Richard Stone's Influence on Cambridge Economics

Stone’s specific contribution to Cambridge economics came from his time as the first Director of the DAE and his subsequent time as Director of the CGP in the DAE. His influence can be seen in three areas that flourished throughout the life of the DAE: the historical research into UK national income statistics, the development of econometrics as a discipline, and the modelling work of the CGP

The historical research^[1] started by Stone in the National Institute and continued in the collection of statistics on consumers’ expenditures in the DAE [56] was later supported by his invitation to Phyllis Deane to join the DAE in 1950. She worked on UK regional statistics, and went on to publish two landmark texts in 1962, on British Economic Growth, 1688-1959 with W.A. Cole and the Abstract of British Historical Statistics with B.R. Mitchell. The work on historical national accounts was continued by Charles Feinstein who joined the DAE in 1958 and who was largely responsible for National Income, Expenditure and Output of the United Kingdom, 1855-1965, published in 1972, which became the standard reference work on the period.

Stone’s influence on econometrics was profound. The DAE under his Directorship soon became a leading international centre for econometrics in the 1940s and 1950s. The Durbin-Watson test for serial correlation in the residuals of regression equations was devised when he was Director of the DAE, as was the Cochrane-Orcutt method for reducing the risk of spurious correlations between variables. The development of econometrics in Cambridge stemmed from this time (see Smith 1998: 88-103). Members of the CGP went on to become internationally acclaimed in applied econometrics, most notably Angus Deaton for demand and inequality analysis and Hashem Pesaran for panel data analysis.

In model building, with the formation of the CGP, Stone and Brown began a distinctive approach that continues to the present day, with five characteristics: (1) structural detail allowing for institutional aspects of behaviour to be represented, (2) consistent aggregation in basing the model on a system of accounts in the form of the SAM, (3) a combination of optimisation and simulation by integrating limited optimisation in production, for example, minimisation of costs, with time-series estimation of the model’s parameters, (4) computable solutions, and (5) both balanced and unbalanced solutions of the models. Although disaggregation is costly in terms of data collection, data processing, computing, and model builders’ time in checking solutions and results, it has the great advantage of representing institutional structures, such as the particular features of agriculture or the electricity industry. The advantage of a computable model is that the results are reproducible and the effects of alternative assumptions can be readily calculated consistently and in great detail.

The original CGP two-model structure of a transient and static model was eventually replaced by a single dynamic model, the Multisectoral Dynamic Model (MDM), developed in the DAE in the late 1970s. The MDM did not rely on the concept of equilibrium (see Barker 1977); rather it relied on simulating the economy based on sets of equations estimated using time series data. After the CGP closed in 1987, one version of the model became, after many theoretical and econometric improvements (including an extension to 20 world regions), the Energy—Environment—Economy Model at the Global level (E3MG). This was developed and used by the Cambridge Centre for Climate Change Mitigation Research (4CMR)^[2] group in Land Economy at Cambridge to generate scenarios for mitigating climate change (see Barker and Crawford-Brown 2015). E3MG is a Post Keynesian, structural model with endogenous technological change.

The work of the CGP also developed into the two Post Keynesian models (MDM and E3ME) at the UK and global scales maintained by Cambridge Econometrics and continuing to the present day. Stone agreed with this evolution: ‘I think growth models should be at least capable of being stable but should not be based on the assumption that equilibrium conditions are always fulfilled’.^[3] In their use of formal econometric techniques to estimate the model, and assumptions of variable degrees of competition and economies of scale across industries, the Cambridge models offer an alternative to the prevalent Computable General Equilibrium methodology based on neoclassical economics.

• [1] See Pesaran (1991: 99-101) for Stone’s own account of the work.
• [2] The 4CMR was formed after the DAE was merged into the Faculty of Economics in 2005 and thoseworking on energy—environment-economy modelling moved to the Department of Land Economy.
• [3] Stone to Tinbergen, 12 February 1981, Richard Stone Papers, King’s College Archive Centre,Cambridge University: Box JRNS 4/10.