A Quantitative‐Historical Approach to British Economic Growth
Abstract and Keywords
The approach to British economic growth is historical in that it compares growth in different periods rather than comparing growth in the UK with that in other countries. The key question of interest is why was growth in the period since World War II better than in other periods since the 1850s. The approach is quantitative in that it is heavily statistical while being well grounded in economic theory. Most of the basic data are presented in a growth accounting framework. This does not imply that the underlying exogenous causes of growth are entirely on the supply side: demand side factors such as those emanating from abroad or from government policy can also be causes of growth.
The subject of this book is the course and causes of British economic growth from the middle of the nineteenth century till 1973. More particularly, our aim is to view in their historical context the years between World War II and 1973. Those were years of unparalleled growth in Britain, as in other countries.
Certain of the questions about British economic growth are questions that can be asked about growth in any country. What was its time pattern? What was the relative importance of the various contributory causes, on the supply side and on the demand side? How did those causes interact with one another and with the process of growth itself? Other questions are particularly important in the British case. Principal among these are questions about the causes and consequences of changes in Britain's relative standing among industrial countries and in its competitive position in international trade, and questions about the effects of wars.
The features of the approach to British economic growth adopted in this book are that it is historical — a study of trends over a long period, not a comparison with other countries' growth in a particular period — and that it is set in quantitative national‐accounting terms.
This chapter explains the scope, method, and arrangement of the book. In the process, it introduces some of the main substantive questions about British economic growth. It discusses the implications of the historical and quantitative national‐accounting aspects of the approach adopted, the plan of the book, and the underlying conceptual framework, and describes some of the main features of British economic growth.
The Historical Aspect of the Approach
The history of past events may be studied either as a prerequisite for understanding the present or as a way of understanding the past itself. (p. 4 ) Our purpose is to do both. We are especially interested in the postwar period, * because it is nearest to us and leads into the present. We shall therefore treat it more fully than earlier periods. In this respect we have not followed the precept of Ranke, that every age is equal in the sight of God. In the spirit of that precept, however, we have regarded earlier periods as being of interest in their own right, as well as for their relevance to the postwar period and to the present. Our interests are thus partly those of the economist, partly those of the historian. †
The period we cover runs from 1856 to 1973. The starting date, 1856, was chosen for statistical reasons, not because that specific year is to be regarded as a watershed. From that date annual statistics on national income and on inputs of labor and capital are available that are in principle internally consistent and comparable with those available from official sources on the postwar period (see Feinstein 1972 ; Feinstein forthcoming). The availability of such statistics is essential for a quantitative approach in national‐accounting terms.
However, the middle of the nineteenth century also makes sense as a starting point for historical reasons. It corresponds roughly to the end of one phase of British economic development: the phase of industrialization, based principally on textiles. From the 1850's onward—in fact probably from a decade or so earlier — there was no significant trend increase in the proportion of the labor force engaged in manufacturing. This is in contrast with the steep rise that occurred in earlier decades of the nineteenth century. It is in contrast also with the rise that occurred in later periods in most of the other advanced countries. 1
By 1856 Britain, with nearly a third of its labor force in manufacturing and with the framework of its railway system completed, had already become a mature industrial country in many respects. It had (probably) the world's highest income per head. The history of British economic growth in the next century and a quarter is the history of the development of the economy from that stage on.
(p. 5 ) The reason why we do not take the story back before the middle of the nineteenth century is not that the events of the previous century are irrelevant to the present, far less that they lack intrinsic interest. On the contrary. Britain's early start on the path of industrialization can with good reason be held responsible for many features of its later development, including features that are still with us. This early start—earlier than other countries by about half a century—took place in the latter half of the eighteenth century and the first half of the nineteenth. Some references to that period will be made at various points in ensuing chapters. However, as we go back in time, data limitations diminish the contribution that can be made by the quantitative national‐accounting method to economic history relative to the contribution that can be made by other methods. 2
Phases in the century and a quarter after 1856 have been variously characterized by economists and historians. Some have seen the whole period, crudely, in terms of continuous (relative) decline, at least after 1870; others have distinguished Kondratieff cycles; all have agreed that the two world wars ushered in major changes. The characterization that emerges from the following pages is indeed one of nearly continuous decline relative to other industrial countries. But as far as the British rate of growth itself is concerned, the pattern is a different one. It is best described as U‐shaped. The rate of growth declined to a low point some time before or around World War I, then increased up to 1973.
The first part of the period, from 1856 to 1913, consist of two phases. The earlier phase culminated in what seemed at the time to be the high‐water mark of Britain's industrial prosperity in the boom of 1872–73. The subsequent 40 years have been commonly regarded as a period of retardation — retardation relative to the growth rate that had been achieved earlier and retardation also relative to what was currently being achieved by the other countries whose competition was increasingly being felt. The timing and extent of this retardation are still unclear. But there is no doubt that the rate of growth of productivity was low, at least after 1900. And there is no doubt also that the 1870's saw a fall in the rate of return on capital that was never subsequently fully made good.
World War I and the years immediately after it were in many respects critical in the history of British economic growth. On the one hand, there took place in 1919–21 the greatest setback to real GDP that the country had experienced since the industrial revolution. On the other hand, those years marked also the beginning of a phase of increase (p. 6 ) in the rate of growth of productivity that persisted for the next half century. It is unfortunate that the statistics for the years 1914–21 are less satisfactory than for any other years within our period, and we shall not be able to analyze the immediate aftermath of World War I as fully as we would have wished. * We have taken 1924 and 1937 as the first and last years of what we normally refer to as the interwar period. This gives us two years that are broadly comparable in terms of the level of activity, and excludes the statistically uncertain and economically abnormal events of 1919–20.
In the interwar period economic growth was not a major topic of public discussion. Unemployment after 1920 never fell below 9 percent, there were violent divergences between the experience of different industries and regions, and international economic relations were in a state of turmoil. Recovery and the restoration of normalcy were the objectives of policy, rather than growth. Not for the first time or the last, however, the preoccupations of public men did not accurately mirror the importance of events. The period was certainly an abnormal one — so much so that for many purposes (e.g. the study of trends in investment) it does not provide a very helpful basis of comparison with the postwar period. But, as is now well known, real income per head of population rose considerably more rapidly than in the prewar years. The contributory causes — industrial, international, and demographic — will be discussed in later chapters. Straight reversal of the causes that had made for the setback of 1919–21 was not paramount among them.
We come finally to the postwar period, 1951–73. Our terminal date, 1973, marks the end of the great secular postwar boom, in Britain as in other countries. We do not carry the story into the long recession that began in that year. Naturally, certain of our remarks on the postwar period will be colored by an awareness of the doubts cast by the recession on whether the previous rate of growth will in the long run be capable of being restored, let alone improved.
In the postwar period the rate of growth of GDP was higher than in earlier periods. This was notwithstanding that the rate of growth of the labor force was considerably lower than in earlier periods. The rate of growth of GDP per worker was thus much higher than in any previous period (not excluding periods before 1856). “There is little doubt that (p. 7 ) the British people would have regarded the post‐war record of their economy as highly satisfactory if the United Kingdom had been an isolated state. But, while British growth was above its earlier rates, growth rates of the other large countries of Western Europe were much higher still” (Denison 1968 : 232). Hence arises the persistent dilemma in the study of British economic growth in the postwar period: is the task to explain why Britain did so well or to explain why it did so badly?
The historical approach causes the first of these two questions to bulk more largely in this book than it does in most discussion of Britain's postwar economic performance: our subject is trends in the British economy over time, not comparison of those trends with trends in other countries. The latter would require an entirely different array of evidence; but we hope that our findings, in conjunction with research on other countries, will prove indirectly relevant to the second question. Comparisons between periods and comparisons between countries are complementary. Why, say, Britain's growth rate in the 1960's differed from Germany's is one question, and why Britain's growth rate in the 1960's differed from its growth rate in the 1930's is quite another. Ideally it should be possible to embrace the answers to both questions in a single grand model of economic behavior. As an intermediate enterprise, both types of question are worth asking. Which type of question is more helpful toward a better general understanding or toward the resolution of questions of policy is not self‐evident. Much depends on the specific question at issue. Some of the underlying patterns of behavior are likely to be more nearly constant over time, others over space. On the one hand, different countries at the same date will have access to the same pool of knowledge and be subject to the same world economic environment. On the other hand, a given country at dates separated by a generation will have many of the same people in the labor force — to say nothing of the children they have brought up and molded — and it will have some of the same stock of physical capital. More important, perhaps, there will be continuity in its institutions, economic, social and political, and this continuity will persist over periods longer than a generation.
The Quantitative Aspect of the Approach
This book is economic history based on the national accounts and on data on inputs. The advantages of the national‐accounting approach do not need to be labored. Gross national product (GNP) and its related measures are in principle comprehensive and at the same time (p. 8 ) susceptible of disaggregation, by sector and in other ways. They describe the performance of the economy as a whole, not just those aspects of it on which we happen to have particularly good information. They provide a unifying basis for the study of economic change. Being quantitative, they give a clear meaning to the concepts of growth or decline. *
The corresponding disadvantage is that the statistical data are by no means fully reliable. It is important to bear this in mind. In general it is our belief that the data are adequate for the purposes for which we have used them. But the possibility cannot be ruled out that future data revisions will alter some of our conclusions qualitatively, not just in detail. This possibility exists with all types of evidence.
A reminder of the fallibility of the data is provided by the imperfect consilience of the three indicators of GDP (income, expenditure, and output). Experience in handling the data reveals that the choice between such alternative indicators usually makes more difference to the results than is made by the choice of ways to manipulate them. In the particular case of the three GDP indicators, we avoid the problem by taking as benchmarks only years where the discrepancies are unimportant, and by using a geometric mean of the three indicators wherever possible.
In general the output statistics are more reliable for goods than for services, and all series are more reliable for recent times than for earlier times. The absolute margin of error in some series is large. † Margins of error in growth rates are probably less, insofar as the same sources of statistical error persist over time. But small differences between periods or sectors in growth rates cannot be regarded as significant. This is a particularly important matter in the case of the United Kingdom, where average growth rates have been fairly low, so that a small (p. 9 ) difference in growth rates may be sizable relative to the average growth rate. For example, the much‐bruited slowdown in the rate of growth of output per man‐year between 1856–73 and 1873–1913 was only a slowdown from an annual growth rate of 1.3 percent to an annual rate of 0.9 percent. * This is to be regarded as a significant difference; a difference of 0.4 percent in the annual growth rate, compounded over the 40 years between 1873 and 1913, would have made a difference of 17 percent in the level in the final year. Hence this is a phenomenon that is worth trying to explain. But it is not sensible to get excited by differences of one‐ or two‐tenths of a percentage point. This is a warning we pronounce here once and for all: it might be desirable but would be tedious to reiterate it at frequent intervals throughout the book. Our own practice may sometimes seem to disregard the point. The chief reason for doing so is in connection with decomposition. If an annual percentage growth rate (or a difference between growth rates) of, say, 1.0 is to be decomposed additively into half a dozen sources, most of them are likely to turn out to be pretty small numbers. In such contexts we sometimes follow the practice of giving percentage growth rates to two decimal points, for purposes of comparison, while recognizing that they have no pretense to that degree of accuracy.
The problem just described is that the data may be inaccurate indicators of what they purport to measure. It may also happen that what they purport to measure is not exactly what would be appropriate from a theoretical point of view. An example is provided by statistics on the capital stock, where the gross measure and the net measure both have conceptual deficiencies (to some extent in opposite directions; see Appendix F ). In this instance we give both measures when dealing with the economy as a whole, but only the gross measure when dealing with sectors (because net series are less reliable statistically and indeed, for that reason, are not published at all for individual sectors in official statistics for the postwar period).
In order to try to understand trends in the variables measured in the national accounts, it is necessary to have regard also to many other kinds of information. This we have done. However, the quantitative national‐accounting approach has inevitably affected the scale on which different aspects of British economic growth are here treated. We shall have a great deal to say about investment, sectoral shifts, and international trade, but less about topics that are in principle at least as (p. 10 ) important, such as technology, entrepreneurship, and industrial relations. These are the subject of a gigantic literature at a very micro level, in industry studies, in business histories, and so on. But much of it is lacking in the historical dimension, and all of it is difficult to translate into overall measures in national‐accounting terms. Discussion of these topics is included at various points, and we have not fought shy of drawing conclusions necessary for the formation of a view about what was important and what was not. However, much that relates to them has had to remain in unallocated residual sources of growth.
Quantitative economic history may imply to some people the use of sophisticated statistical or econometric methods. There are a number of reasons why these are not to be found here.
First, we are interested in trends over periods, not in cycles. It is as easy to see the relationship between the trend growth rates of a number of associated variables by writing them down in a table or plotting them on a chart, as by estimating a regression equation. And it does not misleadingly suggest a precise relationship.
Second, the number of degrees of freedom for estimating trends as opposed to cycles is small. It would be increased if we were to assume that the structural model was the same in each period. But it does not strike us as a plausible assumption that the same relationships between observable variables persisted over the period of 117 years from 1856 to 1973, or even that they changed over time in a way that could be captured in a single econometric model. Though there is continuity between successive periods, we view them as phases in a dynamic process, not as samples from a universe of observations.
Third, it is difficult to have much confidence in econometrically estimated relationships. This is partly because of data inadequacies and the well‐known problems of linear regression. 3 But it is also because econometrics cannot always distinguish between two (or more) theoretically distinct models with markedly different implications; they may both (or all) be consistent with the data. 4 Econometrics can shed considerable light on economic relationships; but it can also mislead. A less formal approach to the data reduces the risk that all the emphasis is placed on a single explanation, which may in reality be false.
Our quantitative techniques are essentially very simple. We compare trends over time in related variables by examining growth rates and changes in percentage shares. These are usually calculated between cyclically comparable years. For many series annual data are (p. 11 ) also shown in charts. One device we have frequently found illuminating is to decompose variables into constituent parts (in the manner of MV = PT). Such tautological decomposition does not as such identify causal elements, any more than the estimation of regression equations does. Its usefulness depends on its appropriateness as a basis for decomposition.
To sum up on both historical and quantitative aspects, our treatment is heavily statistical and, we trust, well grounded in economic theory. It is more in the intellectual spirit of the economic historian than in that of the econometrician in accepting that there may be changes in the structural model between periods. Indeed, a major concern throughout is to identify and explain differences between periods. Throughout, we are conscious of those unavoidable limitations of our knowledge that Lundberg referred to in his characterization of the work of Kuznets: “All sets of findings necessitate more exact knowledge of the ‘mechanism’ involved than that which lies at hand. Hence the explanations do not turn out to be much more than ‘suggestions,’ e.g. in regard to those factors that have determined observed trends or fluctuations in the development: ‘The result is, at best, a sketch of a possible but untested association between the findings’ ” (Lundberg 1971 : 460, quoting Kuznets 1961a : 6).
Plan of the Book
The book is arranged in five parts, as follows.
The structure of Parts Two and Three is based on the foundation of the growth‐accounting method first made familiar by the work of such writers as Abramovitz ( 1956 ), Kendrick ( 1961 ), and Denison ( 1962 ) on the empirical side and Solow ( 1957 ) on the theoretical side. (This method is further discussed in the next section.)
Part Two (Chapter 3–7 ) is concerned with the economy as a whole. The sources of growth of real national income are decomposed, by the growth‐accounting method, into the growth of factor inputs and the growth of factor productivity. Chapter 3–5 describe the growth of factor inputs. They deal respectively with the growth of labor input measured in man‐hours, the growth of labor quality, and the growth of the capital stock. The rate of growth of total factor input (TFI) is defined as the weighted average of the rates of growth of labor and (p. 12 ) capital, where the weights are the distributive shares of the factors in national income. Distributive shares are therefore dealt with next, in Chapter 6 . The rate of growth of total factor productivity (TFP), equal to the excess of the rate of growth of GDP over the rate of growth of TFI, is discussed in Chapter 7 .
Part Three (Chapters 8 and 9 ) deals with TFI and TFP by sector. Chapter 8 studies the growth of TFI and TFP in the principal sectors of the British economy and Chapter 9 the effects of the shifts over time in the relative importance of the different sectors.
The decomposition of the sources of growth yielded by the growth‐accounting method is a decomposition into proximate sources only. It does not as such explain what caused factor inputs and productivity to move as they did. Parts Two and Three contain a full description of the trends in the proximate sources and discuss some of the underlying causes. The remaining causes are discussed in Part Four . The principle of arrangement has been broadly to include in Parts Two and Three analytical matter relating to topics that can reasonably be treated on their own and to postpone to Part Four matters where everything interacts. Thus the treatment of labor input in Chapters 3 and 4 in Part Two includes almost all we have to say on that subject; but since the rate of growth of the capital stock was, in our view, endogenous to the growth process, and moreover, affected demand as well as supply, we have confined the treatment in Part Two (Chapter 5 ) largely to a description of trends and have postponed the main discussion of underlying causal relationships to Part Four (Chapters 11–13 ). Part Four also deals with demand (Chapter 10 ) and with international trade and payments (Chapters 14 and 15 ). These are two related classes of consideration that cut across the growth‐accounting categories and are liable to affect both inputs and output; more will be said about their relation to the growth‐accounting approach in the final section of the present chapter.
The above outline is designed merely to make clear to the reader the logic of the book's arrangement. A fuller indication of the ground covered may be had from the synopses that we have prefixed to each chapter.
Causation and Explanation
We have been bandying about such expressions as source, proximate source, cause, and explanation. Everyone uses such expressions. But they are not free from ambiguities. The ambiguities matter. They (p. 13 ) affect the questions it is sensible to ask, and they can affect the answers — not least statistical answers. So a few further remarks are needed here.
If one could draw a complete deterministic map of knowledge, no event would be an ultimate cause — each event could be traced back to something else. The search for causes would then be the attempt to draw the map; it would be the search for the laws of motion of the entire system. Only very foolhardy persons set themselves such an objective. What economists usually do is try to set up a limited model of the system's laws of motion, embracing some aspects only and relegating the rest to the category of exogenous (most historians regard even that as too ambitious). Matters may be exogenous because they fall outside the confines of the model in time (initial conditions), in space (overseas events), or in scope (technological, political, or sociocultural aspects). Exogeneity is an attribute of the chosen framework of thought, not an attribute of the events themselves. *
The search for causes thus amounts to (1) identification of the behavioral relationships in the system, within the chosen limits of the model; and (2) identification of relevant exogenous events. Given the structure of the model, and given the fact of a certain exogenous event, the consequences of that event are determined; the consequences are the joint result of the exogenous event and the structure of the system.
The most fundamental source of disagreements about explanations of economic growth is disagreement about the structure of the system — which relationships were weak, which strong, and so on. Disagreement on this point may imply disagreement about the appropriate scope of the model as well, that is to say, about which classes of even can reasonably be treated as exogenous (e.g. sociocultural changes).
Even if a model is agreed on, there are a number of different types of questions that may be asked of it. This was noted above in connection with comparisons of periods as opposed to comparisons of countries; even supposing the same model of behavior is valid for all relevant countries and periods, the points of difference between countries regarding initial conditions and other exogenous influences will not be the same as the points of difference between periods. Questions about (p. 14 ) differences between countries and between periods, moreover, are not the only possible ones. Two other types are commonly asked:
(1) Counterfactual questions can be asked about the effects of hypothetical events that did not occur in any time or place. If one lists six causes of growth over a particular period and gives them quantitative magnitudes, one is implicitly making six counterfactual assertions about what the rate of growth would have been had only one cause been present at a time. This procedure creates no problems of principle if the causes are exogenous events. If the causes postulated are not wholly exogenous, serious difficulties can arise. One cannot postulate counterfactual assumptions that are inconsistent with the logic of the system. No quantitative answer is possible to the question “What would have happened to the rate of growth if investment had been higher and everything else had remained unchanged,” because if literally everything else had remained unchanged, investment would not have been higher. In order to consider counterfactually the consequences of an endogenous variable like investment having been different from what it was, it is necessary to specify the supposed circumstances that would have led it to be different. Hence the need to go behind the proximate sources of growth, whether on the supply side or on the demand side.
(2) Entirely different again are questions relating to current policy. What steps the government could now take to alter the rate of growth (or for that matter the level of unemployment or the rate of inflation) is quite a different question from how the present state of affairs came about. Questions about the effects of past government policies are a particular kind of counterfactual question. We shall generally follow the practice of treating government policies as exogenous, while recognizing that on a broader definition of the scope of the model the actions of governments are no less endogenous than those of other economic agents.
The foregoing remarks have been made in order to indicate some of the pitfalls to be avoided and in order to explain the notions of causation that have been in our minds in writing this book. These remarks may appear to promise a program that is more ambitious and more rigorous than anything we shall actually carry out. We shall not produce a formal model of the working of the British economy over time, nor shall we define our scope so sharply as to make a firm distinction between what is exogenous and what is not. However, in Chapter 15 we do try to isolate the effects of influences that were exogenous in the sense of originating from abroad. Chapter 17 contains an outline (not (p. 15 ) in formal terms) of the underlying model of causal interactions that appears to us appropriate, though the area of our ignorance is one of the chief features we are conscious of.
The growth‐accounting method used in Parts Two and Three is a way of carrying further the familiar breakdown of the growth of output into the growth of labor input and the growth of labor productivity. To this end it takes account of the contribution to the growth of output made by increases in capital input, as well as by increases in labor input. It also seeks to take into account the contribution made by improvement in the quality of the labor force, though this is more difficult to measure. The relative contributions made by the various inputs are estimated on the assumption that the remuneration of the factors of production is in proportion (not necessarily equal) to their marginal products — hence the weighting by distributive shares. (This assumption can be relaxed when there are specific reasons for doing so.) The rate of growth of TFP is then the contribution to the growth of output that comes from sources other than increases in measured inputs. Often referred to as “the residual,” it includes any contribution that may arise from increasing returns to scale and from the effects of technical progress and advances in knowledge, of shifts in resources between sectors, and of changes in the extent of obstacles to more efficient use of resources (e.g. restrictive practices on the part of management or trade unions). It will also reflect any errors in the measurement of inputs and output, and in the specification of the relationship between them.
The growth‐accounting approach has been subject to a number of criticisms. Some of the more technical problems are referred to in Chapter 7 , where the underlying logic and the procedures are discussed more fully. More general objections have also been raised by some economists. Some see it as a futile attempt to cut up a seamless robe; this seems to us obscurantist. Whether it is the best way to cut up the robe is more fairly debatable. Our view is that the approach is useful, but only as a first step. It cannot be relied on to give answers to counterfactual questions. However, it is an orderly way of arraying the facts, and that is an important part of the job.
Ours are not the first estimates to have been published of rates of growth of TFI and TFP in the British economy. We do not claim that our results are necessarily to be preferred to those of other authors, though we have tried to make them as well based as possible. Some (p. 16 ) writers have carried out (chiefly for the postwar period) more detailed sectoral breakdowns than we have, and others have made international comparisons that are outside our scope. 5 The estimates here presented are, to the best of our belief, the first that have been published for such a long period as 1856–1973, covering both the economy as a whole and its sectors.
One of the main questions that has been asked in the literature of growth accounting is the relative importance of increases in factor input (TFI) on the one hand and of improvements in the way factors are used (TFP) on the other. 6 Where the distinction is drawn between the two is to some extent arbitrary (the matter is further discussed in Chapter 7 ). We shall see, however, that when account is taken of all elements of factor input that can reasonably be quantified, the rate of growth of TFP was far more important in the British case than the rate of growth of TFI as a source of differences between periods in the rate of growth of GDP. In particular, the faster‐than‐previous growth of GDP in the postwar period was more than wholly accounted for by a faster‐than‐previous growth of TFP (the rate of growth of TFI being lower than in earlier periods). On the other hand over the average of the whole period 1856–1973 TFI made a considerably larger contribution to growth than TFP (1.4 percent compared with 0.5 percent). Between 1873 and 1913 there was no increase in TFP at all. Since technological advance (not directly or indirectly included in TFI) was plainly taking place in some sectors at that time, it follows that there was actual retrogression in other sources of TFP. This gives some prima facie support to the hypothesis of failures in entrepreneurship or in industrial relations put forward by such writers as Phelps Brown and Browne ( 1968 ) and Landes ( 1969 ), but disputed by econometric historians such as McCloskey and Sandberg ( 1971 ).
It is no surprise to find that the rate of growth of TFP was persistently higher in some sectors than in others. More interesting is the finding that in certain periods changes in the rate of growth of TFP in some untypical sectors had important effects on the rate of growth of TFP in the economy as a whole (commerce in the interwar period and across World War II, agriculture in the last quarter of the nineteenth century). However, the main conclusion that emerges from the sectoral data is that changes between periods in the rate of growth of TFP were usually similar throughout the economy. The causes making for these changes were evidently pervasive. In a sense, therefore, the study of the sectoral data leads to negative conclusions.
Also negative are the findings about the effects on the growth of TFP (p. 17 ) of changes in the relative weights of different sectors—manufacturing, agriculture, services (including government), and so on. Insofar as we are able to measure these effects, they turn out to be relatively small in most periods. This may, of course, be related to the fact that, as mentioned at the beginning of this chapter, the net shift of labor into manufacturing had already taken place by the middle of the nineteenth century. It is conceivable that, by comparison with other countries, the very absence in the United Kingdom of large effects of structural shifts was itself a matter of importance.
Supply, Demand, and Foreign Trade and Payments
The problem of distinguishing proximate from underlying causes perhaps arises most prominently in connection with the question of the relative importance of supply‐side forces and demand‐side forces in determining the rate of economic growth in the United Kingdom. Was growth in the postwar period faster than in the earlier periods because demand was stronger or because technology was advancing more rapidly? Was postwar growth in Britain slower than in other countries because of a government‐imposed stop‐go cycle in demand or because of the attitudes of management and labor? Closely related is the question of the role of foreign trade and payments.
One extreme position is that expressed by Say's Law, that the supply of labor and of know‐how is exogenous and creates its own demand (at least apart from cycles). This is what is assumed in the elementary neoclassical growth model. The opposite extreme position is that factor inputs and/or productivity adjust themselves so as to permit the production of whatever is demanded (at least within reason). Intermediate positions are possible, and more appealing, in which interaction is allowed for between forces on the supply side and forces on the demand side.
The use of the growth‐accounting framework may appear to prejudge the issue in favor of the primacy of exogenous forces on the supply side. However, this is not the case. As far as that framework is concerned, demand is one of a number of possible influences on trends in factor input and in TFP, and its importance relative to other influences is a matter for study.
An alternative framework is to decompose the growth of income into the growth of the constituent elements of expenditure: C + I + G + X − M. (This is done in Chapter 10 .) The use of such a framework might appear to prejudge the issue in favor of the primacy of exogenous causes on the demand side; but that too is not the case. It leaves entirely (p. 18 ) open the question why the constitutents of expenditure moved over time in the way that they did. The underlying explanation could be either on the demand side or on the supply side, or a mixture of the two.
We shall be concerned with demand only insofar as it affected trends in real GDP or in inputs and productivity. Business cycles as such do not come within our purview, nor do movements in the general price level, though we shall find it necessary to say a certain amount about both of them. Our standard method of abstracting from cyclical movements will be to focus attention on growth rates between cyclical peaks. The treatment of demand will be in a Keynesian framework, but this framework will not be used in such a way as to preclude the recognition of influences or processes commonly regarded as non‐Keynesian, such as crowding‐out and the level of real wages.
In the British economy foreign trade and payments have been among the most important influences on the level of demand and also on its composition. Many authors have regarded the country's international position as crucial in determining its rate of growth. They have pointed to the retardation of growth in the face of foreign competition in 1873–1913; to the absolute fall in income across World War I associated with the fall in exports; to the failure after the return to gold in 1925 to achieve a boom comparable to America's or Germany's; and to the constraints imposed by the stop‐go cycle in the postwar period. Kaldor ( 1966 ) and W. A. Lewis ( 1978 ) have seen the problems of the British economy as resembling the problems of a region, dependent for its prosperity on attracting an adequate share of the world demand for manufactures. Others have reached similar conclusions from rather simpler premises, unconnected with the composition of demand: they have seen growth as demand‐constrained and demand as balance‐of‐payments‐constrained. This view, expanded into the doctrine of vicious and virtuous circles (Lamfalussy 1963 ; Beckerman et al. 1965 ), was especially popular in the early 1960's and had some influence on the National Plan of 1965.
Opposed to all these is the doctrine, espoused broadly speaking by the Brookings Report (Caves et al. 1968 ), that the problems experienced by the United Kingdom in its foreign trade and payments have been principally a manifestation of supply‐side constraints.
The conclusions that will emerge in later chapters, both on foreign trade and on demand more generally, are of an eclectic kind. The facts cannot be accommodated to monocausal explanations. Exogenous influences on growth from the side of demand or foreign trade were important, (p. 19 ) but they were not all‐important, and they were more important in some periods than in others (for example, they contributed more to the improvement in the rate of growth of TFP that occurred after World War II than they did to the improvement that occurred after World War I). Eclecticism on this point, as on others, is not indecision. Nor does eclecticism amount to the implausible hypothesis that there were lots of exogenous forces all accidentally pulling in the same direction. It is a recognition of the complexity of the interactions within the system, both between periods and between Britain and the rest of the world.
Events on the demand side that may be treated as exogenous in consideration of economic growth in a single country include such matters as government policy, growth in other countries, and international economic arrangements. Though exogenous from one point of view, these are all firmly economic. Hence demand‐dominated theories of growth can be fairly self‐sufficient within the discipline of economics. Exogenous events on the supply side include a larger proportion of matters commonly regarded as outside the scope of economics, or at best on its fringe—demographic, technological, institutional, and attitudinal. Moreover, such matters affect the structure of the system as well as being a source of exogenous shocks. This is why treatments of British economic growth that emphasize the supply side tend to lapse into what has been called “a welter of amateur sociology.” The amateurishness is not entirely the fault of the economists, since their colleagues the professional sociologists have tended not to be very interested in these questions.
Institutions and attitudes that affect TFP do not come into existence or change for arbitrary reasons. Though they may be influenced in part by wars and other events that can fairly be regarded as exogenous, they are likely to be strongly influenced by long‐run economic developments. To treat them as fully exogenous is not satisfactory. Unfortunately, even the relevant taxonomy is still in its infancy, though attention is increasingly converging on it (Leibenstein 1979 ). The prospects of measurement or of a theory in historical terms are even more distant. We do believe that these supply‐side forces were important in the British case, but for the reasons just stated we lack the means to deal with them adequately. Our discussion is therefore necessarily incomplete. Proper treatment of the missing topics must await a Volume II, which we hope may someday be written by other hands.
(*) Throughout this book this expression will be used to refer to the years 1951–73 (the period 1945–51 being regarded as one of postwar recovery). Where “postwar” is used more generally for the periods after both world wars, the context will make that clear.
(†) “The historian is concerned with the past, in its relation to the present; the economist is concerned with the present, and for the sake of the present with the past” (Hicks 1979 : 4). Some historians would regard even the first part of that sentence as conceding too much concern with the present. “The historian . . . is concerned with the later event only in so far as it throws light on the part of the past he is studying. It is the cardinal error to reverse this process and study the past for the light it throws on the present” (Elton 1967 : 66).
(2.) Some national‐accounting estimates are available for the 18th and early 19th centuries; in particular, for real product and national income in Deane & Cole 1962 : 40–82, 154–73, and Deane 1968 ; and for capital stock in Feinstein 1978 . Though these estimates are valuable, the foundations on which they rest are very unreliable, and the series are not comprehensive, not available on an annual basis, and not fully comparable with those for 1856 onward. For a critical comment on the 18th‐century estimates, see Crafts 1976 .
(*) There are also statistical problems with the years 1939–47, but these matter less because the behavior of the British economy was not so peculiar in the aftermath of World War II as it was in the aftermath of World War I. The periods of the two wars themselves are deliberately not much discussed in this book, as being obviously sui generis.
(*) In the 19th century the word more commonly used was progress, with its vaguer and more debatable connotations. Thus the title of G. R. Porter's great statistical compendium, published in 1851, was Progress of the Nation. The word growth does not appear in the index of Marshall's Priniciples. Still more comprehensive indicators than GNP can, of course, be devised, taking account of leisure, environmental quality, and so on. The work done on such indicators suggests that although their actual growth rates are quite different from those of GNP, they do not make a great difference to relative growth rates (Nordhaus & Tobin 1972 ; Beckerman 1980 ). These enlargements of the GNP concept still rest on the same theoretical basis, which is ultimately that of revealed preference. For some purposes this basis is inadequate: neither GNP nor enlarged versions of it purport to measure such larger concepts as happiness or social harmony.
(*) Throughout the book, growth rates are annual percentage compound rates calculated between peak years of business cycles, unless otherwise stated.
(3.) Hendry ( 1980 ) summarizes the list of problems first given by Keynes ( 1939b ) in a comment following his review of Tinbergen's work in business cycles, and then adds another six problems to it.
(4.) Comparing the policy implications of analysis based on different models of the UK economy in the 1960's and 1970's, Goodhart ( 1978 : 188) says: “Looking at exactly the same economy, and even using on occasions very similar structural equations, different modellers come to totally different policy conclusions because of their fundamental perceptions about the working of the economy. Econometrics has not, at least so far, provided any alternative for basic judgement, only some quantitative dressing and support for such judgements.”
(*) There may admittedly be events that any reasonable model would have to treat as exogenous, like the want of the horseshoe nail that led to the loss of an empire. Such events are not likely to be important for long‐run economic growth, though they can certainly be significant for short‐term economic movements (harvest variations, for example).
(5.) See in particular Reddaway & Smith 1960 , Salter 1966 , and Wragg & Robertson 1978 for industry‐by‐industry data; and Denison 1967 , Panic 1976 , and Christensen, Cummings & Jorgenson 1978 for international comparisons. (This list is not exhaustive.) The estimates given by these authors differ in various respects from one another and from ours in regard to coverage, method, and choice of data.
(6.) Most early work on growth accounting pointed to the predominance of TFP. This was challenged in later work, notably by Jorgenson and his associates. See Jorgenson & Griliches 1967 ; Denison 1969 ; Denison's exchange with Jorgenson and Griliches ( 1972 ); Christensen, Cummings & Jorgenson 1978 .