Journal Issue

The Effects of Economic Development on the Base of a Sales Tax: A Case Study of Colombia

International Monetary Fund. Research Dept.
Published Date:
January 1968
  • ShareShare
Show Summary Details

In fiscal studies, as in other areas of economic analysis, the principal focus in recent years has shifted to development. As more and more government operations are expected to make some positive contribution to economic development, analysis of tax and expenditure instruments must be extended to an examination of their effects on economic development as well.

Even as these fiscal instruments are used to promote growth, however, they are themselves affected by the economic and social changes which accompany development. Accordingly, intelligent selection of workable fiscal measures appropriate to various stages of development requires a better understanding of how fiscal possibilities change as the development process unfolds.

In recent years some attention has been devoted to the differences in tax structures of countries at different levels of development. Professor Harley Hinrichs, for example, has noted the sequence of government dependence upon traditional direct taxes on property and gross income, taxes on foreign trade, internal indirect taxes, and modern direct taxes on net income, at succeedingly higher stages of development.1 What has not been fully investigated, however, is why these changes occur and how the underlying possibilities for taxation—which may or may not be used—are altered in the course of development.

One possible approach to this question lies in the study of changes in the base upon which each tax must be levied. The characteristics of a base may provide a specific indicator for the diverse effects of economic developments upon the possibilities for taxation. Because the base of a tax can exist even when the tax is not utilized, it may be studied as one fiscal dimension of the economy when the tax itself cannot.

Following this approach, and utilizing several new methods of analysis, this paper examines one major part of tax structure alteration—the changing possibilities for the application of a sales tax.2 The study is carried out primarily in the context of historical developments in one country, Colombia, which has 18 million people, an area of 400,000 square miles (roughly twice the size of France), and per capita income of about US$325 in 1966. No sales tax was imposed in Colombia until the end of the period studied (1925–65), so that this paper focuses upon the base of a hypothetical sales tax, defined essentially along the lines of the 1965 tax. To provide background for study of this base, an account is presented of two earlier consumption tax episodes as well as of the sales tax levied in 1965. The paper then considers changes over the 40-year period in the size of the base, the incidence of a tax upon it, the adequacy of actual coverage of the base by customs and sumptuary excises, and accessibility of the base to the collection of a possible tax upon it. The findings indicate substantial changes in some, but not all, of these characteristics.

I. Three Consumption Taxes

While the main concern of this paper is with the base underlying possible sales taxation, three episodes in consumption taxation over the past half century offer revealing insights into the effects of changes in the base during this period. The first two episodes, interestingly enough, occurred in wartime—in 1914 and 1942—when the sharp reduction of imports and customs revenues generated some of the same pressures for greater domestic taxation that arose again at later stages of development. The third episode, embodying a far broader sales tax than earlier efforts, points to a number of significant effects of intervening economic developments which later analysis explores.


With the reduction of imports that accompanied the outbreak of World War I in 1914, Colombian customs receipts—which furnished about three fourths of the Government’s ordinary revenue—dropped markedly, falling by 38 per cent between 1913 and 1915. To counter this decline, while restricting expenditures and borrowing from various local sources, the Government imposed a consumption tax on various luxury items. Carrying specific charges which for some products became proportional at higher prices, this tax applied to wines and liquors, beer, patent medicines, cosmetics, imported tobacco and tobacco products, and matches.3 Few of these items were produced within the country, however, and the authorities eventually found it much simpler to make the tax on imported items a part of the customs tariff.4 As a result, McQueen in Colombian Public Finance (page 29) states

By law No. 83 of 1922 and decree 208 of 1923 the imported articles in these categories, comprising the large volume of those paying the tax, were made subject to new customs duties, and the consumption tax proper was restricted to articles produced in the country—limiting the tax, as a practical matter, to beer, matches, and playing cards.

Collection of the remaining tax, which under provision of a 1923 law was scheduled to be transferred to the Departments, came through contracts with the factories for lump-sum payments to cover production during specified periods. The consumption tax yielded only between 1 per cent and 2 per cent of total national government revenues in 1923, roughly the magnitude of the new income tax that was imposed in 1918. Given the importance of customs and public enterprise revenues, however, this amounted to about 13 per cent of all revenues derived from domestic taxes. This early consumption tax, therefore, ended with the transformation of most of the tax into increased customs duties and the transfer of sumptuary levies on domestic goods to the Departments.5


A second consumption tax episode occurred some two decades later, when supply and shipping shortages accompanying the U.S. entry into World War II once more reduced the flow of Colombia’s imports and customs revenues. The intervening years had seen the growth of a number of domestic manufacturing industries, however, and a somewhat different solution was chosen this time to meet the needs for revenue. Rather than redressing the drop of almost 40 per cent in imports and customs receipts by increasing the tax or customs rates on remaining imports, the Government decided to bring the output of the domestic manufacturing industry into the revenue base. By Decree No. 1361 of June 6, 1942, a tax of 3 per cent was imposed upon the domestic sales by textile factories (producing yarn or cloth of cotton, wool, or silk), sugar mills, breweries, and cement factories. Payment of the tax was to be based upon declarations of sales filed within 15 days after the close of each calendar quarter.

Because of the concentration of manufacturing production in a relatively few industries (in marked contrast to later stages of development), the imposition of this limited consumption tax—in effect, selective excise levies on a very few products—promised to provide coverage of a large portion of total manufacturing production and to be comparatively easy to administer. Production in these four industries (textiles, sugar, beer, and cement) accounted for 40 per cent of total manufacturing production in 1940 and, together with the tobacco and liquor industries (already taxed through departmental excises and liquor monopolies), represented 64 per cent of the total. (The corresponding figures for production in these industries in 1963 were 23 per cent for these four branches of manufacture and 27 per cent if distilled liquors and tobacco were included (see Table 1).)

Table 1.Colombia: Relative Importance in 1940 and 1963 of Manufacturing Industries Taxable Under Decree 1361 of 1942 and of Those Covered by Excises and Fiscal Monopolies(Production in thousands of pesos)
Number of


in Survey
Per Cent

of Total
ProductionPer Cent

of Total
ProductionPer Cent

of Total
Taxable under Decree 13611298.475,28239.64,580,49823.0
Distilled liquors271.718,5229.8382,0461.9
Covered by excises or fiscal monopolies1197.746,73324.7887,7774.4
Total covered24816.1122,01564.35,468,27527.4
Total production in survey1,541100.0189,653100.019,955,151100.0
Sources: For 1940: Departamento Administrativo Nacional de Estadística (DANE), Anuario General de Estadística, 1940 (Bogotá, 1941), p. 52. For 1963: DANE, Anuario General de Estadística, 1963 (Bogotá, 1965), pp. 612 and 662–84.

The ease with which this broad coverage could be obtained was increased by the relatively small number of large firms which predominated in the sugar, beer, textile, and cement industries. The 66 taxpaying firms in these four industries accounted for 69 per cent of income tax payments by all manufacturing companies in 1940 (see Table 2), while 9 of the 40 textile firms included in this number were responsible for nine tenths of all textile production.6

Table 2.Colombia: Income Tax, Net Wealth Tax, and Excess Profits Tax Contributions by Various Manufacturing Industries, 1940
Number Of
















(Cols. 4+5+6)

(In thousands of pesos)
Subtotal (Decree 1361 Industries)6633,64316,1112,3356654123,412
All Manufacturing Companies28866,47225,4203,3601,0485344,942
All Companies10,9783,7502,12216,850
(As per cent of manufacturing companies)
Subtotal (Decree 1361 Industries)22.950.663.469.563.577.269.0
All Manufacturing Companies100.0100.0100.0100.0100.0100.0100.0
(As per cent of all companies)
Decree 1361 Industries21.317.719.420.2
Source: Contraloria General De La RepúBlica, DireccióN Nacional de EstadíStica, EstadíStica Fiscal Y Administrativa, 1942–1943, Anexo al Informe Financiero Del Contralor General De La RepúBlica (Bogotá, 1945, hereinafter Cited As EstadíStica Fiscal Y Administrativa), Pp. 43–57.

Difficulties arose because of the regressivity of the tax, however, as the composition of the 1942 domestic manufacturing base imposed a markedly regressive character upon any tax that it could support. This regressivity reflected the effects which import substitution had exerted on the distribution of the tax burden. Because goods with a mass market were the first to attract domestic manufacture, the substitution of nontaxed domestic production for imports had reduced a part of the revenue system which had rested primarily on the poorer classes. To some extent, of course, this lighter burden might be offset by the higher prices charged by less efficient domestic manufacturers enjoying tariff protection. The return of these mass-consumed articles to the revenue base—through the taxation of domestic manufacture proposed in 1942—had corresponding regressive implications for the distribution of the tax burden.

Dissatisfaction was quick to appear. By October 1942, the Government presented a new fiscal and economic plan to Congress, which soon approved it. This plan rescinded the taxes created by Decree No. 1361 “which bear so heavily upon the popular consumption,” with the exception of that on beer, which was raised to 5 per cent and was transferred to the Departments. In place of these taxes, mandatory bond purchases were imposed upon exports and incoming capital, and income, stamp, and inheritance taxes were increased.7 Income tax collections, enhanced by this addition (and augmented since 1936 by surcharges for net wealth and excess profits), soon surpassed customs revenues to become the Government’s leading source of revenue (see Table 3). Sumptuary excise taxes on beer, liquor, and tobacco, the major source of departmental revenues, dropped from 22.5 per cent of consolidated current government revenues in 1940 to 16.0 per cent in 1963, as the shrinking importance of alcoholic beverages and tobacco in national manufacturing production weakened departmental finances (see Table 4). The remainder of Colombia’s manufacturing industry, meanwhile, was to remain free from a broad sales levy for another 23 years.

Table 3.Colombia: Customs, Income Tax, and Total Current Revenues of the National Government, 1915–65(Amounts in thousands of pesos)
YearCustoms1Income Tax2Total Current RevenuesAs Per Cent of Total

Current Revenues
CustomsIncome tax
Sources: For 1915–25: Charles A. McQueen, Colombian Public Finance (U.S. Department of Commerce, Washington, 1926), pp. 20 and 24. For 1926–43: Estadística Fiscal y Administrativa, 1942–1943 (Bogotá, 1945), pp. ix-x. For 1944–48: United Nations, Department of Economic Affairs, Public Finance Information Papers: Colombia, ST/ECA/SGR.A/2, March 1950, pp. 28–31. For 1949: Estadística Fiscal y Administrativa, 1949 (Bogotá, 1951), pp. 3–8. For 1950–65: Dirección Nacional del Presupuesto, Boletín No. 53 (Bogotá, December 1966), pp. 89–99.
Table 4.Colombia: Composition of Consolidated Current Government Revenues, 1925, 1940, and 19631
Thousands of PesosPer Cent of Total Current RevenuesThousands of PesosPer Cent of Total Current RevenuesThousands of PesosPer Cent of Total Current Revenues
National government
Income and net wealth tax7321.021,95814.91,566,00029.8
Liquor and beer7,82911.018,72212.7583,00011.1
Social security4370,0007.0
Total current revenues71,267100.0147,420100.05,261,000100.0
Sources: For 1925: Charles A. McQueen, Colombian Public Finance (U.S. Department of Commerce, Washington, 1926), pp. 24–31, 80, and 88. For 1940: Estadística Fiscal y Administrativa, 1942–1943 (Bogotá, 1945), pp. 2–7, 64, and 155–233; for municipalities, DANE, Anuario General de Estadística, 1941 (Bogotá, 1942), p. 169. For 1963: Data from the Treasury and the Banco de la República.

1965 8

When a sales tax was enacted, late in 1963 to take effect at the beginning of 1965, it was under the pressure of rising development expenditure needs, a cyclical decline in real customs receipts, and the reduced growth of income tax collections under the liberalized exemptions provided by the revenue-sacrificing reform that had been adopted in 1960, at the cyclical peak of imports and customs revenues. To meet the resulting revenue needs, the legislature voted on August 20, 1963 to increase stamp and inheritance taxes, require government bond purchases by banks, levy a surtax on the income tax for 2 years, and institute a general sales tax.9 Under the new law the President was authorized “to establish national taxes on the sales of finished articles by manufacturers or importers at rates varying between 3 and 10 per cent … with exemptions for food items of popular consumption, school books, drugs, and exports.” (Law 21, Article 1 (6).) The resulting general sales tax at the manufacturers’ and importers’ level was established by Decree No. 3288 of December 30, 1963, to go into effect a full year later, on January 1, 1965.10 In January 1965 opposition to the sales tax culminated in threats of a general strike, leading the Government to refer the problem to a Grand Commission of Inquiry, which was set up to consider a number of tax and economic questions, and to issue a revised regulatory decree, No. 377, on February 25, 1965. This was to stand with little change until mid-1966, when several changes in procedures and rates were decreed.

Under these regulations the tax was levied upon the sale or exchange11 of finished articles (defined as those which are the result of an industrial process and are ready for use or further processing), including capital goods but excluding services, food, medicine, school books, exports, gasoline (which is taxed separately), and—to give full scope to the exemption of food—insecticides, fertilizers, seeds, and animal feed. Taxable articles were divided into four rate categories carrying rates of 3 per cent, 5 per cent, 8 per cent, and 10 per cent.12While a number of imported articles were taxed at higher rates than similar products produced domestically, imports from countries in the Latin American Free Trade Association—which accounted for 8 per cent of all Colombian imports in 1965 13—were taxed as though produced domestically. With effect from July 1, 1966, the 8 per cent and 10 per cent rates were raised to 15 per cent, and the 5 per cent rate to 8 per cent, while the 3 per cent rate remained unchanged.14

Under Decree No. 377, finished articles sold to other manufacturers for subsequent processing or to exporters for export were exempted from the tax, upon presentation to the seller of a statement on intended use. In mid-1966, however, this exemption was restricted to exporters and manufacturers of exempt products.15 Other manufacturers were required to pay the tax but were permitted to take a credit for the tax paid on purchases of materials physically incorporated in the products, though not for other materials or capital equipment. On goods sold to persons or firms economically related (by marriage, consanguinity, or 50 per cent common ownership), the tax could be delayed until sale to third parties or could be paid, at the seller’s option, and deducted from the tax liability on the later sale. In either instance, the sale or transfer was to be reported on the bimonthly sales declaration.16

All importers and manufacturers were required to register and to submit declarations of their sales (or lack of taxable sales) and taxes due, along with payment. In July 1966 this requirement was extended to manufacturers of exempt goods as well, and declarations were required for all sales periods even if there were no sales or sales only of exempt products.17 At first, sales declarations were due within a month and a half after each bimonthly sales reporting period; later they were due a month after the end of the period. The tax on imports was generally collected not at customs but upon subsequent sales by the importer, with some exceptions, such as direct importation by non-registered firms or individuals.18 Exemptions from customs duties did not imply exemption from the sales tax. After some initial ambiguity, sales of domestic liquor, lotions, and colognes produced by departmental monopolies were exempted from the sales tax.19

Implementation of the sales tax began with only minimal enforcement efforts during the first year, and an unexpectedly small number of registrants was recorded. Compared with a total of 8,230 nonfood manufacturing establishments reported by the Statistics Department in 1962,20 registrations of both importers and manufacturers who were subject to the tax reached only 5,137 by the end of 1965; further enforcement efforts, however, increased registrations to 8,250 by September 31, 1966.21 In mid-1965, data for four major regional offices showed that only about two out of three registrants were filing the required sales declaration (see Table 5). Data for the Bogotá region alone for the full year 1965 showed that some 77 per cent of registrants filed reports, but this included late filing through August 1966 (see Table 6).

Table 5.Colombia: Registrations and Sales Tax Declarations Due on June 15, 1965 and Filed on Time or Between June 16 and July 31, 1965 in Four Major Regional Offices
BogotáMedellínCaliBarranquillaTotal of Four Offices
(Number of registrations or declarations)
Declarations filed1,2185254132662,422
On time9363633412041,844
Late (June 16–July 31)2821627262578
Registrants not filing declarations6472422551501,294
(In per cent of total registrations)
Declarations filed6568.461.863.965.2
On time50.¿47.351.049.049.6
Late (June 16–July 31)
Registrants not filing declarations34.731.638.236.134.8
(In per cent of total declarations filed)
Declarations filed100.0100.0100.0100.0100.0
On time76.869.182.676.776.1
Late (June 16-July 31)23.230.917.423.323.9
Source: Data from Ministry of Finance, Division of National Taxes.
Table 6.Colombia: Taxpayers, Registrants, and Sales Tax Collections by Type of Manufacture and Importat Bogotá Tax Administration, 1965(Collections and payments in thousands of pesos)
RegistrantsCollectionsAverage Payment per Taxpayer
Number of TaxpayersNumberPer cent paying taxTotalPer cent of total
Type of Industry
Mines, oilfields, and quarries4580.06260.4156.5
Manufacturing industry
Wood, except furniture212875.01,3250.863.1
Wooden furniture284463.69410.633.6
Paper and paper products223170.91,2220.855.5
Printing, publishing, etc.182572.01690.19.4
Leather, except shoes193161.22380.112.5
Rubber products151978.93780.225.2
Chemicals and chemical products14619475.69,4695.864.9
Carbon and petroleum derivatives66100.05,4883.4914.7
Nonmetallic minerals819090.09,6655.9119.3
Basic metal industries283873.610,0136.1357.6
Metal products, except machinery and transport equipment8912471.85,0423.156.7
Nonelectric machinery303976.91,4610.948.7
Electrical and refrigeration equipment517270.810,2226.2200.4
Transport materials263476.41,6391.063.0
Various manufactures9211480.73,9042.442.4
Retail trade122254.58770.573.1
Types of Import
Agricultural and industrial machinery919793.85,7403.563.1
Transport equipment and parts17221379.811,9687.369.6
Electrical and scientific equipment and materials809782.57,3284.591.6
Textiles and clothing2
Chemical and pharmaceutical products203652.213.5328.3676.6
Food and liquor284168.31.8891.267.5
Others not classified11916472.55,8963.649.6
Office equipment192382.61,3090.868.9
Hardware and refrigeration and air-conditioning equipment799087.73,3352.042.2
Source: Ministry of Finance, Division of National Taxes and Budget Office.

Despite early problems in securing compliance, collections of the sales tax for the first year (covering sales in the calendar year 1965 and payments through February 1966) totaled Col$392 million, equivalent to about 10 per cent of national government revenues in 1965. Some 63 per cent of collections, and 78 per cent of the base, was at the rate of 3 per cent. At an average rate of 3.73 per cent, the tax was collected on a base of Col$10.5 billion—18 per cent of Colombia’s gross domestic product (Col$58.8 billion) for 1965 (see Table 7). Data for collections in Bogotá—roughly half of the national total—for July-August 1965 indicated that 30 per cent of the collections came from importers and 70 per cent from domestic manufacturers. Imported merchandise was more heavily represented in the high tax rate categories, however, as imports constituted roughly 60 per cent of the sales in the 8 per cent and 10 per cent categories (see Table 8). Collections at the 5 per cent rate were almost exclusively from domestic manufacturers, reflecting in part the dominant position of payments for beer, which alone accounted for 21 per cent of collections in the Bogotá region during 1965 (see Table 6). Data on the size of tax payments in Bogotá indicate an extreme degree of concentration of collections. While two thirds of the sales taxpayers accounted for only 5 per cent of total collections, 47 per cent of all the collections in Bogotá came from the 20 top firms (see Table 9). Though the Bogotá collection data cover only half of the national total, include both imports and domestic manufactures, and are affected by the rate of the tax as well as the base, they roughly resemble the pattern of concentration that is evident for the domestic portion of the tax base represented by manufacturing production in 1962 (see Table 10 and Chart 1).

Table 7.Colombia: Collections and Implied Sales Base of Sales Tax by Rate Category, 1965 and First Half 1966
Collections1Implied Sales Base
Rate CategoryTotalPer cent of totalTotalPer cent of total
(per cent)(millions of pesos)(per cent)(millions of pesos)(per cent)
1965 2 (Average Rate 3 = 3.73 per cent)
First Half 1966 4 (Average Rate 3 = 4.12 per cent)
Source: Ministry of Finance, Division of National Taxes.
Table 8.Colombia: Base of Sales Tax Payments by Importers and Domestic Manufacturers, at Bogotä Tax Adminstration, July–August 1965
Rate Category

(per cent)
Base of Sales Tax Payments
By importersBy domestic manufacturersTotal
(In thousands of pesos)
(In per cent of tax base by type of taxpayer)
(In per cent of tax base in rate category)
Source: Finance Ministry, Division of National Taxes and Budget Office.
Table 9.Colombia: Distribution of Sales Tax Payments in Bogotá Tax District, 1965
Amount of Tax Paid (pesos)TaxpayersTax Payments
NumberPer cent of totalCumulative per centAmount (thousand pesos)Per cent of totalCumulative per cent
Less than 1,00016410.310.3740.040.04
1,000 to 9,99955034.544.82,4341.51.5
10,000 to 19,99922213.958.73,1882.03.5
20,000 to 29,9991368.567.23,2832.05.5
30,000 to 39,999795.072.22,7351.77.2
40,000 to 49,999644.076.22,8451.78.9
50.000 to 99,9991509.485.610,6796.515.4
100,000 to 199,999976.191.713,1068.023.4
200,000 to 499,999805.096.725,96615.939.3
500,000 to 999,999332.098.721,64913.252.5
1,000,000 to 5,999,999171.199.837,89223.275.7
6,000,000 and over30.2100.039,71824.3100.0
Source: Finance Ministry, Division of National Taxes and Budget Office.
Table 10.Colombia: Distribution of Production Among Manufacturing Establishments Ranked by Number of Employees, 1962 1
Number of

Manufacturing EstablishmentsGross Production

(million pesos)
Per cent

of total


per cent
Number of establishmentsPer cent of totalCumulative per cent
Less than 53,45431.231.22881.81.8
200 or more2171.9100.07,09745.4100.0
Source: DANE, Anuario General de Estadística, 1962 (Bogotá, 1964), p. 760.

Chart 1.Colombia: Distribution of Production Among Manufacturing Establishments Ranked by Number of Employees, 1962, and of Bogotá Sales Tax Collections, 1965

Per Cent of Manufacturing Establishments, 1962, or of Bogoti Sales Taxpayers, 1965

During 1966, sales tax collections increased sharply, totaling Col$644 million. This gain reflected primarily the rise in imports—particularly automobile imports—to a 6-year peak and the generally higher tax rates on imports. This was apparent in the rate distribution of collections during the first half of 1966, when the share of collections in the 8 per cent and 10 per cent rate categories was approximately double that in 1965 (see Table 7). Higher collections were attributable also to the statutory rate increases enacted in mid-1966 and to more effective measures for enforcing the tax.

II. Changes in the Base of a Possible Sales Tax

Between the imposition of the consumption tax in 1914 and the inauguration of the sales tax in 1965, important changes were taking place in the Colombian economy—changes which altered the size, composition, and accessibility of the base upon which a tax system could be imposed. Part II of this paper departs from the actual development of the tax system to examine the changes which occurred in the economic base underlying a possible sales tax.

To measure only economic changes in the base for a sales tax, rather than the effects of definitional differences in tax legislation, the base considered for the entire period is essentially that defined in the sales tax inaugurated in Colombia in 1965. The base studied comprises all expenditures except those for food, medicine, and services. While this differs from some sales taxes adopted in other countries, it is not an uncommon tax base definition. Food is often excluded by sales tax legislation for reasons of equity, and services are often excluded because of the administrative difficulties that they entail.

The base studied differs somewhat from the base for the 1965 sales tax, which includes some double counting of machinery and materials not physically incorporated in the final product. Under the 1965 law these items are taxed on purchase by manufacturers and again when included—without offsetting tax credit—in the taxable value of manufacturers’ sales. There are differences, too, in the exact dimensions of the tax base studied when it is examined from various points of view. The tax base relevant to an examination of private consumption expenditures, excluding business expenditures, comprises retail sales, or the aggregate of the value added at all levels through the retail stage. The tax base examined from an aggregate, national production point of view, including business expenditures, falls somewhere between the value of sales by manufacturers and by wholesalers (as does the 1965 sale tax base). It embraces the value added in industry at the manufacturers’ sales level and the value of some imports, which may be closer to the wholesale level. A tax base at the retail level, to be sure, is larger by the amount of the retail markup than one at the wholesale level, and a base at the wholesale level exceeds that at the manufacturers’ level by the amount of the wholesale markup. In general, however, though divergencies may result from differences between the wholesale and retail markups of different products or over the course of development, the analyses of the base measured at one level are applicable to the base measured at other levels as well.

For the 40 years 1925–65, changes in four aspects of the hypothetical sales tax base are examined: (1) its elasticity, or growth in relation to changes in total income or expenditures; (2) the distribution of the base and the resulting regressivity or progressivity of a tax on it; (3) the adequacy of coverage of the base through the customs house; and (4) the accessibility of the base to possible tax collection at the manufacturers’, wholesalers’, and retailers’ levels. While the context of this study remains that of Colombian development over the past 40 years, evidence from other countries is adduced where necessary to fill in the outlines of tax base changes in Colombia.


A number of elements suggest that the base of a general sales tax covering all expenditures but food, medicine, and services does not constitute the same percentage of either personal expenditures or gross national income throughout the development process. At the lowest levels of development, a significant portion of the population is engaged in subsistence living, that is, much of their consumption consists of goods that they produce themselves. An important part of consumption, therefore, does not pass through the regular marketing mechanism and remains outside the hypothetical sales tax base. While in a developed country the sales tax base of rural families is proportionately smaller because of home-grown food, which is of no consequence if the sales tax exempts food,22 in a developing country rural consumption of nonmarket goods extends to many nonfood items as well. In Colombia in 1953–54, for example, rural per capita consumption was about one third that in urban areas,23 but per capita retail sales outside the departmental capitals—in effect, in the rural areas—was only about one sixth the level inside the capital cities, and nonfood retail sales were only one eighth.24

Besides the effects of subsistence living upon the flow of goods through market channels, the relative size of the sales tax base is affected by changes in the composition of consumption as per capita income rises. The portion of total personal consumption expenditures which is devoted to each category of consumption, such as food or clothing, varies with the level of income and affects how much of total consumption is to be inside or outside the sales tax base. This is true not only of families but also of countries at different levels of per capita national income.

The influence of income levels upon the portion of personal consumption expenditures that is within the sales tax base may be examined by applying the definition of the tax base used here—excluding food and services, medicine being rather difficult to segregate—to consumption patterns in a cross section of countries with different income levels. (This comparison does not affect, of course, that part of the tax base which comprises business expenditures rather than personal consumption expenditures.) The results of such an examination suggest that the percentage of private consumption expenditures embraced by this sales tax base rises appreciably as per capita national income rises.

Using data for the 1950’s compiled by Professor Simon Kuznets from UN sources for 27 countries arranged in four groups, consumption categories have been classified as predominantly inside or outside the tax base (see Table 11).25 This classification indicates that the portion of private consumption expenditures included within the sales tax base rises with increases in income, going from 27.0 per cent in the group of countries with an average per capita net national product of US$116 a year to 37.6 per cent in countries with an average of US$1,110. This increase in the percentage of consumption expenditures that is within the sales tax base as per capita income rises may be viewed in terms of the two major exemptions, food and services. Food consumption decreases as per capita income rises, from more than 60 per cent of the personal consumption expenditures in the very least developed countries to less than 30 per cent in the wealthiest countries. The long-run income elasticity of the food-consumption function is about 0.7, so that as income rises by 10 per cent, food expenditures would rise by only 7 per cent.26 Services, on the other hand, represent a more heterogeneous collection of expenditures with less clear-cut behavior in the aggregate. Between the lowest and highest income categories in Table 11, however, the services component—tax-exempt categories other than food—shows a rising trend, with an income elasticity of consumption of about 1.2, so that service expenditures rise by 12 per cent for each 10 per cent increase in per capita income. The combination of these two components brings a decline in exemptions as incomes rise, the food component declining more rapidly than the service component grows. This is so because food expenditures are far greater than service expenditures in the lower income category and because the elasticity of food consumption is further below 1 than the elasticity of service consumption is above it. (1.0–0.7=–0.3; 1.2–1.0 = 0.2.) As incomes continue to rise, however, the percentage of food in total expenditures grows steadily smaller, and the services percentage grows steadily larger. Increasingly, therefore, the rise in the service component offsets more and more of the decline in the food component. The decline in exemptions as a percentage of consumption becomes progressively smaller, and, at the highest levels of per capita income, exemptions should actually rise. Assuming constant income elasticities for these categories of consumption, the turning point comes when

Table 11.Percentage of Personal Consumption Expenditures Taxable Under a Hypothetical Sales Tax Exempting Food and Services in Countries at Various Income Levels During the 1950’s(In per cent of total personal consumption expenditures)
(Per capita income categories)
Tax-exempt expenditures
Rent and water10.
Fuel and light3.
Household operation3.
Transportation and communications
(excluding automobiles 1)
Recreation and amusement6.
Miscellaneous services5.
Taxable expenditures
Personal care and health4.
Total private consumption expenditures100.0100.0100.0100.0
Taxable as per cent of adjusted private
consumption expenditures37.636.431.327.0
Private consumption expenditures as
per cent of GNP65.370.073.173.8
Taxable expenditures as per cent of GNP24.625.522.919.9
Average per capita net national product
(1953 U.S. dollars)1,110552232116
Average per capita GNP (1953 U.S. dollars)1,217606247124
Source: Simon Kuznets, “Quantitative Aspects of the Economic Growth of Nations: VII, The Share and Structure of Consumption,” Economic Development and Cultural Change, Vol. X, No. 2, Part II (January 1962), p. 24. Based on data for 27 countries, grouped as follows. Group 1: Australia, Belgium, Canada, Sweden, United Kingdom, United States. Group 2: Finland, France, Netherlands, Norway, Austria, Ireland, Puerto Rico. Group 3: Italy, Panama, Union of South Africa, Dominican Republic, Jamaica, Japan. Group 4: Ceylon, Ecuador, Ghana, Honduras, Peru, Rhodesia-Nyasaland, Taiwan, and South Korea.

when S=services, F=food, eF=income elasticity of the food consumption function, and es=income elasticity of the consumption function for services.27 With income elasticities of consumption of 0.7 for food and 1.2 for services, the tax base would cease rising and begin to decline as a per cent of total expenditures when services reached 1.5 times the value of food expenditures. This point is already reached in the United States, for which the services-to-food ratio for 1950–59 is 1.67, using adjusted Kuznets data.28

Another influence on the growth of the sales tax base may be the importance of nonpersonal expenditures in gross national product (GNP). With a sales tax restricted to personal expenditures (which is not the case in Colombia or with the base studied here), the declining share of personal consumption expenditures in GNP tends to reduce the sales tax base as a per cent of GNP in the highest income category (see Table 11).

These influences of personal consumption expenditures on the sales tax base are reflected in the elasticities calculated for the base over the full income range. Between the lowest and highest income categories in Table 11, the income elasticity of the sales tax base is thus 1.15 and the elasticity of the tax base with respect to expenditures is 1.22.29 Both income and expenditure elasticities level off as income rises, falling to about 1.00 as they reach the highest income categories. The elasticities between categories, derived from Table 11, are as follows:

Movement of per capita net national
product in 1953 U.S. dollars from116232552116
Expenditure elasticity of tax base1.
Income elasticity of tax base1.211.120.951.15

If the comparison of consumption patterns in countries at different income levels offers a valid indication, the growth in per capita income within a country may be expected to produce a rise in the sales tax base which is more than proportionate.

This expectation appears to be confirmed by data for Colombia over the past 40 years. These data, however, focus not on changes in personal consumption patterns over this period but on the relationship between income and nonfood imports and production. Two sets of data for income and production are available for this period, one covering the period 1925–54 prepared by the UN Economic Commission for Latin America (ECLA) and the other covering the years from 1950 onward prepared by the Colombian central bank. From these data it has been possible to calculate a reasonable approximation of the hypothetical sales tax base by adding the value of nonfood imports to the value added by nonfood manufacturing industry within Colombia. Though this figure for the sales tax base includes the value of medicine and excludes the value of nonfood domestic raw materials used by the manufacturing industry, it is believed to offer a fairly good approximation, on the whole, of movements in the hypothetical sales tax base over the period 1925–63.30

Use of a time series to examine long-term changes in the sales tax base, however, reveals an important element that is not apparent from the cross-section analysis of countries at different income levels. This is the effect of shorter, cyclical fluctuations on the percentage of income that is included in the sales tax base. In Colombia such cyclical variations in the hypothetical sales tax base have been quite pronounced and have been restricted to the import component (see Table 12 and Chart 2). Influenced by the depression in the 1930’s, World War II, and postwar variations in coffee export receipts and foreign credits, Colombian imports—as a per cent of gross domestic product (GDP)—have reached periodic peaks since 1925 at intervals of 10 years, 10 years, 7 years, 6 years, and 6 years. Whether through income generation, monetary expansion and contraction, or direct trade restrictions and controls, the sporadic requiting of pent-up demand for imports of capital goods and of durable consumer goods has been a regular feature of the Colombian economy. Though a part of these fluctuations was no doubt absorbed by inventory changes, there has also been appreciable cyclical variation in actual final expenditures as well.31 The import component of the hypothetical sales base, in consequence, shows wide fluctuations, somewhat dampened since World War II, around a level trend line of about 10 or 11 per cent of GDP (see Tables 13 and 14 and Chart 3).

Table 12.Colombia: Total Imports and Import Components as Per Cent of Gross Domestic Product, 1925–64

Raw Materials

and Intermediate

Fuels and




Sources: GDP for 1925–49 calculated from GDP in 1950 pesos and annual GDP deflators, ECLA, Analyses and Projections of Economic Development, III: The Economic Development of Colombia, Statistical Annex (in Spanish, Geneva, 1957, hereinafter referred to as Statistical Annex), Tables 4 and 32. GDP for 1950–64 from Banco de la República, Cuentas Nacionales, with 1964 a preliminary estimate. Composition of imports for 1925–49: Statistical Annex, Table 30. Composition of imports for 1950–57: Departamento Administrativo de Planeación y Servicios Técnicos, Informe al Congreso Nacional, 1961–1963, Table 6 A, p. 135. Composition of imports for 1958–64: Based on data from Departamento Administrativo de Planeación y Servicios Técnicos. To obtain comparability between the two GDP series used, ECLA data for 1925–49 have been increased by 15.77 per cent, their average shortfall for the 5-year period (1950–54) during which the two series overlap.

Chart 2.Colombia: Components of Imports as a Per Cent of Gross Domestic Product, 1925–66

Per Cent of GDP
Table 13.Colombia: Components of Hypothetical Sales Tax Base—Value Added in Nonfood Manufacturing and Nonfood Imports, 1925–63
YearValue Added in Nonfood ManufacturingNonfood ImportsTotalValue Added in Nonfood ManufacturingNonfood ImportsTotal
(In millions of current pesos) (In per cent of total)
Sources: Value added in nonfood manufacturing, 1925–53: Based on data from ECLA, Analyses and Projections of Economic Development, III: The Economic Development of Colombia (Geneva, 1957) and Statistical Annex. Value added in total manufacturing: From Statistical Annex, Table 1; for value added in food manufacture, ratio of value added to production for each branch of food manufacture in 1953 (Analyses and Projections of Economic Development, III: The Economic Development of Colombia, p. 419) was used and weighted by changing composition of food production each year 1925–53 (p. 421) to calculate ratio of value added to production for all food manufacture. This was applied to the index of the quantum of food production, Statistical Annex, Table 131; values were converted to current pesos using deflators in Statistical Annex, Table 32. Value added in nonfood manufacturing, 1954–63: Based on Banco de la República, Cuentas Nacionales. Nonfood imports: From data in League of Nations, Memorandum on International Trade and Balances of Payments, 1926–1928, Vol. III (Geneva, 1930), p. 75; League of Nations, International Trade Statistics, 1931 and 1932 (Geneva, 1933), p. 72; League of Nations, International Trade Statistics, 1933 (Geneva, 1934), p. 72; Contraloria General de la República, Dirección Nacional de Estadística, Anuario de Comercio Exterior (Bogotá, various years); United Nations, Yearbook of International Trade Statistics (New York, various years).
Table 14.Colombia: Changes in Hypothetical Sales Tax Base as Per Cent of Gross Domestic Product and in Per Capita Gross National Income, 1925–63
YearValue Added

in Nonfood


Total Sales

Tax Base
Per Capita

Gross National

(In per cent of GDP)(In 1953 U.S. dollars)
Sources: Cols. 2, 3, and 4: Tax base from Table 13 and GDP from Table 19. Col. 5: For 1925–49, from Statistical Annex, Table 4. For 1950–63, from Banco de la República, Cuentas Nacionales. Conversion to 1953 dollars based on average import rate shown in Joint Tax Program of the Organization of American States and the Inter-American Development Bank, Fiscal Survey of Colombia (Baltimore, 1965, hereinafter referred to as Joint Tax Program, OAS/IDB), p. 164. To obtain comparability, ECLA data for GDP and GNI were expanded by their average shortfall from Banco de la República data during the 5-year period (1950–54) during which the two series overlap.

Chart 3.Colombia: Domestic and Import Components of a Hypothetical Sales Tax Base as a Per Cent of Gross Domestic Product, 1925-63

Per Cent of GDP

The domestic component—value added in nonfood manufacturing—shows a quite contrary pattern, moving steadily upward from about 4 per cent of GDP in the 1920’s to roughly 15 per cent in the early 1960’s, with interruptions only in the depression and war years (see Chart 3).32 The total hypothetical sales tax base shows the effects of the wide fluctuations in its import component, so that on a year-by-year basis no relationship exists between the rise in per capita gross national income (GNI) and the hypothetical sales tax base as a per cent of GDP (see Chart 4).

Chart 4.Colombia: Per Capita Gross National Income in 1953 U.S. Dollars and Hypothetical Sales Tax Base as a Per Cent of Gross Domestic Product, 1925–63

Over the long run, however, and abstracting from the influence of short-term import fluctuations, increases in Colombian per capita GNI were accompanied by increases in the hypothetical sales tax base as a per cent of GDP. To remove the influence of import cycles from the data, a moving average was taken of both per capita income and the tax base, varying with the length of the import cycle.33 The moving average of the tax base showed a continuous upward trend over the period, from 13.8 per cent of GDP in 1925–34 to 25.1 per cent in 1957–63, while per capita GNI rose from US$122 to US$216. A chart of both moving averages since the 1920’s showed parallel growth between per capita income and the tax base as a per cent of GDP (see Chart 5). While the averages moved at about the same pace through the early 1930’s, the tax base rose more steeply through the remainder of the 1930’s, and per capita income rose more steeply in the postwar decade. Variations may be attributable to the pace of industrialization or to the shift of population to the urban centers or to other factors. In any case, the removal of cyclical fluctuations from the data for Colombia over the past 40 years seems to point quite clearly to the same concomitant increase in per capita income and in the ratio of the sales tax base to GDP that is evident in a cross-section analysis of countries with various per capita incomes.

Chart 5.Colombia: Moving Averages of Hypothetical Sales Tax Base as a Per Cent of Gross Domestic Product and Per Capita Gross National Income, 1930–60

These movements may be compared with those calculated for the cross section of countries in different income brackets. Using moving averages for the sales tax base and GDP,34 the income elasticity of the sales tax base emerges as 1.30 from 1925–34 to 1945–54, the full period for which ECLA data exist, and as 1.36 for the period as a whole.35

These longer run elasticities are somewhat higher than the elasticity of 1.21 calculated for the cross section of countries in the per capita income range of US$116 to US$232 and may reflect a number of differences. The cross-section data are based on family expenditures, and therefore represent personal expenditures at the retail level. The hypothetical sales tax base calculated for Colombia over the past 40 years is based on the value of goods passing through customs and on the value of goods leaving the factory. It is not clear whether retail and manufacturing values may diverge, as the margin or value added in the distribution channels may or may not remain relatively constant over the period. The production measure of the tax base, finally, includes not only personal expenditures but also nonpersonal expenditures, which are covered by the 1965 Colombian sales tax and may have grown more rapidly than nonexempt personal expenditures. Other influences at work may include a steeper rise in urbanization in Colombia than in the cross section of countries used and the possibly quite significant differences between historical developments in consumption patterns and those revealed by comparison at one point in time.


Quite distinct from the size of the hypothetical sales tax base and its growth relative to income is the distribution of the base and the resulting incidence of a tax upon it. The taxation of the domestic manufacturing base alone at the earlier stages of development, as examination of the 1942 Colombian experience noted, was no doubt regressive in character because of the basic mass consumption on which such industry must concentrate to find a large enough market. Whether a tax on all sales excluding food, services, and medicine would be regressive, however, is quite a separate question, depending upon whether the tax is passed forward to the consumer and, if so, upon what proportions of consumers’ total expenditures and income are subject to the tax—and at what rates—in various income brackets.

Whether a general sales tax is passed forward to consumers, absorbed by the seller, or shifted backward to the factors of production remains an unsettled issue. Predominant opinion is rather critical of any general backward shifting of the tax to factors of production.36 As regards the import component of the sales tax base, moreover, the relatively small size of the Colombian market would give little chance of any backward shifting of the tax abroad. The forward shifting of the tax burden to consumers hinges upon the interaction of market forces, that is, the demand and supply elasticities at work in the Colombian economy. While no analysis is attempted here of this question, there are a number of elements on the Colombian scene—heavy tariff protection, quantitative import controls, periodic upward price adjustments, and imperfections in market structures—which would seem to facilitate forward shifting. It would appear reasonable to accept as a working hypothesis at least, therefore, the distribution of a sales tax burden which forward shifting would entail.37

Incidence of the tax would depend in the first place upon the composition of expenditures by the consumers to whom the tax would be passed.38 Fortunately, rather complete, though not recent, data on the consumption patterns of urban families are contained in Colombia’s 1953 consumer expenditure study, upon which the cost of living index is presently based. Separate consumption patterns are calculated for two classes of urban families: the lower-class worker families, with a median income in 1953 of 350 pesos a month derived primarily from wages, and the more middle-class employee families, with a median income in 1953 of 650 pesos a month derived mainly from salaries.39 These two groups are believed to maintain different consumption patterns, and separate indices are calculated for their cost of living and for their income from employment in manufacturing industries.

To approximate the distribution of the hypothetical sales tax base, expenditures for the tax-exempt categories—food, services, and medicine—have been examined for each income bracket for both employee and worker families, using a weighted average of data for the seven cities in the sample. The results are presented in Tables 15 and 16.

Table 15.Colombia: Per Cent of Family Expenditures (Based on 1953 Consumer Expenditure Survey) Comprising Food, Medicine, and Services by Income Brackets1
Monthly Income in 1953 Pesos
Under 100100–199200–299300–399400–499500–599600–699700–799800–899900–9991,000–1,4991,500 and more
Employees(Per cent of total expenditures)
(Per cent of families in sample)
Distribution of families in each income bracket
Source: DANE, Economía y Estadística (Bogotá), November 1958, pp. 178–205.
Table 16.Colombia: Hypothetical Sales Tax Base as Per Cent of Employee and Worker Family Income and Expenditures Reported in 1953 Consumer Expenditure Survey
Monthly Income in 1953 Pesos
Under 100100–199200–299300–399400–499500–599600–699700–799800–899900–9991,000–1,4991,500 and overWeighted average
Tax base as per cent
of expenditures25.024.724.224.321.923.623.823.622.723.425.723.81
Expenditures as per
cent of income126.4121.5102.6100.093.298.793.988.284.873.864.285.64
Tax base as per cent
of income31.630.024.824.320.423.322.320.819.217.316.520.24
Tax base as per cent
of expenditures22.220.822.722.824.124.524.326.624.226.425.823.36
Expenditures as per
cent of income150.9121.9101.593.588.279.683.276.773.475.163.791.31
Tax base as per cent
of income33.525.323.021.321.319.520.219.917.819.816.421.33
(Per cent of families in sample)
Distribution of families
in each income bracket
Source: DANE, Economía y Estadística (Bogotá), November 1958, pp. 35 and 178–205.

As with the elasticity of the sales tax base with respect to changes in per capita income over time or between countries, the distribution of the hypothetical sales tax base among those in various income brackets is mainly the result of the relative changes in expenditures for food and services, the food percentage of expenditures declining and the services component rising in successively higher income brackets. Thus, for worker families, in successively higher income categories the food percentage declines more rapidly than the service component rises, particularly around the median income bracket. As a result, total exemptions are disproportionately lower and the tax base disproportionately higher, with respect to total expenditures, for worker families who are above the median income than for those below. For employee families, on the other hand, the rise in the service component as income rises fully offsets the decline in the percentage spent for food, so that the tax base remains roughly proportional to expenditures, or, if anything, slightly less than proportional as income rises. The range of variation in the proportion of family expenditures within the tax base is from about 22 per cent to 26 per cent, or about 17 per cent of the average. Though the over-all average for both employee and worker families is nearly the same, there are consistent differences between the taxable percentage of expenditures for both groups in almost every income bracket. In the lower income brackets (up to 500 pesos a month), the tax base is smaller for worker families than for employee families because of relatively heavy worker expenditures for food. In the upper income brackets, the tax base is smaller for employee families than for worker families because of heavier employee expenditures for services. While the middle-class employee families spend perhaps 6 per cent less of their total expenditures on food than do worker families, worker families devote less of their spending to services than do employee families—services being closely linked at this stage of development with the maintenance of a respectably substantial household establishment.40 These findings underline the importance of expenditures for services, which at early stages of development may completely offset the tax exemption for expenditures on food granted for equity reasons to reduce the relative burden on those in the lower income brackets. As national income rises and the service component grows in importance, however, significant changes may occur in the character and income-bracket distribution of expenditures for services, domestic services giving way to taxable household appliances, for example, and untaxed public utility transportation yielding to the taxable private automobile.41 While some of the elements that make expenditures for services more important to upper-income brackets at this stage of development may recede, there are no grounds for concluding what pattern subsequent expenditures for services will follow.

The sales tax liabilities of each consuming unit are affected not only by the effects of the composition of expenditures but also by the portion of income which is not spent but is saved. In fact, the influence of savings on the ratio of the tax base to income is in this case greater than the influence of the composition of consumption. Income data collected in Colombia’s consumer expenditure survey for 1953 imply heavy dissaving in both worker and employee families in the lower income brackets and substantial savings near the top of the income scale, particularly among employee families. The result is a pronounced regressivity of the tax base with respect to income. The base declines from more than 30 per cent of income for the lowest income bracket to about 16 per cent for the highest bracket among employee families, and from about 33 per cent for the lowest income bracket to 16 per cent for the highest bracket among worker families (see Table 16).

A number of reservations may be stated with regard to these conclusions on the incidence of the tax, reservations based on variations in family size, the validity of current income as a measure of incidence, and the cyclical character of Colombian consumption patterns. With regard to family size, it might be argued that the foregoing analysis of the tax base for households in different income brackets fails to take account of variations in the number of persons per family and thus does not deal with the tax per person or the inequities of a heavier sales tax burden on large than on small families. These differences and inequities are minimized or eliminated, however, by the tax-exemption of food purchases. A study of sales tax incidence on typical U.S. expenditure patterns found that while a tax base which included food imposed a greater tax on large than on small families, exclusion of food from the tax base equalized the tax for families of different sizes.42 In fact, the exclusion of food from a sales tax base may be expected to approximate a personal exemption increasing with family size, since larger families have generally been found to devote a larger proportion of their expenditures to food. A British study of wage and clerical workers’ expenditures by family size in 1937–38 found that the percentage of total expenditures allocated to food at given income levels was from 6 to 11 percentage points higher for families with three children than for families with no children.43 The problem of differences in family size and uneven sales tax incidence should not arise, therefore, where there is a food tax-exemption, as in the Colombian tax.

Another objection to these conclusions on incidence might be based on the hypothesis that “permanent income” is the main determinant of consumption and ability to pay. From this point of view it might be argued that the current incomes reported in the consumer expenditure survey for 1953 reflect only differences in the respondents’ transitory incomes, which have little, if any, influence on the consumption pattern that is being measured or on the ability to pay, and that permanent income would make a better measure of the incidence of the tax. It may well be that the selection of a stable sample of families that were willing to answer questions for the 1953 survey operated to narrow the variations in permanent income of participants so that a disproportionate amount of saving and dissaving appeared in the incomes of low-income and high-income families, reflecting incomes temporarily above or below the norms. On the other hand, the relatively great temporal stability of income position in Colombia would be expected to reduce the transitory elements of income.44 Whether permanent or temporary income forms a better basis for judging incidence, however, remains—like the choice between income or wealth—an open question. Should the concept of permanent income be judged preferable, more validity would attach to the distribution of the base with respect to total consumption than to income.45

Of perhaps greater significance is the matter of cyclical variations in Colombian patterns of consumption. Because of Colombia’s continued dependence on imports for capital goods and most durable consumer goods, this portion of consumer expenditures has been greatly affected by the regular variations in imports. While total imports have followed a pattern of peaks and troughs, imports of capital and consumer goods have fluctuated even more sharply. The periodic tightening and relaxation of import controls in response to exchange availabilities has been reflected in the change in the consumption of imports over the cycle (see Chart 2), so that many durable goods become available only in the peak import years. This has been true of automobiles, for example, as a chart of the highly cyclical age structure of automobiles compared with the U.S. pattern reveals (see Chart 6).

Chart 6.Colombia: Age Structure of Cars Operating in Colombia on June 30, 1965, and in the United States on December 31, 1964

One result of such fluctuation is cyclical variation in the distribution of a sales tax base. The concentration of important durable goods purchases in only particular years of a cycle must detract from the validity of any estimate of consumer expenditures based on a shorter period. While the peaks of Colombian automobile purchase are 6 years apart, the consumer expenditure survey for 1953 covered expenditures during only 1 year.46 In other circumstances, where fluctuations are less marked, annual variations in durable purchases may have little effect on consumer expenditure patterns. Where yearly changes are so pronounced, however, and the fluctuating expenditures are probably concentrated in upper income families, the distribution of the sales tax burden may vary significantly from year to year, and a consumer expenditure study conducted at only one part of the cycle may offer only a partial guide to the tax’s incidence.47 It would seem reasonable to expect a more progressive incidence of the tax in the peak import years of heavy expenditures for durables,48 a presumption supported by the shift in the composition of the tax base in the peak import year 1966, which is described below.

The base is not the sole determinant of the regressivity of a sales tax. A government may seek to alter the regressivity of a sales tax also by applying differentially higher rates to luxuries, that is, to those goods which are consumed in disproportionately larger amounts by higher income groups. This was so with the Colombian sales tax of 1965, which, in addition to the basic rate of 3 per cent, carried higher rates for specified products. To evaluate the effect of such differential rates on the incidence of a sales tax, a series of calculations has been carried out here, based on data in Colombia’s consumer expenditure survey for 1953. The tax which would have been paid by families in each income bracket of urban employee families was calculated on the basis of the consumption pattern for 1953, assuming full forward shifting (by no means a settled conclusion) and applying to retail expenditures the relevant 3 per cent, 5 per cent, 8 per cent, or 10 per cent rates in effect in 1965.49 Using the approximation of the tax base calculated above—to include all expenditures other than food, medicine, and services—the average tax rate on taxable expenditures was calculated for each income bracket (see Table 17).

Table 17.Colombia: Base, Rate, and Incidence of 1965 Sales Tax (Based on Data in 1953 Consumer Expenditure Survey) on Various 1953 Employee Income Brackets1(Weighted average of seven cities)
Monthly Income in 1953 Pesos
100–199200–299300–399400–499500–599600–699700–799800–899900–9991,000–1,4991,500 and more
Approximate tax base (Per cent of total expenditures not excluded by exemptions for food, medicine, and services)25.024.724.024.321.923.723.723.722.723.425.7
Average rate on taxable expenditures (in per cent)2.883.203.213.333.383.333.463.593.483.553.74
Tax as per cent of average total expenditures0.720.790.770.810.740.790.820.850.790.830.96
Tax as per cent of average income0.910.960.790.810.690.780.770.750.670.60
Sources: DANE, Economía y Estadística (Bogotá, November 1958), pp. 178–90, and 207–220 and text and regulations of sales tax law in effect in 1965.

These calculations indicate that, in the aggregate, the choice of commodities for taxation at the higher rates does seem to have embraced luxury items, that is, items with an income elasticity of consumption greater than 1. Thus, the average tax rate applicable to expenditures in the tax base shows a clear tendency to rise, albeit moderately, with increases in employees’income 50 (see Table 17, line 2). As a result, while the distribution of the tax base among employee families rises somewhat less than proportionately with expenditures, the tax itself is sufficiently modified by the differential rates so as to rise somewhat more than proportionately with expenditures. Because of the substantial variation in the portion of income saved or dissaved, however, the tax remains moderately regressive with respect to income, varying from 0.96 per cent of income for one of the lowest income brackets to 0.60 per cent for one of the highest.

These figures shed some light on the incidence of a hypothetical sales tax on the 1953 pattern of income and consumption of urban employee families. A number of changes in the economy occurred between 1953 and the imposition of the sales tax in 1965, however, and, while data for incidence of the tax in 1965 do not exist, it is of some interest to compare the consumption pattern underlying the 1953 findings with the expenditures reflected in the data for actual collection of the tax in each rate category in 1965 and in the first half of 1966.51 This is shown in Table 18.

Table 18.Colombia: Expenditures by Tax Rate Category—Comparison of Patterns Derived from 1953 Expenditure Survey for Employee Families and from Sales Tax Collections in 1965 and First Half 1966(In per cent of total expenditures in tax base)
Rate CategoryEmployee Families,Sales Tax
(per cent)19531965First half 1966
Sources: Analysis described in footnote 49 and data in Table 7.

This comparison shows a general resemblance between the 1965 pattern and the 1953 employee pattern, with a number of significant differences. While the 3 per cent portion of the tax base was almost the same in both patterns, the 5 per cent rate category was substantially higher and the 8 per cent and 10 per cent rate categories were substantially lower in 1965. The strong showing of the 5 per cent category may be attributable to a high degree of compliance by the large beer companies, which were already accustomed to paying a departmental tax on beer sales. Payments by beverage companies in Bogotá amounted to some Col$34 million during calendar 1965 (Table 6), or more than 11 per cent of the national total. Also, there were tax collections in some areas for at least part of 1965 on sales by departmental liquor monopolies, which were excluded from the tax base in the calculations for 1953.

The lower figures for 1965 for goods in the 8 per cent and 10 per cent categories suggest (besides the possibility of disproportionate tax avoidance) that by that year the general consumption of these luxury goods had not yet reached the proportions found among employee families in 1953 or that a strong cyclical influence was depressing consumption of these goods in 1965. Data on collections for the first six months of 1966, when imports were reaching a cyclical peak, support the latter explanation, underlining the importance of cyclical influences. During this period, primarily because of heavy imports of automobiles and other durables, the proportion of 8 per cent and 10 per cent category goods in the tax base rose to 13.1 per cent—double the percentage (6.6) for 1965 and distinctly above the percentage (10.4) for employee families for 1953, a year in which imports were high but not quite at the peak.

These cyclical shifts in the base of the tax signal parallel variations in its incidence. Given the moderating effects of the higher rates on regressivity, the tax would appear to have been more regressive in 1965 than the pattern calculated for 1953 employees but less regressive than this pattern in the first half of 1966. As the high-rate categories recede from their 1966 levels with the passing of the import peak, the tax may be expected once again to become more regressive than the calculated 1953 pattern. This shift is at least partly offset, however, by the statutory rate increases enacted in mid-1966 for goods above the 3 per cent category.

The incidence of the sales tax, like that of any tax, can best be viewed in conjunction with the remainder of a country’s tax system. An over-all evaluation of the entire Colombian tax system before the sales tax was enacted is contained in a study directed by Professor Milton Taylor. Covering Colombian taxes at all levels of government in 1961, it found only a mild degree of progression for the total tax system—with the over-all tax rate varying from 10.9 per cent of income for the lowest quartile of income receivers to 9.4 per cent, 11.1 per cent, and 12.7 per cent, respectively, for the successively higher quartiles. As a result, Professor Taylor concluded, the highly unequal distribution of income before taxes—the upper quartile had 65 per cent of the total income of natural persons while the lowest quartile received only 5 per cent—remains more or less the same after taxes.52 Against this background, concern with the regressivity of the sales tax and efforts to moderate it is of some significance.

Adequacy of customs coverage

Alongside the several influences on the growth of the hypothetical sales tax base and variation in its distribution, other factors were at work on an important administrative characteristic of the tax base: its relationship with the customs house, where collection of a tax can proceed far more easily than in the private channels of domestic trade.

In the earlier stages of development, when little domestic industry exists, almost all the hypothetical sales tax base may consist of imports. Colombian authorities found between 1914 and 1923 that when little, if any, domestic manufacture occurs a sales tax can best be collected through customs, either as an import surcharge or simply (when tariffs are not bound by treaty) through revised customs rates. As development progressed, however, and took the form of import substitution—the production domestically of goods that were formerly imported—two important changes occurred. First, the portion of the hypothetical sales tax base that passed through the customs house declined, and, second, the customs house itself encountered new difficulties in taxing many of the goods that passed through it. In this section of the paper an attempt is made to measure and to evaluate these two phenomena as they occurred in Colombia over the period 1925–64.

Consideration of the first phenomenon, the decline in the share of the hypothetical sales tax base that passes through the customs house, is based upon an examination of the composition of the tax base itself. The base may be viewed as having three components. The first component passes through customs and consists of the value of nonfood imports, while the other two, which do not pass through customs, are the value added by nonfood domestic manufacturing industry and the value of the manufacturing industry’s purchases of raw materials, intermediate goods, and services from other sectors of the domestic economy. By drawing upon a number of sources, it has been possible to calculate the value of nonfood imports and the value added by nonfood domestic manufacturing industry each year over the period 1925-63 (see Table 13). Unfortunately, construction of a similar series for the third component (manufacturing industry’s purchases of raw materials, intermediate goods, and services from the rest of the economy) was not possible because of inadequate data. Tentative calculations for 1940, 1956, and 1962 indicated the magnitude of this component to be from 10 per cent to 13 per cent of the value added by nonfood domestic manufacturing industry.53 Disregarding the third component of the base for lack of convincing data on whether it has increased or decreased, the hypothetical sales tax base has been taken as the sum of only the first two components: nonfood imports and value added by the Colombian nonfood manufacturing industry.

The portion of this hypothetical sales tax base that passed through the customs house (that is, nonfood imports) declined from about 73 per cent during 1925–29 to 46 per cent in 1958–61, and to only 40 per cent in 1962–63. Data from collections of the tax in the Bogotá tax district in July-August 1965 indicate a 30–70 split of the tax base between importers and domestic manufacturers (see Table 8). The tax base of manufacturers’ sales, however, includes not only value added by manufacturers but also imported goods incorporated in the value of production. Such goods could be covered at the customs and are therefore included here in the import component.

This long-run decline in the import component has not been continuous. Recurrent fluctuations from 6 years to 10 years in duration have worked a strong influence on Colombian imports. Balance of payments influences—the depression in the 1930’s, World War II, and the rise and fall of coffee prices and foreign credits in the 1950’s and 1960’s—brought wide fluctuations in the import component of the hypothetical sales tax base (see Charts 2 and 3).54 The domestic component, nonfood manufacturing, meanwhile continued its steadier and more rapid growth. Though at several points external circumstances reduced imports to exceedingly low levels, consumption being sustained only by the running down of capital equipment and stocks, domestic manufacturing did not expand more rapidly to supply the missing imported goods. During these periods, domestic industry in fact slowed its rate of growth, reflecting the absence of an adequate domestic capital goods industry and a shortage of the imported capital goods upon which expansion of manufacture depended (see Chart 3). This was quite plainly the case during the depression and again in World War II, when for the first time the import component of the hypothetical sales tax base declined to less than its domestic component and the Colombian Government sought abortively to tax the domestic portion of the base.

In the long run, however, it was not through the reduction of imports that the composition of the base was altered, as imports recovered their previous level after each of these crises. Over the entire period, imports fluctuated around a generally level trend, as a percentage of GDP. Rather it was through the expansion of domestic manufacture that the predominant position of the customs house in respect to the hypothetical sales tax base was lost. Nonfood manufacturing rose from 4 per cent of GDP in the late 1920’s to 15 per cent in the early 1960’s, raising from 27 per cent to 60 per cent the portion of the base that did not pass through customs.

The second phenomenon, the difficulties encountered by the customs house in taxing the portion of the tax base passing through it, is attributable to two separate but interrelated aspects of the development process: the changes in the structure of the economy which development brings and the adoption of government policies aimed at promoting further development.

While the changing structure of the economy over 40 years of development had little effect on the level of imports, as a percentage of GDP, it exerted an appreciable effect upon the composition of imports. As the import substitution process progressed, domestic manufacture and assembly operations took up the production of many manufactured consumer goods which had been imported previously. This process increased the importation of raw materials, intermediate products, and—because domestic industry never significantly extended into capital goods production—capital goods as well. Government import and payment control, allocating scarce foreign exchange to capital equipment for development and to raw materials to keep existing industries in operation, carried this process further.

By the early 1960’s manufactured consumer goods had fallen to only 7 per cent of total imports, from about 43 per cent in the late 1920’s (see Table 19). It may be noted that Colombia’s experience, though rather extreme as a result of government import controls, corresponds to that of other countries. A recent study of import composition and development, covering single countries over time and many countries at different stages of development, found that, as per capita income grew, imports of manufactured consumer goods fell as a percentage of total imports. At the same time, imports of intermediate products rose, and capital goods imports, at least in developing countries and in the smaller developed countries, also rose.55 Incidental to this process, however, are important effects on the role of the customs house. Because in Colombia, as in many other countries, customs rates on manufactured consumer goods have been far higher than those on raw materials, intermediate products, and capital goods, this shift in the composition of imports has exerted a depressing effect on customs receipts. Fewer goods have entered at the high rates, and more goods have entered at the low rates. The magnitude of the revenue loss which is attributable directly to this change in import composition depends not only upon the extent of the shift between import categories but also upon the differentials between customs rates in different import categories. Fortunately, data on both these factors are available for Colombia.

Table 19.Colombia: Composition of Imports and Effect on Average Customs Rate, 1925–64
YearConsumer GoodsRaw Materials

and Intermediate

Fuels and




Average Rate

for Total
Average customs rate, Sept. 1959
(Per cent of total imports)
Sources: Weighted average rates of theoretical incidence of customs duties and other duties or charges of equivalent effect on the c.i.f. value of imports, on September 30, 1959, based on imports during 1956–58: Santiago Macario, “Protectionism and Industrialization in Latin America,” ECLA, Economic Bulletin for Latin America, March 1964, p. 69. For composition of imports, 1925–49: Statistical Annex, Table 30; 1950–57: Departamento Administrativo de Planeación y Servicios Técnicos, Informe al Congreso Nacional, 1961–63, Table 6 A, p. 135. 1958–64: Based on data from Departamento Administrativo de Planeación y Servicios Técnicos.

A measure of average rates for each major import category in Colombia was worked out by ECLA on the basis of the tariff schedule that was in effect in September 1959, weighting the rate on each item by its actual importation in the period 1956–58. The results of these calculations showed rates on manufactured consumer goods to be twice as high as those on raw materials and intermediate products and about three times as high as those on capital goods. Assuming that the relative position of these rates had some long-term stability, and that there was no significant shift in the mix of imports within each category, it is possible to estimate the effects of changes in import composition on total customs revenues. By weighting these rates by the percentage of total annual imports in their respective categories for each year between 1925 and 1963, one may calculate how the weighted average rate for all imports would vary from year to year as a result of changes in the composition of imports alone (see Table 19). While the level of each year’s average rate which these calculations yield is of little validity in itself, depending as it does on the tariff in effect in September 1959, the movement of this weighted average rate from year to year depends only on the relative position of the rates for the different import categories in 1959, and so may have some validity. The movement of this calculated, weighted average rate over the entire period indicates that, apart from any changes in the individual tariff rates, the shift in the composition of imports from high-duty categories to low-duty categories reduced customs revenues by about 30 per cent between the late 1920’s and the early 1960’s. This long-term process was not entirely irreversible, however, particularly by strong cyclical influences, as customs revenue for 1966 demonstrated.

In 1966, liberalization of government import controls accompanying devaluation of the exchange rate and a series of tariff increases brought a sharp increase in customs collections. Reflecting primarily the heavy importation of automobiles at an average customs rate of 250 per cent, the average effective customs rate rose from 15.06 per cent in 1965 to 22.56 per cent in 1966, the highest level since the 1930’s (see Table 20). While no data for importation by categories in 1966 are yet available, developments in that year point to the importance of cyclical shifts in the composition of imports. Alongside the long secular shift in the composition of imports, there has been a tendency for raw materials and intermediate products to remain relatively stable over the cycle as a percentage of GDP, while capital and manufactured consumer goods bear a disproportionate bulk of the fluctuations. While these cyclical increases in capital and consumer goods may have been offsetting in their revenue effects in earlier cycles, the later decline in consumer goods and their volatility at recent import peaks has brought the erratic movement in the average customs rate, which is so dramatically illustrated by the 1966 experience. Over the longer run, however, this cyclical volatility has been superimposed upon an erosion of the average customs rate by the shift from high-tariff imports to low-tariff imports.

Table 20.Colombia: Gross Domestic Product, Imports, and Customs, 1925–66(Cols. 1, 2, and 3 in millions of current pesos)
YearGDP1(1)Imports (2)Customs (3)Imports as Per Cent of GDP (4)Customs as Per Cent of GDP (5)Customs as Per Cent of Imports (6)
Sources: Col. 1: GDP for 1925–49 calculated from GDP in 1950 pesos and annual GDP deflators, Statistical Annex, Tables 4 and 32. GDP for 1950–66 from Banco de la República, Cuentas Nacionales, with some preliminary estimates. Col. 2: Contraloria General de la República, Dirección Nacional de Estadística, Anuario de Comercio Exterior (Bogotá, 1940), p. 3 (1930–36, f.o.b. plus 12 per cent); Joint Tax Program, OAS/IDB, op. cit., p. 161; Data from Colombian Treasury. Col. 3: Includes surcharges but excludes export duties. 1925–43 from Estadística Fiscal y Administrativa, 1942–1943 (Bogotá, 1945), p. 3; 1944–48, United Nations, Department of Economic Affairs, Public Finance Information Papers: Colombia, ST/ECA/SER.A/2, March 1950, pp. 28–29; 1949–61, Joint Tax Program, OAS/IDB, op. cit., p. 161; 1962–66, data from Colombian Treasury.

This erosion of revenues accompanying the shift in import composition reflects in part the difficulty of defining and administering distinctions among intermediate goods according to their different end-uses. It is also the product, however, of growing pressures to protect the new manufacturing or assembling industries. Concern with development, indeed, generates a number of policies which may serve to detract from the Government’s effective use of the customs house as a revenue instrument.

Some of these policies are reflected in the growing exemption of imports from customs duties. This growth of customs exemptions was not a gradual process. After remaining at about 10 per cent of all customs due from 1941 through the mid-1950’s, the value of exemptions rose quite rapidly and was from 30 per cent to 40 per cent of the total by the early 1960’s (see Table 21).

Table 21.Colombia: Effects of Import/GDP Ratio, Composition of Imports, Customs Exemptions, and Level of Customs Rates Upon Customs Receipts as Per Cent of GDP, 1925–64(As per cent of 1925–29 average; 1925–29 average = 100)


Effect of Import


(at 1959 rates)

Per Cent of Customs

Duties not Exempted

Effect of Customs Rate

Levels (Residuals)



Sources: Col. 1: Table 20. Col. 2: Table 19. Col. 3: DANE, Boletín Mensual de Estadística (Bogotá), September 1966, p. 39; Ministerio de Hacienda y Crédito Público, Memoria de Hacienda (Bogotá, 1946), Vol. I, p. 123; and Contraloría General de la República, Dirección Nacional de Estadística, Anuario de Comercio Exterior (Bogotá, 1943), p. xx. Col. 4: Calculated from Cols. 1, 2, 3, and 5. Col. 5: Table 20.

The purposes of these exemptions tell something of the policy considerations with which the Colombian Government has been concerned. The special importation of agricultural products and supplies by customs-exempt public or private agencies accounted for over half of all exemptions both in 1941, when it represented almost 7 per cent of all customs due, and in the late 1950’s, when it was over 15 per cent of all duties (see Table 22). As these special imports leveled off in the early 1960’s, considerably more customs exemptions were granted for aid shipments from the U.S. Government, which rose to over 10 per cent of all duties, and for imports of the public power companies, financed largely with foreign credits, which rose to over 5 per cent of all customs due.

Table 22.Colombia: Customs Exemptions as Per Cent of Total Customs Due, 1941 and 1958–65
Importing Entities194119581959196019611962196319641965
Agriculture and supply entities6.8616.4014.9811.4210.528.353.879.744.00
U.S. aid4.773.194.676.4712.4911.349.963.21
Power companies0.120.591.223.493.395.485.016.25
Government steel and petroleum entities1.301.521.490.871.931.291.021.65
National railroads0.390.030.130.580.520.320.171.830.71
Private enterprises0.
National Government ministries0.713.721.602.542.463.195.862.423.58
Sources: Contraloria General de la República, Dirección Nacional de Estadística, Anuario de Comercio Exterior (Bogotá, 1943), pp. 654–59; DANE, Boletín Mensual de Estadística (Bogotá), October 1959, p. 39, and September 1966, pp. 39–40.

Imports by private entities under investment incentive statutes or special concessions were of relatively less importance, amounting to from 1 per cent to 3 per cent of total customs due in most years but still substantially above the 0.3 per cent recorded for 1941. Customs exemptions on imports by the National Government itself rose from 0.7 per cent of all customs duties in 1941 to over 5 per cent because of heavy imports by the War and Public Works Ministries in the mid-1950’s and averaged about 3 per cent after 1958. While some exemptions, such as those for goods imported under foreign credits, represented net additions to the import flow (without which customs duties would not have been paid in any instance), others reflect the convergence upon the customs house of a variety of policy objectives alongside which the sacrifice of customs revenues was judged of secondary importance. All these elements, however, point to a shift in the role of imports and government in Colombia, for which the exemption of 30 per cent to 40 per cent of all customs dues is but the outward manifestation.56

A final element affecting the adequacy of customs house coverage of the hypothetical sales tax base is the actual level of tariff rates on each item. Because the Colombian tariff schedule was primarily specific in character until 1951 and did not become entirely ad valorem until 1965, the level of rates as a percentage of import value was adversely affected over the years by any rise in prices. No measure is available of the arithmetic average of all customs rates in Colombia during this period (as distinct from an average weighted by the actual composition of imports), and its meaning would in any case be of problematic significance. Movements in this average level of rates, however, would have considerable importance as a measure of the impact on specific customs duties of price changes and tariff reforms. With the data on imports, customs collections, customs exemptions, and import composition (and the assumptions on which this item is based), it is possible to calculate an index approximating movements in the average level of rates for the period 1925–64 57 (see Table 21). This calculation indicates that with the notable exception of the depression years, when prices declined, the average level of rates has followed a cyclical pattern of its own in which the erosive effect of rising prices on specific duties is corrected by periodic tariff reforms. As a result, the level of rates has varied from some 40 per cent below the average of 1925–29 to about 50 per cent above that average. By 1964, the average level was in the upper half of this range (137 per cent of the 1925–29 average), having declined somewhat since the tariff reform in 1959. The rate level served to offset, therefore, some of the negative effects on customs house operations that were exerted by shifts in import composition and by customs exemptions.

A combination of many factors, therefore, over the years reduced effective customs house coverage of the hypothetical sales tax base. While the import level as a percentage of GDP showed only cyclical variations around a rather stationary long-run trend, this stable percentage of GDP represented a declining portion of the hypothetical sales tax base. At the same time, the percentage of the import value that was collected in customs duties declined by 45 per cent between the late 1920’s and 1964, dropping from 22.5 per cent to 12.5 per cent (see Table 20), reflecting heavy exemptions and a shift in composition that was only partially offset by increased rate levels (see Table 21).58 The deterioration in the adequacy of customs coverage of the hypothetical sales tax base may be gauged on the basis of the simplifying assumption that the average effective customs rate for all imports applies to the import component of the sales tax base as well. On this basis, it may be concluded that whereas the customs house collected a tax of 16.5 per cent on the hypothetical sales tax base in 1925–29, it collected a levy of only 6.3 per cent in the period 1960–63.

Accessibility of the domestic base

While the import component of the hypothetical sales tax base was diminishing in relative importance, changing in composition, and becoming more difficult to tax effectively at the customs house, changes were taking place also in the domestic component.

Growth of the domestic base was accompanied by important changes in its composition.59 At the earlier stages of development, manufacturing production was concentrated in the few major industries that were able to find a sufficiently large domestic market. Some 64 per cent of all manufacturing production in 1940, for example, was concentrated in six industries: textiles, sugar, beer, cement, tobacco, and liquor (see Table 1). While convenient for administering a possible tax, the concentration of the early industries on basic consumer goods made them extremely regressive and open to a great deal of popular opposition, as the Colombian Government discovered in 1942. Sumptuary grounds provided a justification for taxing alcoholic beverages and tobacco, however, and these alone accounted in 1940 for 37 per cent of domestic manufacturing production.60 A considerable portion of the domestic component of the hypothetical sales tax base, therefore, was within relatively easy administrative reach through the simpler expedient of an excise tax. With the growth of domestic industry, however, these excises, like customs, became less and less adequate in their coverage. By 1963 the value of alcoholic beverage and tobacco output had declined to only 10 per cent of Colombia’s total manufacturing production (see Table 23).

Table 23.Colombia: Composition of Manufacturing Production, 1940 and 1963(Value in thousands of pesos)
Value of ProductionPer Cent of Total

Manufacturing Production
Clothing and textiles40,0513,904,80621.119.6
Beer and distilled liquors41,8021,418,40522.17.1
Other beverages4,190526,5682.22.6
Cement products and other nonmetallic minerals6,908952,7183.64.8
Cigars and cigarettes28,211505,73114.92.5
Metal and machinery manufactures5,6021,988,6642.910.0
Chemical and pharmaceutical products5,0621,697,0942.78.5
Leather and leather products7,326263,3783.91.3
Rubber products590542,2510.32.7
Paper products1,500612,7810.83.1
Other manufactures8,3572,829,40714.414.2
Sources: DANE, Anuario General de Estadística, 1940 (Bogotá, 1941), p. 52, and Anuario General de Estadística, 1963 (Bogotá, 1965), pp. 612 and 662–84.

As the domestic manufacturing base broadened beyond the coverage of these excises, the structure of domestic industry and marketing took on greater relevance as determinants of more general government access to the hypothetical sales tax base. It is of some significance, therefore, to determine what changes occurred in the criteria most conducive to greater accessibility of the base: the size of manufacturing, wholesale, and retail establishments and the concentration of production or trade in relatively few units.

Information on the pattern of production in Colombian manufacturing industry indicates a marked polarity in the size of firms and a high degree of concentration. While a multitude of small establishments was present in most industries, a sizable fraction of total production was in the hands of a few big firms that were considerably larger than the others. In 1940, more than half of total manufacturing employment was concentrated in 5 per cent of the firms—those employing 100 or more workers—while only 9 per cent of employment was distributed among more than 60 per cent of the establishments—those having less than 10 employees (see Table 24). In the early 1950’s one survey of Colombian manufacturing industry recorded 3 large, heavily capitalized cotton textile companies; 3 establishments providing nearly one third of national productive capacity of woven fabrics; 3 fique mills absorbing nearly 60 per cent of the national production of fique fiber (for sacking); 170 soap manufacturers, of which 12 firms produced 52 per cent of the total output; 2 companies producing over half the national output of rubbersoled shoes; 1 company accounting for 40 per cent of national leather production; only 6 of the 603 footwear manufacturers equipped to produce 100,000 or more pairs a year; and 1 firm supplying most of the cigarette output.61

Table 24.Colombia: Distribution of Manufacturing Establishments by Number of Workers Employed, 1940 and 1962
Number of Workers

per Establishment
Number of EstablishmentsPer Cent of TotalNumber of EmployeesPer Cent of Total
Less than 109776,81863.461.53,29032,8708.911.9
100 or more794715.14.320,259144,03354.852.1
Sources: DANE, Anuario General de Estadística, 1940 (Bogotá, 1941), p. 39, and Anuario General de Estadística, 1962 (Bogotá, 1964), p. 760.

The course of development brought little change in this pattern. Production by artisans declined from 27 per cent of the manufacturing plus artisanal total in 1925 to 20 per cent in 1940 and to 13 per cent in the early 1960’s,62 reducing one obstacle to accessibility of the domestic portion of the hypothetical sales tax base. But the average size of establishments in the manufacturing industry and the concentration of employment among them hardly changed at all. By 1962 the percentage of firms with less than 10 employees was still 61.5 per cent, virtually the same as the 63.4 per cent in 1940, and their share of total employment was 12 per cent, compared with 9 per cent in 1940 (see Table 24). A similar stability marked the share of the largest firms. This pronounced concentration of output is reflected also in actual collections of the sales tax, data for Bogotá payments in 1965 showing that two thirds of the taxpayers accounted for only 5.5 per cent of total payments and that 20 taxpayers accounted for 47 per cent (see Table 9). Quite apart from the over-all growth of domestic manufacture and the expansion of industries that were not covered by sumptuary excises, therefore, development brought no change whatever in the size distribution of Colombian manufacturing enterprises nor, consequently, in the accessibility of the hypothetical sales tax base at the manufacturers’ level.

Information on any changes in the accessibility of the hypothetical sales tax base at the wholesale level over this period is quite meager in comparison. It has been suggested on the basis of studies of a number of other countries, however, that changes in the channels of distribution follow a characteristic pattern during the development process. At the earliest stages, when markets are local and small, distribution channels are quite short. They grow longer, with more links or intermediaries in the chain, as development progresses because of the distances involved, the shortage of adequate capital to carry stocks through the distribution stage, and the manufacturers’ concentration on production alone. In later stages, “as producers gain financial strength, however, the need for separate functional middlemen declines, and the number of links in the channel correspondingly falls.” 63 In Colombia, with a manufacturing industry polarized between the few large, well-capitalized firms and the mass of small producers, the second and third stages of marketing development have probably coexisted for some time. Large manufacturers in many lines sell through agents, regional branches, or even retail outlets, while the small producers sell their output to traditional wholesalers.

The composition of the wholesale trading community in the period 1953–54, for which data are available, is quite the opposite of that in manufacturing. There is no profusion of small firms nor the concentration of sales in a few large establishments which characterizes manufacturing. Instead there are relatively few wholesalers—4,847, compared with 8,425 manufacturers 64—and the trade is more evenly distributed among them. While there is no information as to the persistence of this relationship, as long as it exists it has significant implications for the relative accessibility of the hypothetical sales tax base at the wholesale and manufacturing levels. Collection of a possible sales tax through the largest manufacturing firms could cover a substantial fraction of total manufacturing sales. Any attempt to approach complete—80 per cent or 90 per cent—coverage, however, would require collection efforts among the mass of small establishments as well. In a number of lines, consequently, collection at the wholesale level could provide more complete coverage of the base through a smaller number of firms. This is evidently the case for the textile, shoe, and clothing industry and for the furniture and appliances category, for which the number of manufacturing, wholesale, and retail firms necessary to afford 80 per cent or 90 per cent of total coverage have been calculated in Table 25 on the basis of Charts 7 and 8. Ninety per cent coverage of shoes, textiles, and clothing, for example, could be afforded by collections from 3,614 retailers, or 427 manufacturers, or 250 wholesalers.65

Table 25.Colombia: Retail, Wholesale, and Manufacturing Establishments Accounting for 80 Per Cent and 90 Per Cent of Sales or Value Added in Textiles, Shoes, and Clothing and in Furniture and Appliances, 1953–54
80 Per Cent of Total90 Per Cent of Total
Textiles, shoes, and clothing
Number of establishments2,1011841343,614250427
Per cent of all establishments25395435316
Furniture and appliances
Number of establishments4125417461275332
Per cent of all establishments213731325259
Sources: Based on data from the 1954 Census of Commerce and from the 1953 Census of Manufactures.

Chart 7.Colombia: Concentration of Manufacturing, Wholesale, and Retail Trade in Textiles, Shoes, and Clothing, 1953–54

Chart 8.Colombia: Concentration of Manufacturing, Wholesale, and Retail Trade in Furniture and Appliances, 1953–54

On the basis of the 1953–54 structure of manufacture and marketing, therefore, access to the domestic portion of the hypothetical sales tax base involved fewer firms at the wholesale level than at the manufacturing or retail levels. As long as distribution by a number of large manufacturing firms bypassed the wholesale level, however, coverage of such manufacturers’ sales to retailers would be necessary as well. 66 The complexity of this additional administrative problem would depend upon whether such vertically integrated operations were indeed restricted to the few, largest manufacturers. Over the long run, the effects of development on the accessibility of the hypothetical sales tax base at the wholesale level would appear to reflect the combination of two influences: the difficulties inherent in the proliferation of intermediaries and the benefits accompanying their subsequent reduction, and the complete bypassing of the wholesale community by the few large industrial firms present in Colombia throughout the period of development studied.

A far simpler trend seems to mark developments at the retail level. Although, as for wholesale trade, no information is available about changes in the structure of retail trade over the period 1925–65, there are rather clear indications of the direction that such changes followed. Studies of a number of other countries have suggested that as development progresses the average size of retail stores increases (along with a reduction in the amount of service provided and an increase in the variety of goods that each store carries).67 This implies increasing accessibility of the sales tax base at the retail level. While no data on changes in the size of retail establishments are available for Colombia, there is information on another feature of retail establishments which bears on their accessibility for sales tax collection—their operation as business, rather than as household, undertakings.

Professor John Due has suggested that a good indication of the ability to collect a sales tax may be gained from

the extent to which business activity at various levels is carried out by what may be called commercial undertakings, with the business financial transactions separate from those of the family, or by essentially household units, such as small-scale artisan production, very small retail establishments, and the like. 68

A measure of such household operation is available in Colombia’s 1954 census of commerce, which gives the number and total sales of retail establishments having no paid employees. While this census presents no data on movements over time, it does contain information on retail sales inside and outside the capital cities of the Departments. On the basis of statistics for 12 of Colombia’s 17 Departments, representing 65 per cent of all retail establishments and sales in the country, an analysis was made of data on such family stores. This analysis indicates that the main influence on the share of total sales channeled through family stores is not the level of per capita income or per capita sales but urbanization (see Tables 26 and 27). The percentage of total nonfood sales coming through the family stores (without paid employees) was 28.3 outside the departmental capitals but only 6.4 within. If urbanization is indeed a key determinant of this aspect of access to the hypothetical sales tax base at the retail level, then progress in this direction has probably been considerable over the past few decades. For the 12 Departments studied, the population residing in the departmental capitals rose from 10.6 per cent in 1918 to 21.5 per cent in 1951 and to 31.6 per cent in mid-1964.69 In the declining importance of the family store, therefore, and perhaps in the increase in the average size of a store, the progress of Colombian development has probably been accompanied by enhanced accessibility of the hypothetical sales tax base at the retail level.

Table 26.Colombia: Characteristics of Retail Sales by Department, Inside and Outside Departmental Capitals, 1954
DepartmentPer Capita Sales (pesos per year)Food and Beverage Store Sales as Per Cent of Total SalesPer Cent of Sales in Stores Without Paid EmployeesPer Cent of Nonfood, Nonbeverage Store Sales in Stores Without Paid EmployeesPopulation Distribution
Per cent outside capital
Number inside capitalIn municipios of under 10,000In municipios of under 20,000Inside capitalOutside capitalInside capitalOutside capitalInside capitalOutside capitalInside capitalOutside capital
Norte de Santander81012631.659.231.373.412.256.495,15053.983.8
Valle del Cauca87327741.255.319.737.53.616.7284,1869.839.9
All Departments 180214333.254.821.846.16.428.31,651,157
Sources: Data from the 1954 Census of Commerce and Services and the 1951 population census.
Table 27.Colombia: Distribution of Retail Sales Inside and Outside Departmental Capitals in Twelve Departments, 1954
DepartmentsInside CapitalsOutside CapitalsDepartmentsInside CapitalsOutside Capitals
(In thousands of pesos)(Per cent)
Total sales2,187,5851,324,763862,822100.060.639.4
Food sales917,959440,115477,844100.047.952.1
Nonfood sales1,269,626884,648384,978100.069.730.3
Sales without paid help686,238288,316397,922100.042.058.0
Sales with paid help1,501,3471,036,447464,900100.069.031.0
Nonfood sales without paid help165,23156,182109,049100.034.066.0
Nonfood sales with paid help1,104,395828,466275,929100.075.025.0
1951 population (people)7,675,0491,651,1576,023,892100.021.578.5
Per capita sales (pesos)285802143
Per capita food sales (pesos)12026779
Per capita nonfood sales (pesos)16553864
As per cent of total sales
Total sales100.0100.0100.0
Food sales42.033.255.4
Nonfood sales58.066.844.6
Sales without paid help31.421.846.1
Sales with paid help68.778.253.9
Food sales without paid help as per cent of all food sales56.852.760.5
Nonfood sales without paid help as per cent of all nonfood sales13.06.428.3
Source: Table 26.

III. Summary and Conclusion

A number of elements emerge from this examination of changes occurring in a hypothetical sales tax base—including all expenditures but food and services—during the course of development.

1. Data for personal consumption expenditures (PCE) in a cross section of countries suggest that because food declines from more than 60 per cent of PCE to less than 30 per cent as per capita national income rises, while services rise from less than 20 per cent of PCE to more than 35 per cent, the tax base—embracing the remaining, unexempt expenditures—follows a characteristic pattern. It increases from 27 per cent of PCE in the poorer countries (those having an average per capita national income equivalent to US$11670) to about 37 per cent in the richer countries (US$1,100). The income elasticity of the tax base is about 1.2 through the US$550 category, but it levels off and declines to below 1.0 at the top of the per capita income scale. The same trend characterizes changes in the tax base as a per cent of GDP, but this rises only half as steeply if only personal expenditures are covered by the tax base.

2. This trend in the hypothetical sales tax base as a per cent of GDP is found also in Colombia over a 40-year period in which per capita national income rose from US$125 to US$215. These measurements, however, are founded not on personal consumption patterns but on national production aggregates, with the tax base estimated on the basis of nonfood imports plus the value added in nonfood domestic manufacturing industry. Moreover, because Colombia was characterized by pronounced import fluctuations during this period, in order to find the trend it was necessary to remove such fluctuations by the use of a moving average. As a result, while per capita income increased from US$125 to US$215 over the 40-year period, the hypothetical sales tax base rose from 14 per cent of GDP to 25 per cent, at an income elasticity of about 1.3. The historical rise in the base, measured from the production side, is thus quite close to that suggested by the comparison of personal consumption patterns in different countries during the 1950’s.

3. Calculations on the possible incidence of a sales tax on this base, assuming that there was a forward shifting to consumers, were made on the basis of expenditure data for 1953 for the various income brackets of two urban groups: working class families and middle-class employee families. Expenditures for services were so important to upper income families at this stage of development that they offset fully the tax exemption on food that was granted for equity reasons to relieve the burden on lower income families. As a result, while the tax base rose more than proportionately with respect to total expenditures for working class families, it was roughly proportional, or slightly less than proportional, to total expenditures for middle class families in successively higher income brackets. A comparison of the taxable percentage of expenditures for working class and middle class families in each income bracket showed that the tax base was smaller for workers than for employees in each bracket that was below 500 pesos a month because of the larger food expenditures by workers and smaller for employees than for workers in each bracket that was above 500 pesos because of the larger expenditures for services by employees. For both groups, because of substantial dissaving in the lower income brackets and substantial saving in the upper brackets, the tax base was regressive with respect to income.

This regressivity of the base was reduced by the effects of the differentially higher rates applied to selected products under the sales tax in 1965. Distribution of both the tax base and the differential rates varies considerably over the course of import fluctuations, however. Because automobiles and other imported durable goods purchased by wealthier families have become available only with the relaxation of import controls every 6 years or so, a sales tax would be less regressive in years when imports and purchases of durables were at their peak but more regressive in other years, as is evident in the shifting rate structure of collections between 1965 and 1966.

4. To evaluate the (administratively simpler) alternative of taxing the hypothetical sales tax base through the customs tariff, the adequacy of customs coverage of this base was examined. With some cyclical variations, the adequacy of customs coverage of the sales tax base deteriorated substantially over this period. As domestic industry grew, the portion of the tax base that passed through the customs house (nonfood imports) declined from 73 per cent in the late 1920’s to 40 per cent in the early 1960’s. Though imports, aside from cyclical fluctuations, remained at about the same level of GDP, the domestic component of the base (estimated on the basis of value added by nonfood domestic manufacturing industry), rose from 4 per cent of GDP to 15 per cent.

The customs house encountered difficulties also in taxing that part of the base that still passed through it. These difficulties reflected structural changes in the economy, leading to a shift in import composition and the adoption of government policies to promote further development. The shift—enforced by import controls—away from higher duty consumer goods and toward lower duty raw materials and capital goods is estimated to have reduced customs revenues by about 30 per cent over the 40-year period. The exemption of imports from customs duties rose sharply from about 10 per cent of all customs due in the 1940’s and early 1950’s to between 30 per cent and 40 per cent in the early 1960’s. Relatively little of this increase represented exemptions granted to private industry, the greater portion being for special imports of agricultural and other supplies and later for foreign aid, public power companies, and government ministries. The level of customs duties varied widely over the period, as the specific rates were eroded by price increases and were adjusted periodically through major customs reforms. By 1964, however, average rates on individual items were probably about 25 per cent above their 1925–29 average.

Under the influence of shifting composition of imports and the exemption of imports from the payment of customs duties, which was only partly offset by changing rates, the percentage of import value collected in customs duties declined from 22.5 per cent in the late 1920’s to 12.5 per cent in 1964. With the accompanying decline in the portion of the base that passed through customs, it may be estimated that the tax of 16.5 per cent which had been collected on the hypothetical sales tax base by the customs house in the late 1920’s had shrunk to 6.3 per cent by the early 1960’s.

5. Developments in the Colombian economy over the past 40 years brought changes also in the accessibility of the domestic portion of the hypothetical sales tax base at the manufacturing, wholesale, or retail level. The changing composition of domestic industry reduced the adequacy of sumptuary excise taxes as a possible alternative means of taxing the base, as beverages and tobacco declined from about 40 per cent of manufacturing production earlier in the period to only 12 per cent by 1962.

For domestic production as a whole, the importance of hard-to-tax production by artisans declined during this period. To the extent that larger establishments and the concentration of sales in fewer hands tend to increase the accessibility of the base to possible collection of a tax, however, it is not at all clear that progress has been made at the manufacturing level. Most Colombian manufacturing industries are characterized by a high degree of concentration—with a multitude of small firms at one extreme and a few large firms, producing perhaps one third to two thirds of total output, at the other. Over the past two decades, neither the 60-odd per cent of all firms that have less than 10 workers nor the 50-odd per cent of employment which is concentrated among the 5 per cent of firms that have 100 or more workers has changed at all. To extend coverage beyond perhaps two thirds of the hypothetical sales tax base in most lines of manufacturing, the collection of a tax at the manufacturers’ level would have to include not only the largest firms but also the mass of small manufacturing establishments.

In the wholesale trade, data for 1954 show far fewer wholesalers than manufacturers in Colombia and the volume of wholesale sales to be more evenly distributed among them. The experience of other countries leads one to expect a lengthening of trade channels—and a proliferation of hard-to-tax intermediaries—during most of the development process followed by a shortening of the channels as producers, and others, amass capital and integrate vertically into distribution. The contrasting extremes of Colombian manufacturing industry, however, apparently brought about a coexistence of these two stages. Some of the larger manufacturing firms have long maintained their own distribution channels of agents and branches, bypassing independent wholesalers. On the basis of data for 1953–54 for wholesalers and manufacturers, therefore, while coverage of the hypothetical sales tax base could probably be achieved through fewer firms if collected at the wholesale level rather than at the manufacturers’ level, it would require supplementary coverage of those, presumably large, manufacturing firms that bypass the wholesale level.

Among retail stores, aside from any increase in size which experience in other countries would suggest accompanies development, there was apparently considerable progress in Colombia in another factor that promotes accessibility—the declining importance of the family store. Data from Colombia’s 1954 census of commerce indicate that the importance of sales at nonfood family stores depended not upon differences in per capita income but upon urbanization. The strong movement of population to Colombia’s cities over the past decades, accordingly, may reasonably be judged to have reduced the percentage of nonfood retail sales through family stores and to have moved the tax base closer to accessible retail collection.

For Colombia, the sum of detailed findings on all aspects of the hypothetical sales tax base over the past 40 years suggests that an increase occurred in the possibilities for a sales tax and a decline in the adequacy of customs and sumptuary excises. Aside from Colombia, if a general lesson is to be learned from this study, it is that changes in the course of economic development give rise to new tax possibilities and open gaps in the once-adequate coverage of existing taxes. In any developing country, therefore, evaluation of the tax system must be based not only upon careful examination of the tax system itself but also upon its correspondence to changing conditions in the underlying economy. This paper indicates that, in evaluating the underlying economy, examination of the base for each potential tax can provide a useful means of analysis.

Incidence du développement économique sur l’assiette de la taxe sur le chiffre d’affaires : une étude sur la Colombie


Pour rechercher les particularités du développement propres à influer sur les possibilités d’application d’un impôt et en examiner l’évolution aux divers stades de ce développement, il convient, estime l’auteur, de faire porter principalement l’analyse économique sur l’assiette de chaque impôt et son évolution, étant donné que la matière imposable peut exister alors même que l’impôt n’est pas prélevé. Partant de ce point de vue, l’article examine en Colombie l’évolution, entre 1925 et 1965, de l’assiette d’une hypothétique taxe sur le chiffre d’affaires dont seraient restés exonérés l’alimentation et les services. Si pour mesurer la matière imposable on fait la somme de la valeur globale des importations de produits non comestibles et de la valeur ajoutée par l’industrie nationale de transformation non alimentaire, on s’aperçoit qu’au cours de ces années elle a passé de 14 à 25 pour cent du produit intérieur brut (PIB), avec une élasticité-revenu de 1,3. A titre de comparaison, on trouve pour les années 50 une élasticité-revenu de 1,2 calculée par groupes de pays, la matière imposable étant calculée en pourcentage des dépenses de consommation des ménages. Selon une enquête effectuée en 1953 sur les dépenses des families urbaines en Colombie, l’incidence de la taxe hypothétique est à peu près proportionnelle aux dépenses, mais régressive par rapport au revenu, car la place plus grande tenue par les services dans les dépenses des families aisées compense tres largement l’exonération relativement plus importante des denrées alimentaires chez les families de revenu modeste. De 1925 à 1965, les possibilités de perception indirecte de la taxe par voie douanière sont tombées, en termes de l’équivalent d’une taxe, de 16,5 à 6,3 pour cent de la matière imposable. Cette baisse est due à la progression de l’industrie manufacturière non alimentaire qui est passée de 4 pour cent à 15 pour cent du PIB (les importations de denrées non comestibles représentant une part constante du PIB), au fait que la part des produits exonérés est passée à 30 pour cent des importations, aux modifications de la composition des importations qui ont réduit les recettes douanières de 30 pour cent. L’augmentation moyenne de 25 pour cent des tarifs douaniers n’a compensé qu’en faible partie l’effet de ces divers éléments. L’accessibilité de la matière imposable à une éventuelle perception de la taxe ne s’est guère modifiée au niveau de l’entreprise industrielle (la distribution des entreprises manufacturiéres d’après leur grandeur restant inchangée); elle a progressé au niveau du commerce de détail (l’urbanisation entraînant une diminution de l’importance des commerces familiaux); au niveau du commerce de gros, elle s’est probablement améliorée.

Los efectos que el desarrollo económico surte en la base imponible de un impuesto a las ventas: un estudio sobre Colombia


A fin de determinar qué características del desarrollo influyen sobre las posibilidades de tributación y de qué modo dichas características varían en diferentes etapas del desarrollo, el autor propone que en los análisis económicos se concentre la atención en los cambios que ocurran en la base imponible de cada impuesto, ya que la base puede existir aunque no se la grave con impuesto alguno. Este trabajo procede así con respecto a la base de un impuesto hipotético sobre las ventas, del cual se excluyen los alimentos y los servicios, examinando las alteraciones ocurridas en dicha base en Colombia durante el período 1925–65. La base imponible calculada como la suma del valor de las importaciones de artículos no alimenticios y del valor agregado por las manufacturas no alimenticias producidas en el país fue aumentando del 14 por ciento del producto interno bruto (PIB) hasta constituir el 25 por ciento, con una elasticidad ingreso de 1,3. Esta última es comparable con la elasticidad ingreso de 1,2 representada por la misma base imponible calculada como porcentaje de los gastos de consumo personal en el caso de una selección representativa de países en los años de la década 1950. Utilizando una encuesta de 1953 sobre los gastos realizados por las familias urbanas de Colombia, la incidencia del impuesto hipotético resulta aproximadamente proporcional a los gastos pero regresiva con respecto al ingreso, ya que el hecho de que fuesen mayores los gastos por servicios efectuados por las familias más pudientes contrarrestó con creces el que las exenciones por alimentos fueran mayores en el caso de las familias de situación inferior. Durante el período 1925–65 disminuyó el grado de suficiencia de los derechos de aduana como sustitutivos de un impuesto sobre las ventas, pasando a ser sólo el 6,3 por ciento de la base del impuesto hipotético en vez del 16,5 por ciento. Ello se debe a que la fabricación de productos no alimenticios en el país se elevó del 4 por ciento del PIB al 15 por ciento (el porcentaje de las importaciones no alimenticias que forma parte del PIB permaneció inalterado), al hecho de que las exenciones a la importación se elevaron al 30 por ciento, a un cambio en la composición de las importaciones que redundó en que las recaudaciones aduaneras disminuyeron en un 30 por ciento aproximadamente, y a un aumento del 25 por ciento en el tipo medio de los derechos de aduana, el cual actuó como contrapeso. Las posibilidades que la base ofreció para la percepción de impuestos permanecieron igual que antes al nivel de los fabricantes (dado que la distribución de empresas según magnitud no varió), mejoraron al nivel de los minoristas (puesto que la urbanización reduce la importancia de los pequeños establecimientos de familia), y al nivel del comercio mayorista probablemente también mejoraron.

Jonathan Levin, Senior Economist in the Fiscal Affairs Department, was educated at Columbia University, the Fletcher School of Law and Diplomacy, and the University of Strasbourg (France). He has held positions at the International Tax Program of Harvard University, the Federal Reserve Bank of New York, and the U.S. Securities and Exchange Commission. He is the author of The Export Economies.

Harley H. Hinrichs, A General Theory of Tax Structure Change During Economic Development (International Tax Program, Harvard Law School, Cambridge, Massachusetts, 1966), p. 99.

Professor John Due defines a sales tax as “a levy imposed upon the sales, or elements incidental to the sales, such as receipts from them, of all or a wide range of commodities.” John F. Due, Sales Taxation (University of Illinois Press, Urbana, 1957), p. 3.

Laws No. 126 of 1914 and No. 69 of 1917.

One description of Colombian commerce and industry in the early 1920’s stated: “There has been very little industrial development in Colombia and there are only a few manufacturing plants. In the Atlantic coastal region there are several sugar mills of importance, two textile mills, and a meat-packing plant; and at Cali a textile mill, while throughout the country are found small breweries, tanneries, flour mills, and shoe factories which cater to the local demand.” U.S. Department of Commerce, Colombia, Commerce and Industries 1922 and 1923, Trade Information Bulletin No. 223 (Washington, 1924), pp. 15–16.

C.A. McQueen, Colombian Public Finance (U.S. Department of Commerce, Washington, 1926), pp. 8, 24, and 29.

U.S. Tariff Commission, Mining and Manufacturing Industries in Colombia (Washington, 1949), p. 11.

Banco de la República, Revista (Bogotá), October 1942, pp. 397–403, and December 1943, p. 483.

This section has benefited from my reading of Richard M. Bird’s Sales Taxation in Colombia: Tax Policy and Development Planning (The Center for International Affairs, Report No. 36, Development Advisory Service, Economic Development Series, Harvard University, Cambridge, Massachusetts, 1966).

The sales tax provisions embodied congressional reshaping of a Finance Ministry request for the combination of a personal expenditure tax and a special manufacturers’ excise tax on luxury items (Carlos Sanz de Santamaría, Memoria de Hacienda presentada al Congreso Nacional de 1963 (Bogotá, 1963), Tomo Segundo, pp. 35–36, y Tomo Tercero, p. 57.

Article 204 of the Constitution provides that “No indirect tax … shall take effect until six months after the promulgation of the law establishing the same,” but Article 205, which was in effect at the time, adds “This provision … shall not limit the extraordinary powers of the Government when it is invested with them.” Cited in Seymour W. Wurfel, Foreign Enterprise in Colombia: Laws and Policies (University of North Carolina Press, Chapel Hill, 1965), p. 411. The delay has been attributed to the desire to have the sales tax take effect after the expiration of the income tax surtax that was voted for 1963 and 1964.

Added by Decree No. 1595 of June 24, 1966, to prevent evasion through exchange of taxable goods for exempt goods subsequently sold with no tax paid.

Primarily as follows: at 10 per cent—jewelry, automobiles, tobacco, playing cards, and imports of liquor, tableware, or canned foods; at 8 per cent—radios, television sets, phonographs, records, small electric appliances, cameras, films, and arms; at 5 per cent—cosmetics, pens, large electric appliances, trucks, motorcycles, sports equipment, some textiles, and domestic liquors and tableware; at 3 per cent—all remaining articles not specifically exempt.

Departamento Administrativo Nacional de Estadística (DANE), Boletín Mensual de Estadística (Bogotá), April 1966, pp. 57 and 73.

Decree No. 1595 of June 24, 1966. Domestic cigarettes remained at 10 per cent, however, while the rate on locally made tableware was reduced from 5 per cent to 3 per cent and that on locally made chewing gum was reduced from 10 per cent to 3 per cent.

Decree No. 1595 of June 24, 1966, and Decree No. 1881 of July 25, 1966.

Decree No. 2807 of November 12, 1966.

Decree No. 1881 of July 25, 1966.

Also, importers who act as agents or representatives for firms abroad were authorized to pay the sales tax at customs on the base calculated as the import value, plus customs charges, plus a 15 per cent markup.

During 1965 the tax was collected in some Departments. By Decree No. 2073 of August 6, 1965, however, the tax was to be collected and retained by the Departments and used for public health and education. Subsequently, Decree No. 1595 of June 24, 1966 completely exempted the products of departmental monopolies.

DANE, Anuario General de Estadística, 1962 (Bogotá, 1964), pp. 756–62.

Ministry of Finance, Division of National Taxes.

Reed R. Hansen, “An Empirical Analysis of the Retail Sales Tax With Policy Recommendations,” National Tax Journal, March 1962, pp. 5–6 and 12.

United Nations, Economic Commission for Latin America (ECLA), Analyses and Projections of Economic Development, III: The Economic Development of Colombia (Geneva, 1957), p. 26.

This classification remains inexact to the extent that detail was not available to permit separation of the taxable goods included in the nontaxable household operation, recreation and amusement, and fuel and light categories or the tax-exempt medicine and services included in the taxable personal care and health category.

Robert D. Stevens, Elasticity of Food Consumption Associated With Changes in Income in Developing Countries (U.S. Department of Agriculture, Foreign Agricultural Economic Report No. 23, Washington, March 1965), pp. iii, 18, and 24–25.

This may be shown as follows:

Let C = total consumption, and E = exemptions

then EC will decrease untill EC=ΔEΔC

Now E = F + S

therefore, at the turning point,


Also, if unitary elasticity of total consumption is assumed,


where eF and eS are the income elasticities of consumption for food and services, respectively.

Substituting in (1), and rearranging,


Thus, S—eSS = eFF –F

so that SF=1eFeS1 at the turning point.

Simon Kuznets, “Quantitative Aspects of the Economic Growth of Nations: VII, The Share and Structure of Consumption,” Economic Development and Cultural Change, Vol. X, No. 2, Part II (January 1962), pp. 76–77.

Following Kuznets, op. cit., p. 25, elasticity is taken as the ratio of the percentage change in the tax base to the percentage change in per capita income or expenditures (both percentages to the base of the geometric mean of initial and reference values).

The export of nonfood manufactures from Colombia was not significant during the period under study.

Whether additions to inventory are promptly included in the tax base depends on whether the sales tax is collected at customs or at subsequent wholesale or retail sale. By measuring the tax base to include imports, it is assumed in the discussion that follows that the tax is collected at customs or that the lags which would otherwise ensue are not significant for the points discussed.

Apparently the shortage of necessary capital goods imports and intermediate parts prevented domestic industry from growing to replace cutoff foreign supplies during these periods.

For the tax base, a 10-year moving average was used between 1925 and 1945, a 7-year average for 1946 to 1951, and a 6-year average thereafter. For per capita GNI, a 10-year average was used for 1925 to 1946, a 7-year average to 1951, and a 6-year average thereafter.

GDP data were adjusted in line with the 15.77 per cent shortfall of the ECLA series below the central bank’s series.

This finds interesting comparison in the work of David George Davies, who examines data for state sales taxes in the United States (some of which cover food and selected services) and concludes that the data “reject the hypothesis that sales taxes have been inelastic with respect to income” in 25 of the 28 States analyzed. “The Sensitivity of Consumption Taxes to Fluctuations in Income,” National Tax Journal, September 1962, p. 290. See, however, the discussion above on the decline in the income elasticity of the tax base at higher income levels, pages 51–52.

For a summary of the arguments, see Daniel C. Morgan, Jr., “Reappraisal of Sales Taxation: Some Recent Arguments,” National Tax Journal, March 1963, pp. 97–100.

For a discussion of the broader questions of incidence, taking into account the replacement of other taxes and the distribution of benefits from government expenditures, see Richard A. Musgrave, The Theory of Public Finance: A Study in Public Economy (New York, 1959), pp. 205–31.

There is no information on the composition of expenditures by business units. Since these expenditures are taxable only for materials or equipment not physically incorporated in the final product, they may be unevenly distributed among manufacturing industries, bearing most heavily on those products whose manufacture is most heavily capitalized.

DANE, Economía y Estadística (Bogotá), November 1958, p. 35.

Kuznets attaches some significance to the pressures upon employees, particularly at the higher levels, to maintain a respectably appropriate scale of living as a means to success and attributes to this a lower savings propensity for employees than for independent entrepreneurs. Kuznets, op. cit., p. 51. Working class families may very well be free of these pressures.

Harold G. Vatter and Robert L. Thompson, “Consumer Asset Formation and Economic Growth—the United States Case,” The Economic Journal, June 1966, p. 322, fn. 1.

Reed R. Hansen, op. cit., pp. 1–13.

J.A. Nicholson, “Variations in Working Class Family Expenditure,” in Journal of the Royal Statistical Society, Series A, Pt. IV (1939), pp. 359 ff., and Frank Lorimer and Herbert Rohrback, “Economics of the Family Relative to Number of Children,” Milbank Memorial Fund Quarterly, XVIII, No. 2 (April 1940), both cited in Faith Williams, “International Comparisons of Patterns of Family Consumption,” p. 275 in Lincoln H. Clark, ed., Consumer Behavior, Research in Consumer Reactions (New York, 1958).

Milton Friedman, A Theory of the Consumption Function (Princeton University Press, 1957), pp. 50–52 and 57.

It should be noted also that data for income, and for savings, in the 1953 expenditure survey are not of the same high quality as those for expenditures. It was felt that any attempt to verify all income items and to reconcile them with the carefully verified account of family expenditures might have led the interviewers to adjust expenditure items as necessary to achieve a fictitious balance. DANE, Economía y Estadística (Bogotá), November 1958, p. 74.

In the Colombian survey, daily expenditures were recorded in detail for a month on the basis of daily visits. Expenditures by category were recorded from memory for the extensive period of 1 year and for a supplementary period of six months, during which monthly visits were made to the home. DANE, Economía y Estadística (Bogotá), November 1958, pp. 9 and 69.

The treatment of durable consumer purchases would vary also according to whether purchases were financed from past savings or through installment payments, which might be allocated by the expenditure survey to the category of goods originally purchased. One would not in any case expect the installment financing of automobile purchases to extend the full 6 years between purchases, so that the later years of a cycle would find purchases of such durables absent or underrepresented.

This would depend, of course, upon the effect of such expenditures on remaining expenditures. A historical analysis of the U.S. demand for durable consumer goods concludes that “the rising demand for the newer types of durable products in the ’twenties and ’fifties occurred within each period at the relative expense of outlays for other consumer goods, both durable and non-durable….” Harold G. Vatter and Robert L. Thompson, op. cit., p. 321.

For each of the seven cities, almost 700 consumption items were classified according to the rate applicable under the 1965 sales tax. The average tax rate for each of 15 major expenditure categories was then calculated for each city and was applied to the expenditures shown for each income bracket in each city—which were published in terms of major expenditure categories rather than in itemized detail. The resulting tax was then calculated as a per cent of average total expenditures in each income bracket and as a per cent of income at the middle of the bracket. The result for each of the seven cities was then weighted by the appropriate weights to obtain the national average for each income bracket.

Both the level and progressivity of the rates may be somewhat understated because, lacking adequate detail in the itemized list of consumer expenditures, some of the items subject to the higher tax rates, particularly imported articles, were aggregated with other articles at the lower rates. This, however, may parallel the practices of some taxpayers as well.

It must be noted that the tax collection data cover some business expenditures while 1953 survey data are restricted to personal consumption expenditures. Business expenditures, which would generally be in the 3 per cent category, were estimated by Richard Bird to account for perhaps 10 per cent of the 1965 sales tax base.

Joint Tax Program of the Organization of American States and the Inter-American Development Bank, Fiscal Survey of Colombia (Baltimore, 1965, hereinafter referred to as Joint Tax Program, OAS/IDB), pp. 227–28.

Using data from DANE, Anuario General de Estadística, 1940 (Bogotá, 1941), pp. 45–49 and 52; from Consejo Nacional de Política Económica y Planeación, Departamento Administrativo de Planeación y Servicios Técnicos, Plan General de Desarrollo Económico y Social, II Parte, Industria, 1962, p. 33; and from the Finance Ministry.

See above, pages 53 and 68–70, and the author’s “El Ciclo de las Importaciones y la Política Fiscal en Colombia,” Banco de la República, Revista (Bogotá), June 1967, pp. 742–47.

Nassau A. Adams, “Import Structure and Economic Growth: A Comparison of Cross-Section and Time-Series Data,” Economic Development and Cultural Change, Vol. 15, No. 2, Part 1 (January 1967), pp. 143–62.

See also the analysis of Colombian customs exemptions in Joint Tax Program, OAS/IDB, op. cit., pp. 164–70.

This is based on the following relationship: C = Customs collections, M = Imports, L = Arithmetic average level of rates, U = Percentage of customs unexempted, and W = Weighted average rate for all imports.

Now, by definition,

Ct=MtWtUt, and


where ri is the tariff rate of the item i, and Wi is the weight of this item in total imports, and t refers to the year for which Wt is calculated. In Table 19, a calculation of the weighted average of tariff rates in 1959 (rio) using current weights for each year t was presented. Let this average be Wt*


It will be assumed that the level of rates in the different import categories changed proportionately and that no significant shift occurred in the mix of imports within each category, so that Lt moves in correspondence with Wt/Wt*.

Letting letters with a bar stand for index values (base 1925–29 average) and substituting LtWt* for Wt


The individual components are presented in Table 21. In calculating Lt, for lack of data on exemptions, customs collections are taken to be 100 per cent of customs due for the years before 1940.

Although this deterioration was reversed by the tariff reform of 1965 and the cyclical relaxation of import restrictions in 1966, there are indications that this may be only a temporary interruption.

While manufacturing output is used here to approximate several characteristics of the domestic component of the tax base, it must be noted that this output differs from the actual domestic component by the amount of imported materials incorporated in the value of output, which may vary from industry to industry.

For a description of alcoholic beverage consumption in Colombia, see Bernard Romero Rojas, “El Consumo de Bebidas Alcohólicas en Colombia,” in DANE, Economía y Estadística (Bogotá), May 1956, pp. 50–75.

U.S. Department of Commerce, Office of International Trade, Investment in Colombia, Conditions and Outlook for United States Investors (Washington, 1953), pp. 88–98.

United Nations, ECLA, Analyses and Projections of Economic Development, III: The Economic Development of Colombia, Statistical Annex (in Spanish, Geneva, 1957), Table 1, and unpublished estimates by Colombia’s Planning Department.

Reed Moyer, Marketing in Economic Development (Institute for International Business Management Studies, Occasional Paper in International Business No. 1, Michigan State University, East Lansing, 1965), pp. 35–38.

Excluding food. DANE, Anuario General de Estadística, 1957 (Bogotá, 1958), p. 538, and Censo Nacional de Comercio y Servicios, 1954 (Bogotá, 1957), p. 41.

This is not to suggest that other influences, too, may not affect the choice of levels at which a tax might best be collected.

Such a tax on sales by both wholesalers and manufacturers to retail stores is recommended by George E. Lent, “Manufacturers’ v. Wholesalers’ Sales Tax Base,” Taxes, August 1958, pp. 573–82 and 601.

John Fayerweather, International Marketing (Foundations of Marketing Series, Englewood Cliffs, N.J., 1965), pp. 63–68.

John F. Due, “Administrative Criteria in the Establishment of Sales and Excise Tax Structure,” in Joint Tax Program of the Organization of American States, the Inter-American Development Bank, and the Economic Commission for Latin America, Problems of Tax Administration in Latin America (Baltimore, 1965), p. 417.

Calculations for 1918 are based on 11 of the Departments (Córdoba not yet having become a department), the percentage distribution between a capital and the remainder of a department in 1918 weighted by the departmental population of 1905; see DANE, Economía y Estadística (Bogotá), June 1957, pp. 108–109. Data for 1951 are from Table 27, and those for 1964 are from DANE, Boletín Mensual de Estadística (Bogotá), October 1965, p. 15.

References in this section are to 1953 U.S. dollars.

Other Resources Citing This Publication