The econometric methodology required to measure the effects of economic policy using `causal' or backward-looking models is well established, but more sophisticated models such as National Institute model of the UK economy can also be solved in a `non-causal' or forward-looking mode. In Discussion Paper No. 526, Research Fellow Michael Artis, Robin Bladen-Hovell and Yue Ma use this model in both modes to analyse the effects of policy on the UK economy during 1974-9, the term of office of the last Labour administration and a period of considerable economic stress.

They distinguish the anticipated from the unanticipated effects of policy by first specifying `reference paths' along which policy instruments are hypothetically set at their `neutral' values: a constant treasury-bill rate at the level prevailing at the beginning of the period; a constant rate of income tax with a fully-indexed structure of allowances; and public expenditure growth in line with previous experience. Solving the model in `consistent-expectations' mode assuming that agents expected policy to be neutral yields results representing perfect foresight of the reference policy. Another simulation with policy variables set at the levels that they estimate agents actually anticipated yields another perfect-foresight path, and the difference between the two solutions estimates the effects of `anticipated policy'. A third simulation over the policy variables' actual historical values, with the expected variables fixed at the values generated by the second perfect-foresight simulation yields the `unanticipated' component of the effect of policy.

Artis, Bladen-Hovell and Ma find that fiscal policy raised output relative to the reference path until 1979; and its anticipated effect was consistently positive but its unanticipated effect predominantly negative. Expectations of government expenditures remained above the reference path until the 1976 policy switch, when actual outturns fell below expectations and remained so. Agents expected tax rates above the reference path after the first year, but they were low relative to the historical outturn after the mid-term switch. Fiscal policy added more than 3% to inflation on average, with an impact on the exchange rate amounting to a depreciation of some 20% by 1976. Monetary policy was expansionary relative to the reference path and was expected to be so throughout the period.

Comparing these results with those derived under the alternative assumption of adaptive expectations yields very similar assessments of GDP effects an initial expansionary phase and a 1976 watershed in both cases. Inflation effects come through more quickly when the forward-looking solutions are used; and the broad characteristics of the exchange rate and balance- of-payments responses are similar, although there is considerable disagreement on detail.

The Measurement of Policy Effects in a Non-Causal Model: An Application to Economic Policy in the UK, 1974-79M J Artis, R Bladen-Hovell and Y Ma

Discussion Paper No. 526, April 1991 (IM)