In the last part of his recent book, Carmelo Mesa- Lago (2000) provides a systematic comparison of the policies followed by Chile, Cuba and Costa Rica (Chapter 1) and of the results thereby obtained over the relevant periods for each country (Chapter 2). This is accomplished, of course, by building on the information developed in the previous three parts of the book which address the policies pursued and the results obtained for each country individually. Since others will discuss the individual parts, I will focus exclusively on the comparative issues.
It is quite difficult to perform a systematic comparison of policies and outcomes across the three very different “models” of development represented by these countries. To start with the word “model” is deceiving in that, at least in economics, it normally implies a far more logically coherent and consistent framework that can be associated with any of these “models” as practiced by these three countries, or any other ones for that matter. Out of this actual messy reality it is then necessary to extract some common measuring rods for comparing both policies and outcomes. Since policies are difficult to quantify and must be described qualitatively, while outcomes are easier to quantify, one would think that comparing policies is harder than comparing outcomes. This is true, but it has a redeeming feature.
One is likely to be more guarded and cautious in comparing policies than outcomes and, thus, the ratio of insights to disagreements generated by policy comparisons is likely to be higher than for outcomes comparisons. For this reason, I will argue here that the grounds Carmelo establishes for comparisons of policies are more likely to stand the test of time than some of the ones he establishes for comparisons of performance or outcomes. Despite their qualitative nature, the “vague” measuring rods for policies selected by Carmelo are indicators of a particular important or relevant condition or situation without a value judgement necessarily attached to every policy indicator as to whether more of it is good or bad. On the other hand, many of the “precise” measuring rods for outcomes selected by Carmelo, despite their quantitative nature, are not necessarily pointed in the right direction, because they all entail the value judgement that more of the indicator is always good or always bad, and this may be true for some but not for all or even most of his indicators.
Our discussion starts by identifying the main measuring rods for policies used in the book. The role of the state, according to Carmelo, can be measured in two dimensions: in terms of its power to implement and change policies and in terms of its degree of intervention in the economy. He points out that in the case of these three countries, the continuity of policies has been greater in Costa Rica than in Cuba or Chile. Yet the role of the state in the first dimension has been much greater in both Chile and Cuba (in the second dimension, Costa Rica stands between Cuba and Chile). Continuity of policy as a measuring rod is described as follows: “Continuity in policy should be beneficial for performance in the long run, unless such a policy is inadequate, but then it would probably be corrected.” Thus, continuity of policies is one solid basis for comparison but more is not necessarily better in all circumstances. Moreover, Carmelo attributes the greater degree of continuity of policies in Costa Rica than in Cuba and Chile to its democratic and pluralistic characteristics. It is hard to disagree with this insight even if you do not find Carmelo’s particular arguments entirely persuasive.
Carmelo describes the differences between the three models in terms of goals, economic organization, development strategies, outcomes and social costs. Only the second, third and fifth of these can be viewed as policies, and only the last one has the connotation that the less of it the better. For instance, whether more of an outward or inward strategy is better depends on circumstances. There are different points of view in the economics profession as to which policy is better and when it should be applied, 2 but there is widespread agreement that it is an important factor in determining outcomes. Carmelo simply compares the strategies adopted by the three countries without passing direct judgement on them, which is what one wants to see in this case.
Finally, Carmelo identifies three crucial policy issues on which to compare the models.
Would it have been feasible to avoid or reduce social costs of each model by changing some of their policies? Carmelo answers this question affirmatively and proceeds to document the answer for all three models. Were the policies compatible or did they involve trade-offs? Here the answer is yes and no for all three countries. That is, in some cases trade-offs were avoided due to favorable economic circumstances that harder times made impossible to avoid in other cases. Last but not least, how did these models do with respect to the crisis of the 1980s? Carmelo’s answer is that initially Chile and Costa Rica were negatively affected but Cuba was not. The answer to this last question may be controversial,3 but the merit of raising the question as a proper basis for policy comparisons is not.
When we turn to the analysis of performance or outcomes, the situation becomes better and worse simultaneously. It becomes better in that now indicators of performance can be measured precisely. Furthermore, Carmelo does as good or better job than anyone in ensuring the quality of the individual measurement from the point of view of the reliability of the data. Nonetheless, things also turn for the worse because many of these very well measured indicators are not really measures of performance or outcomes. In particular of the 14 economic indicators, 7 would be unacceptable to most economists as unambiguous indicators of good or bad performance or of a high or a low level of development by themselves. Yet Carmelo treats each one of them as such (p. 594) in order to add them up as part of the process of constructing aggregate indexes that lead to overall rankings of various types (absolute, relative and combined) for the three countries.
More specifically, I am referring to the seven following indicators: Gross Domestic Investment (as a percentage of Gross Domestic Product, GDP); Composition of Output; Export Concentration; Import Composition; Trade Partner Concentration; Volume and Balance of Trade per Capita; and Composition of the Labor Force. These seven measures are dubious as indicators of performance in that having more or less of any of them is meaningless as an indicator of performance in the absence of additional information or caveats. Hence, it is logically inconsistent and somewhat misleading to include them into an additive performance index.
For instance, Composition of Output (best performance is lowest agricultural share and highest industrial share) and Composition of the Labor Force (best performance is lowest agricultural share) are two different ways of measuring the same concept, namely the structure of output. While we observe that on average the share of agriculture (industry) is lower (higher) with the level of development, it does not follow that at any particular time and for any particular country a higher share of industry, for example, implies a higher level of development. It may simply imply a higher level of stupidity among the country’s policy makers in force feeding industrialization or it may imply a resource endowment more conducive to the development of an industrial base. This is especially relevant for these countries, since Carmelo demonstrates that at least two of them relied on policies aimed at promoting industrialization at various stages.
Gross Domestic Investment (GDI) is an unreliable indicator of performance because it is an input into the true or output indicator, namely higher growth. Carmelo is somewhat aware of this issue and throws in the caveat that he is assuming investment is done efficiently (p. 569). But there is another aspect of the issue that arises by definition4 which Carmelo ignores—namely, the effectiveness of investment as an input into higher growth is dramatically affected by the share of the investment financed out of domestic savings (Gross Domestic Savings, GDS) rather than out of foreign savings. That is, 40% GDI/ GDP with a 40% ratio of GDS/GDP implies very different growth potential than 40% GDI/GDP with a 20% ratio of GDS to GDP. Incidentally, gross domestic savings are much harder to measure accurately and show up nowhere in the comparisons.
The four remaining indicators relate to the external sector. Volume of trade per capita [(X +M)/POP5] plus balance of trade per capita [(X-M)/POP] reduces to 2 times exports per capita, [i.e., (X+M +X-M)/ POP = 2X/POP] as the indicator. If we view this indicator as an index of openness and we view openness as a determinant of growth of GDP in a proportional fashion, then this is an index of performance but it is redundant once we include growth itself as an indicator. Import Composition is measured by a lower share of imports of agricultural products and fuel indicating higher levels of performance or development. By this measure, Japan must be one of the least developed countries in the world!
Export Concentration, for example measured as the share of total exports of the major export product, is viewed as an inverse indicator of performance. A caveat acknowledging that Saudi Arabia is doing fine, despite its export concentration, attributes this to oil being a strategic, high-priced export, while other products, specifically sugar, are not. This caveat is less convincing when one remembers than in the late 18th century and early 19th century, sugar was referred to as white gold while oil was referred to as the devil’s shit. Actually, the underlying issue here is that variability of an input can translate into variability of the output and the latter will have negative welfare consequences if consumers are risk averse. The relevant indicator of performance in this case, however, is the variability of the output indicator not an index of an input that may or may not be the major contributor to variability of the output indicator at any one time. Similarly, trade partner concentration, measured as the percentage of exports going to the largest trading partner, is viewed as a direct indicator of bad performance. On this basis the leaders of Canada and Mexico must be out of their minds, since NAFTA is the best mechanism for raising their already extremely high score on this indicator of bad “performance” to its maximum.
Two of the remaining seven indicators are measures of borrowing: either from economic agents in a future period (the fiscal balance) or from foreigners (the external debt per capita). They have the characteristic that whether they are good or bad depends on the stage of the business cycle or the stage of development of a country, respectively, and on how well the borrower uses the resources borrowed.6 Two others, the real wage and open unemployment rates, are indicators of aspects of the distribution of income or poverty incidence. If one had good income distribution or poverty incidence data, they would be unnecessary. Nonetheless, we usually do not have such data and much less so in this case. Hence, by themselves they perform a very useful role in making comparisons. Similarly, inflation is useful as an indicator of performance if you cannot measure other things well, either prices or GDP or both; otherwise, it becomes controversial. For instance, there is an old literature in which it is argued that a little bit of inflation is good for growth for various reasons while acknowledging that high inflation is detrimental for growth.7
The last two indicators, GDP or GSP growth and GDP per capita or GNP per capita, are conceptually appropriate as indicators of performance by themselves. Undoubtedly the greater the growth of GDP or GSP the greater the economy’s capacity to produce and the greater GDP per capita or GNP per capita the greater the capacity to produce per person and the capacity to consume per person, respectively. It is in the construction and evaluation of the reliability of data underlying individual indexes that Carmelo is at his best and these two indexes provide excellent examples. In the case of these series his meticulousness is particularly valuable, because of the difficulty of comparing two different accounting systems, i.e., the capitalist and the socialist, and the problems with discontinuities in the underlying series for Cuba. Nonetheless, it serves no useful purpose to add these two indicators of performance in the construction of an overall index due to the considerable amount of overlap or double counting that would exist. Of course, this does not detract from the usefulness of each indicator by itself.
Summing up our discussion of economic indicators, there is a long tradition in Economics of evaluating overall country performance by looking at GDP per capita or GNP per capita. It is the best summary measure we have because it minimizes or eliminates the two main problems identified above with most of Carmelo’s indicators: either more of the indicator is not always good or bad or there is considerable overlap or double counting with other indicators. The one widely recognized and accepted problem with this measure, especially in the development literature, is that it does not capture movements in the income distribution. Considerable attention has been devoted to the evaluation of situations where this measure and income distribution measures yield different evaluations of performance, for example Fields (1980). Constructing a linear additive index of the two dimensions has never been recognized as a solution. For instance, Paes Barros and Mendoza (1995) show that, for certain classes of welfare functions that exhibit aversion to inequality, if per capita income increases and poverty incidence decreases, welfare increases regardless of what happens to other inequality measures. What is the usefulness of these complex procedures if simply adding up indicators were deemed to be a satisfactory approach? Their usefulness arises because a simple adding up is viewed as unsatisfactory or misleading.
More recently, in the development literature there has been a movement to consider dimensions of performance other than the economic one under the influence of Sen’s (1987) concept of capabilities, for example Betancourt (1996). Two dimensions stressed by Sen have been education and health and an additional dimension stressed by Dasgupta (1993) has been control over one’s life or freedom to engage or participate in various activities of a civic, political or economic nature. How to measure these non-economic dimensions has been the subject of less research and discussion than the economic dimension. Nonetheless similar principles apply. Carmelo is aware of this literature and provides six indicators of performance that fall into these areas under the heading of social standards.
All six of the indicators satisfy the criterion that having more (or in some cases less) of the indicator by itself is always good. In the case of education, the measures used by Carmelo are illiteracy and educational enrollment in secondary and higher education. One can argue that the first one is a measure (an inverse one) of quantity and the second one of quality so that there is little overlap or double counting in the two measures. The main issue in adding these two is why use equal weights, which Carmelo does. The UN Human Development Index, for example, uses unequal weights in adding up similar measures. The choice of weights is arbitrary. With respect to the health measures, Carmelo uses three indicators: infant mortality, life expectancy and the rate of five contagious diseases. Here one problem is overlap or double counting of a health indicator when adding the third measure to the first two. Another one is why give all three indicators equal weights? From a measurement point of view, the use of life expectancy and infant mortality as different indicators of health can be justified on the grounds that the latter is determined by different factors and picks up quality dimensions of health by focusing on a particularly vulnerable sub-sample. Nevertheless, it is difficult to find a rationale for why the third measure does not contain a considerable overlap or double counting with respect to the previous two as a health indicator. This is the case regardless of how valuable or interesting the measure may be by itself. Finally, the last indicator of the 20 in Carmelo’s aggregate indexes is the number of housing units. Again this is an interesting piece of information by itself but its role in an aggregate index of performance is less clear. After all, housing expenditures are included in GDP so there is double counting with that indicator.
Incidentally, Carmelo considers other indicators in some detail but excludes them from the combined index for various reasons. The latter can perhaps be described as difficulty of establishing comparability at various points in time suitable for the analysis in the book. Nonetheless, some of these indicators are interesting by themselves regardless of their inclusion in an overall index. Similarly, and particularly noteworthy in our context, Carmelo discusses indexes of economic freedom and of civic and political liberties in his comparisons with other international rankings (pp. 604-605). Not surprisingly, Cuba is at the bottom of both types of indexes while Chile is at the top in the first one and Costa Rica is at the top in the second one.
Carmelo concludes this part with a brief update from 1994, when most systematic comparisons end, to the most recent year feasible. This section contains also his assessment of the viability of the three models.
Summing up, this book does many things and it does most of them very well. The emphasis on adding up problems in our discussion has to be viewed in the context of the more general contributions in the book. Furthermore, if we look at the substantive conclusions derived from the aggregate indexes with respect to how these three different countries perform, the qualitative answers given by the aggregate indexes, despite their conceptual flaws, are broadly consistent with alternative ways of comparing the three countries using the individual indicators presented by Carmelo. Namely, Cuba does poorly economically and adequately in terms of health and education. For example, as Juan Belt points out in his comment, one of the most intriguing pieces of information in this book is provided by infant mortality rates in the three countries between 1960 and 1992 (table V.21). Cuba starts in 1960 with an infant mortality rate of 35.9 per 1000 and improves to 10.2 per 1000 by 1992; Chile and Costa Rica start in 1960 at 119.5 and 74.3, respectively, and improve to 13.9 and 13.7, respectively, by 1992.
1. I would like to thank J. Pérez-López for his comments on an earlier version.
2. Even the policy of import substituting industrialization (ISI), which has been severely criticized in most settings, has been recently given legitimacy in certain settings by a prominent member of the profession in a mainstream journal, Bruton (1998).
3. For instance, since important factors in bringing about the crisis of the 1980s were the oil price shocks of 1973 and 1979, one could argue that the reason Cuba was not affected, namely subsidized Soviet oil, disappeared in 1989 and Cuba was eventually affected by the same factors. Similarly, one could also argue that the recovery from the crisis started earlier in Chile and Costa Rica because, among other reasons, they adjusted first.
4. It is well known that national income accounting requires GDI = GDS + FS, where GDS is gross domestic savings and FS is foreign savings.
5. X is exports, M is imports, and POP is population.
6. Mesa-Lago acknowledges the second point with respect to the external debt.
7. While this literature is old, it is still relied upon in modern textbooks on development, e.g., Perkins, Radelet, Snodgrass, Gillis and Roemer (2001, Chapter 13).