Crawl Across the Ocean

Tuesday, December 09, 2008

As Expected

Offered without comment, ok maybe one (at the bottom of the post)....

---
From the CBC, October 19, 2006:

"Economy seen cooling till 2008: Bank of Canada - The 2008 growth forecast remained unchanged at 2.8 per cent."

---
From the CBC, October 10, 2007:

"Economic growth is expected to cool to 2.3 per cent in 2008 — down from its previous forecast of 2.6 per cent"


---
From the CBC, January 24, 2008:

"Bank of Canada lowers 2008 GDP estimate: Growth to pick up later in the year and into 2009 ... In its latest monetary policy update issued Thursday, the central bank now says GDP will rise by 1.8 per cent in 2008, down from its October estimate of 2.3 per cent. "We will come through 2008 fine," Bank of Canada governor David Dodge said at a news conference."


---

From the CBC, July 15, 2008:

"...the bank said it expects economic growth in Canada this year of only one per cent. Growth is expected to rise to 2.3 per cent in 2009 and 3.3 per cent in 2010.

That is down from the lowered outlook the central bank offered in April, when it said it expected 1.4 per cent growth for this year. In January, the bank said it was looking for 2008 growth of 1.8 per cent.

The bank also sounded a warning over its outlook for inflation, saying that commodity prices continue to outstrip expectations.

"Assuming energy prices follow current futures prices over the projection period, total [Consumer Price Index] inflation is projected to rise temporarily above four per cent, peaking in the first quarter of 2009," the bank said."

The bank sees energy prices subsequently stabilizing and inflation is projected to return to its two per cent target in the second half of 2009.

Many economists see the central bank remaining on the sidelines for several months."


---

From the CBC, October 23, 2008:

"the Bank of Canada said growth is expected to pick up over the remainder of 2009 and to shoot to "above-potential" in 2010 ...

Asked if Canada was headed for recession, Carney would only say that economic performance will be sluggish for the next few quarters.



---
From the CBC today:

"While Canada's economy evolved largely as expected during the summer and early autumn, it is now entering a recession as a result of the weakness in global economic activity," the bank [of Canada] said.


---

As expected!! You can take these clowns seriously if you like, but you do so at your own peril, they certainly don't take their audience very seriously...

Labels: ,

9 Comments:

  • I don't get it. As the data change, forecasts *should* change.

    By Blogger Stephen Gordon, at 4:27 AM  

  • Spoken like a true economist!

    As the data change, forecasts should *not* change, because if they were accurate forecasts, they would have predicted the change in the data!

    Of course, forecasts should change when you realize they are wrong, but that's not really a sign of good forecasting.

    It is one thing if your forecast is generally accurate and you fine tune it slightly as more data accumulates.

    It is another thing if your forecast is way off and you keep gradually changing it so that eventually your forecast converges with reality as it happens.

    It is yet another thing if your forecasts evolves as in my last example, and you conclude by saying that things occurred as expected.

    This moves from bad forecasting -which is excusable - to dishonesty - which is less excusable, since even the humblest of humans is capable of honesty, while not even the greatest have proven capable of forecasting GDP growth 12 months in advance.

    Imagine if I predicted the Jays to win 100 games next year. Then, after a lousy April, I predicted 90 wins. Then, after a lousy May and June, I predicted 80 wins. Then after a bad July and August, I predicted 70 wins. Finally, at the end of the season, I predict that they will win 65 games, as they in fact have just done. Then I predict that although this year turned out as I expected, the Jays are not likely to do very well next year.

    Clear enough?

    By Blogger Declan, at 6:31 PM  

  • Along the same lines, forecasts should indeed change as new information appears, but if error is due to a poor model rather than random noise, then any forecast becomes suspect. How do you obtain reliable error bounds if the problem is poor model specification?

    While it's true that all models are wrong, models that yield consistently (and significantly) wrong forecasts certainly aren't very useful.

    By Blogger Josh, at 7:46 PM  

  • We need to distinguish between new data that should have been foreseen by the model and new data that is outside the scope of the model.

    If, for example, Canada's growth came in so far below forecast because of a sudden outbreak of disease or a war or we got hit by a meteorite, then you'd say, OK, we didn't expect the model to see that coming.

    But I can't think of any such events that have occurred between now and late 2006 that should have affected the forecast. Everything that has happened economically so far has been (to my mind) within the realm of marcoeconomics and therefore should have been foreseen by an effective macroeconomic model.

    By Blogger Declan, at 7:54 PM  

  • You have very, very high standards. Does anyone satisfy them?

    Economic forecasts - as in climatology, meteorology and indeed field where the data are non-experimental - are model-based. These models are by their very nature approximations to an unfathomably complex reality, so errors are unavoidable.

    By Blogger Stephen Gordon, at 5:42 PM  

  • But I can't think of any such events that have occurred between now and late 2006 that should have affected the forecast. Everything that has happened economically so far has been (to my mind) within the realm of marcoeconomics and therefore should have been foreseen by an effective macroeconomic model.

    Are you saying that in 2006, the Bank of Canada should have been able to predict when Lehman Brothers would collapse and it should have been able to predict how Treasury would have handled the file?

    By Blogger Stephen Gordon, at 5:55 PM  

  • Hey, if you want to concede that macroeconomic forecasts are useless because it is impossible to predict what's going to happen with the economy, then fine by me :)

    But maybe they should stop making them then?

    Obviously predicting that Lehman Brothers would collapse is much too specific for a macroeconomic forecast, but they should have seen that the collapse of the credit bubble would mean a bad year for the Canadian economy that would only get worse heading into 2009, in my opinion.

    I wrote a post on January 5 commenting that we faced, "oil up at $100 a barrel (in USD), the US Dollar dropping, the US Housing mortgage in freefall, the global financial industry (in particular the US) facing a severe credit crunch and capital shortages, the Central banks taking unprecedented measures to keep markets operating, a global economy filled with bubbles and economies overdue for a recession"

    The Bank of Canada made a forecast on January 24th that GDP growth would pick up in the second half of the year and that we would come through the year fine, and they raised their forecast for 2009.

    Even if they can't get the forecast right, they could at least admit when they get it wrong and explain what it was they missed rather than pretending they were right all along - that was the part that provoked my indignant post.

    Are my standards so high? - these are supposed to be the best economic minds in the country...

    By Blogger Declan, at 6:44 PM  

  • You have very, very high standards. Does anyone satisfy them?

    I'm quite sure that my opinion reflects those of many statisticians.

    Economic forecasts - as in climatology, meteorology and indeed field where the data are non-experimental - are model-based. These models are by their very nature approximations to an unfathomably complex reality, so errors are unavoidable.

    Yet a field like meteorology or climatology is based on testable physics and geophysics. Weather is not "unfathomably" complex - stochastic daily forecasts include such nice caveats as the probability of precipitation. The question is, if macroeconomic forecasts are unreliable at best, are the models themselves irredeemably flawed or are reliable forecasts of (political) economies simply impossible?

    By Blogger Josh, at 8:21 PM  

  • Before I went to bed, Environment Canada was announcing 10-15 cm of snow. This morning, the forecast said 15-25 cm. *shrug*

    In its policy reports and its interest rate announcements, the Bank does indeed say what it missed and why that might affect its policy stance.

    Forecasting is hard; forecasting the timing of the arrival of a recession is pretty much impossible. There simply isn't a good leading indicator for recessions. We could all see issues that would lead to a recession eventually. But no-one could say exactly when and how things would go pear-shaped. And there was always the possibility that untenable situations could unwind gradually on their own.

    By Blogger Stephen Gordon, at 6:28 AM  

Post a Comment

<< Home