How a Country Catches-Up:
A Growth Accounting Studyfor the Portuguese Economy (1951-1973)

Luciano Amaral, European University Institute, Florence

This is much of a technical paper. And, as technical papers should be, a very boring one indeed. In it, along with a growth accounting study for the Portuguese economy between 1951 and 1973, a painstaking description of the methods followed is presented. As growth accounting results are very much sensitive to the assumptions, the forms of measurement and the concepts used, I try to provide and explain them here. Enough information is given for readers to be fully aware of what is being measured, how the choices made affected the results and how different choices could have produced different results. Only in the last few pages some interpretation is attempted, along with some rather speculative considerations related to the Portuguese growth process after World War II. These considerations should be seen as mere suggestions or rather orientations for future research, as they are dependent on work still to be done.
I take as self-evident the interest of a growth-accounting study for the period of fastest growth in Portuguese economic history. Through growth accounting some partial answers can be given to questions such as what were the main driving forces behind the growth process and what was the efficiency with which they were used. Such answers are, however, of limited nature. As it became common to label them, they deal just with the "proximate" sources of growth rather than with "ultimate" sources. "Proximate" sources are vague concepts such as capital, human capital, labour and total factor productivity. My results support the view that the Portuguese growth process in this period was mainly driven by capital accumulation, followed by total factor productivity and, lastly, by human capital accumulation. Total factor productivity is seen here as something more than simply a measure of technical progress. It is interpreted as some measure of the efficiency with which the factors of production (capital, human capital and labour) were used. The results show that total factor productivity became increasingly more important throughout the period. I try to link this with the degree of openness of the economy and suggest that the efficiency of the economy as well as its technological updating were in a certain way dependent on the process of opening that took course in the period.

*The general principle of growth accounting is that the output of an economy can be expressed as some function of its inputs. Underlying all growth accounting exercises is, thus, a certain production function framework. The most common production function forms imply strong assumptions both on the relationship between output and inputs, and on the relationship between the inputs themselves. Namely, they assume either a constant elasticity of output to inputs, or a constant elasticity of substitution between inputs, or both. This has strong implications for the perception of the way inputs determine output and for the perception of the way technology acts in the economy.
Recent developments in the theory of production have improved on these shortcomings. Nowadays it is possible to loose both assumptions, namely by using a transcendental logarithmic production function (or, more simply, translog production function) framework. This is the framework used in this paper.
These developments led to the construction of adequate empirical instruments. Whereas for the Cobb-Douglas production function it is enough to use the highly restrictive Laspeyres or Paasche index numbers to measure inputs and their contribution to output, in the more flexible function used in this paper other solutions are necessary. I will use a version of the Divisia Index developed by Frank Gollop and Dave Jorgenson.

*In growth accounting the sources of output are the factors of production (labour, capital and, in more recent studies, human capital) and a residual of unexplained elements. As it became common to label them by growth accountants, these are just "proximate" sources of growth, as opposed to "ultimate" sources. "Ultimate" are big and unquantifiable sources such as culture, institutions, social structure and other (still unquantifiable but of minor importance) that are estimated to have some impact in the behaviour of economies. When one tries a growth account, as in this paper I do, one is not, thus, dismissing "ultimate" sources as unimportant for growth, but only keeping the purpose of analysis limited to those things that are of a quantifiable nature.
Even inside this limited scope, apart from inevitable technical difficulties, there are several problems of concept and interpretation. In the forty years of its existence the growth accounting tradition has dealt with these problems in several ways. The first great discovery of growth accountants was that growth in the United States in the 20th century could be attributed only in a small proportion to capital and labour inputs. The authors of these first studies, after having accounted for the contribution of labour and capital to output, found themselves with enormous residuals, that sometimes could explain 90% of the growth rate. That is, conventional and quantifiable inputs were found to be responsible for only 10% of the growth of the American economy in the 20th century. So important was this residual that eventually it was transformed in the main actor of growth accounting. It would of course be contradictory to call "residual" such an important factor and a name was found to replace the first one. Nowadays, this residual is more or less universally called total factor productivity, or, as an acronym, TFP. TFP is, thus, the productivity of the joint factors of production, capital and labour. It is the part of the growth of an economy that cannot be allocated to the growth of physical inputs.
The simple change of name did not mean a better understanding of the phenomenon and, in spite of several attempts to further explore the challenging residual, TFP still is, in the famous expression of Moses Abramovitz, a "measure of our ignorance".
Other things could be said about the limits of growth accounting. One of Moses Abramovitz's favourites, and a very important one indeed, is the absence of complementarity between the sources. In growth accounting the sources act singularly, i.e. capital (say) contributes, irrespective of labour's contribution, by a certain estimated amount to the growth of the economy. Abramovitz's point is that this is a very simplified way of understanding how an economy works. To give an example, capital embodies a certain amount of technology, technology can foster investment, education can attract and create technology, which in turn can be embodied in capital, and so on. Nothing of the kind is taken into consideration in growth accounting, and this of course, limits its interest.
Two things, in my opinion, have come to the rescue of the growth-accounting approach as a valuable field of research. First, the discovery that in several developing countries capital was the major source of growth, whereas TFP was of minor importance. This restored some confidence in the meaning of such concepts. When put in comparative perspective the fact that growth in some country is led by capital accumulation, while in another one TFP plays a larger role reveals something of the nature of both growth processes. TFP still has various meanings: that the country is using more modern technology, that this technology is being better used, that firm management is better, that economic policy is better, and so on. But this, when put in contrast with a situation where physical inputs play the largest part, means at least that factors are being used in a more efficient way. That is, one can still take some interesting information out of big and vague aggregate factors such as capital, labour and TFP.
A second element of importance in the rescue of growth accounting came with the role attributed to human capital by the new growth theory. As, in addition to this, a systematic correlation between educational levels and growth has been found in several empirical studies, recent growth accounts have tried to include human capital as a separate input. That is what I do in this paper and, as it became traditional to do it, human capital is identified with formal education. There are of course immense problems - technical, theoretical, of interpretation - with the measurement of this new variable. I try to deal with them in this paper.
These considerations are meant only to specify the limits of explanations based in growth-accounting, as well as my degree of belief in them. By providing the figures I do in this paper I am not trying to fully explain the process of growth as it took course in Portugal between the 1950s and the 1970s, for they can simply give me not only a "proximate" description of the process but also a very rough one. Nevertheless, with this study the Portuguese process of growth in the post-war period will be, I hope, slightly better understood. Some of its major features can be pointed out and some controlled interrogations of what is still left to explain be set forth. There is still much lying outside of the scope of such an analysis. It may, however, be considered as a sufficiently rich starting point paving the way for future research.

*The account is presented in three subperiods (1951-58, 1959-65 and 1966-73) and its main results are: a) capital is the most important input in any of the subperiods; b) human capital has always a strong contribution, which, however, is second to capital only in the first period, in the next ones being superseded by the contribution of TFP; c) capital accumulation increases its importance is a very consistent way from subperiod to subperiod; d) the contribution of human capital increases from the first to the second period, but stabilizes from the second to the third; e) the contribution of TFP is, in the period taken as a whole, the second most important, but it shows a pattern of persistent rise in importance from one subperiod to the next, so much so that in the last one its contribution is close to capital's contribution.
From what we know about the evolution of schooling in Portugal in this period, the contribution of education (human capital) to growth is not exactly surprising. After all illiteracy rates fell from close to 50% in the 1940s to close to 20% in the 1970s; close to universal enrollment in primary education was achieved as soon as the beginning of the 1960s; and secondary education grew at a very fast pace throughout the period.
The fact that capital had the most important contribution to output is not surprising either, due to the potential associated with capital accumulation. Investment in human capital requires much longer gestation periods than investment in physical capital. Had the rate of investment been similar in physical and in human capital, the former would for sure grow at a much faster pace than the latter. Also, countries at the stage of "development" Portugal was after World War II have in general very low capital to labour ratios. This creates an important potential ground for catch-up, which can be (relatively) easily explored.
These are, of course, just potential conditions. A second set of causes relates to the actual events occurred during the period. The economic policy followed by the government after the war was essentially directed to the development of capital-intensive industrial projects. Basic metals, chemical industries, electricity, naval construction and repair, cars, were some of the sectors particularly favoured by the state, especially until the beginning of the 1960s. In addition to this, the government tried to create the necessary conditions for fast investment. One of the main vehicles used was a "cheap money" policy, reflected in the discount rate of the Bank of Portugal, that was kept at very low levels throughout the period. As a result the importance of investment in aggregate national demand grew at a very fast speed.
More interesting to interpret is the evolution of TFP. It steadily grew in importance all along the period, at the expense of both human and physical capital contributions. What is relatively surprising for a country at Portugal's level of development is that in the last sub-period (1966-1973) the contribution of TFP is almost the same as the contribution of capital. TFP is not understood in this paper in the narrow sense of technological updating of the economy. TFP is a sort of catch-all factor, where everything that is not possible to attribute quantitatively to physical inputs is stored. It should, thus, be best seen as a measure of efficiency of the economy, as well as, in a certain way, a measure of what in Abramovitzian terms we could call the interdependence between the sources of growth.
To give an example, the fact that human capital has a lower contribution than physical capital does not really mean that human capital was of less importance to growth. Human capital is essential to absorb and adapt technology, as it is essential to enhance learning. Indeed, only in the presence of an educated labour force can an economy use complex technologies and transform them according to its particular needs. Nothing of the kind is directly captured in the sort of study I have attempted here. It is possible to presume a sensitivity of TFP to all this, but it is, nevertheless, just an indirect measure of the phenomenon.
As it is also an indirect measure of other important things. Contrary to what sometimes seems to be a widespread opinion, it is not enough for an economy to jointly accumulate physical and human capital in order to experience fast growth. The European socialist economies are a good example of this. They show how a highly qualified labour force associated with sky-rocketing levels of investment were able to produce amazing growth spurts rapidly followed by incredibly disappointing performances. The effect of physical and human capital accumulation can only be totally felt when put in an appropriate environment. Openness seems to be an indispensable component of that environment. Several facts in the economic history of Portugal after 1945 seem to support this idea.
The opening of the Portuguese economy started very early in the post-war period, with the entry to OEEC in 1948. The countries participating in the organization abolished or reduced a whole set of trade barriers. The participation in the European Payments Union, which introduced the multilateralization of payments in substitution for the usual bilateral agreements, eased even more the flow of commodities through borders. This was, however, still incipient openness. Real openness came between 1960 and 1962, when Portugal successively joined EFTA, the World Bank, IMF and GATT.
With all this, not only technology could flow more easily into the national borders, but also Portuguese industry was forced to upgrade its management methods in order to compete successfully, both at home and abroad, with better equipped rivals.
In addition to this, foreign investment entered the country at increasingly faster rates. In the 1950s restrictions to foreign investment were very strong. Until 1973, however, and mainly from 1960 onwards, they were progressively relaxed. This, together with the good performance of the economy, attracted growing amounts of foreign investment, both under the form of participation in national firms and of the installment of multinational firms.
The parallel chronology of openness and growing efficiency (TFP) of the Portuguese economy seems to support the idea of a connection between the two facts. Openness should not, however, be viewed as a growth factor in itself. It is not enough for an economy simply to be open to experience high growth. If it is not matched by the right conditions, openness can indeed have backward effects. This is where physical and, mainly, human capital enter back in the picture. Were it not for their development and the opening of the economy would not have had the mentioned effects. It was the growing qualification of the work force, linked to a strong capital formation that, when put in contact with the golden age environment of open and expanding markets, put Portugal on a path of high growth.
This picture calls for qualification. Although progressively more open, the Portuguese economy had until the mid-1960s very strict regulatory instruments affecting such important things as technology and investment. Not only the government favoured capital-intensive projects, but also, whenever possible, it tried to impose the use of national intermediate and capital goods instead of imported ones. As it is generally known, one of the most important ways of importing technology is precisely through intermediate and capital goods. This suggests that factor accumulation was until late directed to inefficient projects. Only after these restrictions were relaxed was Portugal able to fully benefit from the potential opportunities offered by the international environment of the golden age.
All these considerations should be seen as largely speculative and dependent on future research. In the form they are presented here they are nothing more than mere informed suggestions pointing the path through which my future work should go.