2. Cross-section versus time-series

In the sub-section ‘Cross-section versus time-series’, Chang (Reference Chang2011) describes various episodes of different countries in which these countries showed historically low growth rates during periods characterized by liberalized markets. In my view, such an approach proofs nearly to nothing, because it does no justice to other factors relevant for economic growth. One could say that the analysis suffers from a missing variables bias. Take the case of Korea, Chang's second example. Chang claims that in South Korea per capita income growth was higher during the decades before the liberalization measures were implemented than during the period these measures were in place. These measures were introduced as a part of the International Monetary Fund (IMF) programs resulting from the Asian crisis of 1997.

Chang ‘forgets’ to inform us about the performance of other countries during this period. From 1999 to 2008 annual per capita growth in Korea was slightly more than 5% whereas many industrial countries showed a growth rate of about 2%. Hence, Chang disregards factors characterising the worldwide trend during the periods concerned. Moreover, in 1997 per capita income was already relatively high in South Korea, which reduces the possibilities for a further improvement. That is why growth equations always contain the initial level of income per capita. The expected coefficient is negative; the higher initial income, the lower the growth rate over the subsequent period is expected to be.

If Chang's approach is not valid, how then should one perform a time-series approach? First of all, the study's objective and its plausible determining factors should be formulated clearly. Matteo Marini (Reference Marini2004) is a good example in this respect. He distinguishes three stages of economic development: underdevelopment, transition and developed. Thereafter he formulates hypotheses about dominant attitudes – different cultures – to be relevant for each phase of development. A second feature of a good time-series analysis of the role of institutions explicitly takes into account that institutions do not change that often. Hence, the study should encompass a relatively long period so that institutional changes can be observed. However, as the above case of South Korea illustrates, this still leaves the possibility that changes in economic performance are invalidly ascribed to the changes in institutions. A third characteristic is, therefore, that the study is not confined to one country. Including various countries introduces variation in institutional structure, and thus the possibility to disentangle the influence of institutions and institutional changes from those of other variables. Finally, the causes of institutional change and their effects are frequently of a non-economic nature. Hence, researchers should be open to information from different disciplines and sources, and willing to use various methods. These could include regressions analyses, historical narratives and comparative historical studies.

It is difficult to accomplish a study that meets all the criteria I have mentioned above. Alberto Alesina and Edward Glaeser's (Reference Alesina and Glaeser2004) study of the way poverty is reduced in the USA and Europe comes close to such an ideal analysis. It studies the data on redistribution policies and social welfare in these countries. From there it investigates the plausibility of various explanations for the systematic differences between the countries. Explanations considered are economic, political (including political institutions), race and ideology. The methods encompass regression analyses, comparative history and the analysis of surveys.

3. Culture and institutions

At various places Chang refers to a relation between values and institutions ‘the very definition of a free market depends on whether an observer accepts the political and ethical values embodied in the institutions that gird the market’ (Chang, Reference Chang2011: 6). In the section ‘Against the fatalism of the climate-culture school’ he attacks the climate-culture school which according to him ‘thinks that all institutions in a country are permeated by one “tradition”’ (ibid., 19). This approach can ‘easily degenerate into ex post justification’ (ibid.). Thereafter, Chang substantiates his claims by presenting examples, which illustrate that a particular world view (Confucianism and Islam) or nationality (German and Japanese) are sometimes regarded as pro-development and during other periods as against economic prosperity. From these examples one can easily draw the conclusion that values, institutions and economic development are not related to each other. My criticism is that such examples illustrate only that opinions differ over time and place but are of no help for investigating the relation between values, institutions and economic development. Such an analysis requires a much more systematic approach.

Such a systematic approach is provided by authors belonging to what I now label the ‘culture matters’ approach. In its most general form these authors argue that values (culture), institutions and economic performance influence each other. Causality can run both ways. Several studies have implicitly or explicitly applied this framework for explaining cross-country difference in institutions and for relating episodes of stagnation and growth to differences in attitudes (for a review of cross-country comparative studies, see De Jong, Reference De Jong2009: chapters 5 and 6).

The degree of centralization is one aspect by which institutions across industrialized countries differ. In some countries, France being a good example, both private and public organizations have a hierarchical structure. Firms' ownership is concentrated and political decision making is centralized as well. At the other extreme are countries with decentralized organisations. They are characterized by dispersed private ownership and in some cases (the USA, for example) an aversion against any centralization of policy and regulation. The degree of centralization of political and private organizations corresponds with the countries' score on Hofstede's power distance index. People in societies with a large power distance believe that inequality is existential: a majority of people perceives that there should be an order of inequality in which everybody has a rightful place, and such an order provides the best protection for everyone. Fellow citizens are a potential threat to one's power and rarely can be trusted; cooperation is thus difficult to sustain. People in societies with a small power distance believe that hierarchy in society is an inequality of roles established for convenience. People of different power levels feel less threatened and more prepared to trust others and engage in cooperation with them. The scores on power distance correlate negatively with cross-country data on local public expenditure (employment) as a percentage of total public expenditure (employment) as provided in Page and Goldsmith (Reference Page, Goldsmith, Page and Goldsmith1987: 157). This illustrates the embeddedness of institutions. This does not mean that institutions or culture cannot change but that changes in one have to take account of the other.

The relevance of institutions' embeddedness comes also to the fore when foreigners provide advice for transition or developing countries. Iwona Sobis and Michiel De Vries (Reference Sobis and de Vries2009) studied the Western advice to an employment agency in the Polish city of Lodz. They found that according to the Polish officials the advice by the Swedish, German and Danish was highly useful. In their view these advisors paid attention to their needs and suggested solutions without referring to the organizations in their own country. They regarded these differences to be too large and the situation in a transition country to be too much different from that in an industrialized country. The French, however, tried to promote their own bureaucratic model, while pragmatic ideas were needed. Moreover, ‘the communication between the consultant and the personnel was obscure and arrogant’ (ibid., 96). As a result, the French advice was not adopted.

The examples given by Chang refer to culture and economic development. His implicit message is that values, institutions and economic development are unrelated. Others have studied these relationships in a more systematic way. In particular, attention has been paid to the cultural attitude during the transition period from an underdeveloped to a developed industrialized economy. Several authors have pointed out that such a period is characterized by a need for achievement (McClelland, Reference McClelland1961), thrift and an orientation towards the future. Hofstede finds that this future orientation is positively correlated with a subsequent period of growth in poor countries (Hofstede et al., Reference Hofstede, Hofstede and Minkov2010: 265). So, a systematic investigation finds a relation between values and periods of economic development, although these values are not exclusively associated with a particular religion or nation as is suggested in Chang's examples.