Why most economic forecasts are useless

Worthwhile Canadian Initiative writes:

Suppose instead that the model has a consistent one period lag, so Y(t)=R(X(t-1)). And suppose that Statistics Canada reports all data on X immediately. Now we can use the model for genuine forecasting of the future. Statistics Canada tells us what X is today, and we use the model to tell us what Y will be one period in the future. But the model can tell us nothing about what Y will be two periods in the future, because Statistics Canada can't tell us what X will be one period in the future. And yet I keep hearing about model-based forecasts for one, two, three, four, etc., periods in the future.

Steve Malpezzi, Walter Barnes and I wrote a paper some years ago that tried to come to grips with this very issue.  The context was office market forecasts, and we worked on developing confidence intervals for estimates of future office demand for a number of metropolitan areas.   In developing the confidence interval, we used a technique from a 1971 Martin Feldstein Econometrica paper that takes into account the fact that to forecast Ys, one needs to forecast Xs as well.

To make a long story short, after a couple of quarters, the confidence intervals blow up (I don't have the paper at home, but when I go in to the office this week I will scan some pictures from it), meaning that after a few quarters, metropolitan office demand could be just about anything.  Local office markets are much less complicated than an entire national economy.

Scariest thing I have learned over the past 48 hours.

Sometime within the next five years, half of LA's city budget will go to pensions.

Ugly University Buildings I have Known

I have spent a fair amount of time at six universities as a student or faculty member.  Five have astonishingly ugly buildings.

We begin with Mather Hall (below left) at Harvard: the Robert Taylor Homes of college dorms.  On the right is USC's Hoffman Hall, which was actually designed by a great architect, IM Pei.  It shows we all have bad days.

Next we have the Academic Center at George Washington (below left).   The photo makes it look nicer than it is.  Huntsman Hall at Penn is the new Wharton Building.  It is a little, er, out of scale for the surrounding neighborhood, but if Philadelphia ever returns to an agriculture based economy, it will have a really nice silo.  

Finally we have the Humanities Building at Wisconsin.  My understanding is that this ironically named building will soon be torn down.  That will certainly produce addition from subtraction.




The Indian School of Business was designed by John Portman, and is simply beautiful.

Housing Inventories

The bad news for the US housing market: despite strong resales in April, the country had about 8.2  months of inventory. For real house prices to stay stable, inventories need to be in the four to six month range, and because inflation is nearly non-existent for the moment, downward pressure on real prices means downward pressure on nominal prices as well. This could be a problem for a housing market that had relied so heavily on FHA loans, which have lax downpayment requirements.

Things here in California are better:

The data come from CAR.  Inventory under $500K is pretty thin, meaning that even if there is a shadow inventory that comes on line, California should be able to avoid much in the way of further price declines.  Just as interesting to me is that while the $750K+ inventory is still pretty large, it has shrunk pretty dramatically.  I actually wonder how these houses are getting financed--are there that many affluent buyers with cash?  When I talk with lenders, they are telling me that to get a decent rate at $1 million+, buyers need at least 25 percent down, and sometimes more.

Sarah Ritchie reminds me that today the Brooklyn Bridge is 127 years old.

Every time I visit New York, I am stunned at what a remarkable human accomplishment it is. I think there is a pretty good chance that among the most important contributors to that accomplishment are its bridges and tunnels.

I cannot think of any city in the world with remotely as many impressive bridges. The names roll off the tongue: Brooklyn, Manhattan, Williamsberg, 59th Street, George Washington, Verrazano, Throg's Neck and Whitestone. London, Paris, Rome and Seoul have very nice bridges, too, but because their rivers are so much narrower than the East (yes, I know it's not really a river) and Hudson, the bridges don't quite so stir my imagination. San Francisco's bridges span great distances, but there are only two that are impressive (the San Matao and Dumberton bridges are just strips of pavements on pillars, and the Richmond Bridge is, well, "interesting").

So happy birthday to Brooklyn Bridge, the first of a wonderful family.

Annoying anti-car headline of the day

From Richard Florida's twitter feed, I get:

One Hour Spent Driving = 20 Minutes Lost Life Expectancy:

So let us think what this means for people who drive one hour per day every day for 60 years. Expected life expectancy is reduced by 60 years times 365 days per year times one hour times 1/3 hour of life lost per hour of driving. This all comes to 7227 hours, or about 300 days. So driving every day for one hour means we lose 10 months of life expectancy (move these numbers around as you wish).

But what if we weren't able to drive at all (and buses and shared-rides vans count as forms of driving)? I am guessing we would be much poorer--mobility has at least something to do with our affluence. Maybe we wouldn't be as well nourished. Maybe we would face more economic stress. I can't be certain, but I would be willing to bet that if we stopped driving altogether, our life expectancy would fall.

I have long supported Pigou taxes on the negative externalities created by automobiles. I support subsidies for transit as a matter of social justice. But do I think cars have provided a net benefit to living standards and life expectancy? Sure!

David Barker comments on the growth path of GDP

He writes:

I just did a quick Chow test to see if there is a structural break in per capita GDP growth between 1935 and 2009 and there is not.

This is just a growth over time model (log GDP on time), and I also checked consumption and disposable income. If I did it right, there are no breaks - not even close. So you are right about Wallison, but one can't draw the opposite conclusion either.

I think in general it is difficult to draw casual inferences about macroeconomic data. More specifically, I agree with David that there is not sufficient statistical evidence to ascribe a cause to the relatively weak performance in growth after 1980. I do think rampant deregulation of financial institutions has been on net harmful (we seem to have financial crises more frequently now), but we haven't sufficient numbers of data points to establish that fact scientifically.

But it is also true that what Wallison wrote is demonstrably false. Life was not barren in the pre-Reagan years, and it has not been the land of milk-and-honey since. The evidence, limited thought it may be, is consistent with the idea that the New Deal was a good thing. On the other hand, Wallison wears very nice suits.

Peter Wallison: Opinions without Data

He writes in his screed against financial regulation:

In the rapturous days after Barack Obama's victory and the Democratic congressional sweep that accompanied it, House Financial Services Committee Chairman Barney Frank declared that the new Congress would enact a "new New Deal." Few people really thought at the time that he or his party meant this seriously. After all, the original New Deal—as anyone who has read history knows—failed to revive the economy.

Indeed, the modern era of rapid economic growth commenced after both Democratic and Republican presidents undertook to lift costly and stultifying New Deal regulations.

Using the National Income and Products Account, I looked at real annual GDP growth between 1933 and 1980 (the stultifying years) and 1980 to 2009 (the "rapid economic growth" years). Between 1933 and 1980, GDP grew by about 8-fold, or more than 4 percent per year (actually 4.5 percent per year). Between 1980 and 2009, real GDP did slightly better than doubling, or 2.7 percent per year.

I try to respect people whose points-of-view differ from mine, but who decides to let this guy waste ink?

USC: The Spirit of Transformation

The frustrations of Academic Life

Was about to ship a paper off for journal consideration and discovered a mistake. I am just hoping it turns out not to be material.

A little experiment in house price indexes

Among my favorite web site's is Morris Davis'. One feature of the site is a page of data that Morris and his colleagues have developed: among them is an estimate of land prices in the United States.

This morning, I looked at the change in real land prices in the United States. Between 1980 and 1997, real land prices (as deflated by the CPI) increased by 4.4 percent per year. I chose 1980 because Morris told me the data has some problems previous to that year, and 1997 because that was the year of the first inflection point in the movement of house prices (an even more impressive inflection began around 2002). The 4.4 percent rate is thus almost certainly fundamental, and does not reflect any bubble.

The question I wanted to ask is whether land is now over-valued or under-valued based on a 4.4 percent long-term real growth path. Unfortunately, Morris' data ends with 2007. For land values to have returned to a 4.4 percent long-term growth path, land values would have needed to have fallen about 1/3 since the end of 2007.

Actually, it is plausible that they have fallen more than this. From the end of 2009 to the end of 2009, the Case-Shiller house price index fell by 20 percent. But the value of structure changes very slowly, while the value of land changed quite rapidly. Land makes up about 1/3 of house values, implying that for house prices to fall by 20 percent, land prices must have fallen around 60 percent. Let's say that falling labor costs and commodity prices meant that construction costs fell a little bit, and that land value fell by only 40 percent. This still means that relative to its long term fundamental trend, land values overshot on their way down.

[Update. Morris has more recent data on the Lincoln Institute Web Site. His estimate is that land prices fell by 53 percent just between the end of 2007 and the beginning of 2009. This suggests that relative to long term trends, land prices overshot downward.]

Stunning Overbuilding Fact of the Day

I am listening to a presentation at the Homer Hoyt meetings on the condo meltdown in South Florida. Developers planned on building 95,000 units in the city of Miami between 2002 and 2007. In the 2000 census, the whole city had 163,000 units.

How do you produce a 63 day winning streak?

Here is how I have been trying to figure this out. Suppose we wanted to figure out what a daily winning percentage had to be in order to observe a 50 percent probability of a 63 day winning streak. It would be (.5)^(1/63), because the probability of 63 straight wins would be Pr(one win)^(63). It turns out that (.5)^(1/63)=.989, which I will round to .99.

Now lets say a firm has a proprietary trading model that is correct 51 percent of the time. This means that on the average day, it will come out ahead (suppose all trades are $1 trades). But if a trader makes one trade a day, he will close the day ahead only 51 percent of the time. If he makes 100 trades a day, however, while his winning percentage per trade remains the same, put his winning percentage per day goes up a lot. Specifically, the standard error for a daily outcome goes down by 1/10, from sqrt(.51*.49) to sqrt(.51*.49/100), or from about .25 to .025. The chance of finishing the day losing on average is based on how many standard deviations away .5 is from .51. In this case, it does from .01/.25 (or not far at all) to .01/.025, or .4 standard deviations away. In a normally distributed world, this means there is a 65 percent chance of finishing the day ahead, assuming each trade has a .51 batting average and 100 trades per day.

To get to winning 99 percent of days, we need to get the standard error for the day to be sufficiently low that .5 is more 2.4 standard deviations away from .51, so the standard error needs to be .01/2.4 or about .004. So we need to find X such that sqrt((.49*.51)/X)=.004. or X=.25/(.004^2)=15,625 trades per day.

Three big assumptions go into this calculations. First, it assumes a stable model. Over the course of one quarter, this may be reasonable. Second, it assumes a model with a 51 percent winning percentage. This is a huge assumption (I do not know what a reasonable number might be). Third, it assumes normality. This is probably not too bad; we do know that Chebyshev's Inequality says that (1-1/k^2) share of any distribution must be within k standard deviations of the mean. This means that 99 percent of any distribution is within 10 standards deviations, but that is an extreme outcome.

Virginia AG Cuccinelli is not only trying to kill Academic Freedom, he is trying to kill America's economic advantage

Mark Thoma sends me to Barkley Rosser:

This is in today's daily links, but I think it deserves a bit more notice:

Virginia AG Cuccinelli Out To Kill Academic Freedom, by Barkley Rosser: Friday's WaPo reports that Virginia Attorney General Ken Cuccinelli, following up on his efforts to end efforts by state universities and colleges to avoid discriminating against GLBT folks, has decided to interfere directly in scientific research in a criminal way. In particular, Cuccinelli is claiming that climate scientist, Michael Mann of hockey stick fame, engaged in billing fraud with the state while working on this subject while a professor of environmental sciences at the University of Virginia, where he has not been located for some years (now at Penn State). Cuccinelli is demanding all kinds of emails and other materials from the university, apparently attempting to imitate the climategate gang that did this over at East Anglia, only to end up with no fraud being discovered.

I think that some of the critics of Mann's work were correct, but this is an outrage. There is no evidence at all of fraud (and those claiming the email in which he spoke of using a "trick" as evidence for this do not understand or are willfully misrepresenting how this term is used in these situations) on his part, whatever errors he may have made in his study of the hockey stick (and it really does not matter exactly what the temperature was 1000 years ago; I have posted on this here previously). ...

Beyond the fact that I find Cuccinelli's attacks to be morally repugnant, they are also aiming for the heart of what remains of America's economic advantage: its ability to innovate. One of the reasons we are so good at innovation is because we have a research culture like no other nation, a culture that comes from universities that are, by world standards, intellectually free and well supported. The smartest people from all over the world come to the US for its universities, and then they stay to do research and to start businesses. Does the Ayatolla Cuccinelli really want to undermine this? Wait, don't answer that...

Two items in this morning's LA Times bring out my inner conservatism (small though it may be)

(1) The LA City Council is considering forbidding landlords of rent stabilized property from raising rents at all. I am certainly not a Friedmanite, but all credible evidence suggests to me that rent control leads to under-provision and under-maintenance of rental housing, and is distributionally unjust (Henry Pollakowski has shown that the benefits of rent control in New York accrue largely to high income people). But what LA is contemplating is even worse: to tell landlords that they can expect to be allowed to raise their (well-below-market) rents by 3 percent and then renege sends potential investors in Los Angeles the message that they can't rely on anything the city tells them. It may help explain why the city (as opposed to the region) has been unable to create jobs for a long time now.

(2) Al Gore bought a 6500 square foot vacation house in Montecito. I get why he might need to own one large house--if he uses his house for business purposes and has large receptions at it, he needs the space. I even get why a former Vice-President may not be able to fly commercial--the security problems are just too large. But for Al to own two large houses is just rank hypocracy. One of the reasons I can't take the GOP seriously is that it seems permeated by holier-than-thou hypocrites. Even though I generally agree with Al Gore on policy, he does at times make it hard to take him seriously.

Yet another reason why real estate is so interesting

I was listening to the BBC last night on the general election in the UK. One commentator stated that the reason coalition government wouldn't work in the UK is because of the design of the House of Commons: the government is on one side of the house; the opposition directly across on the other. It is all about the building!

What Milton Friedman got wrong

Friedman had two fundamental problems with business regulation. His first is that the business would capture the regulator, and therefore use regulation to establish monopoly power. My field leads me to find this line of argument compelling: real estate developers love (regulatory) barriers to entry that keep competitors from building.

His second, though, is just wrong. He argues that in order to preserve their reputations, businesses will self-regulate. Among other things, this ignores that managers often have short-term horizons. It also ignores that when large businesses implode, they leave victims with whom they never engaged in a transaction in their wake. BP did nothing illegal--how's that reputation thing working out? And having now read a whole lot on Goldman-Abacus (including the SEC complaint, the response on GS's web site, the offering circular, and excellent commentary from James Surowiecki, Yves Smith and others), it is not clear to me that Goldman did anything illegal or actionable (but I could be persuaded to change my mind). It is just that what it did (including investing long in CDS) should be unambiguously illegal and actionable. I can't think of anyone who had a bigger reputation franchise than Goldman.

Could we finally get a Pigou Tax on gasoline now?

Lisa Margonelli in this morning's New York Times.

The Deepwater Horizon spill illustrates that every gallon of gas is a gallon of risks — risks of spills in production and transport, of worker deaths, of asthma-inducing air pollution and of climate change, to name a few. We should print these risks on every gasoline receipt, just as we label smoking’s risks on cigarette packs. And we should throw our newfound political will behind a sweeping commitment to use less gas — build cars that use less oil (or none at all) and figure out better ways to transport Americans.

I think it is safe to say that economists across the political spectrum approve of gasoline taxes. To the extent they are regressive, the revenue they raise could be used to provide better mass transport subsidies for low income people. Maybe now there will be political cover to do something.