Testing the Impact of Value-Added and Global Income Tax Reforms on Korean Tax Incidence in 1976: An Input-Output and Sensitivity Analysis
Author:
Mr. Peter S. Heller
Search for other papers by Mr. Peter S. Heller in
Current site
Google Scholar Close
Publication Date:
01 Jan 1981
eISBN:
9781451946871
Language:
English
Keywords:
SP; SDR; holding; developing country; direct tax; excise tax; personal income; incidence assumption; progressive tax; Tax incidence; Income tax systems; Global; income tax
Download PDF (1.7 MB)

Development planners have become increasingly concerned that economic growth be accompanied by some improvement in the distribution of income. The tax system is recognized as a principal policy instrument through which the narrowing of income differentials can occur. It is therefore of more than academic interest to understand which income groups in fact, bear the burden of taxes. This problem has motivated scholars to analyze the incidence of individual taxes and of tax systems as a whole. 1

Abstract

Development planners have become increasingly concerned that economic growth be accompanied by some improvement in the distribution of income. The tax system is recognized as a principal policy instrument through which the narrowing of income differentials can occur. It is therefore of more than academic interest to understand which income groups in fact, bear the burden of taxes. This problem has motivated scholars to analyze the incidence of individual taxes and of tax systems as a whole. 1

Development planners have become increasingly concerned that economic growth be accompanied by some improvement in the distribution of income. The tax system is recognized as a principal policy instrument through which the narrowing of income differentials can occur. It is therefore of more than academic interest to understand which income groups in fact, bear the burden of taxes. This problem has motivated scholars to analyze the incidence of individual taxes and of tax systems as a whole. 1

This study represents such an effort applied to the tax system of the Republic of Korea for 1976. The study also provides a simulation for 1976 of the distributional impact of the two major tax policy reforms of the last decade. This involved the consolidation, in 1974, of the various schedular taxes on factor incomes through the introduction of a progressive global income tax and the radical simplification, in 1977, of the indirect tax system through the enactment of a value-added tax (VAT) and a special excise tax on consumption. Finally, this study tests the sensitivity of the results to the complexity and detail of its underlying assumptions. Roy W. Bahl has recently completed a study on the change in the incidence of Korean taxes in the early 1970s. 2 The present study relies on a more detailed data base and a more disaggregated set of assumptions. By applying the Bahl assumptions to the 1976 data base, one can evaluate the sensitivity of the results to these alternative assumptions. Significant differences would suggest the need for significant caution in the use of incidence analyses in countries with limited data.

Section I provides a brief description of the assumptions used in estimating the incidence of each tax. Where economic theory does not offer a firm basis for determining the incidence of a tax, a sensitivity analysis has been made of the incidence under alternative scenarios. Section II examines the incidence results for the specific categories of taxes. Section III analyzes the overall burden of the Korean tax system, compares the alternative tax regimes, and evaluates the redistributional impact of the tax system. Section IV compares the results of this study with those of Bahl (1977). Section V provides a brief summary and conclusion. This section concludes with a brief summary of the structure of the Korean tax system.

Korea’s tax structure is heavily reliant on domestic indirect taxes and customs duties. In 1976, these accounted for 47 per cent and 15 per cent of total tax revenues, respectively. The two principal indirect taxes are the VAT and the special consumption tax, both of which were introduced in July 1977. These replaced 8 indirect taxes (including a cascaded, multiple rate turnover tax on business sales; an excise tax on 73 luxury commodities; and separate ad valorem taxes on petroleum products, electricity and natural gas, transport services, tickets to various recreational events, restaurant and hotel services, and textile products). In addition, there is an ad valorem tax on telephone service and on alcoholic beverages, and an implicit indirect tax on tobacco products that is levied through tobacco pricing policy.

Income taxes account for 27 per cent of total revenues and include a tax on personal income—the global income tax—and a corporation profits tax. The global income tax was introduced in early 1975, replacing a system of schedular taxes on incomes from wages and salaries, real estate, capital gains, interest and dividends, and unincorporated businesses. The present tax law does retain a separate schedular tax on capital gains income, retirement income, and most forms of interest and dividend income. Finally, property taxes are evenly divided between the central and local governments. At the former level are the gift and inheritance taxes, the assets revaluation tax, and a stamp tax on real estate transfers. Property taxes at the local governmental level include a tax on the ownership of agricultural land (the farmland tax), motor vehicles, and real estate, as well as two taxes levied at the time of the transferral of real property—the registration tax and the acquisition tax. The remaining local governmental taxes principally affect urban households. The property and city planning taxes are real estate taxes, the former having a progressive rate schedule and the latter a flat tax rate.

I. Methodological Issues in Analysis of Tax Incidence

Tax Incidence Assumptions

The incidence of a tax is measured by its impact on a family’s absolute income level and its relative income position. The implicit comparison is with the incidence that would have arisen if the same amount of revenues had been raised by a proportional income tax—that is, a tax with no impact on the relative income distribution. The study neither measures the excess burdens (e.g., welfare or “deadweight” losses) that may arise from the Korean tax system, nor does it evaluate the distribution of the benefits that taxes finance through public expenditure.

A tax may affect a family’s real income in three ways. First, it may introduce a wedge between the price of a good without the tax and the tax-inclusive price. Taxes may be passed on by the producer, wholesaler, or retailer to the consumer through an increase in the price of the product, the result being a net reduction in the purchasing power of a family’s disposable income. Second, taxes may reduce the factor incomes of a household on its labor or productive assets, thus introducing a wedge between the before- and after-tax incomes from a factor of production. Third, a tax may increase the cost of possession or acquisition of a real or financial asset. This may be borne by the owner of the asset or passed on to the consumer of the services produced by the asset (such as a property tax passed on in higher rent). The important problem is to determine how a particular tax operates to reduce real income. Taxes formally levied on one group may be shifted and ultimately may be borne by another. Table 1 summarizes the assumptions used to impute the incidence of each tax in Korea. 3 This section examines some of the important methodological assumptions used in the analysis.

Table 1.

Korea: Assumption and Allocation Bases for the Shifting of Taxes

article image
article image
article image
article image
article image
article image

For a precise description of the allocation procedure, see Heller (1981), pp. 5-31, with particular reference to Table 3.

This includes grain threshing, rice and barley polishing, charcoal briquettes, noodle making, fertilizer, feeds, silk reeling, cotton ginning, fats and oils, spinning, textiles, paper, pulp, staple fiber, vessels, rolling stock, tramcars, and trolley cars (Korea, Ministry of Finance, Tax System Bureau, Korean Taxation, 1977 (Seoul, 1977), p. 206).

In terms of input-output sectors, this is defined to include the following: wood products and furniture, paper and paper products, inorganic and organic chemicals, drugs and cosmetics, chemical fertilizers, petroleum refining and related products, coal products, rubber products, nonmetallic mineral products, iron and steel, primary iron and steel products, nonferrous metal ingot and primary products, fabricated metal products, nonelectrical machinery, transportation equipment, measuring equipment, medical and optical instruments, and miscellaneous manufacturing and fishery products.

Tax base excludes share of foreign corporation tax payments. Principal sources of data

Bank of Korea, Monthly Economic Statistics, Vol. 31 (No. 12, 1977), pp. 126-70.

Choo, Hakchung, “Probable Size Distribution of Income in Korea: Over Time and By Sectors” (unpublished, Korea Development Institute, 1977). Korean Government Publications

  • Economic Planning Board, Summary of Budget for Fiscal Year 1977 (Seoul, 1977).

  • Economic Planning Board, National Bureau of Statistics, Annual Report on the Family Income and Expenditure Survey, 1976 and 7977 (Seoul, 1977 and 1978).

  • Economic Planning Board, Korea Statistical Yearbook, 1977 (Seoul, 1977).

  • Ministry of Agriculture and Fisheries, Report on the Results of Farm Household Economy Survey, 1977 (Seoul, 1977).

  • Ministry of Finance, Tax System Bureau, Korean Taxation, 1976 and 1977 (Seoul, 1976 and 1977).

  • Ministry of Finance, “Introduction of Value-Added Tax into the Republic of Korea” (unpublished, November 1977).

Office of National Tax Administration, Statistical Yearbook of National Tax, 1977 (Seoul, 1977).

Indirect taxes are assumed to be shifted forward in the price of goods on which they are levied. A household’s tax burden will be in direct proportion to its consumption of the taxed goods. Therefore, the total revenues from an indirect tax on a good are apportioned across income deciles in direct proportion to the share of each decile in the total consumption of that good. The tax burden per household is obtained by dividing the tax revenues imputed to a decile by the number of households in the decile.

It is important to take account of the differentiated rate or exemption structure of any tax that applies to many commodities and to obtain accurate estimates on the consumption across income groups of the particular goods and services subject to tax. For example, some taxes, such as the liquor tax, have higher tax rates on the types of products consumed by upper-income groups; the revenues generated from the tax on high quality liquors may account for the bulk of liquor tax revenues. Without data on each income group’s consumption of each kind of liquor, one could easily overestimate the liquor tax burden of the lower-income groups.

Indirect taxes may be levied on an intermediate input. It is assumed that the producers of such inputs can pass these taxes on in their prices, such that the tax is ultimately borne by the consumer of the final good embodying the intermediate input. 4 To impute the taxes or tariffs on intermediate inputs to the consumers of final goods and services, the inverse matrix of the 1975 Korean input-output table is used to determine the fraction of an intermediate input industry’s sales (and thus of its advalorem tax) flowing to each of the final demand sectors. The total level of an indirect tax on intermediate inputs that is assignable to a final goods sector is derived by summing across the intermediate goods sectors. 5 Addition of the direct taxes on the final demand sector to the imputed intermediate input tax permits a calculation of the total level of indirect taxes assignable to each final goods sector. The share of indirect taxes assignable to each of the final demand sectors can then be determined, and their allocation across income deciles is made as with any other final goods tax. 6 Consumption statistics by income group are collected for each of the final demand sectors.

It is assumed that households bear the burden of indirect taxes only in relation to cash outlays. The principal exceptions are some agricultural commodities, such as cereals, vegetables and fruit, meat and fish, and processed foods. In Korea, a farm household may consume a large part of its production. The farmer is not likely to be able to pass on the taxes embodied in the price of inputs in the price of marketed output.

Concerning income taxes, the ultimate incidence of the corporation tax remains unresolved in the literature on taxation. 7 The four principal hypotheses are that the tax is borne: (i) by the recipients of property income (Assumption 1); (ii) by the recipients of corporate property income (Assumption 2); (iii) by the consumer, as the tax is shifted forward in prices; and (iv) by labor, with the tax shifted backward in the form of lower wages and salaries.

Blending hypotheses (ii), (iii), and (iv), Assumption 3 assumes that the corporation tax burden is in part borne by corporate capital, in part shifted forward to consumers, and in part shifted backward to workers in reduced salaries (see Table 1). The Korean corporation is assumed to have sufficient market power and governmental support to shift a portion of its corporate tax liability away from corporate capital. Trade constrains the corporation from shifting all of it in the form of higher prices. Thus, the corporation shifts part of the cost back on its labor force in the form of lower wages. This assumes that the labor force is effectively “tied” to the corporation and lacks extensive mobility, an assumption not unreasonable in 1976.

In estimating the incidence of the two alternative personal income taxes—the pre-1974 schedular income tax and the post-1974 global income tax—estimates are made of the putative tax liability of the average household in a decile, taking account of the composition of that income as between the different income sources (e.g., wages and salaries, unincorporated business income, interest and dividends, rental income, capital gains). Demographic characteristics and family labor force participation are also considered in terms of the relevance of personal exemptions, deductions, and spouse income treatment. Tax rates are applied according to the relevant schedules for particular types of income. Based on these factors, an average tax liability is calculated per decile household; summing across all deciles, and multiplying by the number of households, one may estimate the total implied level of tax revenues. To the extent that these totals overestimate the actual revenues, all presumptive tax liabilities are scaled down proportionately. 8

With respect to property taxes on real estate, two alternative incidence assumptions are made: (1) that the tax is effectively borne by property income (defined as rental income, interest, dividends, and imputed rental income from owner-occupied housing) (Assumption 1); and (2) that the component of the tax on land is borne by the landowner (given the inelasticity of supply of land) and the remainder by the consumer of housing services (in the form of imputed rent and rents paid), (Assumption 2). 9

Estimation of Distribution of Incomes and Expenditure in Korea

A fundamental requirement for the study of the distribution of the tax burden in any country is the development of comprehensive statistics on the distribution of income and expenditure. Among the issues that must be addressed are: (i) the categorization of groups within the population and the distribution of income across these groups; (ii) the definition of the measure of total income to be allocated across households; (iii) the distribution of each type of factor income across households; and (iv) the distribution of consumption of taxed commodities. Although the methodological assumptions and data sources for this study are outlined in an earlier paper 10 it is useful to review some important conceptual issues.

Concerning (i), the population is distributed across ten income deciles, with each decile divided between urban and rural households. Specifically, all households are ranked according to their income; the poorest 10 per cent of households are grouped in the first, or lowest, decile, the next poorest 10 per cent in the second decile, and so on. Within each of these deciles, one may distinguish between farm and nonfarm households, in order to examine any sectoral differences in the impact of the tax system. Since farm and nonfarm households are concentrated at different points of the income spectrum, the share of farm and nonfarm households will vary across deciles. In fact, there are only nonfarm households in the tenth decile.

In estimating the total income to be distributed across households, a comprehensive measure of household income was sought. This measure would include the sum of a household’s factor earnings, net transfer payments received from the government, and net changes in the value of a household’s assets arising from capital gains or losses. An estimate of a household’s Adjusted Family Income (AFI) was obtained following the approach used by Pechman and Okner in their study of the U.S. tax system. 11 The measure of AFI is defined to include an amount equivalent to the indirect or direct taxes that are assumed to be shifted to the consumer in higher prices.

It would be incorrect to compare burdens that include sales and excise taxes with an income concept that does not include these taxes. Since we are comparing tax burdens under several different shifting assumptions with reference to a proportional income tax, a consistent income basis must be used. To achieve this, family income of all household units is increased proportionately by the ratio of indirect business taxes to family income, on the assumption that the use of indirect taxes does not alter the distribution of factor incomes. 12

Similarly, the allocation of corporation tax payments across households depends upon the choice of incidence assumptions. If the corporation tax is, in fact, borne by corporate property owners, their incomes should be supplemented by the amount of tax per household (as if the corporations passed the tax payments initially to each household, which then paid the tax). If part of the tax is borne by labor, a comparable procedure should be applied for wage earners. If part of the tax is borne by consumers, that part of the tax should be treated as an indirect tax and allocated as income in the above manner. 13 Similarly, property taxes are assumed, in the national income accounts, to be indirect taxes. If such taxes are assumed to be borne by property income, then indirect taxes should be reduced and such property taxes should be allocated to households according to their shares of the property tax burden.

Thus, although the total magnitude of adjusted family income is invariant to the incidence assumptions, there are at least two possible distributions of this income, according to the incidence assumption. Two cases may be distinguished: Income Concept A, which assumes that the corporation tax and property taxes are fully borne by domestic corporate property and general property incomes, respectively; and Income Concept B, which assumes that some corporation taxes are shifted to wage earners and consumers and that the local property and city planning taxes are borne according to housing expenditure.

The most uncertain component of the data used in this study involves the estimation of the distribution of the particular factor incomes. The estimates in this study have been pieced together from the expenditure surveys and from detailed data on taxable income by source and income bracket provided by Korea’s Office of National Tax Administration (ONTA). The estimates used are nevertheless only conjectural.

II. Incidence of Particular Taxes in Korean Tax System

Direct Taxes

The global income tax and the corporation tax account for one fourth of total government revenue and are clearly the most progressive elements of the Korean tax system. Under the most progressive assumptions, the direct tax burden rises from 0.5 per cent of income in the lowest-income (first) decile to almost 4 per cent in the ninth decile and 12 per cent in the highest decile. 14 Under more regressive assumptions, 15 the direct tax burden rises from 2.5 per cent of income in the lowest decile to almost 10 per cent in the highest (see Table 2). The burden is almost wholly on the nonfarm sector, with the burden on farm households amounting to less than 2 per cent of their income.

The global income tax is the most important progressive tax, reflecting a highly progressive rate schedule and generous exemptions, particularly for wage earners and farmers. The two highest nonfarm deciles pay more than 85 per cent of all global income taxes. Through its exemption of farm income, the global income tax helps to narrow the after-tax income differential between farm and nonfarm households.

The earlier schedular income tax was also progressive among nonfarm households, reflecting primarily the progressivity of the schedular taxes on dividends, global income, wages, and business income. 16 To compare the incidence of these alternative tax regimes, one must assume that they generate equal amounts of revenue. Under this assumption, the global income tax reform apparently increased the income tax burden of the richest decile of the population, which rose from 6 per cent to almost 7 per cent. For all other deciles (particularly the ninth), the burden decreased. As a share of income, the tax burden declined by a factor of two for the bottom 50 per cent of nonfarm households, remained almost unchanged for the sixth and eighth deciles, and decreased by 25 per cent for the seventh and ninth deciles. The burden among farm households was lowered throughout all deciles.

Table 2.

Korea: Summary of the Effective Burdens of Different Taxes on Total Income, 1976 1

(As a percentage of total income)

article image
article image

Using, as divisor, total income, Concept A (see Section I). See Table 1 for clarification of the various assumptions referred to in this table.

The incidence of the corporation tax is highly sensitive to the incidence assumption chosen. Assumption 2 is the most progressive (see Table 1); according to this assumption, the corporation tax is not shifted. The tax burden rises from almost zero in the poorest half of the population to 0.5 per cent in the ninth decile and 5 per cent in the tenth decile. This reflects the extreme concentration of corporate property income within the highest nonfarm decile. The tax barely touches the farm sector.

Next in progressivity is Assumption 1, according to which the tax is borne by the recipients of nonrental property income. The tax burden rises from approximately 0.7 per cent to 1 per cent of total income in the lower nine nonfarm deciles, being slightly regressive within this range, and then rises to 4.5 per cent in the richest nonfarm decile.

The partial shifting Assumption 3, according to which a quarter of the tax is borne by corporate capital and the remainder borne equally by labor and consumers, leads to a tax that varies irregularly from 2.2 per cent to 2.4 per cent of income for the first through eighth nonfarm deciles, drops to 2.0 per cent for the ninth nonfarm decile, and then rises to 2.3 per cent for the richest nonfarm decile. The burden on farm households is considerably higher under this alternative assumption. The tax is fully regressive, falling from 1.5 per cent of income in the first farm decile to 0.7 per cent in the tenth (richest) decile.

Indirect Taxes

Domestic

The regressivity of indirect taxes on domestic production and consumption is indicated in Table 2. 17 In the present VAT-inclusive tax system, the burden of indirect taxes on nonfarm households varies erratically from 8.0 per cent to 8.7 per cent of total income for households in the first eight deciles and then drops to 7.7 per cent in the ninth decile and to almost 7 per cent in the tenth decile. The burden among farm households is only slightly less in the lowest two deciles (6.0-6.7 per cent of income) but drops to less than half the burden of the nonfarm sector.

These results emerge despite the high taxation of luxury goods under the special excise tax. The regressivity of taxes on consumption and production arises from the decline in the average propensity to consume as income rises. In this context, the rate of decline in the indirect tax burden in the higher-income deciles of Korea suggests far less regressivity than is normally observed. For example, in the U.S. tax system, the ratio of the indirect tax burden of the lowest to the highest deciles is 2.8:1. The ratio in Korea is 1.2:1 among nonfarm households and 2:1 among farm households. This lower regressivity is apparent for all deciles in Korea. 18

Among the constituent taxes, the value-added tax is regressive, with the burden declining from 5.6 per cent of income in the lowest decile to 3.9 per cent in the highest. The burden is lower in the farm sector than the nonfarm sector, with the relative burden declining in the upper farm deciles. The special excise tax burden varies erratically from 1.2 per cent to 1.4 per cent of income in the first nine nonfarm deciles, rising to 1.7 per cent in the tenth decile. It is regressive for farm households, tapering from 0.7 per cent of income in the first decile to 0.4 per cent in the tenth.

The principal remaining excise taxes are sumptuary taxes on liquor and tobacco. Although both levy higher rates on the higher-quality products consumed by middle- and upper-income brackets, the average amount spent on such goods declines as a share of total income as income rises. This ensures a regressive tax burden. Among urban households, the liquor tax burden is erratically proportional in the bottom eight deciles and falls in the highest two deciles. Among farm households, there is a clearer regressivity in the burden as income rises. For the monopoly profits tax on tobacco, the highest burden for nonfarm households is borne by the fourth decile (2.6 per cent of income). The burden tapers off thereafter to 0.9 per cent in the tenth decile. The burden is slightly lower among farm households and is clearly regressive.

The incidence of the pre-VAT indirect tax regime was primarily determined by three of the eight original taxes: the business tax, the commodity tax, and the petroleum products tax. Their incidence, as well as the effect of substituting the simpler VAT-special excise tax regime, is indicated in Table 2. The reform increased the relative burden on the nonfarm sector, with the result that its tax burden rose by up to 0.8 percentage points in each decile, with increases concentrated in the lowest nine deciles. The reduction in the burden of the farm sector was distributed across all deciles, with a somewhat greater decline for households in the richer deciles. There was a small increase in the regressivity of the tax system, with a slight decrease in the tax burden of households in the highest decile.

These results should be interpreted with considerable caution. The analysis is based on a simulation of the effect of substituting a VAT and a special consumption tax of equal yield for the indirect taxes eliminated in July 1977. If the VAT reform had been introduced in 1976 and had led to more (or less) revenues, the burden of the two tax regimes could have been different. Similarly, it is possible that the relative amounts of revenue derived from the replaced indirect taxes would have changed between 1976 and 1978, which would also have changed the overall distributional impact of the reform.

Customs duties

The burden of customs duties is 4.2 per cent of income for the lowest nonfarm decile, thereafter declining to 2.2 per cent for the highest nonfarm decile. The tax is considerably more regressive in the farm sector, with the burden falling from 4.5 per cent in the first decile to 1.8 per cent in the tenth. As with many other taxes, a significant differential in the relative farm and nonfarm tax burdens appears only for the higher-income deciles.

Property Taxes

If property taxes are borne by property income (Assumption 1), they become unambiguously progressive in the nonfarm sector, rising from 1.3 per cent in the first seven deciles to 3.0 per cent in the tenth. For farm households, the burden rises from 0.6 per cent of income in the lowest decile to almost 2 per cent in the highest. If one assumes that the burden of real-estate-related property taxes is shifted to the consumers of housing services (Assumption 2), the total impact of property taxes is regressive through the seventh nonfarm decile and proportional thereafter. As a share of total income, the tax burden falls from 2.7 per cent in the first nonfarm decile to 1.7 per cent in the tenth. However, property taxes are progressive throughout all deciles for farm households.

Relative Distributional Impact of Alternative Taxes

To predict the impact of changes in the tax system, it is useful to evaluate the fraction of households’ tax liabilities attributable to each tax. If one makes the most progressive corporate and property tax incidence assumptions, indirect taxes account for 87 per cent of the tax burden of nonfarm households in the lowest deciles (see Table 3). Three taxes alone account for 81 per cent—customs duties (29 per cent), the VAT (41 per cent), and the special consumption tax (11 per cent). Since these three taxes already reflect general indirect tax laws structured to achieve some progressivity, a further reduction in the tax burden of the lowest-income groups could only arise by lowering the rate of one of these taxes or by developing some form of negative tax credit through the global income tax. Such changes would have only a minor impact on the overall tax burden of the upper-income groups relative to their incomes. For households in the tenth nonfarm decile, approximately 47 per cent of their tax burden arises from the corporate and global income taxes, while the three indirect taxes mentioned previously account for only 33 per cent of their burden. It is also worth noting that under these assumptions, the corporation tax accounts for a significant share of tax liabilities only in the tenth decile.

If one assumes more regressive incidence assumptions regarding property and corporation taxes, 19 there is a change in tax shares reflecting the increase (decrease) in the overall tax burden of the lower-income (higher-income) groups in the non-farm sector. For households in the poorest nonfarm decile, the corporation tax effectively becomes a sales tax on corporate products and/or a tax on their wage earnings. Indirect taxes (exclusive of the corporation tax) again constitute 71 per cent of the tax liabilities of households in the first decile, with a smaller share deriving from the three indirect taxes mentioned previously. As income rises, the share of the corporation tax falls to 10 per cent of households’ tax liabilities in the highest decile. The global income tax assumes a more important role for the richest 50 per cent of nonfarm households, ultimately accounting for 31 per cent of the tax liability of the households in the richest (tenth) decile.

Table 3.

Korea: Fraction of a Household’s Tax Burden Attributable to Each Tax Under Alternative Incidence Assumptions, 1976 1

(In per cent)

article image
article image

Components may not add to totals shown because of rounding. See Table 1 for clarification of the various assumptions referred to in this table.

Two other points should be mentioned. First, the two sumptuary taxes are principally important for the middle-income deciles. The tax on cigarettes accounts for 14-16 per cent of the tax liability of the fourth nonfarm decile, depending on the choice of incidence assumptions. The liquor tax is particularly burdensome in the third and fourth deciles. Second, although the gift and inheritance taxes are imposed for redistributive reasons, they account for only a tiny share (less than 2 per cent) of the total tax liability of households in the highest-income deciles. This reflects their highly selective incidence within these groups.

The tax burden of the farm sector principally derives from indirect and property taxes. More than 70 per cent of any farm decile’s tax burden derives from indirect taxes. Regardless of incidence assumptions, customs duties and the VAT account for 67-77 per cent of the burden of the lowest-income households and 45-48 per cent of the burden in the highest-income households. The special consumption tax accounts for 7 per cent of the farm sector’s tax burden and the liquor tax for another 6 per cent. With the exception of the lowest decile, the monopoly profits tax absorbs 13-18 per cent of the tax burden in each farm decile. As with the nonfarm sector, the corporate tax may act as a corporate sales tax. It accounts for 7-11 per cent of the farm sector’s total tax burden. Thus, it would be difficult to reduce selectively any particular tax in order to lower significantly the tax burden of the lower farm deciles. Any such policy change would lower the burden across income classes and only lightly reduce the relative burden of the lower deciles. In absolute terms, the bulk of the tax reduction would flow to the higher-income farm households.

Most of the property taxes affecting the farm sector are proportional in their impact. The stamp, registration, acquisition, and property taxes together are responsible for less than 5 per cent of total farm tax liabilities. However, the farmland tax accounts for a significant share of the tax liabilities of the higher deciles, rising from 1 per cent in the second decile, to 5 per cent in the fifth decile, and to 16 per cent in the highest decile. Short of eliminating the agricultural income exemption in the global tax, the farmland tax is the only tax that could add substantially to the progressivity of the tax system in the farm sector.

III. Quantitative Estimates of Tax Incidence

One must be cautious in interpreting the results on the overall incidence of the tax system. The substitution of one tax for another might be considered a marginal change, in that it would not seriously affect the allocation of resources or the distribution of factor incomes. However, a complete change in the tax system would affect the underlying distribution of income, rendering the current incidence results invalid. Therefore, the overall incidence estimates are only an aggregation of the results obtained through partial analysis. They rely on the sensitive assumption that the pretax income distribution would not change if taxes were not levied. 20

Overall Incidence of Korean Tax System

The incidence of the VAT-global income tax regime appears progressive for nonfarm households and regressive for farm households. The relative tax burden on farm households vis-à-vis their urban counterparts is 20-35 per cent lower in the lowest deciles, with the differential widening to more than 130 per cent in the highest farm decile (see Table 4). Specifically, if one makes the most progressive assumptions concerning the property and corporation taxes, the tax burden is 14.5 per cent of income for the lowest nonfarm decile, rising to 19.4 per cent in the ninth decile and 25 per cent in the tenth decile. Among farm households, the burden declines from 12 per cent of income in the lowest decile to 8 per cent in the highest.

The lower burden among farm households reflects (a) their lesser dependence on the market economy and (b) their basic exemption from the impact of most progressive income and property taxes; the latter explains the regressivity of those taxes that do, in fact, affect the farm sector.

If one accepts the alternative assumptions that property taxes are shifted in higher prices for housing services and that corporation taxes are shifted to both consumers and wage earners, the progressivity of the present tax system is significantly lessened; the burden of the lower-income (higher-income) deciles significantly increases (decreases). The burden increases by 2-4 per cent of income for the lowest eight nonfarm deciles, and by 1 per cent of income in the ninth nonfarm decile. In the tenth nonfarm decile, the tax burden falls from 25 per cent to 22 per cent. The burden on farm households increases slightly. The tax system is slightly progressive in the first four nonfarm deciles, proportional in the next three nonfarm deciles, and marginally progressive thereafter. The regressivity of the system among farm households is worsened.

Table 4.

Korea: Incidence of Taxes Under Alternative Incidence Assumptions and Alternative Tax Regimes, 1976

(As a percentage of total income)

article image
article image

In the tax regime with a value-added tax, this includes the special consumption tax, value-added tax, liquor tax, telephone tax, and the defense surtax on excises.

Incidence Assumption A reflects the most progressive set of assumptions concerning the property tax (Assumption 1) and the corporation tax (Assumption 2) in Table 1.

Incidence Assumption B reflects a reasonably regressive set of assumptions concerning the corporation tax (Assumption 3) and the property tax (Assumption 2) in Table 1.

In the tax regime with no value-added tax, this includes the textile tax, petroleum products tax, gas and electricity tax, admissions tax, telephone tax, commodity tax, liquor tax, travel tax, business tax, entertainment tax, and defense surtax on excises.

Which case is the most representative? The second probably is, simply because it does not represent the extreme incidence assumptions. However, the sensitivity of the results to the corporation tax incidence assumptions suggests that whatever the Korean Government’s distributional objectives, reliance on the corporation tax increases the uncertainty concerning their realization.

What has been the impact of the set of structural reforms to the tax system in recent years? The movement to a global income tax slightly increased the progressivity of the income tax among nonfarm households (see Table 4). The burden on the highest nonfarm decile increased by 0.6 per cent of income and decreased for all other deciles. The movement toward progressivity in the nonfarm sector is the net result of two offsetting factors—the tax system became more progressive, to the extent that households in the upper-income deciles moved into higher overall marginal rate brackets on the combined total of their global incomes; and the system became less progressive, to the extent that such households used their allowable exemptions to reduce their burden on sources of income that were not exempted under the schedular system.

The movement to a VAT system shifted a small part of the burden from the farm sector to the nonfarm sector and led to a very small increase in the regressivity of the overall tax system. The largest increase in the nonfarm burden occurred in the lower-income deciles, averaging 0.7 per cent of income (see Table 4). The burden on the farm sector fell by 0.3-0.5 per cent of income. Together, the two tax reforms led to an increase in the tax burden of the nonfarm sector and a decrease in the farm sector’s burden.

Redistributive Impact of Tax System

Table 5 compares the distribution of income before and after taxes in order to determine the distributional impact of the tax system. The pretax distribution of income has several important characteristics: (a) the nonfarm sector’s share of total income exceeds its share of all households—76 per cent compared with 66 per cent; (b) almost half of all income is received by the richest 20 per cent of households, and 30-31 per cent of income is received by the richest 10 per cent of households;21 and (c) the poorest 30 per cent of households receive only 10 per cent of total income.

Table 5.

Korea: Distribution of Households, Income, and Tax Burden Across Sectors Under Alternative Incidence Assumptions, 1976 1

article image

This assumes a regime that includes a global income tax and a value-added tax. Components may not add to totals shown because of rounding.

The principal redistributive impact of the tax system is to raise the after-tax share of the farm sector. Compared with its pretax share of 24 per cent, its after-tax share rises to 26 per cent. The relative gain increases as one moves up the farm-income scale, suggesting a widening of after-tax inequality within the farm sector. This result is reasonably insensitive to the choices of tax regimes or of the particular incidence assumptions. The latter choice primarily determines the degree of redistribution within the nonfarm sector.

If one assumes that property and corporate taxes are borne by the recipients of property income, the principal losers are the households in the three highest-income deciles. The after-tax share of the poorest nonfarm households is 4.4 per cent higher than their pretax share, while the after-tax share of the income of the highest nonfarm decile is more than 9 per cent lower than its pretax share. Between the second and seventh nonfarm deciles, the after-tax share of income is roughly 101-102 per cent of the pretax share. Thus, the tax system does have an important equalizing effect on income distribution in Korea, both within and across sectors.

If one assumes that property and corporation taxes are shifted to consumers and wage earners, there is a 5 per cent reduction in the after-tax share of the tenth-decile households and a small increase—2.4 per cent—in the share of the poorest 50 per cent of households. The latter reflects a 7.6 per cent increase in the share of the poorest 50 per cent of farm households that is associated with a 1 per cent decrease in the share of the poorest 50 per cent of nonfarm households. The structural reforms to the tax system in recent years have had a negligible effect on the after-tax distribution of income. If there has been any effect at all, it has been a slight decrease in the after-tax share of the highest nonfarm decile and a slight increase in the after-tax share of the farm sector.

Distribution of Tax Burden

There is no question that, in absolute terms, the rich pay a significant fraction of total taxes in Korea. The households in the highest decile of the nonfarm sector pay between 36 per cent and 43 per cent of all taxes, depending on the choice of incidence assumptions; those households in the two highest deciles (farm and nonfarm) pay between 52 per cent and 58 per cent (see Table 5). In the regime with the VAT, under the most progressive incidence assumptions, the poorest 30 per cent of households pay less than 9 per cent of all taxes. The entire farm sector pays less than 14 per cent of all taxes. The share paid by the farm sector has decreased only slightly as a consequence of the two recent tax reforms.

Another method of indicating the degree of redistribution through the tax system is to contrast the share of total taxes borne by an income decile with its share of total income. The share of the highest nonfarm decile in total taxes is between 17 per cent and 39 per cent higher than its share of total income before taxes. In contrast, the tax shares of farm deciles are between 24 per cent and 32 per cent lower than their before-tax income shares.

IV. Sensitivity of Results to Choice of Assumptions

The need for information on the distributional impact of alternative taxes raises significant problems in developing countries. The available data is usually inadequate to support complicated, detailed tax incidence analyses. Does the level of sophistication and disaggregation used in the analytical and statistical assumptions significantly affect the incidence results? In this section, we compare the present results with those of an earlier study by Bahl. 22 Since Bahl’s data and results related to 1974, all of Bahl’s assumptions and procedures with respect to the allocation of tax burdens have been applied to the present 1976 data base. Table 6 presents the ratio of the effective tax burden of particular taxes observed in this study relative to the effective tax burden derived using the assumptions in the Bahl (1977) study.

Table 6.

Korea: Ratio of Effective Tax Burden Obtained in Present Study to Results Derived Using the Assumptions in the Bahl (1977) Study

article image

The principal difference between the two studies is in the degree of detail and disaggregation used in the criteria for allocation of tax burdens across deciles. This is particularly obvious from Table 1. For example, Bahl uses nonfood consumption per household as the basis for allocating the burden of such taxes as customs duties, the telephone tax, the business tax, the commodity tax, the travel tax, the admissions tax, and the entertainment and food tax. This has the obvious drawback that the implied incidence pattern for each of these taxes will be identical, despite the fact that these taxes are significantly different in their tax bases, exemptions, and rate structures. Such differences underlie the detailed assumptions used in the present study. For which taxes does the added detail and complexity matter?

While a conclusive answer is impossible, the use of a general consumption-type base, such as nonfood consumption, is only a reasonable proxy for the allocation of taxes with a general base, such as the business tax or import duties. The incidence of taxes focused on a narrow set of commodities or services, particularly those with a particular distributional bias in their rate structure, cannot be accurately assessed using a general consumption base.

At one extreme, the business tax was probably the most general indirect tax in the pre-VAT Korean tax system, since it applied to most intermediate and final goods and services produced in the economy. The present methodology uses a sophisticated, highly disaggregated input-output procedure. Yet the results in Table 6 do not dramatically differ from Bahl’s. His results are slightly less regressive in the nonfarm sector and suggest a lower overall burden in the farm sector, but the difference in results may not warrant the considerably greater computational effort expended in assessing the incidence of this particular tax.

At the other extreme, one may point to the results on the incidence of the two principal sumptuary taxes—the liquor tax and the implicit monopoly profits tax on tobacco products. Both taxes exhibit a highly complicated progressive tax rate structure (explicit for the former, implicit in the pricing structure for the latter). The rates are clearly highest for the specific beverages or tobacco products consumed by upper-income groups; in fact, the implicit tax rates for some tobacco products consumed by the poorest households may be zero or negative. The error that results from not taking account of differences in household consumption patterns in the rate structure is striking. Bahl, using tobacco consumption as the basis for allocating the monopoly profits tax, obtains a far more regressive tax burden distribution, ranging from 3.9 per cent of income in the poorest nonfarm decile to 0.7 per cent in the richest. The results in Table 2 suggest there is no burden on the poorest nonfarm households, since the products they consume realize no implicit profit. The burden rises to 2.0-2.6 per cent of income in the middle-income deciles and then falls to 0.9 per cent in the highest decile. Bahl’s results also severely overestimate the monopoly tax burden of the poorer deciles in the farm sector. In a similar vein, Bahl’s results suggest a much heavier liquor tax burden, by a factor of 100 per cent, on the nonfarm poor and a much lighter burden on farm household sectors than are suggested by the present study.

The commodity tax in the pre-1977 tax regime also illustrates the effect of making imprecise assumptions. The commodity tax consisted of a set of excises levied primarily on luxury goods, with the highest rates on the most luxurious commodities. By using a general consumption-type proxy as the basis for allocating the burden of this tax, Bahl obtained results that suggested a far more regressive tax pattern than actually existed. Differences in results that arise from using more or less specific assumptions also emerge when one compares the incidence results for the petroleum products tax, the travel tax, and the telephone tax, though the differences are less dramatic. The effect of the Government’s specific intent to structure a tax to achieve particular distributional objectives cannot be picked up by statistical analysis if the data are not adequately disaggregated.

Although both studies use similar approaches for the allocation of the global income tax burden, the present study took much closer account of some of the intricacies of the particular Korean tax law and did not rely totally on the general rate and exemption structure. Bahl failed to take account of the exemption of farm household incomes from paddy cultivation—an exemption that constitutes a significant proportion of the total household income of households in most deciles. Most farm households have negligible taxable income when one considers the level of their personal exemptions and deductions. Only to the extent that they have interest income, which is subject to a low schedular tax rate, are they likely to have some income tax liability. Again, unless one specifically takes account of the structural details of a particular tax when assessing its incidence, the results obtained are likely to be far off the mark.

V. Conclusion

The principal results of the study may be briefly summarized. The present tax system appears progressive for nonfarm households and regressive for farm households, regardless of the incidence assumptions chosen. For the population as a whole, the burden of the tax system relative to income is roughly proportional for the first 90 per cent of households and progressive for the top 10 per cent. The global income tax, the gift and inheritance tax, the assets revaluation tax, and the farmland tax are the principal progressive elements of the tax system; indirect taxes—such as the VAT, the special consumption tax, and customs duties—are the regressive elements of the tax system.

A principal impact of the tax system has been to narrow the differential in after-tax incomes between the farm and nonfarm sectors. The tax system imposes a relatively higher burden on the nonfarm sector. This differential tax burden across sectors widens as income rises, so that the richer farm households appear to be the primary gainers from the present tax structure. The tax system also appears to narrow the degree of inequality in the before-tax income distribution. The rich pay a significant fraction of total taxes, with the richest 10 per cent of Korean households paying between 36 and 43 per cent of all taxes, depending on the choice of incidence assumptions. The principal impact of the VAT and global income tax reforms was to increase the relative burden of the nonfarm sector, particularly among households in the poorer deciles and in the highest decile.

Finally, the results suggest that to ensure the accuracy of the incidence results, it is necessary to take account of the detailed structure of a particular tax, in terms of its rates and exemptions. This may prove difficult where a country has only a limited statistical data base. However, in cases where the structure of a tax has been deliberately designed to achieve particular distributional objectives, the use of a general type of tax allocation criterion may yield incidence results that are highly misleading in their policy implications.

REFERENCES

  • Aaron, Henry J., Who Pays the Property Tax? A New View (Brookings Institution, Washington, 1975).

  • Bahl, Roy W., “The Distributional Effects of the Korean Budget During the Modernization Process” (unpublished, Syracuse University, October 1977).

    • Search Google Scholar
    • Export Citation

  • Bird, Richard M., and Luc H. De Wulf, “Taxation and Income Distribution in Latin America: A Critical Review of Empirical Studies,” Staff Papers, Vol. 20 (November 1973), pp. 639–82.

    • Search Google Scholar
    • Export Citation

  • Break, George F., “The Incidence and Economic Effects of Taxation,” in Alan S. Blinder and others, The Economics of Public Finance (Brookings Institution, Washington, 1974), pp. 119–237.

    • Search Google Scholar
    • Export Citation

  • Choo, Hakchung, “Probable Size Distribution of Income in Korea: Over Time and By Sectors” (unpublished, Korea Development Institute, 1977).

    • Search Google Scholar
    • Export Citation

  • De Wulf, Luc H., “Fiscal Incidence Studies in Developing Countries: Survey and Critique,” Staff Papers, Vol. 22 (March 1975), pp. 61–131.

    • Search Google Scholar
    • Export Citation

  • Heller, Peter S., “The Incidence of Taxation in Korea” (unpublished, International Monetary Fund, February 25, 1981).

  • Musgrave, Richard A., and Peggy B. Musgrave, Public Finance in Theory and Practice (New York, Second ed., 1976).

  • Pechman, Joseph A., and Benjamin A. Okner, Who Bears the Tax Burden? (Brookings Institution, Washington, 1974).

*

Mr. Heller, Assistant Chief of the Government Expenditure Analysis Division of the Fiscal Affairs Department, is a graduate of Trinity College and Harvard University.

3

For a more detailed discussion of these assumptions, see Heller (1981), pp. 5-46. This paper may be obtained on request from the author, whose address is Fiscal Affairs Department, International Monetary Fund, Washington, D.C. 20431.

4

This requires the assumption that domestic producers of final goods are able to compete with foreign producers in domestic markets, despite the higher costs that domestic producers must pay for such inputs. Any of three conditions are necessary for full forward shifting of tariffs or domestic taxes on primary and intermediate inputs: (i) the domestic producer is more competitive than the world producer; (ii) such taxes are less than the transport, freight, and insurance costs associated with importation; and (iii) the tariff structure on final goods has been structured to allow sufficient “water” in the tariff to cover these input-related tariffs. In the absence of these conditions, the tax or tariff will be borne by a reduction in the value added of the domestic producer.

5
For example, for an import duty, assume that Di equals the level of import duties on an intermediate input sector i, and dij equals the fraction of the ith intermediate goods sector’s output (and, assuming proportional tariff rates, the fraction of import duties £>,) flowing to the yth final demand sector. Then
Dj*=Σi=1IdijDi fori=1,.⁢.⁢.,I;j=1,…J

where Dj* represents the import duties attributable to inputs used in the production of the yth sector’s output. Total import duties assignable to the jth sector will equal the sum of Dj* and the duties on imports of j. Note that dij coefficients are derived by multiplying the input-output inverse matrix by the vector of final consumption demand for 1975.

6

The shares of such indirect tax revenues allocated to final goods sectors are indicated in Table 3 of Heller (1981), p. 12.

7

See Musgrave and Musgrave (1976), pp. 288-321.

8

For further details, see Heller (1981), pp. 22-27.

10

See the References and Heller (1981), pp. 32-46.

12

Pechman and Okner (1974), p. 12. It is a very strong assumption indeed to assume that in the absence of indirect taxes, the distribution of factor income would remain unchanged. To the extent that indirect taxes do affect the allocation of resources—between consumption goods industries and between consumption and investment—there may be a change in the pattern of demand for factors of production, with consequent changes in the rates of return to these factors. This would affect the distribution of factor income. It should be noted that this adjustment will not affect the relative distribution of the tax burden across deciles or sectors.

13

See Heller (1981), pp. 36-40.

14

See Assumption 2 under the corporation tax in Table 1.

15

See Assumption 3 under the corporation tax in Table 1.

16

Since the schedular tax on real estate income accounted for no more than 4 per cent of total schedular income tax revenues, the choice of assumption on the shifting of these property taxes is not quantitatively significant.

17

Detailed results on all the individual taxes can be found in Heller (1981), pp. 48-50.

19

Assumption 3 for the corporate tax and Assumption 2 for the property tax (see Table 1).

21

It should be noted that the distribution of income is sensitive to the tax incidence assumptions made, simply because the measure of a household’s total income differs according to whether such taxes as the corporation tax or property tax are borne by the owners of corporate capital or are shifted to consumers. In the former case, one must attribute, as part of households’ pretax income, the amount of corporate taxes they will ultimately pay through the corporation in corporate tax liabilities. The difference between income distributions is not very great.

Same Series

Other IMF Content

Other Publishers

  • Collapse
  • Expand
Cover IMF Staff Papers
IMF Staff papers: Volume 28 No. 2
Author:
International Monetary Fund. Research Dept.
Volume/Issue:
Volume 1981: Issue 002
Publisher:
International Monetary Fund
ISBN:
9781451946871
ISSN:
1020-7635
Pages:
208
DOI:
https://doi.org/10.5089/9781451946871.024