The home economy.

Anderton, Bob ; Britton, Andrew ; Soteri, Soterios 等

CHAPTER I. THE HOME ECONOMY

The forecasts were prepared by Bob Anderton, Andrew Britton and Soterios Soteri, but they draw on the work of the whole team engaged in macroeconomic analysis and modelbuilding at the Institute. Parts One and Two of the chapter are written by Andrew Britton, Part Three by Bob Anderton and Soterios Soteri.

PART ONE. RECENT DEVELOPMENTS AND SUMMARY OF THE FORECAST

Seldom has such close attention been paid to the indicators of economic activity and seldom have economic and political commentators been so preoccupied with short-term forecasting. It seems right on this occasion to reinforce our customary warning that the interpretation of recent statistics, and their projection even a few months into the future, is always an imprecise and hazardous exercise. This is true whether one attempts to follow a scientific methodology based on an econometric model or whether one prefers to rely on market sentiment, informed anecdotes or surveys of business confidence.

Recent developments and their interpretation

A somewhat clearer picture is now emerging of the state of the economy in the latter half of last year. The fall in output at the time is accounted for, in the latest national accounts, by lower consumption and fixed investment partly offset by an end to destocking. We would interpret the positive stockbuilding in the non-manufacturing sector in the third and fourth quarters as an involuntary consequence of an unexpected contraction of demand. As such it would point to a further fall in activity in the early part of this year.

The fall in consumer spending occurred despite a continuing rise in real personal disposable incomes. In the latter part of last year rpdi was up 3 per cent on a year earlier, but the whole of that increment was devoted to extra saving. Part of the explanation may lie in the sharp fall in the net wealth of the personal sector, although it would be unusual for this to have such a quick effect on household spending. An additional factor was the state of the housing market, where turnover was exceptionally low and prices falling in real terms. Credit conditions were also getting tighter, with the stock of consumer debt hardly keeping pace with inflation.

In the early months of the year retail sales have fallen further (although the preannouncement of a rise in VAT brought some purchases forward into March). It seems likely that consumer spending fell again in the first quarter. It may seem that the response of consumers to repeated and well-publicised cuts in interest rates is disappointingly slow, but the delay is not out of line with previous experience.

Fixed investment also fell in the latter part of last year. The housing market had been depressed for some time, and even now when turnover is beginning to improve builders are in no hurry to increase new construction. There has also been a very general fall in industrial investment, both in manufacturing and elsewhere.

Our model equation for manufacturing investment now fits the downturn remarkably well. The explanation is based on the fall in capacity utilisation, and on the decline in profits which would already have been evident to businesses by last summer, coupled with excessive borrowing in the late 1980s which left balance sheet positions very fragile - picked up in our model by a measure of |disequilibrium net liquidity'. The negative effects of these variables on investment are likely to intensify this year. The CBI Industrial Trends Survey throughout last year was pointing to a fall in investment, in both buildings and new machinery. There was a further sharp deterioration in the surveys conducted in January and April of this year.

The fall in investment has not been confined to manufacturing industry. Separate figures are no longer available for distribution and other services, but the suggestion is that the strong upward trend in investment by these sectors may now be in reverse, and there is no reason to be confident that it will revive the same pace even when the present recession is at an end.

Business confidence, as measured by the CBI Survey, dipped sharply last October. It is difficult to know to what extent this was a reflection of what was actually happening to industry |on the ground' at that time, or to what extent it reflected the Gulf crisis or the political situation at home. This series is subject to wide variation, sometimes unrelated to the current or immediately future state of the economic cycle.

The reliability of this series as a forward indicator of the cycle is of particular interest at the present time since it recovered sharply in the April survey. (It is usual for business confidence to improve in the Spring but this went beyond the normal seasonal pattern.) A regression equation linking this indicator with the economic cycle as measured by capacity utilisation suggests a lead of about six months. A comparison of turning points indicates a lead of one or two quarters in the 1980 trough, but as much as five quarters at the 1988 peak. The CSO include the series in their longer leading composite indicator, and give it a median lead relative to their reference cycle of 12 months (with a range from -34 to +6 months). All one can say is that the upturn in confidence in the April survey, if it is sustained, probably points to an upturn in the economy some time in the next year.

Apart from this and some similar evidence from other surveys there is little if anything to suggest that the upturn is in sight at all. The rise in unemployment in the first quarter confirms the depth of the recession up to that point and will itself be associated with lower employment earnings and with lower consumer demand. The relatively low level of import volume is consistent with the picture of subdued demand and activity.

The effect of the recession on the underlying rate of inflation is still difficult to detect. The growth of average earnings in the latter half of last year implied by the wage bill and the employment figures is as high as 11 per cent, rather hard to believe at this stage in the cycle. Perhaps one or other of the component series will be revised. The more recent information on settlements shows a wide variety, including some well over 10 per cent, but also some postponements which will bring down the average increase.

The |headline figure' for the retail prices index came down from 10 per cent in the first quarter of last year to 8.7 per cent in the first quarter of this year, and it is bound to be much lower in the second quarter. There is no suggestion as yet of a slowdown in the rise in wholesale prices, which is surprising given the steep fall in capacity utilisation. On the other hand the deflator for fixed investment actually fell in the fourth quarter of last year.

The effective exchange rate index for sterling rose by 6 1/2 per cent between the first and second halves of last year. Mainly for that reason the prices of manufactured exports and imports both fell significantly.

Membership of the exchange rate mechanism has held up sterling against European currencies, but there has been a fall of 13 per cent against the dollar since the fourth quarter of last year, associated with a fall of nearly 5 per cent in the effective exchange rate index. Especially since many world prices are set in dollars in the short term, the effect of this depreciation on prices in this country will be significant within the next few months. On the other hand we are already benefiting from the halving of world crude oil prices between the fourth quarter of last year and the first quarter of this year.

The only change in the Budget this year which is of significance for the short-term forecast was the reduction in the poll tax (costing 4.5bn [pounds] in a full year) and the increase in the rate of VAT (yielding 5.5bn [pounds] in a full year). The estimated effect of the VAT increase on its own would be to raise prices by about 1 1/2 per cent in the first year and reduce output significantly, even in the long run. It is likely however that the economic effects of the poll tax reduction together with the VAT increase will be broadly offsetting.

Short-term interest rates have been progressively reduced from 15 per cent last summer to about 11 per cent (for 3-month inter-bank deposits) in the second quarter of this year. Mortgage interest rates came down from 15 1/2 per cent to about 13 1/2 per cent and are confidently expected to fall further in the next few months.

Policy assumptions

Early in May the 12-month interest rate on Euro-sterling deposits implied a market expectation of short-term rates falling to little below 11 per cent by next Spring. This seems to us a little pessimistic. We are forecasting base rates of 10 1/2 per cent by the end of this year and 10 per cent by next Summer.

This judgement is clearly linked with our forecast of the exchange rate, which is explained in more detail than usual in Part Two below. As the possibility of a realignment some time within the forecast period cannot be ruled out we have to construct our central forecast as the mean of a probability distribution. This shows a gradual depreciation. We would in fact attach a high probability to a constant exchange rate scenario, but this would involve the progressive correction of market expectations. A variant forecast on these lines is described in the Annex to this chapter.

In principle we treat the outcome of the general election in a similar way. The situation now is different from that in the run-up to the 1983 and 1987 elections in that the two largest parties have moved closer together in their statements on macroeconomic policy. The main difference of emphasis is the relative priority of tax cuts and public spending increases if the growth of the economy justifies either. We discussed Labour Party policy at some length in the Review last November, and we do not have enough additional information to construct a formal forecast variant. Instead we have adopted an intermediate assumption with some tax cuts and a modest rise in public spending. This could be interpreted as a fifty-fifty weighting of the probability of a change of government. If so it is the mean of the probability distribution, but not of course its mode.

The conclusion of our medium-term projections is that the scope for either spending increases or tax cuts will be limited. This has been the message of our forecasts for some time now; it is based on the view that the reduction in interest rates resulting from ERM membership will on its own stimulate demand as much as is compatible with keeping inflation in check and with external balance. In our forecast growth averages over 2 1/2 per cent between 1992 and 1998. Tax receipts (especially corporation tax) initially lag behind the recovery, but by the mid-1990s a substantial |growth dividend' can be expected. In forecasting we always assume that public sector pay rises in line with earnings in the private sector. Since the value of public authorities' consumption accounts for about three-quarters of the yield of direct and indirect taxation much of the |dividend' is, on our assumptions, automatically spent in this way. It also seems unrealistic to suppose that the buoyancy of real receipts from national insurance contributions will not be reflected in increases in benefits beyond the minimum required by price indexation.

Summary of the forecast

We now expect GDP to fall by 2 to 2 1/2 per cent this year with the recession just a little longer and deeper than we said in February. The quarterly path will never be known precisely, so precise forecasting would be almost meaningless, but the profile we have in mind would show total output virtually flat for the rest of this year. This means that there may not be clear evidence that the recovery has actually begun until well into next year. Year-on-year growth in 1992 is forecast at about 2 per cent.

Consumer spending should recover by the latter half of this year with real incomes rising a little and the savings ratio beginning to edge down again in response to interest-rate cuts. By next year the growth of consumption could be up to 1 1/2 or 2 per cent again.

The prospect for an upturn in industrial fixed investment is more distant. Our model equation for manufacturing investment predicts a fall of 15 per cent this year, which is compatible with the survey evidence. It then predicts a further fall in 1992, thanks mainly to the lagged effect of falling profits and liquidity. This substantial investment recession will have a lasting effect on the supply side of the economy (see the box on page 27 below). We estimate that the level of manufacturing capacity will be reduced by some 5 per cent at the end of next year and productivity growth in the medium term will be some 1 per cent a year lower as a result of the investment shortfall this year and next.

The outlook for export growth is reasonably good if world trade growth is maintained at 5 per cent this year and 6 per cent next year (see chapter II). Thanks to this, and the subdued level of imports, the balance of payments deficit this year should be much reduced - to about 6 billion [pounds], compared with 12.8 [pounds] billion last year. But the outlook for 1992 and beyond is not so good.

Further falls in mortgage rates will help to reduce the |headline' inflation figure to about 4 per cent by the end of this year, but it is not expected to fall further than that in 1992. The deflator for GDP at factor cost is expected to rise by 7 per cent this year and 5 per cent next year. The fall in underlying inflation should continue into the medium term.

Unemployment can be expected to rise strongly throughout this year, reaching over 2 1/2 million by the Autumn. Thereafter its rise should be much slower, but the actual peak may not occur until late in 1992.

For the medium-term we are again able to show a smooth transition to monetary union in 1997. The rate of inflation settles at 2 to 3 per cent a year which is similar to the rate elsewhere in Europe. This is compatible according to our model with steady growth in GDP at around 2 1/2 per cent a year, with unemployment at about 2 1/2 million, falling very slightly.

PART TWO: FORECASTING THE EXCHANGE RATE

In the past one of the most difficult tasks facing an economist making a short-term forecast has been to anticipate movements in the exchange rate. If we look back over the past 20 years at the record of forecast made by the Institute or by any of the other forecasting teams in this country we find that many of the largest errors in forecasts of inflation and output occur as a result of unanticipated exchange-rate movements. This is not at all surprising. Theory says that the expected gain from holding any two currencies should be the same, so that the predicted change in the exchange rate for any period ahead should be equal to the difference between the interest paid on similar assets of that maturity in the two currencies. As soon as any new information becomes available to the market, the exchange rate will jump to a new value, and the arbitrage conditions will again be satisfied. If a reliable method existed for forecasting exchange-rate movements, the market would soon discover it and make it ineffective.

Now that the UK has joined the exchange-rate mechanism, errors in exchange-rate forecasting should be less of a problem, and as a result forecasts of inflation and output should be more reliable. However, we are now in a period of transition when the possibility of an exchange rate realignment cannot be entirely ruled out. In this section we consider the circumstances in which a realignment could still occur, and interpret the present differential between interest rates in the UK and Germany. Our main forecasts are based, as usual, on the assumption of consistent expectations - that is that market expectations will in fact be realised. But the market expectation is the weighted average of many possible paths for the exchange rate in the future.

Analysing the policymaking process

The choice of an exchange rate within the new regime illustrates the problem of time inconsistency. It pays the authorities to build up a reputation for keeping their commitment to exchange-rate stability even though in general the |optimum' exchange rate will change as new information becomes available or simply with the passage of time. As the commitment of the UK authorities is a recent one, the market is in the process of learning how firm that commitment actually is. Eventually an |expectational equilibrium' may be reached, when the behaviour of the authorities is fully understood by the market, but we are not at that equilibrium yet.

Formally the determination of the exchange rate in these circumstances could be analysed as the interaction of three processes. The first would be the behaviour of the economy (represented in our forecasts by the Institute's macroeconomic model). The second would specify the policymakers' preferences, which could be represented by a welfare function depending on present and future values of variables such as inflation and unemployment, together with a set of parameters which define their relative priorities. In the short term an exchange-rate depreciation adds to inflation, but reduces unemployment. In the long run (if there are no further depreciations) both inflation and unemployment will be unaffected. Because the authorities wish to keep unemployment as low as possible they will be tempted to realign, but if they value their reputation for consistency they will be reluctant to do so.

The third process would be the formation of expectations, and this is the most difficult of the three to describe because it involves learning from experience. Expectations must reflect the subjective probability of future realignments based on the evidence available about the authorities' intentions. These probabilities will be updated each time period in the light of a new observation of the authorities' decision to realign or not to realign. But there is no general rule, specifying how these estimates should be updated in the light of new information, which would be appropriate in all circumstances.

The expected exchange rate for next period must be calculated as the weighted average of two possible outcomes - a realignment or no realignment. But this probability has to be specified for each period of the future before even a one-step forecast can be made, because today's optimum exchange rate must depend on whether the authorities expect to realign again at any period in the future. Thus the distribution of expectations relating to the next period will in general by asymmetric: there will be a concentration of probability at the point of zero change, with the rest of the distribution scattered over a range of points corresponding to the optimum outcome on all possible assumptions about alignments in the future. The weights will reflect the latest estimates of the authorities' preferences.

A realignment will occur when its expected benefits exceed its costs, including the cost of lost reputation. Given continuity (appropriately defined) these net benefits will be zero at the point where realignment occurs, so any actual realignment would provide valuable information about the authorities' preferences. Following realignment a new commitment is made to the new exchange rate, which will hold until it pays to abandon it. The authorities are themselves uncertain as to whether they will realign again or not - but that uncertainty reflects only the uncertainty about the future course of the economy. They are assumed to know their own present and future preferences.

A very simple approach

This rather complex interaction can be simplified by concentrating attention on the issues of commitment by the authorities and learning by the markets in the current situation.

The aim of the UK authorities is to get inflation down, at least to the European average, but this must be at an acceptable cost in terms of unemployment an international competitiveness, which depend on relative prices. So long as our rate of inflation is higher than that in the rest of Europe we are losing competitiveness. Sooner or later our loss of competitiveness will stop; the more credible the commitment to the ERM, the sooner that will happen. But there is in fact a level of competitiveness (and hence of expected unemployment) at which the authorities would give up the struggle and realign the exchange rate. For obvious reasons they are not telling us what that level is. The process of learning in this case consists simply of updating estimates of the critical level.

If the authorities failed to acquire any reputation at all, then competitiveness would suffer until that crucial point was reached. The pound would then be devalued, inflation speed up again and the process would be repeated. (It is possible however, that the rise in unemployment would stop the inflation before the realignment was triggered.)

Credibility improves the chances of avoiding the need for realignments. As competitiveness worsens, but the authorities do not devalue the pound, their credibility improves, and the mean expectation of devaluation is reduced because estimates of the unemployment threshold are revised up. In fact just before the threshold is reached the mean estimate of where that threshold lies will almost certainly indicate a greater toleration for unemployment than the authorities actually have. If the threshold is actually crossed, it will (in this very simple model) be known with certainty from that point on. But so long as the threshold is not crossed the authorities can benefit from the reputation they have built up, which could in fact be based on an over-estimate of their determination not to realign.

Two further complications

At present, the forecasting of the exchange rate is further complicated, both by the approach of a general election, and by the transition to economic and monetary union.

The timing and the outcome of the election are both in principle endogenous. It affects expectations even if the welfare function of the authorities would in fact be the same under any government. If there were a change of government then the markets would not know whether the priorities had changed or not. The process of learning would have to begin all over again. That would of itself the probability of realignment. Equally important is the possibility that the approach of an election would influence the priorities of the government in office. Thus a realignment is unlikely just before an election, and more likely just after, whatever the outcome of the election may be.

The possible approach of EMU provides an endpoint to the authorities' optimisation. If exchange rates are to be |set in stone' at some period in the future, the rate at that time becomes especially important. It is possible that a last realignment will take place at the time when EMU comes into effect. But the likelihood of EMU happening (or of the UK being a full member of it) may itself depend on what exchange-rate policies the UK follows in the meantime. Clearly the general evolution of the ERM towards a more fixed exchange-rate system has been influenced by the wish of most European governments to make the transition to EMU. The French, for example, are keen to demonstrate that they can manage without realignments, so as to make EMU more likely. If the UK authorities want to avoid being in the slow lane of a two-speed Europe, they will be reluctant to realign at all, in case that is seen as a sign of weakness.

Table : Interest-rate differentials

The following were the Euro-currency rates quoted in the Financial Times for 2nd May.

 1 month 3 months 6 months 12 months
Sterling 11-81 11-66 11-38 11-19
D-Mark 8-88 9-06 9-19 9-31
Differential 2-93 2-60 2-19 1-88

We assume that the mean expected depreciation of sterling against the D-Mark was equal to these differentials, so that the expected return from holding both currencies was the same. This may not be exactly true because of transaction costs, or because the market prefers one currency to another for reasons other than the expected return - possibility related to the variances and covariances of currencies which go to make up an optimum portfolio. But we would guess that between sterling and the D-Mark the |arbitrage condition' nowadays holds more-or-less exactly.

The expected depreciation of sterling implied by these interest differentials would be consistent with a probability distribution such that the likelihood of appreciation is very small, the likelihood of no change is very high, and the likelihood of deprecation, even downwards realignment, is not negligible. One scenario consistent with the mean expectation would be a gradual depreciation within the present wide band, followed by a narrowing of the band with a lower central rate. But it should be emphasised that the forecast path is the average of a wide variety of possibility scenarios, not the illustration of a particular central case.

It is quite likely that, in the event, sterling will not in fact depreciate at all in the forecast period. A variant forecast on that assumption is described in the Annex to this chapter. Should this happen, some market participants will be |surprised' and the mean forecast of the exchange rate will rise as these participants learn from their mistakes. Thus in the variant forecast the assumption of |consistent' expectations in its usual strong form does not hold.

PART THREE. THE FORECAST IN DETAIL

Forecasts of expenditure and output (table 1)

The decline in output since the second half of last year has been associated with some interesting developments. Although de-stocking occurred in the last quarter of 1989 and the first of last year, this did not continue into the final six months of 1990. Consequently, given the output downturn, this latter period suggests that involuntary stockbuilding has taken place. An explanation for this may be that private sector decision makers did not expect the recession to be so severe. A second feature of this recession is the rapid slowdown in the growth of manufacturing exports (see Chart 1). However, it is difficult to discern whether this is due to a deterioration in price competitiveness and recently depressed world trade or whether the squeeze on profit margins is reducing supply. Partly for these reasons we are now forecasting substantial de-stocking this year, which extends into 1992, and low export growth for 1991.

We have revised down or forecast for output growth this year and now expect total GDP to decline by around 2 1/4 per cent with a slightly larger fall in non-oil GDP. We expect non-oil GDP to stop falling in the second half of 1991 and to remain fairly flat for the rest of the year. Our forecast of a 10 per cent decline in total investment during 1991 contributes to the depressed activity. The largest fall occurs in manufacturing investment. For 1991 this percentage fall in manufacturing investment is similar to that predicted by the CBI but we are also predicting a further fall next year. This is consistent with a revision of output expectations by the company sector which had been too optimistic. The current CBI survey indicates a substantial increase in excess capacity from the beginning of last year.

Although our forecase of a recovery in total GDP growth next year is similar to the February forecast, the extent of the downturn in 1991 now expected means the level of activity we expect this year is now lower. Hence, unemployment is now expected to reach the higher peak of over 2 3/4 million in 1992. [Tabular Data Omitted]

Personal income and expenditure (table 2)

Real personal disposable income increased at an annual rate of 3 per cent during the final quarter of last year. The savings ratio therefore rose to 10 1/2 per cent as real consumers' expenditure declined by 1 1/2 per cent over the same period. Data for retail sales for the first two months of this year confirm the continuation of a deceleration in consumption but the figure for March was biased upwards due to purchases brought forward to avoid the VAT increase announced in the Budget. Recent figures for consumer credit and bank lending indicate that consumer borrowing was depressed in the first quarter of this year. In addition, the number of new vehicle registrations in March was 6 per cent less than in the previous three months (26 per cent lower than in the same quarter a year ago). However, in contrast, to this data, the Gallup survey indicates that consumer confidence, although remaining at a very low level, has recently recovered slightly.

Our predicted path for consumption has a very important bearing on our forecast as this category of expenditure accounts for more than a half of total domestic demand. Our model equation for consumers' expenditure has been overpredicting over the past few quarters but this residual discrepancy frequently disappears when data are revised. Part of the statistical adjustment, which itself is frequently revised, is included in the data for consumption and this raises further concerns about the accuracy of the data. Nevertheless, we apply the recent negative residual to our forecast for consumers' expenditure.

We expect real consumers' expenditure to fall by more than 1 per cent this year with most of the decline occurring in the first half of 1991. Expenditure on consumer durables will suffer the largest decline of around 8 per cent. This is probably what most forecasters are expecting although the Treasury is far more pessimistic in the budget forecast (predicting a 1 3/4 per cent fall in consumption). The factors determining our predicted decline in consumption are a slowdown in real disposable income growth, lagged adjustment to the 10 per cent fall in real net wealth last year, and continued slow credit growth especially in the first part of 1991. Consumer credit growth will take time to respond to the lower interest rates and the substantial expected falls in inflation mean that real interest rates remain high. Given that we expect real income to rise by almost 2 per cent this year, largely because real wages remain high as price inflation falls more rapidly than wage inflation, this implies a rise in the savings ratio in 1991. Intangible factors such as consumer confidence affect the latter; confidence will be boosted by declining interest rates, falling inflation and the end of the Gulf War but the rapid rise in unemployment should dampen the recovery in confidence. A recovery in consumers' expenditure will probably begin in the latter part of this year, and continue at an accelerating pace, throughout next year. Although real disposable income growth will probably remain subdued in 1992, consumption will be encouraged by cheaper credit and increased real wealth as base rates continue to fall. [Tabular Data Omitted]

Fixed investment and stockbuilding (tables 3 and 4)

Growth in total real investment expenditure came to a halt in the second half of last year. Although investment growth had been decelerating for some time before this, it now seems that capital spending was based on output expectations far more optimistic than those currently held. The abrupt decline in activity from the later half of 1990 onwards has resulted in considerable excess capacity. The latest CBI survey confirms that this is particularly acute in the manufacturing sector. The situation is exacerbated by the financial deficit of the company sector which was probably viewed as sustainable under the previous more optimistic expectations. Therefore, instead of companies reducing their financial deficit via continued output and hence profits growth, the route back towards financial balance will partly be by means of investment expenditure cut backs.

Although firms are forward looking and probably expect activity to recover in the near future, the combination of uncertainty, financial distress and excess capacity will encourage a substantial fall in investment this year. We expect total investment to fall by 10 per cent this year with declines in all categories except the oil sector where safety equipment is being installed. Our new manufacturing investment equation agrees with the CBI survey that investment in this category will decline by around 15 per cent this year. Although profitability and manufacturing output recover next year, the equation predicts there will be a further fall in manufacturing investment next year largely because capacity utilisation remains at a low level.

Housing investment experienced a substantial fall in 1990 and we expect this to continue, but at a more moderate pace, this year. The decline in base rates and mortgage rates so far has resulted in a rise in sales of property but the supply has originated largely from the existing surplus housing stock. Consequently, we expect positive growth in housing investment next year (rather than this year) as mortgage rates continue to decline. Both the business and financial services and distribution sectors will probably experience a subdued recovery next year. However, this may only be enough to arrest the decline in total investment with the result that total capital spending is likely to remain flat in 1992.

Considering the severity of the downturn in activity, the amount of de-stocking has been surprisingly low. It seems that the scale of the decline in demand has left firms with larger than desired levels of stocks. This is confirmed by the CBI survey where stocks of finished goods in particular are much higher than desired. We are forecasting substantial de-stocking in all sectors this year. However, de-stocking will probably continue at a decelerating rate through 1992 as the expected output fall this year will cause stock/output ratios to rise to an undesirably high level. [Tabular Data Omitted]

Balance of payments (table 5 and 6)

The recent 3 billion [pounds] revision to the invisibles trade balance reduced the current account deficit for 1990 to around 12 3/4 billion [pounds] or 2 3/4 per cent of GDP. This amounted to a 7 billion [pounds] improvement in the external balance of the UK compared to 1989 and the stepdown in the visible trade deficit when output declined in the second half of last year is quite pronounced. The beneficial movement in the terms of trade over this period may also partly explain both the improvement over this period and the lack of any further visible deficit reduction in the first quarter of this year. Chart 1 shows that, after performing exceptionally well in the past few years, export volumes of manufactures have not contributed much to the recent decline in the visibles deficit. Given the recession, many economists expected UK manufacturers to switch from the domestic market to overseas but weak world trade growth and a loss of price competitiveness may have prevented this. It may also be the case that the poor financial situation of UK firms has forced cut backs in non-price competitiveness factors such as price-discounting and advertising.

Although the recent upward surge in the US dollar (see chart 2) will improve the overall price competitiveness of UK exports we do not expect a strong response from export volumes as the UK's major export market is now Europe. Germany is our largest single export market and sterling has recently appreciated against the D-Mark. Given these factors and our forecast of subdued world trade growth of below 5 per cent this year, exports of goods and services will probably only grow by around 1 per cent in 1991. Next year should see a substantial recovery in exports of goods and services in response to our expected sterling depreciation against the D-Mark and renewed growth in the UK's export markets. It is this external demand which will in part lead the UK out of recession.

Import volumes of manufactures may decline by around 3 per cent this year in response to large falls as investment, stockbuilding and consumption of durable goods. Imports of raw materials and oil will also fall in line with demand and the visible deficit should be reduced this year to around 11 billion [pounds]. A visible trade deficit of 2 3/4 per cent of GDP when GDP itself is falling by more than 2 per cent indicates that the UK's trade deficit will be around for some years. Next year will see a renewed deterioration in the trade balance as domestic demand recovers.

There is no reason to believe that the invisibles balance will not continue at its revised higher level. The surplus on interest profits and dividends this year will benefit from the recent upsurge in the dollar which increases the sterling returns on UK assets held in the USA. This effect will be bolstered by declining UK interest rate differentials (and hence reduced returns on UK liabilities) and transfers will benefit from foreign contributions to the Gulf War. The surplus on services should remain at around 3 1/2 billion [pounds] for this year and next. Given these factors the current balance will look far more flattering than the visible deficit with a current account deficit of the order of 6 billion [pounds] in 1991 deteriorating to around 8 [pounds] billion next year. [Tabular Data Omitted]

Output and the labour market (tables 7 and 8)

The decline in real GDP since the peak of the economic cycle, in the middle of 1990, has accumulated to approximately 3 per cent. In contrast to the 1980/1 recession major output falls have been experienced in all areas of the economy. The manufacturing and construction sectors have endured the largest falls in output since the middle of 1990, 6 1/2 and 5 1/2 per cent respectively. They are closely followed by the distribution and business service sectors whose corresponding rates are 4 and 2 per cent respectively. The uniformity of the recession amongst sectors during the present downturn seems to be due to the effects of continuous high real interest rates which are in contrast to real interest-rate movements in 1980/1 (when base rates were actually lower than inflation for a time). We expect further, though much more subdued, falls in all non-oil output categories, except for the public sector, until the middle of this year. A flat output profile is then expected until growth is resumed at the end of the year. Year on year growth rates in most sectors however will still be significantly negative.

The workforce in employment data series have recently undergone major revisions on account of the 1989 Census of Employment and the 1990 Labour Force Survey. The revised figures indicate that employment peaked in the second quarter of 1990 and has since fallen by 340,000. Employment falls have lagged behind the decline in output in their usual pro cyclical manner. This is usually interpreted as the effect of labour being a quasi-fixed factor of production, the degree of fixity arising from fixed costs associated with hiring, firing and training employees.

Even through employment falls had been predicted the steep increase in unemployment in March of 113,000, the highest ever recorded, and the subsequent rise in April of 84,000, were higher than previously expected. This is probably due to a combination of two factors. Firstly the recession is now expected to be worse than initially anticipated. Labour hoarding may have taken place due to labour posessing a degree of factor fixity, but once the true extent of the recession revealed itself and firms' expectations adjusted to new information their optimal employment requirement, was drastically reduced, giving rise to the recent large increases in unemployment. The second reason is the sharp squeeze on profit margins. High real interest rates, falling sales and rising unit labour costs have squeezed profits dramatically. A bleak domestic market and a tighter world market have led firms to contain cost increases via reductions in stocks and employment levels since this is a relatively quick way of restoring company sector profitability in a depressed market. Our forecast sees substantial rises in the unemployment total over the near future. We expect the unemployment total to reach 2.6 million by the end of this year and then to continue growing until its peak of 2.8 million in the middle of 1992. A comparison of expected unemployment rates in the UK and other European countries is discussed in Box. 1. The rapidly falling employment levels and the slower rate of decline in output lead us to expect an increase in productivity growth in the latter part of 1991 with a considerable acceleration in 1992.

BOX 1. UNEMPLOYMENT by Paul Gregg

The somewhat deeper recession than previously forecast implies somewhat higher claimant unemployment by the end of this year. Hence we now anticipate claimant unemployment will rise to just under 2-6 million by the fourth quarter and to just under 2.8 million by the third quarter of 1992, which we anticipate will be the peak level. The continued rise in unemployment after the beginnings of a recovery reflects in part the weakness of the recovery in the later half of this year and partly that unemployment lags behind output changes by between 6-9 months. The latest Labour Force Survey results provide evidence that the claimant count understated unemployment on the internationally accepted ILO/OECD definition (i.e. those searching for work in the last four weeks and available to start in the next two). As of Spring 1990 the LFS measure stood 300,000 higher than the claimant count. As table 1 shows this meant that the UK's unemployment rate stood at about one percentage point higher than that indicated by the claimant measure. This difference significantly alters our view of unemployment in the UK relative to our EEC partners. [Tabular Data Omitted]

Wage and price inflation (tables 2, 9 and 10)

The most recent data indicate that the underlying rate of increase of whole economy average earnings is 9 1/4 per cent. Although this is a half per cent reduction since the previous quarter it is disappointing given the substantial falls in wage drift that must have occurred since the downturn began. For example, overtime hours worked in manufacturing have been falling rapidly since the second half of last year. However, the CBI's pay databank shows that manufacturing settlements fell to 8.3 per cent in the first quarter of this year from a growth rate of 8.9 per cent in the previous quarter. Further confirmation that wage inflation is receding is provided by Incomes Data Services who report that two thirds of pay deals monitored by them at the start of the first quarter of this year were above 10 per cent whereas by March only one third were above 10 per cent

The dispersion of pay settlements has increased during the early phase of the downturn. Sectors particularly affected by the recession are offering much lower pay settlements. For example, Thomas Cook staff have experienced temporary pay cuts, pay settlements within the banking industry have been between 5 to 8 per cent and the 5 per cent pay increase in the concrete industry indicates the severity of the recession in the construction industry.

Although average settlements and earnings are moving in a downward direction the current rate of increase is clearly not sustainable. For example, unit labour costs in manufacturing were growing at 12 per cent in February of this year. Settlements should decline through this year as unemployment increases rapidly and retail price inflation continues to fall. We envisage that average earnings will be growing at around 8 per cent by the end of this year. Although the continued increasing slackness in the labour market will put further downward pressure on earnings next year this will be partly offset by strong import price growth (stemming largely from the recent strength of the dollar) and a recovery in productivity. Consequently earnings may still be above 7 per cent at the end of 1992. Throughout our forecast we have assumed that the switch from the poll-tax to VAT will have a neutral effect upon wage inflation, although the increase in the RPI will be reduced.

Manufacturing output prices are currently still accelerating which is disappointing given the sustained previous fall in manufacturing input prices. However, the latest CBI survey firmly indicates that manufacturing prices will soon slow down as the number of companies expecting to increase prices for domestic sales is lower than at any time since the survey began. We are also predicting falls in manufacturing wholesale inflation but do not expect this to occur until the growth in unit labour costs has fallen (in the latter part of 1991) as these costs pressures are the main reason why output price increases are still high now. Lower growth in unit labour costs next year should ensure a further fall in manufacturing output inflation in 1992.

The underlying rate of inflation as represented by the consumer price index (corrected for the poll-tax treatment in the national accounts) should bottom-out by the end of this year. The officially published consumer price index will again be distorted during 1991 by the reduction in the poll-tax from the second quarter of this year. Underlying consumer price inflation will creep up again next year as the rise in import prices will feed directly into consumer goods.

The increase in the retail price index will fall considerably through this year as mortgage interest payments decline in line with our projected base rate falls. By the end of 1991 retail price inflation should be around 4 per cent but this will increase slightly to 4 1/2 per cent next year as underlying inflation is pushed up by rising import prices. [Tabular Data Omitted]

Public sector finance (table 11)

It seems that the public sector was roughly in financial balance for the 1990/1 fiscal year. This compares with a debt repayment of 8 billion [pounds] for the previous fiscal year. Recent research by the Treasury has shown that the PSBR is more pro-cyclical than previously thought and this is consistent with the dramatic deterioration in public finances since the beginning of the recession. Revenue growth will continue to decline as corporation and income tax are depressed by lower economic activity and current grants will increase considerably given the actual and expected rise in unemployment. The March Budget will have an almost neutral effect on the PSBR for the 1991/2 fiscal year although the partial switch from the poll tax to VAT will have a more pro-cyclical negative effect upon revenues. However, transfers from Kuwait etc. in the form of Gulf War payments will slow down the slide into public sector deficit. Given these factors we are expecting a PSBR of approximately 8 billion [pounds] for this fiscal year followed by a further deteriotation to around 14 1/2 billion [pounds] next year.

As explained in Part One, our forecast is based on a 50-50 mixture of Conservative and Labour policies (as the result of the election is uncertain). Therefore, privatisation revenue is set at roughly half that scheduled by the government and public expenditure is somewhat higher. We also assume some limited tax cuts in the medium term. [Tabular Data Omitted]

Medium and long-term forecast (table 12)

It seems that the average growth rate for the UK economy for the rest of this century will be around 2 1/2 per cent. Growth may be slightly above this rate in the mid-1990s if interest rates fall in line with our projections. Given this profile for GDP we expect unemployment to remain around the 2.5 million for some years. UK membership of ERM will dictate that external price competitiveness cannot be improved by exchange-rate depreciation. Therefore, to maintain a sustainable current balance deficit over the longer-term unemployment at this level is necessary to maintain competitiveness by restraining domestic earnings and prices growth. Slow employment growth enables total productivity to grow by approximately 2 per cent in the medium term.

Consumers' expenditure will probably be the major source of demand over the medium-term period. From 1993 onwards the personal sector savings ratio will probably experience a gradual but clear fall. This is partly due to our assumption of monetary union circa 1997 and the corresponding fall in UK interest rates towards German levels (see chart 3). Consumption is therefore largely driven by upward revaluations in net wealth and strong consumer credit growth as base rates fall. Subdued real disposable income growth in the medium-term also contributes to the savings ratio fall. Downward pressure on wage inflation from both the high level of unemployment and the discipline of ERM membership explains much of the slow growth in real incomes.

The balance of payments deficit will probably deteriorate in the mid-1990s. This will be associated with the aforementioned decline in the personal sector savings ratio and the continued dis-saving of the public sector. The PSBR will probably remain at around 2 1/2 per cent until the mid-1990s due to high unemployment and the lower growth rate of GDP compared to the 1980s. In contrast, over the same period, the financial deficit of the company sector will gradually diminish but renewed investment and dividend growth will limit the extent of the recovery in corporate finances.

The medium-term supply-side effects of the expected decline in manufacturing investment caused by the current recession are analysed in Box 2 below. [Tabular Data Omitted]

BOX 2. THE SUPPLY SIDE AND THE RECESSION (*)

The stated objective of the government has been to reduce inflation by maintaining a high level of interest rates. The current contraction in demand provides evidence that this policy is certainly deflationary in the short term but the associated fall in investment will cause longer term damage to the supply side of the economy. We have attempted to quantify these supply side effects by simulating what would have happened to the economy if manufacturing investment had continued to grow at its trend rate rather than declining during the recession as in our main forecast. We have therefore constructed a variant where manufacturing investment grows from the last quarter of 1990 at an annual rate of 3 per cent for three years. We have ensured that the ex ante demand for manufactures is the same as in our main forecast by reducing expenditure on consumer durables by the same amount as the extra investment generated in the variant. Therefore, the simulation captures a scenario where the current recession is generated by a contraction in demand by reducing consumption without damaging manufacturing investment. Given that the UK is now a member of the ERM we have assumed that the exchange rate and interest rates are the same as in our main forecast. The simulation results are expressed as percentage changes from our main forecast and are shown in table B1. B1.

Table B1. Manufacturing investment growing at trend rate of 3 per cent per annum

 1991 1992 1993
Manufacturing investment (volume) 16.6 1992 30.3
Manufacturing capacity utilisation -0.6 29.3 -3.5
Manufacturing capacity (volume) 0.9 -2.2 6.3
Manufacturing output (volume) 0.3 3.5 2.8
Consumer prices 0.7 1.3 -3.7
Average earnings -0.6 -2.4 -1.0
Manufacturing productivity -0.2 -1.2 3.2
Imports of manufactures (volume) 0.1 1.3 -1.1

The National Institute's econometric model embodies a vintage capital model of production. New investment embodies technical progress which will tend to raise the average productivity of labour. The capital stock is defined in terms of vintages of machinery and variations in demand and (hence capacity utilisation) will change the age profile of machinery used and the productivity of workers. In our model, a higher level of manufacturing capacity utilisation will raise both prices and imports. The first effect is capturing the pressure of demand (i.e. overheating) and higher imports result from a deterioration in non-price competitiveness factors (i.e. delivery delays, poor quality control etc.). Higher import also cause substitution away from domestic goods and result in a fall in UK production.

Turning to the simulation results we see that imposing a trend rate of growth for manufacturing investment raises its level by about 30 per cent compared to our main forecast over the period shown. Manufacturing capacity is therefore higher with the result that when the economy recovers from the recession |overheating' is not so prevalent and there is less upward pressure on the price level. Productivity is also higher; in our base forecast older vintages of machinery are brought back into production when output rises whereas in the simulation younger vintages of capital are employed which embody superior technology. Wages benefit from the higher productivity and therefore earnings do not fall as much as prices. However, unit costs fall because of the rise in productivity. As the nominal exchange rate is fixed, price competitiveness improves. Non-price competitiveness benefits from the lower level of capacity utilisation and this increases domestic manufacturing output as UK goods replace those previously supplied from overseas.

Given that this simulation only relates to manufacturing investment, it is fair to assume that the damaging supply side effects of the current recession will be greater than those reported above as we are also expecting substantial falls in non-manufacturing investment this year.

NOTE

(*) This exercise is based upon similar simulations originally performed by Simon Wren-Lewis in |Supply, liquidity and credit: the Institute's domestic econometric macromodel', National Institute Economic Review, November 1988, pp. 32-43.

ANNEX. MODELLING THE EVOLUTION OF STERLING'S CREDIBILITY IN THE ERM

Introduction

Part Two of this chapter has discussed some of the issues involved in forecasting the exchange rate under a policy regime where the authorities have stated their intention to hold the exchange rate within a particular band but the credibility of that policy announcement is not complete. Accordingly, the forecast assumes that the exchange rate depreciates in line with market expectations. This Annex illustrates the implications for the forecast outturn if the government does stick to its intended exchange rate path and the private sector gradually learns about the government's true policy stance.

Despite the government's announced policy of pegging the sterling-D-Mark exchange rate, the main forecast described in this chapter is characterised by a forecast path which reflects the mean expected depreciation of sterling against the D-Mark as given by the differential between sterling and D-Mark short term interest rates. Thus, we are assuming that the uncovered arbitrage condition holds exactly, implying that, ex ante, investors will be indifferent between holding sterling and D-Mark assets. As explained in Part 2, the expected depreciation implied by the interest rate differential will depend upon the probability distribution as perceived by the markets of all possible outcomes for the exchange rate. The base forecast which is assumed to reflect this probability distribution is shown in Chart A1. In what follows, we illustrate how this expected path for the exchange rate may have been arrived at by making a number of sylised assumptions about the underlying probability distribution. In doing this, we are necessarily taking as given the complex economic and political factors considered in Part 2, which actually determine this probability distribution. As will be shown below, the methodology required for computing the mean expected depreciation is itself complicated enough.

Computing the mean expected depreciation

The authorities have entered into a fixed exchange rate regime, the ERM. Let us suppose the announced policy rule is that interest rates will be set at the required level to hold the exchange rate at its target level against the D-Mark; we assume this is set at the forecast 91q2 value of 2.97 D-mark. Further, let us suppose that interest rates can only be changed at the beginning of each year at which time expectations within the private sector will also be set. However, at any time within the year there is a possibility that the authorities will realign by a certain amount with a particular probability. Given the foreign exchange market's assessment of this probability distribution, the mean expected depreciation can be calculated and hence the expected path for interest rates.

The mean expected depreciation of the sterling-D-Mark rate assumed for the first four years in our base forecast is given by the top row of Table A1. What type of probability distribution for expected depreciation might have generated this path? Clearly, the underlying probability distribution will not be unique since the expected path may, for example, be consistent with a large probability of a small depreciation or a small probability of a large one, or some combination of both. To illustrate the methodology, it is instructive to postulate one particular distribution which might plausibly have generated these expectations. This is done by using a probability tree which has been constructed to be consistent with our forecast path for the exchange rate and interest rates: see Table A2. It embodies the following sylised assumptions;

Table : A1. Expected exchange rate depreciation

 (in case of no realignments ex post after year 1)
Expectation formed 1 2 3 4

at beginning year

1 1.8 1.6 1.3 1.0
2 1.56 1.21 0.85
3 1.10 0.69
4 0.55

[Tabular Data Omitted] - in any year the probability distribution is bi-modal in the sense that the exchange rate can either stay fixed or

be realigned by a particular amount.

- if the authorities do choose to realign, they will devalue by 10 per cent.

- in year 1, the probability of a realignment is 18 per cent. For every year that the authorities are observed to

hold the exchange rate fixed (which it is assumed is done by concerted intervention within the ERM), the

probability of a future realignment declines, to under 10 per cent after four years.

- if the authorities do realign by 10 per cent, they are then strongly expected to allow the exchange rate to

depreciate gradually thereafter (by 2 per cent with probability 0.9)

- if the authorities do realign but then hold the exchange rate fixed, the expectation that they will not

depreciate again (i.e. their credibility) builds up gradually. The probability tree contains all the relevant information for calculating expected exchange rate movements. Each |branch' of the tree represents a possible outcome for the exchange rate in a particular year. The two branches at the top represent the alternative possibilities in the first year. The sixteen branches at the bottom of the tree represent the proliferation of possible outcomes by the fourth year. From this we can calculate the ex ante expected devaluation in the exchange rate for the first four years (which will be equal to the required interest rate differential) as expected at the beginning of the first year; see the top row of table A1. For example, the expected depreciation in year 2 as perceived at the beginning of year 1 is given by the probability of a realignment of 10 per cent, given that there was no realignment in the first year (0.82 X 0.156 X 10) plus the probability of a depreciation of 2 per cent in year 2 which would happen if there had been a realignment in year 1 (0.18 X 0.9 X 2). This sums to 1.6 per cent as given in table A1.

Evolving credibility: stylised example

Of course, in this particular example, the path implied by the central expectation will never occur so after one year expectations will have been wrong ex post. The probability tree then informs us how to revise our expectations of future exchange rate movements depending on which branch of the probability tree we have |travelled along'. To give a specific example of how expectations may evolve, suppose the authorities chose not to realign during the first four years. In our example, this happens to be the outcome which is most likely to occur, happening with probably 0.58. Table A1 illustrates how exchange rate expectations are sequentially revised over the future as the private sector effectively learns about the government's policy intentions. For example, at the beginning of the forecast period, the exchange rate is expected to decline by 1 per cent in the fourth year of the forecast. However, by the time the fourth year begins and the authorities have been observed to hold the exchange rate fixed, the expected depreciation for that year is then revised downwards to 0.55 per cent (see the fourth row of table A1). This may be interpreted as an increase in credibility of the authorities' fixed exchange rate commitment which will consequently allow interest rates to be cut in the fourth year by more than was envisaged at the beginning of the forecast period.

In fact, this example is of more than academic interest. By computing this path where credibility evolves gradually, we may be putting ourselves in the position of the Treasury forecasters who may know with certainty (for the sake of argument) that the exchange rate will not be realigned. This is a specific example of the general case where forecasters hold different expectations to the markets. The interesting corollary to this is that even if the Treasury were using the Institute model and had the same information regarding all variables, they should be producing a different forecast from ourselves because of their superior information on the true intended policy stance.

The evolution of credibility in the forecast - or - |A Treasury's Eye View'

Having illustrated the methodological issues involved in deriving an ex post forecast where the private sector learns about government behaviour, it is now possible to illustrate the implications of this for the forecast itself. As in the stylised example given above, it is interesting to illustrate the ex post forecast outcome if the government does not allow the sterling-D-Mark exchange rate to depreciate despite the fact that the market is discounting such a fall. As noted above, by doing this forecast variant we are perhaps putting ourselves in the position of Treasury forecasters who know that the exchange rate will not be realigned but who are confronted by a private sector which has a mean expectation of a non-zero depreciation (note that in our base forecast we are implicitly assuming that we have the same information about government intentions as the rest of the market).

To undertake this exercise on the forecast itself is slightly more complicated than the stylised example for a number of reasons;

- the updating of expectations is carried out every quarter. Rather than computing a probability tree on a

quarterly basis (which would have 65536 branches by the end of the fourth year) we adopt an updating rule

for the expected depreciation which approximately retains the same properties as that used in the stylised

example. Starting from the ex ante mean expected depreciation shown in chart A1, it is assumed that for

every period the government is observed to hold the sterling-D-Mark exchange rate flat, so the depreciation

expected thereafter will be progressively reduced relative to its base value until it has complete credibility by

the end of the decade. For example after two years holding the sterling-D-Mark rate fixed, the mean

expected depreciation for every period thereafter is assumed to be reduced by 25 per cent, whereas after

four years, it has been approximately halved (see chart A1). As the expected depreciation falls, so interest

rates decline in line.

- as exchange rate policy is defined with respect to the sterling-D-Mark rate, we need to re-calculate the

implied effect on the sterling-dollar and the sterling effective exchange rate every time that the sterling-D-Mark

expectation or outturn differs (note that the effective exchange rate will fall even when the D-Mark rate

is constant because of the assumed forecast appreciation of the dollar against the D-Mark).

- since the exercise is undertaken on a model of the whole economy we need to consider the expectations of

other parts of the private sector such as wage bargainers, as well as the foreign exchange markets. We

assume, as in the base forecast itself, that the expectations held by these different groups do not differ, that

is they perceive the same probability distribution of realignments (for the implications of making a different

assumption see Miller and Sutherland (1990)).

The actual implementation of this type of ex post forecast is also more complicated than the procedure used to derive the base case. This is because the usual consistent expectations assumption can not be adopted since expectations are continually being falsified. As a consequence, a sequential or rolling forecast procedure needs to be adopted. This technique is necessary whenever expectations are formed in a forward-looking manner and when the information set on which those expectations are conditioned alters unexpectedly within the forecast period; in this case the |news' which is obtained during the forecast horizon is that the exchange rate has not moved in line with expectations and hence that the subsequent expected path must change. Thus, the actual forecast procedure adopted is as follows;

(1) Start from the forecast base beginning in 91Q3. (2) Hold expectations constant but run the forecast with the sterling-D-Mark rate flat for 91Q3. (3) Update exchange rate expectations and the required interest rate profile from 91Q4 onwards, and

recompute the forecast from 91Q4 onwards with expectations set consistently again. (4) Repeat (2) but with the exchange rate flat in 91Q4. (5) and so on.

Since the expected depreciation is relatively small after 1995 with the expected convergence towards monetary union, the rolling forecast procedure is only carried out for four years. A fuller description of how this |stacked solution' technique is adopted in a stochastic simulations context, see Ireland and Westaway (1990).

We can now determine how much difference it makes to the forecast outcome if the government sticks to its exchange rate commitment and the private sector gradually perceives this. Charts A1-A3 compare the ex post outcomes of the sequential forecast, which we will refer to as variant A, with the original ex ante prediction, our base forecast. Period by period, variant A shows the sterling-D-Mark rate remaining constant for the first four years (chart A2) despite the original expectation that it would depreciate, as reflected in the short term interest rate differential (see chart A1). As a result, variant A displays ex post inconsistency of the exchange rate path with the interest rate differential over the first four years. Subsequently, expectational consistency does hold, but the expected depreciation after 95Q2 has been lessened because of the enhanced credibility of the authorities' exchange rate commitment. Chart A3 clearly demonstrates the advantages to the government of holding the exchange rate at its target level as in variant A. Inflation is some 1 1/2 per cent lower over the first four years as a direct consequence of lower import prices caused by the |unexpectedly' lower exchange rate. Indeed, since inflation undershoots European levels, it might appear that the authorities could afford, ex post, to relax the stance of policy as a |reward' for their commitment to exchange-rate stability. This case should not be overstated, however, since some degree of low inflation relative to our competitors may still be necessary to restore the real exchange rate to its sustainable equilibrium level (which in previous work we have argued is below prevailing levels, see Wren-Lewis et al, (1990)).

It is important to note that most of the gains in inflation performance in variant A occur because of the direct effect of the higher exchange rate. Interestingly, since the period-by-period revision to future depreciation is modest the |credibility' effects caused by gradual learning manifested in lower expectations of depreciation over the future seem relatively small. But this is not to say that the lack of credibility of policy in the ex ante forecast and in variant A is not costly. Chart A3 also illustrates how the inflation outcome would differ if the exchange rate policy of the authorities had been foreseen correctly. This is shown as variant B which shows an improved inflation outturn relative to Variant A since the implications of the higher exchange rate are anticipated by forward-looking agents in labour and product markets who revise down their inflation expectations accordingly. In fact, the direct expectational effect on inflation is even larger than this but that improvement is partly offset by the effects of lower interest rates which necessarily have to fall to German levels if the exchange rate target is completely credible; the lower interest rates also produce higher output than in variant A but this effect is relatively unimportant and short lived.

By quantifying the unequivocal superiority of variant B over variant A and even more so over the base case, this exercise illustrates clearly how important it is that the authorities should be able to persuade the markets of their intentions to hold the exchange rate fixed. It also casts light on the question of the validity of the so called |Walters critique' of UK membership of the ERM which suggested that the policy could not be sustained because of the disruptive effects on domestic demand of lowering interest rates too quickly (see Walters (1990)). Our conclusion, which has been confirmed by recent experience of sterling within the ERM, suggests first that the |Walters critique' oversimplified the problem by ignoring the issue of evolving credibility. Second, it indicates that even if credibility of the authorities' ERM commitment were total the expectational effects on inflation would outweigh the potentially harmful effects of lower interest rates.

What are the main conclusions from this exercise? First, it illustrates clearly that forecasts of future exchange rate movements based on the mean expectation may be misleading for a forecaster who holds significantly different views about government intentions from the rest of the markets. Second, it has demonstrated the appropriate forecasting methodology in these circumstances and for the more general case where expectations evolve during the forecast horizon. Finally, and perhaps most importantly, it has illustrated that consideration of the evolution of the credibility of sterling within the ERM has an important bearing on the immediate prospects for the UK economy. In particular, it shows that the forecast path for inflation will be lower than we are currently predicting if the government manages to hold the exchange rate around its current level rather than depreciating in line with market expectations. Moreover, this inflation outcome will be improved still further if both wage bargainers and foreign exchange operators can be persuaded of the government's resolve to stick to this policy.

REFERENCES

Ireland, J. and Westaway, P.F., (1990) |Stochastic simulation and forecast uncertainty in a forward-looking model', National

Institute Discussion Paper No. 183 Miller, M. and Sutherland, A. (1990) |The |Walters critique' of the EMS: A Case of Inconsistent expectations', CEPR

Discussion Paper No. 480. Walters, A. (1990) |Sterling in Danger', London, Fontana. Wren-Lewis, S. Westaway, P., Soteri, S. and Barrell, R., (1990) |Choosing the rate: an analysis of the optimum level of entry

for sterling in the ERM', National Institute Discussion Paper no. 171, also Manchester School forthcoming.

PHOTO : Chart 1. UK export of manufactures

PHOTO : Chart 2. UK exchange rates

PHOTO : Chart 3. UK and German interest rates

PHOTO : Chart A1. UK - German short-term interest-rate differential

PHOTO : Chart A2. Sterling-D-Mark exchange rate

PHOTO : Chart A3. Annual price inflation (consumer price index)