Has the North-South divide come to an end? - Prospects for regional unemployment.

Wilkinson, David

Introduction

The large increase in unemployment in the last two years and the very different patterns of regional unemployment growth as compared to the last recession have brought about renewed interest in the distribution of unemployment. In the last two years the variation in regional unemployment rates has diminished significantly such that by August of this year all regions in Great Britain had an unemployment rate somewhere between the 11.4 percent in the North and the 7.8 per cent in East Anglia. Ignoring these two outliers, the remaining eight regions in Great Britain all had unemployment rates between 8.9 and 10.8 per cent.

This represents remarkable convergence of unemployment rates in a relatively short period of time. In March 1990, when unemployment was just beginning to rise, the regional spread of unemployment was much greater. Both the North and Scotland had an unemployment rate in excess of 8 per cent, whilst the South East and East Anglia had unemployment well below 4 per cent. It is the South East that has seen the most rapid increase in unemployment now standing at more than 2 1/2 times its level of March 1990, whilst the increase nationwide has been around 75 per cent.

This note serves to introduce a new model of regional unemployment which has been developed at the National Institute over the last year in order to forecast claimant unemployment in the eleven standard regions of the UK. The underlying properties of the model will be discussed in some detail, although for a more thorough account of the model there will be a more technical summary in Gregg and Wilkinson (1992).

The note develops as follows. Section 1 considers the current position and recent trends in regional unemployment growth. Section 2 lays out the modelling strategy to be used and looks at the model in more detail. Section 3 looks at some simulations using the model, of policy options following the sterling devaluation, and in doing so demonstrates the main properties of the model. Finally, section 4 looks at the forecast for regional unemployment consistent with the forecast presented in the UK Economy section of this Review, and considers the prospects of recovery in the eleven standard regions of the UK.

1. The current position and recent trends in regional unemployment growth

The latest recession has been markedly different from the recession of the early eighties in terms of the regional dispersion of unemployment growth. Then, rises in unemployment were far more evenly spread across the country, whilst the growth in unemployment in the last two years has been predominantly in the South East, East Anglia and South West. Unemployment in the North of England, Scotland and Northern Ireland, although remaining high, has showed relatively moderate increases since 1990.

Table 1.1 shows unemployment rates across all regions of the UK in July 1986, when UK unemployment was at its peak at 3.1 million; March 1990, when it was at its trough of 1.6 million; and the current position in August of this year when unemployment is a little over 2.8 million. If we ignore Northern Ireland, which historically has an unusually high unemployment rate, and consider Great Britain, then the current variation in regional unemployment rates is significantly lower than both 2 1/2 years ago and six years ago.

The lowest rate remains in East Anglia, but the South East has shown a significant increase in unemployment relative to the rest of the UK. In July 1986 and March 1990, Wales, Yorkshire and Humberside, the North West, the North and Scotland all had an unemployment rate well in excess of the UK rate, but by August 1992, Scotland and Wales have a rate below that of the UK and Yorkshire and Humberside has the same rate as the UK. The North and North West still have high unemployment, but the difference between their rates and the UK rate has declined.

Table 1.1 A comparison of regional unemployment rates July
1986, March 1990 and August 1992

per cent

 Unemployment

Region July 1986 March 1990 August 1992

South East 8.3 3.6 9.5
East Anglia 8.2 3.5 7.8
South West 9.6 4.1 9.4
West Midlands 12.7 5.8 10.8
East Midlands 10.0 4.9 8.9
Wales 14.0 6.4 9.8
Yorkshire & Humberside 12.6 6.8 9.9
North West 14.1 7.5 10.6
North 15.2 8.5 11.4
Scotland 13.4 8.2 9.6
Northern Ireland 17.7 14.1 14.9
UK 11.2 5.6 9.9
GB 11.0 5.4 9.8

Source: Department of Employment, Employment Gazette

Table 1.2 shows the growth in unemployment since March 1990. The South East, East Anglia and South West, have faced the greatest increase in unemployment, with the number of unemployed having more than doubled in the last 2 1/2 years. The East and West Midlands, which had the next lowest unemployment rates in 1990 have also faced relatively large increases in unemployment, both regions experiencing increases of around 80 per cent. The pattern continues through the regions. Wales and Yorkshire and Humberside faced an increase of around 50 per cent in unemployment since March 1990, while the remaining regions all have shown increases in unemployment well below 50 per cent.

The major increase in unemployment since March 1990 has clearly come in the regions which had the lowest unemployment rates at that time, and the most significant rise in unemployment has been in the South East. The South, in general, continues to face the largest rises in unemployment, and further convergence of regional unemployment rates is likely over the coming months.

2. The model

The model involves a three stage process with two adjustment mechanisms. Due to the complexities of the first two stages and the adjustment mechanisms, these steps are only undertaken for three super-regions. These regions are the South East, Central and Northern and are chosen so as to be of roughly equal size. Due to the size of the South East it is both a standard region and a super-region while the Central and Northern Regions both consist of five standard regions. Central comprises of East Anglia, the South West, East and West Midlands and Wales; whilst Northern comprises of Yorkshire and Humberside, the North West, the North, Scotland and Northern Ireland.

Table 1.2. Unemployment growth since March 1990

 thousands

 Unemployment

 March August Percentage
Region 1990 1992 increase

South East 339.3 868.6 156.0
East Anglia 34.5 78.9 128.7
South West 90.0 213.1 136.7
West Midlands 148.9 273.0 83.3
East Midlands 95.0 175.2 84.4
Wales 83.9 128.0 52.6
Yorks & Humberside 157.5 237.1 50.5
North West 232.8 324.8 39.5
North 119.7 157.4 31.5
Scotland 205.0 243.0 18.5
Northern Ireland 98.5 108.6 10.3
UK 1606.6 2807.5 74.7
GB 1508.1 2698.9 79.0

Source: Department of Employment, Employment Gazette.

Step 1 estimates regional employment for the three super-regions using the industrial disaggregation available in the National Institute Domestic Economic Model (NIMOD). Step 2 then derives regional unemployment estimates from the regional employment inputs above, regional population and several economy-wide unemployment determinants; and step 3 then disaggregates regional unemployment in the super-regions into unemployment for the eleven standard regions using a simple shift share approach. The two adjustment mechanisms considered are relative regional wages, which influences employment; and net inter-regional migration flows which clearly influence population and hence unemployment.

2.1. Employment

The first stage in our analysis is to estimate regional employment for the three super-regions identified above using the industry breakdown of employment in NIMOD. We consider four industry groups: Manufacturing, Distribution and Business Services, Public Services and a residual category which consists of Agriculture, Energy, Construction and Transport.

We jointly estimate regional employment, with each region showing different responses to changes in industry employment. To explain additional regional variation in employment, terms covering relative wages in the regions and aggregate personal tax and mortgage payments as a proportion of personal income are included in a way such that they do not violate the industrial/regional employment identity. The equation is estimated from the fourth quarter of 1974 up to the second quarter of 1991 and takes the following form:

|Delta~|E.sub.it~ = ||Sigma~.sub.j~||Alpha~.sub.ij~ |Delta~|E.sub.jt~

+ ||Sigma~.sub.j~ ||Beta~.sub.ij~ |E.sub.jt-1~ + ||Gamma~.sub.0~ |E.sub.it-1~

+ ||Delta~.sub.0~ |W.sub.it-1~ + ||Delta~.sub.i~ |Delta~ |W.sub.it-1~

+ ||Sigma~.sub.i~ M|P.sub.t-1~ + ||Mu~.sub.i~ T|X.sub.t-1~ (2-1)

with the following restrictions:

||Sigma~.sub.i~||Beta~.sub.ij~ = - ||Gamma~.sub.0~ for all j (2.2)

||Alpha~.sub.ij~ = -||Beta~.sub.ij~/||Gamma~.sub.o~ for all i and j (2.3)

||Sigma~.sub.i~||Sigma~.sub.i~ = 0 (2.4)

||Sigma~.sub.i~||Mu~.sub.i~ = 0 (2.5)

|E.sub.i~ is employment in region i, |E.sub.j~ is UK employment in industry j; with j corresponding to the four industry groups described above, |W.sub.i~ is relative wages in region i, MP is UK mortgage payments as a proportion of personal income, TX is UK personal taxation as a proportion of personal income and subscript t represents time.

The sum of industry employment is therefore constrained to equal the sum of regional employment, and the impact on regional employment of a change in industry employment, with fixed relative wages, mortgage and tax payments, is the same in both the short run and long run.

The mortgage and tax payments are designed to pick up deductions from income which might otherwise have been directed towards consumption and hence increase employment in a region. They are already built into the determination of aggregate employment hence we consider regionally specific variations in the effects of tax and mortgage payments on employment, which must by definition sum to zero across the regions, and hence the sum of the |Sigma~ and |Mu~ coefficients across the regions must be zero.

The model shows that lower taxes and lower interest rates, and hence mortgage payments, are both of greatest benefit to employment in the South East, and less beneficial to employment in the North. The Central Region benefits from lower taxes having a small, but positive effect on the distribution of regional employment, but lower interest rates imply that employment in Central Britain is lower than would be expected if we had just considered a simple nationwide effect.

In contrast, relative wages are not incorporated into the aggregate model, and hence are modelled as additional effects. The |delta~ coefficients are the same across the three regions, but this time it is the actual relative wages that sum to zero to maintain the industrial/regional employment identity.

The short run and long run impact of relative wages are of opposite sign. High wages in the short run will attract more labour supply into the region and raise spending power and hence increase regional employment. However, in the long run high wages may become prohibitively costly for firms and reduce competitiveness and make relocation elsewhere, where wages are relatively lower, more attractive, in that case regional employment will ultimately be lower.

Relative wages, mortgage and personal tax payments provide additional regional variation around the industrial composition effects on regional employment, but it will be the industry composition of employment that will be the main determinant of regional employment in the medium term. However, macro-economic policy will have profound implications for the distribution of employment in the short term.

In the past decade the most significant trends in industry employment have been the rapid increase in Distribution and Business Services and Public Services employment, and the continued decline of employment in manufacturing. From its trough in 1983 to its peak in 1990 total employment including self employment increased by nearly 3 million or 12.6 per cent. Table 2.1 shows that out of this 3 million, around two million was an increase in Distribution and Business Services employment, and a further million came in Public Services employment. In this climate of rapidly increasing employment there were falls in Manufacturing employment of around 350,000 and only a modest increase in other employment.

The majority of the increase in employment took place in the Central and South East regions, with a slightly smaller increase in the Northern super-region. The boom in Business Services and Distribution is then clearly associated with a boom in employment in the South East and Central Britain, and we may also expect the decline in manufacturing to have contributed to the smaller rises in employment in the Northern super-region.

The fall in employment since the middle of 1990 has also been quite rapid, but the industrial pattern of declining employment is quite different from the boom. In the 18 months to the end of 1991 total employment in the UK has fallen by just over a million. Again Table 2.1 indicates that the fall in Public Services employment has been virtually negligible, but the falls elsewhere have been quite evenly distributed. The most significant turn around has come in Business Services and Distribution; the industries with the fastest growing employment up to 1990, have faced a significant contraction in employment in the recession. Manufacturing employment continued to fall significantly throughout the decade, but the annual rate of decline has shown some increase during the recession.

The major contrast between this recession and the recession of the early 1980s is that employment in the South East has been the hardest hit. Well over a half the fall in employment has come from the South East and it seems that the jobs boom in the region from 1983 onwards has been reversed with many of the jobs created in this period now having disappeared. The Central Region has come out of the decade with a very rapid increase in employment between 1983 and 1990, and a relatively small contraction in employment thereafter. The Northern super-region has faced the smallest decline in employment, but in the preceeding period employment growth was much weaker than the other regions.

The model provides us with indications of how each region would respond to changes in the industrial composition of employment. We can calculate the long run TABULAR DATA OMITTED effect on each region of an increase in industry employment for any industry group. The results in Table 2.2 show the effects on regional employment of an increase in industry employment of 100 in any of the four industry groups. Clearly an increase in industry employment of 100 must be matched by a total increase of 100 in regional employment.

Increasing employment in Manufacturing will have the greatest effect on employment in the Northern regions, with around a half the increase occurring in these regions. There will also be quite a substantial effect on employment in the Central regions, but the effect on employment in the South East is relatively small.

Changes in employment in the other industry groups have a much greater effect on employment in the South East, with an increase of 100 in employment in Business Services and Distribution, Public Services or the other employment category increasing employment in the South East by 44, 52 and 37 respectively. Business Services and Distribution employment also has a relatively strong impact on employment in the Central regions, but only a weak effect in the North. Public Services and other employment have a stronger effect in the Northern regions and only a weak effect in the Central regions.

2.2 Unemployment

The estimation of regional unemployment has the same basis as that used in NIMOD to estimate UK unemployment. It is based on the contrasting ability of different employee types to claim benefit once they cease to be employed, (see Gregg 1992). The ability to claim benefit is likely to vary according to a person's past employment history, gender and other sources of income and wealth.

Table 2.2. The effects of an increase of 100 units in industry
employment on regional employment

 Industries

Regions Manu- Public Distribution Other
 facturing services and Business
 Services

South East 13 52 44 37
Central 37 20 40 21
Northern 50 28 16 41

Four employment categories are considered in the model: male employees, full-time female employees, part-time female employees and the self employed. Also included are regional population and the effects of Special Employment Measures (SEM's) and two variables associated with the introduction of Restart interviews. The equation is estimated from the third quarter of 1975 upto the end of 1990 and takes the following form:

|Delta~|U.sub.it~ = ||Alpha~.sub.1~ |Delta~|U.sub.it-1~ + ||Alpha~.sub.2i~ |P.sub.it-1~ + ||Sigma~.sub.j~||Beta~.sub.j~ |E.sub.ijt-1~

+ ||Beta~.sub.5~ |Delta~M|E.sub.it~ + ||Beta~.sub.6~ |Delta~F|E.sub.it~ + ||Gamma~.sub.1i~ |Delta~SE|M.sub.t~

+ ||Gamma~.sub.2i~ E|E.sub.it-1~ + ||Gamma~.sub.3i~ D86 (2.6)

where |U.sub.i~ and |P.sub.i~ are unemployment and population in region i, |E.sub.ij~ is employment of type j in region i with j corresponding to the four employment types described above. M|E.sub.i~ and F|E.sub.i~ are male and female employment in region i,(1) SEM is employment in special employment measures, consisting of the Enterprise Allowance Scheme, Job Release Scheme, Job Share Scheme, Employment Training and the Job Training Scheme. E|E.sub.i~ is the difference between current employment in region i and employment in the third quarter of 1986, but is zero prior to this. D86 is a dummy variable for the period from the third quarter of 1986 to the second quarter of 1987 increasing over this period from 0.25 to 1 in intervals of 0.25.

There is a more or less one for one relationship between changes in the claimant count and changes in male employment, a slightly weaker relationship with full-time female employment, and changes in part-time female employment and self employment have a very much smaller impact on the claimant count. We assume that the effect on the claimant count of changes in different employment categories is the same across all regions, so the |Beta~ coefficients in equation 2.6 are the same for all regions.

Two key features of labour market participation in recent years have been the large increases in self employment and female employment, in particular part-time female employment, and the decline in male employment. From 1983 to the peak in employment in the middle of 1990, employment rose by nearly 3 million. Table 2.3 shows that self employment increased by just over a million in this period and the number of female employees increased by around 1.8 million. The rise in male employees was negligible.

The subsequent decline in employment, however, has been mostly in male employment. Over a half of the 1.1 million fall in employment since the middle of 1990 has been male employees, whilst the decline in female employees has been quite modest. Out of a fall in female employment of around 290,000 nearly all has been full-time employment, with the number of part-time female employees remaining virtually unchanged.

The pattern of increasing female and part-time employment in economic booms and declining male employment in recessions is likely to lead to persistently high claimant unemployment. The eligibility to claim TABULAR DATA OMITTED benefit is generally much higher for men than women and for full-time rather than part-time employees and hence we see very rapid increases in claimant unemployment in recession and a relatively small decline in a boom.

Changes in regional population, with given employment, are found to have a positive influence on regional unemployment such that just over two fifths of any increase in population will end up on the claimant count.

Special Employment Measures and the effects of changes in benefit entitlement associated with the introduction of Restart interviews in 1986 may well be expected to have different impacts across the regions. The duration composition of unemployment is radically different across regions, and many of the above measures are specifically targetted at the long-term unemployed.

Restart interviews were introduced to improve the job search activities of the longer term unemployed, but also allow a more rigorous assessment of claimants' availability for work and hence eligibility for benefit (Philpott 1990). The latter effect will lead to a reduction in the numbers of registered unemployed in the form of a 'shake-out' of claimants, modelled here by the dummy variable D86, whilst the improvement in job search activities will produce a reduction in claimant unemployment relating to better knowledge of opportunities for jobs or training when they become available. It seems that Restart has raised the proportion of jobs being taken by claimants as opposed to non-claimants, and raised the numbers leaving the count without jobs to go to. This job search effect is modelled by an extra employment variable for the period after the introduction of Restart which indicates a closer relationship between changes in employment and changes in the claimant count. Improved job search activity needs to be matched by opportunities for employment or training, so the number of places on the various special employment measures also needs to be taken into account.

The impact of both Restart and Special Employment Measures is found to be significantly stronger in the Northern regions with a slightly weaker effect in the Central regions and quite a modest impact on the South East. This reflects the greater build up of long-term unemployed in the North, Midlands and Wales throughout the eighties and the targetting of SEMs and Restart at the long-term unemployed.

2.3. Relative wages

Relative wages across the regions are incorporated into the determination of regional employment in the model and have shown quite significant adjustments in the past decade. In this simple model changes in relative wages are determined by changes in relative regional unemployment rates. As unemployment in a region increases relative to the UK, then relative wages in that region will decline as high unemployment reduces wage pressure within that region (Jackman and Savouri 1991 and Layard, Nickell and Jackman 1991). Equation 2.7 is estimated from quarter 3 of 1973 to the second quarter of 1991 and has the following form:

|W.sub.it~ = ||Alpha~.sub.i~ + |Beta~ |W.sub.it-1~ + |Gamma~ |U.sub.it~ (2.7)

where |W.sub.it~ is relative wages in region i and |U.sub.i~ is the relative unemployment in region i, subscript t represents time.

In the early 1980s the Northern and Central regions faced large increases in unemployment compared to the South East which, was ultimately followed by a decline in relative wages in these regions, offset by an improvement in the South East. From 1981 to 1989 regional wages relative to the UK declined by around 4 per cent in the Central and Northern regions, from 6 and 4 per cent below the UK level in 1981 to 10 and 8 per cent below the UK in 1989.

More recently the opposite has occurred. The South East has experienced relatively large increases in unemployment in the last two years and we have already seen a stabilisation of relative wages in the South East at around 18 per cent above the UK average. It is expected that relative wages will ultimately decline in the South East as wage pressure declines with increased unemployment.

There is also a regional fixed effect incorporated into the model which implies that in the absence of any relative unemployment effects, relative wages will remain above the UK level in the South East and Northern regions, but below in the Central regions. This fixed effect picks up any unchanging regional characteristics that cause wages to vary across the regions.

2.4 Net migration flows

Net migration flows are very small compared to total population in each of the regions, hence like relative wage movements, the effects of migration flows take time to have any significant effect on regional population and hence regional unemployment. The Central regions which have experienced the largest number of net inward migrations in recent years, typically face a net flow of less than 10,000 per quarter. This is very small compared to a population of working age in excess of 10 million, but over a period of a decade these movements can be quite significant. A net inflow of 5,000 per quarter will increase population by 200,000 in ten years.

Migration flows are mainly determined by economic prospects across the regions (Pissarides and Wadsworth 1989, and Pissarides and McMaster 1990). Better employment prospects and higher wages in a region will attract more people into that region, hence we relate net inward migration flows to changes in relative wages and relative unemployment rates. Again there are other determinants of net migration flows which we are unable to model specifically and hence are incorporated into a fixed migration effect. This fixed migration effect implies net flows out of the South East and Northern regions into the Central regions.

The South East has remained the most prosperous region for many years, with relatively high wages and low unemployment, and hence has attracted significant net inflows up until the late 1980s. More recently, other factors like the high cost of living, pollution and over crowding have been dominant and net flows have tended to be out of the South East. The recent rises in unemployment, and the decline in relative wages are likely to increase the number of outflows over the next couple of years.

The Northern and Central regions have been less prosperous with low wages and particularly high unemployment in the Northern regions. Together with fixed net outflows, this has led to very large outflows from the Northern regions, although the outflow rate has declined more recently with a relative improvement in wages and the rate of unemployment. The Central regions have benefitted from better living conditions and have managed to attract a steady net flow of people into the region despite a decline in relative wages at the start of the 1980s. For the rest of the decade conditions have been stable with unemployment in line with the UK rate and no significant change in relative wages, hence net migrations have continued to flow into the Central regions.

2.5 The standard regions

The final step in the model is to breakdown the Northern and Central super-regions into their five standard regions which together with the South East make up the eleven standard regions of the UK. This is done using simple non-economic relationships relating the share of unemployment in each standard-region to the level and change in unemployment in the three super-regions.

High unemployment in the South East means that given unemployment in the Central Region, the share in the South West and East Anglia will be relatively high, and shares elsewhere will be low. This reflects geographical proximity, the reliance on similar industries in these regions as well as similar experiences in house price movements. Conversely, if unemployment in the Northern regions is high, then the share of Central unemployment in the South West and East Anglia will be low and the share in Wales and the East and West Midlands will be high. This is largely due to the Midlands and Wales being predominantly manufacturing centres similar to the Northern regions, whilst the South West and East Anglia rely more heavily on the service sector.

The Northern regions are found to be far more independent from the rest of the economy, with much weaker relationships existing between unemployment in the Northern regions and the South East and Central regions, than were found for the standard Central regions. The relationship between unemployment in the Midlands and Wales and the Northern regions is matched by a slightly weaker relationship between the North, North West and Scotland and the Central regions. In addition there is a relatively strong correlation between unemployment in Yorkshire and Humberside and unemployment in the South East, partly reflecting the reliance on similar industries as there has been quite a significant number of business relocations from the South East to parts of the Yorkshire and Humberside region.

Northern Ireland shows little in common with any of the other regions. It is isolated geographically, and a weak positive relationship with the South East most likely stems from the large presence of public sector employment in both regions. The unemployment rate in Northern Ireland has been consistently above that of any region in Great Britain, and it is very unlikely that this will change over the next decade.

3. Simulations

The best way to illustrate the workings of the model and to see the likely outcome of any shifts in policy is by way of simulation. The recent collapse of the value of sterling was a major shock to the economy, and any subsequent change in policy in response to it is likely to have a considerable impact on the forecast described below. For more detail on the impact of devaluation on the UK economy see Pain 1992.

The first simulation we shall consider is a cut in interest rates of 2 per cent below the base assumed in the forecast described in Table 10 in the UK economy chapter, we will then go on to consider the effects of a further 10 per cent devaluation of the pound. Devaluation and interest rate cuts are likely to have significant inflationary consequences, so we also consider two policy options directed at the control of inflationary pressure. First we consider a |pounds~2 billion a year cut in public expenditure and second a one point rise in the basic rate of taxation, also equivalent to around a |pounds~2 billion a year increase in personal taxation. All shocks take place from the fourth quarter of 1992.

3.1 The impact of a 2 point cut in interest rates

An interest rate reduction affects the economy via the exchange rate and the influence of real interest rates on both companies and household expenditure. Taking out the exchange rate effect, which we will consider below, we look at the direct effect on regional demand.

The main impact is to ease the problems of mortgage debt and reduce the cost of finance to industry. The reduction in mortgage interest payments will provide the greatest boost to the South East economy since the mortgage debt problem is most serious in this region. Around a half of the increase in employment generated by the increase in demand occurs in the South East and consequently the fall in unemployment relative to its base level is predominantly in the South East.

Lower interest rates also mean that the cost of borrowing to industry is reduced, providing a boost throughout the economy, but particularly to manufacturing. Any increase in manufacturing employment will be predominantly in the Northern regions, hence we also expect lower interest rates will mean significantly lower unemployment in the Northern regions. The Central regions will benefit from both of these effects, although the overall impact on unemployment is likely to be somewhat lower than experienced in both the North and South East.

3.2 The economic impact of a 10 per cent sterling devaluation

In contrast to an interest rate cut, the effects of a permanent 10 per cent devaluation in sterling against all currencies is quite rapid. In the first year there is a substantial reduction in unemployment, and by the end of the second year virtually the full effect has been felt.

Table 3.1. The impact of a 2 point cut in interest rates on
regional unemployment

 1993 1994 1995 1996

South East -19 -41 -53 -57
Central -7 -13 -20 -26
Northern -12 -27 -39 -46
UK -37 -81 -113 -130

All figures are year end change from base levels in thousands.

There is an initial boost to output concentrated in manufacturing, as the volume of exports increases significantly in the next year. Yet again this boost to manufacturing will have a major impact on employment in the Northern regions and hence falls in unemployment in the first couple of years are also concentrated in the North.

There will ultimately be knock on effects into all industries and a boost in confidence which could bring forward some investment expenditures which will be of benefit to all regions. By the end of 1995 the impact on manufacturing employment is beginning to dwindle and a more general boost to demand takes over. Therefore from 1996 onwards the falls in unemployment are far more evenly distributed across the regions.

3.3 The impact of a |pounds~2 billion a year cut in public expenditure in 1993

We can only consider a general cut in public expenditure, as there has not yet been a clear indication as to where specific cuts are likely to fall. Consequently, in the long term, the effect on unemployment is quite evenly distributed across the regions.

The main impact is to reduce public sector employment by around 100,000 from the middle of 1994. The model then predicts that in the long run most of this fall in employment will be in the South East and Northern regions, with more modest falls in the Central region. There will also be quite significant falls in manufacturing and other employment, as demand from the public sector for construction and major developments of infrastructure declines.

The regional impact of specifically targetted cuts may be somewhat different to the effects described above, with the emphasis more on pay restraint the South East may suffer from even larger falls in employment. However, a general spending cut, probably best thought of as a cut in expenditure on health and education, leads to quite an evenly distributed increase in unemployment across all regions in the long run.

Table 3.2. The impact of a 10 per cent sterling devaluation on
regional unemployment

 1993 1994 1995 1996

South East -23 -52 -66 -62
Central -24 -50 -58 -54
Northern -34 -78 -92 -81
UK -82 -179 -216 -196

All figures are year end change from base levels in thousands.
Table 3.3. The impact of a |pounds~2 billion a year cut in
public expenditure on regional unemployment

 1993 1994 1995 1996

South East +32 +49 +49 +46
Central +15 +21 +21 +21
Northern +26 +44 +46 +43
UK +73 +114 +116 +110

All figures are year end changes from base levels in thousands.

3.4 The impact of a 1 point rise in the basic rate of taxation

An increase in personal taxation will lead to a reduction in personal disposable income throughout the economy, but the impact is likely to be felt the most in regions with relatively high income. The model predicts that the reduction in disposable income will reduce employment predominantly in the South East with a more moderate impact on employment in the Central regions and very little effect in the Northern regions.

Higher taxes feed through to reduced consumption and ultimately lower employment, but there is likely to be some delay before the full effect is felt on unemployment. Comparing Tables 3.3 and 3.4 it is clear that a cut in public expenditure has a much larger impact on unemployment in the first few years than an equivalent increase in taxation. The longer term effects are more or less equal.

One difference between the spending cut and a rise in taxes is the impact on the regions. Cutting expenditure has a rapid effect on employment in the South East and Northern regions, whilst a tax rise has a gradually increasing effect on unemployment which is predominantly in the South East. The South East is worst hit under both measures, but the Northern regions remains relatively unaffected by an increase in taxation, whereas the Central regions are least affected by a cut in public expenditure.

Table 3.4. The impact of a 1 point rise in the basic rate of
taxation on regional unemployment

 1993 1994 1995 1996

South East +8 +18 +25 +30
Central +5 +12 +17 +21
Northern +2 +6 +12 +17
UK +15 +36 +54 +68

All figures are year end change from base levels in thousands.

4. The forecast

4.1 Unemployment in the UK

After a rapid decline in unemployment growth from April of last year we have just begun to experience a re-acceleration of unemployment growth in the third quarter of 1992. Rises of the order of 25 to 30,000 per month may be expected to continue for the remainder of the year and a peak in excess of 3 million is likely towards the end of 1993. The recent sterling devaluation and interest rate cuts will produce a major stimulus to the economy, but this is likely to be offset, at least in the short run, by a decline in business and consumer confidence which will make further large job losses likely in the next few months. Several major companies have announced significant job losses as they assess their labour requirements over the next couple of years and it is probable that unemployment growth will remain significant throughout 1993.

4.2 Unemployment in the regions

Unemployment is expected to rise well into 1993 in most regions of the UK, with an early downturn likely only in the Northern regions and Wales. The remaining Central regions and the South East will face rising unemployment possibly until the final quarter of 1993.

The recent devaluation of sterling will provide the greatest boost to the manufacturing sector of the economy, and as noted in section 2.1, about a half of any increase in manufacturing employment translates into an increase in employment in the Northern regions. The recent cut in interest rates will reduce the level of TABULAR DATA OMITTED mortgage payments, and ease the pressure on debt repayment. Given the higher level of mortgage debt in the South East, this is likely to provide a significant boost to consumption and ultimately employment in this region. Even with the reduction in interest rates we still expect the South East to face the largest rises in unemployment over the next twelve months and hence, as we can see from Chart 4.1, further convergence of regional unemployment rates is likely. The peak in South East unemployment is likely to be close to a million at the end of 1993 by which time the variation in unemployment rates will be at a minimum, with less than 1/2 a per cent covering the rates in the three regional groupings.

The problems of the South East are the problems of the economy as a whole amplified by the enormous level of debt and the dramatic fall in house prices in the region. Since the middle of 1989 house prices in the South East have fallen by around 30 per cent, compared to a 16 per cent fall in house prices nationally. The problems of the housing market have clearly delayed any upturn in the South East, and have severely hampered recovery throughout the UK.

The large fall in house prices has led to a huge number of households being in possession of properties whose value is below the value of their mortgage. This climate has led many people to reduce their debts rather than increase expenditure. The situation is likely to deteriorate for some time, particularly in the South, and further postponement of increased consumption is likely until households feel more comfortable with their level of debt. Compounding these problems is a fear of job losses leading to a vicious spiral of increasing unemployment, increased repossessions, falling house prices and increased debt to asset value ratios.

East Anglia and the South West have along with the South East experienced the largest falls in house prices in the last three years, and have suffered greatly from the problems of mortgage debt. These regions are likely to face further problems, and an economic upturn is most likely to follow on from a recovery in the South East. This upturn is not expected until the middle of next year and hence the South West and East Anglia are likely to face rising unemployment throughout 1993. They will be the last two regions to experience a downturn in unemployment, possibly slightly later than the South East, reflecting the need for a strong recovery in the South East to filter through to the other Southern regions.

In contrast, Wales looks to be one of the strongest regions at the moment, and along with the Northern standard region seems to be the region most likely to face an early downturn in unemployment. Wales has benefitted enormously from a large number of business relocations in the last few years, with many firms trying to avoid the high wage and office costs in London and the South East, and being attracted by a highly skilled labour force, relatively low wages and relatively cheap office space.

The East and West Midlands cover the middle ground of the Central group of regions, and recovery in these two regions is likely to follow the timing of an overall recovery in the UK, with a downturn in unemployment expected towards the end of 1993. The West Midlands has a similar manufacturing base to the Northern English regions and hence may experience a slightly earlier turn around, whilst the East Midlands has experienced quite large falls in house prices, hence recovery may be slightly delayed. However, both regions are expected to face some decline in unemployment before the end of 1993.

The Northern regions have so far survived the recession much better than the rest of the UK. Only moderate rises in unemployment are expected in the next few months and a downturn in unemployment is expected in most regions in the first half of next year. House prices have remained relatively buoyant in most regions, with only Yorkshire and Humberside facing lower house prices than two years ago. Modest price falls have prevailed throughout the Northern English regions in the last year, but quite strong price increases have persisted in Scotland and Northern Ireland. The problem of mortgage debt has therefore had little or no impact in most Northern regions and hence contributes to the expectation of a much earlier recovery than experienced elsewhere.

The North and North West are expected to join Wales in facing the first falls in unemployment early in 1993, with a downturn in Yorkshire and Humberside following soon after. The North and North West are probably already in the initial stages of recovery and both regions will undoubtably benefit from the improved competitiveness of British goods abroad following the recent sterling devaluation. Yorkshire and Humberside will continue to be hit by a decline in house prices, hence recovery in this region is likely to be more muted as some delay in increased expenditure is inevitable as households continue to reduce their holdings of debt.

Scotland is beginning to be hit relatively hard by the second wave of the recession and has faced a recent re-acceleration of unemployment growth. Until quite recently increases in unemployment in Scotland had been very modest. The unemployment rate has increased very slowly such that Scotland now has a rate below that of the UK, whereas at the start of the recession the unemployment rate in Scotland was the second highest in Great Britain. Although Scotland is now one of the strongest regions in the UK, it is still likely to face rising unemployment for much of 1993, but the unemployment rate in Scotland is expected to remain significantly below that of the UK.

Northern Ireland has, not surprisingly, been found to be a relatively independent region, but it is still expected to rely on quite a strong recovery elsewhere before being able to undergo a sustainable recovery. It has faced by far the smallest percentage increase in unemployment in the last couple of years, and only quite modest increases are expected over the next few months. A turn around in unemployment is likely towards the middle of 1993, probably following soon after a downturn in the other Northern regions.

4.3 The longer-term outlook

The initial recovery is expected to come from the Northern English regions and Wales and is probably already under way. A stronger and more sustained turnaround is expected in the South from the middle of next year. As interest rates come down, the pressures on debt repayment will be eased and ultimately this will lead to higher consumption.

Low interest rates mean that the fall in unemployment from the end of 1993 is likely to be largest in the South East. Combined with the expected growth in employment, particularly in Business Services and Distribution, the prospects for unemployment in the second half of the decade are rather better in the South East than the other regions. By the end of the decade unemployment is likely to fall to around 2.5 million in the UK. Nearly a half of the fall will be from the South East.

Conclusions

Unemployment is expected to rise above 3 million in 1993 with the largest rises likely to be in the South East. The Northern regions, in contrast, will face a relatively early turn round with recovery already in progress in some regions and only moderate rises in unemployment expected in the next six months. The Central regions are expected to follow the middle ground broadly reflecting the recovery experienced in the UK as a whole.

In the medium term the South East is expected to experience the largest falls in unemployment, although unemployment will remain high in all regions. It no longer seems a question of there being a North-South divide in terms of regional unemployment. But we must now consider high unemployment as a national problem with deep seated roots.

NOTE

(1) The various employment categories for the three regions are not estimated, but are approximated by weighting total regional employment by the proportion of national employment in each category. For example, male employment in region i, M|E.sub.i~, is approximated by total employment in region i, |E.sub.i~, multiplied by the proportion of male employment in the UK, PME, i.e. M|E.sub.i~ = |E.sub.i~.PME.

REFERENCES

Gregg, P. (1992), 'Out for the Count--A social scientist's analysis of unemployment statistics in the UK', National Institute Discussion Paper No. 25.

Gregg, P. and Wilkinson, D. (1992), 'A regional model of unemployment: a technical report', (forthcoming).

Jackman, R. and Savouri, S. (1991), 'Regional Wage determination in Great Britain', CEP Discussion Paper No. 47.

Layard, R. Nickell, S. and Jackman, R. (1991), Unemployment: Macroeconomic Performance and the Labour Market, Oxford University Press.

Pain, N. (1992), 'The aftermath of 'Black Wednesday': the economic impact of the sterling devaluation and the base rate cut', National Institute Briefing Note No. 3.

Philpott, J. (1990), 'A solution to long-term unemployment: the job guarantee', London: Employment Institute.

Pissarides, C. and McMaster, I. (1990), 'Regional migration, wages and Unemployment: empirical evidence and implications for policy', Oxford Economic Papers No. 42 pp. 812-31.

Pissarides, C. and Wadsworth, J. (1989), 'Unemployment and the inter-regional mobility of labour', Economic Journal, vol. 99, pp. 739-55.