I took a given month's traffic and divided it by the average of the 12 months around it (5 1/2 months on each side plus the month itself). I repeated this process over the years and came up with the following seasonal adjustments (and used them to create the chart below). I'm fairly comfortable with them, but they are hardly locked in stone.
Jan: 0.9352, Feb: 0.8362, Mar: 0.9136
Apr: 0.9983, May: 1.0315, Jun: 1.0303
Jul: 1.0641, Aug: 1.0993, Sep: 1.0760
Oct: 1.0965, Nov: 0.9960, Dec: 0.9231
July through October are clearly the big months for imports. We've seen three of of the four so far and they are all down year over year (in Los Angeles anyway). The data for October is due to be released soon. I'm not all that optimistic things will be looking better.
I think it is relatively safe to say we aren't growing our container import traffic through the Port of Los Angeles at a rate of 1.01% per month any longer (sorry China!). We made a valiant effort to restart it in 2004 by continuing to keep interest rates phenomenally low (note how 2005 through 2006 follows the line again), but alas, all that nearly free money seems to have broken something. I suppose we can try to restart it yet again by devaluing our currency even more, if we like the thought of $200 oil that is. Maybe we do.
November 8, 2004
A Dollar Warning
The larger worry is that the Bush Treasury, and perhaps Mr. Bush himself, seem to have fallen for the notion that a country can devalue its way to prosperity. This is the patent medicine of the manufacturers' lobby, as well as the kind of economist who has done so much for Argentina, Mexico and other nations over the years. Britain tried this in the 1970s, and had to call in Margaret Thatcher to save the country from sinking to Third World status. The Carter Administration also tried talking down the dollar and ended up inspiring a global run on U.S. assets.
See Also:
U.S. Port Traffic Down
Source Data:
Port of Los Angeles: Statistics
Update: I was not including the current month's data in the denominator when calculating the seasonal adjustments. I should have been since I was trying to see how far each month was off of the average. I've therefore updated the seasonal adjustments and the chart with this correction.
a couple notes.
ReplyDelete1) its interesting you were able to use a natural e for your model. Its traditionally easier just to ln the data, that way percentage increases and decreases are easier to read. Its also easier to have a clean regression that way.
2) It would be interesting if the data was adjusted to the yen exchange rate. A stronger yen, naturally decreases everything that goes through LA.
Thank you for your observations.
ReplyDelete1) I'm just using Excel's exponential growth trend line so it does all the work. Of course, it is interesting that it followed an exponential growth curve of about 1% a month for so long. Clearly that wasn't sustainable.
2) That would also be interesting. Unfortunately, that too doesn't help them or us much. The extrapolated conclusion would be that they ship us just one container someday and we pay a LOT of our money for it. I'm just not sure it would take a billion people to assemble the goods within it. ;)
I should add that my seasonal adjustments did not assume there was exponential growth. I simply chose a KISS (keep it simple, stupid) method that could approximate the adjustments regardless of where the data headed.
ReplyDeleteI only added the exponential growth curve after I had seasonally adjusted the data.
A book could probably be written on how best to adjust for seasonality. It is controversial no matter how it is done. The goal is to find patterns in the data and filter them out. There's no proof that the results will be more accurate though. It is possible that a lot of that noise that's being filtered out was actually perfectly good data (screaming for attention perhaps). There's really no way to know for sure.
In this instance, filtering out some of the [Christmas] trees seems to make the forest a bit easier to see though, so I'm relatively pleased with the results. It also helps determine if the trend really is down (year over year growth taken on its own can be a bit misleading if the previous year was an anomaly).