Understanding the US 2012 Presidential Election Result: How Models Triumphed over Big Data

by quantellia 17. November 2012 16:04

Today, opinions of what happened in the 2012 presidential election are beginning to settle.  Since history is often written from the perspective of the winners, many commentators have attributed the American right’s pre-election false confidence to a sort of blindness or group-think.  But there’s another truth.  Given what was known ahead of the election, the prediction that Romney would win was more defensible than the prediction that Obama would be re-elected.
The video above shows why.
The reason is simple. Take a look at this graph, reflecting data published by the social science data analysis network (SSDAN).
As you can see, the percent of African Americans who voted in 2008 is a clear departure from the historical trend.  It’s reasonable to assume that Republicans concluded that the pattern would revert to historical levels, which given the overwhelming amount of data going back over 30 years, is a completely rational, “big data” conclusion.  (Read Nobel laureate Daniel Kahneman’s “Thinking Fast and Slow” for a clear and deep analysis of this phenomenon, which is called Regression to the Mean).
An alternative explanation, of course, is that in 2008, something exceptional happened. And that pattern continued into 2012.  Predicting the future requires an additional ingredient, beyond simply projecting historical trends:
The Future = Historical Trends + Unique events and decisions happening today
Nassim Taleb defines a “black swan” as an event that is 1) a surprise, that has 2) a big effect, and that 3) history rationalizes with hindsight:  “of course the right was ignoring the obvious”.  The reason Black Swans happen: ignoring the second half of the above equation.    Taleb describes dozens of historical events like this one, where misunderstanding exceptions to historical rules has led to disaster. 
Simply put, the future is not always like the past.  It is essential to know when it is, and isn’t.
Obama’s win qualifies on all three Black Swan criteria.  Since such a high percentage of blacks voted for Obama, this miscalculation can be seen as having misled many Republicans to believe in Romney’s guaranteed election, and perhaps even to have caused the loss through leading to an inappropriately small focus on the needs of blacks and other minorities. What’s missing is the ability to clearly analyze the right-hand side of the equation: how do we know when something new is unique enough that it changes the course of history? 
It is clear that the Democrats were masters of this “secret sauce”.  They built an on-the-ground voter turnout operation to bump the curve upward from its expected value in exactly the places that would affect the Electoral College result.  By being able to focus their resources on exactly where they would contribute most towards achieving victory, the Democrats effectively out-modeled the Republicans. 
Before the book is closed on this election, all who chronicle its history should ensure that the modelers on both sides are appropriately acknowledged.  Because from now on, close elections will be won by campaigns who base their strategic decisions on models that are better than those of their competitors.

Want to know more about how great models can drive value for your organization? Drop me a line at lorien.pratt@quantellia.com and/or visit http://www.quantellia.com to learn more.



General | Politics

Tax Policy and the Economy: Towards Better Understanding and More Productive Discussion

by quantellia 3. November 2012 16:13
It's exhausting: understanding the economy in the midst of this election season is incredibly complex. And political economics tends to be, well, adversarial. Trying to understand the likely outcomes of opposing claims about economic policy—what does each mean for the future?—is maddening. There are SO many moving parts and incomplete facts, not to mention misleading commentary. And now, here we are, at a critical point in time, at an election that could shape the economic destiny of the United States for many years to come. Yet we have few aids, other than history lessons and common sense, to cut through the complexity and grasp the real consequences.
We need something to help. A way to clearly and simply decipher and weigh the alternatives and implications. How can subjective assertions be made more objective? How can relevant details be defined, tested, and considered, say from the perspective of various “what if” scenarios? Could opposing claims be compared and contrasted much more easily?
The answer is yes, there IS something! If interested, read on.
First, some background: Democratic societies are founded on the idea that government is “by the people, for the people.” But today, most are far more complex than their founders could have conceived. Let’s face it, it’s difficult to determine, for example, which tax policy will really lower unemployment, reduce the deficit, or raise GDP. It seems we’re left captive to the opinions of pundits and other “experts” who may have lost their objectivity in partisanship.
The good news: Using the approach suggested here, any reasonable person can recover the power to independently assess political claims. As a result, our society as a whole can recover the kind of transparent and participative democracy that keeps politicians accountable to The People as rightful authority.
An invitation: After watching the video below, join the discussion. Together we can take on many tricky issues facing the world today—starting with, as in this example, economic policy. Only an informed electorate can make the sound judgments necessary to ensure a free society. Help us in enabling a greater understanding of our complex world through the use of advanced technology.
Moving forward: beyond the dull thud of conflicting ideology
How do we create a common ground for those who are passionate both about advancing society, and yet also bring different points of view to the debate? How do we create a level playing field where we can explore different positions, where our goal is shared understanding and achievement of improved outcomes for everyone?
Before proceeding, we’ll confess that we don't have a comprehensive answer. What we do offer is a way to change a way that we discuss these topics.  In particular, an assertion that "doing X will cause Y", or that "doing A will have benefit B" and "will not cause a negative outcome C" should be something that we can test, in such a way that we'll agree on the result.
We should certainly be able to shed more light on questions like this than the dull thud that we often feel from futile collisions of conflicting ideologies.
The right tools can help us argue constructively about complex things
Here’s an analogy.  Let’s imagine two engineers arguing about the best design for a new airplane wing.  One engineer thinks it should have an upward curving shape. Another thinks the wing should have a winglet at the end.  Both agree that the goal is better aerodynamics, fuel efficiency, and safety.  They won’t spend too much time in ideology, rather at some point both will agree that a wind tunnel simulation will help them to reach common ground, and that once it exists, the behavior of the wind tunnel model can be both the arbiter of the disagreement, and also bring into sharp focus the precise points upon which they disagree.
With model in hand, the discussion shifts from whether upward curves are better than winglets per se, and towards a more constructive analysis of the details of both proposals, and whether they (by themselves or in combination) produce the expected results.   A key dynamic within this argument is both sides’ recognition that the model is imperfect, yet nonetheless informative. It provides a forum to understand the implications of the two choices on aerodynamics, fuel efficiency, and safety.   Modeling makes explicit assumptions that each side has about the others’ proposal, and provides a forum to test and correct those assumptions. 
Altogether, this approach shifts the tone of the argument from loudly proposing one position or the other to a collaborative exercise where the model is tweaked to achieve a goal.
Our economy is far more complex than that airplane wing, and we have far more at stake.  We propose that we can learn from the airplane engineers, using the same sort of tools and reasoning to overcome the complexity inherent in understanding how policy is likely to affect the macro economy.
Agreeing to goals and levers
So how do we go about providing this small incremental, yet valuable, improvement?  We suggest the following approach.
First, both sides must be able to agree on what is a favorable outcome.  As an example, let us consider what we are trying to achieve through a tax policy.  Consider the following:
  1. A successful tax policy will result in an improvement in GDP
  2. A successful tax policy will reduce unemployment
If both sides can agree to these goals, then we have already made progress. Furthermore, we have a better basis for moving forward rather than making loose assertions about who is a job creator, what is fair, and how the economy has reacted in the past when we tried different ideas.
Second, both sides should agree upon what courses of action are available to them.  Let us consider the shape of the progressive tax scale (the tax rate levied on income earners at different income levels) as a “lever” of this system, over which we have control.
Now for the hard part.  Both sides make assertions about how changing tax policy will affect the outcomes that we have agreed to care about.  On the table today are two general ideas: policies where higher-income earners pay higher taxes, or a “trickle down” approach where higher-income earners pay less.   Ultimately this will result in a decision: to choose a policy that leans towards one or the other approach.  
Instead of leaving these discussions to the exclusive domains of economists, how can intelligent non-experts nonetheless form opinions and engage in informed, participatory discourse?  This is our challenge.
Our goal: improving discourse through better understanding
We propose that a solution is to work towards a common understanding of how the macro economy behaves, involving both sides in the discussion. The goal: to achieve consensus such that both sides a) understand how the system behaves, and can then b) apply the policy decisions to the model to observe how it affects outcomes.
Of course, this is ambitious to say the least.  The United States economy is a vastly complex system, with more variables than we could possibly hope to capture, along with interactions more tangled than we can easily represent or understand.  As if that's not difficult enough, there's also a considerable dose of random behavior in this system. All of this leads to a situation where it is impossible to predict the future.
Understanding the system, not predicting the future
So we’re not proposing that we build a precise model of the economy with any reasonable predictive ability.  What we are proposing is that we use these techniques to move towards a much more sophisticated discussion, with greater granularity than we have seen so far.   It is time, right now, to end empty debates as to whether rich people are "job creators" or whether a given tax policy is "fair.”  Rather, let's discuss whether a given tax policy is likely to stimulate demand within a given income bracket, and how will businesses react. What force will such a policy exert, either upward or downward, on either employment or GDP?  
It’s not about the numbers
Since models always appear to present precise numbers, it is always tempting, if not irresistible, to disregard a model whose numbers don't match reality.  Indeed, if our goal was to score a prediction bulls-eye, then any difference between numbers and reality would be a drawback.   But this is not our purpose.
Again, simply observing how something simple like taxation can ripple through an economy is vastly more enlightening, and is likely to lead to overwhelmingly better decision-making than adhering to ideological positions.
The World Modeler simulation
We have built a simulation in World Modeler which is by no means a precise model of the United States or, for that matter, any other economy. However, its structural features are broadly correct.
The direction that the modeled economy will take based on different kinds of policy decisions more or less matches the direction that the US economy or other industrialized economies typically take when similar measures are taken.
Over time, if such a model is allowed to live and grow, and if constructive criticism including ideas from experts, comparisons to history, and continuously improving data, are applied, then for the reasons we've mentioned earlier, while the model will never be an accurate predictor of an economy, it will, hopefully, evolve to a level where it provides us substantive guidance in how to create economic policies that benefit everyone.
It accomplishes this through:
  1. Alignment on all sides
  2. A much more sophisticated level of discussion
  3. A common reference point that all parties can agree can be used as an arbiter
The details
So let us continue with a concrete example of the kind of analysis that we are proposing.  As begun above, we have established agreement on both sides that policy decisions about the national economy should attempt to increase GDP while lowering unemployment. As a next step, let's add a new element: the idea that the economy consists of three major groups of actors: businesses, consumers, and the government. 
These actors have complex microeconomic internals but broadly interact with each other as follows:
  1. The government collects revenues from businesses and consumers through taxes. Some of that revenue is paid to consumers for their labor and paid to businesses for their goods and services.
  2. Businesses consume labor from consumers, raw material from other businesses, and provide goods and services that  the government, consumers ,and other businesses exchange for money.
  3. Finally, consumers receive money in exchange for labor, some of which they pay to the government in taxes, some of which they save or invest, and some of which they spend, consuming goods and services from businesses.
In the kind of discussion we propose, "the economy" is the equilibrium that these three components reach as the system finds its natural operating level through their interaction.  The discussion about tax policy is one concerning which forces the respective proposed changes exert on that equilibrium.

We can run the analysis at this level, with just these three interactions.  Let's suppose this three-part system is in equilibrium at a given unemployment rate, and at a given GDP.  What will be the impact of lowering taxes on high-income earners?  Any consumer is characterized by the degree to which they consume and the amount they save.  High-income earners have generally reached the peak of their consumption, so giving them more money is unlikely to significantly increase consumption.  It will, however, increase their ability to save and invest.

On the business side this has little effect.  Business activity is driven largely by consumption, and so more money in the pockets of high-income earners will have little impact.  More money savings will lead to greater investment.  When this investment is in businesses, it tends to lead to greater efficiencies and reductions in employment.

From the government's perspective, decreasing taxes to the rich decreases government revenues, which decreases its ability to employ.

Just this simple analysis - right or wrong (which is not our focus, as above) - begins to provide a mechanism by which we can assess the value of various policies, and to move beyond ideological adherence to one or the other position.

Our first-order model
To take this to the next step, we've built a simple first-order model of an industrial economy, as illustrated in the video and model snapshots below.  Using World Modeler, we can run a simulation of this situation, apply various tax changes, and allow a computer to do the hard work of analyzing how a tax policy affects the various factors. The video below shows results of some initial experiments.

It is important to emphasize, again, that the point of this model is not to assert that one policy is better than another.  Rather, our goal is to provide a forum in which these issues can be understood and evaluated at a level of clarity, precision, and granularity that allows those with different points of view to find common ground over time.
If there’s enough interest, we’ll be going through the details of these sub-models in future blog postings.  Please post in the comment fields to vote for a more detailed treatment. We can also send you a trial version of the software with the model included, which we are offering on an open source basis.   Sign up at http://www.worldmodeler.com or email directly to us.
Crowdsourced political intelligence
As a final aspirational statement, social networks and the concept of crowdsourcing provide an opportunity for models that begin as very rough-hewn and approximate to benefit from the incremental input of expert and others' opinions over time.
We envisage a movement for sharing models that is similar to open source, where small open-source movements originally built basic tools which over time evolved into complex software packages that rival the best that the commercial software sector has to offer, through the selfless and passionate interest of expert individuals.
A challenge and a vision
The challenge proposed here is to make this level of analysis accessible to a population that is not comprised solely of actuarial analysts and econometricians.  To do this, we must make quantitative analysis more visual and intuitive than is typically presented in these cultures. 
This article is, therefore, a call to action to take what is arguably one of our most important national debate topics, and to make it as accessible as possible to people who are affected by the outcome.

We welcome any and all suggestions as to how to move forward.  If you’ve read this far, you’re in a precious minority of those with the energy to make it happen.
What does success look like in this endeavor? At some point in the future, the graphics that media outlets use are beyond pictures donkeys and elephants butting heads, even beyond clever data visualizations.  Instead, they provide a dynamic yet understandable representation of how policies proposed by different sides of the debate affect different aspects of the United States economy.


The opinions expressed herein are my own personal opinions and do not represent my employer's view in anyway.

© Copyright 2023 The World Modeler Blog