Korean National Income Based on a Chain Index: 1953~2010

Korea’s national income statistics have been compiled by the Bank of Korea since 1953. However, there is a break in the time series. The current time series (1970 onward) is based on the ‘1993 SNA (System of National Accounts)’ suggested by the UN, and the previous time series (1953~1970) was based on the ‘1953 SNA’. The difference between the previous and current time series is 4.8% in 1970 when the two series overlap. The difference is even greater in terms of comparisons across industries. In addition, it has now become even more difficult to connect the current and the previous time series because, in 2009, the Bank of Korea introduced a chain weighted method for calculating the current time series (1970 onward). Under the chain weighted method, the time series underwent substantial modification; for instance, the economic growth rate during 1970~2005 is 0.9%p higher than the rate under the general method. This paper applies chain weighted values and the ‘1993 SNA’ to the previous time series (1953~1970) by utilizing various national account manuals published by the UN and previous Korean input-output tables in order to calculate a long term time series from 1953 to 2010 based on the same criteria as the current time series (1970 onward). In the revised time series, it appears that 1953 GDP at current basic prices is 3.5% higher and the growth rate for the period of 1953~1970 is 1.5%p higher each year than under the previous time series. Under the revised time series the size of the Korean economy as of 2010 is 50-fold bigger than that of 1953. In terms of industries, manufacturing and SOC show significant expansion whereas the extent of that of the service industry is relatively small.


Ⅰ. Introduction
The Bank of Korea has announced Korean national income statistics since 1953. However, there is a break in the time series in 1970. Seeking to reduce the differences in classifications and scope by economic activities between the two time series, Nak-Nyeon Kim (2009) applied basic prices as in the current time series to the former time series (1953~1970), the first time such a technique had been employed using Korean data.
However, the criteria of the current and former time series match only partially and no methods of eliminating the time series disarray caused by the chain weighted method are presented because the calculation is based on the data prior to its application.
The break in the Korean national income statistics in 1970 is caused by several revisions in the national income estimation manual having been undertaken since the first release of national income statistics in 1958. The first national income statistics of Korea were based on the '1953 SNA (System of National Accounts)' 1 recommended by the UN. Since then, the manual has been amended a number of times to accord with '1968 SNA', '1993 SNA' and so on, reflecting improvements in statistical methods, the integration of the Korean economy into the global economy and changes in the government's roles.
As the UN manual has changed several times, previous time series should also have been amended based on the latest manual, the '1993 SNA', but revision of the Korean national income statistics has been undertaken only for the period from 1970 onward, thereby giving rise to a break in the current Korean national income statistics, which are divided into the current time series for the period of 1970 onward based on the '1993 SNA' and the previous time series for the period of 1953~1970.
In the case of 1970 when there is overlap between the two series of statistics, the difference between the current and previous time series is 4.8% when comparison is based on total GDP, and the difference is even bigger in regard to comparison between industries in the two time series. In other words, it has been difficult to examine changes in the Korean economy on the basis of a consistent standard. In addition, the introduction of the chain weighted method by the Bank of Korea for the calculation of the current time series in 2009 made the connection between the 韓國開發硏究 / 2012. Ⅲ current and former time series even more problematic, and caused the statistics to differ from the long-term time series based on the fixed weighted method employed by Kim Nak-Nyeon. The introduction of chain weighted method brought about extensive changes in the time series. For instance, the annual economic growth rate during 1970~2005 is 0.9%p higher on average under the chain weighted method.
This paper sets out to compile a long-term time series for the period of 1953~2010 for the first time. We do so by using coherent criteria and utilizing the UN national account manuals and previous input-output tables of Korea. Efforts are directed particularly toward identifying the characteristics of chain weighted method and suggesting practical methods for its utilization.
This paper is composed as follows. Chapter Ⅱ looks into existing national account data and identifies the characteristic of each component data set. This chapter also tracks data relevant to national income statistics such as the inputoutput tables, because the value added of the input-output tables is the same as national income statistics. Chapter Ⅲ explores ways to connect the former and current time series, and establishes a time series for the national income statistics for the entire 58-year period since 1953. Chapter Ⅳ reviews the connection of the two time series and undertakes an analysis of the differences between the time series before and after the connection. The chapter also attempts an overview of the changes in the Korean economy over the last 58 years using the connected data. Chapter Ⅴ summarizes the research.

Ⅱ. National income statistics in Korea
In 1957, the Bank of Korea was designated as the official compilation agency for national income statistics. In 1958, it announced national income statistics for the period of 1953~1957 based on the System of National Accounts and Supporting Tables (hereinafter referred to as '1953 SNA') recommended by the UN. These were the first officially recognized national income statistics of Korea. As shown in <Table 1>, the statistics were compiled with 1955 as the base year, and thereafter the base year was changed every 5 years.  19641953~1963 " 196519671953~1966 " 197019721953~1971 " 197519781953~1977 " 198019871970~1985 1968SNA 198519891970~1987 " 199019941970~1992 " 199519991970~1997 " (partial introduction of 1993  The '1953 SNA' differs from the '1993 SNA' in terms of its basic system, scope and processing method. Regarding the basic system, the 1953 SNA has individual accounting systems such as those for national income, the input-output table and the money flow table whereas the 1993 SNA consolidates all these accounting systems in order to improve connectivity among statistics. The scope of fixed asset formation is expand to include intangible fixed capital and military supplies transferable to civilian use. And the government's characteristics are changed from being a consumer to being both a consumer and producer. Therefore, the scope of government fixed capital formation, which was limited to government buildings, is expanded to include expenses on machinery and equipment, military supplies transferable to civilian use, and Social Overhead Capital (SOC).
In addition to the differences in the estimation manual, which is the biggest cause of the discontinuity in the time series, the availability of basic statistics, coming out of new production also contribute to this discontinuity. The followings sections set out the characteristics of the former and current time series, and inputoutput tables that would help make up for the discontinuity in the time series. The aggregate of the lower real value matches that of the higher real value The aggregate of the lower real value does not match that of the higher real value

Previous Time Series (Fixed-Base-Year GDP): 1953~1970
The GDP statistics compiled in the previous time series (1953~1970) were revised for the last time with 1975 as the base year, and they comprise the agriculture and fishery, mining, manufacturing and service industries. Among the relevant publications, 'National Income in Korea', issued in 1982, is the final edition.
Major statistics are recorded in the Economic Statistics System (ECOS) of the Bank of Korea. The time series is compiled based on the '1953 SNA', wherein each industry's output is multiplied by the value added ratio, thereby estimating total value added. Constant value added is estimated by either extending the base year value added or converting current value added, rather than employing the currently used double deflation method. 2

Current Time Series (Chain Weighted GDP): 1970~2010
The GDP statistics for 1970 onwards have been announced every five years since 1980, and the final version of the GDP statistics for the current time series (1970~2010) with 2005 as a reference year was compiled in 2009. This time series is calculated, unlike previous statistics, based on the '1993 SNA' by using the chain weighted method. The output amount and the value added of each industry are calculated individually, with value added being broken down into detailed items (compensation of employees, operating surplus, depreciation of fixed capital and net taxes on production). And realization at constant value is by the double deflation method.

Previous Input-output Tables
Input-output tables and national income statistics are inseparably wedded to each other. However, there is a difference between the two. Input-output tables show both intermediate input and value added, whereas national income statistics exclude intermediate input, or the value of inter-industrial transactions. Accordingly, theoretically speaking, the value added or final expenditure amount of the two statistics should be the same.
The first official Korean input-output tables were produced for the year 1960 in 1962. Thereafter, the Bank of Korea itself compiled input-output tables for the benchmark years of 1963, 1966, 1970, 1975, 1980, 1985, 1990, 1995, 2000, 2005 and update versions. These statistics are available at the Bank of Korea's ECOS and provide useful information with respect to how to deal with value added when comparing and analyzing differences between the current and former time series of the national income statistics.

Ⅲ. Method of Connecting the Previous and Current Time Series (1953∼2010)
Generally, GDP data are revised to include additional value added that has emerged since compilation or following a change in the estimation method or scope of national income.
Regarding the method, there is still controversy over whether to update the data at uniform ratios by going back to the initial period, or to revise gradually without changing the initial data values. For instance, revising the data to the initial period will be difficult as new industries emerge after the compilation process. However, if the method of compiling statistics changes, as it did in the US 3 , the initial data values need to be changed too.

Principles of Connection and Classification of Economic Activities
The following principles of connection are established in light of the data availability of the former time series with a 1975 base year, the current time series with a 2005 base year, and the national account manual recommended by international organizations like the UN (1953 and1993 SNA).

1) Establishing Previous Time Series Based on Basic Prices
The value added of each industry in the previous time series GDP statistics (1953~1970) was calculated based on market prices, whereas the current time series GDP statistics (1970 onward) were established based on basic prices in which net taxes on production (taxes -subsidy) is deducted from the market price. Accordingly, the former time series first needs to be drawn up based on basic prices in order to be connected with current time series. There are two methods of doing this. The first is to deduct net taxes on production from the value at market prices.
The second is to estimate basic prices by summing up factor cost (compensation of employees + operating surplus + depreciation of fixed capital + net taxes on production). The inadequacy of basic data for estimating net product tax before 1970 makes the first method unworkable. The second method of estimating basic prices in consideration of factor cost can be adopted if the GDP data for each industry from the 1982 edition of Korean national income are utilized. 4 Therefore, this paper employs the second method in order to convert the criterion from market prices to basic prices. GDP based on factor cost appears smaller than that based on market prices by an amount equivalent to net indirect tax. 5 In the 1982 edition of 'Korean National Income', GDP based on factor cost is defined as the sum of compensation of employees, operating surplus and depreciation of fixed capital. To calculate total GDP, the net indirect tax of the former time series is converted to net product tax through partial adjustment, and then the net product tax is added to the GDP figure based on basic prices from the former time series. GNI is derived by adding net factor receipts from abroad to total GDP at market prices.

2) Bringing Concepts and Estimation Methods into Line
First, the method for estimating the government service of the former time series is changed. The former time series recognizes the government as a consumer, and 196 韓國開發硏究 / 2012. Ⅲ therefore identifies government services as final consumption and considers only factor cost as added value. The current time series recognizes the government as a producer as well as a consumer, and therefore identifies final consumption after estimating the output of government services. 6 In addition, the scope of fixed capital formation in the government sector is expanded. The former time series considers only fixed capital formation in buildings in its calculation, whereas the current time series considers fixed capital formation in buildings as well as in machinery, equipments, roads and military facilities with non-destructive purposes in its calculation. Accordingly, the scope of fixed capital formation in the former time series is expanded to that of the current time series. As a result, the total sum GDP of the revised time series is higher than that of the former time series in which output and value added are estimated at cost.
Second, taxes on imported goods are separated from the retail and wholesale industries of the former time series and added to net taxes on product. Taxes on imported goods, which are included in retail and wholesale industry in the former time series, are hear stripped out from them.
Third, the classification of the house ownership industry is matched. The previous time series considers house ownership as an independent industry, whereas the current time series includes it in the real estate industry. Hence, the former classification standard criterion is changed to correspond to the criterion of the current time series.
Fourth, the communications industry is separated out. In the previous time series, transport, storage and communication are in the same classification. To match the two time series, communication is therefore separated out and classified with business and personal services.

3) Classification by Economic Activities
Industries are classified into 11 categories in <Table 3>.

1) Analysing Differentiating Factors between Former and Current Time Series
This paper looks into factors differentiating the two time series after bringing them into line by converting the market prices of the former time series to base prices of the current time series. The differences mainly comes from ① the expanded scope of fixed capital formation in the manufacturing sector, ② the estimated volume of home ownership, and ③ the inclusion or exclusion of some elements of fixed capital formation in public administration and defense.
The total changed-value of GDP at basic prices amounts to 116.3 billion won in 1970 when the current and previous time series overlap. Out of this, the fixed capital formation in the manufacturing industry is 32.1 billion won, owneroccupied housing in the real estate is 42.9 billion won, the public administration and defense is 34.4 billion won, respectively accounting for 27.6%, 36.9% and 29.6% of the total.
Regarding the construction, its value under the previous time series appears to be about 10 % lower in 1970 than under the current time series, which seems mainly attributable to differences in estimation methods. The Bank of Korea estimated the previous time series by working out construction cost, before 198 韓國開發硏究 / 2012. Ⅲ multiplying by the value-added rate. 7 Currently, the Bank of Korea collects data on the construction costs of construction companies, and value-added rate of each construction project is estimated through a more detailed look at the compensation of employees, operating surpluses, the depreciation of fixed capital and so on.
Regarding the public administration and defense sector, the depreciation of fixed capital accounts for 2.3% in the value added of the 1960 Input-Output table and 3.4% in the value added of the 1966 Input-Output Table. In 2008, the ratio rose to 30.6% as the scope of the depreciation of fixed capital had widened; hence we raised the base value of the previous time series by 25.6%, which represented the size of the gap between the two time series.

a. Production (at current prices)
There are two types of method for connecting up the previous time series, which is based on different criteria, like Korean national income: either fixing or changing the initial values. The former was extended by interpolation without changing 1953 initial value. Kim (2009), who conducted a research study on methods of connecting the time series of Korean national statistics, adopted the method of fixing initial values. The focus of his research was placed on bringing the scope of the statistics into line rather than on matching the existing time series. The problem, however, was that the difference in the pre-and post-1970 values widened considerably.
Meantime, the Bureau of Economic Analysis of the US Department of Commerce changes initial values when the cause of difference lies in definitions or in changes of classification and statistical methods. 8 This paper adopts the method of changing the initial values if the cause of the difference lies in definitions or in changes of classifications and statistical methods, and it uses the method of fixing initial values and extending by interpolation in other cases. Detailed revisions on individual industries are made in light of methods of compiling national income statistics in the past, and SNA of the United Nations.
The method of connecting the current and former time series on the basis of current prices involves four main steps. First, the methods of processing or changing scope of the two time series should be matched in order to connect the current and former time series. The differences between value added at market prices and value added based on factor cost are used in order to separate out taxes on imported goods from the retail and wholesale industries. Second, interpolation is employed to deal with the still remaining difference after the processing methods of the two time series are brought into line. However, increase or decrease in inventories, net factor income, and statistical discrepancy are the same as those in the previous time series. Third, the total sum of value added based on basic prices was made by the sum of industry's GDP, which was the revised time series for individual industries.
Fourth, the conversion of GDP at basic prices to GDP at market prices requires net product taxes. However, in 1970, when the net indirect taxes of the former time series and the net product taxes of current time series overlap, net indirect taxies are 5.2% larger than net product taxes. The reason for this seems to be that indirect taxes include more taxes than product taxes. Accordingly, the former time series is matched with the current time series through the reduction of an amount equivalent to the excluded amount from the former time series.
GNI is calculated by adding net factor income from abroad to newly calculated GDP based on market prices; and the result only relates to current prices.

b. Production (at constant prices)
The chain weighted method should be applied to the current GDP statistics time series of production 9 in order to obtain a new economic growth rate using the revised national income statistics. In order to get the real GDP of the current time series, the previous time series should be recompiled based on 2005 prices because the reference year of current GDP is the year of 2005. This process requires a deflator (current value added/constant value added based on 2005 data). Theoretically, the deflator with the 2005 reference year can be obtained by using 9 The US and Japan estimate and announce their economic growth rates based on expenditure data.
However, Korea has done so based on production activity data since the 1950s. Nathan Associates, Inc., who came to Korea after the Korean War to consult on reconstruction and development, publishing a report titled "An Economic Program for Korean Reconstruction, 1954", noted that "Korean GDP, as a tool to draw up economic development plans, was estimated focusing on production activity data. Since fundamental statistics were almost non-existent in Korea, easily accessible estimation and data processing methods were adopted in consideration of production activity data." The Bank of Korea (2009c), p.44.

200
韓國開發硏究 / 2012. Ⅲ the deflators in the current and former time series. However, this method is valid only when the current value of the corresponding year in the current time series, whose reference year is 2005, matches the current value of the corresponding year in the previous time series, whose base year is 1975. When current value changes due to variation in scope, as in this paper, the deflator referenced on 2005 can be projected backward until 1953 by reflecting the rate of increase of the deflator based on 1975.
Real value added by individual industries can be obtained through deflators calculated by this method. And gross value added, obtained by summating all the real value added by industries, is a constant value resulting from the previously adopted fixed weighted method. However, the national income statistics were compiled based on the chain weighted method in 2009 by the Bank of Korea, and therefore, gross value added should be reevaluated based on the chain index.
One can turn constant GDP at basic price estimated by this process into constant GDP at market price by adding constant net product tax to constant GDP at basic prices. The constant prices of product tax or product subsidy are calculated by multiplying the real output of the year under comparison by the tax rate or subsidy rate of the reference year. Since output amount data before 1970 are lacking, the adjusted data were estimated by utilizing the rate of increase in each industry's value added, real net product taxes, current product taxes and product subsidies in 1970.

c. Expenditure
Expenditure is estimated from the viewpoint of consumers, not producers. Hence, expenditure is based on market prices including net product taxes. However, matrix data with a product amounts and product taxes are not available. Hence, this paper includes in expenditure the changed amount of value added at basic prices, which is estimated from production activities, without consideration of product taxes. As a result, finance, insurance, real estate and business services show a relative large expansion, whereas construction, accommodation and food services all shrink.
These differences between the previous and the current time series in production activity are allocated to expenditure as below. First, housing related services, accommodation and food service are categorized simply as private consumption; public administration and defense as government consumption; and construction as fixed capital formation. Transport and storage services and others are allocated in <Table 4> in proportion to final demand by industry using the domestic transaction

Comparison between Time Series Before and After the Revision
<Table 5> compares the current value added time series before and after the revision. Annual figures show that the total GDP of revised time series rose by 3.5% in 1953, by 2.7% in 1956, and by as much as 4.8% in 1970. This reflects the fact that the rate of increase in each industry varies from year to year. <Table 6> shows figures for revised real GDP. Real GDP (at basic prices) in 1953 with reference year 2005 prices is 17 trillion KRW, and increases by an annual average of 7.6% during 1953~1970 to amount to 58.9 trillion KRW in 1970. Shown by industry, the real GDP of the agriculture, forestry and fishery industry was 6.8 trillion KRW in 1953 and it less than doubled to 12.5 trillion KRW in 1970, whereas the real GDP of the manufacturing industry increased by more than 10 times from 0.4 trillion KRW in 1953 to 5.1 trillion KRW in 1970.
The economic growth rate changes in line with the revision of real GDP. <Table 7> shows the difference in growth rates between the current and former time series. In 1954, the GDP growth rate of the revised time series is 1.2%p higher than that of the former time series. It is 3.0%p higher than the former time series in 1968, and 202 韓國開發硏究 / 2012. Ⅲ <Table 5> Comparison between Time Series Before and After the Revision (at basic prices) increases by an average of 1.5%p during 1954~1970 which is mainly attributable to the revision of current GDP, and to the change of the previously-used fixed weighted index to a chain weighted index. According to index theory, substitution bias 10 arising from the fixed weight method is greatly reduced when the chain weighted method is applied, in which process, in general, the growth rate before the reference year is revised upward whereas that after the reference year is revised downward. With the introduction of the chain weighted method, the annual economic growth rate in Korea during 1971~2005 is seen to increase by an average of 0.9%p.
The growth rate of the former time series increases after the revision because the real volume growth rate is over-estimated in the fixed weighted method in which 10 Refers to a phenomenon in which a gap appears between real index and fixed weighted index which uses a fixed commodity basket or weighted value because the fixed weighted index fails to reflect changes in relative prices and quantity system that appears as time goes by. Note: 1) Finance, insurance, real estate and business services.
the high weighted value of the reference year and the high volume growth rate of the comparison year are applied. For instance, the share of manufacturing products in industry was very small in the 1950s, but since then it has increased significantly, whereas that of agriculture, fishery and forestry products has moved along a path in quite the opposite direction.

Characteristics of the Korean Economy Seen from the Perspective of the Long Term Time Series
[ Figure 3] shows economic growth trends during the 58 years after the Korean war until 2010 using the revised time series. The growth rate was low right after the 204 韓國開發硏究 / 2012. Ⅲ <Table 7> Differences in Economic Growth Rates between the Current and Former Time Series war, but was very high in the 1960s backed by the active economic development policy of the government utilizing foreign loans. In the early 1980s, the growth rate turned negative due to the second oil shock that erupted in the late 1970s, political instability and so on. However, the rate turned positive in the mid-1980s thanks to the three lows -low interest rates, low exchange rates and low oil prices.
In the 1990s, the growth rate gradually declined, and in 1998, it shifted back into negative territory due to the Asian financial crisis. Although the rate rebounded somewhat thereafter, the overall growth rate has remained around 4~5%.
A similar trend appears in <Table 8>, which shows the economic growth rate of Korea for each period. In the 1950s, in the aftermath of the Korean War, the growth rate stayed at 5.6%; however, it increased continuously to 9.2% in the 1960s, and to 10.3% in the 1970s. And then, the rate gradually fell back to 8.6% in the 1980s, to 6.7% in the 1990s, and to 4.6% in the 2000s, which means that the economy is on the path to maturity.
Changes in the Korean industrial structure estimated based on the long term time series show that the most static industries among Korean industries over the last half a century have been public administration/defense and agriculture/forestry/ fishery. The agriculture/forestry/fishery and public administration/defense industries increased merely 4.2 times and by 4.1 times from 1953 to 2010, respectively; while the manufacturing sector increased by 664.3 times over the same period. In particular, social overhead capital (SOC) in forms such as electricity, gas and water supply increased by a massive 1,349.4 times, thereby satisfying demand from industry as well as the private sector and making a 206 韓國開發硏究 / 2012. Ⅲ <Table 10> Individual Industry Weights in All-industry (at current value) significant contribution to enhancing the quality of life. At the same time, the speed of growth over the past six decades differs from industry to industry, which has given rise to significant changes in the industrial structure of the Korean economy. In 1953, for instance, agriculture/forestry/fishery accounted for 46.9% of all-industry production when based on current value while the manufacturing sector made up just 7.4%.
However, the ratio of the manufacturing sector went up to 18.5% in 1970, and this growth momentum continued with the ratio rising to 24.6% in 198024.6% in , 26.6% in 199024.6% in and 28.3% in 200024.6% in . Nonetheless, it lost dynamism thereafter, retreating to 27.7% in 2009. In the case of social and personal services, the sector grew from 3. 8% in 1953 to 8.4% in 1970, to 12.6% in 1990 and to 18.9% in 2009, indicating the constantly expanding ratio of the service industry as it relates to the quality of life.
In contrast, public administration and defense stood at 7.4% in 1953, and at 9.2% of all-industry in 1960, but it has been more or less constant at around 5~6% since 1960. The mining sector declined continuously from 1.1% in 1953 to 0.2% in 2009.
All these movements mean that the economic structure of Korea has significantly changed over the last half-century.

Ⅴ. Conclusions
There is a break between the current and former time series in 1970. This paper set out to make the connection between the current and former time series on the basis of the 1993 SNA and by utilizing past national income data, input-output tables and so on. This enabled us to look into the long-term economic trends since the Korean War with a more coherent set of criteria. The characteristics of the connected time series are as follows.
First, the time entire series has uniform compilation and industrial classification criteria for the entire 58 years. When the criteria of current time series are applied to the former time series, current GDP at basic prices increases by 3.5% in 1953.
Second, the chain weighted method, which was introduced by the Bank of Korea in 2009, is applied to obtain the real GDP for years before 1970. The subsequent re-estimation of the growth rate gives an average annual growth rate 1.5%p annually higher during the years from 1954~1969.
Third, the economic growth rate trend estimated based on the revised time series for the last 58 years shows that the Korean economy was accelerating until the 1970s and since then its pace has decelerated somewhat.
Fourth, the size of the Korean economy increased more than 50-fold from 1953 to 2010. However, the growth rate by sector shows that the manufacturing and SOC sectors increased greatly while the service sector expanded on a relatively small scale.