British Leyland not keeping up with the Jones': competition and survival in the UK car market (1971-1998)

 

James Walker, London School of Economics

(J.T.Walker@lse.ac.uk)

Supervisor: Dr Peter Howlett

 

         

1. Introduction

Leonard Lord, the chairman of British Motor Company stated that if “you make proper bloody products, then they will sell themselves”.[1] Lord was not alone in his view that the nature of the products was an issue central to the success of British car manufactures. Considerable contemporary analyses were undertaken on the car industry both by government agencies,[2] and by contemporary researchers,[3] with each of these studies emphasizing that a key failing of the domestic industry’s flagship, British Leyland (BL hereafter),[4] was its inability to provide an adequate “product package” that led to what the Central Policy Review (1975) termed a ‘product-led decline’.

            The Central Policy Review’s concept of ‘product-led decline’ encapsulated three distinct dimensions of competition and technology, and was widely cross-referenced in the aforementioned studies. The first dimension was that BL's cars were uncompetitive relative to rival manufactures in the “product package” they provided. Second, that BL manufactures were unable to fill the domestic market place with a range of products differentiated in ways that reflected consumers' taste and demand patterns. Furthermore, BL was accused of marketing multiple brands and models with qualities that were “too similar” to each other thus not obtaining the potential advantages associated to strategically locating and dispersing its brand’s offerings. Third, that BL's cars were outdated and that BL, unlike its foreign owned counterparts, neglected to regularly upgrade its product ranges by embodying new technological advances.[5]

            While there was a body of contemporary work arguing that innovation and competition played an important role in the declining domestic role in the UK car market, the historical literature has paid limited attention to the products themselves. At best the historical literature makes assertions based on the failed sales success of particular British model(s) focusing on a wide range of alternative factors. While different authors provide different emphases on these factors, each analysis concerns three sets of actors in the market: employees and a poor industrial relations record; government intervention in the industry; and the ability of management to provide a coherent strategy to cope with rising international competition.[6]

Foreman-Peck et al. (1995) have pointed out the literature on the British car industry has taken a two approaches. The first is to list all the plausible ‘factors’ that have influenced the industry and in some cases to attribute a causal relations between them. The second is to take a more ambitious line of picking a single explanation and gathering a selection of evidence consistent with it. The weakness of the second method is however that the scope of the explanation is limited to the particular facet. From an empirical perspective these literature provide a series of anecdotally derived conjectures that a priori are quite compelling but the contribution of each argument is difficult to assess.

This paper operationalises the notion of product led-decline in the evolution of the market to access managerial effectiveness at the strategic level of the firm in relation to the survival of its products. Hence, the singular focus of my research falls into the second approach to examining the industry.  However, the nature and design of models and model variants is determined in the design and development phases of model development through the interaction of design, engineering, and marketing department of a car firm with oversight and resources directed by management. It is unlikely that the mainstream arguments concerning the fate of BL, other then a lack of management acumen, have a direct role in the product creation and strategic location processes. Poor employee relations and resulting labour disputes were confined to the shop floor and so do not have an important role in determining the departments involved with designing the products themselves. Nor is there is no evidence of Government intervening in the running of the firm as can be seen through the comments successive heads of the company. Indeed Stokes, BL's Chairman between 1967-75, pointed out that “When running the business we had very little pressure from government”. Similarly, Edwardes, Chairman and CEO between 1977-1983 took on the role on the condition that he had complete autonomy.[7]

            My focus on product-led decline serves a duel purpose. First, it provides the first all-encompassing examination of phenomena in the UK car market. Second, and more generally, the analysis allows me to examine the forces shaping the precipitous decline in British industrial competitiveness that scholars of British economic history have termed the ‘British disease’ in the industry which has taken the mantle as the ultimate metaphor of the phenomena in the post-war era.[8]

            To analyse the roles of competition and innovation in shaping the survival of the products of BL and other participants in the UK car market considerable information on the evolution of the prices of a complete panel of new cars models and model variants between 1971-98 is brought to bear. I utilise an extensive newly constructed data set that was assembled from trade publications that provide a means to capture shifts in the structure of the market and to concentrate on the upgrading though the incorporation of about 120 embodied product characteristics. Those data are combined with information on the firm, brand, and market niche of each model. A novel aspect of the data set is that I capture characteristics that make up the set of model variants over time allowing me to chart individual histories of the over 600 models that are marketed during the period examined.

            The remainder of the outline to the paper is organised as follows. Section 3 examines the contrasting experience of British Leyland to that of the market as a whole through the evolution of the entry and exit of models, market structure, and innovation, which form the three building blocks of the analysis. Evidence of PLD is tested in Section 4 - Appendix 1 defines how measures of product-led-decline are calculated. Finally, Section 5 contains a concluding discussion.

 

2. British Leyland and the Evolution of the Market

 

Figure 1: Launch and discontinuation of models, 1971-98

 

A: All market participants                

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

B: British Leyland

 

 

 

           

 

 

 

 

 

 

 

 

 

 

While there had been a steady growth in car imports prior to 1970, the stranglehold of domestic producers in the UK car market was abruptly released during the 1970s and early 1980s as is reflected in the unprecedented growth in import penetration.[9] Examining the survival of models, or in BL’s case their tendency towards extinction, according to the product-led decline hypothesis was determined by the effects of innovation and competition both within BL’s product range and relative to the offerings of her rivals. Before turning to the empirical methodology and analysis I briefly examine the evolution of the three key ingredients to research product-led decline that concern: product survival, competition, and product innovation, evaluating BL within the context of the evolving market.

Figure 1 illustrates the entry and exit patterns of car models in the UK market between 1971 and 1998 used to derive the survival time of models from data derived from Parkers’ Guide to New and Used Prices and the Motorists’ Guide to New and Used Car Prices. Model entry rates dominated exit rates for the majority of the period with a net expansion in the number of models in the market occurring in 21 of the 28 years. The upward trend in the stock of new models marketed was punctuated due to a rise in exit and a fall of entry between 1974 and 1976, following the first oil crisis in 1973, and due to a reduction in entry in the late 1980s. In addition, entry and exit rates follow a quite regular pattern reflecting the mature nature of the market under analysis.

The second panel of Figure 1 examines the entry and exit of BL’s models and illustrates that the firm moved in the opposite direction to the market with exit dominating entry in eighteen of the nineteen years prior to the completion of rationalisation of the firm’s model range in 1987.

Figure 2 plots annual counts for four tiers of aggregation (firm, brand, model and model variants) within the UK market. While there was little change in the number of firms and brands operating in the market the graph illustrates a dramatic expansion in models, whose numbers roughly triple. More spectacular than the expansion of models marketed over the period is in the precipitous rise in the number of model variants that expand by a multiple of six. It is perhaps not surprising to find that there was an expansion in the number of models marketed in the early 1970s as European firms expanded their presence in the market. That these trends were highly persistent throughout the period, while the stock of brands and firms remained almost constant, suggests that there has been a structural shift in the product strategies of firms towards a greater cannibalisation of their products within a more fragmented market.[10] The rationalisation at British Leyland directly contradicted those trends.[11]

To reasonably capture technological innovation and “quality upgrading” for a product as complex and multi-dimensional as the car required me to record a wide selection of innovations that occurred over the extended snap shot examined.  Appendix 2 provides a time line of innovations for 125 embodied attributes as well as the brand that first included each new attribute as “standard” in at least one model variant.[12] Technological innovation has clearly been an on-going feature of the car industry as it has evolved over the post-war period. Appendix 2 indicates that technological leadership was limited to a small sub-set of firms. European firms account for over 75 per cent of product innovation introductions, with the four top

 

Figure 2: Counts of firms, brands, models and model variants

 

A: All market participants               

 

 

 

 

 

 

 

 

 

 

 

 

 

 

           

B: British Leyland

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

innovating firms being  BMW, Mercedes; Rolls Royce and Jaguar, which account for 35 per cent of introductions. Japanese firms comprise 16 per cent and US-based firms for 8 per cent of introductions, respectively. While European firms dominate innovation introduction, a noteworthy feature of Appendix 2 is there is a considerable shift between UK and continental producers over the period. Prior to 1970, UK manufactures introduced 60 per cent of the new innovations in the sample, but this proportion fell steadily from the 1970s to 13 per cent in the 1990s, while European producers introduce 18 per cent of new product innovations prior to 1970 but account for 80 per cent of the total in the 1990s.

            It is also of note that, other then Jaguar and Land Rover, the brands that made up BL were rarely the first to incorporate new product technologies, providing some indication that the technological capabilities of BL mass producing brands was limited.[13] Since the technological expertise at Land Rover was specific to 4-by-4 vehicles, Jaguar was the viable candidate to obtain technological insights from. Indeed, Barber, Vice Chair of BL in 1973 recognised that “If you look at what we were good at in those days, our volume-cars were not up to world standards. If we could have capitalised on Jaguar we could have done something”.[14]                     

The structure of research at BL was however not designed to encourage collaboration and it was a stated policy of the firm to maintain rivalry in engineering systems, methods and products in order to preserved the separate ‘identities’ of the its brands. Whip and Clark (1986), who provide the most detailed study of the firm’s product innovation process, argue that management’s unrealistic market expectations and the firms’ isolated planning, production, and design activities added to problems with the firms’ inadequate engineering capabilities.

There is also evidence that BL’s brand were hostile to collaboration with the engineering director at Jaguar stating that he “used to fight with cars” in his desire to maintain engineering independence from other BL manufactures.[15] A striking example is found in the production of Rover's most the ambitious project, the P6BS, which was developed independently of Jaguar, being shelved six months prior to its launch in 1972.[16]

 

3. Pinning Down Product-Led-Decline: Hypotheses

The first dimension of the product-led-decline hypothesis was that BL's cars were uncompetitive relative to rival manufactures in the ''product package'' they provided.[17] If BL’s cars were uncompetitive we should expect that competition by rivals reduce the survival of British Leyland’s products more then it would affect BL’s rivals. Therefore,

PLD Hypothesis 1a (Market) - Rival Competition: Competition though rivals 'crowding' the product spectrum will reduce the life of a model.

But that,

PLD Hypothesis (British Leyland) 1b - Rival Competition: The effects of competition on the survival British Leyland models were greater then for other market participants.

            As has been determined, the UK car market encompasses a large number of cars, which are differentiated from each other by each car maker. The second of the PLD hypotheses emphasises two distinct levels where British Leyland failed to differentiate its products. The first corresponds to the common-sense notion that differentiating a product reduces the degree of competition from rivals’ offerings allowing firms to avoid the brunt of price competition.

PLD Hypothesis 1b (Market) - Competition Counterforce: Product differentiation, and the degree of that differentiation, by mitigating the degree of competition, prolongs the life of a model.

However, if the degree of quality differentiation at British Leyland was low relative to participants, and hence would not have sheltered BL’s products, and reduced the likelihood that they would remain in the market, then

PLD Hypothesis 1b (British Leyland) - Competition Counterforce: A failure to differentiate meant that BL was not able to avoid the brunt of competition, and hence did not affect the survival of its products.

The second aspect of the PLD hypothesis is subtler, reflecting the multi-brand, multi-model nature of the industry. By choosing to cannibalise their product, a firm trades-off the loss of market share of its incumbent model against an expansion in the combined sales associated with marketing more than one model. The insight of PLD is that it points out that the degree to which firms compete internally, by offering multiple products, can be mitigated through the extent and degree that a firm is able to differentiate its set of product. Hence

PLD Hypothesis 2a (Market): Within Firm Competition: Cannibalisation between models of the same firm reduces the survival of existing models due to within firm competition.

PLD Hypothesis 2b (Market): Within Firm Competitive Counterforce: Successful differentiation of the set of firm’s models will prolong the expected life of existing models.

            The PLD hypothesis however argues that British Leyland as a multi-brand multi-model firm engaged in cannibalisation it did not differentiated the range of its models from each other thus:

PLD Hypothesis 2b (British Leyland): Cannibalisation between models of the same firm reduces the survival of existing models due to within firm competition.

But that

PLD Hypothesis 2b (British Leyland): Within Firm Competitive Counterforce: Successful differentiation of the set of firm’s models will prolong the expected life of existing models.

The third PLD hypothesis argues that British Leyland did not upgrade it products. As I have highlighted earlier, technological progress has been an on-going feature of the market and there have been repeated claims in motoring publications that BL’s products have not kept pace with the market.

Model Upgrading (Market) 3: Quality upgrading by maintaining a model in line with the market standard prolongs the life of products.

 

4. Testing the PLD Hypotheses

To investigate to what extent PLD affected the products of British Leyland relative to those of other market participants I undertake a survival analysis. Survival – in my case the number of periods that a car survives in the market - is modelled as a variable t and is an observation of a random variable T. The random variable is assumed to have a density function, f(t), and a distribution function, F(t), which defines the survival function, S(t)=1-F(t). The survival function shows the probability that the car model survives at least t periods. From the survival the probability of exit each period, commonly termed the hazard function, can be defined as , i.e. the rate at which a spell is completed after duration t conditional that it has survived until t. In this formulation the hazard function is simply a function of time, however it is reasonable to assume that the hazard rate is a function of PLD as well as model specific factors such as the age of the car, the segment where it is marketed, the period when the car was first developed. I estimate, as my workhorse model, the piecewise constant exponential model,

where  is the baseline hazard function, X is a vector of explanatory variables, and  is a vector of parameters. My preference for using the piecewise constant exponential model reflects the flexibility of the model and that I wish to quantify the extent to which the three hypotheses underlying the PLD hold.[18] The data set analysed contains 652 models, over the 28 year ‘spells’, amounting to 4,714 model-spell observations. Correlations between the dependent variable and the key independent variables related to the hypotheses to be tested are provided in Appendix 3.

Maximum likelihood estimates of the model are reported in Appendix 4. The initial specification explores aspects of market structure before turning to the central issues of competition and quality upgrading associated to the PLD hypothesis.

Specification 1 incorporates age, cohort, firm entry, and segment effects. The age effects, which are referenced by models that have been launched for less than one year, follow a non-linear pattern. I also include dummy variables for year of introduction, grouped into four-year cohorts. Although the coefficients suggest that the product-life-cycle of cars to be shortening over time, they are not statistically significant.  Finally I include segment effects taking the mini-segment as the omitted reference group. The coefficients on the established segments (small family, medium, large, luxury and sports) are positive and significant, while the ones of the  "new" segments (jeep and minivan) behave similarly to cars in the mini segment.

Specifications 2-4 assesses the relative impacts of the PLD’s three dimensions on the survival of models at BL between 1971-87 relative to the market as a whole.[19]

The relative effects of internal (cannibalisation) and rival competition in the market and at BL are examined in specification 2.[20] Rivalry can be seen to have had a weakly significant impact on the market, but a stronger effect on the survival of BL’s products with its coefficient being well determined and significantly larger then for the market as a whole. While rival competition was a less important force at the market level cannibalisation was a more pervasive factor. A complication in interpreting the cannibalisation variable is that it may simply capture the effects of product renewal as obsolescent products are discarded. To separate these effects I include a product renew dummy that takes value of one when a brand introduces a new model in a particular year and segment.  Product renewal raises the exit hazard, but its inclusion does not alter the effects of cannibalisation.

That cannibalisation has been a dominant force in the car market is perhaps not surprising given the considerable rise in the number of models and model variants the market characterised earlier in Figure 2. Similarly, internal competition does not appear to have been a driving force of model survival at BL whose model exits were strongly determined by attempts to renew its product range.

Specification 3 tests examines the effects of the counter-forces to competition in the form of how well BL and other markets differentiate their individual models (differentiation_model) and model ranges (differentiate_firm) when the first appear in the market, and the degree to which each model and model range is differentiated over time (dispersion_model and dispersion_firm). The PLD hypothesis suggests that BL failed to both differentiate it products or to provide a sufficient degree of direction relative to the market. The results suggest that this was indeed the case but that it was not the initial differentiation of models, but the ability of the BL to disperse its product range with the positive effect of the firm’s dispersion compared to the negative effects in the market as a whole being significantly lower.

Specification 4 examines hypothesis 3: that an inability at BL’s models to keep pace with the market. I find that while upgrading at BL had the effect of enhancing the life of models the impact was relatively minor.

 

 

5. Concluding discussion

The explanation of the post-war decline of British industry has been a central question and stimulant of historical research. I argue that an important root cause of the demise of Britain’s once dominant car maker, BL, was the poor quality of its products. Without a viable product range to provide to an increasingly discerning market BL’s ability to compete with the ever expanding product ranges of its rivals, even in the market where it should be most competitive, the home market, was limited.

Despite competitive pressure BL attempted maintained its product range but was unable to obviate competitive effects through dispersing and differentiating its products. In a market characterised by increasing product cannibalisation and dispersion of models and model ranges in order to avoid the brunt of competition BL bucked the trend. Finally, a lack of quality upgrading of BL’s products, relative to its competitors, meant that its products survival was halved, leading to costly renewal of products. In an industry where the sunk costs of product are high the ability of a BL to maintain its profitability and market dominance was negligible.

 

References (abbreviated)

Allan, G.C. (1976), The British Disease, London

Central Policy Review (1975), The Future of the British Car Industry

Church, R. (1994), Rise and Decline of the British Motor Vehicle Industry, Basingstoke

Edwardes, M. (1983), Back from the Brink: An Apocalyptic Experience, London: Collins

Foreman-Peck, J., Bowden, S., and A McKinlay (1995), The British Motor Industry, Manchester: Manchester University Press

Hilton, M. (2003), ‘The Fable of the Sheep; or, Private Virtues, Public Vices. The Consumer Revolution of the Twentieth Century’, Past and Present (forthcoming)

Kiefer, N.M. (1988), “Economic Duration Data and Hazard Functions”, Journal of Economic Literature 26, 646-679

The Motorist Guide to New and Used Car Prices, Blackfriars Press Ltd., 1965-1993

National Enterprise Board (1977), Guide Lines

National Enterprise Board (1978), Report on the British Leyland Corporate Plan, up to 1981, and Business Plan for 1978

Parker's Guide to New and Used Car Prices, Parker Mead Ltd., 1972-2001

Porter, P. (1987), The Jaguar Project XJ40, Haynes

Ryder et al. (1977), British Leyland: The Next Decade, (London)

Society of Motor Manufactures and Traders (1965-1973), New Registrations of New Cars by Make and Model Line, Vans and Hackneys by Make and Other Vehicles by Make and Unladen weight: in Great Britain, Northern Ireland and the Isle of Man

SMMT (1973-1979), Motorstat NR2

SMMT (1980-1998), UK New Vehicle Registrations

Stavins, J. (1995), ‘Model Entry and Exit in a Differentiated-Product Industry: The Personal Computer Market’, The Review of Economics and Statistics 77, 571-584.

Walker, J.T. (2003), ‘British Lemons? Quality and Price in the British Car Industry’, unpublished manuscript, LSE

Whip, R. and P. Clark (1986), Innovation and Auto Industry, Frances Pinter (London)

Williams, K., Williams, J. and C. Haslam (1987), The Break Down of Austin-Rover: A Case Study in the Failure of Business Strategy and Industrial Policy, (Leamington Spa)

Williams, K., Williams, J., and D. Thomas (1983), Why are the British so Bad at Manufacturing?, (London, 1983)

Wood, J. (1988), The Wheels of Misfortune: The Rise and Fall of the British Motor Car Industry, (London)

 

 

 

Appendix 1 Method to Calculate Product-Led-Decline Spatial Measures of Quality

 

2.1 Calculating ‘Quality’

To determine overall value of each of the variants, v of each model m produced by firm f in year t, the pooled hedonic regression being specified as

                                       (1)

where z are vertical differentiated attributes. In addition to the attribute-based variables, I have included a set of dummies for time (), segment (), firm (), variables for model and variant age (), and an i.i.d. error term (). Because Equation 1 uses model variants as the unit of aggregation I am able to capture cross-section variation between variants of the same model, and also to exploit the upgrading of models over time through the adoption of new model variants embodying vertically differentiated product innovations and the upgrade of established versions. The quality index is used to derive all the indices used to test the PLD hypothesis is detailed below.

 

2. 2 Calculating Spatial and Temporal Quality Indices

This appendix illustrates the construction of the within and between firm quality based measures used to evaluate the product-led-decline hypotheses. The first involves developing quality indices for between model (‘model differentiation’ and ‘model dispersion’) and within firm strategic location effects (‘firm differentiation’ and ‘firm dispersion’) that are used to evaluate the two dimension of the second PLD hypothesis. The second is a dynamic measure of quality ‘upgrading’ of models over time to evaluate the third PLD hypothesis.

 

Measuring differentiation[21]

I first construct a distance measure for each variant with respect to other rivals’ variants in the segment but excluding own firms models,

 

where  is the rivals’ variant k in segment s at time t-1 (the year before the introduction of the new model), and i is the number of rivals‘ variants. Since a model can have multiple variants, a model-level differentiation index is constructed as the mean of the distance indices of its variants when the model first enters the market,[22]

To construct of the within-firm differentiation variable proceeds in a similar fashion to its model-level equivalent. I first calculate a distance measure with respect to other firms’ own models, 

where  is a firm variant i in segment s at time t-1 (the year before the introduction of the new model), and N is the number of firms’ variants. Since a model can have multiple variants, I construct a within-firm differentiation index as the mean of the distance indices of its variants when the model first enters the market,

 

Measuring dispersion

The proposed hypotheses in Section 2 suggest that dispersion will effect the survival of models at both model and firm levels. To measure both dispersion effects I utilise cross-section variation in the variants’ quality index in each year. To do so I first calculate a model dispersion index as

                     

where  is the average quality index of model m, , and  is the number of variants of model m at time t. Next I calculate the model relative dispersion index (MRDI), which is obtained after normalising  by the segment dispersion index,

so obtaining

Next, the firm relative dispersion index (BRDI) is calculated in a similar fashion: by comparing the within-firm quality dispersion to the total dispersion of variants competing in each segment.

where  is the average quality index of firm f in each segment s,  and  is the total number of variants of the firm in segment s. Hence, the BRDI is calculated as,

using the definition of  in the denominator again. Thus, for each firm, the BRDI encapsulates the dispersion of their own models in a particular segment.

 

Measuring upgrading

The distance-based measure of upgrading by the ability of the firm to enhance its products between periods. Model upgrading is determined by

 

where  and  is the number of variants of model m of firm f at times t and t – 1 respectively. Since the data is taken at the variant level, the model upgrade measure of a model is the difference between the average of variants qualities making up the model between periods. However, while upgrading is a one-way process with larger proportions of “high quality” features being embodied over time, a model’s success in the market is based not only on a firm’s ability to upgrade each model but also on its rival’s ability to embody new features. So, while upgrading within a model is always upward, this need not be the case relative to the market. To capture this I normalise the upgrading rate of models against the general upgrading taking place within the segment where each model resides,

 

where s is the car segment and  is the number of variants of all models in relevant sub-market segment s at time t.[23] Upgrading is therefore defined as

  

 

Appendix 2     Time line of product innovations

Appendix 1 (continued)          Time line of product innovations

 

Appendix 3  Determinents of model survival

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: (1) t statistics in parenthesis (2) British Leyland accesses between 1971-1988, since no effects of the PLD hypothesis were discerned there after.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: (1) t statistics in parenthesis (2) British Leyland accesses between 1971-1988, since no effects of the PLD hypothesis were discerned there after.