Donald Barr, South West RDA/SWO Economy Module, delivers a presentation on how research can help support policy and enhance policy & the key questions researchers need to ask.
2. The rationale for intervening
to develop the economy:
The role of Research
• Look at how research has been used to make better
policy and strategy – the frameworks for intervention
• Consider how different types of evidence are used to
gain different insights
• Learn a bit about export propensity along the way
This session we will:
3. Start with a proposition…
“Exporting is good, the region
doesn’t export much…..
….so we should do something
to improve our exports”
4. “Exporting is good, the region doesn’t
export much: we should do something
to improve our exports”
Exporting
• Is exporting good?
If so why?
• How much does the
region export?
• How much could /
should the region
export and why
doesn’t it?
• What would you
need to do to
improve exports?
Generic
• Test the received wisdom – is it
really a good idea to start with?
• Examine the macro-economic
evidence – is there really a
problem?
• Look in more detail: If there is a
problem what are the underlying
causes?
• What, if anything, can be done
about the underlying causes?
5. What’s so good about exporting?
The literature review – what others have
already established. Exporters have:
• higher productivity
• higher R&D intensity
• higher propensity to innovate
• higher rates of revenue growth
In short they are more competitive but
there is a question of causality…..
6. Discussion 1
A
Not competitive and
not exporting
B
Competitive but
not exporting
C
Not competitive
but exporting
D
Competitive
and exporting
Groups B and C: what sort of firms? What
could be the policy implications?
7. Discussion 2
A
Not competitive and
not exporting
D
Competitive
and exporting
C
Not competitive
but exporting
B
Competitive but
not exporting
?
8. “Exporting is good, the region doesn’t
export much: we should do something
to improve our exports”
Exporting
• Is exporting good?
If so why?
• How much does the
region export?
• How much could /
should the region
export and why
doesn’t it?
• What would you
need to do to
improve exports?
Generic
• Test the received wisdom – is it
really a good idea to start with?
• Examine the macro-economic
evidence – is there really a
problem?
• Look in more detail: If there is a
problem what are the underlying
causes?
• What, if anything, can be done
about the underlying causes?
9. SW performance 1
Index of Exports as a proportion of GVA, 2005
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
North
East
East
Midlands
Wales South
East
East Northern
Ireland
North
West
West
Midlands
Yorkshire
and the
Humber
Scotland London South
West
IndexUK=100
Goods exports and proportion of total GVA:
bottom of the class
10. SW performance 2
Index of Exports as a proportion of goods GVA, 2004
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
London South
East
East North
East
Wales Northern
Ireland
North
West
East
Midlands
West
Midlands
Scotland Yorkshire
and the
Humber
South
West
IndexUK=100
Goods exports as a proportion of goods GVA:
still bottom
11. “Exporting is good, the region doesn’t
export much: we should do something
to improve our exports”
Exporting
• Is exporting good?
If so why?
• How much does the
region export?
• How much could /
should the region
export and why
doesn’t it?
• What would you
need to do to
improve exports?
Generic
• Test the received wisdom – is it
really a good idea to start with?
• Examine the macro-economic
evidence – is there really a
problem?
• Look in more detail: If there is a
problem what are the underlying
causes?
• What, if anything, can be done
about the underlying causes?
12. Where should we be?
3.23314159950.7876291669723824Sub Total
1.7717205300.96930855970288Scientific instruments and controls33
1.471092490.795923374521Furniture, fittings, temp buildings36
0.143223061.1025382232314216Other Transport Equipment35
23.3444546600.3159744190838Road Vehicles34
6.0179608070.172316171323582Electrical Machinery & Appliances.31
3.6318592700.36185830511735Telecomms & Sound Equipment.32
51.2985982230.1728676167630Office Machines & Adp Machines30
1.4620806610.578135361428684Machinery manufacture29
0.992686470.63170030270709Metal manufacture28
2.072465150.6375216119115Metals27
3.774684330.6884732124155Non-metallic materials26
0.923558020.59229281388687Plastics25
1.877785792.801168231416601Chemicals and Pharmaceuticals24
0.0522520.984527346151Electric Current40
4.221240174.0711938729369Fuels
10,23,1
1
0.211354120.52330346631492Textiles17
1.921136541.8811116459281Paper and pulp21
34.253953060.0782411541Cork and wood02,20
4.4411262750.1537088253475Leather goods, footwear18. 19
0.752953981.05409881391754Food and Live Animals0
FactorExports with
SW structure
and SE
propensity
FactorExports with SW
propensity and
SE structure
Actual
exports
Sector descriptionSIC
5432
1
3.23314159950.7876291669723824Sub Total
1.7717205300.96930855970288Scientific instruments and controls33
1.471092490.795923374521Furniture, fittings, temp buildings36
0.143223061.1025382232314216Other Transport Equipment35
23.3444546600.3159744190838Road Vehicles34
6.0179608070.172316171323582Electrical Machinery & Appliances.31
3.6318592700.36185830511735Telecomms & Sound Equipment.32
51.2985982230.1728676167630Office Machines & Adp Machines30
1.4620806610.578135361428684Machinery manufacture29
0.992686470.63170030270709Metal manufacture28
2.072465150.6375216119115Metals27
3.774684330.6884732124155Non-metallic materials26
0.923558020.59229281388687Plastics25
1.877785792.801168231416601Chemicals and Pharmaceuticals24
0.0522520.984527346151Electric Current40
4.221240174.0711938729369Fuels
10,23,1
1
0.211354120.52330346631492Textiles17
1.921136541.8811116459281Paper and pulp21
34.253953060.0782411541Cork and wood02,20
4.4411262750.1537088253475Leather goods, footwear18. 19
0.752953981.05409881391754Food and Live Animals0
FactorExports with
SW structure
and SE
propensity
FactorExports with SW
propensity and
SE structure
Actual
exports
Sector descriptionSIC
5432
1
SW with SE
structure?
Exports fall
by 22%
SW structure
with SE sector
propensity?
Exports treble!
13. Explaining the difference:
Export drivers
• Firm Size and Age,
• Skills Level
• Innovation and R&D
• Co-operation with Others
• Foreign Ownership,
• Capital Intensity
• Agglomeration Effects
• Industry
14. • Community Innovation Survey (CIS), waves 3
and 4
• Inter-Departmental Business Register (IDBR)
• Annual Business Inquiry (ABI)
• CapStock
Explaining the difference:
Data sources
15. Table 4.2: Probability of Exporting (CIS4)
1 2 3 4 5
Sample size 3603 3603 3603 3603 2789
South West
0.523
(0.080)***
0.644
(0.107)***
0.686
(0.114)**
0.712
(0.120)**
0.746
(0.151)
North West
0.668
(0.097)***
0.723
(0.114)**
0.819
(0.130)
0.832
(0.132)
0.895
(0.162)
Yorkshire and Humberside
0.602
(0.093)***
0.701
(0.114)**
0.777
(0.130)
0.780
(0.130)
0.851
(0.163)
North East
0.545
(0.105)***
0.673
(0.131)**
0.736
(0.143)
0.746
(0.147)
0.765
(0.173)
West Midlands
0.560
(0.087)***
0.668
(0.109)**
0.708
(0.119)**
0.699
(0.119)**
0.708
(0.137)*
Wales
0.494
(0.070)***
0.708
(0.107)**
0.754
(0.115)*
0.788
(0.122)
0.795
(0.125)
South East
0.835
(0.098)
0.888
(0.109)
0.945
(0.121)
0.944
(0.122)
0.981
(0.161)
East Midlands
0.538
(0.090)***
0.713
(0.125)*
0.783
(0.139)
0.809
(0.147)
0.841
(0.176)
East
0.758
(0.164)
0.906
(0.203)
0.956
(0.218)
0.984
(0.225)
1.139
(0.292)
London benchmark
Manufacturing
5.080
(0.501)***
3.513
(0.384)***
4.668
(0.528)***
4.188
(0.481)***
3.965
(0.537)***
Construction
0.253
(0.056)***
0.207
(0.048)***
0.245
(0.060)***
0.241
(0.058)***
0.244
(0.073)***
Wholesale
1.006
(0.111)
1.192
(0.140)
1.498
(0.185)***
1.352
(0.171)**
1.448
(0.207)**
Hotels and catering
0.446
(0.088)***
0.354
(0.076)***
0.451
(0.105)***
0.465
(0.108)***
0.408
(0.118)***
Transport
1.095
(0.154)
0.771
(0.120)*
1.006
(0.163)
0.973
(0.159)
1.019
(0.199)
Other benchmark
Firm output –
1.169
(0.054)***
1.154
(0.056)***
1.093
(0.054)*
1.067
(0.061)
Employment –
1.060
(0.046)
1.064
(0.048)
1.011
(0.047)
0.971
(0.052)
Capital –
1.381
(0.047)***
1.343
(0.048)***
1.276
(0.045)***
1.250
(0.051)***
Proportion with degrees – –
1.020
(0.002)***
1.020
(0.002)***
1.020
(0.002)***
European owner – – –
1.430
(0.205)**
1.293
(0.214)
Japanese owner – – –
2.925
(1.188)***
3.009
(1.600)**
American owner – – –
1.804
(0.300)***
1.527
(0.286)**
Other foreign owner – – –
1.285
(0.327)
1.154
(0.300)
UK owner benchmark
Post millennium birth – – – –
0.606
(0.087)***
Innovation (product or process) – – – –
1.764
(0.172)***
Has website – – – –
1.290
(0.147)**
Economic potential – – – –
0.931
(0.071)
% of area that is rural – – – –
0.997
(0.002)*
Llunit
0.959
(0.027)
0.657
(0.030)***
0.694
(0.033)***
0.707
(0.034)***
0.699
(0.038)***
Pseudo R2
0.117 0.189 0.218 0.231 0.240
Hosmer-Lemeshow (Prob.) 1167.85** 3595.30 3648.44 3615.84 2821.99
Log pseudo-likelihood -2112.480 -1940.297 -1871.501 -1839.112 -1406.831
Exclusion of compound vars. (prob.) – – – – –
Notes: In all logistic estimates, the dependent variable is Exporter. Odds ratios are presented with their robust standard errors
presented in parentheses. *, ** and *** signify significance at the 10%, 5% and 1% level respectively. Constants omitted as per
ONS requirements. Source: ONS.
16. Micro economic findings
•Firm size (output and employment)
•Level of capital
•Proportion with degrees
•Nationality of ownership
•Innovation activity
These are the significant drivers of inter-
regional differences in export performance.
Once these are adjusted for the differences
cease to be statistically significant – there
is no special South West factor!
17. “Exporting is good, the region doesn’t
export much: we should do something
to improve our exports”
Exporting
• Is exporting good?
If so why?
• How much does the
region export?
• How much could /
should the region
export and why
doesn’t it?
• What would you
need to do to
improve exports?
Generic
• Test the received wisdom – is it
really a good idea to start with?
• Examine the macro-economic
evidence – is there really a
problem?
• Look in more detail: If there is a
problem what are the underlying
causes?
• What, if anything, can be done
about the underlying causes?
18. • This research has been used to make better policy
and strategy: it demonstrated just how poor the
region’s export performance really was and identified
the weakness to be largely those associated with our
relatively poor productivity.
• It was a good example of how different types of
evidence are used to gain different insights
• And I hope you discovered something about export
propensity along the way
CONCLUSIONS
Notas del editor
Continuing Shane’s theme of intervention to promote economic development but looking at a specific case.
CLICK 1 Specific interventions that promote economic development generally sit within broader strategy or set of strategies. In the past this has been a framework of strategies to delivery the Regional Economic Strategy (RES). In future interventions will sit within local or national strategies.
CLICK 2 The phrase “evidence-based” was the mantra of the previous administration and its prevalence suggested a shift away from something that was less evidence-based.
However, two points to make about things being evidence-based 1) that evidence only get you so far and it is the role of leadership and enterprise to make those calls where the evidence is not clear. The shift from regional to local is just such a call – the evidence is not over-whelming either way. 2) Evidence always has a role to play even in policies and strategies that are not clearly evidence based. For example the challenge pot approach by definition can only select from the proposals put forward so cannot claim to either be meeting the most acute need or finding the best possible solution. But the projects put forward will themselves be backed up by evidence.
CLICK 3
The relationship between evidence and strategy is a bit chicken-and-egg in that the evidence tends to be assembled and focused around an initial starting strategic point. Generally you don’t actually start with the evidence and allow the strategic implications to be revealed – it is more of an iterative process wherein strategy and policy is tested and refined by looking at the evidence. But LEPS could be different?
The challenge is in getting from a general statement of direction to an understanding that will support the actual interventions that might be made.
So that the strategy will often start a loosely evidence-based understanding: for example:
CLICK“Exporting is good, the region doesn’t export much: we should improve exports”
This is obviously topical as it is essentially the entirety of our national growth strategy. We’ll leave aside the problem that this also the recovery programme for all the trade deficit countries and the trade surplus countries have no intention acknowledging, let alone addressing, the global trade imbalance.
Instead we’ll think through how this challenge was addressed in the context of the SW region. For the analytical mind (all present) this prompts a series of questions:
Taking our reasonable proposition we can break it down into a set of questions which need to be answered that will affirm the broad strategy and make it deliverable.
CLICK 1-5
For each of these questions around exporting there is a generic equivalent.
CLICK 6-10
This process of breaking a general statement down into a set of specific questions is the essence of creating a research brief. Some or all of the questions you may be able to answer with the resources (people, skills, evidence) you already have. What you can’t do you may want to commission externally. This is what we did in the case of exports.
Answering the first question: It is very clear from the literature that exporting firms exhibit many characteristics desirable to policy makers. (But note that the research only covers economic performance and says nothing about how exporting correlates with environment and social responsibilities).
For policy-makers correlation is necessary but understanding causality is key. Are firms more competitive because they export? or is a raised level of competitiveness a pre-condition necessary to start exporting?
The answer, according to the research is both. Firm level evidence shows that productivity rises before they start exporting as they reach an internationally competitive level and then there is a step change in productivity once exports commence.
The literature also identifies a set of firm-level characteristics positively associated with exporting: firm size, skills level, co-operation with others, foreign ownership, firm age and capital intensity. Industry and location are also important particularly when they come together in the form of agglomeration effects. These all become important when we try and understand the SW’s RELATIVE export performance.
So we can answer the first question: Yes, exporting does seem to be good!
So, we have two groups of firms those who are not competitive enough and are not exporting and the who are competitive enough and are exporting. But there are two further, less obvious possibilities B and C.
What sort of firms might be in these groups and what could be the policy implications - Spend a couple of minutes discussing these two questions with your neighbours.
B They could be firms on their way to exporting or there could be a some other barriers as like management skills and attitudes. In the SW it has long been suspected that business goals have been an important factor.
C These firms have some advantage that is effectively making them competitive enough. They will include firms in receipt of significant subsidies and those protected by the regulatory environment. These arrangements tend not to be sustainable so unless they induce an improvement in underlying performance they will eventually fail.
The previous diagram has been re-drawn to show the possible flows between the four quadrants.
This research was actually jointly commissioned by SWRDA and UKTI-SW and this diagram is helpful in thinking about who’s responsible for the different elements. Broadly the RDA’s role has been in getting firms / sectors to a higher level of competitiveness and UKTI’s has been about actually getting them across that threshold.
The grey pathway suggests that firms have to go through B to get to D.
The pathway from A to C should not be a policy option because of State Aid rules. But given the huge state investments in some industries e.g. the motor industry it’s hard to believe that this route is not being used.
So the evidence is clear that once you have achieved that objective of competitive and exporting firms this is a good thing.
So the next part of our starting proposition was that the region doesn’t report much. To test this we just need to at the regional trade data which is relatively good, although there are some issues.
By and large, where ever there are taxes or subsidies involved data collection is very good, even if its not always released. For exports HMRC is refunding VAT on goods exports so they have a complete record.
But this could be unfair on region’s with relatively high proportion of GVA accounted for by services…. So compare goods exports with manufacturing GVA….
And for some, like London which moves from near bottom to top, this is the case. But not for the south west.
This picture is very consistent over time with the region ranking last or second to last in every year between 1996 and 2005
There is one big caveat about the official economic data on exports which is that regional data on service exports is incomplete at best.
Official figures which cover about a third of the service sector show the SW export propensity at a little over half of the UK level. A more complete analysis using the Regional Account comes to the same sort of figure. However, with service exports London has a huge effect on the total so the SW does not compare so badly with other regions, usually being ahead of at least some of Y&N, WM and the Devolved bits.
So now we have established that the region's export performance really is something to be concerned about.
The next step is to try to understand the underlying causes. Sometimes what looks like a big issue on paper has a very simple explanation.
For example the DECC data on transport energy consumption show south Gloucestershire with shockingly high per capita figures. But the data is estimates from atmospheric emissions so the traffic on the M4 and M5 is included in GS’s figures.
So the next step is to look for a reasonable explanation for the SW’s export performance.
The obvious question at this point is how much of the region’s performance is simply as a result of our industrial structure? If we don’t have the presence in sectors that export it is rather different to having that presence but not exporting.
This is a relatively straight-forward thing to test as we have sector export data at the regional level. Firstly we can keep the SW’s sector export propensities but re-weight our industrial structure to match that of a region with a good export performance.
Secondly we can keep our structure but chance the sector propensities to match that of the SE.
This simple analysis makes it very clear that the SW’s export performance cannot be explained away by industrial structure. On paper we should be doing much better. So the challenge is to explain why we are not.
So going beyond the macro data sets we need to look the micro level drivers: what are the things that make it more likely that a firm will export?
From the Literature Review we know that there are a number of factors that are positively associated with firms exporting. What we want to know is what role these are actually playing in the region’s export performance.
In an ideal world there would be a detailed business survey which collected all these data on all of these variables with a big enough sample to allow regional and even sub-regional analysis.
In reality we have to make imaginative use of what we have available. Imaginative in the sense of bringing together data gathered in different surveys by using the IDBR as the common link and imaginative in the sense of identifying how the variable collected can be used as proxies for the variables we actually want.
This is the process of econometric analysis where-in a statistical model is used to identify the role that a number of different variable are having on the outcome sought, in our case on export propensity.
Because it involves firm-level data there are obviously issues around confidentiality. The choices available are either to use ONS analysts to do the work or alternatively ONS will allow competent analysts to use their micro data lab in Newport.
The output from this sort of exercise looks like this: a series of ‘odds’ which change as each variable is controlled for. So the first step might be to control for industrial structure and see how significant capital is. This is done for each of the variables in turn until we get to a point where all of the known factors are controlled for and we can see if there is still a difference between the regions.
Having done all of this work it was a little disappointing not to have arrived at some ground-breaking recipe for export success – but this is not surprising really.
Firm size and the level of foreign ownership are not easy to influence at the local but creating conditions for growth and openness external investment are possible at the national level. Local FDI initiatives have a rather mixed track record.
Capital investment, innovation and investment in skills are the bread and butter of raising productivity and hence of economic development. So doing more of the same appears to be the policy needed to improve exports.