I remember seeing the ad pictured in this article as a teenager and being sure that it was implicitly pro-tobacco. I don’t really blame them and perhaps it’s unsurprising, but if the state is actually going to wind up paying higher health care bills as a result of smoking then maybe they need to hire actual professionals to head up their tobacco propaganda office.

You’re right, and I thought about doing it, but I don’t have much experience with GARCH models. Any chance you want to share some code that does it?

Businesses need to make money – no argument there. And customers do need to be accountable for their own money management. However, raising fees on something, somewhere regularly is a short term fix. A typical bank will set revenue projections, then review fees across the board to assist in meeting the objective. Where is innovation to build business? If economies of scale truly apply, fees should be going down…..except when banks make bad decisions, or take risky bets. Recouping falls on the head of the little guy…. I have empathy for all sides but the cycle needs to change.

I agree that if we know events cluster- that they have high auto-correlation, this makes sense. Earthquakes are a naturally occurring phenomena where this might be true.

But the social events – 9/11, market crashes – we don’t know that they have a stable, objective auto-correlation. In fact, 9/11 made it abundantly clear that they don’t have a stable rate at all. What are the odds of 4 full commercial jet planes being hijacked at once, and used all used as missiles?

Up to 9/11 most passengers would not interfere with a hijacked plane. Most hijackings had been for escape reasons, and no-one had used a commercial jet as a missile before. The passengers on flight 93 might not have reacted the way they did, if they did not think one hijacking wasn’t related to another. They immediate assumed a nearly perfect correlation.

Given a coordinated attack across 4 flights, how would we know that the auto-correlation hadn’t just increased? Perhaps bombs, rather than hijackers would be used. Data isn’t sufficient to answer that question.

For the doomed Concorde the situation was unclear immediately after the accident. If the crash was due to debris on the runway, there would be no need to update. If it was due to the age of the aircraft, a design fault or a Y2K problem, then one woiuld have to update .