1.  I have been involved in joint research with my colleagues Gary Gillespie and Kim Swales (Fraser of Allander Institute, University of Strathclyde) on the impact of FDI on the Scottish Economy for some time now, involving the application of our regional modelling framework to issues of inward investment. (These comments draw very liberally from this joint work.) Here I comment briefly on the more policy-relevant aspects of this work. I take it others will have rehearsed the background in terms of the growth in the importance of FDI and I will not repeat this here, but focus simply on the regional impact of FDI.

  2.  There is some evidence to suggest that regional impacts depend on the nature of the direct investment: in particular investment in "greenfield" sites may have more beneficial impacts than those that arise through takeovers and mergers (Ashcroft and Love). Except where explicitly stated otherwise, this note deals with the former. I begin by commenting on analysis of the likely regional impact of FDI and then consider appropriate evaluation methods.

Analysing the regional impact of FDI

  3.  There are two main sources of impact: that on the demand side of the regional economy and the effects on the supply-side. In theory, these effects on the host region could either be adverse or beneficial. For example, a merger could be followed by rationalisation that reduced demands for the host region's output and local "high level" functions (such as R&D). Furthermore, "technology sourcing" could be a motivation for inward investment, so that supposedly beneficial "spillovers" would be unlikely. However, there has been a tendency to assume that, at least for greenfield investments in a UK regional context, there are beneficial demand and supply side impacts, and there is (as yet at least) little evidence to the contrary. (However, see eg Driffield.) The debate tends to concern the precise nature and scale of these benefits.

  4.  However, it should be emphasised that there is a great deal that we do not really know about the impact of inward investment on UK regions, and considerable further research in the area is required if we are to answer the key policy questions with any certainty.

Impacts on the demand-side of the regional economy

  5.  One set of issues that receives particular emphasis traditionally in a regional context relates to the demand-side effects. These refer to the (largely transitory) construction expenditures as well as to the (permanent) stimulus to local exports that is normally associated with inward investment (at least in a Scottish context).

  6.  Our own work on the demand-side effects has tried to extend our knowledge in a number of respects. First, we have sought to develop input-output tables that explicitly disaggregate by ownership. That is to say that foreign-owned and indigenously owned sectors are separately identified in the Input-Output table, which identifies all the inter-firm transactions within the home region. In earlier work this simply distinguished foreign-owned manufacturing from other, but in more recent work we have distinguished four ownership types across five manufacturing sectors. This allows a proper input-output (I-O) analysis for the foreign-owned sector, and readily allows identification of the output, value-added and employment "multipliers" which are rooted in local databases that explicitly recognise the foreign-owned component of manufacturing industries. This contrasts with earlier work, much of which tended to employ multiplier values that are not based on ownership-disaggregated local databases, and indeed may not be based on local data at all.

  7.  Our results emphasise the importance of the different types of multiplier, reflecting the distinctive characteristics of the foreign-owned component of manufacturing. Thus employment multipliers for foreign-owned plants tend to be significantly higher than output multipliers in Scotland, reflecting their higher average wages, capital and imported intermediate intensities. While this work confirms the limited "backward linkages" associated with the foreign-owned sector, it also emphasises the importance of inward investment on Scottish employment.

  8.  In fact we have also been able in recent work, to develop plant-specific I-O databases that allow derivation of plant-specific output, value-added and employment multipliers. That is we were able to identify the relevant multipliers for IBM for example, and compare this to the impact of the entire sub-set. (Standard I-O techniques involving "hypothetical extraction" of the firm/industry are employed.) The striking feature of this is the very wide variation in the value of these that is apparent across individual plants. Not surprisingly, different types of plants have radically different demand impacts. The policy implication is that it is very useful and important to establish key plant characteristics to facilitate improved estimates of the likely local impacts. (We understand that Scottish Enterprise are now using this information in their evaluations).

  9.  A second feature of our treatment of demand is our recognition that the supply side of the host regional economy may, and typically will, be subject to resource constraints at least in the short-run. For example, capital stocks and population are likely to be fixed in the short-run. Thus the demand impact of FDI will generally induce wage and price increases and exert an adverse impact on regional competitiveness, reducing net exports. However, over time capital stocks and population gradually adjust, through investment and migration flows respectively. Ultimately, in-migration and capital accumulation can generate results identical to those associated with regional I-O analyses (augmented for endogenous investment and population), though this depends on an absence of any regional-specific factor of production. Our analysis seeks to track the impacts of FDI through time, and confirms that the measured impact depends importantly on the time horizon that is adopted. Thus the immediate effects may be quite different from the longer-term effects of FDI. The policy implication is that the implicit time horizon of policy-makers may have a significant impact on any evaluation.

  10.  It may be worth emphasising a point implicit in the preceding paragraphs: that it is crucial to adopt a system-wide perspective in evaluating FDI. The effects depend, in particular, on what is assumed about the labour market in the host region. Conventional I-O models assume an entirely passive supply side, motivated perhaps by the notion of a region characterised by substantial (involuntary) unemployment and excess capacity. Our approach (involving the use of a Computable General Equilibrium model, or CGE) recognises that this is not true of all regions in all time periods. However, this does have the important implication that there is no alternative to policy evaluations at least implicitly taking a view about regional wage determination and regional migration. (This can easily be seen by taking the extreme—and unrealistic, except perhaps for an economy like Jersey—case of an entirely fixed regional labour supply. In this case the employment multiplier must be precisely zero.)

Supply-side impacts

  11.  It is generally accepted that positive efficiency differences exist between multinational firms and indigenous plants. It is argued that these efficiency differences generate subsequent "efficiency spillovers" from the high-efficiency FDI sector to the domestic sector. These spillovers are assumed to be the result of positive externalities from international production which are conveyed to indigenous firms through interaction or competition with foreign-owned plants within the region. There are a number of possible transmission mechanisms for such effects, for example, induced productivity through demonstration or competitive effects.

  12.  Efficiency spillovers are not restricted to domestic firms within the particular sector receiving the FDI. Linkages from the FDI to other domestic sectors may provide a route through which spillovers may be transmitted. However, quantifying the scale of these spillovers has proved difficult and controversial. Barrell and Pain is an example of an influential aggregate study that estimated that a 1 per cent rise in the stock of FDI in UK manufacturing ultimately raises manufacturing labour-augmenting technical progress by 0.27 per cent, though they find no evidence of an effect in services.

  13.  Hubert and Pain investigate the robustness of this result by extending the analysis in a number of important respects, including investigation of whether this is simply due to compositional effects. (FDI increases average productivity even in the absence of genuine efficiency spillovers.) The finding of a statistically significant impact seems robust. However, others have arrived at different conclusions (eg Driffield, and Wakelin et al). There is a need for more econometric work on micro databases, especially in a regional context, before any firm conclusions can be drawn from these analyses.

  14.  In Gillespie et al we investigate the likely system-wide ramifications of regional spillover effects of the scale of those estimated for the nation by Barrell and Pain. While these effects do take much longer than the demand effects to build up, the stimulus to indigenous labour efficiency does ultimately have a substantial impact on the host economy, especially in terms of output and value-added. This is good news for those who emphasise the wealth creating character of spatial regeneration policy. The employment impact is rather less significant, however, because the increase in labour efficiency implies that, while the cost of labour in efficiency units falls, less employees are required to produce any given level of output. However, even the employment impact is ultimately non-trivial.

  15.  The demand effects dominate the supply (spillover) impacts in our model simulation experiments, and imply a substantial positive impact on the Scottish economy. This dominance is much more marked in the short-run than in the long-run and spillover effects do ultimately have a significant impact on the Scottish economy as noted above.

Policy and its evaluation

  16.  In terms of the regional policy evaluation process, we would make a number of points. First, there is some reassurance for current evaluation procedures in that the demand effects that they focus on are large and dominate the supply effects. This is particularly true of the short to medium-term horizon of current evaluations.

  17.  However, these evaluations do miss out on a potentially important source of policy-induced regional GDP growth over longer time periods. Ultimately, efficiency spillovers may have significant beneficial impacts on the host region, but current evaluation methods would not pick this up because they are typically demand orientated and short to medium run in their focus.

  18.  In other work we also question the "100 per cent crowding out" assumption that the Treasury tends to make in relation to regional policy expenditures. If regional policies, such as those directed at inward investment, genuinely stimulate regional exports and ultimately facilitate efficiency spillovers then such policies are not a zero sum game.

18 July 2000