⁽¹⁾ This forecast takes no account of any potential impact of the ‘industrial action’ being taken in support of the current Civil Service pay dispute. Between 9 March and 29 April some 600,000 non-industrial man-days had been lost as a result of the dispute (Hansard, 29 April). Payments of substantial portions of the government's revenue have been held up and extra short-term borrowing made necessary. Clearly there will be some impact from this temporary reallocation of funds, but its nature cannot be assessed yet.

⁽²⁾ We have returned to the practice of quoting the average estimate of GDP as the most appropriate single measure. When the CSO changed to quoting the output measure in the January 1980 issue of Economic Trends we followed its example. Now that the average estimate has been reinstated for six months we feel it appropriate to readopt it. The three measures of GDP are shown for the past in statistical appendix table 1 and the output measure for the forecast period in a footnote to table 2, so the effect of the change is clear. This and other difficulties relating to official data sources are discussed in more detail on p. 10

⁽³⁾ The measures announced in the Budget and the associated public expenditure White Paper, Cmnd 8175 are summarised in the Calendar of Economic Events on pp. 62-3.

⁽¹⁾ Moreover the method of separating trend and fluctuation may well differ.

⁽¹⁾ Quoted in D. J. O'Dea, Cyclical Indicators in the Postwar Economy, NIESR Occasional Paper XXVIII, CUP 1975, p. 116.

⁽¹⁾ This is not a behavioural argument and should not be taken to mean that if there had been zero stockbuilding throughout the period then this is the likely consequent path of GDP; stockbuilding is simply subtracted.

⁽²⁾ The statistical problem is made somewhat worse on this occasion because of the absence of some of the data normally published by now (the trade accounts for March, for example).

⁽¹⁾ Our forecast assumed the full increase in the tax on derv would be enacted. The effect of the reduced rise is likely to be negligible at our degree of aggregation.

⁽¹⁾ See for example, R. W. R. Price, ‘Public expenditure: policy and control’, National Institute Economic Review, No. 90, November 1979.

⁽¹⁾ This simple analysis therefore excludes from consideration those over ‘working age’ who do in fact participate in the labour force.

⁽¹⁾ Excluding iron and steel.

⁽¹⁾ Investment in private dwellings is forecast to be almost exactly 10 per cent of total fixed investment, so a 25 per cent rise in the former in 1982 only contributes a boost of 2 1/2 per cent to the total.

⁽¹⁾ Owing to a change in the published trade statistics relating to erratic items it has been necessary to re-estimate part of the export sector of our model. The properties of the new equations are similar to those described in National Institute Discussion Paper No. 28.

⁽¹⁾ This simulation differs from that in the memorandum to the Treasury and Civil Service Committee by S. G. B. Henry and K. Cuthbertson which was based on the February forecast.

⁽²⁾ If all the signs in table 20 were reversed it would show the effects of imposing the actual Budget changes on a world of ‘unchanged policies’.

⁽¹⁾ We are grateful to colleagues at HM Treasury, the London Business School and Liverpool University for helping to check the accuracy of this comparison.

⁽²⁾ We initially wished to include an assessment of the Cambridge Economic Policy Group, CEPG, figures for 1980 as set out in the base projections in Cambridge Economic Policy Review, vol. 6, no. 1, April 1980, but have excluded them here as they form part of a medium-term projection and are not a forecast in the same sense as the others.

⁽³⁾ Official data, particularly for 1980, may be revised in the future and this might alter the comparison of the various teams as discussed below.

⁽⁴⁾ Further details may be found in S. Brooks and K. Cuthbertson, ‘Economic Models and Economic Forecasts’, National Institute Discussion Paper No. 41 April 1981, together with a condensed version published in the National Institute Economic Review, No. 95, February 1981.

⁽¹⁾ It is possible to use an ARIMA time-series model to forecast the statistical adjustment but this is not done by any of the forecasting teams discussed here.

⁽²⁾ It is not clear what assumption was adopted by Liverpool.

⁽³⁾ However even small differences in assumptions about the variable have a substantial effect on the forecast of GDP where it is as large as it was in this case.

⁽¹⁾ See F. Teal and D. R. Osborn ‘An assessment and comparison of two NIESR econometric model forecasts’, National Institute Economic Review, no. 88, May 1979 for estimates of NIESR forecast errors caused by errors in lagged dependent variables.

⁽²⁾ 1979 IV was partly estimated by NIESR. The upward revision to expenditure GDP for 1979 by the CSO between the April 1980 and April 1981 issues of Economic Trends is 1.4 per cent.

⁽³⁾ This arises because many equations have lagged dependent variables. Also the static single equation residuals for the past would have been different and this might have entailed a different ‘forecast’ path for the residuals from that chosen by the forecasters at the time; this may also substantially alter the profile of the forecast (see Brooks and Cuthbertson op. cit.).

⁽⁴⁾ Economic Outlook, vol. 4, no. 5, February 1980, p. 5. Both teams based their public expenditure forecasts on the 1980 White Paper (Cmnd 7841).

⁽⁵⁾ £575 million after allowance for an increase of £325 million in the contingency reserve, close to the NIESR assumption of £500 million additional cuts.

⁽⁶⁾ Economic Outlook, Liverpool Occasional Papers, No. 2., p. 7.

⁽⁷⁾ A 7-11 per cent target range for £M3 had been announced prior to the budget of March 1980 (see Economic Progress Report, December 1979, HM Treasury) and this was adopted by NIESR and LBS for the financial year 1980/81. NIESR envisaged only a moderate fall in the Treasury Bill rate to 15 per cent, while LBS forecast a fall in MLR from 17 per cent in 1980 I to 11 per cent in 1980 IV.

⁽¹⁾ Economic Progress Report, December 1980, HM Treasury.

⁽²⁾ Economic Progress Report, No. 116, December 1979, HM Treasury.

⁽³⁾ Non-durable consumption only: actual change 1.2 per cent.

⁽⁴⁾ The exception here may be the Liverpool model where relative prices do not influence export volumes (see P. Minford, ‘A rational expectations model of the United Kingdom under fixed and floating exchange rates’ in K. Brunner and A. Meltzer (eds.), The State of Macroeconomics, North Holland, Amsterdam (1980) p. 308, equation 26).

⁽⁵⁾ See Brooks and Cuthbertson, op. cit., table 11, p. 22.

⁽¹⁾ Liverpool forecast ‘compromise’ GDP. In all of the (three) Liverpool variants (based on different paths for the PSBR to GDP ratio) output in 1980 was forecast to be unchanged. (See Economic Outlook, Liverpool Occasional Papers, No. 2, 1980.)

⁽²⁾ HMT's latest estimate for the financial year 1980/81 (see FSBR 1981/82, 10 March 1981, p. 28).

⁽³⁾ NIESR error refers to the period 1979 IV to 1980 IV.

⁽¹⁾ In the FSBR 1980/81, HMT seems to treat the 7-11 per cent target range for £M3 as a ‘policy assumption’ (p. 25). Elsewhere it is stated that, ‘the course of the PSBR, based in the assumed growth of GDP and present public expenditure plans (that) should be broadly compatible with the monetary objectives’ (p. 18). We have therefore taken the view that the announced target range for £M3 for 1980/81 is HMT's best estimate of what would happen to the growth of £M3 over this period.

⁽²⁾ Interesting results on the appropriate use of intermediate targets like £M3 and the PSBR-to-GDP ratio have been put forward by Wescott et al., ‘Monetary Policy’, Third Report from the Treasury and Civil Service Committee, Session 1980/81, vol. III. For example they state (p. 65) the following conclusion which is independent of any model or objective function: ‘The pursuit of intermediate targets in conjunction with ultimate targets … leads in general, to inferior optimal solutions.’

⁽³⁾ Recent evidence on the superiority of structural over time-series models for the US economy is given by S. K. McNees, ‘The accuracy of macroeconometric models and forecasts of the US economy’ in P. Ormerod (ed.) Economic Modelling, Heinemann, London, 1979. Evidence for the UK in this area is scanty but see for example J. C. K. Ash and D. J. Smyth, Forecasting the UK Economy, Saxon House, 1973.