Unpleasant Facts and Other Musings

Max Planck said “Science advances one funeral at a time.” He was implying that older scientists are too tied to the status quo to accept evidence challenging the view of the world that they helped shape. This means that younger scientists are often the key to advancing the frontier of knowledge, so I was very interested in the Core Economics blog post on youth creativity along with their links to an interesting WSJ journal article on the declining scientific productivity of younger scientists and a related paper on science policy by Benjamin F. Jones. 

The WSJ article highlights the inverted U of productivity, where a person’s creative productivity peaks sometime between the ages of 25 and 50.  For physicists, mathematicians and poets, the age of peak productivity has historically been closer to 25 than 50 as fluid intelligence peaks earlier.  For biologists, historians, philosophers and novelists, the peak age may be closer to 50 than 25, as crystallized intelligence is more important than fluid intelligence in these fields.  The current NIH grant system tends to give grant money to established scientists, which some worry means that many scientists are getting money only when they above the historic age of peak productivity causing the relative productivity of younger scientists to drop. 

In the paper, “As Science Evolves, How can Science Policy?” Benjamin Jones tries to paint these changes as a natural reaction to the changing state of the process of science and thus as nothing to be alarmed about. His main point is that a scientific advancement pushes forward the boundaries of knowledge, the natural result is for scientists to be more productive later in life and to be more productive in teams as each member can contribute their specialized knowledge. He is not worried as much about whether or not the grey ceiling and bureaucracy in academia is preventing progress, but sees progress by on average older people and teams as confirmation that the frontier of knowledge is harder to reach.

It is very possible that the expanding frontier of knowledge is making science more difficult for younger people, but the knowledge barrier might not be scientific but social and political.  The evidence presented in the paper can be used to suggest that the bureaucratic obstacles are getting more difficult to navigate. The first place the paper looked at was the increasing age of noble prize winning scientists’ first great achievement and inventors’ first parents.  Both of these have increased since 1900.  Both of these may be due to an increased cost in money rather than time to get to the frontier of knowledge.  Advances in the frontier often put specialized tools farther outside the reach of amateurs. More recently, it may be due to the lack of personal control many scientists who depend on grants have over their research until they get their own grants and labs at a later age.

The paper also focused on evidence of the increased effectiveness of teamwork. Rather than a sign of scientific specialization, increased teamwork can be a sign that certain people are specializing in some aspect of the bureaucratic process. (The paper also shows that between 1975 and 1993, individuals working on teams of patent applications are more likely to jump fields in patent applications than solo inventors while in 1975 they were both equally likely to jump fields. This evidence could suggest an increased role of specialization, or it could suggest that on the margin would-be patent trolls have found it much easier to monetize if they work in teams instead of individually.)

Jones also found that papers published by teams have a higher probability of a large amount of citations.  A hypothetical example of non-scientific specialization could work as follows: Maybe one member is really good at getting grant money, while the other is the one with the ideas and the experimentation process (the one who might otherwise be a solo author), another is good at writing up the paper in the proper form and another is good at getting people interested in continuing their line of inquiry, thus earning citations and recognition for their work. If citations are often a tit for tat social game that academics must learn to master and that is improved by having more connections, which is another barrier in the frontier of knowledge that is relatively unrelated to the accumulated advances science.  This team advantage has reversed the solo advantage in a number of fields since the 1950’s, so these fields may be evidence that once fertile fields have become more and more politicized.

There is definitely something going on in the changing process of science, but to focus solely on the frontier problem as the explanation for recent changes papers over important structural problems. Treating recent changes as the new status quo and implementing policies with goals such as encouraging more teamwork, like the paper suggests, may make it even harder for the scientist to out compete the bureaucrat... in science.

The money spent on newspaper advertising as a percent of GDP has been declining since the 1950’s.  To some degree, this slow decline is expected as the industry gets more efficient and advertising demand takes up fewer resources relative to the rest of the economy. However, when looking at newspaper advertising revenue divided by GDP, there is clearly something more going on than a gradual improvement in efficiency.

Source: GDP data from the BEA, Newspaper ad data from the Newspaper Association of America. In 2003 forward online newspaper ad revenue was included in the calculation.

To put the drop from 2003 through 2009 in perspective, here is a chart of Google’s revenue (99% of which is from advertising) versus the advertising revenue of the entire newspaper industry:  These numbers are before netting out the cost of acquiring this revenue.

Google doesn’t incorporate all of new media, but it is interesting that one company in the same general industry of advertising is about to eclipse an entire industry. The death of newspaper ad revenue has been on horizon for a while. Newspaper ad revenue has been declining for retail and classifieds since 2000. Classified ad revenue has been hit the hardest by the free competition on the internet are at one third of their 2000 peak. The competing retail ads aren’t quite free, so they have declined at a slower rate and newspaper retail ad revenue is at 50% of their peak. Despite the competition from the internet, national advertising and online ad revenue combined to keep pushing up newspaper revenue until it peaked in 2005. Right now, the one bright spot for newspapers is online ad revenue, which has increased from 2.6% of total revenue when it was first measured to 10% today. Unfortunately, online revenue is still very small and it is not enough to support the industry in its current form.

The high value of Palin’s 2010 Republican nominee Intrade contract is probably more representative of the wishful thinking of certain Palin supporters than an unbiased prediction of future probabilities. This happened in the last election with Gore’s VP contract, so Palin supporters probably shouldn’t get too excited about this market’s prediction.

Andrew Gelman links to an interesting piece by Dave Armstrong on the Heritage Foundation Economic Freedom Index. I have my own issues with some of the subtleties of the index, but the approach taken is a good example of how a quantitative approach can sometimes miss the qualitative point altogether.  In the analysis of the components of the economic freedom, the assumption of unidimensionality is made.

“Unidimensionality - that there is only one underlying source of variation. Here we are assuming that each indicator is an imperfect indicator of Economic Freedom and that once we take account of economic freedom, there is only random variation left over.”

By making this assumption, he can then decide that factors that aren't correlated to the common factor of the other variables do not actually measure economic freedom.  In the subsequent analysis, fiscal freedom, a measure of taxes, and government spending are found to be uncorrelated with the common factor of variables. This is where a quick qualitative check would have been helpful. The amount of money the government controls and taxes from the private sector is a priori related to economic freedom. Instead of assuming that government spending and fiscal freedom are not related to economic freedom, the assumption of unidimensionality should be reassessed. 

The new index created with the unidimensionality assumption that takes out government spending and taxes could be very interesting as a rule of law index, it just isn't an index of economic freedom.

The quantitative approach does yield some useful insights, as the article does make a good point about statistical significance of the change in rankings. The United States index score is still well within its historical range, so the United States's move off of the list of most free countries is not really statistically significant. The most relevant part of the ranking change is that over the ten year period certain countries have become more free, but that is a relative and not absolute US decline.

One of the aspects of the coming capital gains tax raise is that stocks that pay out their earnings in dividends become relatively more attractive when compared to stocks that retain their earnings or give money back to shareholders via buybacks and so are more likely to return value to shareholders via capital gains rather than dividends.

Low dividend/High Dividend Paying Stocks, unadjusted for dividends:

Note on creation of indices: (Done with Palantir Finance)

S&P 500 companies

Excluding lowest decile of 11 month return to filter out failing companies

High dividend stocks: Top Quartile of High Dividend Paying Stocks, current average dividend yield of almost 5%.

Low Dividend Paying Stocks: Bottom quartile, average dividend yield close to 0%

The big decline in 2000 through 2002 was the unraveling of the tech bubble.  The dip in 2008 looks cyclical, as companies at the closest to the epicenter of the boom and bust are less likely to be the ones returning money to shareholders. In the same way, the companies that outperform at the end of a crash are the companies who barely survived, and companies that barely survive are less likely to be distributing needed cash to shareholders.

Looking at the markets fall, there are three negative events that stand out as possible explanations: The ash cloud from Iceland’s volcano is still spreading over Europe, the University of Michigan released a lower than expected consumer confidence number and the SEC suing is Goldman Sachs (Suspiciously, on a day when there is news about how they ignored Allen Sanford’s Fraud in 1997).

1. The Ash Cloud: This seems back and should theoretically lower consumer spending and oil consumption temporarily, but the market didn’t really distinguish between European and US stocks today.

Lacking historical minutely data, I looked at the market close to market open futures data to determine if the U. of Michigan numbers have any market impact:

The impact looks rather negligible, perhaps because there are other more timely indicators of consumer confidence and divergences in this survey relative to expectations is as likely as not to noise. At the very least, any signal seems to be drowned out by the noise from earnings releases or other market data.

3. Today Goldman Sachs’s market capitalization fell from over 100 billion dollars to a little over 87 billion dollars on news that the SEC is suing them for fraud related to how they represented CDOs. No one thinks that Goldman is going to lose 13 billion dollars in the lawsuit, but because of the chance that it may be a tip of the iceberg type of event or that this negative news will hurt their credibility and therefore their business the market’s move may not be too excessive. Other financial stocks fell in sympathy, along will most other risk assets, either because of fears that these banks did the same, hedging by Goldman who knew of the lawsuit ahead of time, or because their regulation will be harsher if financial companies are perceived in a more negative light.

I’m glad that I don’t have to explain market moves every day, as usually “more buyers than sellers” or vice versa makes more sense than the majority of news stories coming out about market drivers.

Over the past few years, it has been interesting to see the United States slowly lose its top spot in the Index of Economic Freedom. In 1995 the United States was in 4th place, behind Hong Kong, Singapore and UK. In 2010 the United States is in 8th place, behind Australia, New Zealand, Ireland, Switzerland and Canada but no longer behind the United Kingdom, which fell to 11th place. This change is mostly due to other countries improving, although the United States has fallen from a score of 90 on corruption to 73 in 2010.

It is good to have an understanding of underlying relationships between assets. Knowing that gold goes up when the dollar goes down or that airlines and consumer discretionary stocks do not benefit from higher oil prices is important for someone trying to understand the market.  However, sometimes these relationships are given an importance that masks what is really going on.  Howard Simons has an article up at Minyanville that makes this mistake.

“The current partial contribution of crude oil prices to the S&P 1500 Supercomposite remains positive, 0.59% on a net weighted beta at last calculation. The reason for this is simple: The positive impact across the Energy and Basic Materials sectors outweighs the negative impact across the Consumer Discretionary, Consumer Staples, and Health Care sectors. 

He is right that higher oil prices are generally better for Energy and Basic Materials sectors than for the others. Energy companies selling oil and oil services make more money when oil is higher, and oil is highly correlated to the price of many other commodities that materials companies make money selling.  However, implying that the S&P 1500 was driven up despite the negative effect of oil on consumer discretionary stocks is proved false by simple data analysis.

The chart below shows the weekly percent change of oil versus the weekly percent change of the AMEX Consumer Discretionary Select Index from March 2009.

The relationship between oil and consumer discretionary stocks is positive, as both have been rallying due to more positive economic and financial expectations.  The more intuitive relationship of higher oil prices mean consumers have less money to spend on everything else is masked by the more extreme financial market effects. Simons does recognize this factor when he discusses that global economic growth is partially behind the high correlation between crude and the S&P 500, but it doesn’t make his earlier statement about sectors any more correct. 

The below chart looks at crude oil weekly returns vs. the relative beta adjusted weekly return of the consumer discretionary stocks (Consumer Discretionary weekly return times beta to S&P500 minus S&P 500 weekly return). Since October 1^st, even this relationship has been slightly positive.

The chart below shows how the relationship between crude oil and the S&P 500 is variable over time (The rolling 26 week T-stat of crude oil explaining the S&P 500). Theoretically, the relationship is positive when crude oil and the stock market are being driven by economic growth and it is negative when oil is acting more of a constraint on growth due to supply restrictions. When the oil squeeze comes, this chart will fall below zero.

The charts were made with tools from Palantir Finance, which released some good news today.