Friday, October 28, 2011

My Hero!

Keen-eyed followers of this blog may have noticed that if you view my complete profile you'll learn that one of my interests is "staying at home". Actually it's a (male side of the) family tradition - though my son, Matt, seems to be bucking the trend!

As accurate as this profile is, I do have a small confession to make. I stole the line about staying at home from Professor Sir Richard Stone - specifically, from his listing in Who's Who when he was still alive.
Now, I'd hate you to leap to the conclusion that my time spent staying at home is devoted to scanning the pages of Who's Who on the off-chance of uncovering something a tad off-beat. Apart from anything else, the length of our family's "Honey Do..." list would preclude this. No - there's actually an interesting story behind all of this, as well as an opportunity to pay tribute to one of the founding fathers of econometrics.

Knighted in 1978, John Richard Nicholas Stone (1913-1991) was awarded the Nobel prize in Economic Sciences in 1984,
"for having made fundamental contributions to the development of systems of national accounts and hence greatly improved the basis for empirical economic analysis".
More specifically, Richard Stone worked with James Meade and John Maynard Keynes to produce the first National Accounts for the U.K.  The 1941 Budget was accompanied by three tables based on this work, with something bordering on a "disclaimer" from the Chancellor of the Exchequer when he presented the Budget:
"The Chancellor in his budget speech emphasised that the publication of official estimates of national income and expenditure should not be regarded as setting a precedent" [!!!!!; DG]
This last quote is from Stone's autobiography, which you can read here.

What was especially noteworthy about the national accounts developed by Stone and his colleagues in the U.K. is that they were of the "double-entry" type that we now take totally for granted. In this respect they differed from the early efforts in the U.S. and Canada around the same time. It's difficult now to conceive of macroeconometric modelling without reliable national accounts data, but such data are really a relatively recent "invention".

Indeed, you may be surprised to learn that the availablility of quarterly national accounts came very late in the piece in many countries - New Zealand, for example. The early macroeconometric model-building that was conducted by the Reserve Bank in that country in the late 1960s to mid 1970s was hampered by the absence of quarterly accounts. This was a particular problem, as a quarterly forecasting (and simulation) model was the first priority! Needless to say, a great deal of time was spent on some extremely creative, thorough, and painstaking data construction.

Having been part of that early modelling exercise, I've always felt entitled to describe N.Z. as the "only under-developed country where it's safe to drink the water".

Richard Stone gave us far more than proper national accounts, though. His work on the estimation of systems of demand equations was pioneering, culminating with two important volumes co-authored with Deryck Rowe and others in 1954 and 1966. Stone and Rowe (1957) undertook some of the most important early econometric analysis of the demand for durable goods; and Stone gave us the familiar "Linear Expenditure System" (Stone, 1954), based on the maximization of the Klein-Rubin (direct) utlity function. The LES was the primary workhorse for empirical demand analysis until the introduction of the "Almost Ideal Demand System" by Deaton and Muellbauer (1980).

However, what endears Stone to me most (apart from his impeccable tase in hobbies) is the role that he played as Director of the Department of Applied Economics (set up at Keynes' instigation) at the University of Cambridge, from 1945 to 1955.

In the  mid 1980's, Max King and I (King and Giles, 1987) assembled a festchrift to honour Donald Cochrane. Don was our Dean of Economics and Politics at Monash University. The volume was originally intended to be in honour of Don's retirement, but sadly it became a memorial as he died prematurely in 1983. Don was Australian, but spent time in England as a fighter pilot during World War II, and completed his Ph.D. under Ston'es supervision at Cambridge.

Ther are some wonderful inter-connections involving several "household names" in econometrics, and at the centre of them all is Stone and the DAE. Max King describes this admirably in the following terms:

"Stone was soon able to attract to his new department a number of talented young researchers whose work was to have a great impact on the practice of econometrics.

There was the work of Cochrane and Orcutt (1949) which succeeded in bringing the serial correlation problem to the attention of econometricians. Donald Cochrane was a PhD student in the Department, while Guy Orcutt joined the staff in 1946 as  a senior research worker. At about this time Klein (1947) published an econometric model of the U.S. economy and used the von Neumann ratioto  test the independence of its disturbances. His testing revealed evidence of lack of independence in very few equations. Cochrane and Orcutt argued that it is likely that the disturbances of econometric models are positively correlated. They gave a number of theoretical reasons to syupport their case, and examined the results of the application of the von Neumann ratio to the residuals from four econometric models including two of Klein's models. They also performed a small simulation experiment which cast doubt on the reliability of the von Neumann ratio as a test  for correlated disturbances in econometric models. For the particular models they considered,  their experiment showed the von Neumann ratio to be biased towards accepting randomness and that this bias increases with the number of regrerssors.

Cochrane and Orcutt also discussed the cost of ignoring serial correlation, thus highlighting the need for a reliable test for correlated disturbances. Their 'tentative' method of dealing with AR(1) disturbances, now known as the Cochrane-Orcutt transformation, became widely used. In 1950, two of their Cambridge colleagues published a solution to the testing problem that Cochrane and Orcutt posed.

In the late 1940's James Durbin was a postgraduate student studying mathematical statistics at Cambridge. Because he had expressed an interest in economic applications, he was attached to the Department of Applied Economics. At about this time, T. W. Anderson was  a short term visitor to the Department and Geoffrey Watson was a PhD student at North Carolina under R. L. Anderson's guidance. R. L. Anderson had suggested to Watson that he consider deriving the null distribution of the von Neumann ratio applied to OLS regression. Watson joined the Department of Applied Economics in 1949 and collaborated with Durbin on this project while the latter was employed as a research worker in the Department. Their collaboration continued for a time after Durbin took up an appointment at the London School of Economics."
[ King, 1987, pp. 22-23.]

There a few other comments that I should add.

There was an even earlier connection between Geoff Watson and Don Cochrane. Geoff was also an Australian, and they both completed their undergraduate degrees at the University of Melbourne.

Klein was also in England towards the end of the period that all of this was going on, but just "up the road" ("down the road"?) at Oxford. He spent time there between 1954 and 1958 as a consequence of McCarthyism in the U.S..

In the festschrift  that Max and I put together, we were fortunate in being able to include original contributions from Richard Stone, Guy Orcutt, and Geoff Watson. If we'd been able to get a paper from Jim Durbin we'd have had a perfect batting average!

It was while researching the background for that book that I discovered Richard Stone's professed passion  for "staying at home". Imagine what he could have achieved from there with today's technology!


Note: The links to the following references will be helpful only if your computer's IP address gives you access to the electronic versions of the publications in question. That's why a written References section is provided.

References

Deaton, A. and J. Muellbauer (1980). An almost ideal demand system. American Economic Review, 70, 312-326.

King, M. L. (1987). Testing for autocorrelation in linear regression models: A survey. In M. L. King and D. E. A. Giles (eds.), Specification Analysis in the Linear Model (In Honour of Donald Cochrane). Routledge & Kegan Paul, London, 19-73.

King, M. L. and D. E. A. Giles, eds. (1987). Specification Analysis in the Linear Model (In Honour of Donald Cochrane). Routledge & Kegan Paul, London.

Stone, R.  (1954). Linear expenditure systems and demand analysis: An application to the pattern of British demand". Economic Journal, 64, 511-527.

Stone, R. and D. A. Rowe (1957). The market demand for durable goods. Econometrica, 25, 423-443.


Stone, R. and D. A. Rowe (1966). The Measurement of Consumers’ Expenditure and Behaviour in the United Kingdom, 1920–38, Vol. 2, Cambridge University Press, Cambridge. 

Stone, R., D. A. Rowe, W. J. Corlett, R. Hurstfield and M. Potter (1954). The Measurement of Consumers’ Expenditure and Behaviour in the United Kingdom, 1920–38, Vol. 1, Cambridge University Press, Cambridge. 


© 2011, David E. Giles

No comments:

Post a Comment