Links

1. Teacher efficiency is analyzed in LA by the LA Times.  It turns out that what teachers are paid for are not related to their teaching effectivity.  The next step towards sanity would be to create these links, and to find some ways to actually fire the poorly performing teachers.

2. A survey from Fidelity suggests that a lot of people are borrowing against their 401(k)'s.  This is another factor keeping consumer spending high relative to private sector income.

3.  Tyler Cowen looks at whether it was the fiscal policy of Australia that has helped their economy avoid a deep recession. He mentions Australia's connection with China's economic demand, which seems to be the dominant variable. However, he leaves out their high trend inflation, which helps them avoid problems other central banks reach when they hit the zero bound.

4. A caricature of futurists. Not entirely inaccurate.

Inflation uncertainty

Tyler Cowen linked to two posts highlighting how high monetary policy uncertainty is today due to the high levels of inflation variability. The striking thing about these posts is that at our low inflation rate, the standard deviation of inflation is just as high as it was during the 70’s.

Using the month on month changes of inflation (bottom of top chart) instead of year on year changes (top of top chart) we see inflation variability spiking in mid 2006 rather than early 2009. Furthermore, the comparison the blogs make to the 70’s glosses over that by very similar measures, inflation uncertainty was much higher in the 50’s.

Of course, this may just be due to the crazy adjustment from the post war economy, so the policy uncertainty of the 50’s was probably low despite high the variability of inflation.

The high level of inflation volatility and therefore policy uncertainty can help explain why there are deficit hawks and gold bugs with nominal ten year yields at 2.6%.  Ignoring the increasing deficit, the US treasury is going to have to roll over 2.5 trillion dollars worth of debt* in the next year alone, so maybe there is something to be uncertain about.

*To look this up for yourself, go here. Download the excel file and look at marketable debt. There are over 1.7 trillion dollars worth of Treasury bills and .74 trillion worth of Treasury Notes payable in the next year. The amount of inflation adjusted notes coming due is under 40 billion and there are no long term treasury bonds coming due.

Links & Critiques

1. Michael Pettis has a post on why he thinks it unlikely that China will be able to rebalance their consumption. The main flaw is his analogy to Japan's economy. One of the main reasons Japan had trouble rebalancing was that Japan had almost fully converged with other developed economies except there were a lot of micro-inefficiencies in Japanese consumption markets. When houses are built one at a time and the retail market is still dominated by mom and pop stores the cost of consumption is rather high, reducing a household's willingness to consumer versus save on the margin. Growth stopped in Japan more because the convergence process was done and Japan couldn't overtake the rest of the world any more in places it was ahead.  To argue that the readjustment in China is going to have the same effect as it did in Japan, it would have to be assumed that China's economy has reached its steady state and will no longer catch up with the world.

2. Arnold Kling looks at momentum in employment. This is another reason why markets care more about the headline number.

3. Some school districts don't understand bulk discounts. No wonder they are running out of money.

More wasted money

The House recently passed a party line vote to spend more money on education and Medicaid. The $10 billion dollars to the states comes with many strings attached, notably that they do not get the money if they decided to cut spending on education as a share of total revenues. I’ve touched on the irrelevance of medical spending past a certain level before, so let’s look at the effectiveness of education spending.

Source: “Does Spending More on Education Improve Academic Achievement?” by Dan Lips, Shanea Watkins and John Fleming

Of course, the chart is somewhat misleading.  Education scores are range bound while spending is not, but it is notable that they often seem to be going in opposite directions*. 

Using additional cross country data from NationMaster, we see that government spending on education beyond a certain point doesn’t seem help very much anywhere in the world:

 

Total education spending as a percent of GDP isn’t any more correlated with scientific literacy, but at least the correlation isn’t negative.

 

So the government may be wasting money here, which just gives us more evidence that  α is very high.

*At younger ages there is some evidence of improved math scores, but these improvements are negligible by age 17. Click through the article for more charts.

And the links came to enlighten us all

1. Tyler Cowen tells a good story at his 2009 TEDx Midatlantic talk. (HT: Patri)  His point that thinking in terms of a narrative often makes things too simple, and many existing narratives are misleading.  

On non-fiction: "Nonfiction is in a sense the new fiction. The book may happen to say true things, but again everything is taking the same form as these stories" 

On a useful heuristic: "As a good rule of thumb just imagine that every time you are tell yourself a good versus evil story, you are basically lowering your IQ by 10 points or more."

On books about cognitive biases: "None of these books identify what to me is the single central most important way we screw up. And that is, we tell ourselves too many stories or we are too easily seduced by stories. And why don't these books tell us that? It is because the books themselves are all about stories."

Listen to the whole thing, as they say.

2. Bryan Caplan discusses the rising return to hours put into college education.  This research is a big part of why I don't believe the premise from other science research that things are changing mainly because it is harder to reach the frontier of science. Science is changing because on the margin education is no longer about education and research is no longer about research.

3. Scott Sumner has already reacted to the Fed meeting later today.  If the market doesn't go up, he doesn't like it.  Of course, I'm left to wonder whether he would be that upset if the market stays relatively neutral, since it appears that some form of quantitative easing or reinvestment in mortgage bonds, or cut in the interest paid on excess reserves has already been priced into the market.

4. Greg Mankiw points out some unfortunate estimates on the future performance of Medicare. The health care sector is one of the least efficient parts of society, yet it doesn't seem like the United States is on any path close to slowing down the amount of resources wasted there.

Comparing NFP Surprises (Delayed)

The nonfarm payroll numbers were released by the BLS today (Note: This was written on Friday, but the post was delayed to a DNS attack on Posterous). The headline numbers were much worse than expected, with total jump numbers disappointing by 66 thousand workers (131k jobs lost) and the private sector disappointing expectations by 19 thousand workers (Only 71k jobs gained).  On the other hand, average hourly earnings and the average weekly hours of all employees was slightly higher than expected at 0.2% MoM for earnings instead of 0.1% and 34.2 hours worked instead of 34.1 hours. 

The markets reacted to this data by selling off.  This isn't surprising, as the headline number is what everyone looks at. However, it is interesting to note that the payroll surprises in hourly earnings and average weekly hours amount to be equivalent to an earnings increase of more than 500k in new jobs.

Something is going on here. 

1. The headline number is still more important because it is a better predictor of future economic activity.
2. Average weekly hours and hourly earnings are not that meaningful or accurate as more and more of the economically important jobs are no longer based on hourly earnings or weekly hours.
3. Average weekly hours and hourly earnings have rounding issues where the actual surprise only looks large because not enough significant figures are used for the estimates/reports. Maybe economists thought the hourly pay would go up by .013% and it ended up going up by .018%, which is half the effect that the headline numbers suggest).
4. Other revisions are more important than the headline numbers.

The most important of these factors is number 2, as the reasons behind the variations in hours such as the fluctuations in pay to overtime manufacturing workers is definitely less significant in an economy with fewer manufacturing jobs relative to the total number of jobs.

Vague Fed

Arnold Kling complains about vague fed policy.

Everyone is acting as if in order to maintain the Fed's independence, the Fed must be allowed to be vague about its targets, vague about how it might achieve those targets, secretive about how it thinks its actions influence those targets, and ad hoc in its approach to deciding when to take action. I would suggest re-examining such assumptions.

Part the Fed's focus in non-panic times is spent making sure they don't surprise the markets too much. In order to avoid surprising the markets, the Fed has to be vague in order to slowly walk the market towards their plans, which they are probably still in the process of figuring out.

"My suggestion is that, if you get asked those questions, just say we're examining nontraditional methods and there are many different ways in which we can address the issue. I would be as nonspecific as you know how to be. The major reason is that I don't think we will know until we start to address the issue." - Greenspan in 2003, discussing what the Fed might do as interest rates approach the zero bound

If they stated their shifting views in real time the market might react in an extreme manner when the message changed, and unexpected sources of volatility are generally harmful to a leveraged economy. Furthermore, if the message changed too often, there would eventually be points where the market might ignore the Fed due to the low signal to noise ratio. Then the Fed would lose control of the very useful tool of managing future expectations. So in order to keep this tool, they have to be very sparing with their use of clear messaging.

Links from last week

1. Falkenblog presents a business cycle theory based on mimicry. Basically, in every boom there are entrepreneurs who know how to make money and then other overconfident entrepreneurs who are merely copying some measurable quality of these entrepreneurs.  Each time, the traits being copied are somewhat different, and because these qualities have worked in the past it means that quantitative investors can "prove" what they are investing in will work (Housing prices never go down!). At the same time the differences in each cycle make sure economic models will not correctly predict how the cycle will play out.  It will be a very good framework for thinking about things during the next boom. Read the whole thing.

2. Robin Hansen provides reasons for an ambitious person to be somewhat overconfident if they want to be an executive. He doesn't think he is doing this, but it is the conclusion that I reach from reading the data he provides. CEOs, doctors and lawyers are all overconfident, so these are probably useful signals to send if joining their ranks is a personal goal. Along the same lines, this is why people in finance are inflationary because they do not make money in a deflationary environment.

3. Macroblog notes that the correlation between median CPI & the M2 is back.  He also has a post up discussing the interest paid on excess reserves. First, he notes that because the interest paid is small it didn't effect the choice between deciding to accumulate reserves vs. lend to the private sector by very much. Towards the end of the article, he notes that paying interest on excess reserves is to make sure that banks hold more reserves so that the payment system runs efficiently.  This contradiction is a little confusing to me. Regardless, the point in the middle of the post about a zero interest rate on reserves being unlikely because this would interfere with the liquidity of the money market market is interesting because it highlights that many of the more unique propositions like setting a slightly negative interest on reserves to encourage lending are beyond the pale at this point.

4. Starcraft 2 came out this past week. Rather than discuss whether or not it is a good idea to bet against South Korea of the performance of the US tech sector for the next quarter, I'll point you to this comic.

Rounding

I remember learning the rules about rounding significant figures when the variable to be rounded is a 5 (Round to even numbers, not odd numbers), but my teachers never explained why this was done. While trying to verify whether odds or evens were preferred, I learned the true reason behind why the seemingly weird rounding system exists.

Fantasies about fives
Students are sometimes told to increment the least significant digit by 1 if it is odd, and to leave it unchanged if it is even. One wonders if this reflects some idea that even numbers are somehow “better” than odd ones! (The ancient superstition is just the opposite, that only the odd numbers are "lucky".)
In fact, you could do it equally the other way around, incrementing only the even numbers. If you are only rounding a single number, it doesn’t really matter what you do. However, when you are rounding a series of numbers that will be used in a calculation, if you treated each first-nonsignificant 5 in the same way, you would be over- or underestimating the value of the rounded number, thus accumulating round-off error. Since there are equal numbers of even and odd digits, incrementing only the one kind will keep this kind of error from building up.
You could do just as well, of course, by flipping a coin!

Over 10 years later the rule finally makes sense to me, but I might decide try the coin flipping approach instead the next time this comes up.

Wordsum Sample Test

Over at MR, Alex Tabarrok highlighted a chart made by Razib Khan about a Wordsum test (a test given in the GSS survey that correlates with IQ) and those who drink alcohol.  In the comments, Razib pointed us towards the GSS browser to test a commenter’s theory that drinking alcohol matters less among those that have completed college.

Attended college:

Didn’t attend college:

After looking at this post, one of my friends wanted to see an actual Wordsum test. Here is one illustrative example that I found in a paper:

From: Huang, Min-Hsiung, Hauser, Robert M. Trends in Black-White Test Score Differentials: II. The WORDSUM Vocabulary Test (1996) (page 30 of the pdf)