Content of revision
Whereas this revision does lead to some substantial changes as compared to the previous version of the KOF Globalisation Index, we are still constrained by a number of factors in the selection of the variables. While we wish to consider as many individual variables as possible to portray the multifaceted concept of globalization and to exploit their variation when using principal component analysis, we rely on variables with a broad coverage that are updated regularly. We need variables that cover basically all countries in the world from 1970 onwards. This excludes many variables that are, for example, collected for OECD countries only. Because we continue to release an update of the KOF Globalisation Index on a yearly basis, we need variables of which we know that these will be regularly updated and published. Furthermore, using principal components analysis to determine the weights of the individual variables requires a minimum number of variables. As we do not want a single variable to dominate one particular sub-dimension, we employ the rule that for each sub-dimension we need at least three variables. Distinguishing between de facto and de jure globalization in every sub-dimension (trade, financial, interpersonal, informational, cultural and political) requires a minimum set of 36 variables.
Distinction between de facto and de jure globalization
The new KOF Globalisation Index distinguishes between de facto and de jure measures of globalization. We compute a separate index for the de facto and de jure economic, social and political dimensions of globalization. On the sub-dimensional level, we compute a separate index for de facto and de jure trade, financial, interpersonal, informational and cultural globalization. Some variables in the 2007 version of the KOF Globalisation Index measure the possibility for information exchange between international actors. These variables include access to television and internet and are reclassified as de jure indicators in the revised KOF Globalisation Index. Many new variables, especially measuring de jure globalization, are introduced.
Distinction between trade and financial globalization
The economic dimension of the revised KOF Globalisation Index consists of the two sub-dimensions: trade globalization and financial globalization. We consider this a key advantage over the 2007 version of the KOF Globalisation Index and other globalization indices. The distinction between trade and financial liberalization has already been employed in previous studies. Jaumotte et al. (2013), for example, examine how trade and financial globalization influence income distribution within a country. The results show that trade globalization was negatively associated with income inequality, and financial globalization was positively associated with income inequality.Footnote 4 Kose et al. (2009) find that both trade and financial globalization influence the nexus between output volatility and growth. The effect tends to be stronger for trade globalization. Other studies, however, show that trade and financial globalization go hand in hand (see, for example, Lane and Milesi-Ferretti 2008).
Measuring cultural globalization more broadly
Another feature of the revised KOF Globalisation Index is to measure cultural globalization more broadly. We include more variables that do not rely on individual value concepts. The original selection of variables in the 2002 version of the KOF Globalisation Index was based on an understanding of cultural globalization based on Saich (2000), which defines modern cultural globalization largely as the dispersion of American values. It was measured by the number of McDonald’s restaurants in a country. The focus on American values was somewhat relaxed in the 2007 version of the index by including the number of IKEA stores and trade in books as additional variables. In any event, the KOF Globalisation Index has been criticized for rather measuring Westernization than cultural globalization in general. In the revised version of the KOF Globalisation Index, we include three new de facto variables measuring cultural globalization, of which none relies on an individual value concept. These variables measure trade in cultural goods, trademark applications of non-residents and trade in personal, cultural and recreational services. The variables McDonald’s restaurants and IKEA stores are still included in the index.
Time-varying weights for the aggregation
The revised KOF Globalisation Index includes time-varying weighting of the individual variables in the aggregation process. As in the 2007 version of the KOF Globalisation Index, we use principal component analysis to determine the weights of the individual variables for the lowest aggregation level of the index. However, we no longer use the full sample years to determine time-invariant weights, but instead apply principal component analysis on rolling windows of 10 years to calculate time-varying weights. This procedure has the advantage of letting the weights adjust over the years to account for changes in the role of individual variables in serving as proxies for globalization.
We reassess the ability of some variables contained in the 2007 KOF Globalisation Index to measure flows of information and communication over the entire time span. Some variables in the 2007 version of the index, such as international letters, trade in newspapers and trade in books, are strongly affected by the digitalization and the internet and are gradually replaced by different information and communication channels. Ideally, we would like to include variables measuring those new channels alongside with the variables measuring the traditional channels to capture some of the substitution between the two variables over time. However, when no measures for those new means of communication are readily available for many countries, we exclude some of the previous variables. Still including these variables would indicate that social globalization is decreasing, while in fact only the means of communication are changing.
Dimensions of the index and variable selection
De facto economic globalization
De facto trade globalization
The sub-dimension de facto trade globalization refers to the exchange of goods and services over long distances. This is measured using the variables exports and imports of goods and exports and imports of services, both measured as a share of GDP. To account for the geographical distribution of trade linkages, we include a variable that measures trade partner diversity. It is computed as the inverse of the average over the Herfindahl-Hirschmann trade partner concentration index for exports and imports of goods. That is, indexing countries by i and their trade partners by j, the Herfindahl-Hirschmann index is equal to \( HH{I}_i={\sum}_{j=1}^n{\left({a}_j^i\right)}^2 \), where \( {a}_j^i \) is the share of trade partner j in country i’s exports or imports. The more dispersed exports and imports of a country over different trade partners are, the lower HHIi and the higher the value of our variable. The variable trade partner diversity favors countries whose export and import structure is globally oriented as compared to countries that primarily trade regionally. Due to data limitations, we compute trade partner diversity only for trade in goods.
De facto financial globalization
De facto financial globalization is measured by capital flows and stocks of foreign assets and liabilities. We thus use a quantity-based measure as opposed to a price-based or news-based measure of financial globalization (Baele et al. 2004). Kose et al. (2009) propose to focus on the sum of stocks of foreign assets and liabilities instead of flows to mitigate the problem of volatility and measurement errors in the flow variables. Following Lane and Milesi-Ferretti (2007, 2018), we include variables on foreign direct investments, portfolio investments, international debt and international reserves (excluding gold). All variables are calculated as the sum of stocks of assets and liabilities and normalized by GDP. We also include the sum of primary income payments and receipts as a share of GDP. It comprises earnings and payments arising mainly from the cross-border provision of labor and capital. For historical values of all stock variables, we rely on the updated and extended dataset External Wealth of Nations by Lane and Milesi-Ferretti (2018), which comprises information about the composition of the international financial position of a large sample of countries.
De facto social globalization
De facto interpersonal globalization
This sub-dimension captures direct interactions among citizens living in different countries. The interaction can occur by the means of personal calls across borders. Personal calls are measured as international voice traffic in minutes per capita using fixed or mobile telephones. Personal contact with foreign citizens is the most likely form of direct interactions, which we measure using three variables. Migration, measured as the stock of foreign-born persons in a country, is the most persistent form, while tourism and foreign students (both counted as inbound and outbound) can be regarded as some form of temporary migration. Finally, international transfers paid and received always include some sort of personal interactions. All variables are normalized by domestic population.Footnote 5
De facto informational globalization
While interpersonal globalization is intended to capture personal interactions, informational globalization is meant to measure the actual flow of ideas, knowledge and images. De facto informational globalization is measured using three variables. Internet bandwidth measures the used capacity of international internet bandwidth and serves as a proxy for international digital information in- and outflows. International patents, measured as the stock of patent applications made by non-residents, describe international flows of technology, scientific knowledge and related information (OECD 2010). High technology exports describe flows of technological and scientific information. While international patents mainly describe an inflow of information, high technology exports mainly represent the outflow of information. All variables are normalized by domestic population.
De facto cultural globalization
Cultural globalization is the most difficult dimension to measure. Following Saich (2000) and Dreher et al. (2008), it refers to some extent to the domination of U.S. cultural products, measured using the number of McDonald’s restaurants. The definition has been expanded to western countries being trendsetters in much of the cultural realm, represented by the introduction of the number of IKEA stores to the index. In the revised version of the index, the definition is extended and new variables that do not follow an individual value concept are introduced. The stock of trademark applications by non-residents, representing the exchange of foreign trademarks, are introduced. The variable is conceptually very close to McDonald’s restaurants or IKEA stores. However, it does not focus on American or any other individual culture. We also include two variables that describe the transmission of cultural values by the means of sharing cultural goods and services. Following Disdier et al. (2010), we include trade in cultural goods based on the definition by UNESCO (2009). Following Hellmanzik and Schmitz (2015), we introduce trade in personal, cultural and recreational services, a subcomponent in the Balance of Payments. It includes, for example, services related to provision of cultural goods such as production of motion pictures or musical records, organization of sport events or operation of museums. Both variables are measured as the sum of exports and imports and normalized by domestic population.
De facto political globalization
De facto political globalization captures the diffusion of government policies. It is measured using the variables participation in UN Peacekeeping missions, the number of embassies and international NGOs in a country. The presence of embassies implies foreigners acting in their home countries’ interest. Hence, it is an indication of how much a government accepts foreign sovereign governmental influence and resources. International NGOs are counted as the number of internationally oriented NGOs active in a country. Similar to an embassy, the presence of international NGOs involves the influence of foreigners with political or social motives in one’s own territory, which is interpreted as political influence from abroad.
De jure economic globalization
De jure trade globalization
The sub-dimension de jure trade globalization relates closely to the sub-dimension economic restriction in the 2007 version of the KOF Globalisation Index (Dreher et al. 2008). It refers to policies that facilitate and promote trade flows between countries. It is measured using variables on trade regulation, trade taxes, tariff rates and free trade agreements. Trade regulation includes the average of two subcomponents: prevalence of non-tariff trade barriersFootnote 6 and compliance costs of exporting. The variable trade taxes measures the income of taxes on international trade as a share of total income in a country. The variable tariff rates refers to the unweighted mean of tariff rates. The variables trade regulation, trade taxes and tariff rates are calculated as the inverse of the normalized values such that higher values relate to a higher level of de jure trade globalization. Free trade agreements refer to the stock of multilateral and bilateral free trade agreements.
De jure financial globalization
The sub-dimension de jure financial globalization measures the openness of a country to international financial flows and investments. The IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER) is the primary source for most measures of de jure financial globalization (see, for example, Quinn et al. 2011). It measures the openness of the capital account of a country. We include the most widely used index based on the AREAER reports: the Chinn-Ito index (Chinn and Ito 2006, 2008).Footnote 7 The second variable measures investment restrictions based on the WEF Global Competitiveness Report. To account for policies that are potentially favorable to capital flows, we include the number of international investment agreements, which covers bilateral investment agreements and treaties with investment provisions.
De jure social globalization
De jure interpersonal globalization
De jure interpersonal globalization refers to policies and resources that enables direct interactions among people living in different countries. The variables we have chosen are conceptually close to the ones we use for de facto interpersonal globalization. We use the number of mobile phone and telephone subscriptions per capita. Movement of people across borders, such as migrants, tourists or students, count for a substantial amount of de facto personal contact. The variable freedom to visit represents restrictions on international travel. It measures the percentage of countries for which a country requires a visa from foreign visitors. Travel visas, alongside passports, are key control instruments of population movements by modern states (Czaika et al. 2018). The number of airports hosting international flights, normalized by population, is a measure for international connectivity.
De jure informational globalization
De jure informational globalization refers to the ability to share information across countries. It is measured by the number of television sets per capita. It is also measured by the number of people having access to the internet. Additionally, the press freedom index captures the availability of news related information.Footnote 8 The index aims at portraying media independence and assessing the degree of print, broadcast, and digital media freedom.Footnote 9
De jure cultural globalization
The sub-index de jure cultural globalization refers to openness towards and the ability to understand and adopt foreign cultural influences. It is inspired by the GlobalIndex (Raab et al. 2008). The authors justify their choice of variables by highlighting their key role in quantifying the spread of common values of rationalism and hence cultural assimilation across the world. Three factors are important to measure the ability of understanding and accepting foreign cultural values. A great part of today’s international culture is influenced by an egalitarian view on the role of woman in society. Consequently, we assume that having an equally egalitarian view intensifies cultural assimilation. As an approximation of such views, we include the gender parity index on gross primary school enrolment. It is an indication of parity of boys and girls and as such a strong indicator of the equality of men and women. Secondly, as a measure of education that is assumed to foster the spread of common values, we use the human capital index calculated in the Penn World Tables. Third, we include the civil freedom index, an assessment of civil liberties published in the freedom of the world report. It quantifies aspects of civil freedom such as expression and belief, associational and organizational rights, rule of law and personal autonomy and individual rights. We assume these are important elements supporting cultural globalization.
De jure political globalization
De jure political globalization refers to the ability to engage in international political cooperation. It is measured using the number of multilateral treaties signed since 1945, the number of memberships in international organizations and a measure for the treaty partner diversity. The number of treaties and memberships in international organizations describe the communication and meetings of negotiators, with the intention to influence future relationships. They therefore rather characterize the willingness of creating networks than actual manifestation of flows. Partner constellations are informative when examining how a country influences global politics. Having the same number of treaties with a smaller number of partners rather reveals strong individual relationships than willingness to create global political networks. The variable treaty diversity measures the concentration of partners in international treaties. We use bilateral and not international treaties because we believe that negotiating a bilateral treaty indicates that each party was actively involved, which is not necessarily the case for international treaties. Because there is no encompassing database on bilateral treaties, we use bilateral investment treaties.
Method of calculation
The 2018 KOF Globalisation Index is based on 43 individual variables, which are aggregated to a de facto and a de jure index of five sub-dimensions (trade, financial, interpersonal, informational and cultural globalization), three dimensions (economic, social and political globalization) and one total index. We distinguish between 18 different indices if we maintain the distinction between de facto and de jure. We also report the overall index for the total, for each of the three dimensions, and each of the five sub-dimensions. The overall index is calculated as the average of the de facto and the de jure index. This increases the total number of indices to 27. This broad variety allows researchers to fine-tune towards those dimensions that they consider relevant for the individual research question they would like to examine.
Imputation of missing data
The KOF Globalisation Index is calculated on a yearly basis from 1970 to 2016 and for 203 countries and territories. The selection of countries and territories relies on the definitions by the World Bank. However, not all variables are available for all countries and years. Missing observations within a series are imputed using linear interpolation. Missing observations at the beginning or the end of a series are substituted by the closest observation available. Specifically, this implies that we carry the last non-missing observation backwards in case of missing observations at the beginning of a series and forward in case of missing observations at the end of a series. Data coverage increases for most variables over time, which means that imputation of missing data is more prevalent in earlier years of our sample.Footnote 10
Normalizing the data
Normalizing the data implies that each variable is transformed to an index with a scale from one to one hundred, where one hundred is assigned to the observation with the highest value across the whole sample of countries and years. The remaining observations are ranked according to the percentiles of the distribution. This panel normalization is different to annual normalization, where observations are normalized within a given year only. Panel normalization is not sensitive to outliers, which is a clear advantage over the original series. The disadvantage is that changes in the data in any year possibly affect the index value of countries in all years.
Determining the weights
We employ principal components analysis on a 10-year rolling window to determine time-varying weights for the individual variables. We use observations for t-10 until t-1 to compute the weights for time t. The weights for the years 1970 to 1979 are set equal to the weights of the year 1980, given the shorter time window.Footnote 11 Principal components analysis partitions the variance of the variables in each sub-group and the weights are determined in a way that maximizes the variation of the resulting principal component. We calculate the weights using the entire sample of countries. By applying time-varying weights as opposed to fixed weights determined over the observations of all years, we account for structural changes in the relevance of individual variables to capture globalization over time. While the weights of individual variables vary over years, the weights of the sub-indices are determined by giving equal weights to each component and are held fixed over the sample period. Economic globalization includes trade and financial globalization, both of which receive a weight of 50% within the economic dimension. Social globalization consists of interpersonal, informational and cultural globalization, each of them contributing a third to the social globalization index. Economic, social and political globalization are aggregated to the Globalisation Index using again equal weights. The overall globalization indices are calculated as the average of the de facto and the de jure indices. Table 1 shows the weights of the different levels of the indices.
Aggregation to indices
Once the weights are determined, the aggregation consists of adding up individual weighted variables to the desired level of aggregation. Each aggregation level is calculated from the individual variables instead of using the aggregated lower-level indices. Doing so has the advantage that variables can be used in higher aggregation levels of the index, even if the value of a sub-index is not reported because of missing data. A disadvantage is that the higher ordered dimensions can only be replicated using lower ordered dimensions for countries for which all variables are available. Observations of indices are reported missing if more than 40% of the underlying variables are missing or at least two out of three sub-indices cannot be calculated.