Quibbling With Quiggin
John Quiggin writes something absolutely conventional and accepted by 99% of macroeconomists and I throw a cow.
Sometimes I bore myself but I must object to
“The big problems arose out of the Phillips curve, as I’ve discussed before. First it was oversold as a stable trade-off by Keynesians [citation needed].”
Which Keynesians when ? We agree that those Keynesians do not at all include Phillips or Samuelson and Solow in 1960. Such a claim should be supported by citations. I think it just will not do to say that everyone knows it is true. There are many things which are generally agreed to be true and demonstrably false. I refer you to James Forder at Oxford.
http://www.economics.ox.ac.uk/index.php/academic/james-forder
He claims to have looked and looked or Keynesians selling the Phillips curve as a stable tradeoff. I think he found a report to the Canadian commission on price stability or something.
I will just google.
Also. by the way, I ask what does the PIH fit which aren’t fit by a model of habit formation and myopia ? Is there any evidence that expectable changes in aggregate income affect aggregate consumption ? I am quite serious and ask for information.
Reposting my comment from CT.
@Quiggin vs Waldmann
Oddly enough, I can point to claims of a stable PC after 1968, but not before. It is almost as if Friedman’s strawman argument made his critics harden their position thus breathing life into the strawman — a sort of self-fulfilling prophecy.
1. James Tobin (1972):
Inflation and Unemployment
Cowles Foundation Paper 361
American Economic Review, 62, 1972
2. Gardner Ackley, who was a member and head of the CEA in the 1960s:
“However, the notion that there was a *stable* Phillips curve tradeoff did not capture the support of the economists most closely associated with formulating monetary and fiscal policies, at least in the United States. The writer believes that a careful study of the Annual Reports of the Council of Economic Advisers for the years 1961-1968 will discover no reference to any such stable trade-off.”
However, he continues …
“However, it did creep into the 1969 Annual Report — see Chart 8, page 95″
Macroeconomics: Theory and Policy
Gardner Ackley (1978)
Pg. 440, fn. 15
It would be interesting to explore the question whether Friedman 1968 was actually the genesis of the belief in a stable PC — by making his opponents harden their position.
Thanks. I should stress that I am a consumer of intellectual history. I don’t read many old economics articles (I will take the 5th on whether I read new economics articles). I think another possible explanation for the appearance of belief in a stable Phillips curve in 69 and 72 is that the data from the 60’s looked a whole lot like the idealized Phillips curves drawn for illustration. This shouldn’t have convinced anyone of anything as both a stable long term Phillips curve and an expectations augmented Phillips curve fit the data equally well. But that doesn’t mean it didn’t.
Or maybe, as you suggest, Friedman provoked Keynesians. I think that it is possible to find out by asking people what the reaction to Friedman 1968 was like.
“Is there any evidence that expectable changes in aggregate income affect aggregate consumption ?”
Well, there’s this paper by Eric French, Taylor Kelley, and An Qi:
http://www.chicagofed.org/digital_assets/publications/economic_perspectives/2013/1Q2013_part2_french_kelley_qi.pdf
From the conclusion:
“This article documents the decline in aggregate consumption during and after the Great Recession. It also explores the relationship between the decline in consumption and the decline in consumers’ expectations about their future income. The analysis uses microeconomic data from the Michigan Surveys of Consumers to study expected income growth. These data show that expected income growth declined significantly during the Great Recession for all age, income, and education groups. It is the worst drop ever observed in these data, and it has not yet fully recovered to pre-recession levels. Furthermore, we show that expected income growth is a strong predictor of actual future income and consumption growth. For this reason, forecasts of near-term consumption and income growth using these data suggest sluggish income and consumption growth over the next year.
Policymakers are still debating which actions, if any, should be taken to stimulate the economy. Although this article does not give any clear direction on the path that should be taken, the results discussed here suggest that actions undertaken to stimulate the economy are unlikely to lead to an overheated, high-inflation economy in the near future.”
Thanks but I said “expectable” that is “as would be expected by an agent with rational expectations”. Low expected income growth from t to t+1 is significantly associated with rapid consumption growth from t to t+1. Huh ? Well low expected growth is very strongly correlated with a pleasant surprise (and vice versa).
http://ideas.repec.org/a/taf/applec/v45y2013i28p4004-4021.html
Now note that a can be correlated with b and b with c but a not correlated with c. Notably, professional forecasters predict an eventual end to the current output gap — that is faster than normal income growth. Low expected income growth after a sharp GDP decline might explain the depth of recessions. But it has much less than nothing to do with the PIH, since such sharp decline has since WWII (till now perhaps) been followed by faster than average growth.
We have had (since WWII and I don’t say we have “had” anything in the future) a pattern which is further from the PIH than a static IS_LM type consumption function. subjective expected income growth and sample average conditional average income growth moving in opposite directions.
Oh @Mark thanks again. I will read the paper to which you linked. It looks extremely interesting. So thanks.
Thanks a third time. I have now read well the tables in the paper anyway They do not at all answer the question I asked.
This reply is long. The point is that the regressions reported by French Kelley and Qi have nothing to do with the question I asked in my post.
Decades ago, the results would have been considered striking evidence *against* the PIH. Now they correspond to a well known stylized fact which (like all conceivable facts) can be reconciled with rational utility maximization.
Now I asked the question in a little off topic aside (I think I will do a whole post). But my question can be illustrated by an extremely simple example
In a regression of consumption on disposable income and some measure of expectable future income growth, is the coefficient on expectable future income growth positive and statistically significant ?
All of the regressions (note I didn’t type” figures & regressions”) with consumption on the left had side were regressions of the future rate of growth of consumption on something and expected real income growth. For the simplest illustrations of the PIH with utility additively separable in consumption and everything else and time separable in consumption, a significant coefficient on expected future income growht (and the one on lagged consumption growth) are rejections of the null permanent income and separability joint hypothesis.
They don’t address my question (I consider the PIH rejected so the glass isn’t full and ask if it is half full or entirely empty).
I note that the regressions wtih achieved growth as dependent variable reject the joint null of rational expecations and a quadratic loss function (that is they reject the null that the subjective expectation from the Michigan survey is equal to the objective conditional mean).
The regressions in which subjective expected income helps forecast income include only lagged income growth (the sign of the coefficient is as I predicted above — people over extrapolate recent changes). It is possible to make best forecasts of real income growth using long time series and regress the residuals on expected income growth. This is a valid test of the null that the survey respondents have nothing to add (the coefficient would not be 1 under rational expectations and quadratic loss functions, but it is possible to test the null that it is 0). Also residuals of expectations on a few key variables can be used (should increase power against the null).