Economists, traders, policy-makers and even medical researchers seek out correlations, and then look to “prove” or “disprove” causal relations among the correlated elements of the systems they observe.
What did he say?
In simpler terms, if you observe A and then most of the time observe B afterward, and if you only observe B following A, you might conclude or at least theorize that A causes B.
No guarantees, of course. Especially if you haven’t been able to define a mechanism by which condition A changes the state of the system so that B becomes possible, likely or even inevitable.
Let’s look at a case where correlation didn’t necessarily define causality. It cost investors more than a trillion dollars, and has cost taxpayers hundreds of billions (so far).
I remember seeing a former colleague, Michael Youngblood, at one of the industry conferences in 2005. He was presenting his mortgage default prediction model, and discussing the variables that drove that model.
The single variable with the highest correlation with future defaults in a metropolitan area (88%, as I recall) was new job formation. Makes sense. If a local economy is creating new jobs, people will come to that area to work, and even people who get into trouble on their mortgages will be able to sell their homes to the newcomers. Of course, the plenitude of jobs means that those who lose a job can usually find a new job, as well.
Michael’s work showed a fantastic correlation — if a local economy was creating new jobs above trend, 12 months later it would have a decline in mortgage defaults and credit losses.
Now let’s look at the statistical underpinning for these models. Most people’s exposure to “confidence intervals” from statistics is limited to hearing that Gallup or other pollsters questioned 800 or 1,000 “likely voters” to come up with their measure of political preference in upcoming elections, and that when they poll that number of people, they say their results are “within a range of plus or minus three percent.”
But the data set Michael and other mortgage researchers had at their disposal was orders of magnitude larger — roughly 100 million mortgages observed every month over a number of years. Even down at the local SMSA level (Standard Metropolitan Statistical Area), there were tens to hundreds of thousands mortgages in the data set and millions of observations. That many observations and an 88% correlation leads to an infinitesimal confidence interval.
Throw on a couple of other factors (inflation, mortgage rates, seasonality, tax rates, for example) and you have yourself a pretty robust model. Add in house prices, and you can even make your model better, predicting loss severity on those mortgages that do go to foreclosure.
For all the years from WW2 until 2006, an economic lifetime, single family homes increased in price on a national basis. There were localized dips year-over-year, like in Southern California when Cold War military spending was cut back in the 1990’s, or in the Oil Patch following the break in OPEC’s power brought about by massive North Sea and Russian oil exports.
An especially tough example was Detroit and a stretch of northern Illinois, Indiana and Ohio where the Big Three auto makers and car parts manufacturers were going out of business or cutting payrolls to the bone while automating. In those areas, defaults stayed persistently high and house prices locally declined from 2001 through 2007, even while the rest of the nation was on its last parabolic mania of housing bubble price increases.
In each of the examples above, a decline in job formation or meaningful job losses preceded the increase in foreclosures and the decline in house prices. The theory of unemployment driving defaults and house price declines seemed pretty well proven.
Until the current nationwide housing price decline, researchers thought of house prices as an effect, not a cause. Unfortunately, there is a tipping point at which the house price began to drive increased defaults, especially for “responsible” borrowers who took out affordable fixed rate mortgages and paid 20% or more in down payments.
Those borrowers, the ones who documented their income, paid 20% down, and took an affordable fixed rate mortgage are now the largest group of defaulters. They are also the loans that made up nearly 100% of the credit exposure held by Fannie Mae and Freddie Mac.
Those people have ordinary lives. They need to sell their houses every seven to ten years on average (varies by region), due to new jobs, geographic transfers in their current jobs, lost jobs, divorce, death, growing families, etc. That’s about 10% to 15% of them each year. If they can’t recoup their purchase price and get their original down payment out when they sell, they will defer selling as long as they practically can. If they can’t recoup even enough to pay off the mortgage, their life situations end up pushing them into default. There are literally millions of families in this situation today, and I have a hard time condemning them for “buying too much house,” a common complaint of the innumerate but oh-so-sure “free market” crowd.
Today most mortgage researchers have replaced job formation with house price declines as the primary driver for future defaults. Like the job formation data that led to the pre-collapse models, the correlation between dropping house prices and future default is pretty high (not as high as 88%), but high enough to be considered as an explanatory variable. It leads defaults, but again not as much as unemployment did. Needless to say, today’s unemployment being double what it was five years ago also adds to the foreclosure picture.
So which variable is actually causing defaults? It seems that when a house has declined to the point that the borrower is 5%, 10% or more “underwater” the house price drives defaults. Maybe it’s just life happening and the large pool of borrowers, but the bottom line is that in regions where most of the borrowers are underwater on their mortgage loans, that variable seems to dominate.
I fully expect the primary causation of defaults to revert to joblessness once the housing market stabilizes and returns to the long-term price path defined by material and labor costs.
I bring all this up because I think I see another example in the works, and it will likely be a big surprise and disappointment following the next couple of elections. Worse yet, it may cost us even more than being wrong about housing prices and mortgages.
So what are the competing correlations I think exist in our political system?
Encumbency vs. concentrated campaign money as cause for bad government.
Many of us are in the mood to “throw all the bums out” because we don’t like how they’ve handled our government, almost at every level. Naturally we feel that deals have been cut by representatives who no longer represent us, who have “perks” like health care insurance most of us can’t get, retirement plans that blow away Social Security, and a system that has given them raises over the past decade while most working people actually make less today at the same jobs we held a decade ago. In fact, one of the quickest paths to serious annual income over the past ten years has been to be a senior staff employee for a Senator or Congressman. The average paycheck for those people when they become lobbyists recently topped $700,000 per year. Never mind what an actual Congressman or Senator can make.
There’s plenty to support the idea that getting a new crop of politicians to replace the current crop will improve things, but a few glaring red lights that say it might be worse.
Most of the studies that rate incumbency as the predictor of election outcomes don’t look at money. Money spent advertising and paying for “boots on the ground” efforts like phone banks and informal taxi service on election day actually has a higher correlation with success in an election than incumbency. The reason the two different “causes” get conflated is that incumbents are hooked into the donation apparatus of their parties and lobbyists, they have free advertising by being newsworthy in their districts, they can carry over money from prior campaigns, and they have the Franking privilege (which matters less and less as we move away from physical mail to e-mail).
I can’t be the only one who saw Clinton’s appointment of Robert Rubin and double appointment of Alan Greenspan as clear indications that the modern Democratic Party was just as “owned” by corporations, hedge funds, banks and major wealthy people as its Republican counterpart.
Then the money flow changed. It started to come from small contributors on the internet giving directly to candidates. None of them would have any particular influence, but as soon as their money threatened to or did outweigh the concentrated contributions controlled by lobbyists, political party apparatchiks or by wealthy individuals, the ability to milk the system for concentrated profit was threatened.
Some of my hard-core partisan Republican friends (yes, I still have a few that haven’t rejected me for my apostasy) still nurse a sense of victimhood because the Obama campaign had more money to spend than McCain, and McCain “had to” go back on his public campaign finance pledge when Obama did. The fact is that Obama harnessed the broad power of the internet to reach enough small donors that the likes of Enron’s Ken Lay or even the Koch brothers (the old power structure) didn’t have the firepower to buy the election.
The corporations, hedge funds, banks, etc. were losing control. That didn’t sit well with the lobbyists and their corporate sponsors, who had done so well co-opting both parties to make their corporate masters more profitable.
That’s why I think the 5-4 right wing majority of the Supreme Court went way outside judicial conduct standards when they decided to send Citizen’s United back to re-write and expand their case rather than ruling on the already pretty bold claim they faced in March of 2009. It was as if Roberts & Co. wanted to write a new law and felt that the lawyer arguing the case hadn’t reached far enough, so they told the lawyer how to argue a much larger case. They got what they wanted, which was the ability for corporations, unions, and anyone else (since the donations could be anonymous) to spend as much as they wanted in any political campaign in America.
Now we’re hearing that third-party spending (not direct campaign or even party money) is running 10-1 in favor of radical pro-corporate candidates, especially in small states where the full power of an individual Senator is decided by a few hundred thousand voters. In tiny states like Alaska and Delaware (Joe Miller in Alaska got just over 60,000 votes and Christine O’Donnell in Delaware just over 30,000), hundreds of thousands of dollars controlled by professional lobbyists poured in to swing the primaries to unexpected winners that just happen to support eliminating laws that don’t favor corporations, laws that established Medicare, unemployment insurance, Social Security, the VA hospital system and minimum wages.
I did a purely financial analysis of the optionality of campaign contributions last January when the Citizen’s United case was announced with its pre-ordained corporate favoring result. It seems that a number of anonymous large contributors have figured out the same thing, and they are buying the “cheap” government calls the Supremes handed them, all without exposing who they are and how much they are spending. It would be money-laundering to support a criminal enterprise if buying elections was illegal, but it’s not.
Do you think for a red-hot second that the people who paid for the ads won’t be able to whisper in the ear of newly-elected Senators, Representatives and Congressmen that they were behind the ads that put the newcomer into office? Do you think that Wilson, O’Donnell, Angle or Paul will actually be able to write laws? Rand Paul has already told his coal miner constituents that he thinks mine safety standards should be strictly up to the mining companies, since he doesn’t understand that business. Who will Rand Paul turn to for language in any new laws when the current regulations are reviewed because of the recent spate of fatalities?
I’m betting that the new flood of money, not incumbency, will swing a lot individual races.
We’re likely to end up with a bunch of tyros in office, totally dependent on their corporate sponsors to write any new legislation. What could possibly go wrong with that?
We’ll just have to see whether these political call options pay off. If they do, I expect that we as taxpayers will be paying trillions of dollars more to those billionaires and corporations that are buying the next Congress now, and maybe the White House in two years.
You’d think we would have learned our lesson from the last corporate-controlled government, the one that set us up for financial meltdown through unregulated derivatives, pushed subprime mortgages through unregulated non-bank financial companies, and ran a fiscal surplus into trillion dollar annual deficits as far as the eye can see.
By the way, what did they do with the six decades worth of payroll tax surpluses that was collected? They didn’t spend them on Social Security or Medicare, but to listen to the current crop of faux conservatives, that’s the problem. Since when is services owed for payments received a problem created by the people who paid for those services, rather than the people who spent the money elsewhere?
I guess that damage was not severe enough. Now we need a crop of Hoover/Coolidge throwbacks to show us how bad it can be.