From Kaiser Fung, Harvard Business Review Blog
In their best-selling 2013 book Big Data: A Revolution That Will Transform How We Live, Work and Think, authors Viktor Mayer-Schönberger and Kenneth Cukier selected Google Flu Trends (GFT) as the lede of chapter one. They explained how Google’s algorithm mined five years of web logs, containing hundreds of billions of searches, and created a predictive model utilizing 45 search terms that “proved to be a more useful and timely indicator [of flu] than government statistics with their natural reporting lags.”
Unfortunately, no. The first sign of trouble emerged in 2009, shortly after GFT launched, when it completely missed the swine flu pandemic. Last year, Nature reported that Flu Trends overestimated by 50% the peak Christmas season flu of 2012. Last week came the most damning evaluation yet. In Science, a team of Harvard-affiliated researchers published their findings that GFT has over-estimated the prevalence of flu for 100 out of the last 108 weeks; it’s been wrong since August 2011. The Science article further points out that a simplistic forecasting model—a model as basic as one that predicts the temperature by looking at recent-past temperatures—would have forecasted flu better than GFT.
In short, you wouldn’t have needed big data at all to do better than Google Flu Trends. Ouch.
In fact, GFT’s poor track record is hardly a secret to big data and GFT followers like me, and it points to a little bit of a big problem in the big data business that many of us have been discussing: Data validity is being consistently overstated. As the Harvard researchers warn: “The core challenge is that most big data that have received popular attention are not the output of instruments designed to produce valid and reliable data amenable for scientific analysis.”
This week's Economist magazine talked about the rise of shadow banking, or rather the continued and growing presence of it. While I have read extensively and will provide some of my views on this, I'm just going to point out something interesting that I've thought of:
Lending to households or consumers who may be not very 'financially-capable' is an action by a bank that can be viewed very differently by different parties at different times. Some will say that doing so is financially irresponsible - these borrowers do not really have the ability to return these loans. Many then conclude that the financial institutions are nothing but liars and cheaters, especially with stories of apparently unsuspecting families accepting loans that would be impossible to repay in the long run.
While the stories have their validity, this opinion however has incomplete reasoning behind it. Why would a profit seeking bank do so? Excluding parties with mandates to do so (like Freddie Mac and co), the banks often do so because these borrowers use collateral (such as real estate) that is deemed stable and growing in value. The accurate pricing of both risk and hence value of these assets is hence a big question - a static analysis of real estate prior to the burst of the 2007-8 bubble may suggest real estate is a fast appreciating and historically fairly stable asset. However a dynamic and perhaps less quantitative look at real estate will reveal that there has been a momentum of speculative demand pushing speculative demand for real estate, and that soon without real demand or with a minor shock, housing prices could just collapse like a pack of cards. While I would say that I'm dissatisfied with the lack of a quantitative and rigorous method to evaluate these assets in a dynamic way such that predictions can be made and a thorough understanding of the forces behind asset prices can be accomplished, it is not an overstatement that current financial models are insufficient in the regard of assessing shifts and changes in the market due to other parties - a bit like game theory - as I suspect that most models simplify the effects of other market players as 'efficient'.
Anyway, the other view of lending to borrowers who are less credit worthy is that it is actually a good policy because the market is inefficient in not providing them the loans because of potential mis pricing of risk and positive externalities such as community benefits from increased house ownership. For me, the arguments about the positive effects of house ownership can only make sense with the tools to weigh the other side of the equation - the pricing of risk and assets.
One of the reasons why the invisible hand seems invisible is that sometimes it's not there. Joseph Stiglitz
Overheard at Spanish classes at Las Lilas yesterday:
A: Como ir a Dhoby Ghaut?
B: Tomas la metro! Metro de Singapur es muy bien.
Of course, I was corrected after that.
What is so dangerous about a can of coke that requires it to be taxed by the government?
Nothing really. Not if it is just one can. The problem lies not in soft drinks per se but in the lifestyle and over-consumption of it. Obviously the government cannot really tax "over-consumption" of soft drinks (there are too many vendors and ways to bypass the rules by simply getting your healthier friends to buy as part of their quota), not in our present society at least.
Hence, the second-best measure is to put a "fat tax" or a blanket tax on all soft drinks (or just on the upsized ones), so that people will be discouraged from drinking soft drinks. In America, the "fat tax" is specifically targeted at lower income groups who tend to over-drink these high calories drinks as part of their unhealthy but cheap diet.
As suggested, the fact that these individuals drink high-caloric soft drinks suggests that demand is pretty inelastic (There aren't many cheap alternatives out there and, hey, soft drinks are really quite nice to drink...). This means that the tax will fall mostly on the consumers, who aren't that wealthy to start of with. Analysts have suggested, and I tend to agree, that such a tax may not be that effective in reducing the quantity consumed by poorer folks in the short-run due to the income-substitution effect dominating the price-substitution effect (Giffen good effect). Essentially, not-so wealthy people may perceive that their income is even lower (it is in terms of purchasing power of soft drinks) and hence have a higher tendency to consume inferior goods like soft drinks. This effect counteracts the price-substitution effect caused by the higher prices of soft drinks. In the end, demand may still fall, but not by much.
In the long-run, however, it is plausible that beverage manufacturers or new entrants into the market will start to develop non-soft drinks that target the needs and demands of these social-economic classes of society. It is important therefore for the government to identify and clearly specific the content materials of soft drinks targeted by the tax. For example, a tax on corn syrup (which is reported to be the cause of unhealthy sugar content in soft drinks) will spur manufacturers to use healthier alternatives such as cane sugar.
An interesting point about corn syrup made by Willson was that corn is heavily subsidised as a crop in the United States. The heavy use of corn syrup (a cheap but unhealthy source of sugar) in the United States suggests that the real cause of the sugar problems may be the government subsidies on corn farming in the first place.