Transcript Inter-regional Mobility of Entrepreneurial

Inter-regional Mobility of
Entrepreneurial SMEs
James Foreman-Peck and Tom Nicholls
Welsh Institute for Research in Economics and Development , Cardiff University
URESG Cardiff 26 September 2013
Identifying entrepreneurial SMEs
• Entrepreneurial firms moving between early
innovative and mature product life cycle phases
are likely to be spatially mobile
• Expanding, well-managed enterprises will be
prone to move from high cost (congestion, wages
and rent) locations where they innovated to
• low cost (smaller, less agglomerated) places
suitable for standardised production
• The spatial pattern: from core to periphery or v.v.?
• Spatial consequences: ‘trickle out’ or boosting
regional disparities?
Data set
• Business Structure Database (BSD) which is intended to
identify all but the very smallest enterprises.
• Consequently included in each year chosen for study
here are about two million cases (Barnes and Martin
2002).
• Our selection from this data set consists of all
enterprises employing fewer than 250 persons.
• These enterprises are the smallest combination of legal
units that is an organisation producing goods or
services and that benefits from ‘a certain degree of
autonomy in decision-making’ (ONS 2006, p. 7).
Relative Labour Productivity Index (RLP)
• Turnover as the output measure (in the absence of firm level
price deflators).
• To allow comparisons across different industrial sectors,
estimate each SME’s productivity relative to the industry
mean.
• This removes industry-specific factors -differences in the
capital stock and bought in materials.
• An index number greater than one indicates higher than the
industry average productivity. For enterprise i in industry j,
and where LP is labour productivity and ALP is average
industry labour productivity;
• RLPij = LPij / ALPj
• Industries are defined at the 3-digit level of the UK SIC 1992
classification
• We also construct an industry-relative enterprise-level
employment index (rsizeE) as a measure of size.
Facts
• Inter-regional moves (affected 1.5% of SMEs 2004-6)
are a small proportion of SME exits (23% )
• London the largest exporter- 2.9% net c 1.3% (The
industrial core region exports employment to the
periphery:Hypothesis )
• Wales and Scotland the smallest
• All regions except London and North West were net
importers
• ‘Neighbour’ effects; SMEs appear most likely to move
to regions that are adjacent. From Wales SMEs tend to move
to the West Midlands and North West
Table 1 - SME relocation and exit frequencies by UK region 2004-6
Region
2004
Wales
N. East
York. &
Hum.
N. West
W. Mid.
E. Mid.
S. West
Scot.
E. Eng.
S. East
London
Total
Relocating
(2006)
%
relocating
(2006)
Net
*
imports
(2006)
Net
imports
as a % of
2004
total
67,453
643
0.75%
372
37,777
431
0.86%
107,658
1,366
144,796
No
change
or exit
(2006)
Exit
(2006)
%
exiting
(2006)
Total
(2004)
0.43%
17,565
20.5%
85,661
59
0.12%
11,909
23.8%
50,117
0.97%
77
0.05%
31,966
22.7%
140,990
1,787
0.93%
-130
-0.07%
44,542
23.3%
191,125
122,789
1,830
1.14%
116
0.07%
35,720
22.3%
160,339
101,855
1,891
1.41%
242
0.18%
30,375
22.6%
134,121
142,975
2,106
1.14%
857
0.46%
40,147
21.7%
185,228
101,316
575
0.44%
289
0.22%
29,474
22.4%
131,365
149,744
3,431
1.74%
617
0.31%
43,647
22.2%
196,822
243,532
5,503
1.69%
1,378
0.42%
75,874
23.4%
324,909
209,719
8,596
2.90%
-3,877
-1.31%
26.4%
296,611
1,429,614
28,159
1.48%
78,296
439,51
5
23.2%
1,897,288
*(Moving into region – moving out of region)
Source: ONS, authors’ calculations
Employment Expansion
• Entrepreneurial SME may be expected to be growing
rapidly and so rather larger than this average; hence table
limited to enterprises employing more than 10 in 2004.
• The growth of employment among this group between
2004-7 averaged just over 9 percent while that among
enterprises that relocated by 2006 was more than double,
an average of over 21 percent.
• Fastest growing movers originated in London and in the
North West. Can afford to expand employment by
relocating to lower wages and rents regions.
• Restricting the sample to those employing more than 20
the percentages are 8 for the region as a whole and 14.6 for
movers.
Table 3
Growth of Employment 2004-7 (Enterprises employing >10, 2004)
North East
North West
Yorks
East Midl
West Midl
East Engl
London
S East
S West
Wales
Scotland
N Ireland
Total
Total Region
Mean
0.1203
0.0912
0.0913
0.1545
0.0500
0.0582
0.1211
0.0895
0.0614
0.0536
0.1162
0.0620
SE
0.0145
0.0192
0.0129
0.0756
0.0107
0.0081
0.0162
0.0195
0.0093
0.0114
0.0247
0.0115
Relocators Out 2006
Mean
SE
0.0181
0.1940
1.9368
1.8811
0.0673
0.0914
0.0789
0.1038
0.1111
0.1135
-0.1046
0.0531
0.2785
0.1176
-0.0649
0.0585
0.0970
0.0991
0.0429
0.1529
-0.0240
0.0978
-0.7590
0.0933
0.0907
0.0075
0.2168
0.1377
Increased Relative Labour Productivity
• Performance of movers also on average is stronger (24
percent 2004-7) than among the region as a whole (16
percent).
• The geographical pattern is rather different from that
for employment growth, however.
• Relocators from London and the North West are no
longer the fastest, though they both grow more
strongly than stayers in those regions. Moving to lower
wage areas reduces profit-maximising labour
productivity (hypothesis?)
• For enterprises employing more than 20, movers
average 30 percent RLP growth and the regional
average is 18 percent.
Table 4 Relative Labour Productivity Growth 2004-7 (Enterprises with more 10 employees 2004)
North East
North West
Yorks
East Midl
West Midl
East Engl
London
S East
S West
Wales
Scotland
N Ireland
Total
Total Region
Mean
SE
0.1463
0.1654
0.1619
0.1460
0.1656
0.1461
0.1880
0.1512
0.1474
0.1697
0.1735
0.1656
0.0103
0.0061
0.0067
0.0069
0.0066
0.0065
0.0074
0.0056
0.0068
0.0097
0.0075
0.0103
0.1615
0.0021
Relocators Out 2006
Mean
SE
0.4134
0.1456
0.2111
0.0975
0.2151
0.0998
0.1406
0.1286
0.3258
0.1026
0.3151
0.0736
0.2260
0.0534
0.2938
0.0630
0.0804
0.1044
0.0774
0.2363
0.3536
0.1713
0.2108
0.6066
0.2453
0.0276
Relocation estimating equation
Pr is probability, Rt+1 = 1 if the firm has relocated in year t+1 (2006), Φ(.) is the
distribution (probit) function;
• Pr(Rt+1=1) = F(g0 Location t-1+ g1Aget + g2ln(RLPt-1) + g21 ln(RLPt-1)*ln(rsizeEt-1) +
g22(Locationt-1*ln(RLPt-1))+ g3(rsizeEt-1) +g31ln(rsizeEt-1)2 + g4(Legal form)t-1 +
g5(Number of plants)t-1 + g6Takeoverst-1 +61(Takeovert*ln(RLPt-1)) +
g7ln(Takeovert*Locationt-1) + g71(Takeovert*ln(RLPt-1)*Locationt-1) … (1)
•
•
•
•
•
•
•
Hypothesis 1 is the probability of relocation falls with enterprise age, g1 < 0.
Hypothesis 2a, mobility chances increase with productivity, is that g2 > 0,
supplemented by a specification that allows the productivity effect to vary with the
location and the size of the firm.
Hypothesis 3, g3 > 0, is that larger firms are more likely to be spatially mobile, but if
g31<0 the effect diminishes and may even be reversed as size (employment) increases.
Hypothesis 4 (g4) is that more personal forms of ownership make for lower mobility.
Hypothesis 5 is that partial mobility is more likely than full mobility so when the
number of plants in the enterprise exceeds 1 relocation chances increase, g5> 0.
Hypothesis 6 is that a takeover raises the chances of the firm subsequently moving, g6
> 0, also allowing that these chances vary with the productivity of the target (as for
instance predicted by Q theory (Jovanocic and Rousseau 2002).
Hypothesis 7, takeovers affect relocation differentially according to the target location,
is tested by whether g7 ≠ 0. Takeover effects also vary with target productivity.
N. East
-0.2693***
Table 5- Probit regression estimates of relocation
N. West
-0.2823***
York. & Hum.
-0.2332***
E. Mid.
-0.0887***
W. Mid.
-0.1891***
E. Eng.
-0.0390***
S. West
-0.1734***
Wales*ln(RLP)
0.0424***
Variable
Dep var relocation
Full sample
Takeover
0.1191***
Scot.*ln(RLP)
-0.0141
Ln(RLP)
0.0291***
N. East*ln(RLP)
0.0213
Ln(RLP)^2
0.0108***
N. West*ln(RLP)
0.0240**
York. & Hum.*ln(RLP)
0.0176
Age 2 to 4
0.0101
E. Mid.*ln(RLP)
0.0182*
Age 5 to 9
-0.0424***
W. Mid.*ln(RLP)
0.0159*
Age 10 to 19
-0.1345***
E. Eng.*ln(RLP)
0.0238***
Age 20+ years
-0.2684***
S. West*ln(RLP)
0.0096
Wales*takeover
0.3401**
Ln(local unit)
0.1007***
Scot.*takeover
0.5586***
Ln(remp)
0.0191***
N. East*takeover
0.4861***
Ln(remp)^2
-0.0060***
N. West*takeover
0.2223***
Company
0.2045***
York. & Hum.*takeover
0.1793**
E. Mid.*takeover
0.0454
Partnership
-0.0788***
W. Mid.*takeover
0.2568***
Wales
-0.2555***
E. Eng.*takeover
0.0930
Scot.
-0.5263***
S. West*takeover
0.2182**
Takeover*ln(RLP)
0.1191
Industry
Y
N
Pseudo R
1,897,288
2
Log-likelihood
0.05
-139,117
• Consistent with hypothesis 1 the coefficients indicate
that older SMEs are less likely to relocate. A 2-4 year
old enterprise is twice as likely to move as one that is
over twenty years old. Likely to be due to life cycle
effects or that newer firms have fewer local links,
ensuring that relocation is less costly
• Hypothesis 2a is confirmed by the positive and
significant coefficient on productivity (RLP) but to
establish the full productivity impact on relocation the
interactions terms g21 and g22 must be included.
probability of relocation is around 0.4 to 1.8 percent
over the three year period considered, depending on
location and productivity.
• Relocation chances rise most strongly with productivity
for firms in the East of England. Elsewhere the effect of
productivity is rather mild.
• SMEs that are based either in London or the South East are
most likely to relocate (including migration between these
two regions). This is indicated by the negative coefficient for
the region variables and is due to the effects of the
firm/plant/product life cycle and the ‘core’ position of these
regions.
• Regions with the next highest probability of relocation are the
East Midlands and East of England. The region with the lowest
predicted probability of relocation is Scotland at all levels of
productivity.
• Confirming hypothesis 3, size (employment) is positive and
significant in the model. The positive relationship between
relative SME size and the probability of relocation has no
tendency to diminish with enterprise size.
• Hypothesis 4, that more personal forms of business
organisation are less likely to move, is borne out. The legal
form coefficients of company and partnership are statistically
significant . As expected, companies have the highest
probability of relocating, followed by sole proprietors and
then partnerships.
• Hypothesis 5, that partial relocation is more likely than full
relocation for a firm is confirmed in Table 5. The number of plants
has a positive relationship with the probability of relocating.
• The takeover variable in the relocation model is significant and
positive (hypothesis 6).
• Some of the region-takeover variables are also significant and
positive (hypothesis 7) but the productivity-takeover interaction is
not. For all regions takeovers increase the chances of relocation .
• The largest effect is for SMEs in the North East (2.78 percent) and
Scotland (1.97 percent). The smallest marginal effect is for SMEs in
the East Midlands (0.45 percent), East England (0.49 percent) and
London and the South East (0.55 percent).
• Whereas a takeover in Wales triples the chances of relocation away
from Wales, a London takeover only increases the probability of
movement out of London by one third. This may be a source of
policy concern in the periphery. But takeovers also have other
impacts on productivity and survival chances of SMEs, which act in
the opposite direction to relocation effects for peripheral regions
(Foreman-Peck and Nicholls 2013).
‘Differences in differences’
equation for impact of
relocation on employment or productivity
• Where ∆ denotes the difference between the value in 2004 and
2007, and Y =employment or Relative Labour Productivity;
∆Y = β0 Relocation t-1+ β1Aget + β2ln(RLPt-1) + β22(RelocationtE
1*ln(RLPt-1))+ β3(rsize t-1) + β4(Legal form)t-1 + β5(Number of plants)t-1
+ β6Takeoverst-1 + β7ln(Relocationt*Employmentt-1) +
β71(Relocationt*Locationt-1) + β8 Location
…(2)
• For expanding firms we expect β0>0 and β0 + β22 ln(RLPt-1) + β7ln
Employmentt-1 + β71 Locationt-1>0, H2b.
• Also for growing SMEs, if London functions as the spatial core
region we expect positive contributions of SME net exports to
employment so long as neoclassical ‘trickle down’ is occurring,
• H8b; β71Relocationt+ β8 >0,
when Location indexes London location in 2004 and London differs
from the national economy as a whole, where relocation is
concerned.
Productivity
• Among SMEs with positive employment growth between 2004
and 2007 the direct effect for those that relocated was to
expand labour productivity by 8.5 percent more than those that
did not, in line with the lifecycle of the entrepreneurial firm.
• This conclusion must be qualified by the significant interaction
terms- with employment and with productivity; larger
employers that moved raised employment, though the opposite
effect is found for more productive SMEs.
• Ignoring the London effect for movers, at the (growing) sample
means the net effect of mobility is slightly larger at 10.2
percent, {(+.0822 +(.00194*5.4109) –(.0808*-.06292)) =) 0.098
and (100*(exp(.098)-1)=10.2.}
• But on average larger (employment increasing) firms – those
employing more than 20- raised productivity when relocating
by much more, by (100*(exp(0.258) -1)=) 29.4 percent (the
interaction terms were not statistically significant).
.
•
• Employment expansion proportionately bigger
over the range of firm size tested.
• Relocating SMEs that have positive employment
growth raise employment by (0.129→) 13.8
percent compared to growing firms that do not
move.
• Those with more than twenty employees create
(0.199→) 22 percent more jobs, while those
employing more than 50 enhance employment
by (0.301→) 35 percent apparently as a result of
relocating.
Relocation Effects on Employment and Labour
Productivity for SMEs Expanding Employment 2004-7:
OLS Regressions
Dep. Var.
Relocation06
Reloc* emp04
Reloc*RLP
Reloc*lond
Reloc*nw
N
adj. R-sq
Log Difference in Relative Labour
Productivity 2004-7 by initial size
All
Emp04>20
Emp04>50
(1)
0.0822***
(0.0106)
0.00194***
(0.000581)
-0.0808***
(-0.011)
-0.0908***
(-0.0148)
-0.0522
(-0.0273)
(2)
0.258**
(0.0886)
-0.00021
(-0.00096)
-0.0944
(-0.0569)
-0.221**
(-0.084)
-0.0407
(-0.195)
(3)
0.138
(0.187)
0.000399
(0.00145)
-0.235**
(-0.0871)
-0.321*
(-0.141)
-0.256
(-0.159)
699428
0.205
40687
0.225
14172
0.248
Log Difference in Employment 20042007 by initial size
All
Emp04>20
Emp04>50
(4)
(5)
(6)
0.129***
0.199***
0.301***
(0.0117)
(0.0451)
(0.0875)
0.000344 -0.000688
-0.00147
(0.000391) (-0.000511) (-0.000803)
-0.022
0.06
0.0645
(-0.012)
(0.0319)
(0.0602)
-0.0407*
0.137
0.123
(-0.0188)
(0.0745)
(0.124)
0.0023
-0.0216
0.0771
-0.0419
(-0.0907)
(0.145)
334209
0.159
25866
0.104
9197
0.095
Some specific regional effects
• London-based SMEs increased employment and
productivity relative to all other expanding firms except
Scotland’s .
• But contrary to the descriptive statistics, those Londonbased enterprises that moved did so by rather less.
• Nonetheless they boosted employment for the average
mover by (0.129-0.0407=0.088→) 9.2 percent, which in
view of the great volume of London’s SME net exports,
must constitute a significant contribution to reducing
regional disparities.
• This conclusion confirms hypothesis 8b
• SMEs in Wales raised relative productivity between 2004
and 2007 by less than any other UK region.
Contraction
• Just as mobility helps expansion it also permits or is
associated with stronger contraction.
• Among contracting firms those that move reduce
employment by more (though this effect cannot be
found for larger SMEs).
• Whereas cutting employment may be helpful for
profits, reducing productivity is not.
• Presumably mobility for smaller enterprises - the
majority - causes unanticipated problems of
downward adjustments
• whereas in fact for firms employing more than 50 there
is no statistically significant relocation effect on
productivity.
Relocation Effects on Employment and Labour
Productivity for SMEs Contracting Employment
2004-7: OLS Regressions
(1)
Dep. Var.
(2)
(3)
Log Difference in Employment 2004-7 by initial
size
(4)
(5)
(6)
Log Difference in Relative Labour Productivity2004-7 by
initial size
All
empl04>20
empl04>50
All
empl04>20
empl04>50,
-0.162***
-0.497***
-0.412
-0.167***
-0.281***
-0.0912
(-0.0119)
(-0.0875)
(-0.211)
(-0.0131)
(-0.0765)
(-0.0967)
-0.00330***
0.000298
-0.000094
-0.000863
0.00167*
0.0000335
(-0.000634)
(0.000854)
(-0.00138)
(-0.00047)
(0.000743)
(0.000792)
-0.0245**
-0.0244
-0.0192
-0.0434***
-0.0607
-0.033
(-0.00837)
(-0.0383)
(-0.0552)
(-0.013)
(-0.0446)
(-0.0819)
0.0816***
0.205*
0.248
0.0395*
-0.0934
0.0383
(0.0183)
(0.0984)
(0.166)
(0.0198)
(-0.116)
(0.119)
0.00794
0.229
0.065
0.0712
0.0386
0.0286
(0.0362)
(0.153)
(0.239)
(0.039)
(0.094)
(0.116)
N
244649
28655
10286
588016
21895
6643
adj. R-sq
0.068
0.171
0.213
0.071
0.101
0.109
Relocation06
Reloc*em04
Reloc* RLP
Reloc*lond
Reloc*nw
Some specific regional effects
• For the average contracting SME, moving out of
London is associated with employment and
productivity expansion, consistent with achieving
cost or other advantages from their new sites.
• In the North East of England and Wales
contracting SMEs reduced employment by the
least, other things being equal, suggesting
competition was less pressing in these peripheral
regions.
Conclusion
• SMEs that relocate are more productive, relatively larger and
younger.
• Spatially mobile enterprises are more probably registered as
companies, taken-over and located in London and the South East
(core locations).
• In addition these expanding businesses become more productive
and employ more workers after moving than other growing SMEs
that stay put.
• Relocation is also a strategy for contracting enterprises, but not
necessarily a helpful one.
• Among contracting firms those that move reduce employment and
productivity by more, though this effect only pertains to smaller
companies.
• Since London and the South East is the highest income and
productivity region, their net export of employment-expanding
SMEs must tend to reduce regional disparities
• Acknowledgments This research was supported by the
Economic and Social Research Council and Welsh Assembly
Government (Grant Number PTA-040-2006-00004). The
work contains statistical data from the Office of National
• Statistics (ONS), which is Crown copyright and reproduced
with the permission of the controller of Her Majesty’s
Stationery Office and Queen’s Printer for Scotland. The use
of the ONS statistical data in this work does not imply the
endorsement of the ONS in relation to the interpretation or
analysis of the statistical data. This work uses research data
sets which may not exactly reproduce National Statistics
aggregates.