United Kingdom: Recent Economic Developments

This paper reviews economic developments in the United Kingdom during 1991–95. Following a brisk expansion in 1994, when the economy achieved a rare combination of above-trend output growth and historically low inflation, the pace of economic activity in the United Kingdom has moderated in the course of 1995. Real GDP growth picked up to an annual rate of 4 percent in the first half of 1994, moderating to 3½ percent in the second half. Real GDP growth slowed further to just more than 2 percent in the first half of 1995.

Abstract

This paper reviews economic developments in the United Kingdom during 1991–95. Following a brisk expansion in 1994, when the economy achieved a rare combination of above-trend output growth and historically low inflation, the pace of economic activity in the United Kingdom has moderated in the course of 1995. Real GDP growth picked up to an annual rate of 4 percent in the first half of 1994, moderating to 3½ percent in the second half. Real GDP growth slowed further to just more than 2 percent in the first half of 1995.

V. The Output Gap and Inflation 1/

This chapter focuses on the relationship between the output gap and the rate of inflation, and in particular in documenting the “speed limit” effect (i.e., the relationship between changes in the output gap and inflationary pressure).

1. Defining the “speed limit” effect

The conventional approach to explaining inflation in industrial countries relies on the concept of the output gap, namely, the extent of slack available in an economy as represented by the excess of potential capacity with respect to actual output. On this approach, a positive (negative) output gap--i.e., actual output above (below) potential--indicates an excess of aggregate demand (supply) and will tend to increase (decrease) the rate of inflation. An elaboration of this approach is to include also the change in the output gap as a determinant of inflation. Positive (negative) changes in the output gap represent a reduction (increase) in the degree of slack and that on its own will stimulate (subdue) inflationary pressures. Under this formulation the inflation rate can be represented by:

π=β1gap+β2Δgap+otherterms(3)

where π is the rate of inflation, gap is the output gap and Δgap is the change in the output gap. 2/

Isolating the two effects of the output gap on inflation, it is possible to define a trade-off between the gap level and its change such that, ceteris paribus, the inflation rate is equal to the contribution arising only from the other terms:

Δgap=β1β2gap.(4)

This trade-off defines a “speed limit”, that is, a change in the output gap such that the inflationary (deflationary) effects from the gap change are fully offset by the deflationary (inflationary) effects from the gap level term. In order to calculate the “speed limit”, inflation equations were estimated.

2. Estimating the inflation equations

An inflation equation equivalent to equation (1), was estimated. The dependent variable was the RPIX (denoted as PX), estimated as a function of the output gap (YGAP), import prices (PM) and unit labor costs (ULC). 1/ Quarterly data for the period 1982-94 were used in estimating the inflation equations.

Before estimating the inflation equations, the time-series properties of the different series were studied for the 1976-94 period. Dickey-Fuller (DF) and Augmented Dickey-Fuller (ADF) tests confirmed that the log of RPIX, import prices and unit labor costs contain a unit root (Table 5.1.) The first differences of these variables appear to be stationary, although ADF tests ran with two and three lags did not reject the null hypothesis of a unit root in the log difference of unit labor costs. At the same time, the ADF unit root test (with three lags) corroborated what is true by construction, namely that the output gap is stationary in levels.

Table 5.1.

Unit Root Tests, 1976-94.

article image

The fact that the three price indices appear integrated of order one leads to the natural possibility that they are cointegrated. The existence of a long-nan relationship between retail prices, import prices and unit labor costs was the basis for the static equation in an earlier model. 2/ This issue is explored in Table 5.2, where the Johansen maximum likelihood method is pursued. F-tests on retained regressors suggested that the suitable lag-length was only two lags for the system. Both the trace statistic and the maximum eigenvalue statistic rejected the null hypothesis of a zero rank for the long-run matrix, but failed to reject the null of a unitary rank--indicating a single cointegrating vector for the system.

Looking at the first eigenvector--standardized on LPX--suggests the possibility that the coefficients on import prices and unit labor costs add up to one. In other words, the system displays static homogeneity in prices. This restriction was imposed and tested for in the lower panel of Table 5.2, after imposing a unitary rank for the long-run matrix. A likelihood-ratio test did not reject the imposed restriction of static homogeneity (x2 of 2.6). Interestingly enough, the share corresponding to import prices in the cointegrating vector (0.286) is estimated to be substantially higher than previous estimations. 3/ On the basis of this restricted cointegrated vector, an error correction was calculated (and denoted by ecpx) to be included in the estimation of the inflation equations.

Table 5.2.

Cointegration Analysis, 1975-94

article image

Two methods were used in the estimation of the inflation equations: Recursive Least Squares and Recursive Instrumental Variables. The latter takes into account possible endogeneity of right-hand-side variables. In particular (as in the earlier version of the model), unit labor costs were instrumented with a mismatch variable based on the ratio of vacancies to unemployment (MMV), retail prices (P), average underlying earnings (W), the unemployment rate (U), whole-economy productivity (PR) and lagged unit labor costs. After numerous specification searches, the final version of the estimated inflation equations for 1982-94 are presented in Table 5.3.

Table 5.3.

Inflation Equations, 1982-94 Dependent variable: D4LPX

article image

Overall, the inflation equation estimated by RLS appears quite satisfactory: all the variables have the expected signs and most of them are significant at conventional significance levels. Moreover, the fit of the equation is remarkable, as verified by the large R2 and the plot of the actual and fitted values (Chart 5.1). All the diagnosis tests suggest that the specification is rather acceptable (displaying perhaps some autoregressive conditional heteroskedasticity). The RIV estimation yields roughly similar results, although the specification test suggests that the structural model may not encompass the unrestricted reduced form.

CHART 5.1
CHART 5.1

UNITED KINGDOM INFLATION EQUATION, 1982-94

RLS Estimation

Citation: IMF Staff Country Reports 1995, 130; 10.5089/9781451814057.002.A005

Source: Staff calculations.

The recursive estimates of the RLS coefficients throw additional information, especially with respect to the period 1990-92, which corresponds to the period during which sterling was part of the ERM (Chart 5.2). Although almost all coefficients ended-up significant when the whole 1982-94 sample period was used in their estimation, during 1990 most of the coefficients showed marked instability in their estimated values. For example, the coefficient corresponding to the lagged inflation rate jumped from around 0.4 to 0.5, and similar behavior is observed in the case of import prices. The coefficients on the error correction term, unit labor costs (lagged four periods), and the output gap change displayed lower estimates, while the one on the lagged value of the output gap increased its standard error to become insignificant.

CHART 5.2
CHART 5.2

UNITED KINGDOM RECURSIVE COEFFICIENTS 1/

Citation: IMF Staff Country Reports 1995, 130; 10.5089/9781451814057.002.A005

Source: Staff calculations.1/ RLS estimated coefficients plus/minus twice their standard errors.
CHART 5.3
CHART 5.3

UNITED KINGDOM RECURSIVE RESIDUALS AND CHOW TESTS 1/

Citation: IMF Staff Country Reports 1995, 130; 10.5089/9781451814057.002.A005

Source: Staff calculations.1/ RLS residuals and structural change F-tests.

The possibility of a structural break around 1990 is reinforced by the behavior of one-step recursive residuals and Chow tests (Chart 5.3). Therefore, the models were estimated again for the period 1982-89 (Table 5.4.). Not surprisingly, the models display a better fit and specification over this restricted time period. As a further check on the possible structural break after 1990, both estimated models were run to provide forecasts for the out-of-sample period 1990-94. The null hypothesis of no structural change in any parameter between the sample and forecast period is strongly rejected (x2 of 355.6). The RLS model underpredicted the inflation rate consistently over the forecasting period, although curiously enough, from 1991 onwards the actual inflation rate lies within conventional forecast errors (Chart 5.4).

Table 5.4.

Inflation Equations, 1982-89 Dependent variable: D4LPX

article image
CHART 5.4
CHART 5.4

UNITED KINGDOM INFLATION EQUATION, 1982–89

RLS Estimation and Forecast 1/

Citation: IMF Staff Country Reports 1995, 130; 10.5089/9781451814057.002.A005

Source: Staff calculations.1/ Shaded area indicates forecast plus/minus twice the forecast error.

3. Estimating the “speed limit”

Given the estimations presented above, it is possible to calculate the “speed limit” effect according to equation (2). The “speed limit” defined in terms of the change in the output gap is conditioned on a given amount of slack and, thus, will vary along the business cycle. Table 5.5 presents the slope of the trade-off between the output gap change and the output gap level, calculated as in equation (2) as the ratio of the coefficient on the output gap level to the coefficient on the output gap change.

Table 5.5.

Estimated “Speed Limit” Trade-offs

article image

Although these calculations refer to point estimates, it is interesting to note that there is some variation in the implied trade-offs. The flattest estimate corresponds to the RLS estimation for 1989-94 (a slope of -0.75) while the highest estimate was obtained for the RIV for the same sample period (a slope of -0.91). Allowing for the standard errors of the two coefficients involved in the calculation of this trade-off would obviously yield a less precise estimate of the “speed limit”. In this regard, it is important to notice that the coefficient on the lagged output gap exhibited a larger standard error than the coefficient on the output gap change. Moreover, as seen in Chart 5.2, the coefficient on the lagged output gap in the RLS estimation became statistically undistinguishable from zero during 1990-93, which in turn would also reduce the “speed limit” trade-off for this period. This is, of course, corroborated by the fact that the calculated trade-off for the RLS estimation on the 1982-94 period is smaller than the trade-off for 1982-89.

Subject to this and other caveats, the estimated “speed limit” trade-off corresponding to the RLS estimation for the 1982-94 period-- together with other iso-inflation lines--is graphed in Chart 5.5, where the yearly change in the output gap is plotted against the level of the output gap lagged one year. Along the negatively sloped line labeled “Inflation - 0” the inflationary impact of the output gap level is exactly offset by a matching change in the gap.

The evolution of the United Kingdom during 1994-95 is also shown in the chart. 1/ The economy was growing so briskly during the first half of 1994 that, despite the sizeable output gap, the inflationary impact arising from changes in the output gap in all likelihood were close to surpassing the deflationary impact from the output gap level, thus being in the neighborhood of the “speed limit” in the second quarter. Since then the pace of expansion has moderated, with the economy experimenting a pause in the absorption of slack in mid-1995.

Chart 5.5
Chart 5.5

United Kingdom: Speed Limit Effect

(Gap change and gap level trade-off)

Citation: IMF Staff Country Reports 1995, 130; 10.5089/9781451814057.002.A005

Sources: CSO, Economic Trends; and staff estimates.

STATISTICAL APPENDIX

Table A1.

United Kingdom: Real Output and Its Major Components at Constant Factor Cost

(Percentage change over preceding year)

article image
Source: Central Statistical Office, Economic Trends.

Includes oil and gas extraction.

Based on output data.

Table A2.

United Kingdom: Labor Market Indicators

article image
Source: Department of Employment, Employment Gazette.

Not adjusted for seasonal variation.

Estimates of the self-employed, with or without employees, are based on labor force surveys for data through 1990, and on Department of Employment estimates thereafter.

Great Britain, percent of total.

Claimants basis.

Over 52 weeks.

Table A3.

United Kingdom: Selected National Accounts Aggregates at 1990 Market Prices

article image
Sources: Central Statistical Office, Economic Trends.

Half yearly and quarterly levels at seasonally adjusted annual rates or changes from a year ago.

An unweighted average of expenditure, income, and output estimates.

Contribution to growth of GDP (average estimate).

Average measure in billions of pounds.

Table A4.

United Kingdom: Selected Personal Sector Data

(In percent of GDP)

article image
Sources: CSO, Financial Statistics. Economic Trends: and staff estimates.

New dwellings, mortgage approved. Data for December of each year.

Table A5.

United Kingdom: Components of Personal Income

article image
Source: Central Statistical Office, United Kingdom National Accounts.

Half yearly and quarterly levels at seasonally adjusted annual rates or changes from a year ago.

In 1990 prices, deflated by the implied deflator for consumers’ expenditure.

Relative to personal disposable income.

Contribution to growth in disposable income.

Table A6.

United Kingdom: Selected Financial Statistics Industrial and Commercial Companies

(In percent of GDP)

article image
Source: CSO, Financial Statistics.

First two quarters at an annual rate.

Net of stock appreciation.

Bank borrowing and other loans less bank deposits.

Large companies.

Ratio of interest payments to post-tax income.

Table A7.

United Kingdom: Selected Indicators of Investment Activity

article image
Source: Central Statistical Office, United Kingdom National Accounts.

Half yearly and quarterly levels at seasonally adjusted annual rates or changes from a year ago.