UNIVERSITY OF GRONINGEN
Faculty of Economics and Business
Msc. International Economics and Business
Revealed Comparative Advantage
A comparison of the Balassa index and
the value added approach
Dylan M. O. Jong
S1991361
jongdylan@gmail.com
7 July 2014
2
Abstract
Generally, literature uses the Balassa index (Balassa, 1965) as a measure for assessing the comparative advantage of industries (Utkulu and Seymen, 2004) as this is assumed to correspond closely with the value a specific industry adds. However, now globalization has entered a new phase (Baldwin, 2006) – in which inter-industry trade becomes increasingly important – this measure loses its accuracy according to Timmer et al. (2013). This thesis examines the accuracy and validity of measuring the Balassa index in today’s increasingly fragmenting world. It proves that the Balassa index is significantly different from RCA based on value added. This could have major implications for policy making, location and investment decisions of companies and the academic world.
3
1. Introduction
The comparative advantage of specific industries among nations has been discussed for many years and maintains its relevance today. Governments base and adjust their policies partly on the comparative advantage levels of industries. A country is said to have a comparative advantage in a specific industry when it is performing relatively better in that specific industry compared to other countries (Balassa, 1965). Similarly, a country is said to have a comparative disadvantage in a certain industry when it is performing worse in a specific industry than other countries (Balassa, 1965). Hence, every country has comparative
advantages in some industries and comparative disadvantages in others.
Measuring comparative advantage is challenging as relative prices under autarky cannot be observed (Balassa, 1989). However, in 1965, Balassa (1965) introduced a measure that could ‘reveal’ a country’s comparative advantage in certain industries. This revealed comparative advantage (RCA) is also known as the Balassa index (Balassa, 1965). The Balassa index (1965) measures the comparative advantage of a nation in a specific industry by comparing the gross export levels of countries. Balassa (1965) argues that one can reveal the comparative advantage by examining the post-trade patterns of countries as it “reflects
relative costs as well as differences in non-price factors.” The Balassa index is still used by
many studies as a measure for RCA (e.g. Algieri et al., 2011; Qineti et al., 2009; Fertö and Hubbard, 2004; 2002).
4
the other countries get a lower RCA level because of the higher total amount of industry and country exports.
Two errors occur when using the Balassa index (Timmer et al., 2013):
1. Foreign intermediate inputs are not taken into account: Assume country A is producing all parts for a car in its home market. Subsequently, it exports all parts to country B for assembly – the incentive being lower wages for instance. After country B has assembled the car, the completed car is then exported to a distribution center in country C. In this case, the export value for country A would represent the value that country A adds by producing the car parts. Assume country A adds a value of 5. For country B however, the export value does not solely represent the value that the country itself adds to the product (as it also included the value country A added), which leads to an overstatement of the comparative advantage. The completed car likely has a higher value than the separate parts: assume its value is 6. Now it seems country B is adding a value of 6, as the export value for the completed car amounts 6. This would imply that country B is adding more value to the car than country A did. However, when subtracting the value country A added by producing the car parts, we find that country B in fact only added a value of 1.
2. The influence of domestic final demand is neglected: When we take the car-example again, we now assume that the car is not only assembled but is also sold in country B. Still, the total export of country A would be 5. However, the total export of country B now decreases to 0, which would imply the country does not add any value to the car. As a result, the additions of country B with a value of 1, are not taken into account when measuring the RCA when looking at total exports.
Timmer et al. (2013) therefore advice to “focus on activities that are directly and indirectly
involved in production of final manufacturing goods and measure their value added.” Taking
value added as a basis for measuring the RCA of a specific industry within a country, provides us a much more accurate picture as it does take into account (1) foreign intermediate
inputs, (2) and the domestic final demand. Hence, this approach is considered to be much
more accurate than the approach based on gross exports (Timmer et al., 2013). Nevertheless, a direct comparison between these two measures has not been performed yet.
5
by measuring the RCA based on the value that is added within a country. This thesis poses the following research question:
To what extent does the Balassa index differ from the RCA based on value added? Is the Balassa index still an accurate measure?
This thesis is organized as follows: in section 2, a literature review is presented in which the hypotheses are also introduced. Section 3 presents the methodology. Section 4 will present the data along with a short description. In section 5 the results are presented, after which a conclusion is drawn from them. The final section discusses the implications for further research. Subsequently, section 8 displays the references and section 9 is the appendix.
2. Literature
In this section we provide an overview of the effects of globalization, as presented by Baldwin (2006), that have led to a large degree of fragmentation in production processes. Subsequently, several studies that have done empirical research on fragmentation are discussed. Furthermore, the measurement of comparative advantage is discussed.
2.1 Fragmentation
During the last two centuries, the world has become increasingly globalized. Baldwin (2006) divides this globalization period into two: the first- and the second unbundling. The first unbundling occurred in two waves – one from roughly 1850 to 1914, the other from the 1960s to the present (Baldwin and Martin, 1999). Baldwin (2006) groups the impact of the first unbundling into six stylized facts. First, the Industrialization/Deindustrialization; in the first wave, the Western Europe and the US industrialized while East-Asia (mainly India and China) deindustrialized. In the second wave, the East Asia industrialized while Western Europe and the US deindustrialized. Secondly, there was the International divergence/convergence; the first wave saw Western Europe and the US incomes diverge
6
The fourth fact is Growth Take-Off. Sometime before the first wave started, the Industrial Revolution led to extensive modern growth in Western Europe and the US, while East-Asia lagged behind in terms of per capita income. In the second wave of the first unbundling, income convergence led to spectacular growth in East-Asia, where growth in Western Europe and the US stagnated. Urbanization is the fifth stylized fact of the first unbundling. While some of world’s largest cities resided in East-Asia before the first unbundling started, after the first wave cities in Western Europe and the US started emerging rapidly. In the second wave however, East-Asian cities took over the lead again in terms of growth rates. Lastly, income
inequality rose during the second wave of the first unbundling of globalization as well as
unemployment rates. Most importantly, inventions such as railroads and steamships lowered transportation costs and led the separation of production and consumption: products no longer needed to be produced and consumed in the same region or country. Production and consumption now became separate concepts. Now something produced in one country, could be transported to another country to be consumed. As production took place in one area, it was argued that gross exports provide a quite accurate measure in terms of RCA during the first unbundling (Balassa, 1989). As products were still often produced from scratch in one country, gross export includes all the value that is added by that particular country.
7
more specializing in certain activities rather than in certain industries. Baldwin and Lopez-Gonzalez (2013), argue that the production activities are not fragmented globally but rather mostly just regionally. The article speaks of the regions as Factory Asia, Factory North America and Factory Europe. This finding has been empirically investigated by Los, Timmer and de Vries (2014) by extending the fragmentation measure that was introduced by Feenstra and Hanson (1999). They firstly find that the value chains have indeed become more internationally fragmented. Secondly, they find “that global fragmentation of value chains
has progressed much faster than regional fragmentation”, indicating that fragmentation has
occurred more at a global level compared to the regional level as was proposed by Baldwin and Lopez-Gonzalez (2013).
Because of fragmentation, it becomes questionable whether it is possible to measure the comparative advantage of nations with the Balassa Index (1965), based on gross exports (Timmer et al., 2013). The problem that fragmentation brings for calculating the comparative advantage of nations is that part of the exports is actually not produced domestically, but consists of foreign inputs. In today’s world we therefore need to focus on activities that are directly and indirectly involved in production of final manufacturing goods and measure their value added (Timmer et al., 2013).
2.2 Comparative advantage
The two most well known theories on comparative advantage are the Ricardian theory and the Heckscher-Ohlin (H-O) theory (Utkulu and Seymen, 2004). They differ in the sense that the Ricardian theory argues that differences in technology across countries results in comparative advantage while the H-O theory argues that differences in factor prices result in comparative advantage (Utkulu and Seymen, 2004). It has been difficult to measure comparative advantage and test the H-O theory because one cannot observe the relative prices under autarky (Balassa, 1989). Balassa therefore suggested to take the RCA by looking at the trade patterns. Even though the Balassa index could be considered the most well known RCA measure, it was not the first RCA measure. Liesner (1958) was the first one to come up with the RCA literature. His proposed RCA formula was as follows:
8
where X represents exports, i is the country, j is the industry and n is a set of countries. Balassa (1965) extended this formula in the following way:
(2)
where again, X represents exports, i represents a country, j represents an industry and n represents a set of countries. Also in the formula is now: t, which represents a set of industries. A country has a revealed comparative advantage in a specific industry when its RCA is higher than 1. Similarly, a country has a revealed comparative disadvantage when its RCA is lower than 1.
There have been multiple articles written on the stability and consistency of the measures of the RCA index (e.g. Yeats, 1985; Hinloopen and van Marrewijk, 2001). Yeats (1985) found that the RCA index faces difficulties with the distribution of country index values within different industries. He argues that if for a given industry the associated country index values are all highly concentrated in a range slightly above or below unity, the nation with the greatest comparative advantage in the industry may have a relatively low RCA index value (Yeats, 1985). Conversely, if production and exports of a second industry are highly concentrated in a relatively few countries, it is possible that a nation which does not have the greatest comparative advantage, may still have a very high index value (Yeats, 1985). Hinloopen and van Marrewijk (2001) found that the formula for calculating the RCA values is well defined as the distribution changes very little from one period to another. Furthermore, they find that the distribution across countries differs considerable, which makes the RCA a good measure for comparing countries (Hinloopen and van Marrewijk, 2001). The fact that the RCA value is bounded from below and not on the upper-side makes it difficult to compare the differences between below and above unity. This is a minor flaw that the RCA has.
Greenaway and Milner (1993) and Vollrath (1991) have proposed additions or alternatives to the Balassa index. Greenaway and Milner (1993) argue that the imports should also be taken into account. Taking imports into account solves the problem of intermediate
inputs (error 1, see introduction). However, it does not solve the problem of domestic final demand (error 2, see introduction) Furthermore, it led to some ambiguities around the zero
9
lead the values to become symmetric through the origin (Utkulu and Seyman). The RTA is calculated as the difference between the relative export advantage. These alternative measures eliminate some of the double accounting problems, but not all (Utkulu and Seymen, 2004). Furthermore, they do not take into account the final domestic demand, which leads to an error (see introduction). For this thesis, the RCA based on value added is compared to the Balassa index. We chose the Balassa index as it is the most well known, widely accepted and frequently used RCA measure (Utkulu and Seymen, 2004).
Current literature points to the fact that due to more fragmentation and the emergence of global value chains, it becomes more and more incorrect to assume that the Balassa index (1965) – based on gross exports – provides an accurate and similar view to the RCA based on value added (Timmer at al., 2013). Therefore, we expect significantly different RCA levels for the Balassa index compared to the RCA based on value added. We therefore hypothesize:
Hypothesis I: During/after the second unbundling RCA based on value added provides a more accurate picture of the competitiveness of a country than RCA based on
exports.
In the next section, we will introduce the methodology.
3. Methodology
As mentioned, this thesis compares the Balassa index with the RCA based on value added, using the following equation:
(2)
where, X represents exports, i represents a country, j represents an industry, n represents a set of countries and t represents a set of industries.
A comparable formula is used for calculating the RCA based on value added. However, instead of using gross exports, value added is used as a basis. This results in the following equation:
10
where, VA represents exports, i represents a country, j represents an industry, n represents a set of countries and t represents a set of industries.
By taking the value added, the two errors (see introduction) are no longer a problem. The error of the foreign intermediate inputs is not causing any trouble when using this RCA measurement because these inputs are not seen as value that is created in the specific country. It therefore does not lead to double-accounting. Furthermore, the second error – final domestic
demand – is taken care of as well, because the focus is no longer just on export but on the value that a country adds.
To compare the results of the RCA based on value added with the Balassa Index, we used multiple tests and graphical plots. We started with a plot of the empirical cumulative distribution function of both variables in order to get an overview of how both variables are distributed compared to each other. As the graph is only meant as an indication, we will only plot the graph of all years pooled together. With the distributional graph in mind, we will measure for differences between the two RCA groups. We will first determine whether the two groups have equal variances in order to see if the two groups are similarly distributed. After we have established whether or not the two samples are varying similarly, we will compare their means.
Ruxton (2006) argues that “if you want to compare the central tendency of two
populations based on samples of unrelated data, then the unequal variance t-test should always be used in preference to the student’s t-test or Mann-Whitney U test”. Ruxton (2006)
11
nominal rate when there was a higher variance together with a higher sample size (Coombs et al., 1996). These results are in accordance with similar studies on these two tests (e.g., Zimmerman and Zumbo, 1993). Multiple studies found that even when the population variances are equal, the power of the unequal variances t-test still is similar to that of the student’s t-test (Moser et al., 1989; Moser and Stevens, 1992; Coombs et al., 1996). Ruxton (2006) argues that the unequal variance t-test performs as well as, or better than the student’s t-test when looking at the control of both the Type 1 and Type 2. Besides the student’s t-test and the unequal variances t-test, a third option is the Mann-Whitney U test. Zimmerman and Zumbo (1993) found that in terms of Type 1 errors, the unequal variance t-test performs just as good as the Mann-Whitney U test when variances are equal and considerably better when variances are unequal. They therefore suggest that the unequal variance t-test can effectively replace the Mann-Whitney U test (Zimmerman and Zumbo, 1993). In case there is evidence of non-normality in one of the two variables, it is advised to first rank the observations before using the unequal variance t-test (Ruxton, 2006). We therefore perform a Shapiro-Wilk test to check for non-normality, after which we will decide whether or not to rank the observations before performing the unequal variances t-test.
12
13
Unfortunately, this table ‘only’ goes up and until a sample size of 350. However, Hinloopen and van Marrewijk (2005) have developed a rule-of-thumb that one can use in order to derive the critical values. This is shown in table 1.
Table 1: Rule-of-thumb for calculating the critical values for the HM index
Critical percentile: ln(cp) = β0 + β1ln(n) Percentile 90 95 97.5 99 β0 0.3503 0.5018 0.6230 0.7508 (-0.0008) (-0.0006) (-0.0005) (-0.0007) β1 -0.5008 -0.5005 -0.5001 -0.4994 (-0.0002) (-0.0001) (-0.0001) (-0.0001) R2 0.9999 0.9999 0.9999 0.9999 Source: Hinloopen and van Marrewijk (2005)
By using this rule-of-thumb, we can derive the critical values for performing hypotheses testing for differences between the two variables.
4. Data
The quality of the results always largely depend on the quality of the data. For this research, the World-Input-Output Database (WIOD) is used. WIOD is a publicly available database that provides time-series of input-output tables for 40 countries worldwide for the period 1995-2011.1 The WIOD complies with national accounts and international trade statistics. The database is constructed in monetary terms (millions US dollars). As we work with ratios, there is no need for deflating. Even though WIOD consists of 40 countries in general, the current analysis is confined to 39 counties listed in Appendix 1, as the export levels are not available for the rest-of-the-world. The countries that are used in this research together make up for 85 percent of world’s GDP in 2008 (Timmer et al., 2013). In addition, the WIOD consists of 35 different sector (Appendix 2). Together, the 40 countries and 35 industries make up for 1,400 RCA values for both RCA groups per year. The total amount per group for all 17 years amounts 23,800 observations. A summary of the data is presented in table 2:
14
Table 2: Descriptive statistics of RCA based on value added (V) and the Balassa index (E)
Variable N Mean SD Min Max Variable N Mean SD Min Max
V1995 1400 1.11 0.87 0.00 8.06 E1995 1400 1.41 2.98 -0.59 68.12 V1996 1400 1.10 0.84 0.00 7.65 E1996 1400 1.39 2.90 -0.17 68.39 V1997 1400 1.10 0.87 0.00 8.18 E1997 1400 1.40 3.00 -0.21 74.08 V1998 1400 1.10 0.88 -0.09 9.85 E1998 1400 1.38 3.01 -0.80 75.96 V1999 1400 1.11 0.89 0.00 8.68 E1999 1400 1.38 3.01 -0.84 73.75 V2000 1400 1.13 0.92 0.00 9.83 E2000 1400 1.36 2.91 -1.06 69.47 V2001 1400 1.13 0.94 0.00 12.40 E2001 1400 1.32 2.76 -1.25 67.24 V2002 1400 1.13 0.92 0.00 8.18 E2002 1400 1.30 2.69 -2.57 64.61 V2003 1400 1.12 0.91 0.00 9.79 E2003 1400 1.28 2.70 -3.96 67.94 V2004 1400 1.11 0.91 0.00 11.74 E2004 1400 1.26 2.43 -0.71 50.27 V2005 1400 1.10 0.91 -0.03 13.49 E2005 1400 1.25 2.41 -1.08 44.33 V2006 1400 1.08 0.88 0.00 13.91 E2006 1400 1.25 2.23 -4.83 40.49 V2007 1400 1.07 0.87 0.00 15.23 E2007 1400 1.25 2.21 -0.63 39.24 V2008 1400 1.05 0.84 0.00 15.49 E2008 1400 1.24 2.18 -0.86 40.61 V2009 1400 1.05 0.83 0.00 12.47 E2009 1400 1.25 2.10 -0.25 35.76 V2010 1400 1.04 0.82 0.00 11.16 E2010 1400 1.28 2.14 -0.22 33.69 V2011 1400 1.03 0.82 0.00 10.93 E2011 1400 1.27 2.17 -0.20 34.94 V_All 23800 1.09 0.88 -0.09 15.49 E_All 23800 1.31 2.60 -4.83 75.96 * SD: standard deviation
Considering the descriptive statistics we can observe several differences. First of all, the means for the Balassa index are somewhat higher for all years. Secondly, the standard deviation is much higher for the Balassa index compared to the RCA based on value added. Thirdly, we see that the minimum values for the RCA based on value added is mostly zero, with only a two negative numbers. Whereas, the Balassa index has negative values in all years. The value added has been calculated in the WIOD by taking the compensation for labor and capital services together, which indicates the value added by the use of domestic labor and capital services to the intermediate inputs. (Timmer, 2012). This leads to a total of two negative values in total, for which no explanation is given. The negative numbers from the Balassa index presented in table 2, are the result of negative export values in the WIOD. The negative export values exist in the WIOD because “the exports to the rest-of-the-world
15
make up for 1.5 percent of the total values for the Balassa index. We have chosen to not omit the negative values as they still carry information. Furthermore, have not replaced them with a value of zero because they do not influence the outcome significantly. Fourthly, we see that the Balassa index has higher maximum values compared to the RCA based on value added.
5. Results
In this section, we discuss the results for the tests and plots that we have introduced in the methodology section. Following the methodology, we first plotted the cumulative distribution function in order to get an overview of how both cumulative distributions are positioned relatively to each other (figure 1).
Figure 1: Cumulative distribution plot of the RCA based on value added and the Balassa index.
16
value added (around 15). One cannot draw major conclusions from this graph but it does give an overview of how the two variables are distributed compared to each other.
Subsequently, we tested for equal variances between the two types of RCA. The results are shown in table 3. As can be seen in the table, the two groups do not have equal variances for all years separately, neither for all years pooled together.
Table 3: F-test for the RCA based on value added and the Balassa index.
Year F-statistic p-value Year F-statistic p-value
1995 0.0856 0.0000 2004 0.1396 0.0000 1996 0.0842 0.0000 2005 0.1421 0.0000 1997 0.0838 0.0000 2006 0.1533 0.0000 1998 0.0852 0.0000 2007 0.1535 0.0000 1999 0.0875 0.0000 2008 0.1496 0.0000 2000 0.0998 0.0000 2009 0.1569 0.0000 2001 0.1154 0.0000 2010 0.1477 0.0000 2002 0.1169 0.0000 2011 0.1425 0.0000 2003 0.1131 0.0000 All 0.1139 0.0000
In order to determine whether or not to rank the observations before performing the unequal variance t-test, we have to test for possible non-normality. This has been done by using the Shapiro-Wilk test. Normality was rejected for all variables with a p-value of 0.0000. Following these results, we will rank all observations before performing the unequal variance t-test. The results of the unequal variance t-test are shown in table 4.
Table 4: Unequal variance t-test for RCA based on value added and the Balassa index.
Year t-statistic p-value Year t-statistic p-value
17
From table 4, we find that the means for all years separately are statistically different from each other at any level greater than 3.34 percent. Furthermore, all years pooled together leads to an absolute significant difference with a 0.0 percent probability that both means are similar. It can thus be concluded that the two variables have significantly different means.
After having identified that the two types of RCA have different variances and means for all years separately as well as all years pooled together, we will now turn to the two basic graphical methods for comparing distribution functions.
Figure 2: Q-Q plot for RCA based on value added and the Balassa index.
Firstly, we will show the Q-Q plot for all years pooled together (figure 2). Clearly, the Q-Q plot (blue) is not even close to being similar to the reference line (red). Only for a very small part at a Balassa Index of around 0 to 3 it matches the reference line more or less. The rest of the values are clearly off. As mentioned before, outliers could have led to divert the attention away from the majority of observations.
18
The two functions intersect at a value of around 0.7, after which the RCA based on value added distribution function is slightly above the Balassa index.
Figure 3: P-P plot for RCA based on value added and the Balassa index.
Besides showing the P-P plot, figure 3 also shows the final part of the analysis of differences between the two variables, the HM. From all P-P plots and the table with the rule-of-thumb for calculating the critical values, the results and hypotheses tests can be derived (table 5).
As can be seen in the table, the HM values are larger for all confidence levels for all years separately. As the HM indexes are larger than the confidence levels, it is found that the RCA based on value added and the Balassa index are significantly different for all years separately at a 99 percent confidence level. For all years pooled together, the significant difference exists at a confidence level of 97.5 percent and is very close to being significant at a 99 percent confidence level (HM: 0.0137 to critical value: 0.0138).
19
Table 5: Results of the hypothesis test for RCA based on value added and the Balassa index. Percentile Year N HM 90 95 97.5 99 1995 1400 0.2203 0.0377 0.0440 0.0498 0.0569 1996 1400 0.2130 0.0377 0.0440 0.0498 0.0569 1997 1400 0.2137 0.0377 0.0440 0.0498 0.0569 1998 1400 0.2156 0.0377 0.0440 0.0498 0.0569 1999 1400 0.2163 0.0377 0.0440 0.0498 0.0569 2000 1400 0.2061 0.0377 0.0440 0.0498 0.0569 2001 1400 0.2091 0.0377 0.0440 0.0498 0.0569 2002 1400 0.2077 0.0377 0.0440 0.0498 0.0569 2003 1400 0.2119 0.0377 0.0440 0.0498 0.0569 2004 1400 0.2080 0.0377 0.0440 0.0498 0.0569 2005 1400 0.2015 0.0377 0.0440 0.0498 0.0569 2006 1400 0.1963 0.0377 0.0440 0.0498 0.0569 2007 1400 0.1977 0.0377 0.0440 0.0498 0.0569 2008 1400 0.1871 0.0377 0.0440 0.0498 0.0569 2009 1400 0.1824 0.0377 0.0440 0.0498 0.0569 2010 1400 0.1745 0.0377 0.0440 0.0498 0.0569 2011 1400 0.1745 0.0377 0.0440 0.0498 0.0569 All 23800 0.0137 0.0091 0.0107 0.0121 0.0138
6. Conclusion
20
This result could have major implications in many fields. For example, we should question whether previous research – which made use of the Balassa index (1965) when assessing comparative advantage levels of nations in certain industries after the start of the second unbundling – still remains valid. In the discussion section we discuss the major differences between the two measurements by analyzing Cyprus and the Netherlands for the year 2011. The examples show what great consequences it has, when choosing the Balassa index over the RCA based on value added after the start of fragmentation. In addition, these results could have major implications politically as well. As the RCA based on value added differs substantially from the Balassa index (Balassa, 1965), different industries are considered having a comparative advantage. When developing (national) policies or strategies, governments and companies should not use the Balassa index anymore, as this will result in wrong decision making.
7. Discussion
With these results, it is clear that we can no longer consider the Balassa index (Balassa, 1965) to be accurate in today’s fragmenting world. The RCA based on value added represents reality in a much better way and is significantly different from the Balassa index. It is therefore not advisable to base predictions on the previously commonly used Balassa index. Furthermore, the fact that the differences are already highly significant in 1995 leads one to believe that it has not been correct to measure the comparative advantage of nations by looking at the gross exports for some time. This means that research which have used the gross exports as a measure of comparative advantage for assessing a period after the start of the second unbundling should be questioned.
21
squaring the differences in log for all values. Subsequently, all differences for all sectors were summed up for each country, which leads to a total value of differences for each country. We chose to analyze the country with the biggest differences as this would be an interesting case to analyze. For both countries, we have taken the top 10 sectors with the most comparative advantages for both the Balassa index as well as the RCA based on value added. Subsequently, we have compared the ranks of these sectors between the two measures. We will start with discussing the Balassa index of Cyprus (table 6).
Table 6: Balassa index for Cyprus in 2011
Top 10: Balassa index for Cyprus in 2011
Rank BI ISIC Industry BI Rank RCA Δ
1 63 Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
16.66 2
-1
2 50 Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
16.62 10
-8
3 52 Retail Trade, Except of Motor Vehicles and Motor- cycles; Repair of Household Goods
13.77 14
-11
4 N Health and Social Work 7.87 20 -16
5 F Construction 4.66 7 -2
6 61 Water Transport 4.55 5 1
7 AtB Agriculture, Hunting, Forestry and Fishing 2.81 22 -15
8 62 Air Transport 2.78 9 -1
9 64 Post and Telecommunications 2.72 16 -7
10 51 Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles
2.34 18
-8
Table 6 shows that there are major differences between the Balassa index and the RCA based on value added. The biggest difference in rank is in the ‘Health and Social Work’ industry, which is ranked number 4 for the Balassa index whereas it is ranked number 20 for the RCA based on value added.
22
in the ‘Hotel and Restaurants’ and the ‘Real Estate Activities’. All of which were not deemed as having a comparative advantage in the country of Cyprus when we would focus solely on the gross exports. These three industries clearly suffer from the final domestic demand error that has been explained in the introduction. The fact that these industries have low exports does not mean that they do not have a comparative advantage.
Table 7: RCA based on value added for Cyprus in 2011
Top 10: RCA based on value added for Cyprus in 2011
Rank RCA ISIC Industry RCA Rank BI Δ
1 P Private Households with Employed Persons 5.84 34 -33
2 63 Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
2.71 1
1
3 H Hotels and Restaurants 2.27 31 -28
4 M Education 1.84 11 -7
5 61 Water Transport 1.54 6 -1
6 70 Real Estate Activities 1.49 33 -27
7 F Construction 1.48 5 2
8 20 Wood and Products of Wood and Cork 1.47 15 -7
9 62 Air Transport 1.46 8 1
10 50 Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
1.42 2
8
23
Table 8: Balassa index for the Netherlands in 2011
Top 10: Balassa index for the Netherlands in 2011
Rank BI ISIC Industry BI Rank RCA Δ
1 F Construction 2.37 14 -13
2 23 Coke, Refined Petroleum and Nuclear Fuel 2.35 34 -32
3 15t16 Food, Beverages and Tobacco 2.27 31 -28
4 L Public Admin and Defence; Compulsory Social Security 2.11 15 -11
5 M Education 2.06 5 0
6 63 Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
1.97 7
-1
7 24 Chemicals and Chemical Products 1.89 6 1
8 50 Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
1.85 4
4
9 AtB Agriculture, Hunting, Forestry and Fishing 1.75 28 -19
10 71t74 Renting of M&Eq and Other Business Activities 1.70 9 1
When looking at table 9, representing the RCA based on value added for the Netherlands in 2011, we again see major differences. As was the case with Cyprus, we see that the ‘Private Households with Employed Persons’ has the biggest comparative advantage for the Netherlands. Other industries that are ranked much higher compared to the Balassa index, are the industries: ‘Manufacturing, Nec; Recycling’, ‘Health and Social Work’ and ‘Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles’. Again, this could have major implications for governmental policies and companies’ strategies. For instance, companies could choose to locate certain activities of their value chain in a specific country because of its comparative advantage. When using the Balassa index, certain industries in a country would show a comparative advantage, whereas they actually have a comparative disadvantage when looking at the value added. This means that they would not locate their operations in the best location, because of the two errors – Final
domestic demand and foreign intermediate inputs - that the Balassa index has. Based on all
24
Table 9: RCA based on value added for Cyprus in 2011
Top 10: RCA based on value added for the Netherlands in 2011
Rank RCA ISIC Industry RCA Rank BI Δ
1 P Private Households with Employed Persons 2.31 35 -34
2 36t37 Manufacturing, Nec; Recycling 1.89 25 -23
3 N Health and Social Work 1.74 31 -28
4 50 Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
1.44 8
-4
5 M Education 1.42 5 0
6 24 Chemicals and Chemical Products 1.39 7 -1
7 63 Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
1.38 6
1
8 51 Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles
1.33 24
-16
9 71t74 Renting of M&Eq and Other Business Activities 1.30 10 -1
25
8. References
Algieri, B., Aquino, A., & Succurro, M. (2011). Going “green”: Trade specialisation dynamics in the solar photovoltaic sector. Energy Policy, 39(11), 7275-7283. Balassa, B. (1965). Trade liberalisation and “revealed” comparative advantage1. The
Manchester School, 33(2), 99-123.
Balassa, B. A. (1989). Comparative advantage, trade policy and economic development Harvester Wheatsheaf Hemel Hemstead.
Baldwin, R. (2006). Globalisation: The great unbundling (s). Economic Council of Finland,
20(2006), 5-47.
Baldwin, R., & Lopez-Gonzalez, J. (2013). Supply-Chain Trade: A Portrait of Global
Patterns and several Testable Hypotheses,
Bems, R., Johnson, R. C., & Yi, K. (2011). Vertical linkages and the collapse of global trade.
The American Economic Review, 101(3), 308-317.
Blinder, A. S. (2006). Offshoring: The next industrial revolution? Foreign Affairs, , 113-128. Brakman, S., Inklaar, R., & Van Marrewijk, C. (2013). Structural change in OECD
comparative advantage. The Journal of International Trade & Economic Development,
22(6), 817-838.
Coombs, W. T., Algina, J., & Oltman, D. O. (1996). Univariate and multivariate omnibus hypothesis tests selected to control type I error rates when population variances are not necessarily equal. Review of Educational Research, 66(2), 137-179.
Feenstra, R. C., & Hanson, G. H. (1999). The impact of outsourcing and high-technology capital on wages: Estimates for the united states, 1979–1990. The Quarterly Journal of
Economics, 114(3), 907-940.
Fertö, I., & Hubbard, L. J. (2003). Revealed comparative advantage and competitiveness in hungarian agri–food sectors. The World Economy, 26(2), 247-259.
Fertő, I., & Hubbard, L. (2004). The dynamics of agri-food trade patterns-the accession countries’ case. Seventy-Eighth Annual Conference of the Agricultural Economics
Society, Imperial College, London, April, pp. 2-4.
Greenaway, D., & Milner, C. (1993). Trade and industrial policy in developing countries: A
manual of policy analysis University of Michigan Press.
Grossman, G. M., & Rossi-Hansberg, E. (2006). The rise of offshoring: It’s not wine for cloth anymore. The New Economic Geography: Effects and Policy Implications, , 59-102. Hinloopen, J., & Van Marrewijk, C. (2001). On the empirical distribution of the balassa
26
Hinloopen, J., & van Marrewijk, C. (2004). PP Plots and the Harmonic Mass Index: An
Application to Comparative Advantage,
Hinloopen, J., & Van Marrewijk, C. (2005). Comparing Distributions: The Harmonic Mass
Index,
Holmgren, E. B. (1995). The PP plot as a method for comparing treatment effects. Journal of
the American Statistical Association, 90(429), 360-365.
Hummels, D., Ishii, J., & Yi, K. (2001). The nature and growth of vertical specialization in world trade. Journal of International Economics, 54(1), 75-96.
Isard, W. (1951). Interregional and regional input-output analysis: A model of a space-economy. The Review of Economics and Statistics, , 318-328.
Johnson, R. C. (2014). Five facts about value-added exports and implications for
macroeconomics and trade research. The Journal of Economic Perspectives, 28(2), 119-142.
Johnson, R. C., & Noguera, G. (2012). Fragmentation and Trade in Value Added Over Four
Decades,
Johnson, R. C., & Noguera, G. (2012). Accounting for intermediates: Production sharing and trade in value added. Journal of International Economics, 86(2), 224-236.
Koopman, R., Wang, Z., & Wei, S. (2012). Tracing Value-Added and Double Counting in
Gross Exports,
Liesner, H. H. (1958). The european common market and british industry. The Economic
Journal, , 302-316.
Los, B., Timmer, M. P., & Vries, G. J. (2014). How global are global value chains? a new approach to measure international fragmentation. Journal of Regional Science,
Miller, R. E. (1966). Interregional feedback effects in input‐output models: Some preliminary results. Papers in Regional Science, 17(1), 105-125.
Moser, B. K., & Stevens, G. R. (1992). Homogeneity of variance in the two-sample means test. The American Statistician, 46(1), 19-21.
Moser, B. K., Stevens, G. R., & Watts, C. L. (1989). The two-sample t test versus
satterthwaite's approximate F test. Communications in Statistics-Theory and Methods,
18(11), 3963-3975.
Qineti, A., Rajcaniova, M., & Matejkova, E. (2009). The competitiveness and comparative advantage of the slovak and the EU agri-food trade with russia and ukraine. Agric.Econ.–
27
Ruxton, G. D. (2006). The unequal variance test is an underused alternative to student's t-test and the Mann–Whitney U t-test. Behavioral Ecology, 17(4), 688-690.
Schott, P. K. (2001). One Size Fits all? Heckscher-Ohlin Specialization in Global Production, Timmer, M. P., Los, B., Stehrer, R., & Vries, G. J. (2013). Fragmentation, incomes and jobs:
An analysis of european competitiveness. Economic Policy, 28(76), 613-661. Timmer, M., Erumban, A., Francois, J., Genty, A., Gouma, R., Los, B., et al. (2012). The
world input-output database (WIOD): Contents, sources and methods. WIOD
Background Document Available at Www.Wiod.Org,
Trefler, D., & Zhu, S. C. (2010). The structure of factor content predictions. Journal of
International Economics, 82(2), 195-207.
Utkulu, U., & Seymen, D. (2004). Revealed comparative advantage and competitiveness: Evidence for turkey vis-à-vis the EU/15. Nottingham: European Trade Study Group, Vollrath, T. L. (1991). A theoretical evaluation of alternative trade intensity measures of
revealed comparative advantage. Weltwirtschaftliches Archiv, 127(2), 265-280. White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct
test for heteroskedasticity. Econometrica: Journal of the Econometric Society, , 817-838. Wilk, M. B., & Gnanadesikan, R. (1968). Probability plotting methods for the analysis for the
analysis of data. Biometrika, 55(1), 1-17.
Yeats, A. J. (1985). On the appropriate interpretation of the revealed comparative advantage index: Implications of a methodology based on industry sector analysis.
Weltwirtschaftliches Archiv, 121(1), 61-73.
Zimmerman, D. W., & Zumbo, B. D. (1993). Rank transformations and the power of the students t-test for non-normal populations with unequal variances. Canadian Journal of
28
9. Appendices
Appendix 1: List of countries represented in this research (WIOD) European
Union North America
Asia and Pacific
Austria Germany Netherlands Canada China Belgium Greece Poland United States India
Bulgaria Hungary Portugal Japan
Cyprus Ireland Romania Latin America South Korea Czech Republic Italy Slovak Republic Brazil Australia
Denmark Latvia Slovenia Mexico Taiwan
Estonia Lithuania Spain Turkey
Finland Luxembourg Sweden Indonesia
29
Appendix 2: List of industries represented in this research (WIOD) ISIC Industry
AtB Agriculture, Hunting, Forestry and Fishing
C Mining and Quarrying
15t16 Food, Beverages and Tobacco
17t18 Textiles and Textile Products
19 Leather, Leather and Footwear
20 Wood and Products of Wood and Cork
21t22 Pulp, Paper, Paper , Printing and Publishing
23 Coke, Refined Petroleum and Nuclear Fuel
24 Chemicals and Chemical Products
25 Rubber and Plastics
26 Other Non-Metallic Mineral
27t28 Basic Metals and Fabricated Metal
29 Machinery, Nec
30t33 Electrical and Optical Equipment
34t35 Transport Equipment
36t37 Manufacturing, Nec; Recycling
E Electricity, Gas and Water Supply
F Construction
50 Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of Fuel
51 Wholesale Trade and Commission Trade, Except of Motor Vehicles and Motorcycles
52 Retail Trade, Except of Motor Vehicles and Motorcycles; Repair of Household Goods
H Hotels and Restaurants
60 Inland Transport
61 Water Transport
62 Air Transport
63 Other Supporting and Auxiliary Transport Activities; Activities of Travel Agencies
64 Post and Telecommunications
J Financial Intermediation
70 Real Estate Activities
71t74 Renting of M&Eq and Other Business Activities
L Public Admin and Defence; Compulsory Social Security
M Education
N Health and Social Work
O Other Community, Social and Personal Services
