Chapter

4 Improving the Administration of the Colombian Income Tax, 1986-88

Editor(s):
Richard Bird, and Milka Casanegra de Jantscher
Published Date:
September 1992
Share
  • ShareShare
Show Summary Details
Author(s)
Charles E. McLure, Jr., and Santiago Pardo R. 

Over the past thirty years, the Colombian income tax has undergone more or less constant change as the result of efforts by some to reform it and efforts of others to reverse that process. On balance, the system has been improved greatly, and Colombia now has one of the best income taxes of any developing country.1 Progress, however, has often lagged in one important area, that of the administration of the tax.2 This is especially troubling since, in the words of Milka Casanegra, an expert from the Fiscal Affairs Department of the IMF, in developing countries “tax administration is tax policy.”3

From late 1986 to the end of 1988 Colombia was again engaged in a fundamental reform of its income tax system. While the sweeping changes in tax structure—especially the elimination of the taxation of dividends and the movement to a comprehensive system of inflation adjustment—have drawn the most attention, there have also been important administrative changes that should greatly improve the implementation of the tax; furthermore, some of the structural changes will simplify tax administration.4

This paper describes the administrative improvements that were made during 1986-88. Section I describes the structural changes that were introduced to simplify compliance and administration. Section II describes the administrative changes per se that were made during the period. And Section III comments on the general applicability of lessons from the Colombian experience for other developing countries.

I. Improving Administration Through Structural Reform

As in most developing countries, Colombia has traditionally collected most of its individual income tax through withholding on wages and salaries. In 1970, for example, such withholding represented 70 percent of all individual income tax revenue. Colombia does, however, include income from virtually all sources in the tax base of its income tax, which is a global tax, not a schedular one. Like most countries, Colombia has traditionally employed deductions and/or credits to adjust tax liability to the economic circumstances of taxpayers—notably marital status, number of dependents, and expenditures thought either to reflect diminished taxpaying ability or to deserve public support through the tax system. Credits have been allowed for residential rent, health care, educational expenses, and mortgage interest. These deductions and credits, as well as multiple employment, the earning of nonlabor income, and the ceding of income to spouses, have made it impossible for amounts withheld on wages and salaries to reflect closely the ultimate liability of individual taxpayers, except in relatively simple cases (one job; no nonlabor income; no itemized deductions). As a result, many taxpayers have been obligated to file returns, either to report liabilities not covered by withholding or to claim refunds for excess amounts withheld.

In common with most developing countries, Colombia lacked the administrative capability to monitor compliance by the vast majority of individual taxpayers; this was clearly true for itemized deductions, and even for marital status and personal exemptions.5 All that was really possible was a check of forms for mathematical errors, a process for which the term “desk audit” would be too ambitious. As a result, many taxpayers failed to file returns as required by law; for them withholding was a final tax. Many who did file claimed exemptions and deductions for which they were not eligible. On the other hand, claiming refunds was a laborious and uncertain process.

In 1986 the Government decided to adapt the individual tax system to administrative reality. It eliminated the personal exemptions for the taxpayer, spouse, and dependents; it repealed a form of income splitting that allowed the taxpayer (commonly the husband) to “cede” a limited amount of income to the spouse (commonly the wife) for income tax purposes; and it eliminated many of the itemized deductions.6 (The only itemized deduction remaining is that for mortgage interest on the principal residence of the taxpayer. The Government wanted to repeal the mortgage deduction, but the Congress exerted political pressure to keep it.)

Repealing the personal exemption for dependents might appear to make the income tax unfair. But this exemption was virtually impossible to control, since taxpayers could claim nonexistent dependents to gain the maximum benefits from this deduction, with little risk of detection. The Government believed that eliminating the deduction would create less inequity than retaining it. Evidence from polls taken since the 1986 reforms suggests that the general public accepts this view.

With these changes, withholding would be much more accurate for most taxpayers with incomes largely from wages and salaries; the primary exceptions would involve multiple employment and income not subject to withholding on wages and salaries.7 Withholding was made a final tax for those meeting the following criteria: at least 80 percent of total income is from wages and salaries subject to withholding; other income has been subject to withholding; net wealth is not more than C$15 million (US$25,000); there is no obligation to collect sales tax; and total income is less than C$8 million (US$13,000).8

It is estimated that this change eliminated the need for 1.5 million individual taxpayers (out of 2.2 million) to file tax returns, thereby greatly simplifying compliance and administration. Moreover, it removed a factor that erodes the social fabric of a country: Income tax laws that are known to be impossible to administer invite the taxpayer to cheat in self-defense, knowing that everyone else is cheating. Perhaps equally important, elimination of the need to file tax returns, where filing is largely pointless, improves taxpayer morale.

Beginning in 1979 Colombia has steadily extended the scope of withholding beyond wages and salaries. In that year, withholding was applied to interest payments. Two years later it was extended to honorariums and commissions, and in 1983 to services. In an important move made in 1985, the Government began to apply withholding to all payments for purchases made by corporations, limited liability companies, and the Government. Initially set at ½ of 1 percent, the rate applied to purchases was raised to 1 percent in the 1986 reforms. The current withholding rates are as follows:

Wages and salaries0 to 30 percent
Honorariums and commissions7 percent
Interest7 percent
Services4 percent
Rents2 percent
Purchases1 percent

The differences in these rates are intended to reflect differences in ratios of net income to receipts. Thus, for example, it is assumed that the costs of generating a peso of honorariums and commissions are less than the costs of generating a peso of sales of merchandise, and that the ratio of income to receipts is correspondingly greater.

All amounts withheld are credited against ultimate tax liability on global income for those filing tax returns. Only for those filing no returns (primarily low-income recipients of wages and salaries) is withholding a final tax.

In 1989 withholding accounted for 70 percent of all revenue from the individual income tax. Of this, withholding on wages and salaries accounted for 21 percent, and withholding on other payments 49 percent. Overall, in 1989, withholding generated 50 percent of all income tax revenues (Table 1).

Table 1.Colombia: Withheld Taxes as a Percent of All Income Tax Collections
YearPercent Withheld
197514.5
197615.8
197718.2
197818.3
197922.6
198027.3
198129.4
198231.3
198327.7
198432.3
198538.0
198641.9
198749.2
198851.8
198949.7
Source: Directorate of National Taxes.
Source: Directorate of National Taxes.

The 1986 reforms included many other changes that simplified compliance and administration. Though these are far too numerous to discuss in detail, or even to list, the following are worthy of mention.

First, many forms of labor income that were formerly exempt were made taxable.9 This eliminated the need for rules drawing lines between exempt and taxable income.

Second, Colombia repealed a provision under which imputed income from owner-occupied housing was taxed. Though attractive in theory, this rule was not administrable.

Third, in order to “integrate” the corporate and personal income taxes and reverse the perceived “decapitalization” of the economy, Colombia eliminated the taxation of dividends. For administrative reasons, this route was chosen over conceptually preferable techniques such as a deduction for dividends paid, different rates on corporate income that is distributed and retained, and shareholder credits for dividends received.

Fourth, the rates applied to the income of corporations (formerly 40 percent) and limited liability companies (formerly 20 percent) were unified at 30 percent, and the top individual rate was reduced from 49 percent to 30 percent.

Finally, it should be noted that not all the 1986-88 reforms simplified the tax system. For example, the adoption of a comprehensive system of inflation adjustment for the measurement of income from business and capital adds to the burden of compliance and administration. This makes it doubly important that simplification and administrative improvements be adopted where possible.

II. Administrative Reforms

The 1986 reform legislation gave the Government legal authority (via the grant of “extraordinary powers” allowed under the Constitution) to restructure the tax administration and to change the procedural law governing relations between taxpayers and the fiscal authorities. This section describes the changes made under that authority.

The starting point for the exercise of the extraordinary powers was an analysis of the primary objectives of the Directorate of National Taxes. The objectives identified were the auditing of taxpayers on a current basis and the collection of amounts due. Further analysis showed that the first of these goals was being sacrificed to the second and that the second was not being discharged satisfactorily. This realization provided the foundation for the reforms described below.

Elimination of Certificates of Tax Payment

Traditionally, Colombia has placed great emphasis on paz y salvos (tax clearance certificates) stating that the taxpayer has no outstanding tax liabilities. These certificates have been required to leave the country, to transfer real estate, to register an automobile, to export or import, to enter into contracts with the Government, and so on. The idea is that the use of this technique will guarantee that those who have tax liabilities will meet them.

There are several problems with this approach. At best it helps to identify taxpayers and to collect taxes once liability has been established. It does nothing to ensure that correct liabilities are established—which can be accomplished (in the case of taxpayers for whom withholding is not a final tax) only if adequate trained personnel are devoted to auditing.

Even worse, the tax clearance certificate system did not even achieve the purpose of ensuring payment of known liabilities. There was an active market in these certificates, some of them initially obtained legitimately by taxpayers who had no use for them (as they did not contemplate foreign travel, the transfer of real estate, contracts with the Government, and so on) and some of them counterfeit.10 Only honest taxpayers needed to go to the trouble and expense of obtaining the certificate by legal means; for them the time and expense required to comply with the law could be considerable indeed.

Besides being ineffective and unfair, the tax clearance certificate system absorbed an enormous amount of administrative resources; some 60 percent of the 6,000 employees of the tax administration were involved in collecting taxes and issuing certificates. Moreover, the system created a false sense of security that the income tax was being administered effectively.

Again, the Government decided to face reality. It simply made no sense to devote so much effort to certifying that established liabilities were being met, and so little to establishing the liabilities. In February 1988 the Government eliminated the tax clearance certificate system for all taxes collected by the Directorate of National Taxes, thereby freeing up 3,600 employees for more productive activities, notably auditing with greater care the reduced number of taxpayers still required to file tax returns.

Collection of Revenue and Data Processing

As is the case in many developing countries, Colombia has long had problems in processing both tax payments and the data needed for tax administration and policy analysis. Part of the data processing problem has been the lack of legal authority to require that large corporate taxpayers (including limited liability companies, an important form of business organization in Colombia) provide data in machine-readable forms.

In addition, the 1974 reforms required that taxpayers file copies of many documents supporting claims for deductions and credits. In theory this information assists in the audit of those supplying the goods and services for which deductions and credits are claimed, as well as the taxpayer in question. In practice, the result has been quite different. The Directorate of National Taxes has been physically overwhelmed by the avalanche of paper produced by this system. In some cases large taxpayers have used trucks to deliver their tax returns, together with boxes of the supporting documents required by law, implicitly challenging the directorate to try to find the discrepancies between the amounts reported and their legitimate deductions and credits. Rather than assisting in the audit and cross-checking of the tax returns of buyers and sellers, the filing requirement actually impeded these tasks; at best the documents filed could be used only for auditing the taxpayer, not for cross-checking the returns of others. Moreover, the quantity of documentation required slowed down the processing of information from returns.

Auditing has been hampered by the practices of some unscrupulous taxpayers who have exploited an overly generous interpretation of a provision of the law intended to protect taxpayers’ rights. The law provides that a taxpayer whose return has been audited and corrected cannot be subject to audit again on the same return. Some taxpayers have deliberately included glaring errors on the first page of their returns (for example, omission of the taxpayer identification number of suppliers, without which no deductions are allowed), hoping that the errors will be found and corrected, making them immune from auditing that might lead to the detection of evasion based on far more serious errors buried deeper in the tax return.

Collection of taxes also involved serious problems. Collection officials in the Tax Directorate could conveniently “lose” returns accompanied by cash, or they could alter amounts of payments reported on returns; in either case they could pocket the money with relatively little fear that such pilferage would be detected by the lax internal controls. Yet another possibility was that the collecting officer, in collusion with taxpayers, would issue receipts for payment not actually made; this would allow the taxpayer to support false assertions that tax liabilities had been met.

In yet a third compromise with reality, the Government greatly simplified the tax return and eliminated the requirement that supporting documents be filed with tax returns. Now such documents must be retained and made available to the tax authorities on demand, but the information they contain need only be reported in summary form on actual tax returns. Moreover, large taxpayers (chosen by the tax administration on the basis of the size of their liabilities) must now provide information on magnetic media that can be processed by computer. As noted above, the 1986 reforms reduced drastically the number of individual taxpayers who file under the new system. Auditing the relatively few who do file will be easier.

Collection techniques have also been changed fundamentally. Since February 1988 the tax administration no longer handles either the initial receipt of tax returns or tax collections. Rather, all returns and tax payments (for all taxes under the jurisdiction of the Directorate of National Taxes—that is, income tax, value-added tax, stamp taxes, and so on) are now made entirely and solely to banks, which act as collection agents for the Government. This allows reliance on the well-developed internal controls of the banks. Moreover, banks are required to process and transcribe onto magnetic media all the data on tax returns and revenue receipts and forward them to the Tax Directorate in machine-readable form. To compensate the banks for their expense, the Government allows the interest-free use of the funds collected for 18 to 25 days before they must be credited to government accounts. As part of the arrangement, banks must process returns with no tax liability, as well as those with liabilities.11

This change has improved service to taxpayers, as well as tax administration. Whereas declarations and payments had previously been accepted by 900 government tax offices, they are now accepted by the 3,000 offices of banks located throughout the country.

Within 15 days after tax returns are presented, the Government has data in machine-readable form. Thus, cases can be selected for audit and collection follow-up on a current basis. This contrasts sharply with experience under the old system when data processing could take as much as two years from the time of filing of returns. Such delays create a number of serious problems. Tax collections decline in real value, especially when inflation is rapid. Even worse, chances of collection are reduced for a number of reasons: taxpayers may go bankrupt; there is a greater likelihood of collection being foreclosed by the statute of limitations (two years from the date of filing); and the “scent” of evasion is much colder when auditors attempt to do their work.

As a result of this step, an additional 400 government employees formerly engaged in receiving tax returns and tax payments have been made superfluous. Elimination of these jobs saves an estimated C$1 billion a year over and above amounts saved through avoidance of pilferage.

Use of Supplementary Information

The fiscal authorities now have access to information that is invaluable in identifying taxpayers and assessing liabilities. First, banks must file information returns identifying individuals or companies having accounts with more than C$120 million in transactions (the sum of debits and credits) annually.12 This has proved to be an effective means of identifying wealthy “taxpayers” who have never filed returns.

Credit cards issued by banks and others provide another important source of information. Issuers of credit cards must identify (1) all businesses that have charges to credit cards in excess of C$10 million a year and (2) all holders of cards who charge more than C$1 million to their card in one year. Thus far there does not appear to be much problem with a switch to cash payments, perhaps because of security problems associated with cash payment.13 On the other hand, a taxpayer can circumvent these rules by having several credit cards, perhaps at different banks.

Finally, notaries public are obligated to inform the fiscal authorities of all real estate transactions in excess of C$10 million.

These measures have been exceptionally effective. For the first time the tax administration has begun to audit those who have never filed. (Experience suggests that the worst mistake a taxpayer can make is to file a first return and thus become known to the fiscal authorities.) In 1990 the tax administration identified 26,560 taxpayers who had never filed a return; 15,815 of those were summoned and 4,607 filed a return before the end of the year with a tax of C$600 million (US$1 million).

Also, in that year, the tax administration visited 24,650 taxpayers who did not issue invoices as required; afterward 17,948 of those corrected their behavior. After a second visit, 125 taxpayers did not cooperate. As a penalty, their businesses were temporarily closed for one or two days. (See also below.)

The auditing capacity of the tax administration has been increased by the decisions to eliminate tax clearance certificates and use banks to collect both revenues and information. As shown in Table 2, in the last three years the results of auditing have increased dramatically. In 1987 the fraction of the income tax collected due to auditing represented only 1.6 percent of the total income tax revenue; in 1989 it represented 15.7 percent. While this experience could be given a variety of interpretations (for example, unchanged auditing effort and worse voluntary compliance), the proper interpretation is almost certainly improved audit performance.

Table 2.Colombia: Effectiveness of Audit(In millions of U.S. dollars)
198719881989
Income tax collections
Total collections1,335.41,426.91,534.0
Result of audit21.6143.8240.9
(In percent)(1.6)(10.1)(15.7)
Sales tax collections
Total collections717.5762.4797.1
Result of audit18.925.852.6
(In percent)(2.7)(3.5)(7.1)
Source: Directorate of National Taxes.
Source: Directorate of National Taxes.

Processing Refunds

Refunds pose problems for the tax administration of most developing countries. Colombia does attempt to meet its obligation to make refunds of overpayments where there is no evidence that the taxpayer has understated his liability. This is essential if there is to be an effective system of widespread withholding. Unfortunately the processing of tax refunds creates an opportunity for dishonest tax officials to extort money from taxpayers applying for refunds. The technique is simple: paperwork is “lost” or moved along slowly in the absence of bribery.

New administrative regulations designed to eliminate this abuse allow the dismissal of an employee who does not move paper through the system within a set period of time—ordinarily 30 days from the time the claim is presented; 5 days in the case of a claim for an “expedited” refund, supported by a warranty from a bank or an insurance company. (There are, of course, safeguards to protect employees who express to their superiors legitimate concerns about the propriety of claims for refunds.) The system of expedited refunds has largely solved the problem of getting refunds to those who have large excess amounts withheld (and of exporters under the value-added tax). The primary remaining problem involves the many small taxpayers who may lack the sophistication or resources to request an expedited refund. There is still some opportunity for extortion in these cases. A system of automatic refunds of small amounts (for example, less than C$50,000 or about US$83) with random spot checks would eliminate delays and opportunities for bribery—but not without some risk of lost revenues due to fraudulent claims for refunds. Such an approach would free further resources for auditing.

Morale of Taxpayers and the Tax Administration

Both the reforms just described—in processing collections and refunds—have led to an improvement of taxpayer morale—at least, the morale of honest taxpayers. Rather than facing the prospect of dealing with dishonest officials who will demand bribes or otherwise cause trouble for the taxpayer (such as by “losing” the record of his payment), the taxpayer can be relatively confident that payments will be handled properly by the receiving bank and that claims for refunds will be processed expeditiously by officials who lack the leverage to demand bribes.

The various reforms described above have also had a salutary effect on the honesty and morale of the tax administration. In the prereform system, corrupt officials had many opportunities to demand bribes, both in the collection process and in the processing of refunds. Because of laws regulating public employment, such officials could not be discharged without proof of illegal activity. Yet graft and corruption were difficult to detect and prove because controls were inadequate. Worse, the presence of corrupt colleagues undermined the morale of those inclined to be honest, causing the spread of discontent and dishonesty.

The reduction in opportunities to extract bribes from taxpayers has changed this environment. Without the possibility of supplementing their incomes through bribes, some tax officials—presumably those who were corrupt—have resigned. The officials who remain are inevitably more honest than those who have moved on to greener pastures.

Penalty Provisions

Before 1986 certain violations—notably “inexactitude,” that is, omission of income and overstatement of costs—carried a penalty equal to twice the amount by which the tax had been understated. But the penalty did not become due until the final resolution of liability, after the taxpayer had exhausted all appeals. Moreover, penalties were not indexed for inflation. Even worse, penalties did not carry interest because of the interpretation of a constitutional provision that prohibited the tax authorities from collecting two penalties based on the same violation. With interest rates at 25-40 percent a year, there was a substantial incentive to contest assessments, since the real present value of a penalty postponed several years was reduced substantially.14 Finally, the recent history of repeated amnesties made it virtually certain that liabilities would be forgiven even before they were established. Under the prior system, ten-year delays in settlement were not unusual.

Tax lawyers were the primary beneficiaries of this process. Thus, there was room for reform that would benefit both taxpayers and the government, at the expense of tax lawyers.

New rules have been adopted to encourage quicker settlements. The basic sanction has been reduced to 160 percent of the amount at issue, but it can be reduced still further by quick admission of liability by the taxpayer. A mistake not yet discovered by the tax administration can be corrected by paying only a 10 percent penalty (formerly the penalty was 100 percent). If the taxpayer agrees to settle at the time of the initial field audit by the tax authorities, the penalty is 20 percent (formerly 200 percent) of the underpayment. If a formal demand for supplementary payment made by the tax administration is accepted by the taxpayer before the case goes to court, the penalty is 40 percent. If the taxpayer agrees to the increased assessment after the case goes to court, but before the final judicial determination of liability, the sanction is 80 percent.

The result of this revision of the penalty system has been greater willingness to settle claims quickly. Disputes on some 80 percent to 90 percent of audited returns have been settled with sanctions of 20 percent to 40 percent. In 1990, of 3,248 audits, 2,640 taxpayers had corrected their returns within two to three months, resulting in increased tax collection of C$2.3 billion (US$3.9 million). Under the law prior to 1986, this process would have taken eight or nine years of controversy and might have yielded less additional revenue.

Under another form of sanction, the Directorate of National Taxes has been authorized to close a business that does not issue receipts to customers or that is found to have more than one set of accounts. (Procedural safeguards for taxpayers include the requirement that there be a formal written accusation, to which the taxpayer has a specified time within which to prove innocence of the charges.) There has been television coverage of the tax administration closing shops for one day and posting public notices reading “closed for tax evasion.”

Organizational Changes

In addition to the types of reforms just described, important changes were made in the organization and conduct of the Directorate of National Taxes. There were too many administrative layers between the Director General of National Taxes and the taxpayer, which reduced the effectiveness of vertical control. Moreover, the responsibilities assigned to the various divisions in the directorate were so broad that expertise suffered in some cases. A streamlining of the organization reduced the number of administrative levels, increased control, and allowed more qualified personnel to deal with particular problems.

Since relatively few large taxpayers generate a widely disproportionate amount of revenue, special administrative offices were established to deal with such taxpayers in the three largest cities, Bogota, Medellin, and Cali.15 In other cities, special units within local tax offices were created to deal with large taxpayers. This change is intended to provide better service to taxpayers, as well as to improve tax compliance.

Tax administration has long been favored by politicians as a means of rewarding supporters with government positions. Such political appointees are generally poorly qualified, and some are dishonest. To combat this practice—which also exists in other agencies—a career civil service has been created for the public sector, including the tax administration. With hiring and promotion based explicitly on merit, it should be easier to resist unqualified political appointments, as well as to weed out unqualified employees by preventing their promotion. This new system also allows greater ease of dismissal of officials hired after 1988.

A National Tax School has been created to provide instruction in tax law for both the public and private sectors. Inexpensive day-long seminars have brought together tax administrators, practitioners, and business employees; 2,000-3,000 persons have attended such seminars at one time. Besides serving an important instructional function, such seminars provide valuable interaction that leads to a two-way process of learning—about taxpayer problems and about the goals of the tax administration. Such interaction can help dispel the feeling of mutual distrust that all too often characterizes relations between taxpayers and tax administrators.

III. Lessons for Other Countries

Colombia’s experiences with problems of administering its income tax system have valuable lessons for other developing countries. As with the discussion of the previous two sections, these conveniently fall into two categories, implications of tax structure for tax administration and broad issues of tax administration.

Implications of Tax Structure

The tax systems of many developing countries closely resemble those of the advanced countries that colonized or otherwise exercised hegemony over them. (These sometimes seem to be “frozen in time,” containing features since discarded by the developed countries.) Thus, former British colonies have tax systems that resemble that of the United Kingdom, francophone countries follow the French system, and the tax systems of Latin America have been strongly influenced by the U.S. tax system. Unfortunately, not enough attention has been paid to the fundamental question of whether the tax structure of a developed country is appropriate for a developing country, considering especially differences in administrative capabilities, as well as cultural attitudes toward taxation. For present purposes, we focus only on limited features of the personal income tax.

The standard paradigm for personal income taxation is similar to the following: Add together income from all sources. Subtract deductions for personal exemptions and itemized deductions. (Alternatively, allow analogous tax credits once tax has been calculated; this distinction is not important for present purposes.) Then apply the rate structure to calculate tax liability. Employ withholding on wages and salaries in order to ensure collection on a “pay-as-you-earn” basis. In some countries, the individual is the taxpaying unit, in others the married couple or the family unit. In some instances, the rate structure depends on marital status, by means of income splitting. With this paradigm in mind we can see where problems arise.

The theoretical justification for deductions for personal exemptions is a venerable and attractive one. There is no taxpaying ability until income exceeds a certain minimum level; deductions for personal exemptions prevent the taxation of income below the level of the exemption. Since the minimum level below which there is no taxpaying ability is clearly related to family size, personal exemptions are commonly allowed for the taxpayer’s spouse, children, and perhaps other dependents.

In developed countries that have good records on vital statistics, there is relatively little difficulty monitoring claims for personal exemptions. In principle, in some developing countries the relevant information on vital statistics could also be made known to the tax authorities; in Trinidad and Tobago, for example, the Department of Inland Revenue is responsible for vital statistics. But in many developing countries, the communication and use of these statistics is quite inadequate. In such cases, the provision of deductions or credits for all dependents is an open invitation to cheating; in essence, it creates a tax on honesty. Under these conditions, it is far from clear that personal exemptions make the tax system more equitable, rather than less.

Many countries allow itemized deductions. Although such deductions often show an amazing variety of response to local customs and political forces, there is a rather standard “core” group that is found in many countries. The deductions include interest on mortgages on owner-occupied housing, certain taxes (especially those on real estate), medical expenses, charitable contributions, and various forms of saving (life insurance, purchase of mutual funds, and so on). Many of these deductions can be questioned on grounds of public policy, but that is not our objective in this paper.16 Our concern is whether they can be administered satisfactorily.

As with personal exemptions, many of these expenditures could, in principle, be subject to audit; most mortgage interest is paid to financial institutions, as are commonly deductible payments to institutions holding the savings of taxpayers; taxes are paid to governments; and medical payments are made to hospitals and to doctors, who can be required to register and issue receipts. In fact, it is difficult to cross-check such information unless evidence of payment—commonly copies of receipts—is attached to the taxpayer’s return; after all, it is only in the last few years that the United States has begun to require that financial institutions and state and local governments file information returns covering interest and taxes received. Falsification of receipts is a potential problem unless either (1) the amounts involved are so small that the deductions could just as well be repealed or (2) safeguards are taken to prevent falsification, in which case administrative costs skyrocket. Given the dubious policy grounds for many itemized deductions, repeal seems an attractive option.

The case for the elimination of both personal exemptions and itemized deductions is strengthened once we consider their implications for with-holding. Given the levels of literacy and of legal and economic sophistication found in most developing countries, it is desirable that withholding be a final tax for those with little nonlabor income. Yet this is possible only if withholding takes account of the taxpayers’ personal exemptions and itemized deductions. In principle, the employer’s information on personal exemptions could be communicated by the government, as it is in Jamaica, and therefore be accurate—or at least as accurate as the government’s information. But we have seen above that the government does not necessarily possess these data.

The problem is far worse for itemized deductions. At least the vital statistics do not change during the year for most taxpayers, and the changes that do occur are objectively clear and easy to verify, at least in principle. By comparison, itemized deductions generally are not known even by the taxpayer at the beginning of the year, and they can be known accurately and certainly by the fiscal authorities, only with great difficulty, after the year has passed. This information, which is needed for withholding, is not likely to be known accurately by the employer during the year, even if the employee has a good estimate of the relevant expenditures. Moreover, unlike the government, the employer has little direct incentive to attempt to verify that the itemized deductions of its employees are legitimate. In short, if taxpayers are actually to claim itemized deductions, there is little chance that withholding can be a final tax. Of course, what commonly happens in practice is that many taxpayers forgo filing returns and claiming itemized deductions, so withholding is de facto a final tax.

The approach used in Venezuela illustrates the problem. At the beginning of the year taxpayers indicate to their employers the number of personal exemptions and estimate their itemized deductions. Employers then base withholding on this information. If personal exemptions or estimated deductions change during the year, the employee is to inform the employer, who calculates the implied change in tax liability and alters the amount withheld to spread the resulting change in liability over the remaining wage and salary payments for the year. Finally, the taxpayer is obligated to file a return at the end of the year, reporting actual figures. This system is clearly quite cumbersome for both taxpayers and employers; moreover, it is easily abused. It appears that far fewer taxpayers file returns than are legally required to do so.

The upshot of this discussion is fairly obvious. In many developing countries, neither personal exemptions nor itemized deductions can be monitored with enough precision that we can be sure that they contribute to equity among taxpayers. Moreover, many itemized deductions lack justification on policy grounds. Finally, personal exemptions and itemized deductions make it very difficult to design a satisfactory system of withholding. As a result, either scarce resources are squandered on complicated withholding systems and the filing and auditing of returns, or else the equity objectives underlying these provisions are not realized. All things considered, it does not seem that these provisions involve enough net benefit to justify their continued use by many developing countries. Their elimination—and the possibility that withholding could be a final tax for many taxpayers—would raise taxpayer morale and would free scarce administrative resources for more productive uses.

The choice between individual filing and filing by married couples or family units generally reflects deeply held social views; we do not comment on these. But the choice also has important economic and administrative implications. First, making the couple or the family the taxpaying unit has adverse incentive effects, since any income earned by secondary workers is taxed at marginal rates determined by the income of the primary earner. Second, and more relevant for present purposes, this choice complicates withholding, since ultimate tax liability depends on the aggregate income of all members of the taxpaying unit. Accurate withholding is simpler under individual filing, since ultimate liability depends only on the income of the individual.17

Administrative Reforms

The type of administrative reforms implemented in Colombia seems to be widely applicable to other developing countries. One important lesson of the Colombian experience is that tax clearance certificates are valuable to taxpayers if they can be counterfeited or obtained illegally—perhaps even more valuable than money. Yet they are often printed with no safeguards against counterfeiting and theft. The point is not simply to print these certificates under the same control as bank notes. Rather, the lesson is that if the tax administration can certify the discharge of tax liability, it can also upgrade collection for everyone to the point that such certificates are unnecessary.

In all developing countries, owners of small businesses and noncommercial professionals are notoriously difficult to tax; indeed, experience from many countries indicates that the tax administration may not even know the identity of some such potential taxpayers. Cross-checking the membership lists of professional societies against tax returns can help identify professionals; this technique was used effectively in Jamaica.18 Tax administrators can also ascertain that owners of relatively large businesses have filed tax returns. But there may remain a residual of relatively well-to-do persons who do not file. Use of information on bank accounts, credit cards, and transactions in real estate can help to locate such potential taxpayers.

The tax administrations of most such countries probably suffer from the same types of corruption that has plagued the Colombian system. Thus it is important to eliminate opportunities for corrupt officials to practice extortion and theft. Using the banking system to collect taxes is one such approach; imposing sanctions against undue delays of refunds is another.

While many tax administrations are understaffed and underpaid, many—including some where understaffing and low pay are problems—are saddled with employees known to be dishonest, who cannot be dismissed for lack of evidence. Eliminating opportunities for extortion can help to force them out of positions they would not otherwise voluntarily leave.

Under the laws of some countries, only written documents can be used as evidence. This is sometimes interpreted as implying that copies of such documents must be filed with tax returns. The resulting flow of paper to the fiscal authorities is acute if there are no legal requirements that data be provided in machine-readable form, even by large taxpayers. Colombia’s experience illustrates both this problem and its possible solution.

Data processing can also be improved by using the banking system to process the data from tax returns, as well as to collect money. In most countries, the banking system is among the most highly computerized sectors of the economy; it is thus generally likely to be far more capable of processing tax return data than are the fiscal authorities, which are commonly understaffed, underpaid, and not computerized.

The structure and application of penalties are also a common source of problems in developing countries. In some cases, the statute of limitations is written in such a way that a taxpayer can escape liability entirely by employing delaying tactics. This was true in Colombia during the 1960s. More commonly, the provisions for penalties and interest are so generous to the taxpayer that delay is attractive, even if it is certain that penalty and interest will eventually be assessed. Again, the Colombian experience in reducing the incentives for delay may have instructive lessons for other countries. Finally, even if, in principle, penalties and interest are adequate, they will have little effect unless they are applied consistently and fairly. This has also been a problem in Colombia.

Many of the reforms discussed here can improve the image of the tax system and the tax administration. These include requiring fewer ordinary citizens to file tax returns, elimination of tax clearance certificates (especially where the system is abused), fewer requirements to provide copies of documents with the tax return, dealing with efficient banks instead of inefficient public tax offices, more accurate public record keeping and more timely refunds, fewer dishonest tax officials, and quicker settlement of disputes. Such improvements may be among the most important benefits of tax reform. Taxpayers are not likely to meet their obligations fully when the tax system is seen to be unfair and the tax administration is thought to be corrupt or inept or both. This is especially true where there is a strong distrust of government. An improved image will not induce taxpayers to trust the government or to pay taxes with glee. But it will almost certainly help to reduce distrust and improve compliance.

Comments

Vázquez-Caro Jaime

The purpose of these comments is threefold: (1) to describe the political, economic, and fiscal environment that led to the Colombian tax reforms in the 1980s; (2) to challenge the underlying approach used by the authors to define and organize the various administrative categories; and (3) to examine certain substantive aspects of tax policy and to propose possible explanations for developments to date.

The authors discuss what has been accomplished on the basis of what could be called the income tax’s “manageability” in terms of its substantive and procedural features. In addition, they examine the organizational strategy, which was based on a greater recourse to the private sector—as a withholding instrument for the bulk of taxes—and the banking system as the principal teller which, in addition, performed the initial processing of tax returns. This strategy was intended to simplify anything that could pose administrative difficulties and to privatize activities that had previously failed. After explaining the characteristics of the reforms, the authors draw conclusions in the form of lessons that might apply to other countries.

The Climate for Tax Reform in Colombia

Colombia’s macroeconomic policy has changed gradually. By Latin American standards, it has opened up its economy to the rest of the world cautiously and has undertaken conservative external credit and monetary policies. Much the same has occurred with regard to its tax system, which was reformed after many background studies.

After the 1974-76 tax reforms, Charles McLure and Malcolm Gillis, who were members of the Musgrave mission of 1968, returned to Colombia as World Bank consultants to make their assessment. They were surprised at the manner in which the Colombian working group had adapted the recommendations they made in the late 1960s. In 1974, the administrative aspects (of which I was in charge at the time) were not regarded with the same interest as the economic and the design aspects of the tax reforms. As a result, the Gillis and McLure study (1977) described the Colombian tax system as among the most advanced in Latin America. Economic reasoning at the time accepted the duality of solid tax theory and ineffective tax administrative practices.

Between 1977 and 1981, very little happened in the country with respect to taxes. During that period favorable external circumstances generated nontax revenue (from foreign exchange operations of the Central Bank) on a scale matching total value-added tax (VAT) receipts. Instead of adopting offsetting fiscal measures when this revenue dried up in 1980, the country resorted, between late 1980 and 1982, to foreign credit, which doubled the external debt but allowed the ailing tax system to breathe easier. Once the supply of external credit was exhausted, Colombia financed its deficit by issuing new money, thereby prolonging the fiscal lethargy still more.

Tax policy began to be redefined only when the scale and monetary implications of the fiscal deficit had demonstrated the urgency to consider the issue of taxes seriously. The earlier magical period of public finance gave way to a period of fiscal crisis. During the 1980s the signal given to the Colombian tax authorities—both from within and from the outside—was to increase tax collections. There was no alternative. The directive was clear.1 Like the rest of Latin America, during the debt crisis of the 1980s, Colombia was forced to generate fiscal surpluses to meet its external debt repayment obligations. The only long-term solution was for the tax system to meet these imbalances and maintain current public spending levels.

These macroeconomic imperatives became clear and the microeconomic framework used to set policy was simplified during the 1970s. The consensus among economists was to lower tax rates and to simplify tax laws. More and more people were convinced that a plethora of tax incentives had made it impossible to control the tax evasion masked within tax avoidance. Moreover, even the very existence of the income tax was questioned because of its effects on economic efficiency. The VAT was assigned a prominent role in many countries because of its neutrality and for its potential for generating revenue. Furthermore, an awareness that procedural legislation had to be streamlined to increase effectiveness had already taken hold in the country. Colombia has been a pioneer in developing procedural frameworks based on legal efficiency. The procedural reforms enacted in 1967 (Decree-Law No. 1366), 1974 (Decree-Law No. 2348), and 1977 (Law No. 52) provided an important framework for the conceptual development of a national tax structure for the economic development of the 1980s.

The Colombian income tax endured despite the growing popularity and importance of VAT. The modifications made were primarily aimed at achieving compatibility with the new economic doctrines and the new emphasis on administrative viability. These doctrines improved the conditions for tax administration implicitly, because they were based on notions of simplification and lowering of tax rates.

The informal sector continues to account for a large share of the Colombian economy. The fact that widespread smuggling is tolerated leads not only to tax evasion on imports but also on VAT as many taxpayers’ transactions occur under the table. The counterweight to this structural trend is that during the last decade Colombia’s economy has been characterized by an affluence of dollars as a result of the drug trade. Tax amnesties (in 1982, 1984, 1986, and 1990) have played a central role in promoting the legalization of this new money. In all likelihood, these amnesties have had an impact on income tax elasticity, but there has been no research on this subject.

The Importance of Tax Administration

In the medium and long terms, the success of a tax system depends less on the soundness of the laws governing taxation and procedures than on the administrative practices that emerge during the course of its development. The administrative implementation of any tax system is critical to its long-term survival. Casanegra’s (1990) fitting remark, “in developing countries, tax administration is tax policy,” applies not only to developing countries. I believe that the secret to success in the developed countries has been this emphasis on implementing tax laws. Enacting legislation in the absence of administrative planning is useless, and also frustrating for those in charge because their systems are actually better “designed” for failure than for success. An indicator of the implicit effectiveness of tax legislation might perhaps be constructed by identifying the specific provisions in the laws that are inconsistent with the implementation of the system, the “monkey wrench” indicator. For instance, studies of Argentine tax laws show that in some of these laws one out of every five provisions contradicts the rest of the text. In the case of the Colombian reform process, one of the major accomplishments of tax policymakers and administrators during the 1980s was the deliberate removal of the monkey wrenches that had been thrown into the tax system.

Administrative analysis of each separate tax is not easily performed, because in many cases the administrative units are concerned with such overall aspects as identifying taxpayers, systems for recording and monitoring a taxpayer’s total obligation, control of tax evasion in general, custody of the supporting documentation pertaining to tax arrears, and tasks related to improving tax compliance.

Obviously each tax will require a specific administrative approach that will depend on the degree of procedural and administrative integration of the overall system. One approach has a separate tax identification number for each tax combined with a single current account for each taxpayer, specific collection system for each tax, and an integrated audit system, as was the case in the Argentine system until recently. Another system, used in Colombia until a few years ago, uses a unique tax identification number for all taxes and separate accounts and auditing schemes for each tax. The possible combinations of administrative strategies are countless.

Examining the administration of a tax requires separating the integrated administrative aspects from those aspects that are specific to that tax. In Colombia, for example, for many years the trend has been toward integrating collection systems for the VAT and the income tax. Tax audit is performed separately for each tax, however, and a taxpayer can be visited by more than one tax inspector, each of whom is looking for evasion of different taxes. In Argentina, on the other hand, until quite recently each tax had a separate account but auditing tended to be integrated.

The study fails to consider these distinctions. After explaining the specific aspects of the income tax, the authors discuss the modifications made to all taxes as if they were dealing only with the income tax. To keep matters straight, I suggest the classification used in Table 1, which lists the various ways of organizing procedures according to the sequence in which they occur and the functions observed in the Colombian system. The table indicates that procedures can be grouped as “atoms,” that is. individually by tax. or as “molecules,” that is, encompassing several taxes. In practice, because laws on procedures are not defined systemically, procedural design is often incoherent in many countries.

Table 1.Colombia: Administrative Procedures
ProcedureBy Type of TaxCommon to All Taxes
Identificationx
Withholdingx
Tax filingx
Paymentx
Amended returnx
Investigationx
Discussionx
Accountingx
Collectionsx
Tax amnestyxx

Critique of the Study’s Conclusions

What are the main points made in the McLure and Pardo study?

First, the role of the income tax. The income tax continues to play a decisive role in the structure of Colombian tax revenue. As noted above, the reforms preserved the major revenue-generating taxes. The income tax and the VAT underwent a parallel evolution. Following developments that seemed to indicate that the VAT would overshadow the income tax, the latter recovered its leading role. This occurred despite government policies favoring the VAT. Much the same occurred in Mexico. More recently, Ecuador has also undertaken to strengthen both taxes. Curiously, Colombia did away with the wealth tax which had supplemented the income tax.2

The generalization of the VAT in 1983 and the sales tax, its predecessor, in 1965, had always been intended to supplement income tax revenue. The trend followed by some Latin American countries in the 1980s was to rely mainly on the VAT and to reduce the role of the income tax. In Argentina, for example, the tax structure continues to include a complex income tax that collects little. Chile kept the income tax but with less revenue-generating importance. Uruguay virtually abolished the income tax in the early 1980s, although schedular tax schemes have recently been reintroduced. In Colombia, the relative importance of both taxes has been kept in balance. Experiences in Argentina and Uruguay suggest that neither tax should be used exclusively. As the income tax has been practically eliminated, governments have been forced to raise VAT rates on a regular basis to levels that may be undesirable from both economic and administrative viewpoints.

Second, the timely consideration of administrative aspects. The Colombian tax system has been greatly improved by an approach to reform that sought to integrate in a coherent fashion the substantive aspects of the reform process—taxes themselves—with the administrative process and organization. It was decided to do whatever was needed to make the system work. Rather than seek new solutions, implementation was the main preoccupation. For the first time in the country’s history, the decision to enforce the tax system was accompanied by changes to the laws aimed at facilitating and simplifying administrative procedures. Establishing a sound tax administration was set as a parallel objective to tax structure reform. In brief, the administrative aspects of the tax system were taken seriously. For years, Colombia had suffered setbacks that prevented an operational tax system from being developed. At the political level, taxes were continually redefined, while the tax administration maintained rigid management systems. The measures described in the study attest to the progress made in the case of the income tax. The demands imposed by “what should have been done” were eased, and the focus was put on doing what could be done.

In particular, progress was made in extending the inflation adjustments introduced to a certain extent in the 1974 reform, the tax treatments of business partners and businesses were integrated in a simple manner, and tax rates were lowered to levels in line with U.S. rates. These modifications to the system brought it closer to the realities of the economy and overall enforcement became somewhat easier. The standardization of tax paid by most individuals was unquestionably a reform in which administrative considerations were taken into account to the greatest degree. In a sense, the Colombian strategy consisted in striving for what is possible while accepting that perfection cannot be attained. This acceptance resulted from the recognition by those who analyzed the system of the need for simplifying tax legislation.

Clearly, the Colombian system was improved with the easing of demands on the administration and the elimination of procedures previously considered to be necessary safeguards. Operations became smoother, and previously unchallenged procedures were changed. An assessment of what occurred is incomplete without an analysis of the total cost of operating the system. This cost should include the cost to the private sector. The arrangements with the banks imply an opportunity cost to the budget, which should be measured in terms of its public debt cost. The widespread levying of taxes at source has a financial impact on liquidity and implies administrative processing of tax withholding forms by those that prepare them. Despite noticeable improvements and taxpayer satisfaction with the new system, it should be noted that higher administrative costs are being incurred because of the need to remunerate banks for processing tax information and cash handling and because more work is required from withholders.

What are the flaws in the analysis?

The authors fail to emphasize sufficiently what I consider to be of paramount importance for tax reform: the legal design of a penalties system based on rewarding compliance. The penalties set up in Colombia over the years were so inconsistent that in many cases they actually encouraged noncompliance. Even worse, they became a weapon in the hands of officials intent on corrupting the system. The fact that tax inspectors can come to an agreement with taxpayers on assessment and impose penalties whose severity increases according to delays caused by taxpayers’ opposition, has restored leverage to the tax administration.

The problem posed by paz y salvos (tax clearance certificates) was part of a broader diagnosis performed in the 1970s that failed to result in action, because of political factors and not because of a lack of understanding of the issue among those who participated in the reforms of the 1970s. Moreover, the study may leave the mistaken impression that tax clearance certificates were an invention of the 1974 reforms and that the 1980s were spent trying to solve the problems of the 1970s. Tax clearance certificates were a product of the 1960s that remained in force as testimony to administrative inability to set up and manage an accurate taxpayer current account that could have been the basis for an effective collection system. The problems of tax procedures encompassed much more than just the tax clearance certificates.

The study also criticizes the tax administration for requesting huge amounts of information that were never actually used. With the implementation of the 1974 tax reform, magnetic media could be legally used for submitting information. At the time, it would have been out of the question to make such a requirement mandatory. Therefore, submitting tons of information on paper was the taxpayer’s privilege. The fact that much of this information was unused is more a reflection of carelessness and lack of purpose than of any redundancy. Moreover, the Directorate of National Taxes has always reserved the right to require additional information. Tax reform was not needed to modify this aspect of compliance. Simply eliminating the annexes would have sufficed.

The decisions made were based more on conjecture than on a comprehensive evaluation of the various options. Income tax withholding on wages and salaries could have been monitored together with payroll taxes. Checks on withholding at source appear to have structural weaknesses because employers are not required to report amounts withheld for each employee, which thereby precludes cross-checking between withholders and taxpayers. This situation allows for fraudulent behavior from both sides. Withholders may pocket the withheld amounts and taxpayers can file claiming a tax refund.

What are the points on which I disagree with the study by McLure and Pardo?

A simpler tax cannot be justified on grounds that Colombia is a developing country, if the technology for supporting the administration of more equitable tax systems is available. To a certain extent the pragmatic approach used in making the decisions was based on the constraints of computerization during the 1970s and ignores the fact that the hardware of the 1980s has tremendous potential and that Colombia has adequate human resources for designing an appropriate system. Paradoxically, in the 1970s we may have made the opposite mistake: we had great faith in the computers that were being developed and tended to overestimate their potential.

I do not share the view that administrative problems justified the elimination of deductions for children and other dependents and for health expenses. It is true that nonexistent spouses and children were made up for “tax purposes.” This occurred because the administration never tried to keep the situation under control. No one seems to have remembered that the Cajas de Compensatión Familiar (Family Allowance Funds) maintain an updated roster of all dependent children of employees and that the Directorate of National Taxes could have obtained this information on magnetic media. Moreover, keeping the deduction for housing expenses is unfair and undermines the streamlining effort.

I do not share the view that the cause of the computer systems problems lies in the organizational structure of the Ministry of Finance, in which a separate Systems and Information Center operates as a Directorate-General having the same level as the Directorate of National Taxes. The problems encountered with the computer systems were not organizational but rather technological: mainframes used obsolete languages, and procedures were based on standards used in the late 1960s. Incidentally, I have found this to be true also in Argentina, Brazil, Chile, Ecuador, Mexico, and Venezuela.3 In Colombia, redefining the current account and achieving viable alternatives in data processing were more important than organizational changes.

I have serious reservations regarding the consolidation of the tax return and the payment slip in one document. This was not necessary to solve the current account problem. Because of the consolidation, some information useful for audit had to be sacrificed, including information that previously had been processed on applications such as “SACUATO,” which was the first real attempt by the Tax Directorate to come up with a clear and systematic picture of taxpayers. As opposed to tax collection, audit requires more data, which are used in more varied ways. This approach essentially creates a hybrid for managing the continuous tax collecting process, and audits, which are a discontinuous or discrete process. Revenue collection is hampered because of data checks performed for the auditing process, that is, information other than payment of the tax requires substantially greater validation and processing than is required by the mere entering of the payment amount. Before the current account is fully updated in Colombia, a new payment cycle is already beginning, which, by definition, once again renders the account obsolete.

For two reasons, it is unlikely the new tax system will be successful in the long run. First, the system is based on a few taxpayers who, in addition to being responsible for the collection, payment, and paperwork of the taxes of third parties, are required to file and make payment on their own taxes. The administration is no longer administering. Most taxpayers are excluded from the system. The system relies on huge penalties on withholders-at-source who fail to comply. In other words, the system relies on the administration’s capacity for monitoring compliance of these withholders in the medium term. The problem is that although the elimination of tax returns has reduced the number of persons the administration must deal with, the transactions that must be closely watched and that are taxable are still the same. The periodic filing of tax returns marks the close of the accounting period, the double entry of the system. I disagree with the view that the capacity of the present computerized system is insufficient to process the tax returns of all taxpayers.

I disagree also with the way the banks are remunerated for processing tax return data. The system is a poor one for the following reasons:

  • It results in a “Tanzi effect,” which benefits banks; the greater the increase in inflation, the greater the loss to the Government.

  • To encourage tax payments from major taxpayers, banks give them loans with low interest rates, thus effectively reducing the taxes paid by major taxpayers.

  • Finally, at the time the banks must hand over the taxes to the Government, the fierce battle for liquidity that ensues distorts interest rates.

In broad terms, it can be said that Colombia took a major step forward when it decided to give a leading role to administrative considerations when formulating tax policy. Unfortunately I believe too much importance has been given to the public side of administration; the result is that the private sector now carries a heavier load—including banks, withholders, large-scale informants—and there are shortcomings in the fairness of the system.

On the whole, the study provides an excessively optimistic picture of the Colombian reform process of the 1980s. It also presents it as the only available alternative. I do not share these views. The merits of the process were, as the authors note, its ability to implement new efforts. The pragmatism shown by the reformers in objectively assessing the situation permitted progress to be made. This was possible because the reformers also took it upon themselves to create the conditions necessary for progress.

Richard Gordon

Charles McLure and Santiago Pardo R. have written an excellent paper. They describe a number of important techniques used by one Latin American country to improve administration of the income tax, most particularly as it affects individuals.

The following comments expand on a few of the points raised by the authors that relate primarily to substantive tax law. More specifically, these comments will address (1) administrative aspects of the design of substantive tax law and (2) some techniques for enforcing that substantive law. The comments are designed to add to McLure and Pardo’s suggestions for improving income tax administration in developing countries that are engaged in substantial income tax reform, particularly in the area of personal tax reform.

Design of the Substantive Income Tax Law

The first key to effective income tax administration lies in the design of the substantive income tax law itself. McLure and Pardo describe how certain changes in the Colombian income tax law greatly eased the job of administration. The most successful of these changes resulted in three important benefits: (1) they reduced the complexity of the system; (2) they reduced the dependence of administrators on those facts that were difficult to verify; and (3) they reduced the number of individuals who were required to file returns.

In effect, perhaps the most important role of the first two benefits, that is, simplicity and reliability, is to permit the last, that is, reducing the number of filers. The fewer the people who must themselves file returns (including those whose final income tax is remitted by their employer and who do not have to file returns themselves), the fewer the returns that must be processed, the fewer the refunds that must be made, and often the fewer the audits that will be required.

Complexity

As McLure and Pardo correctly point out, reducing the theoretical “complexity” of a system may not always make it easier to administer. The theoretically least “complex” system would tax a flat percentage of every person’s “global income,” that is, the actual net accretion to wealth of an individual over a specific period of time. Many tax policy experts would contend that such a system would be optimal from an economic perspective. Of course, determining actual global income would be virtually impossible.

Nevertheless, at the outset, complexities should generally be avoided when designing an income tax. “Complexities” should only be added when there are compelling policy reasons to do so, or when tax administration would be greatly simplified.

When looking at the initial design of a tax system, one of the most important aspects of tax simplicity is treating all taxpayers and all income as similarly as possible. By so doing, incentives for shifting income from taxpayer to taxpayer, or from type to type, are reduced. This also reduces problems regarding defining different types of taxpayers or income.

In most cases, treating different taxpayers as similarly as possible requires keeping the number of marginal tax brackets relatively small. Also, in most cases, treating most different forms of income alike requires eliminating deductions and credits that are not a legitimate part of determining real income.

For policy reasons, however, different taxpayers cannot always be treated alike. Most tax jurisdictions conclude that a zero bracket, and at least a small number of graduated rates, are desirable. Although graduated rates complicate administration, they nearly always constitute a compelling policy reason for violating simplicity.

Treating all forms of income alike can also result in substantial administrative problems. For example, capital gains can usually be taxed effectively only on realization. Because the timing of realization is often under the control of the taxpayer, deductibility of capital losses must normally be limited in some way to net capital gains. If not, taxpayers would tend to realize losses and not gains.

There are also substantial administrative problems, in addition to policy ones, in treating all taxpayers alike. Individual taxpayers at the higher end of the income spectrum normally owe more in tax than do individual taxpayers at the bottom. It is therefore usually preferable for tax administrations to expend more resources in evaluating and confirming the income of an upper-income individual than that of a lower-income person.

In sum, the simplification of income tax administration often requires making exceptions to the rules of taxing all taxpayers and incomes alike.

McLure and Pardo suggest that reducing the tax administrator’s dependence on facts that are difficult to verify, and reducing the number of individuals who are required to file returns, are important factors in improving the ease of administration. Accomplishing these two objectives sometimes requires making exceptions to the “treat alike” rule (although in some of the examples cited by the authors, it also means sticking to the “treat alike” rule).

Approximation

One of the most important methods of accomplishing McLure and Pardo’s goal is to substitute approximations of certain types of income for actual income, and to restrict the applicability of those approximations (or restrict their effect) to certain taxpayers. As discussed above, while tax policy theory suggests that an income tax should be based upon the taxpayer’s global income, the accurate measurement of global income can often be difficult. In general, the greater the accuracy of measurement, the more emphasis will be placed on facts that will be difficult to verify and that will vary from individual to individual. Other things being equal, the rougher the approximation of income, the easier the system will be to administer.

Individuals will normally have income from various sources. The process of approximating income and collecting tax will involve two steps. The first will be to make relatively accurate assessments of some types of income. The second will be to approximate other types of income. In many instances, “approximation” will involve exempting certain amounts of certain types of income (see Bird (1983, p. 4)).

In order to make accurate assessments of income with ease, that income must be relatively easy to identify and verify. Tax on that income must also be relatively easy to collect. As will be discussed below regarding employment income, for most taxpayers, virtually all of their income will fulfil these criteria.

It should be noted, however, that it will not always be easy to identify and verify all substantial sources of income. In some cases, large amounts of income will not be easy to identify. In those instances either approximation (and simplified collection techniques) or relatively difficult, expensive, and time-consuming evaluation and collection techniques will be required.

Of particular importance is the fact that there are fewer taxpayers at the top of the income pyramid than at the bottom. Relatively speaking, by limiting income approximations to taxpayers from the relatively lower-income groups, and by making more accurate income evaluations of taxpayers from the relatively higher-income groups, the number of taxpayers whom the tax administration would need to examine carefully would be limited. Therefore, in the trade-off between accuracy and ease of administration, income approximations should be limited to (1) smaller amounts of income; (2) a smaller part of a taxpayer’s total income; and (3) to the lower-income earners. On the other hand, more accurate measurement should be made of income that is (1) relatively easier to assess and verify; (2) relatively large; (3) constitutes the bulk of a taxpayer’s income; and (4) is earned by wealthier taxpayers.

When referring to “income,” it should be understood that income is equal to gross income minus deductions. Those issues regarding verification and approximation of gross income also apply to deductions. Deductions are discussed in greater detail below.

Wage Income

Certain types of income are relatively easy to identify and verify, and so can be taxed easily. Such income does not need to be approximated. The most important is wage income. Depending upon how much of an economy is organized into entities, and the extent to which the organized sector is concentrated in relatively larger firms, wages can constitute a substantial portion of an economy’s personal income. At the lower end of the economic spectrum, they often constitute virtually all of a taxpayer’s income.

Salary income is usually easier to verify than other types of income. Companies can be required not only to report gross wages to the tax authorities but also to calculate the amount of tax actually due. If the taxpayer has no other income or deductions (a big if, which is addressed below), the company’s calculation will equal the employee’s final tax due. The company can then remit the taxpayer’s final income tax due as part of a wage withholding system, making the filing of a return by the employee unnecessary. In effect, the taxpayer need have no contact with the tax authorities at all. The employer acts as a “mini-assessing authority” for the government (see Bird (1983, p. 6)).

Audits at the company level are easier and more efficient to perform than audits at the level of the individual taxpayer. Deductions at the company level for wages paid are linked to the amount of wages declared as income to the employee, and the relative ease of audit means that substantial tax penalties for misreporting employee wage income have a deterrent effect.

Therefore, companies should be required to compute, withhold, and remit income taxes due on the wage income of employees. Which entities should count as “companies” for withholding purposes is an empirical question. Often, the answer is any entity registered or incorporated, or any entity with more than a small number of employees, or any entity with a capital value or turnover in excess of a small amount.

The obvious problem with treating withholding tax on wages as a final tax lies in the phrase, “if the taxpayer has no other income or deductions.” However, in most cases the lion’s share of most taxpayers’ income will be employment income, and deductions will be few. The treatment of nonwage income, and of deductions, is addressed further below.

Another problem arises when an employee has more than one salaried employment. In a system without graduated rates (including a zero rate) of income tax, this would not be a problem. However, the existence of graduated rates means that a single taxpayer would often owe more tax on the same income than if it were earned by two different taxpayers.

In systems that maintain substantially graduated rates, and where multiple wage employment is common, this problem can be alleviated by requiring the taxpayer to provide information to his employers about his other employment. Although this complicates the system somewhat, the use of a unique taxpayer identification numbering system and simple cross-checking may make the administrative burden less onerous. How effective this technique would be will depend on the particular features of the economy, including the degree of concentration of the wage economy into a limited number of firms, the extent to which salaried individuals work at different jobs, the completeness of taxpayer identification records, and the degree of computerization of the tax administration.

McLure and Pardo discuss certain taxpayer-specific criteria that some jurisdictions use for specific tax policy reasons to reduce tax, such as the number of dependent children. If supplied by the taxpayer, these criteria could be used by the employer to compute total tax owed. As discussed earlier in this paper, distinctions among different taxpayers should be avoided wherever possible. This is particularly true when, as McLure and Pardo point out, the information is hard for the employer, and the tax authorities, to verify.

In certain circumstances, however, policy planners will insist on deductions being based on such information. One way of reducing problems is to disallow cumulative deductions beyond some amount. Limiting deductions in this way, which is really a variation of income approximation, is discussed below. When a deduction is based on information that is difficult to verify, a specific limit on the amount of deduction can also be helpful. For example, a deduction could be allowed for each child, up to a limit of two. It might then simply be assumed that everyone will declare two deductions.

Business Income

It is much more difficult to estimate the income of farmers, sole proprietors, and small partnerships. When a taxpayer’s total income consists of only a small amount of business income, it might make sense to exempt such income from tax. Excluding income from the tax base is merely another method of approximating income. The issues involved in excluding minimum types of income is discussed below.

Taxing business income in excess of the exempt amount is more problematic than taxing employment income. It requires some contact between tax administrators and taxpayers, since there is no easily identifiable third party to assess the amount of tax due and to collect it.

A possible alternative is to enact withholding on certain types of payments to the taxpayer. Such withholding would not only serve as a mechanism for ensuring that tax payments are remitted but would also serve as an approximation of tax due by the taxpayer. Estimates can be made as to the average profit made on a particular input (for example, purchases) or output (for example, sales of services). Tax can then be withheld on a gross basis, with a rate that yields an amount approximately equal to the tax on the estimated profit. The sum of gross withholding taxes could be treated as a final tax. Which inputs and/or outputs should be subject to such withholding depends on the particular facts and circumstances unique to the economy. However, inputs and/or outputs of businesses that normally have income that is relatively greater than those at the bottom of the income spectrum, but that are particularly hard to tax, might be good subjects for such withholding taxes. Such businesses might include the major professions, as well as contractors (see Gison (1984)).

Estimates of business income should be relied upon less when such income is relatively large or constitutes a large percentage of the taxpayer’s income, or constitutes income of taxpayers at the upper level of the income spectrum. Of course, the exemption discussed earlier accomplishes some of this. Gross withholding taxes should not serve as final tax on large amounts of business income, although they may serve as a minimum tax in some cases.

Taxpayers with sufficiently large amounts of business income should be required to keep detailed books of account. The income of such taxpayers would, if large enough, justify the difficult and time-consuming audits necessary to verify the taxable income computed by the taxpayer.

For taxpayers whose business income lies between the exempt amount and the amount for which books of account would be required, it will often make sense to make approximations of income more accurate than withholding but less complicated than keeping books of account. Such approximations would be based upon simple and relatively easily verified facts, such as a forfait-style system. These approximations could also serve as a minimum tax for the books of account cases. In some instances, one might want to give taxpayers the opportunity to challenge the forfait-type approximations.

The cut-off amounts for exemption, approximation, and books of account systems, as well as the methods of implementation to be used, need to be determined by an empirical analysis of the facts in the particular economy.

Exclusions

The first and most obvious type of exclusion is the zero tax bracket. Excluding the lowest-income earners from the income tax system entirely can eliminate a large number of filers. However, in excluding these taxpayers, attention must be paid to the operation of any withholding system. As noted above, withholding can serve as a method of approximating income tax due as well as a method of ensuring collections. Withholding on payments made to tax-exempt taxpayers would still act as final gross taxes, an issue discussed in greater detail below. The refunding of such withholding would necessarily complicate tax administration and require the exempt taxpayers to file.

One of the advantages of salary income is that the company can determine the amount of gross wages to be paid to an employee over the course of the year. However, income received from outside sources cannot easily be taken into account by the employer.

This problem can be dealt with by having the taxpayer report all income to the employer. The employer could then adjust the withholding, using a pay-as-you-earn (PAYE) system. In such systems, the employer, in effect, becomes the tax administrator.

But such a system creates a number of problems. Verifying that the employee has in fact reported all income to the employer can be difficult indeed. Moreover, a substantial amount of additional paperwork would be involved between employee, employer, and tax administration.

Excluding some such income from the income tax base would be an administratively easier method of solving the problem, as suggested above with regard to business income. Employees at the lower end of the income scale are unlikely to have much income outside of their salary or part-time business income. By excluding such income from tax, only the employees’ salary income would be subject to tax. This would greatly increase ease of administration.

Probably the most common form of such income is bank interest. As noted above, part-time business income may also be important. In some economies, although in few, if any, developing countries, income from dividends and capital gains may also be important, particularly where mutual funds are a common form of investment for taxpayers at the relatively lower end of the income scale.

Other forms of nonlabor income, such as rent or royalties, are likely to be earned only by a relatively small number of people, or are likely to be earned only by upper-income wage earners. In the case of such income, the administrative benefits of exclusion are unlikely to outweigh other considerations; they therefore need not normally be subject to an exclusion. The one exception to this rule might be imputed rent on owner-occupied housing.

Relying again on the principles of income approximation outlined above, nonlabor income should be excluded (1) only if it is small; (2) only to the extent to which it constitutes a relatively small portion of the taxpayer’s income; and (3) so that, where possible, it primarily affects taxpayers at the lower end of the income spectrum. The amount of exclusion should be set low enough to satisfy the above criteria, but high enough to ensure that most taxpayers do not need to file returns. Lower-income taxpayers for whom nonlabor income constitutes a substantial percentage of total income would still be taxed on a substantial portion of that nonlabor income. Upper-income taxpayers, whose income normally includes a substantial amount of nonlabor income, would be taxed on most (relatively speaking) of that nonlabor income. Estimation would be largely confined to lower-income taxpayers and would apply only to a relatively small proportion of the lower-income taxpayer’s income. Administrative resources would be concentrated on more accurate evaluation of the real income of the few upper-income taxpayers.

In most instances, a single fixed amount of exclusion should be set for the sum of interest, dividends, and capital gains. In other instances, separate amounts might be set for each category. The optimal nature and size of fixed amounts can be determined only by empirical analysis.

Deductions

As with gross income, deductions of expenses related to the earning of income, as well as deductions for special incentives, create two problems, Deductions may be difficult to verify, and, by allowing deductions, income computation is made more complex. Also, as with gross income, deductions can be estimated and limited, particularly with regard to the smaller portion of a taxpayer’s income and to relatively lower-income earners.

First, as discussed above regarding PAYE systems, deductions that have little or no economic basis (meaning that they are not legitimate expenses in the earning of income) should be disallowed, particularly deductions that are difficult to verify. In some instances, “tax expenditures” that have valid social goals could be replaced with spending targeted to the intended groups of beneficiaries. They can also be limited in other ways, as suggested earlier.

Some other deductions might also be denied, even though they may have some economic validity. These might include deductions that are likely to have a personal component, or a component that is hard to value and therefore prone to abuse. Expenses relating to passive or investment income, commuting expenses, and entertainment expenses usually fall into this category.

As with the exclusion of nonlabor income, only deductions in excess of a certain amount should be taken into account by the tax system. The rationale of such a system is virtually identical to the rationale for excluding minimum amounts of nonlabor income.

In most instances, a single fixed amount can be set for the sum of all allowable deductions. In other instances, some of which are discussed below, separate limitations might be set for different categories of deductions.

While the exclusion of income from tax is generally likely to be popular with taxpayers, excluding deductions is likely not to be. Therefore, it may make political sense to provide a general deduction for taxpayers, up to the excluded amount. Rate brackets can be adjusted accordingly.

Withholding

The objectives of withholding have already been discussed. In principle, withholding rates must be set based upon factual information as to (1) what a particular taxpayer’s liability should be and (2) what rate of withholding is likely most closely to approximate that liability.

First, it is necessary to determine what income should be subject to withholding. As noted above, employment income in firms with over a certain number of employees, as well as certain other payments, should be subject to withholding.

In a system with graduated rates, taxpayers’ liabilities should differ based upon their total income (see Dixon (1985, pp. 39-40)). Withholding rates should, therefore, vary on the basis of the personal circumstances of any particular taxpayer. The lowest-income taxpayer should suffer no withholding, while the highest-income taxpayers should suffer relatively higher withholding.

With regard to wage income, the personal circumstances of a taxpayer can be taken into account. Withholding at the company level can be based upon the total amount of wages that the taxpayer will earn. But with regard to other types of withholding, it would be administratively difficult to vary withholding based upon the personal circumstances of the individual taxpayer.

Determining the rate of withholding likely to approximate a taxpayer’s final liability is also difficult. An income tax is, by its very nature, based on net income, not upon gross amounts. One cannot tell with any degree of certainty what a taxpayer’s final income will be by looking at any gross amount that contributes to the computation of that income, whether the gross amount be interest, dividends, or gross input purchases.

Some factual analysis can help set withholding amounts. In general, for most individual taxpayers, interest and dividend income is net income; in a pure global income tax system, interest and dividends would properly be taxed at the taxpayer’s marginal rate. The fact that withheld amounts will be final taxes primarily for taxpayers at the lower end of the income scale suggests, however, that these amounts should be less than at the top taxpayer’s marginal rate.

The rate of the most typical taxpayer would probably be appropriate for most interest income. The rate for dividend withholding might possibly be set higher, if it is likelier that dividend recipients would be higher-income taxpayers than interest recipients (see Jenkins and others (1991, p. 10)). However, such differentiation might set up a disincentive for equity investment or create an incentive to disguise dividends as interest.

Withholding on other forms of periodic income would also vary depending on circumstances. Withholding on inputs or outputs in a business enterprise can be even more problematic. Empirical analysis is of even greater importance here.

Loss Limitations

Several other approximations can help make an income tax system more easily administrable. One, discussed earlier, relates to the usual “realization” requirement before capital gains or losses are taken into income. Because taxpayers may control realization, limits on the deductibility of capital losses—allowing them only to be deducted up to the amount of capital gains—restricts the ability of the taxpayer to realize only losses.

Other limitations include restricting the deductibility of business costs and expenses only against business income, passive losses against passive income, or foreign losses against foreign income. In these instances, limitations on losses make it easier for a tax administration to avoid repeated detailed investigations as to the economic validity of the losses.

Some Techniques for Enforcing the Substantive Law

Many of the substantive tax law changes outlined above involve the use of income approximation, particularly for those at the lower end of the income scale. Different types of approximations, based on exclusions, withholding, and certain types of loss limitations, are based both on the form and the amount of income. In general, the goal is to reduce the information needed to assess tax due, and, in turn, to make it unnecessary for large numbers of taxpayers to file returns.

Verification

Verification of the relevant information will always be necessary. For lower incomes, this will often mean verification of employment income, which was discussed above. But it is also necessary to verify who actually is at the lower end of income distribution.

Withholding allows tax authorities to collect at least some of the tax that is likely to be owed to the government. In some cases, it also allows for a final approximation of income tax due. Of equal importance is the reporting of various income-producing transactions, particularly wages and passive income. Such reporting can help tax administrations to verify how much income should be subject to exclusion and whether any additional tax is due. In most instances, transactions subject to reporting requirements will also be transactions subject to withholding. This may not always be the case, however.

The use of taxpayer identification numbers, and the reporting of income by identification number to the tax authority, can play an important role in this process, even in developing countries. A taxpayer identification system can allow the matching of income, particularly employment, interest, and dividend income. This matching can determine whether or not a taxpayer has exceeded various exclusion amounts and whether he should have filed a tax return. As McLure and Pardo point out, the services of banks can often be used in assisting the tax administration in this task.

Reporting of types of transactions involved in business income, along with any withholding, can also help identify taxpayers who should be filing returns. The reporting of transactions by identification number can be useful when verifying the returns of taxpayers who exceed the amounts for exclusions and who should be filing.

As McLure and Pardo note, there is always the danger that tax administrations might be flooded with information that they cannot efficiently use. They discuss the benefits of the requirement that, whenever possible, information be supplied in machine-readable form. What information should be reported, as suggested above in the various discussions regarding withholding, would depend on the facts and circumstances of the particular economy.

As McLure and Pardo also discuss, there are other sources of information, from bank and credit card accounts to notices of real estate transactions. These may reveal whether taxpayers have been concealing income that is statutorily subject to tax and whether they (and perhaps withholding agents) should have filed returns.

Penalties

I have written elsewhere in detail about the design of taxpayer penalties (Gordon (1990)). Only a few points are emphasized here.

Penalties for failure to carry out a legally required act should be designed to encourage the taxpayer to carry out that act. Optimal penalties are often expressed in terms of probability of detection and potential tax savings (Mansfield (1987, pp. 18-19)). But, as with any aspect of tax administration, the most effective penalties are those that relate to activities most easily discovered and verified. Therefore, they should be used sparingly to increase the likelihood that transactions that are difficult to discover or verify are brought into the tax net.

Some of the most effective penalties are those that apply to withholding agents: employers, companies, and banks. Because the transactions are relatively easy to define, because the amount of tax owed is relatively easy to evaluate, and because the taxpayers are therefore relatively easy to audit, stiff penalties can act as a substantial deterrent. With regard to most withholding agents, investigation for failure to carry out withholding and reporting requirements is often quick and easily resolved.

Interest charged on late payments is not a “penalty” at all, but an adjustment designed to ensure that neither taxpayer nor tax authority gains the time value of money from a delayed payment or refund. However, an interest-like penalty can be used to encourage taxpayers to settle disputes. McLure and Pardo discuss a sliding scale of penalties depending on when taxpayers settle claims. Another possibility is to base the applicable penalty on a multiple of the applicable interest rate. This would ensure that for each day a taxpayer failed to settle, his penalty would increase at a rate greater than the cost of money.

Conclusions

The design of the substantive income tax law can play a crucial role in determining the ease of tax administration. To ensure a well-designed tax law, it is necessary to identify effective ways of easing implementation and to scrutinize carefully the facts unique to the particular economy. Careful legislative and administrative drafting are also required. As the process of tax reform continues, the process of identification of methods, examination of facts, and drafting of provisions will continue to improve. The work of McLure and Pardo contributes substantially to this process.

REFERENCES

    AlmJamesRoy Bahl and Matthew Murray “Income Tax Evasion,” in The Jamaican Tax Reformed. by Roy Bahl (Cambridge, Massachusetts: Lincoln Institute of Land Policy1990).

    BahlRoy “The Political Economy of the Jamaican Tax Reform,” in Tax Reform in Developing Countriesed. by Malcolm Gillis (Durham, North Carolina: Duke University Press1989).

    BirdRichard M. “The Administrative Dimension of Tax Reform in Developing Countries,” in Tax Reform in Developing Countriesed. by Malcolm Gillis (Durham, North Carolina: Duke University Press1989).

    Casanegra de JantscherMilka “Administering the VAT,” in Value-Added Tax in Developing Countriesed. by Malcolm GillisCarl S. Shoup and Gerardo P. Sicat (Washington: World Bank1990).

    GillisMalcolm and Charles E. McLure Jr.La Reforma Tributaria Colombiana de 1974 (Bogotá: Biblioteca Banco Popular1977).

    McLureCharlesE. Jr. “Income and Complementary Taxes” (unpublished1982).

    McLureCharlesE. Jr. “Analysis and Reform of the Colombian Tax System,” in Tax Reform in Developing Countriesed. by MalcolmGillis (Durham, North Carolina: Duke University Press1989).

    McLureCharlesE. Jr. John MuttiVictor Thuronyi and George R. ZodrowThe Taxation of Income from Business and Capital in Colombia (Bogotá: Ministerio de Hacienda y Crédito Público, 1988; Durham, North Carolina: Duke University Press1990).

    McLureCharles E. Jr. and George R. Zodrow “Tax Reform in Colombia: Process and Results,” in Tax Policy in Developing Countriesed. by Javad Khalilzadeh-Shirazi and Anwar Shah (Washington: World Bank1991).

    MusgraveRichard A. and Malcolm GillisFiscal Reform for Colombia: The Final Report and Staff Papers of the Colombian Commission on Tax Reform (Cambridge, Massachusetts: Harvard University International Tax Program1971).

    Perry R. Guillermo and Mauricio Cárdenas S.Diez Años de Reformas Tributarias en Colombia (Bogota: Empresa Editorial, Universidad Nacional de Colombia1986).

    UrrutiaMiguel “The Politics of Fiscal Policy in Colombia,” in The Political Economy of Fiscal Policyed. by Miguel UrrutiaShinichi Ichimura and Setsuko Yukawa (Tokyo: United Nations University1989).

For detailed discussion and evaluation of the recent history of tax policy in Colombia, as well as references to literature on that topic, see McLure (1989) and McLure and Zodrow (1991). On the landmark 1974 reforms, see Perry and Cardenas (1986) and Urrutia (1989).

McLure (1982) called for a high-level commission of experts—analogous to the 1968-69 Musgrave Commission on tax policy—to examine tax administration.

Casanegra de Jantscher (1990, p. 179). McLure (1982) noted that “the finest tax structure can be subverted by tax avoidance made possible by carelessly drafted statutes and regulations and especially by tax evasion facilitated by poor tax administration.”

The structural changes are described and analyzed in McLure and others (199).

In the pre-1986 system, most taxpayers subject only to withholding claimed the largest number of personal exemptions possible—“three or more.” Those filing returns could claim even larger numbers of dependents. Though the extent of cheating has not been documented, that it was widespread is part of the conventional wisdom in Colombia.

The net wealth tax was also eliminated under the 1988 reforms, to come into effect in 1992. This was not done for administrative reasons, but to lower the tax-induced incentives to invest abroad. This change reduced the progressivity of the system. The net wealth tax is applied to the wealth of individuals, on a base consisting of assets minus liabilities. In 1989 nearly 700,000 individuals were subject to this tax; these constitute the universe of income tax filers. The rate is progressive, ranging from zero to 1.8 percent. The 1987 revenues from this tax were C$34,000 million, or 8 percent of total collections from the income and complementary taxes (C$427,000 million). In 1989, C$48,000 million was collected, compared with total income tax revenue of C$590,000 million. As the tabulation below shows, net wealth tax collection is concentrated in the 20 percent of taxpayers with net worth in excess of C$10 million, who pay 80 percent of the tax.

Ranges of Net Wealth (In Colombian pesos)Number of TaxpayersAccumulated PercentageNet Wealth Tax (In Colombian pesos)Accumulated Percentage
0–5,000,000330,59348.71,3742.8
5,000,000–10,000,000221,34181.27,70618.9
10,000,000–50,000,000114,64298.318,97358.3
50,000,000 or above11,338100.020,067100.0
Total677,91448,120

Eliminating the filing requirement for those who have net wealth of less than C$15,000 million and meet the above-mentioned conditions should cause a maximum loss of revenues estimated at 1.6 percent of all income and complementary tax collections. The loss is presumably even less, since many taxpayers with net wealth of less than C$15,000 million receive income from nonlabor sources and thus are required to file an income tax return and pay income and net wealth taxes.

The taxation of dividends was also eliminated. This is unimportant for most Colombian taxpayers. Interest income is subject to withholding at a rate of 7 percent; amounts withheld can be credited against tax liability on global income.

These limits are for the 1990 tax year; at the time of enactment in 1986, the two limits were C$6 million and C$4 million, respectively. In December 1990, the exchange rate was roughly C$600 per US$1.

The 1974 reform had eliminated most exemptions for nonlabor income. For constitutional reasons, exemptions for labor income could not be eliminated under the emergency procedures used at that time. Thus, payments such as vacation pay, “primas” (the thirteenth month salary), severance pay, pensions, and per diem payments had remained exempt.

False tax clearance certificates were said to sell for C$10,000, the equivalent of about US$16.

In Chile, by comparison, banks accept returns only from those with tax liability. Moreover, Chilean banks do not process data on tax returns.

This figure was initially set at C$6 million, which was so low that for five months—until it was revised to C$120 million—the reform threatened to induce a financial panic as depositors withdrew funds to avoid detection by the fiscal authorities.

It appears, however, that the use of credit cards is decreasing for other reasons, notably a government requirement that 30 percent of the initial balance must be paid in the first month after debt is incurred.

At an after-tax interest rate of 20 percent a year, a three-year postponement reduces present value by 42 percent.

The largest 2,000 taxpayers in the country—all companies—account for 80 percent of all revenue. Some 800 of these (accounting for 50-55 percent of all revenue) are located in Bogota. There are 40,000 other companies, as well as 2.5 million individual taxpayers.

For an analysis of the deductions allowed in Colombia in the late 1960s, see Musgrave and Gillis (1971); on the credits allowed under the 1974 reforms, see Gillis and McLure (1977).

On the other hand, the need to allocate capital income among members of a family complicates matters.

See Alm, Bahl, and Murray (1990); this experience is summarized in Bahl (1989).

REFERENCES

    BirdRichard M. “Income Tax Reform in Developing Countries: The Administrative Dimension,”Bulletin for International Fiscal DocumentationVol. 37No. 1 (1983) pp. 314.

    DixonDaryl “Tax Avoidance and Withholding Tax,”Australian Tax ForumVol. 2No. 1 (Autumn1985). pp. 3352.

    GisonCornelio C. “The Philippine Expanded Withholding Tax System,”Asian-Pacific Tax and Investment BulletinVol. 2No. 3 (1984) pp. 95100.

    GordonRichard K. Jr. “Income Tax Compliance and Sanctions in Developing Countries,” in Taxation in Developing Countriesed. by Richard M. Bird and Oliver Oldman (Baltimore: Johns Hopkins University Press4th ed. 1990).

    JenkinsGlennJames Owens and IanRoxan “Income Tax Reform for Zambia: Administration and Policy” (unpublished; Cambridge, Massachusetts: International Tax Program, Harvard University1991).

    MansfieldCharles Y. “Tax Administration in Developing Countries: An Economic Perspective,” IMF Working Paper No. 87/42 (Washington: International Monetary FundJune1987).

Although these fiscal problems created the conditions that brought about reform, it should be noted that the best condition for changing taxes is certainly not during periods with the largest fiscal deficits. Perhaps the lost period (1977-82) would have been much more appropriate for modifying the tax system, given the extrabudgetary considerations.

The stamp lax, an administrative nightmare which generates very little revenue, was inexplicably maintained.

In Mexico, data processing systems began to be changed in 1988 and the process has now been completed.

    Other Resources Citing This Publication