Pollution Abatement Expenditure, Productive Efficiency and ...



Does Plant Ownership Affect The Level Of Pollution Abatement Expenditure?

Alan Collins

Department of Economics, University of Portsmouth, Locksway Road, Milton, Southsea, Hampshire, PO4 8JF, U.K.

Email: alan.collins@port.ac.uk

Richard I.D.Harris

Department of Economics and Finance

University of Durham, UK

Abstract

This paper considers a number of hypotheses. Primarily amongst them is the notion that foreign owned plants spend more on pollution abatement than domestically owned plants after controlling for productive efficiency and cognisant of the prevailing regulatory regime. The evidence drawn upon in the first econometric assessment of this contention is plant level data from the UK metal manufacturing industry. In essence, this study directly estimates the influence of ownership and efficiency characteristics in firms’ decisions regarding whether to spend or not on pollution control and how much to spend. To explore these themes a two stage econometric exercise was undertaken on a hitherto unused source of environmental data, namely the UK Annual Business Inquiry Respondents Database (or ARD). A Heckman-type sample selection model was estimated to examine the probability of abatement expenditure being made or not and also to explain how much was spent on each of the principal means of pollution control. The results suggest that older plants were more likely not to incur any expenditure on process or post-production pollution capital expenditure. Plants that were non-EU foreign owned were generally more likely to spend on pollution abatement than UK plants. Likewise, in the main, the more efficient firms and the more capital-intensive firms were also more likely to spend on pollution control than UK-owned firms. The significance or otherwise of a wide range of other factors was also explored and reported on.

JEL-L JEL-O JEL-D

Keywords

Pollution abatement Efficiency UK manufacturing

a Support from the Economic and Social Research Council (reference no. R000222602) is gratefully acknowledged, as is permission from the Office for National Statistics at Newport, South Wales, to use the Annual Business Inquiry Respondents Database.

I Introduction

An extensive body of theoretical studies continues to evolve and shed light on firms’ likely responses to alternative environmental regulatory instruments and regimes[1]. An interesting recent strand of this work has begun to examine in particular firms’ technology choice with respect to the linkages between profit maximisation and the imposition of specific pollution abatement instruments (Kort et al. 1991). This is an important policy focus for regulators and the business community in the context of increasingly aggressive environmental policy objectives, where the scope for policy substitution has been raised. This relates to the idea that environmental policy objectives may be attained by (i) pure productive efficiency enhancing means and (ii) direct pollution abatement enhancing means. It may be, however, as in the U.K and elsewhere, that environmental regulatory bodies do not have as yet the human capital resources to contemplate the environmental potential of the productive efficiency enhancing means. In the context of the USA Gray and Shadbegian (1995) found a negative relationship between a plant’s pollution abatement costs and its total factor productivity. Yet, it is reasonable to posit that more efficient plants may have less need to engage in pollution abatement expenditure (PAE), since, by virtue of their greater efficiency with regard to the use of resource inputs, they intrinsically generate lower levels of pollution. Perhaps contrarily, it is also reasonable to think that efficient plants will be amongst the most keen to seize worthwhile resource input minimizing opportunities when they arise. Clearly, this is likely to be associated with higher levels of some particular types of PAE, such as process-based capital expenditure, as opposed to arguably cruder end-of-pipe solutions.

Despite the extensive theoretical activity in this area, relatively little econometric work has taken place that moves us towards injecting more of an empirical dimension in support of this developing firm-environment literature. The non-case study based empirical work that has taken place has primarily focused on evidence drawn from the USA. Such work has usefully exploited the annual Pollution Abatement Costs and Expenditures Survey (Barbera and McConnell 1986). Whilst metals have recently been the subject of macro-scale environment-economy interaction modelling (Guinee et al. 1999), this particular study contributes to the corpus of firm-environment research at the industry level. This work empirically examines the linkages between plant ownership, productive efficiency and the decision to engage in pollution abatement. More specifically, we address the following question. After controlling for differences in productive efficiency, do domestic or foreign-owned plants spend more on pollution abatement?

Evidence is drawn from the specific context of the UK metal manufacturing industry over the period 1991-1994. Other UK industries over this time period would have presented equally satisfactory sources of evidence to explore the hypotheses we set out, operating as they did under the same environmental regulatory regime. Using this evidence, for those plants found to be engaging in pollution abatement, this study also presents the first econometric study to consider what determines their actual level of PAE in the four main categories of pollution control. These are, namely, process based capital expenditure, post-production capital expenditure (end-of-pipe solutions), current expenditure (using the firm’s own staff), and through payments to others (i.e. contracting out some pollution control functions). Yet, as we have already emphasised, plausible reasoning could readily be devised to articulate both positive and negative influences of productive efficiency on the undertaking of PAE.

This paper extends the firm-environment literature in two key aspects. First, we believe our study provides the first non-US econometric study of PAE relevant to a major pollution intensive industry. This is furnished by exploitation, for the first time by economists, of the environmental data contained within the Annual Business Inquiry Respondents Database (ARD) of the UK Annual Census of Production (now known as the Annual Business Inquiry). Second, this study is the first to explicitly consider the hypothesis that foreign plant ownership raises the probability that a plant will engage in pollution abatement activity. Many studies have investigated or hypothesised how national or cultural differences may influence firm profitability, productivity, and levels of research and development amongst other things[2]. Hence, there might also reasonably be expected to feature some environmental performance implications arising from such differences. For example, given that the stringency of environmental regulatory regimes varies significantly in different countries one might expect there to be environmental benefits aspects to the positive externalities generally said to arise from the presence of foreign -owned (FO) plants in a particular host country (Blomstrom and Kokko 1998). Indeed there is a large body of work[3] that has sought to suggest, identify and help explain a wide range of positive effects on industry-wide or general manufacturing sector productivity, as a direct consequence of such increasing FO firm penetration.

The principal findings of this particular study are that increasing efficiency levels leads to a small increase in the probability of making expenditures on pollution abatement investment in the production process but very substantially increases the probability of zero expenditure on direct staff and operating costs relating to pollution control. It is also found that US-owned plants are more likely to incur some spending on post-production assets but are significantly less likely to spend on other forms of pollution control. EU-owned plants, however, are more likely to spend on pollution abatement than UK plants. Plants owned by enterprises from Australia, New Zealand, South Africa and Canada have a higher probability of incurring some expenditure on pollution abatement. Furthermore, and probably reflecting actual or perceived variations in regulatory enforcement effort, plants in less populated areas are far more likely not to engage in any PAEs.

The paper is organized in the following way. In Section II theoretical issues are considered relevant to the linkages between ownership status, efficiency and the regulatory regime in the time period under study, and our main hypotheses are posited. Section III presents some background contextual information relating to the UK metal manufacturing industry over this period. This descriptive analysis suggests a number of hypotheses that are tested in the subsequent econometric phase of the study. Section IV sketches an outline of the modelling strategy and econometric model employed to examine the influence of productive efficiency and ownership status on PAE. Section V presents the results with brief concluding remarks offered in Section VI.

II Ownership, Efficiency and Capital Intensity with Technology-Forcing Environmental Standards

There are a number of theoretical reasons why one might expect differences between domestically and foreign owned plants with respect to their level of efficiency and accordingly their environmental performance. These relate to mainstream productive efficiency reasons and more specific resource productivity based explanations. These are considered in turn.

Irrespective of the fact that FO-plants are more likely to be younger, explanations for higher FO plant productivity (as in Figure 1) relates to a combination of two sources, namely, labour productivity and superior technology that will tend to be more environmentally benign. Labour productivity in the FO plant may, however, also be higher because more output-per-employee emerges due to the use of a superior technology. The superior technology explanation can embrace more than simply a swifter rate of engineering or scientific advance in FO firms. It may also incorporate “soft technology” aspects with superior managerial and production organisation practices (e.g. Total Quality Control). Thus superior technology enables firms with the same level of capital-per-worker, to produce more output per unit-of-labour (position FO1 in Figure 1). The other extreme is that higher labour productivity may simply arise because a plant uses more capital-per-worker (position FO2), so while labour productivity is higher, capital productivity is lower. The evidence (at least for the UK) suggests that the superior technology explanation is likely to predominate (see Harris, 1999ab).

Yet in the face of uniform environmental regulations across plants and a competitive market, variations in PAE should in large part be explained by differences in plant efficiency. Contingent on their overall stringency, more efficient plants should require less PAE to satisfy any given environmental regulations. However, in terms of the dynamics underlying the competitive process, efficient firms may also be the most likely to lead in the adoption of resource minimizing and hence cost-saving production techniques. The latter view implies a greater tendency towards voluntary overcompliance (Arora and Gangopadhyay 1995) by efficient firms with respect to environmental regulations. This effect has been observed in a developing country context. For example, Eskeland and Harrison (1997) present evidence that foreign-owned plants in four developing countries were significantly less polluting than comparable domestic plants.

This overcompliance effect could be inferred from systematically greater PAE in efficient firms than in inefficient plants. Overcompliance has been explained in terms of its possible role as an element in a non-price competitive strategy (Kirchhoff 2000). In this sense it can be exploited in some consumer markets as a source of competitive advantage on the grounds of quality differentials. Indeed when such markets are also characterized by highly asymmetric information between firms and consumers, there also arises the potential for “greenwash” i.e. where firms lie about their environmental performance (Kirchhoff 2000). However, in the context of intermediate goods markets such as metal manufacturing, this is a less convincing explanation for overcompliance. More likely is an explanation based on a process of passive evolution linked principally to the notion of a sunk cost or path dependency argument

Turning to Figure 2, it is likely that each technology choice will be associated with differing levels of resource productivity (Y/R). Resource productivity can be increased by recovering more of the potential residuals discharge from the production process to serve as output. In general, Technology 2 in Figure 2 could offer greater resource productivity than Technology 1 based on a number of processes including wholly in-plant recycling of raw materials, the use of generated heat from production as an energy source, and re-use of waste materials as another product line. If a technology forcing environmental regulation was introduced to try to induce a higher level of resource productivity, then this could require a shift to a level of technology superior to Technology 1. (Note, Figure 2 also suggests that greater resource productivity is also associated in heavy industry with greater capital intensity.)

In the context of the metal manufacturing industry over the relevant data time period, the prevailing regulatory instrument across all plants was indeed a technology-forcing standard determined by reference to an ambiguous and inefficient guiding doctrine – BATNEEC – Best Available Technology Not Entailing Excessive Cost. This was formally introduced in UK statute within the 1990 Environmental Protection Act. On this basis there was greater regulatory pressure to install a capital stock within newer plants that would be closer in function to the “best available technology” (BATD) level in the domestic market. However, under this guiding principle, environmental regulators could be minded to tolerate some level of departure from this BATD standard in older established plants, where it might generate an excessive “corporate burden” (Pearce and Brisson 1993). By implication this involved the environmental regulators forming an implicit view as to an acceptable rate of return for the firm. The older established domestic firm could thus meet the requirement for greater resource productivity whilst retaining most existing capital. This could be undertaken by augmenting the existing capital stock with additional discrete pollution abatement orientated capital. Alternatively, the firm could re-assign existing staff or hire new staff to engage exclusively in pollution abatement related tasks. This would inevitably reduce labour productivity. Whichever option or combination is applied, let this departure from the domestic level of BATD be represented by Technology S in Figure 2.

Accordingly, it is likely that FO firms have a systematic tendency to overcomply (Y/RFO2 > Y/RUK-owned) and ‘overspend’ on pollution abatement. This can arise from a combination of (i) higher mainstream production capital intensity (K/LFO1 > K/LUK-owned) which will generally support greater residual recovery in heavy industry, or, (ii) because the transplanted production technology intrinsically embodies a given higher level of resource productivity. This would accord with the notion that FO firms may have experience of stricter environmental regulation in their home country (say where BATDomestic < BATForeign). Hence, for this reason they are, at least in the short run, locked into an environmentally superior technology. Even in the long run such firms would have to make a judgement concerning the extent to which they would meet expected future levels of stringency of environmental regulations and set that against any benefits from relaxing resource productivity (reducing overcompliance). Given that one would generally expect environmental regulations to be increasingly stringent over time, then overcompliance, especially by FO firms, seems likely to persist even in the long run.

Thus, the arguments presented here suggest that factors such as foreign-ownership, capital-intensity, efficiency, and the age of the plant are likely to be important in determining PAE[4]. More explicitly, distilled from the above discussion, premised largely on the overcompliance explanation, the main hypotheses that are tested in the subsequent econometric phase are:

i) FO plants engage in greater PAE than domestically owned firms.

ii) More capital intensive plants engage in greater PAE.

iii) More efficient firms engage in greater PAE.

iv) Older plants engage in greater end-of-pipe (capital augmenting) PAE.

In addressing a uniform technology-forcing standard, PAE decisions can also reasonably be viewed as a sequential decision process. First, firms decide whether they need to spend or not on pollution control. They also need to decide what level and what type of pollution abatement expenditures they wish to make. Given that certain types of pollution abatement expenditures are intrinsically more expensive than others, then it seems reasonable to suppose that the level and type of expenditure decision could be considered jointly. To test whether different forms of pollution abatement expenditure are complements or substitutes to each other would require the estimation of a simultaneous model. However, we do not have the econometric tools to estimate a simultaneous Heckman model (see below for details), and furthermore we lack prior information that would allow us to impose some structure on such a model (i.e., which variables should enter which equation in the 2-stage Heckman approach in order to identify the system). Thus, as a first attempt we have resorted in the econometric phase to estimating a reduced-form version of such a structural system. Yet to inform such model development it is first necessary to broadly appreciate the significance and scale of the focus of this study – the UK metal manufacturing industry.

III UK Metal Manufacturing Industry 1991-4: Background

Metal manufacture and use is vital to the social and economic prosperity of the entire globe, and most nations participate to some degree in its manufacture (Roberts 1996). In the UK the metal manufacturing industry comprises a presence in both iron and steel (ferrous metal) and non-ferrous metal manufacture. Of the former, this includes manufacture of basic products such as steel tubes and steel wire as well as drawing, cold rolling and forming of steel to be used in the manufacture of other products. In terms of non-ferrous metals this primarily comprises the manufacture of aluminium and aluminium alloys, copper, brass and other copper alloys. There are also some plants manufacturing some other non-ferrous metals and their alloys. The source, nature and method of assembly of the metal industry panel data available from the ARD used in this and the next section are described in the Appendix.

By way of critical assessment of the data it should be noted that it was collected as part of the UK government's Annual Business Inquiry that forms the basis of the 'official' statistics used to measure output and costs in each industry. The government use a stratified sampling procedure to ensure that the data collected achieve good coverage of each industry, and since employment information is

available for each plant (whether included in the annual inquiry or not) it is possible to weight the data to obtain nationally representative figures. As such, the data we use is likely to be both accurate (in

terms of point estimates of pollution expenditures) and contain sufficient coverage of the industry to make its use statistically robust when testing the types of hypotheses suggested in the previous section.

Metal manufacturing has long been a significant source of environmental pollution (Braennvall et al. 1999). It poses considerable health risks to both workers (Comba et al. 1992, De et al. 1995, Maynard et al. 1997, Moulin et al. 1998), and the public (Baxter et al. 1996, Guinee et al. 1999). Its role in diminishing the quality of the physical environment has also been the subject of much scrutiny (Tremmel 1992, Dudka and Adriano 1997, Guinee et al. 1999). In contrast with most other industrial sectors the waste residuals produced in metal manufacturing are largely non-dissipative (i.e. not immediately or gradually dispersed into air, water or soil in the course of their normal use) (Kneese et al. 1970). The residuals comprise bulky solids (e.g. slag), much particulate matter, gaseous emissions from energy conversion, and much liquid waste resulting from cleaning or “pickling” the metal during fabrication to reduce oxide scales when the metal has contact with air. The principal pickling agent for steel is sulphuric acid, but other acids such as hydrochloric, nitric and hydrofluoric are also used. Slag may be re-used in road construction aggregates and in concrete manufacture. A substantial volume of the particulates produced in the foundries as “flue dust” can be recovered “by wet scrubbing” and other precipitation processes due to their relatively high metal contents. Other particulates such as soot from coke and coal burning (used in reheating furnaces and rolling mills) can also be captured by various forms of carbon filters and scrubbers. Most of the liquid waste acids can be neutralized with lime but recovery has been considered problematic (Marquardt and Nagel 1992). That said, from the liquid wastes in most plants it has been possible to generate commercial grade ferrous chloride solution for use in flocculation processes in water treatment plants. These processes also apply in some non-ferrous metal manufacture (copper and brass mills) but in addition with regard to copper, lead and zinc, some very concentrated and highly toxic sulphur dioxide fumes are also generated. Some of this sulphur may now, however, be recovered as commercial grade sulphuric acid.

Thus, it can be seen that environmental pollution is a major ‘output’ of the metals industries. Before considering PAE in this section and by way of context, it is instructive to look at the pattern of PAE across the UK manufacturing sector as a whole. In this way any distinctive features of metal manufacturing can be drawn out. As a consequence of the heavy pollution potential of this industry considered above, the declared expenditure per plant on managing waste residuals is relatively high, only exceeded on average by three other sectors (see Figure 3). In the manufacturing sector as a whole, of those plants that spend on pollution control, payments to others to manage and dispose of their waste dominates over this time period as the prime means of dealing with waste residuals (Figure 4). Recasting this picture in terms of plant ownership categories, UK manufacturing plants seem to lag behind in PAE with regard to all means of pollution control except payments to others (Figure 5). Turning now to the metal industry specifically (Figure 6), over the period 1991-4 average annual spending by plant on current staff for pollution control seems to have risen significantly from just over £10,000 to approaching £35,000 (although there is some evidence to suggest this may have been offset by a decline in process-based capital expenditure). Nevertheless, the figures for this and the other means of pollution control (such as process capital expenditure) still seem remarkably small given the scale and nature of the production processes being undertaken. The small magnitudes of these declared levels of PAE (in the context of this pollution intensive industry) lend some weight to the view that pollution control is inextricably bound up with the mainstream production process. In essence then it is possible to view plant efficiency, in addition to regulatory stringency, as a key driver of the level of PAE.

Viewing this pattern of expenditure by plant ownership category (see Figure 7) shows some interesting deviations from the national picture. Of those plants that do spend, those plants from the Commonwealth block (Australia, Canada and South Africa) tend to dominate in terms of overall PAE. These are countries where strong vertical relations amongst firms can be expected since they are countries where much of the metal ore deposits are extracted. Hence, there may be some PAE spillover arising from linkages with the metalliferous ore extraction industry. In contrast to the wider manufacturing sector, of those plants that do spend, UK plants no longer lag behind all other ownership groups in that US owned plants declare less PAE. The rest of Europe still dominates the UK in overall term in PAE with the specific and surprising exception of process based capital expenditure, where the UK even exceeds average European expenditure. This descriptive analysis does not, however, take account of plant variations in productive efficiency, and perhaps underplays the fact that there are a large number of firms that do not spend directly on PAE at all over the period 1991-4.

IV Econometric Model

In order to discover the strength of some of the relationships between PAE (by type) and factors such as efficiency, foreign ownership and capital intensity, we posit a simple 2-stage model that assumes the variables in wit (below) determine whether a plant spends/does not spend on pollution control. This comprises:

wit = (ln EFF, ln GVA, ln AGE, ln KL, EU, US, AUS)it

+ ln DENt + 4-digit SIC industry dummies [1]

where[5] EFF is a measure of plant level technical efficiency;

GVA is real gross-value-added;

AGE is the age of the plant;

KL is the capital-to-labour ratio;

EU, US, AUS are dummy variables coded 1 if the plant is owned by an EU, US or Australasian/Canadian/South African enterprise; and

DEN is population density of the Local Authority District in which the plant is located.

The DEN variable proxies for variations in regulatory stringency and the perceived extent of pollution hazard. Strictly speaking, this departs from the approach used by Gray and Shadbegian (1998) in the USA (though they too have previously considered the use of population density for this purpose in that geographical context). This departure relates to different institutional environmental regulatory frameworks. Within the USA, the federal Environmental Protection Agency (EPA) takes the lead role in environmental regulation, however, state agencies are also strongly involved in the setting and enforcement of environmental standards. Accordingly, differences by state have been found in environmental regulatory stringency. By using electorally based proxies as a measure of regulatory stringency, Gray and Shadbegian (1998) examined (in one industry) how this impacts on firms’ investment decisions. In the specific context of the United Kingdom, however, sub-federal agencies do not have as heavy a role. Accordingly, such electoral proxies are not really valid in this particular national context. This is not to say, however, that the UK Environment Agency does not over time and spatially, vary in its level of regulatory stringency. It may, for example, concentrate its enforcement efforts in more heavily populated areas, where they may quite reasonably perceive the risks of environmental pollution on public health to be higher than in less populated areas.

The variables in the second-stage model that determine the amount spent (if PAE>0) comprise:

xit = (ln GVA, EU, US, AUS)it + t + 4-digit SIC industry dummies [2]

where t is a time trend.

In essence, by this approach, it is assumed that various plant level characteristics determine whether it is in the interests of the plant to actually spend anything on pollution abatement. If the answer is 'yes' then the scale of output (and thus presumably pollution), ownership, time and industry effects determine the volume of spending. The dependent variables comprise zit = 1 if the plant spent anything on pollution control in time t; and yit is real[6] expenditure on pollution abatement by plant i in time t.

The approach used is based on the standard Heckman (1979) sample selection model where the selection mechanism comprises:

[pic]= ((wit + uit, zit = 1 if [pic]> 0 and 0 otherwise, where prob(zit = 1) = ((((wit) [3]

where ( is the density function and the regression model comprises:

yit = ((xit + (it observed only if zit = 1, and (uit, (it) ~ bivariate normal [0, 0, 1, ((, (]. [4]

Thus we wish to estimate the following model that is based on an efficient and unbiased estimator of ( when yit is observed only when zit = 1, i.e.,

E[yit | zit = 1] = ((xit + (((((((wit) [5]

where ( is the correlation between (u, (), and ((((wit) = (((((wit)/[1(((((wit)] is the inverse Mills ratio which is obtained from estimating the selection model[7] (where ( is the probability density function associated with the first stage of the Heckman model, i.e. equation [3]).

The variable names, definitions, and basic descriptive statistics are set out in Table 1. In this study, the measure of plant level technical efficiency (or more accurately technical inefficiency) is measured via a stochastic frontier production function that allows each plant to have different levels of efficiency in different years. Full details are presented in Harris (1999b). Regarding the relevance of our approach, we are clearly using existing (and econometrically appropriate) methods in order to test specific hypotheses not usually tested because of lack of data. To devise new methods for testing, while clearly an advance, is beyond the scope of the present paper.

V. Results

The results discussed below are based on an exploratory econometric analysis of the ARD data and as such its is important to be aware of the limitations of the models estimated. Ideally one would wish to consider the determinants of the different types of PAE simultaneously. Given limited information about the fuller structural relationships and the fact that the econometric tools to estimate the full structural model are currently unavailable, this study has resorted to estimating reduced-form models. It is suggested in this study, however, that this first cut approach is not inconsistent with a profit maximising model where firms maximize profit subject to producing goods and ‘bads’ (environmental pollution) (see, for example, Kneese et al. 1970, Hernandez-Sancho et al 2000).

The results presented in Table 2 enable us to make some comments concerning PAE and in respect of the four types of pollution control undertaken. First, the results suggest that by increasing efficiency levels by one standard deviation – see Table 2b (and the results referring to spend/not spend) for details – this would increase the probability of incurring expenditure on assets used in the production process to minimise residuals by over 3 per cent and decrease the probability of spending anything on payments to others by 4.8 per cent (Table 2d). Further, it would also decrease the probability of spending on direct staff, material and operating costs relating to pollution control by 6.8 per cent (Table 2c). These results are generally supportive of the characterization of efficient firms as being among the most keen to introduce resource input minimizing techniques, but (because they are more efficient) they are less in need of undertaking current expenditures on pollution abatement.

With respect to the results on whether to spend or not spend in Table 2, increasing production increases the probability of spending on all forms of pollution abatement. For example, a standard deviation increase in real GVA increases the probability of incurring expenditure on payment to others relating to pollution control by over 3 per cent. As to how much is spent (cf. the second block of results in Table 2), the elasticity of pollution control spending with respect to the amount produced ranges from 0.8 to 0.97, implying that as output increases the amount spent on pollution abatement increases in a similar proportion. It is also possible to observe that older plants are more likely to spend nothing on pollution abatement. A standard deviation increase in the age of plants increases the probability of incurring no process or post-production capital expenditure by approximately 6-8 per cent depending on PAE type.

In accordance with the simple model set out earlier, those plants with greater capital-intensity are more likely to spend on pollution control. By illustration, a standard deviation increase in the K/L ratio decreases the probability of incurring no production process capital expenditure by nearly 7 per cent.

Perhaps indicative of greater stringency and enforcement of environmental regulations nearer or in urban settlements, plants located in less populated areas are more likely not to spend on pollution control. These results show that a standard deviation increase in the population density increases the probability of incurring no expenditure on direct staff, material and operating costs relating to pollution control in the metal industry by some 3.8 per cent

Turning now to the effects of plant ownership status, the results show (ceteris paribus.) no statistically significant effect for EU-owned plant ownership with respect to spending on pollution abatement (vis a vis UK-owned plants). This is despite the seemingly stronger profile given to the issue of environmental quality in the politico-legal framework of continental Europe. For example, Krol and Steil (1997) noted that in relation to some German non-ferrous metal plants, they had to address at the Federal level alone 233 laws, 549 directives and 498 administrative regulations with environmentally relevant rules. In addition, there were at the time 330 EU regulations and several thousands of state and municipal regulations. Alternatively, this result may be indicative of success in increasing harmonization of important aspects of EU environmental policy given that the UK is a member of the EU bloc. Plants owned by enterprises from Australia, New Zealand, South Africa and Canada have a higher probability of incurring some expenditure on pollution abatement. In particular, they are over 12 per cent more likely to spend on assets for post-production pollution control and waste management, when compared to UK-owned plants, and between 14 and 27 per cent more likely to spend on payments to others and expenditure on direct staff, material and operating costs, respectively

Against a backdrop of a regulatory regime driven by adherence to the BATNEEC principle, this evidence is suggestive of systematically greater incidences of voluntary overcompliance being more likely in Commonwealth FO-plants than UK-owned plants, but show rather mixed results for the US with respect to PAE type. Whilst US-owned plants are 11.8 per cent more likely to incur some spending on post-production assets, they are, ceteris paribus, significantly less likely to spend on other forms of control (e.g., nearly 30 per cent less likely to make payments to others). This seems to characterize US plants as being more likely to favour “techno-fix” and in-house solutions when addressing environmental problems.

For those plants that have positive PAE spending, EU plants spend 95%-106% more than UK owned plants on post-production expenditure and payments to others, respectively, but between 20 and 77% less than UK-owned plants on process capital expenditure and direct expenditure, respectively. It is not entirely clear why this should be so, though it may simply reflect an historical regulatory emphasis given to these types of pollution abatement methods in the UK. Commonwealth plants have some 95-100% higher expenditure than UK-owned plants with respect to current expenditure. As with the results for the probability of spending anything, US-owned plants are likely to spend more on post-production pollution capital expenditure but less on other forms of pollution abatement. Hence, any arguments concerning plant ownership and pollution control really needs to take into account the nature of the PAE. Otherwise one would risk masking distinctive patterns, with respect to different emphasises by ownership category, on particular approaches to pollution control.

There are some major differences across plants in different sectors of the metal industry. For example, in terms of whether any expenditure takes place, copper, brass and brass alloy production is associated with particularly high probabilities of incurring any PAE when compared to the reference group (iron and steel manufacturing). In large part, this can be related to the need to control the more toxic fumes (such as high concentration sulphur dioxide) associated with this sub-sector of the metal industry. Non-ferrous metals also generally have a higher probability of PAE (and to a lesser extent steel tubes and aluminium and alloys). In contrast, steel wire and other drawing, cold rolling and cold forming of steel often have lower probabilities of incurring expenditure vis a vis the iron and steel reference group.

Finally, in terms of how much is spent, the time dummy is negative in most cases, implying that plants were finding other ways to meet pollution control targets other than through paying for such control, and probably also linked to the impact of the 1991-93 recession which was particularly severe.

VI. Concluding Remarks and Summary

This paper has explored the determinants of the amount directly spent in metal manufacturing plants on pollution control. In particular, the influence of ownership and efficiency characteristics were considered as key determinants of the decision whether to spend or not on pollution control and how much to spend. To explore these themes a two stage econometric exercise was undertaken on a hitherto unused source of environmental data, namely the UK Annual Business Inquiry Respondents Database (or ARD). A Heckman (1979) sample selection model was estimated to examine the probability of abatement expenditure being made or not and also to explain how much was spent on each of the principal means of pollution control.

Operating within the prevailing environmental regulatory regime of the time – a period which followed the highly ambiguous BATNEEC doctrine - the results suggest that over the period 1991-4, older plants were more likely not to incur any expenditure on process or post-production pollution capital expenditure. In accordance with our primary hypothesis, plants that were non-EU foreign owned were indeed generally more likely to spend on pollution abatement than UK plants.[8] Should pollution abatement objectives feature more highly on the Government policy agenda then this finding would point to addressing some policy domain overlaps. In particular, it would suggest environmental regulators should be actively supportive of greater levels of foreign direct investment (through acquisitions), from particular country blocs. Likewise, in the main, the more efficient firms and the more capital-intensive firms were also more likely to spend on pollution control than domestically-owned firms. Accordingly, given the significance of efficiency in this environmental policy arena, a key policy implication that warrants attention is for industrial and environmental regulators to explore the potential benefits from policy integration in the framing of environmental objectives. The evidence presented in this paper suggests that industrial and environmental policy could be more usefully considered as complementary, especially with regard to pollution abatement current expenditures. The significance or otherwise of a wide range of other factors was also explored and reported on including the PAE dampening effects of lower perceived regulatory enforcement levels.

References

Aghion, P. and P. Howitt (1998) Endogenous Growth Theory, MIT Press, Cambridge.

Arora, S., and Gangopadhyay, S. (1995) Toward a theoretical model of voluntary overcompliance. Journal of Economic Behavior & Organization 28(3) pp.289-309.

Barbera, A.J. and McConnell (1986) Effects of Pollution Control on Industry Productivity: a Factor Demand Approach. Journal of Industrial Economics XXXV(2) 161-172.

Baudry, M. (2000) Joint Management of Emission Abatement and Technological Innovation for Stock Externalities, Environmental & Resource Economics, 16(2), 161-183.

Baxter, M.S., Mackenzie, A.B., East, B.W., and Scott, E.M. (1996) Natural Decay Series Radionuclides in and Around a Large Metal Refinery Journal Of Environmental Radioactivity 32(1-2) 115-133.

Blomstrom, M. and Kokko, A. (1998) Multinational Corporations and Spillovers, Journal of Economic Surveys, 12, 247-278.

Braennvall, M.L., Bindler, R., Renberg, I., Emteryd, O., Bartnicki, J. and Billstroem, K. (1999) The

Medieval Metal Industry was the Cradle of Modern Large-scale Atmospheric Lead Pollution in Northern Europe. Environmental Science & Technology 33(24) 4391-4395.

Buckley, P.J. (1983) New Theories of International Business: Some Unresolved Issues, In: M. Casson (ed.) The Growth of International Business, George Allen & Unwin, London.

Caves, R.E. (1974) Multinational Firms, Competition, and productivity in Host-Country Markets, Economica, 41, 176-93.

Comba, P., Barbieri, P.G., Battista, G., Belli, S., Ponterio, F., Zanetti, D., and Axelson, O. (1992) Cancer of the Nose and Paranasal Sinuses in the Metal Industry: a Case-Control Study. British Journal of Industrial Medicine 49(3) 193-196.

Cornwell, L. and Costanza, R. (1994) An Experimental-Analysis of the Effectiveness of an Environmental Assurance Bonding System on Player Behavior in a Simulated Firm. Ecological Economics 11(3) 213-226.

Damania, D. (1996) Pollution Taxes and Pollution Abatement in an Oligopoly Supergame, Journal of Environmental Economics and Management, 1996, 30(3)323-336.

Dudka, S. and Adriano, D.C. (1997) Environmental Impacts of Metal Ore Mining and Processing: a Review. Journal of Environmental Quality 26(3) 590-602.

Dunning J. (1958) American Investment in British Manufacturing Industry George Allen & Unwin, London

Dunning, J. (1977) Trade, Location of Economic Activity and Multinational Enterprise: a Search for an Eclectic Approach, In: B. Ohlin, P.O. Hesselborn and P.M. Wijkman (eds.) The International Allocation of Economic Activity, London, McMillan.

Edwards, S. (1998) Openness, Productivity, and Growth: What Do We Really Know? Economic Journal 108(447) 383-98.

Eskeland, G.S. and Harrison, A.E. (1997) Moving to Greener Pastures? Multinationals and the Pollution–haven Hypothesis, Policy Research Working Paper 1744, Policy Research Department, Public Economics Division, The World Bank, Washington D.C.

Fredriksson, P.G. (1998) Environmental policy choice: Pollution abatement subsidies Resource and

Energy Economics 20(1)51-63.

Globerman, S. (1979) Foreign Direct Investment and Spillover Efficiency Benefits in Canadian Manufacturing Industries, Canadian Journal of Economics, 12 42-56.

Goulder, L.H., Parry, I.W.H., Williams, R.C,, and Burtraw, D. (1999) The Cost-Effectiveness of Alternative Instruments for Environmental Protection in a Second-Best Setting. Journal of Public Economics 72(3) 329-360.

Gray, W.B., and Shadbegian R.J. (1995) Pollution Abatement Costs, Regulation and Plant-Level Productivity, National Bureau of Economic Research Working paper 4994 NBER: Cambridge, MA.

Gray, W.B., and Shadbegian R.J. (1998) Environmental Regulation, Investment Timing and Technology Choice Journal of Industrial Economics XLVI(2) 235-256.

Grossman, G.M. and E. Helpman (1991) Trade, Knowledge Spillovers, and Growth, European Economic Review, 35, 517-26.

Guinee, J.B., van den Bergh, J.C.J.M., Boelens, J., Fraanje, P.J., Huppes, G., Kandelaars, P.P.A.A.H.,

Lexmond, T.M., Moolenaar, S.W., Olsthoorn, A.A., De Haes, H.A.U., Verkuijlen, E., and van der Voet, E. (1999) Evaluation of risks of metal flows and accumulation in economy and environment. Ecological Economics 30(1), 47-66.

Harris, R.I.D. (1999a) Using the ARD Establishment Level Data to Look at Foreign Ownership and Productivity in the United Kingdom - A Comment mimeograph (available at )

Harris, R.I.D. (1999b) Efficiency in UK Manufacturing 1974-1994, mimeograph (available at

)

Harris, R.I.D. and S. Drinkwater (2000) UK Plant and Machinery Capital Stocks and Plant Closures, Oxford Bulletin of Economics and Statistics, 62, 239-261

Hernandez-Sancho, F., A. Picazo-Tadeo and E. Reig-Martinez (2000) Efficiency and Environmental Regulation: An Application to Spanish Wooden Goods and Furnishings Industry. Environmental & Resource Economics 15, 365-378.

Hymer, S.H. (1976) The International Operations of National Firms, Lexington Books, Lexington, Mass.

Johnson, H.G. (1970) The Efficiency and Welfare Implications of the International Corporation, in C.P. Kindleberger (ed.) The International Corporation, MIT Press, Mass.

Johnson, H.G. (1975) Technology and Economic Interdependence, Macmillan, London.

Jung, C.H. Krutilla, K. Boyd, R. (1996) Incentives for advanced pollution abatement technology at the industry level: An evaluation of policy alternatives. Journal of Environmental Economics and Management 30(1) 95-111.

Kirchhoff, S. (2000) Green business and blue angels - A Model of Voluntary Overcompliance with

Asymmetric Information. Environmental & Resource Economics 15(4), 403-420.

Kindleberger, C.P. (1969) American Business Abroad, Yale University Press, New Haven.

Kneese, A.V., Ayres, R.U., and d’Arge, R.C. (1970) Economics and the Environment : A Materials Balance Approach Resources for the Future: Washington D.C.

Kort, P.M., van Loon, P.J.J.M. and Luptacik, M. (1991) Optimal Dynamic Environmental Policies of a Profit Maximizing Firm Journal of Economics 54(3), 195-225.

Krol, W. and Steil, H.U. (1997) International Activities in Environmental Protection and the Effects on the German Nonferrous Metal Industry: a Review. Metall 51(3), 126-136. [in German]

Krugman, P.R. (1991a) Increasing Returns and Economic Geography, Journal of Political Economy, 99, 483-499

Krugman, P.R. (1991b) Geography and Trade, MIT Press, London.

Laffont, J.J. and Tirole, J. (1994) Environmental Policy, Compliance and Innovation European Economic Review, 38(3-4) 555-562.

Laffont, J.J. and Tirole, J. (1996) Pollution Permits and Compliance Strategies Journal of Public Economics, 1996 62(1-2)85-125.

Marquardt, K. and Nagel, R. (1992) Treatment of Wastewater from the Metal-Working Industry. Chemieingenieurtechnik 64(1), 1-5. [in German]

Maynard, A.D, Northage, C., Hemingway, M., and Bradley, S.D. (1997) Measurement of Short-Term Exposure to Airborne Soluble Platinum in the Platinum Industry. Annals of Occupational Hygiene, 41(1) 77-94.

Moulin, J.J., Wild, P., Romazini, S., Lasfargues, G., Peltier, A., Bozec, C., Deguerry, P., Pellet, F. and Perdrix, A. (1998) Lung Cancer Risk in Hard-Metal Workers American Journal of Epidemiology 48(3), 241-248.

Pearce, D. and Brisson, I. (1993) BATNEEC: The Economics of Technology-Based Environmental Standards with a U.K. Case Illustration. Oxford Review of Economic Policy 9(4), 24-39.

Roberts, M.C. (1996) Metal Use and the World Economy. Resources Policy 22(3) 183-196.

Schwabe, K.A., (1999) The effects of separability on incentive-based instrument performance.

Economics Letters 63(3), 377-380.

Tremmel, G. (1992) Soil Contamination and Rehabilitation of Polluted Abandoned Sites of the Metal

Industry. Metall 46(9) 947-949. [in German]

Venables, A. (1994) Trade Policy under Imperfect Competition: a Numerical Assessment In: P. Krugman and A. Smith. (eds) Empirical Studies of Strategic Trade Policy, Chicago University Press, Chicago.

Vernon, R. (1966) International Investment and International Trade in the Product Cycle, Quarterly Journal of Economics, 80 190-207.

Vernon, R. (1979) The Product Life Cycle Hypothesis in a New International Environment, Oxford

Bulletin of Economics and Statistics 41 255-67.

Appendix

The Data

The UK metal industry panel data used is comprised of the individual records of the Annual Census of Production (ACOP) (now the Annual Business Inquiry). They are available from the UK Office for National Statistics (ONS) branch located in Newport, South Wales. For each year there are two files that can be merged to produce plant level data. One file covers the sample of establishments[9], known as the ‘selected’ file, who were asked questions about financial matters (e.g. amounts spent on capital expenditure, including any pre-production expenditure). The other file contains information (such as employment and ownership structure) on ‘non-selected’ establishments (the remainder of the population). Establishment level data can be ‘spread back’ to plants using employment shares and the unique reference number allocated to each plant.

Using plant-level estimates of capital expenditure (on plant and machinery) based on acquisitions less disposals and including pre-production expenditure, it is possible to estimate the capital stock for each plant. Further, it is possible to do this using the same methods (and length-of-life assumptions) as those used by the ONS when they calculate the ‘official’ estimates for the UK. Plant and machinery price deflators, supplied by the ONS were applied to the data, to produce real gross investment in plant and machinery by industry (see Harris and Drinkwater, 2000, for a discussion).

Estimates of gross value-added, were converted to real prices using 4-digit indices of producer prices (inputs and outputs) provided by the ONS (i.e., we double-deflated using gross output and intermediate outputs to obtain real gross value-added).

Regarding employment data, this was extracted from the individual records of the ACOP. These estimates (together with the estimates for capital expenditure and output) were aggregated to the industry level and compared to the published estimates in the various annual reports of the ACOP. Typically, the margin of difference between the two estimates for ALL manufacturing industry was in the region of 1%. Where differences did occur, this is likely to be due to the fact that the individual returns database can have additional records added after the ACOP summary tables are compiled. We used a more detailed procedure to obtain population weights (based on industries at the 4-digit level sub-divided into size bands where this was possible) and in addition some errors (such as duplicate cases) were discovered in the ACOP database.

Table 1: Definitions of variables used and (weighted) mean and standard deviation values

|Variables |Definition |Mean |S.D. |

|Proportion spending on: | | |

|Post-production capex |Assets used for post-production pollution control |0.096 |0.295 |

| |and waste management | | |

|Process capex |Assets used in production which through improved |0.080 |0.271 |

| |technology reduce pollution | | |

|Current (own staff) |Spending on direct staff, material and operating |0.132 |0.338 |

| |costs for pollution control, treatment and | | |

| |monitoring and waste reduction and management | | |

|Payments to others |Payments to others for treatment and disposal of |0.244 |0.429 |

| |liquid and solid waste | | |

|Total |Any of the above |0.326 |0.469 |

|Amount (log of £’000) spent on: | | |

|Post-production capex |Assets used for post-production pollution control |1.283 |2.233 |

| |and waste management | | |

|Process capex |Assets used in production which through improved |1.297 |2.430 |

| |technology reduce pollution | | |

|Current (own staff) |Spending on direct staff, material and operating |0.906 |2.483 |

| |costs for pollution control, treatment and | | |

| |monitoring and waste reduction and management | | |

|Payments to others |Payments to others for treatment and disposal of |1.038 |2.108 |

| |liquid and solid waste | | |

|Total |Any of the above |1.619 |2.316 |

|Independent variables | | | |

|ln EFF |Frontier production function estimate of technical |-0.622 |0.642 |

| |efficiency (see Harris, 1999b) | | |

|ln GVA |Real gross-value-added in £m 1990 prices |-0.887 |1.827 |

|ln AGE |the age of the plant (i.e. t minus year opened + 1,|2.024 |1.076 |

| |with all plants opening 0 |

| |[pic] |z-value |[pic]x100 |[pic] |z-value |

|(a) Post-production capital expenditure | | | |

|ln EFF |0.031 |0.39 |0.58 |( |( |

|ln GVA |0.102 |3.77 |1.90 |0.798 |19.68 |

|ln AGE |-0.447 |-7.61 |-8.28 |( |( |

|ln KL |0.303 |7.29 |5.62 |( |( |

|ln DEN |-0.090 |-3.01 |-1.68 |( |( |

|t |( |( |( |-0.134 |-1.64 |

|EU |0.131 |0.77 |2.58 |0.950 |2.63 |

|US |0.523 |3.67 |11.84 |0.554 |1.66 |

|AUS |0.526 |3.30 |12.11 |-0.305 |-0.96 |

|SIC 2220 |0.498 |3.48 |10.80 |0.474 |1.40 |

|SIC 2234 |-0.076 |-0.54 |-1.37 |0.118 |0.31 |

|SIC 2235 |-0.576 |-2.15 |-8.16 |1.373 |1.80 |

|SIC 2245 |0.542 |4.16 |11.98 |0.804 |2.59 |

|SIC 2246 |0.998 |6.14 |25.24 |0.733 |2.14 |

|SIC 2247 |0.806 |5.82 |19.45 |1.202 |3.77 |

|Constant |1.040 |3.36 |( |2.334 |4.39 |

|( |( |( |( |-0.556 |-2.88 |

|( |( |( |( |1.405 |3.58 |

|( |( |( |( |-0.781 |-2.80 |

| | | | | | |

|Log L |-1061.355 | |N |1912 | |

|[pic] |0.132 | |Censored N |1608 | |

Cont…

| | | | | | |

|(b) Process capital expenditure | | | |

|ln EFF |0.165 |1.89 |3.07 |( |( |

|ln GVA |0.055 |1.94 |1.03 |0.892 |23.65 |

|ln AGE |-0.451 |-7.31 |-8.38 |( |( |

|ln KL |0.363 |8.30 |6.74 |( |( |

|ln DEN |-0.149 |-4.59 |-2.77 |( |( |

|t |( |( |( |-0.203 |-2.76 |

|EU |0.113 |0.67 |2.20 |-0.196 |-0.55 |

|US |-0.800 |-3.74 |-10.33 |-1.558 |-2.43 |

|AUS |0.246 |1.48 |5.08 |0.109 |0.35 |

|SIC 2220 |0.139 |0.94 |2.70 |0.753 |2.53 |

|SIC 2234 |-0.722 |-4.60 |-10.67 |1.666 |3.13 |

|SIC 2235 |0.308 |1.84 |6.40 |0.919 |3.07 |

|SIC 2245 |-0.092 |-0.65 |-1.65 |1.758 |5.43 |

|SIC 2246 |0.236 |1.28 |4.77 |2.094 |6.29 |

|SIC 2247 |0.137 |0.88 |2.68 |1.850 |6.83 |

|Constant |2.021 |6.30 |( |2.024 |4.16 |

|( |( |( |( |-0.505 |-1.92 |

|( |( |( |( |1.330 |2.67 |

|( |( |( |( |-0.671 |-1.90 |

| | | | | | |

|Log L |-1044.655 |N |1912 | | |

|[pic] |0.132 |Censored N |1627 | | |

| | | | |

|(c) Current expenditure on staff, etc. | | | |

|ln EFF |-0.277 |-4.19 |-6.84 |( |( |

|ln GVA |0.110 |4.31 |2.72 |0.908 |27.48 |

|ln AGE |-0.282 |-5.30 |-6.97 |( |( |

|ln KL |0.192 |5.34 |4.74 |( |( |

|ln DEN |-0.152 |-5.38 |-3.77 |( |( |

|t |( |( |( |-0.067 |1.13 |

|EU |0.200 |1.33 |5.24 |-0.773 |-2.90 |

|US |-0.632 |-3.72 |-12.73 |-0.457 |-1.08 |

|AUS |0.885 |5.80 |26.66 |0.954 |4.86 |

|SIC 2220 |0.045 |0.34 |1.11 |-0.909 |-3.65 |

|SIC 2234 |-0.701 |-5.41 |-15.25 |1.018 |2.84 |

|SIC 2235 |-0.557 |-2.96 |-11.60 |0.586 |1.26 |

|SIC 2245 |0.051 |0.42 |1.26 |0.394 |1.78 |

|SIC 2246 |1.017 |6.51 |30.79 |-0.692 |-3.44 |

|SIC 2247 |0.549 |3.95 |15.33 |0.825 |4.03 |

|Constant |0.934 |3.40 |( |2.374 |7.42 |

|( |( |( |( |-0.553 |-3.02 |

|( |( |( |( |1.374 |4.05 |

|( |( |( |( |-0.760 |-3.02 |

| | | | | | |

|Log L |-1502.077 |n |1912 | | |

|[pic] |0.214 |Censored N |1430 | | |

Cont.

|( d ) Payments to others | | | |

|ln EFF |-0.147 |-2.39 |-4.77 |( |( |

|ln GVA |0.104 |4.46 |3.40 |0.967 |34.77 |

|ln AGE |-0.195 |-4.18 |-6.37 |( |( |

|ln KL |0.142 |4.49 |4.62 |( |( |

|ln DEN |-0.038 |-1.55 |-1.23 |( |( |

|t |( |( |( |-0.157 |-3.88 |

|EU |0.371 |0.25 |1.21 |1.058 |5.58 |

|US |-1.125 |-6.08 |-29.05 |-0.896 |-2.26 |

|AUS |0.420 |2.83 |14.20 |1.008 |5.14 |

|SIC 2220 |0.639 |5.20 |21.72 |0.464 |2.67 |

|SIC 2234 |-0.478 |-4.15 |-15.67 |0.195 |0.91 |

|SIC 2235 |0.060 |0.43 |1.99 |0.395 |1.85 |

|SIC 2245 |0.700 |6.08 |24.17 |0.479 |2.77 |

|SIC 2246 |0.639 |4.29 |21.63 |0.981 |5.17 |

|SIC 2247 |0.500 |3.80 |16.92 |0.449 |2.49 |

|Constant |0.440 |1.91 |( |0.344 |1.44 |

|( |( |( |( |-0.624 |-5.52 |

|( |( |( |( |1.289 |14.72 |

|( |( |( |( |-0.805 |4.08 |

| | | | | | |

|Log L |-2198.781 |N |1912 | | |

|[pic] |0.375 |Censored N |1082 | | |

| | | | | | |

Figure 1; Technology Choice, Ownership and Productivity

Figure 2: Ownership, Resource Productivity and Capital Intensity

Figure 3: Average (real) pollution control expenditure per plant*, 1991-1994, by industry: all manufacturing

* includes only plants with positive expenditure Data weighted by population weights

Note, the (population weighted) estimate of total spending in all industries was £738.1 million p.a. for the 1991-94 period.

Figure 4: Average (real) expenditure p.a. on pollution expenditure control* (and percentage of plants with positive expenditure), 1991-1994: all manufacturing

* includes only plants with positive expenditure Data weighted by population weights

Figure 5: Average (real) expenditure per plant*, 1991-1994, by type and ownership group: all manufacturing

* includes only plants with positive expenditure Data weighted by population weights

Figure 6: Average (real) expenditure p.a. on pollution expenditure control* (and percentage of plants with positive expenditure), 1991-1994: metal industry

* includes only plants with positive expenditure Data weighted by population weights

Figure 7: Average (real) expenditure per plant*, 1991-1994, by type and ownership group: metal industry

* includes only plants with positive expenditure Data weighted by population weights

-----------------------

[1] See, for example Cornwell and Costanza (1994), Laffont and Tirole (1994, 1996), Damania (1996), Jung et al. (1996), Fredriksson (1998), Goulder et al. (1999), Schwabe (1999), Baudry (2000).

[2] See, for example, Dunning (1958, 1977), Vernon (1966, 1979), Kindleberger (1969), Caves (1974), Johnson (1970,1975) Hymer (1976), Buckley (1983).

[3] See, for example, Globerman (1979), Krugman (1991ab), Grossman and Helpman (1991), Venables (1994), Edwards (1998), Aghion and Howitt (1998)

[4] Another potential influence is the location of the plant - this is discussed in section 4 when the model for estimation is presented.

[5] Variables are formally defined in Table 1

[6] Actually expenditure was deflated by the 4-digit producer price index for the industry to which the plant belonged.

[7] Note, the impact of sample selection is therefore obtained via the two estimated coefficients (((. Typically, this is reported in most studies (and in most econometrics packages) via a single term, usually denoted as ( (=(((). This is also reported below, as well as the separate terms that comprise it.

[8] While the raw data shows that EU-owned plants spent more on PAE, after controlling for such factors as age and capital-intensity, the econometric model estimated shows that EU-owned plants are no more likely to incur any expenditure than are UK-owned plants.

[9] Establishments are either single plants or they make a return that covers several plants –details and definitions are provided in the introductory notes for each annual census.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download