I think you're reading it incorrectly. In CT that's what the last poll was. Because of how he presents the data (chance of Republican victory) the CT % isn't there, but based on other races with the same profile, my guess is that it'd come in at 0%, actually.

Intrade is something rathr different than a statistical model based on polls - it's the market judgement of a bunch of betters/investors who take polls as only on input, then run it through their judgement.

I'd add that I don't necessarily buy this model, just thought it was interesting.

read it incorrectly. There are indeed two types of %. The chance of winning and poll results. The winning probability is calculated from the overlapping uncertainty in the results of the polls. He uses the filter to adjust the poll results.

Intrade is probably the skill weighted average of political consults having their models and access to the secret polls too.

I recommend a regression model to guess the winner in election races.

-- 101 ways to avoid the subjunctive mood

My latest knowledge is:
Average budget for a senate race is 7 000 000$, for a house race 1 000 000$.

In 98% of the races the incumbent wins, because they can change election district borders. Is the figure correct?

-- 101 ways to avoid the subjunctive mood

I imagine he'll get over his embarassment by pointing out that he was one of the only economists on record to predict the GDP of the third quarter almost spot on. Roubini's record over the last year has been better than the vast majority of economists.

As for the stock market - eh, we'll see. I haven't noticed that anyone has a very good record of doing that and I have no faith in concensus opinions about the market.

I'll add that I have a lot more faith in the yield curve, which is unambiguously inverted, than I do in the stock market, as predictive mechanisms. The stock market is often wrong. The yeild curve is rarely wrong.

I honestly don't know whether 2007 will technically be a recession year. I suspect it may avoid a technical recession, but feel like hell. For most ordinary Americans, however, it'll be a recession, no matter what NBER decides, and after all the screwing around that has gone with inflation numbers, I wouldn't be surprised if it really is a recession, whatever the official GDP numbers come in at.

I imagine he'll get over his embarrassment by pointing out that he was one of the only economists on record to predict the GDP of the third quarter almost spot on.

If you had read The Agonist, you would have noticed that he was correct because of a statistical glitch.

Somehow free and publicly available GDP forecasts tend to be noise, but somehow the stock market can predict recessions. I see some discrepancy here.

-- 101 ways to avoid the subjunctive mood

What? I didn't say the stockmarket predicts recessions. I said that yield curve predicts recessions pretty damn well. The stock market has had a lot more false negatives than the yield curve.

As for Roubini's predictions - it doesn't matter how you slice it, he was closer to correct than the majority of economists, because his numbers were lower. No matter which glitch your sarcastic gibe refers to, the effect of removing said glitch is to lower the GDP, not make it higher, which would have made the consensus forecast closer to correct, rather than his.

Roubini isn't perfect (who is other than you?) However, he isn't a flake or a quack, and he isn't paid to make happy talk the way most institutional economists are.

The correct link, by the way, is here:

http://www.rgemonitor.com/blog/roubini/144686/

To clear the air from the spin that one is increasingly hearing it is useful to ask a simple factual question: what is the relation between stock markets and recessions? So, for a moment, let us leave aside the issue of whether my recession call is correct or not. And let us assume, for the sake of the pure logical argument, that a recession is coming and then ask the question: if we will have a recession, what will happen to the stock market? So, you do not have to believe in a recessionary hard landing to consider this specific question. You just need to ask yourself the simple question of what happens to stock prices when recessions do come. (Thus, for now I will aside the question of what happens to the stock market when you get a “soft landing” or “hard landing” short of a recession. I will consider this question in a future discussion)

Luckily we have enough data series on previous recession and stock prices to give an answer to this question. Consider the charts that are shown below. They present the percentage change in that S&P500 index around the last six U.S. recessions (i.e. starting with 1970), i.e. in the months before the start of a recession, in the months during a recession and in the months after it. The vertical lines in each charts represents the peak of the business cycle (i.e. the beginning of a recession) and its trough (end of a recession). On average the stock market does not change much between the peak and the trough of the business cycle: on average the fall is only 0.4% between peak and trough; in some recessions – such as the 1974-1975 one - the peak-to-trough fall is much deeper (-13%) but in others – such as the 1980 one – stock prices actually rose 5.8% between peak and trough; so -0.4% is an average for all recessions.

This may seem like a relatively small adjustment but the peak-to-trough comparison is deceptive. It is deceptive because, usually, the stock market starts to fall before a recession starts (i.e. before the business cycle peak), then it falls very sharply during the first stage of a recession, and then in starts to recover in the late stages of a recession before the recession has reached its bottom (i.e. before the trough of the recession). Specifically, the stock market falls from the peak in the business cycle to its lowest level during a recession averages 17.5%; and in every one of these six recessions you have the same pattern: initially stock prices sharply fall as the economy enters a recession. Then, the recovery of the stock market starts before the trough of the business cycle has occurred, i.e. before the economy has gotten out of a recession.

Notice also that, in most episodes, the stock market peaks a few months before the actual start of the recession and starts falling even before the formal start of the recession (i.e. before the peak of the business cycle). Since stock prices almost always start to fall a few months before the recession has formally started – as signals of an impending slowdown and possible recession are already mounting even before a recession is formally triggered and thus priced in the stock market – the cumulative fall in stock prices from their pre-recession peak to their bottom level in the actual recession is well above the 17.5% figure for the stock price fall from the start of a recession to the lowest level of such stock prices during a recession. This average fall in stock prices from pre-recession peak to into-recession bottom is actually close to 28%, an extremely severe and sharp bearish downfall.

In other terms, the peak-to-trough average flat behavior of the stock market hides a much sharper fall in the stock market before a recession and during the first half or so of a recession, followed by a relatively sharp recovery in the late stages of a recession. This pattern makes total sense as equity prices are forward looking and, at any point in time, they reflect all available information about the expected path of current and future dividends/earning and interest rates. The stock market starts to fall before a recession has formally started because the closer you get to the peak of the business cycle when the macro news on growth become increasingly weaker, the higher is the probability that a recession will occur and will thus drag down profits. So, a forward looking equity market peaks before the peak of the business cycle and starts falling before the actual recession has started. That is why stock prices tend to be a good – if imperfect - leading indicator of the business cycle.

The fall in the stock market from the peak of the business cycle to the market lowest level in the recession was 21.0% in the 1970 recession, 33.88% in the 1974-75 recession, 10.6% in the 1980 recession, 18.2% in the 1981-82 recession, 14.6% in the 1990 recession, 10.3% in the 2001 recession. In most recession, as discussed above, the stock market peaks before the recession and starts to fall even before the recession has formally started. In the 1970 episode the stock market peaked 9 months before the recession and fell 12% even before the recession started. In the 1974-75 episode, the stock market peaked 12 months before the start of the recession and fell 23% even before the recession formally started in December 1973 with a good half of this pre-recession drop right after the beginning of the Yom Kippur war that led to Arab oil embargo. An exception is the 1980 episode when the stock market was actually rising in the few months before the start of the recession in February 1980. In the 1981-82 case, the stock market peaked four months before the onset of the recession and then fell already about 4% before the recession actually started. In the 1990 case, the stock market peaked two month before the recession and fell about 2% before the formal start of the recession. In the 2001 episode, the S&P peaked about seven months before the start of the recession in March 2001 and then fall by 31% even before the recession started (the peak of the Nasdaq was, of course even earlier, in March 2000 a full year before the formal onset of the recession).

Of course, in the economic history of the US in the last few decades sometimes stock prices have fallen and a recession has not materialized, i.e .stock markets are not a perfect and uniquely correct leading indicator of a recession. But, and this is more important in the context of the question asked above, any time a recession did occur, the stock market actually sharply fell. So, the issue here is not whether the stock market may at times provide false alarms or incorrect signals of the business cycle; of course, as it is well known, it does at times provide false signals. The issue is whether hard landing and beginning of recessions are associated with sharply falling stock prices. And the simple and unequivocal answer is that recession lead to bearish stock markets where the peak in the economy to the trough in the stock market (as separate from the economic peak-to-trough that lags the one of asset prices) is about 17.5% and where the peak-to-trough in the stock market (i.e. the pre-recession peak to the into-recession bottom of the stock market) is about 28%, i.e a very clear, sharp and deep bear market. So, factually hard landings and recessions do lead to falling stock prices and bear stock markets. So, the recent market buzz and chatter about hard landings and recessions being good for the stock market is utter nonsense based on actual data from decades of US business cycles and repeated recession episodes.

Of course, once a recession has triggered a severe bear market, at some point – before the bottom of the recession – the stock market does start to recover. The fact that the stock market recovers before the trough of the business cycle is reach is also logical and based on the forward looking nature of stock prices: even before a recession has ended the rate of economic activity fall tends to increase: in early stage of a recession the first derivative of output is negative (negative growth) while the second derivative shows an acceleration of the rate of economic contraction. In later stages of a recession, the first derivative is still negative but the second derivative shows a slower rate at which the economy is contracting and signals that the trough of the business cycle may be close, i.e. there is incoming light at the end of the recession tunnel. Thus, for forward looking stock prices it is not necessary to wait until the recession is over for such prices to recover: once the evidence is building up that the worst stage of a recession is close to be over and that the trough – bottom of the downturn – will be reached soon (i.e. the probability that the recession will be over soon is increasing) then the stock markets starts to recover: i.e. stock prices reach their trough before the trough of the business cycle.