The Nobel Prize

Robert F. Engle (b. 1942)

Robert Engle was born in Syracuse, New York, USA in 1942. He grew up near Philadelphia, Pennsylvania and attended Williams College; he graduated from Williams with a BS in 1964, achieving highest honours in physics. From a young age he had a strong interest in science – his father had a PhD in chemistry – and Engle elected to continue with physics in graduate school at Cornell University. He was awarded an MS by Cornell in 1966. However, even before starting graduate school, he felt his attachment to physics was weakening. At Williams, Engle had filled up his senior- year programme with an elective in economics that he had enjoyed and, before completing his MS, he approached the Economics Department at Cornell to ask about the possibility of changing disciplines. The department offered him a PhD fellowship and he accepted it, taking undergraduate classes in economics while he completed his master’s in physics. Cornell awarded Engle a PhD in economics in 1969. In an interview in Econometric Theory, Engle reflected on the advantages a background in physics can lend to an econometrician given the emphasis on the integration of theory and data common to both subjects; however, interestingly, he has also acknowledged the difficulties of his late discipline switch: ‘It was probably ten years before I really absorbed the economic way of thinking’ (see Diebold, 2003; Nobel Foundation, 2004).

Engle’s first academic post was in 1969 as assistant, later associate, professor at the Massachusetts Institute of Technology (MIT). At Cornell he had developed an interest in time-series econometrics but this was not much of a focus for the economists at MIT and Engle found his place there ‘complicated’ (Nobel Foundation, 2004). Involvement with the Boston Redevelopment Authority allowed him to develop his skills in economic modelling and he became an urban economist. With the encouragement of his fellow Laureate Clive Granger, Engle left MIT in 1975 for the University of California at San Diego (UCSD), where he was appointed associate professor, and where he continued to teach urban economics. At the time Granger was professor at UCSD. Engle was promoted to Professor of Economics at UCSD in 1977. In 1999 he moved as professor to the Stern School of Business at New York University, and since 2003 he has been Michael Armellino Professor in the Management of Financial Services in the Stern School.

Engle’s honours and awards include fellowships of the Econometric Society (1981), the American Academy of Arts and Sciences (1995), and the American Statistical Association (2000). He has been a research associate of the National Bureau of Economic Research since 1987 and received an Excellence in Teaching Award from the MIT Graduate Economics Association in 1974–75. Jointly with Clive Granger, Engle was awarded the 2003 Nobel Memorial Prize in Economics ‘for methods of analysing economic time series with time-varying volatility (ARCH)’ (Nobel Foundation, 2004).

In honouring Engle with a Nobel Prize, the Royal Swedish Academy of Sciences highlighted the contribution of his 1982 Econometrica paper, ‘Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation’, to econometric method. In this paper, Engle announced his discovery of a model – known by the acronym ARCH – that facilitated the measurement of uncertainty in an economic process, where uncertainty was itself subject to change. In developing ARCH, Engle’s purpose had been to try to test Milton Friedman’s proposition that business cycles were rooted in entrepreneurial uncertainties over future rates of inflation: trepidation about what might happen to prices could make businesses cautious and reluctant to invest, thus prompting an economic downturn. As Engle (2004, p. 406) notes, for Friedman to be right, uncer- ROBERT ENGLE 324 tainty ‘had to be changing over time’. It turned out that, though there was variation in the uncertainty in inflation forecasts for both the UK and, as Engle subsequently found, for the United States, it did not appear to have business-cycle implications in either context. Engle suggests that this may reflect the multiple determination problem common in macroeconomics which renders uncertainty one among several factors of influence. He also points to the differences associated with relatively low-frequency data in macroeconomic time series. However, if the deployment of ARCH in macroeconomics was not particularly fruitful, Engle found the field of finance much more receptive: here, uncertainty over the returns on assets is a clear imperative in decision making, and high-frequency data are readily to hand to underpin the modelling process. Engle (1987a), in collaboration with David Lilien and Russel Robins, is his first application of ARCH to finance.

The balance between risk and reward is central to financial economics. Engle’s ARCH model provides a means to evaluate the volatility of the returns on assets – such as shares – and hence their associated risks. A common approach to risk assessment has been to assess ‘historical volatility’ by using the standard deviation of returns over a given time period (Engle, 2004). But then the question becomes over what period should volatility be measured? The difficulty is that returns are subject to ‘volatility clustering’ (Engle, 2001). In other words, to reprise an earlier point, uncertainty – here over returns – is itself subject to change. For example, for most share indices, large movements in returns are typically followed by further large movements, and small fluctuations tend to be similarly tracked by equally modest changes. The ARCH model resolves the problem of choosing the ‘right’ period over which to measure volatility by taking weighted averages of past squared forecast errors generated by a statistical analysis (Engle, 2004). Such weights assign more importance to recent information – acknowledging clustering – and less to information that is dated. The whole analysis turns on the treatment of heteroscedasticity – meaning that uncertainty is time-varying – not as a statistical problem of least squares but as something that can be tested for and fruitfully modelled (Engle, 2001). Once this has been done it becomes possible to measure and forecast volatility as an aid to risk analysis and as a means of tackling other financial problems such as options pricing, an area to which Engle has also contributed (see Engle et al., 1994a). has been generalised, and its acronym consequently extended to GARCH, by Tim Bollerslev, a student of Engle’s. Although not carrying any new insights, GARCH is useful, and popular, given that it permits relatively parsimonious specification (Nobel Foundation, 2004).

ARCH has spawned a voluminous literature. Notable works by Engle here include explanations of volatility rooted in the impact of turbulence in early-closing markets on markets that close later (see Engle et al., 1990). More recently Engle, with Jeffrey Russell, has proposed the autoregressive conditional duration (ACD) model, which uses the clustering of market trades to forecast the arrival probability of the next trade (Engle and Russell, 1998). Other work by Engle acknowledged by the Royal Swedish Academy includes pivotal research on cointegration with his fellow Nobel Laureate, Clive Granger. Engle and Granger (1987b) ‘is perhaps the most cited paper in the history of econometrics’ (Diebold, 2004, p. 166). Other important papers by Engle include those on exogeneity, for example, Engle et al., (1983). Engle has also jointly edited books on cointegration (Engle and Granger, 1991), and econometrics (Engle and McFadden, 1994b). Finally, Engle (1995) is a collection of papers on ARCH.

Main Published Works
(1982), ‘Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of United Kingdom Inflation’, Econometrica, 50, July, pp. 987–1007.
(1983), ‘Exogeneity’ (with D.F. Hendry and J.-F. Richard), Econometrica, 51, March, pp. 277– 304.
(1987a), ‘Estimating Time-Varying Risk Premia in the Term Structure: the ARCH-M Model’ (with D.M. Lilien and R.P. Robins), Econometrica, 55, March, pp. 391–407.
(1987b), ‘Co-integration and Error-Correction: Representation, Estimation and Testing’ (with C.W.J. Granger), Econometrica, 55, March, pp. 251–76.
(1990) ‘Meteor Showers or Heat Waves – Heteroscedastic Intradaily Volatility in the Foreign- Exchange Market’ (with T. Ito and W.L. Lin), Econometrica, 58, May, pp. 525–42.
(1991), Long-Run Economic Relationships: Readings in Cointegration (ed. with C.W.J. Granger), Oxford: Oxford University Press.
(1994a), ‘Forecasting Volatility and Option Prices of the S&P 500 Index’ (with J. Noh and A. Kane), Journal of Derivatives, 2, Fall, pp. 17–31.
(1994b), Handbook of Econometrics, vol. 4 (ed. with D. McFadden), Amsterdam: North-Holland. (1995), ARCH: Selected Readings (ed.), Oxford: Oxford University Press.
(1998), ‘Autoregressive Conditional Duration: A New Model for Irregularly Spaced Transaction Data’ (with J.R. Russell), Econometrica, 66, September, pp. 1127–62.
(2001), ‘GARCH 101: The Use of ARCH/GARCH Models in Applied Econometrics’, Journal of Economic Perspectives, 15, Fall, pp. 157–68.
(2004), ‘Risk and Volatility: Econometric Models and Financial Practice’, American Economic Review, 94, June, pp. 405–20.

Secondary Literature
Diebold, F.X. (2003), ‘The ET Interview: Professor Robert F. Engle’, Econometric Theory, 19, December, pp. 1159–93.
Diebold, F.X. (2004), ‘The Nobel Memorial Prize for Robert F. Engle’, Scandinavian Journal of Economics, 106 (2), pp. 165–85.
Royal Swedish Academy of Sciences (2004), ‘The Nobel Memorial Prize in Economics 2003’, Scandinavian Journal of Economics, 106 (2), pp. 163–4.
Post a Comment