School of Accounting

1
School of Accounting Economics
PATTERNS and TRENDS in UNEMPLOYMENT in
SCOTLAND 1986-2004 Research Results Scotecon
Grant Number scotecon/360603 John Adams,
Napier University Business School Ray Thomas,
The Open University January 2006
2
School of Accounting Economics
The problem with LAPU Long-term Unemployment As
a Percentage of Unemployment
At a time of rising unemployment the number of
short-term unemployed will be increasing, and
consequently, the percentage of long-term will be
decreasing. This might mistakenly be read as an
improving situation. ..... Conversely when
unemployment is falling the percentage of
long-term unemployment will increase if most of
the slack is taken by those recently out of
work. (from RSS Report on unemployment
statistics published in 1995).
3
School of Accounting Economics
ILO finds the same problem
Changes in the incidence of long-term
unemployment, however, often lag behind changes
in total unemployment. As unemployment increases
at the beginning of a recession, the heightened
flow into unemployment will result first in a
relatively smaller incidence of long-term
unemployed (denominator increasing as the
numerator holds steady). This will usually change
when the recession lengthens and when recovery
begins, as the short-term unemployed usually find
jobs more quickly, leaving behind an increasing
share of long-term unemployed.
http//www.ilo.org/public/english/employment/str
at/kilm/kilm10.htm
4
School of Accounting Economics
LAPU not best denominator

• ILO echoes the RSS position.
• But Eurostat, OECD, ONS all publish LAPU
  statistics
• A better denominator for year-or-more
  unemployment is the number at risk
• A better denominator for any duration group is
  the population at risk

5
School of Accounting Economics

PAR year-or-more unemployment rate PARtgt1
100 Ut gt1 /Ut-1 100 number unemployed for
year or more divided by number unemployed a year
earlier
PAR less-than-year rate PARtltY 100
UltY/E(T-1)T 100 numbers unemployed less than
1 year divided by number of entrants in previous
year

• PAR rates differ in kind from conventional
  unemployment rates
• PAR rates include unemployment in both numerator
  and denominator.
• The denominator of PAR rates is not population
  but time

6
School of Accounting Economics

• Structure of the Presentation
• Population at risk unemployment and exit from
  unemployment rates (1986-04)
• 2. Entry rates and unemployment rates (Period
  1996-2004)
• 3. What has happened since 1996
• 4. Conclusions and policy implications

7
Population at risk unemployment and exit from
unemployment rates
School of Accounting Economics
The first row of figures includes the numbers at
the date of the count of claimants in December
2004 in six three month long durations groups,
from 0-3 months to 15-18 months. The arrow
heads in this first row of figures indicate the
flows within the stock of unemployment over the
three month period October to December 2004.
8
Population at risk unemployment and exit from
unemployment rates
School of Accounting Economics
    Table 1 Numbers of unemployed and PAR exit
rates for duration groups for Parliamentary
Constituency Areas in Scotland in
2004        Duration group - months   Measure
0-3 3-6 6-9 9-12 Year or more   Average
numbers Dec 2004 585 239 118 65
180   Coefficient of Variation 36.2 42.9 47.3
52.0 28.1   Average exit rate Oct-Dec 2004
77.8 59.5 51.3 44.3 32.5   Coefficient of
Variation 9.1 13.1 15.9 22.6 16.8        
Exit rates do not provide the major explanation
of variation in unemployment rates per duration
group NOR do they Explain the spatial variations
in unemployment across Scotland, as we shall see
later.
9
Population at risk unemployment and exit from
unemployment rates (1986-2004)
School of Accounting Economics
Chart 2 shows that the range of variation in
unemployment exit rates is much less than the
range of variation in levels of unemployment.
Claimant unemployment in Scotland in the early
years of the Thatcher regime was well above 300
thousand and the seasonally adjusted series
peaked at 350 thousand in January 1986. By
December 2004 unemployment had fallen to under
100 thousand a quarter of its 1986 level. The
PAR exit rates shown in Chart 2 do not show
anything like that scale of variation. The
important implication is that variation in exit
rates cannot be expected to explain variation in
levels of unemployment. Variations in
unemployment rates, depend upon rates of entry to
unemployment as well as exit rates. This is the
KEY finding of this research ENTRY RATES have
been driving UNEMPLOYMENT RATES not exit
rates!
10
PCAs that have experienced high levels of entry
to unemployment have also experienced low levels
of exits. PCAs with low levels of entry have
experienced high levels of exits. The
geographical distribution of unemployment is
largely the product of the pattern of
distribution of entries to unemployment. But
that pattern is magnified, not mitigated, by the
geographical distribution of exits.
School of Accounting Economics
2. Entry rates and unemployment rates
It is evident that entry rates rather than exit
rates dominated the distribution of unemployment
in Scotland in 2004. Chart 3 confirms the
dominance of entry rates in illustrating a 0.92
correlation between entry rates and unemployment
rates.
11
School of Accounting Economics
2. Entry rates and unemployment rates
12
School of Accounting Economics
2. Entry rates and unemployment rates
It might be thought that the high level of
entrants to unemployment is directly associated
with high levels of exits that the major
explanation of geographical inequality in entry
rates is associated with a recycling of a
limited number of employment opportunities. But
the evidence given by Table 4 below does not
support this kind of explanation. Table 4 gives
statistics for 36 PCAs in the upper (18) and
lower (18) quartiles according to unemployment
rates and average rates of entry and exit.
Table 4 Unemployment rates, entry and exit rates
Upper Lower quartiles of PCAs in Scotland in
2004
UNEMP ENTRANTS EXITS
Only three PCAs in the upper quartile,
Cunninghame South, Greenock and Paisley
North, had exit rates of more than 300 (percent
of number unemployed). Only two PCAs in the
lower quartile had exit rates of less than 300.
13
School of Accounting Economics
3. What has happened between 1996 and 2004?
UK real GDP has grown by 25.8 percent in this
period Questions 1. Is there any evidence for
the convergence hypothesis in Scotland? (see
Chatterji, M., Oxford Review of Economic Policy,
1992) 2. Is there evidence of hysterisis
effects? 3. Do different areas have their own
natural rate of unemployment? (see Marston,
S., Quarterly Journal of Economics, 1985) 4. Is
there a steady-state inter-temporal equilibrium
unemployment rate which is IMMUNE
to macroeconomic policy and local LM policies?
(see Pehkonen, J. Tervo, H., Regional Studies,
1998) 5. Do current statistical systems support
our understanding of unemployment? 6. Do recent
(1996 onwards) and current policies CHANGE the
pattern of unemployment?
14
School of Accounting Economics
3. What has happened since 1996?
Chart 4 below relates entry rates and
unemployment rates in 1996 in the same way as
Chart 3 dealt with 2004. The correlation in 1996
of 0.850 is lower than that for 2004 of 0.917,
but the pattern in 1996 is remarkably similar to
that eight years later.
15
School of Accounting Economics
3. What has happened since 1996?
Both time-series and cross-section data show that
variation in exit from unemployment rates has
been substantially less than variation in
unemployment itself. Does this mean that entry to
unemployment has played the dominant role in
determining levels of unemployment? Table 3 shows
the distribution of unemployment, entrants to
unemployment and exits from unemployment
according to three standard measures of variation
or dispersion. All three measures are
independent of the average levels of
unemployment, and rates of entry and exit.
Table 3 Three measures of variation in the
distribution of unemployment rates and entry and
exit rates among 72 PCAs in Scotland (2004)
Unemp
Gini coefficient
16
School of Accounting Economics
3. What has happened since 1996?
17
School of Accounting Economics
3. What has happened since 1996?
A glaring contrast between two cities
Note the contrast between these two cities that
are only 64 km apart! The correlation between
the two in 2004 was 0.94!
18
School of Accounting Economics
Strong Hysterisis
Very Strong Hysterisis
ELASTICITY ENT/GDP M -.0641 MAX -0.811 MIN
-0.011
(6 WCS)
(19 WCS)
ELASTICITY ENT/GDP M 0.412 MAX 1.064 MIN
0.046
ELASTICITY ENT/GDP M -1.02 MAX -1.641 MIN
-0.553
(11 WCS)
Flexible?
19
School of Accounting Economics
3. What has happened since 1996?
AREA TURBULENCE INDEX (PCAS) entrants that
have changed across the 72 PCAs) (Normally
measured in terms of employment) see Layard, R.
et al (1990).
04 ½ ??? (NEi/NE)? 13.7
96

Interpretation Between 1996 and 2004 the
spatial distribution of entrants to unemployment
CHANGED by just under 14 percent 86 percent of
the SPATIAL STRUCTURE of entrants to unemployment
across Scotland remained CONSTANT This STRUCTURE
has been essentially unresponsive to BOTH UK real
GDP growth AND active LM policies at the local
level.
20
School of Accounting Economics
Although entrant turbulence has been small
(14) it appears MOST of this is accounted for
by only a FEW areas.
The strong rigidity in the Scottish labour market
identified in the 6 previous slides can also be
considered at a macroeconomic level as in the
next diagram
21
School of Accounting Economics
 
U                                                
           
U
DIFFERENTIAL NAIRUS? 10 PCAs gt 1SD gt M (96) are
included in the 13 PCAs gt 1SD gt M (04) 11 PCAs
gt 1SD lt M (96) are included in the 13 PCAs gt 1SD
lt M (04)
AD96
AD04
?U
?(Y/P) 25.8
A/U96 B/U04 C/U96 lt D/U04 C/A lt D/B
CONVERGENCE
13 PCAS
A
13 PCAS
B
gt1SD M
U96
C
NAIRUs
U04
?U
D
12 PCAS
gt1SD M
13 PCAS
Y/P
0
Y/P(96)
Y/P(04)
22
School of Accounting Economics
The STATIC Distribution of Unemployment Across
Scotlands PCAs
 
1996 12 25 21
13
2004 13 25 21 13
 
MEAN 1 SD
MEAN 1 SD
MEAN
TWO PUZZLES 1. WHY IS IT ALMOST A PERFECT NORMAL
DISTRIBUTION? 2. WHY HAS THIS DISTRIBUTION
REMAINED UNCHANGED OVER 9 YEARS (and possibly
for much longer)?
23
School of Accounting Economics
THE WORST AND THE BEST PCAs in terms of
ABSOLUTE UNEMPLOYMENT LEVELS
RATIOS 1996 2004 MEAN
WORST/ MEAN BEST 3.03 3.17 WORST/MEAN 1.82 1.67
BEST/MEAN 0.35 0.37 BEST/WORST 0.19 0.22 Note
The above ratios are totally consistent with the
diagram in SLIDE 21. The ratios are IDENTICAL if
expressed in terms of Unemployment Rates!
(WORST)
(WORST)
(BEST)
(BEST)
24
School of Accounting Economics
SHIFTING NAIRU Although the NAIRU has fallen
substantially over the period (UK wide) 19
of Scotlands PCAs REMAINED within 1 SD of the
1996 unemployment average. This is a level of
unemployment more consistent with a significantly
higher rate of inflation! It is also more
evidence of strong and persistent
hysterisis effects operating across areas
containing OVER 31 of the PWA!
INF RATE
?(W/P) 0 (1996-2004, HMT,NOV05)
P
U96
U04
1 SD
U04
0
Mean Unemployment per PCA
U04 (1,438)
U96 (2,475)
U04 (1,712)
19 PCAS
25
School of Accounting Economics
The Story in MAPS
26
MAP 1 Unemployment rates for PCAs in Scotland
in 2004
27
MAP 2 Entry to unemployment rates for PCAs in
Scotland in 2004
28
MAP 3 Unemployment rates for PCAs in Central
Scotland in 2004
29
School of Accounting Economics
4. Conclusions and policy implications
This research casts significant doubt on the
economic and social inclusion elements of the
Governments New Deal programmes. Although
UNEMPLOYMENT HAS FALLEN SUBSTANTIALLY across the
UK and in Scotland the original research on PAR
rates has led to the telling of another story.
Further research identified persistent and in
some areas even growing inequality in the
geographical distribution of unemployment, and
has revealed that the mainspring of the
persistence of inequalities and of growing
inequalities is the geographical pattern of entry
to unemployment.
30
MAP 4 Entry to unemployment rates for PCAs in
Central Scotland in 2004
31
MAP 5 Changes to unemployment rates for PCAs
in Scotland from 1996 to 2004
32
MAP 6 Changes to unemployment rates for PCAs
in central Scotland from 1996 to 2004
33
School of Accounting Economics
4. Conclusions and policy implications
The lack of attention given to statistics of
entry indicates that the support given by the
Office for National Statistics almost exclusively
to the ILO measure of unemployment is
counterproductive to the understanding and
analysis of labour market processes and the
development of labour market policies. It is
not reasonable to spend up to 10 million a year
on the LFS when that survey does not provide
basic information on the events that LEAD to
unemployment.
34
School of Accounting Economics
4. Conclusions and policy implications
The phenomenon of unemployment being maintained
by entry rates is evident in many urban and
several rural/suburban areas in Scotland, however
the greatest impact of this phenomenon is clearly
in the urban areas, in particular in the West of
the country. Policy needs to focus on stable
job creation and further research is needed into
the churning problem to determine whether
spatially differentiated entry to unemployment is
a matter of social inequality or a function of
the types of jobs certain areas attract - or are
given. We also need to know whether high
levels of entry to unemployment are related to
company recruitment and retention practices
and/or to the very nature of the types of jobs
that are available in certain areas.
35
School of Accounting Economics
4. Conclusions and policy implications
The greater influence of entry rates as compared
with exit rates suggests that in some parts of
Scotland the New Deal has been at best transitory
and at worst cosmetic. The above discussion
highlights a number of areas for future research
that we believe will prove fruitful in gaining a
deeper understanding of the dynamics of the
labour market both in Scotland and across the
United Kingdom.
36
School of Accounting Economics
Between 1996 and 2004 a TOTAL of 2,014,000 JOBS
have been added to the UK Economy
04
03
02
01
00
99
98
97
96
Scotland gained 5.25 of these but has 8.3 of
the PWA!
REAL GDP HAS RISEN BY 25.8
37
School of Accounting Economics
4. Conclusions and policy implications
It would appear that strong hysterisis effects
are present in many PCAs in Scotland, especially
in the West This suggests that persistence of
unemployment rates and even more so, entry rates,
is very high Different areas MAY be responding
to exogenous shocks differently (Very strong
hysterisis to none) The dispersion analysis
suggests that equilibrium unemployment rates
DIFFER between areas (Differential
NAIRUs) There is NO EVIDENCE whatsoever to
support the convergence hypothesis It MAY be
the case that different areas reach a
steady-state unemployment in the long run (the
natural rate differs between them!) The
steady-state (if it exists) represents an
inter-temporal equilibrium rate of unemployment
per area over LONG time periods (A Shifting Mean
at national level maintains the differentials
almost exactly at PCA level!) If it does exist,
this research suggests it is a function of entry
rate persistence, irrespective of macroeconomic
trends All of the above point to further
research needed into the pattern of employment
across Scotland and over time
38
School of Accounting Economics
4. Conclusions and policy implications

• The absolute persistence of ENTRY TO UNEMPLOYMENT
  in SOME areas suggests JOB DESTRUCTION in these
  areas
• Alternatively it suggests a total failure of
  policy intervention to keep people in a job
• Across Scotland the unemployment trends have
  been, almost everywhere a function of AGGREGATE
  DEMAND not of labour supply problems
• Long term unemployment RESPONDS to MACROECONOMIC
  turbulence differently in different areas - NOT
  to local labour market intervention!
• Policy should have focused on DEMAND SHIFTING JOB
  CREATION
• There is no evidence of convergence of
  unemployment across the economy
• It has also become ENDEMIC, INTER-GENERATIONAL
  and PERSISTENT amongst large groups in the
  population for so long now that the spatial
  imbalances might even be described as CULTURALLY
  EMBEDDED!
• CURRENT POLICY FRAMEWORKS ARE NOT ADDRESSING THE
  UNEMPLOYMENT PROBLEM IN SCOTLAND

39
School of Accounting Economics
BACK TO OUR QUESTIONS 1. Is there any evidence
for the convergence hypothesis in Scotland?
NO 2. Is there evidence of hysterisis effects?
YES 3. Do different areas have their own
natural rate of unemployment? PROBABLY 4. Is
there a steady-state inter-temporal equilibrium
unemployment rate which is IMMUNE
to macroeconomic policy and local LM policies?
POSSIBLY 5. Do current statistical systems
support our understanding of unemployment and
employment? NO 6. Do recent and current policies
CHANGE the pattern of unemployment? NO