implementable in our current framework.  This means if it needs a Constitutional amendment, it needs to be popular enough to win the support of 2/3 of both houses and 3/4 of the states.
  • It can’t favor one party over another; not only are reforms that are blatantly partisan wrong, as a practical matter any real reform requires the support of both major parties.
  • It should avoid, where possible, high stakes driven by small changes.  For example, in 2000, Florida’s 25 electoral votes turned on a few hundreds of votes in a few Florida counties.
  • It should reinforce the two-party consensus-driven system that has largely served America well for 223 years.
  • It should exaggerate margins.  Virtually all successful democracies have electoral systems with this property; modest popular vote victories become sweeping mandates.
  • It should lead to a decisive victory.  Under the 12th Amendment, if no candidate wins a majority in the Electoral College, the House of Representatives picks a President from the top three candidates, with each state having one vote.  This has happened only once since the passage of the Amendment, and led to the crippled Administration of John Quincy Adams.

The Threshold Proportional Vote system meets all of these requirements:

  • It’ll force the candidates into campaigning in any state that adopts it.  Only DC, Massachusetts, Rhode Island and Alaska have voted more than 60% for one party in every election since 2000.
  • This is at least as far as our current system, and experimentally tracks the popular vote slightly better; see the next  post.
  • States can adopt this by legislative action today, and in the battleground states it will largely be a wash for each party.  But a Constitutional amendment is also feasible; unlike the movement for a national popular vote, this method preserves the currently electoral advantage of the small states — and no method that does not has any hope of passage.
  • As we’ll see later, this method slightly favored the Republicans in 2004 and 2008, and slightly favored the Democrats in 2000.  In neither case was the difference substantial; this doesn’t hurt or help either party.
  • The Bush v. Gore contest  in 2000 would have been over one, not 25, electoral votes.  Florida essentially tied in 2000; under this system, the count would have been 13-12 one way or another.
  • We’ve deliberately chosen a method where only the top two candidates in a state can win some of  its electoral votes (if this seems unfair, remember it’s only one, now); and by choosing a steep slope, it exaggerates margins: a 1% increase in popular vote translates into a 5% increase in a candidate’s state electoral votes.

OK, great idea.  So what’s wrong with it?

Some of the strengths of this proposal are also its weaknesses, at least from the perspective of political campaigns and interest groups.  There’s an advantage to candidates in the current system: they can focus their money and effort in a relatively small set of states and markets.  A truly national campaign would cost more and be a much more strategic effort.  Right now, candidates can afford to ignore California, because the Democratic candidate is guaranteed 55%-60% of the vote in the state; under the current system,  no electoral votes are at stake.  But under the Threshold Proportional Vote system, the 5% of the vote up for grabs translates into 25% or so of California’s electoral votes — 14 electoral votes, or so.    That’s equivalent to winning Colorado and New Mexico under the current system, and nobody is going to walk away from that.  Under the Threshold Proportional Vote system, every state is a tossup; the whole map is the battleground.  This is also bad for interest groups in the current swing states.  Right now, they exercise outsize influence on the political process, because they represent the favored few who can make a majority in a tight election.  In the Threshold Proportional Vote system, every vote is truly equal.

A more subtle problem is that the Threshold Proportional Vote system requires accurate counts, even for decided elections.  This is particularly the case for large states like California, with many steps between 50% and 60%; today, whether the winning candidate gets 54.9% or 55.1% of the vote doesn’t matter much; under the Threshold Proportional Vote system,  it could mean the difference of an electoral vote.  But other countries (notably Israel) use proportional vote systems, without being beset by counting controversies.

One way to avoid this is to use a tranched rather than smooth system, to make recounts unlikely.  Under a tranched system, a fixed number of electors would be awarded at specific percentages of the popular vote.  If the tranches are adequately spaced, recounts become unlikely.

States have various thresholds for automatic recounts, but 0.5% of the vote is very common.  Choosing a tranche threshold of 2.5% (awarding electors to a candidate at 50%, 52.5%, 55%, 57.5%, and 60%) would lead to a recount in about two in five elections.  Choosing three tranches (50%, 55%, and 60%) would lead to a recount in about one in five.  Of course, bear in mind that the frequency of recounts and the stakes in a recount are inversely related: the more tranche points, the fewer votes at stake in each contest.

In the next post, we’ll look at how the Threshold Proportional Vote system would  work in practice.