Catching up and Falling Behind

Catching up and falling behind?

Explanations for Post-war comparative performance

1

INTRODUCTION

This lecture will consider four approaches to growth over time, and how growth varies between countries. It will provide a theoretical but non-technical summary of these arguments and examine how economists and economic historians have used these explanations.

1.Neo-classical growth theory.

2.The Post-1973 Slowdown in TFP Growth.

3.New Growth Theory.

4.Convergence and catch-up.

5.Institutional Sclerosis.

6.Conclusion.

NEO-CLASSICAL GROWTH THEORY

1.At the heart of the neo-classical approach to the understanding of growth rate differences between countries is the concept of the aggregate production function.

This is a technological relationship which expresses the level of output as a function of the level of inputs, such as labour, capital, and land, together with other factors such as the state of technology, the degree of managerial and labour efficiency, etc.

In practice, the arguments of the production function are normally confined to just three factors, namely labour, capital and technology. The basis of the production function is the individual production process at the plant level, and it is assume that these micro-production functions can be aggregated to produce a well-defined relationship at the aggregate level, i.e. for the whole economy, for manufacturing or for individual industries.

2.In the neo-classical model, the level of output is assumed to increase over time because of technical progress, even if the volume of factor inputs remains constant. Technical progress, or total factor productivity, is a residual.

Total factor productivity (TFP) is crucial to the neo-classical model. TFP is the growth of output less the growth of labour and capital, each weighted by its factor share. The advantage of using TFP over labour productivity as a measure of the rate of technical progress is that it explicitly makes an allowance for the growth of the capital stock. Using the growth of labour productivity as a measure of the rate of change in economic efficiency is that it is clearly only a partial index. For example, a rapid growth of labour productivity could be primarily due to a rapid increase in the amount of machinery with which each employee has to work. The growth of total factor productivity makes an explicit allowance for this.

3.The two original articles on neo-classical growth economics are Solow’s (1957) article entitled ‘Technical Change and the Aggregate Production Function’ and Abramovitz’s (1956) article entitled ‘Resource and Output Trends in the United States since 1870’. Making the usual neo-classical assumptions, Solow found that output per hour doubled in the US between 1909 and 1949, and of this over 80 per cent could be attributable to ‘technical change in the broadest sense’ while less than 20 per cent could be attributed to the growth in the capital-labour ratio.

The seminal full-blown growth accounting study is Denison’s (1967) Why Growth Rates Differ, which was a detailed study of the sources of growth of eight European countries and the United States over the period 1950-62. Denison has since used this approach to examine the productivity slowdown since the early 1970s (1974, 1979, 1984, 1985). The technique has also been used to examine the sources of growth for the UK in a historical context by Matthews, Feinstein and Odling-Smee (1982).

All of these studies, then, are attempting to reduce the residual (which, after all, is nothing more than a measure of our ignorance). Some neo-classical purists have sought only to augment the estimates of the growth of the inputs and to thereby reduce the growth of total factor productivity – see Jorgenson and Griliches (1967) and Jorgenson, Gallop and Fraumeni (1987). Others, especially Denison, have adopted a far more pragmatic approach by also allowing for increasing returns, for disequilibrium factors such as the intersectoral movement of labour.

4.Table 15a reports a summary of Denison’s (1967) estimates of the sources of growth for eleven countries during the Golden Age, although for eight countries the terminal date is 1962. These figures are supplemented with the results of a study by Kendrick (1981) in Table 15b. Kendrick uses Gross Business Product because of the difficulties associated with the measurement of output in the service sector.

We can note that the growth of TFP explains by far the largest part of the growth of output. On the basis of Denison’s results, the growth of TFP (R1) on average accounts for 58 per cent of the growth of output. This ranges from 74 per cent (France) to 40 per cent (Canada). A similar picture emerges from Kendrick’s results.

5.A major innovation of Denison (1962) was to adjust the labour force for quality changes and this has now become standard practice in the growth accounting literature. It has long been recognised (at least by economic historians) that education is not merely a consumption good, but also represents an investment in human capital. General schooling, while not necessarily imparting vocational skills, has the advantage of making the labour force more flexible in that it can obtain greater benefit from any subsequent apprenticeships and on the job training, as well as being more responsive to new ideas and techniques of production. Education thus tends to raise the efficiency of labour by improving its quality.

The post-war period saw a rapid expansion in the amount of schooling received by the average worker. The amount of years spent by the representative person in the education system rose steadily, both through choice as per capita income rose and because of legislation which raised the minimum school-leaving age in a number of countries.

Denison constructed an index of the labour input which takes into account the increasing quality. The estimates of improvement in the quality of the labour force are relatively small when compared with the size of TFP growth in most countries. In the case of the US, Denison estimated that it contributed 0.49 percentage points to the growth of output over 1950-62. The growth of education as not so fast in the other advanced countries over this period. For Northwest Europe, the contribution was 0.23 percentage points – which accounted for 5 per cent of the growth of national income.

6.One of the least understood aspects of the growth process is what determines the pace of technical change. In his 1967 study, Denison circumvented the conceptual problem of how to measure the contribution of technical progress. He simply assumed that the contribution of ‘advances in knowledge’ was the value of the residual of the most advanced country which was the United States, and this contribution to output growth (some 0.8 percentage points per annum) was imputed to all the other countries. The residual defined as ‘the unexplained growth of output’ was therefore zero for the US. Hardly a satisfactory procedure.

A moment’s reflection should be enough to remind us that some of the most spectacular technological advances – such as those generated by the NASA space programme or the Concorde project – were not accomplished without a considerable research cost. Considerable expenditure is devoted to investment in research and development (R&D), whether it is on the salaries of designers and scientists, on their equipment or on the finance of the various prototypes. It is possible to extend the concept of the production function with these expenditures being considered as ‘inputs’ and the value of improvements or new products due to this knowledge as ‘output’. By its very nature, production of new knowledge, per se, is a very risky and uncertain business and much expenditure has to be written off with very little to show for it.

Denison was very sceptical about the importance of R&D in his 1967 study. He says:

The purpose of most – probably the bulk – of R&D expenditures, moreover, is such that it does not affect the growth rate no matter how successful it may be . . . [T]here is no way of knowing whether those fruits of organised research that do affect the measured growth rate contribute 0.1, 0.7, or any intermediate amount to the total contribution of advanced of knowledge. The fact that expenditures for research and development have expanded so much in the United States while estimates of the contribution of advances of knowledge have not, may suggest that organised R&D is not very important to growth. But even this remark must be qualified by noting that we do not know how large has been the increase in R&D expenditures of the types that are relevant to measured growth.

THE POST-1973 SLOWDOWN IN TFP GROWTH

1.Last week we saw that 1973 marks something of a watershed, with the advanced countries (unweighted) average growth per annum of both GDP and manufacturing falling to about one-half of the 1950-73 value.

This has been the most serious macroeconomic disturbance since the Second World War. As Fischer points out, a 4 per cent growth rate per annum of per capita income means that it doubles every 18 years, but if the rate falls to 2 per cent per annum, it takes 35 years to double.

2.The slower growth in output cannot be explained in terms of a fall in the growth of the labour input because there was a commensurate fall in the growth of labour productivity. Moreover, as can be seen from Table 16, TFP growth also fell pari passu, suggesting that the cause could not be attributed to a slowdown of the growth of the combined factor inputs.

3.If we divide the post-1973 period into two sub-periods, 1973-79 and 1979-92 (1979 being the year of the second large increase in oil prices and a peak year in terms of economic activity), 1973-79 showed a widespread decline in TFP growth. In the period 1979-92, the picture is rather more complex with manufacturing industries showing a slight recovery in some countries, but not in others, and service industries experiencing a further fall in growth rates in most economies.

4.Ironically, the original impetus for the growth accounting approach in the 1960s was the attempt to explain why the residual was so large, but since then the focus has shifted to trying to account for why it has been so relatively small since 1973. There are two points worth emphasising:

a.The slowdown affected all the advanced countries, to a greater or lesser extent, simultaneously, from the years 1973/4 onwards (although there is some evidence that the US experienced the slowdown somewhat earlier in the mid-1960s). The obvious implication is that there is likely to be some common factor at work. Nevertheless, much of the detailed and painstaking work in trying to account for the slowdown has been by American economists concerned with the US economy and, while the international aspect may not have been completely overlooked, it has not always been given the prominence it deserves. For example, Denison (1979), after examining a large number of hypotheses, came to the conclusion that there may have been no one single cause: everything just happened to go wrong at once:

Several developments may have combined to slow the advanced in knowledge itself, and others to retard the incorporation of new knowledge into production. Similarly inflation, stagnation, soaring energy prices, high taxes and changing attitudes may have conspired to exert a large adverse impact on the miscellaneous determinants of output that forced the residual series into an actual decline.

b.The slowdown hit all the industries of the various economies to some degree, regardless of whether they were manufacturing or services. This again poses the same question as to why this was the case. One could perhaps make a case that the potential for technical change had become exhausted in, for example, one industry, as the number of inventions dried up. Maddison (1984, p. 70) terms this argument Wolf’s Law after Wolf and it is the thesis that ‘technical progress will ultimately be retarded because we will have exploited the easier innovations, there will be less to discover, and the unknown will be harder to penetrate’. A moment’s reflection, though, suggests that this is not as plausible as it may at first seem as an explanation of the slowdown as a whole. What, a few years ago, may have seemed almost impossible to solve, now, with the advance of knowledge, may become almost child’s play.

NEW GROWTH ECONOMICS

1.In the 1980s, a series of growth models was devised that sought to challenge the existing neo-classical theory of growth.

2.The existing theoretical literature on growth had increasingly become entangled with disputes in capital theory over the legitimacy of aggregating inputs into ‘factors’ of production, or of aggregating production functions themselves.

The theory had also become more complex with the development of multi-sectoral models or models with capital goods disaggregated either by sector or by ‘vintages’.

Nor could the limited data available discriminate among the many variations on the standard model.

3.Two developments were important in explaining the recent revival in growth theory:

a.The ‘re-discovery’ of a well-known fact – convergence of per capita incomes across countries was either limited or absent in the world economy. We will illustrate this in the next section. If exogenous technical progress was commonplace, shouldn’t standards of living tend to convergence, especially given the increasing mobility of capital internationally? It is only recently that growth theorists have become aware of this absence of convergence, even though development economists and economic historians had long been aware of it.

b.Dissatisfaction with existing models arising from a form of ‘intellectual imperialism’ (Boltho and Holtham). If something called technical progress accounted for the steady-state growth rate and for the bulk of observed growth according to the growth accountants, was it not regrettable that economists had nothing to say about what determined the rate of technical progress? Economic tolls were being used to explain many decisions previously thought to be non-economic: economics of crime and punishment, marriage and parenthood. It is Schumpeter’s dictum that economic causes should be sought for economic phenomena that is the motto of the endogenous growth theorists.

4.The rationale of these new growth models, developed initially by Romer and Lucas, attempted to dispense with the assumed exogenous technical progress of the old growth models and to explain steady stage growth endogenously, i.e. within the model. Broadly speaking, new growth economics identifies four different views which emphasise various aspects of the growth process: learning by doing; human capital; research and development (R&D) and public infrastructure. It also stresses the potential for international spillovers in some of these areas.

5.A crucial feature of the old growth models was that capital accumulation is subject to diminishing returns such that in the long-run, the rate of growth is independent of the rate of investment and thus of policies which influence investment. The new growth models abandon this assumption: Scott actually argues that the law of diminishing returns does not apply to capital, even in the long run. If investment and technical change are inextricably interwoven, then capital accumulation will result in a qualitative change in the economic environment.

6.Lucas’s seminal (19988) article puts human capital centre stage. Human capital not only adds to the labour supply, as in Maddison’s growth accounting, but there is also an externality which augments TFP growth. The new growth theory also assumes that the initial human capital stock with which an individual starts is proportional to the existing human capital stock. In other words, it is not necessary to keep re-inventing the wheel. Knowledge does not totally disappear with the death of a worker.

7.A lot of the writing on new growth theory has remained largely theoretical. However, Crafts and Toniolo’s chapter in Economic Growth in Europe Since 1945 does set an agenda for future research.

CONVERGENCE AND CATCH-UP

1.There are many examples of unusually fast growth by follower countries in catching-up the leaders. This view anticipates ultimate convergence of productivity levels, at least among OECD economies, and predicts that growth of TFP and ultimately overall growth will increase with the productivity gap between a country and the leading country of the time.

In this view, the recent slowdown in Europe reflects the closing of the productivity gap between Europe and the US. Abramovitz is one of the key writers on catch-up factors and he argues that catch-up is not automatic nor is its potential always fully realised. He argues that catch-up depends on what he terms ‘social capability’. Differences in social capability are sustained by market failures: inadequate education, capital markets and political economy considerations.

The catch-up thesis is popular among contemporary American scholars, in their attempts to explain why US income levels are being steadily approached by those of other relatively wealthy countries, and why the growth rates of the USA have not been as fast as those of most of its rivals.

2.Figure 1a plots the data by region on a linear scale and suggests no obvious ‘convergence’ within Europe: the tracks of real GDP per head are approximately parallel. West Europe overhauls North Europe, and South Europe overhauls East Europe in the mid-1960s.