8 Improving Tax Compliance

Richard Bird, and Milka Casanegra de Jantscher
Published Date:
September 1992
  • ShareShare
Show Summary Details
Carlos A. Silvani

This paper comments on the experiences of several countries with regard to tax compliance and suggests certain general guidelines for improving it, particularly in countries where there is a relatively high level of noncompliance.

The goal of tax administration is to foster voluntary tax compliance. Penalizing tax evaders or going after delinquent taxpayers are not in themselves the object of tax administration. Voluntary compliance may be encouraged, however, if the administration is successful in establishing a strong prospect that noncompliance will be detected and effectively punished.

Tax compliance will thus be furthered if there is an effective tax administration. An effective tax administration should not be confused with an efficient tax administration. An administration may be efficient in that its tax collection costs are very low, yet at the same time it may be ineffective if it is unable to enforce compliance. The effectiveness of the administration is not the only determinant of the level of voluntary compliance, but it is likely to be the key factor, especially in countries in which there is a high level of noncompliance. Many reasons contribute to evasive behavior, including the degree to which the community supports the government and its economic policies.1 All the same, taxpayers will comply better if they believe that failure to do so will mean assuming a substantial risk of being penalized in a relatively severe fashion. Were this not so, there would be no way of explaining how it is that in Chile, in the late 1970s, with a military government that had limited popular support, there was nevertheless a marked rise in tax compliance. In Argentina, on the other hand, the late 1980s and early 1990s registered the lowest rates of tax compliance in that country’s history, even though the government holding office at that time had the fullest support of the community. In Bolivia, with the advent of a democratic government, tax administration improved. In Colombia, the change is absolutely independent of popular support for the government, and the same could be said of Spain and Portugal. In other words, the level of tax compliance seems relatively independent of the taxpayers’ degree of acceptance of their government and its policies.

The effectiveness of the tax administration is of course not the only determinant of tax compliance. It is enough to note in this connection the negative influence that macroeconomic variables such as the rate of interest and inflation may exert on tax compliance. Nor can it be doubted that a lower cost to taxpayers of complying with the system, its fairness, the simplicity of its laws and procedures, and the services that the tax administration provides to taxpayers are all important factors in expediting and stimulating voluntary compliance. Nevertheless, in countries with a very high degree of noncompliance, the ability of the tax administration to impose effective penalties is perhaps the key to shaping the behavior of taxpayers.

If this is so, then the problem is that of making tax administration effective. Tax administration will be effective if it is able to deal with the following key shortfalls:

(1) Unregistered taxpayers: The first shortfall originates in the gap between potential taxpayers and registered taxpayers.

(2) Stopfiling taxpayers: The second shortfall reflects the difference between registered taxpayers and those who file returns.

(3) Tax evaders: The third is the difference between the tax reported by taxpayers and the potential tax according to the law.

(4) Delinquent taxpayers: The fourth and last gap is the one between the amount of taxes that taxpayers report owing, or that the tax administration may eventually assess, and the tax actually paid by taxpayers.

If tax compliance is to improve, the tax administration must take effective action to deal with these shortfalls. Accordingly, when evaluating a tax administration, objective data must be looked at in order to estimate the degree of effectiveness displayed by the administration in dealing with each of these gaps. And important as it is to learn how effective the administration is in dealing with each gap, the truly key point is to determine the overall degree of effectiveness of the administration. If the administration is able to effectively control only one of these gaps, noncompliance will shift to the gap where the administration exercises weaker control. It is not enough, for instance, to make strong efforts to eliminate the informal economy and register all potential taxpayers if registered taxpayers cannot then be made to file returns and to report and pay the proper amount of tax. This point is particularly important in designing tax policy as well as in allocating tax administration resources.

I. Monitoring Tax Compliance

This section outlines some ideas for narrowing the gaps described above.

Unregistered Taxpayers

Obviously, one prerequisite for learning whether a taxpayer is registered is to have a reliable and efficient single register of taxpayers (SRT). Such a register is needed not only to control this gap but also to provide the basic hinge on which will turn many of the activities to be carried out in order to narrow the other gaps.

Single Register of Taxpayers

The single register of taxpayers should consist of a taxpayer identification number (TIN) and of information associated with each TIN. The data that should be associated with each TIN are name, surnames (distinguishing between paternal and maternal surnames), commercial business name, address, business activity, and type of tax payable by the taxpayer. For control purposes, and depending on the characteristics of the tax system, it will be important in some countries to establish a connection between the TIN of a partnership or a corporation and the TIN of its members, indicating their percentage interest in the entity. Also connected with the TIN of companies (whether sole proprietorships or legal entities) should be the location of branches, if there are any.

Taxpayer Identification Number (TIN)

Ideally, there should be a single TIN capable of identifying the taxpayer in connection with every tax. In some countries, this would not be a realistic solution over the short term, but it should remain a desirable long-term goal. The TIN should be the same number assigned to individuals for purposes of civilian identification—the identity card number, for instance—so long as this number is consistent with all the requirements of a highly reliable TIN. If the TIN is the same as the identity card number, it is easier to cross-check information when monitoring taxpayer compliance. But even where the TIN is the same as, say, the number on the identity card, the tax administration will still have to make sure that (1) no more than one TIN is assigned to any one taxpayer and (2) each TIN is assigned to only one taxpayer. This means that the assignment of TINs, or at least its control, should be centralized.

The TIN should not include any taxpayer characteristic likely to change over time, such as the business activity or the location of the taxpayer’s domicile. Otherwise, a problem might arise because the TIN would need to change along with any of those circumstances. In principle, the best TIN is simply sequential. However, if the TIN is not sequential and some of its digits reflect important information, that information should be permanent and unchanging: the date or place of birth of the individual, for example. As for the TIN of companies, only one TIN should be assigned to each company; branches of the company, therefore, should not have a different TIN from that of the head office.

Keeping Domiciles Current

As with any dynamic data base, keeping the single register of taxpayers current is as important as constructing it. Out of all the data in the register the key piece of information to keep current is domicile, because the administration will lose touch with the taxpayer whose domicile is not known with precision. In Colombia in 1986, for instance, before the revamping of the tax administration, computer printouts of value-added tax (VAT) forms mailed out to taxpayers had to be suspended because more than 20 percent of the forms were being returned by the post office for having the wrong address.

Experience has shown that requiring taxpayers to report a change of address and establishing penalties for those who fail to do so is not enough. In some cases, the most practical solution is to record once a year all addresses reported on, say, tax returns, even though the percentage of taxpayers filing returns with a different address than the one registered in the single register of taxpayers is likely to be under 10 percent. An alternative solution is to require tax returns (VAT or income tax) to be filed with identification labels preprinted by the tax administration. In this case, all that needs to be done is to pick out the corrections made by taxpayers on the preprinted labels. Having the tax administration make available preprinted identification labels is a highly recommended practice not only because it helps to update domiciles but also because it guarantees that the taxpayer will correctly use the TIN assigned under the single register of taxpayers.

Business Activity Code

A business activity code should be connected with each taxpayer. This code should preferably follow the structure of the Uniform International Industrial Code (UIIC) to make it easier to draw comparisons with the country’s national accounts and with international statistics. To begin with, some three hundred different economic activities ought to be enough.2

From time to time, the code of business activity assigned to each major taxpayer should be reviewed. An error resulting from an improper code being assigned to a major taxpayer may cause substantial statistical distortions. Tax returns should ask for (and every so often pick out) the alphabetical description of the business activity to facilitate such review.

Detecting Unregistered Taxpayers

As for detecting unregistered taxpayers, the problem lies primarily in finding the relatively small taxpayers. In countries such as Peru, however, where the informal economy is very large, unregistered taxpayers may also be of medium size. Unregistered taxpayers may be detected chiefly by two means:

(1) The first option involves searches or inspections (batidas) in which a group of officials goes door-to-door within a particular geographical area, checking that all persons or establishments engaged in a taxable business activity in that area are properly registered (see Silvani and Radano, Chapter 2 in this volume).

(2) The second option is to go through registered taxpayers in order to detect their suppliers or customers who are not registered taxpayers. This may be done by checking whether purchase invoices being deducted and sales invoices being issued show TINs for taxpayers registered in the single register of taxpayers. If withholding systems are in place, unregistered taxpayers may also be detected by crosschecking the data against the information supplied by withholding agents with respect to transactions in which taxes have been withheld.

On the other hand, problems arise in some cases because the tax administration massively registers persons who are not, and should not, be taxpayers. In these cases, the problem created by “overregistration” of taxpayers is as bad as, or worse than, that of locating unregistered taxpayers. Examples of this situation were observed some years ago in Argentina, Brazil, Peru, and Portugal.

Registering a person and assigning that person a TIN is not too heavy a burden for the tax administration, as long as the tax obligations incumbent on such persons are precisely defined. Persons who are not active taxpayers and are therefore under no obligation to file tax returns should not be confused with active taxpayers. Otherwise, a large number of taxpayers will appear to have filed no tax returns, even though they have no obligation to do so, and this will needlessly hamper the work of the tax administration.

Taxpayers Filing No Returns

In many countries, the ratio of stopfiling taxpayers to total registered taxpayers is higher than 20 percent. Some examples are noted in Table 1.

Table 1.Ratio of Stopfiling VAT Taxpayers to Total Registered Taxpayers
CountryYearPercentage of Stopfiling VAT Taxpayers
Sri Lanka1990401
Source; IMF, Fiscal Affairs Department.

Relates to the tax on business turnover.

Relates to income tax.

Source; IMF, Fiscal Affairs Department.

Relates to the tax on business turnover.

Relates to income tax.

The ratios set out in Table 1 are far higher than desirable. This is usually due to three factors: (1) the single register of taxpayers is out of date and includes many taxpayers who are no longer in business; (2) some returns are filed with errors in the TIN and, accordingly, persons who have filed show up as stopfiling taxpayers; and (3) the tax administration does not deal systematically with stopfiling taxpayers.

The first two problems may be solved by properly updating the register and distributing preprinted identification labels to taxpayers, as described in the preceding section. As for the third problem, it is important that the administration should take steps to exert its control over major stopfiling taxpayers. In this connection, and in order to define an adequate strategy for monitoring stopfiling taxpayers, it is advisable to rank stopfiling taxpayers by size and by business activity. This is illustrated in Table 2.

Table 2.Chile: Percentage of Stopfiling VAT Taxpayers, 1987
Annual Sales (In thousands of U.S. dollars)Percentage of Stopfiling Taxpayers1
Taxpayer SizeMore thanUp toAgricultureIndustryCommerceAll activities
T113,217 or more004.41.9
T8Sales equal to zero16.223.717.515.6
All taxpayers10.213.913.914.9
Source: IMF Fiscal Affairs Department.

Stopfiling taxpayers mean those who did not file at least one monthly VAT return in 1987.

Source: IMF Fiscal Affairs Department.

Stopfiling taxpayers mean those who did not file at least one monthly VAT return in 1987.

Obviously, priority must be given to monitoring stopfiling taxpayers whose sales volume is highest. Control of these taxpayers must be monthly and should be directed in particular to monitoring major taxes, which are generally the VAT (or sales tax), excises, and taxes withheld by the taxpayer, if these taxes are monthly. Deferring action on these taxpayers may mean that the taxes owed to the treasury will be used by them for financing their own business activities. Such a situation should be detected and immediately corrected, because the accumulation of past due and unpaid taxes may climb to very high figures (particularly where withholding is involved) which may mean that after some time the taxpayer may no longer be able to pay even if he wishes to do so. If this should happen, the problem of delinquency may cease to be the company’s problem and may become, for the most part, the tax administration’s problem.

As for smaller stopfiling taxpayers, control over them could be exercised every six months or every year. Thus, once or twice a year, taxpayers would be required to file any returns that have been omitted and to pay the relevant fines. Imposing and collecting fines is essential in combating this kind of tax delinquency, for it is virtually the only enforcement tool available as a rule to tax administrations in this situation.3

In many countries, interest and penalties must be paid before or simultaneously with the filing of overdue tax returns. This makes it difficult to regularize delinquent taxpayers and diminishes the effectiveness of tax administration. The proper policy is to accept the amount of money that the taxpayer is in each instance ready to pay, billing him later on for any balance that might result from a review of the figures. In any event, whether or not payment of the penalties is demanded as a prerequisite for letting taxpayers file overdue returns, the information system should always monitor the penalties imposed. To require as a condition for accepting an overdue tax return that the taxpayer pay all interest and penalties in full merely causes the taxpayer (1) to lower the tax reported or (2) to remain as a stopfiling taxpayer. Tax returns should include two special boxes for the taxpayer himself to settle the interest and penalties due. This would be consistent with the general principle of self-assessment and the information system could check whether the taxpayer has correctly paid the penalties.

Delinquent Taxpayers

In a manner similar to the above suggestion, the tax administration should take swift and effective action on delinquent taxes, in order to make it clear that tax delinquency is a very expensive source of financing. As a rule, tax delinquency is very concentrated, that is to say, a small percentage of delinquent taxpayers accounts for a high percentage of unpaid taxes. Collection priorities should therefore be set consistently with the amount owed. In addition, priority should be accorded to collecting newer debts because usually it is easier to collect on newer debts. For this reason, every effort should be made to collect before the amount outstanding becomes uncollectible. Furthermore, in order to evaluate the effectiveness of collection, it is advisable that uncollectible debts be written off or at least set apart from files on collectible amounts. Concentrating on collecting the newer and more substantial debts will make it possible over time to reduce the ratio of delinquent debt to total collection.

Consideration should be given to assigning collection of small amounts—say under US$10,000—to private collection agencies. The reason is twofold: (1) private agencies would be paid only if they collect the amount owed, which ought to make them more effective than the tax agency; and (2) tax administrations are always short of resources and, in general, rarely collect small delinquent amounts.

To begin with, the tax administration could delegate to the private sector all out-of-court collection entailing completion of all steps preceding executory process and the enforced collection of the amount due. Second, judicial collection might be delegated—which in some countries would require legislative changes. Be that as it may, it would be advisable to have private sector companies deal with collecting minor delinquent amounts, even if it means paying out in commissions the full proceeds from penalties and even part of the tax collected. The idea is to change the behavior of taxpayers, because they will know that if they fail to pay on time they will inexorably have to pay surcharges on top of the tax.

Delegating to the private sector the collection of minor amounts of delinquent taxes also would save the tax administration money. Just as it is often advisable to turn tax collection over to banks, so it would be worthwhile to place collection of minor delinquent taxes in the hands of specialized agencies. The same idea is behind both propositions: to free up resources in order that the tax administration may concentrate on its basic functions.

As for large delinquent amounts, collection of these should not be left to the private sector. If the fee for collecting the amount is set as a percentage of the taxes recovered, large amounts of delinquent taxes could drive up the cost excessively. Besides, in many countries the largest tax debts are owed by government companies or large companies, and collection of the debt may be connected with the settlement of economic and social problems that lie beyond the scope of tax administration.

Control of Tax Evasion

As with any complex system, the effectiveness of a tax administration is determined by its weakest subsystem. Even if a tax administration is very effective in detecting stopfiling or delinquent taxpayers, the overall effectiveness of the administration will be low if it is unable to control evasion. Noncompliance will then take the form of evasion and will widen the existing gap between the amount of reported taxes and the potential tax foreseen by the law.

In order to gauge the effectiveness of the monitoring function and to determine the best strategy to follow, it is advisable to (1) analyze the data in tax returns filed by taxpayers and (2) compare actual tax receipts with the estimated potential tax receipts.

Analyzing the Data in Tax Returns

Depending on the data processing capacity of the tax administration, it may be possible to obtain statistical data that will provide some idea of taxpayer compliance and, in line with that information, guide the enforcement policy. Below, to illustrate, are some income tax and VAT statistics.

Corporate income tax

Table 3 shows the ratio between profits and reported sales in tax year 1985 in a selected European country for all economic activities.

Table 3.Percentage of Taxpayers According to Profits/Sales Ratio Reported for Income Tax, in a European Country, Tax Year 1985
All Economic Activities
Profits/SalesPercentage of tax payersAccumulaied percentage of taxpayers
From 0 to 1 percent12.537.8
From 1 to 2 percent11.449.2
From 2 to 3 percent16.165.3
From 3 to 4 percent13.478.7
Source: IMF, Fiscal Affairs Department.
Source: IMF, Fiscal Affairs Department.

Table 4 shows the percentage of taxpayers who in tax year 1985 reported profits lower than 3 percent of sales in the same European country, for selected business activities. By analyzing these data, the tax administration can gauge whether they reflect evasive behavior in any particular business activity, and direct its enforcement action accordingly.

Table 4.Percentage of Taxpayers According to Profits Reported for Income Tax in a European Country, Tax Year 1985
Business Activity CodeSelected Business ActivityPercentage of Taxpayers Reporting Profits Lower Than 3 Percent of Sales
3220Manufacture of wearing apparel, except footwear70.08
3311Sawmills and woodworking by mechanical means72.69
3320Manufacture of furniture, except metal and molded plastic furniture75.43
3699Manufacture of other nonmetal mineral products70.97
6105Wholesale trade in tools, hardware, and electrical appliances74.46
6205Retail trade in construction materials, metals, and tools71.77
6206Retail trade in automobiles, motorcycles, and bicycles76.76
Source: IMF. Fiscal Affairs Depart menl.
Source: IMF. Fiscal Affairs Depart menl.

Tables 5 and 6 show income tax data from a selected Latin American country for the tax year 1989.

Table 5.Percentage of Taxpayers According to Profits/Sales Ratio Reported for Income Tax, in a Latin American Country, Tax Year 19891
Profits/SalesPercentage of TaxpayersAccumulated Percentage of TaxpayersProfits/Sales Average Ratio
From 0 to 1 percent6.256.20.4
From 1 to 2 percent6.062.21.1
From 2 to 3 percent5.067.22.4
From 3 to 4 percent3.770.93.4
Source: IMF, Fiscal Affairs Department.

Covers only 7 percent of taxpayers contributing 87 percent of total tax revenues. Banking and insurance activities were not included.

Source: IMF, Fiscal Affairs Department.

Covers only 7 percent of taxpayers contributing 87 percent of total tax revenues. Banking and insurance activities were not included.

Table 6.Percentage of Taxpayers According to Profits/Assets Ratio Reported for Income Tax, in a Latin American Country, Tax Year 19891
Profits/AssetsPercentage of TaxpayersAccumulated Percentage of TaxpayersProfits/Assets Average Ratio
From 0 to 1 percent3.553.50.3
From 1 to 2 percent3.156.61.5
From 2 to 3 percent2.759.32.5
From 3 to 4 percent2.862.13.4
Source: IMF, Fiscal Affairs Department.

Covers only 7 percent of taxpayers contributing 87 percent of total tax revenues. Banking and insurance activities were not included.

Source: IMF, Fiscal Affairs Department.

Covers only 7 percent of taxpayers contributing 87 percent of total tax revenues. Banking and insurance activities were not included.

Tables 3 through 6 are meant merely to illustrate, and no comparison of the data set out in them is intended. In the countries taken as examples, however, this information was used as the basis for selecting the business activities and the taxpayers to be monitored, and in both instances the results were highly positive.


One piece of information frequently used to analyze compliance with the VAT is the markup reported. Markup is the ratio of all sales to all purchases reported for the VAT during a specific period—one year, for instance. A ratio or coefficient equal to or lower than 1 indicates that the taxpayer is reporting taxable annual sales equal to or lower than his taxable annual purchases. In other words, he is saying that he does not have a positive markup or “profit margin” between sales and purchases sufficient to defray the cost of salaries, financial costs, and other expenses not covered by the VAT.

Table 7 shows the markup reported by VAT taxpayers for all business activities in a selected European and in a selected Latin American country.

Table 7.Taxpayer Distribution According to Markup Reported for VAT
Percentage of Taxpayers
Latin American countryEuropean Country
Markup Reported for VAT1PercentageAccumulated percentagePercentageAccumulated percentage
0.01 to 0.504.
0.5 to 1.0024.328.916.124.7
1.0 to 1.1019.148.09.534.2
1.1 to 1.2010.658.69.443.6
1.2 to 1.306.465.07.250.8
1.3 to 1.404.369.35.856.6
1.4 to 1.503.172.44.761.3
1.5 to 1.602.474.83.965.2
1.6 to 1.701.976.73.268.4
1.7 to 1.801.778.42.771.1
1.8 to 1.901.479.82.373.4
1.9 to
2.0 to
Higher than 2.1017.9100.023.0100.0
Source: IMF, Fiscal Affairs Department.

Total yearly sales reported, divided by reported purchases for the same year.

Source: IMF, Fiscal Affairs Department.

Total yearly sales reported, divided by reported purchases for the same year.

As indicated earlier with respect to income tax, Table 7 is intended merely to illustrate the point by using as a specific example the statistical data which in some countries have proved very useful in defining enforcement strategies and guidelines for selecting taxpayers to be audited.

Experience has shown that taxpayers who report a low markup (under 1.10, for instance) have a strong likelihood of turning out to be tax evaders. Experience also indicates, however, that taxpayers with a very low markup (lower than 0.50, for example), can usually furnish reasonable explanations (such as being engaged in the export trade or in making large investments) and the low markup reported cannot necessarily be accounted for as tax evasion.

Estimating Tax Evasion

In some cases, estimating tax evasion may be important because choosing the correct enforcement strategy may depend on the level of evasion prevailing in a particular country. Annex I includes an outline of a methodology that has been used several times to estimate evasion of the VAT.

At one end of the scale are countries that have very low levels of evasion—below 10 percent, for instance. In these cases, the enforcement strategy and the system of penalties ought to be steered toward preserving the existing order, so that taxpayers who are complying do not change their behavior as a result of observing those who evade. The most advisable course of action in these cases is to impose drastic penalties and make an example of tax evaders. “Society must punish the bad in order that the good may remain good.”4

At the other end of the scale may be found those countries with tax evasion levels close to 50 percent. Here, both the problem and the strategy must be totally different, because it is a question of reversing the existing situation. We will deal with this topic below.

II. Guidelines for Tax Audit

This section discusses some general ideas which, taken as a frame of reference, are useful in establishing priorities and planning the allocation of audit resources.5 The audit function is of crucial importance to a tax administration; if it is not reasonably effective, tax administration will not be reasonably effective either.

To narrow the gap between the tax reported by taxpayers and the potential tax defined by law, an adequate audit plan must be put into practice. The plan must provide for broad coverage of the universe of taxpayers and must encompass diverse economic activities and taxpayers of various sizes. As described below, the various forms of noncompliance must be combated in the manner most suitable to each. In addition, a good audit plan requires special programs to prevent non-compliance. To implement such a plan, an adequate share of the tax administration’s manpower must be allocated to this task, but the reality is that many countries devote to tax audit only a small percentage of their tax administration personnel. Many tax administrations assign only 10 percent or 15 percent of their staff to audit. In countries where there is a low level of evasion, on the other hand, the percentage of personnel generally assigned to tax audit is about 40 percent.

The effectiveness of the audit function should not be measured in terms of straight tax receipts derived from additional assessment; what should be gauged is rather to what extent this function contributes to improving tax compliance. Measuring the effectiveness of tax audit by the amount of taxes collected through audits is like evaluating the authorities responsible for enforcing traffic laws on the basis of how much money they collect in fines from drivers. For both the tax administration and the traffic authorities the objective is to secure maximum voluntary compliance (that is, to minimize noncompliance) without regard to the revenues that may be directly obtained through control. It follows from this line of reasoning that the preferred course of action may be to make the presence of a traffic or a tax inspector conspicuous so as to forestall any breach of the law, even if this action is expected to yield no direct collection of revenue.

The audit function should preferably be evaluated in terms of the quantity and quality of audits carried out and the revenues “voluntarily” paid.6 It is advisable to encourage the filing of amended returns (corrections) in which the taxpayer acknowledges the mistakes or omissions made by him and detected through tax audit. Amended returns, being returns filed by the taxpayers themselves, are not subject to the costly and lengthy proceedings involved in tax disputes. Even if the taxpayer is not willing to acknowledge the full amount of the tax claimed by the administration, it is preferable in some cases to accept the filing of “voluntary” amended returns in which the taxpayer acknowledges part of the unpaid tax, rather than to pursue an official assessment reflecting what the tax administration believes to be the full amount of the underpaid tax. Official assessments usually involve a legal dispute and it is often better to reach a reasonable compromise accepted by the taxpayer than to obtain a favorable judgment in a court of law.

Typical Forms of Evasion

Adequate planning of the audit function requires definition of the general guidelines for combating the various forms of tax evasion. Some of the most common forms are described below.

Evasion with Fraud

A taxpayer guilty of evasion with fraud commits tax evasion through fraudulent actions such as forging or falsifying records. Accordingly, such actions go beyond the scope of taxation and into the area of criminal offenses. Fraud should be punished even where no harm is done to the treasury.

In order to combat fraud, special programs should be put in place for the thorough examination of a small number of cases. Taxpayers to be audited are those whose characteristics make it reasonable to suppose, in light of earlier auditing, that they have committed fraud. For evasion with fraud, the strategy should rely on the imposition of very stiff penalties; some countries impose prison terms. The penalty should be very severe and should be made public because it must serve as a deterrent. The chief goal of this strategy is not to punish the person guilty of committing fraud in order that he should not do it again but to make an example of that person.

Evasion Without Fraud

To combat those who qualify only as tax evaders—because they have underreported their taxes without committing fraud in the process—the strategy should be different. The evader should be dissuaded from engaging in this behavior again. Here the purpose is not to punish the tax evader in order to make an example of him to others but to prevent repeat offenses. The evader will change his behavior if he learns, or if he believes, that tax evasion does not pay.

In areas where there is a high level of noncompliance, evasion basically hinges on two variables: (1) the probability of being caught and (2) the penalty for evasion. The way to increase the probability of detecting and punishing tax evaders (that is, to increase the risk inherent in tax evasion) is by developing a strategy based on massive and speedy control. The idea is to take action on some of the specific violations committed by the taxpayer, without looking into whether he may have committed others. This would broaden the audit coverage (the ratio between taxpayers and number of audits), thereby raising the risk implicit in evasion. In countries where the tax compliance level is low, the object of enforcement should be to bring about not perfect compliance by a few taxpayers but better compliance by most taxpayers.

If the penalty structure is reasonably adequate, priority should be given to increasing the likelihood of “effectively” punishing the tax evader. Effectively means that the taxpayer should actually pay the penalty, not that he should be punished on paper through an official tax assessment that will never be paid or that will be paid only in very small part. Furthermore, if the likelihood of effectively punishing evasive behavior is very low, taxpayers will take the risk even if the penalty is very severe.

Tax Avoidance

Tax avoidance is the behavior of those taxpayers who take advantage of legal loopholes in order to underreport their taxes. In other words, they take advantage of circumstances not clearly provided for in the legislation, or deficiently construed, in order to lighten their tax burden. If tax avoidance is to be detected, audit programs must be carried out with that specific end in view. Once having detected the manner in which taxes are being avoided, the problem may be overcome by refining the legal provisions and their interpretation, in other words, by amending the legislation. This kind of underpayment of taxes is sometimes engaged in by large taxpayers who have trained professionals to advise them on the interpretation of legal provisions.

Preventive Action

An audit plan should also provide for preventive action, the aim of which should be to persuade taxpayers that evasion does not pay. This is achieved if the tax administration can show the taxpayer that it has the information and the operational ability to punish him if he evades taxes.

It is advisable to devise enforcement programs whose chief purpose is to let the taxpayer know that the tax administration has information on certain economic transactions such as imports, leases, purchases, sales, and payments for professional services. Here the purpose of a visit by an inspector, or of a letter mailed to the taxpayer, is to show the latter that the tax administration has information which, after the tax becomes due, will be compared with his tax returns in order to detect possible inconsistencies. It is obviously not necessary to give the taxpayer a detailed account of all the information the agency has or does not have. The point is to demonstrate to a significant number of taxpayers, before they commit tax evasion, that the tax administration processes data on their economic transactions and that this information will be used in detecting evasion.

Auditing Large Taxpayers

In general, auditing of large taxpayers should be steered primarily toward detecting the tax avoidance practices they engage in through legal interpretations or transactions with related parties. The likelihood is small that major taxpayers would engage in gross instances of tax evasion as not issuing invoices for their sales or falsifying inventory figures, because their own organization and accounting systems do not allow it.

It needs to be kept in mind that the comments below refer only to the audit of large taxpayers and are not applicable to other enforcement actions of the tax administration such as those aimed at detecting stopfiling and delinquent taxpayers.

It is not a good idea to concentrate on auditing large taxpayers. Some countries fall into the error of overdoing the allocation of resources to auditing large taxpayers because it is presumed that this is where the largest tax payoff is to be found. Although this assumption may be correct, and the productivity for each audited case (the potential additional tax to be collected) may be raised by auditing large taxpayers, total tax receipts from all taxpayers may turn out to be lower if auditing of medium or small taxpayers is neglected. This is so because total tax receipts may grow more by focusing audit on medium and small taxpayers, owing to the persuasive effect that this policy has on all taxpayers as a whole.

In some countries almost the entire audit staff is devoted to auditing large taxpayers. The reasoning behind this is that economic concentration is very high and audit activities should therefore be highly concentrated as well. For instance, because as much as 70 percent of tax revenues are accounted for by 5 percent of taxpayers it is presumed that audit should focus mainly on this 5 percent. Such a policy can lead to the absurdity of concentrating tax receipts even more. Thus, in this example, if the tax administration takes no action with respect to the medium and small taxpayers that make up the remaining 95 percent, a situation may develop where 5 percent of all taxpayers will account for 100 percent of all tax revenues.

Concentration of tax receipts in large taxpayers has two causes: (1) income and production are often concentrated and (2) the high rate of tax evasion by medium and small taxpayers, who for this very reason should be especially targeted for audit.

It has been shown in certain cases that a change in audit strategy intended to tighten control over medium and small taxpayers’ compliance leads to an increase in total tax receipts and that, notwithstanding the high degree of concentration, this rise in total tax revenues originates primarily in medium and small taxpayers. As shown in Table 8, in Chile, for instance, with the improved effectiveness of tax administration that was partly due to a change in audit strategy, there was a real increase of 55.5 percent in VAT receipts during 1979-82. This growth was due to an increase in tax receipts of 16.5 percent from large taxpayers, 86.9 percent from medium taxpayers, and 185.6 per cent from small taxpayers. (See Table 9 for the change in the concentration of VAT receipts in Chile, caused by the improved compliance of medium and small taxpayers.)

Table 8.Chile: Increase in VAT Receipts, 1979—82
Taxpayer Size1Percentage Increase
All taxpayers55.5
Source: Chile. Internal Revenue Service.

Large taxpayers were defined as those with annual sales of more than US$6 million; medium, between US$6 million and US$600,000; and small, under US$600,000.

Source: Chile. Internal Revenue Service.

Large taxpayers were defined as those with annual sales of more than US$6 million; medium, between US$6 million and US$600,000; and small, under US$600,000.

Table 9.Chile: Concentration of VAT Receipts
Percentage of Receipts
Taxpayer Size119791982
Source: Chile, Internal Revenue Service.

Large taxpayers were defined as those with annual sales of more than US$6 million; medium, between US$6 million and US$600,000; and small, under US$600,000.

Source: Chile, Internal Revenue Service.

Large taxpayers were defined as those with annual sales of more than US$6 million; medium, between US$6 million and US$600,000; and small, under US$600,000.

Encouraging Future Voluntary Compliance

When the goal is to raise the prevailing level of tax compliance significantly and permanently, promoting compliance is as important as punishing noncompliance. In this connection it is always self-defeating to declare tax amnesties “for the last time,” because amnesties actually amount to a reward granted to the tax evader. It is desirable, on the other hand, to encourage future voluntary compliance by concentrating audit activities on the current tax year and blocking off previous years from review. If any significant violations are found while going over the current year, then all preceding periods not barred by the statute of limitations would be looked into. Those taxpayers who have evaded taxes will find it in their interest to have a good future performance if by doing so they are able to block off the past. It goes without saying that this behavioral change will come about only if truly effective auditing is established successfully. Systems similar to the above have been used in Argentina, Chile, and Mexico and have been found useful during a transition period, until a reasonable level of compliance can be assured.

The Withholding Problem

In several countries where tax compliance levels are low, various kinds of withholding systems have been introduced in an effort to reverse the prevailing situation. In very broad terms, the premise is that large taxpayers must withhold the tax when they do business with medium or small taxpayers. Thus, for instance, some countries prescribe that if the government or a large company buys from a medium or small taxpayer, the former must not pay to the latter the full amount of the VAT due on the transaction, but must instead withhold a portion of it. Likewise, if a large company sells to a “nonmajor” taxpayer, that company must collect a surcharge which will be a partial payment against the tax payable by the medium or small taxpayer. Although countries such as Argentina, Chile, Colombia, and Ecuador have successfully implemented various withholding systems, it needs to be stressed that withholding should not be looked at as a cure-all for the problem of tax evasion. It is well to remember that every time a tax is withheld it becomes necessary to monitor not only the withholding agent to make sure that the amounts withheld are actually turned over to the treasury but also the “withholdee” to verify that only the amount withheld, and no more, was deducted from the tax owed.

In short, tax administrations that use withholding systems in a general way should evaluate the effectiveness of each kind of withholding. To do this it is necessary to establish at least at the aggregate level (considering all taxpayers) whether the amount collected for each type of withholding is the same, higher, or lower than the amount of deductions taken by taxpayers for each tax withheld. This analysis is crucial because one of the problems observed is that, over time, if withholding is not effectively controlled by the tax administration, the amounts withheld are not turned in by the withholding agents at the proper time and, in addition, taxpayers deduct as amounts withheld larger than those actually withheld.

III. Information Systems

To implement the suggestions made above, an adequate information system is essential. The key is to diagnose properly the situation prevailing in each country and to devise a strategy for introducing an adequate information system. As shown by the experience of many countries, buying computers alone is not enough to improve the effectiveness of tax administration. Table 10 shows the performance of some general indicators in a Latin American country. The situation described below is representative, broadly speaking, of conditions frequently observed in many countries throughout the world, particularly in Latin America and Asia.

Table 10.A Latin American Country: Performance of Some Indicators
1965197519851985 in


to 1965
Tax receipts (in constant currency)3,1068,20210,8793.50
Returns filed551,0001,837,0002,112,0863.80
Total administration expenditure (in constant currency)40.262.2219.75.42
Number of tax administration personnel2,3135,9836,9512.96
Memory of central computer0.61MB2.5MB24MB1,500.0
Number of computer screens0351,000
Memory of peripheral computers003MB
Source: IMF, Fiscal Affairs Department.
Source: IMF, Fiscal Affairs Department.

The figures in Table 10 show rather consistent behavior in such areas as receipts, returns filed, budgeted expenses, and staff of the tax administration. One item, however, stands out clearly: computer equipment.

Because of changing technology and the priority assigned by the authorities to the purchase of computer equipment, computing capacity grew remarkably in the period under consideration. In 1985 the memory of the computer equipment was 1,500 times larger than in 1965 and nearly 10 times larger than in 1975. This comparison, however, underestimates the computing capacity for 1985 because processing speed has risen significantly over the same period. In addition, the tax administration had no teleprocessing screens in 1965, whereas in 1975 it had 35 and by 1985 it had 1,000. Furthermore, in 1985 the administration had peripheral computers whose capacity was larger than that of the central computer in service in 1975.

Unfortunately, despite the growing computing capacity and the modernization efforts made by the tax administration, the latter did not significantly improve its effectiveness in monitoring tax compliance. This may be deduced from the frequency of tax amnesties declared in that country and from the fact that it was still not possible by 1985 to exercise timely control over registered taxpayers who filed no returns or over delinquent taxpayers—situations that are generally easier to control than tax evasion (that is, the failure to report taxes in the tax return). An increase in computing capacity is not in itself enough to guarantee an increase in the effectiveness of the tax administration.

Problems Connected with Information Systems

One common problem is that the information systems in use yield data with errors and are therefore not very reliable. The immediate consequence of this is that the information produced by the computer is not taken seriously either by the tax administration or by taxpayers. To overcome the problems posed by these errors, solutions are sometimes devised that turn out to be inadvisable. These include duplicating procedures, developing parallel manual processes, and retrieving the same data from different computers for different purposes. All this does is aggravate the problem, because such duplications only serve to make the ever-scarce manpower of tax administrations even scarcer. Besides, such systems are divergent, and in the absence of timely correction of errors even more mistakes will be made. Ultimately, the computing system may be rendered totally useless and may be finally disconnected, which is what happened in Uruguay in the early 1980s, or it may be completely redesigned, as in Colombia and Bolivia beginning in 1985.

Another difficulty frequently observed is the rigidity of systems. In many instances systems do not lend themselves easily to legislative changes because they were designed on the assumption that the tax system is unchangeable. Obviously this is not true, and it is therefore necessary to establish systems capable of handling abstract concepts such as debts, credits, rates, forms, interest, fines, and so on. These concepts will always apply regardless of the type of tax being dealt with, and consequently, if the computing system is designed to operate with these concepts, all that will be required when an existing tax is changed or a new one is established will be to change the rules under which these concepts are processed.

In sum, then, the chief problems connected with information systems are as follows:

  • The systems are not very reliable (they produce a considerable number of errors).

  • Control of stopfiling taxpayers is done manually.

  • Control of delinquent taxpayers is done manually.

  • There is no adequate control and procedural follow-up for collecting delinquent taxes.

  • The TIN is not reliable enough.

  • An independent manual system is used to keep record of payments.

  • Processes are duplicated because the same piece of information is transcribed over and over again for different purposes.

  • There are duplicate files for documents.

  • The systems are too rigid and incapable of swiftly adjusting to changes in tax legislation.

  • Statistics are not timely.

  • There are no systems to support enforcement activities.

General Guidelines for Developing Information Systems

In developing an information system, the first priority is to design the frame of reference describing the general characteristics of the systems and procedures used by the tax administration. The key is to define an adequate strategy, setting out with precision the course to be followed. In other words, what is the goal going to be? To avoid the seesawing that sets back the progress of tax administration, charting the direction of that progress is more important than its speed. The frame of reference to be used in developing the information system should be defined before taking other decisions, particularly those connected with the purchase of computing equipment. Otherwise serious errors may be made, as demonstrated in some Eastern European countries that have recently introduced structural reforms in their tax administrations. If there is no frame of reference for developing these systems, a situation may arise where the administration becomes the servant of the computer instead of the other way around.

The frame of reference of the tax administration should answer specific questions such as the following:

  • Should taxpayers send their tax returns to the offices of the tax administration, to banks, to the post office, or to any of these three?

  • Should taxpayers pay their taxes at the offices of the tax administration, at banks, at post offices, or at any one of these three?

  • Should separate forms be required for tax returns and for payments or should a single tax return be designed in such a way as to render a separate payment form unnecessary?

  • Should banks accept a tax return filed without an accompanying payment in cases where no payment is due or there is a tax credit?

  • Should banks accept a tax return unaccompanied by full payment of the taxes due?

  • Should the data be retrieved at the tax administration, at banks, or at private data-entry companies?

  • Where should the data be validated and what should be the general requirements for validation?

  • What is the most adequate level of centralization or decentralization for data entry and processing?

  • What sort of TIN should be used? Should the same TIN be used for all taxes at the same time or should its application be gradually extended to different taxes?

  • Is teleprocessing necessary? In what applications?

  • Is it better to post the data on-line to the current account of taxpayers or to retrieve the data from tax returns in a separate process and then post them to the current account in a subsequent operation?

  • Should a different system be established for monitoring large taxpayers?

  • How should files be organized to keep the returns filed by taxpayers?

  • What information should be asked for in tax returns?

  • What systems should be developed as a first priority?

  • What systems should be developed to support control activities?

  • Should the production of statistics be an independent process or a natural by-product of the information system?

  • What sort of controls should be put in place in order to make certain that the systems are secure? In other words, how to prevent the recorded data from being falsified or becoming known to unauthorized persons?

These questions are merely examples from a much longer list of questions to be appropriately answered by the frame of reference established to guide the development of tax administration systems and procedures.

Audit Support Systems

Systems must be developed to support auditors in the field.7 When visiting a taxpayer, for instance, the auditor should have with him a computer printout showing the data from tax returns filed by the taxpayer. The visit, furthermore, should be based on very specific presumptions of noncompliance: for example, when it has been detected that purchases made from the taxpayer by certain companies exceed the sales or revenue he reported. This can be detected only by developing systems that cross-check data on the economic transactions carried on by taxpayers.

Below are some examples of enforcement support systems that may be profitably developed.

Abstract of Reported Taxes

The abstract of reported taxes should be a computer printout showing the data reported by the taxpayer when filing his tax return. The purpose of this system is to enable the auditor to conduct the audit using the information in the abstract. To guide the auditor, the computer should highlight on the abstract any inconsistencies such as incorrect deductions and arithmetical errors. The abstract should also include ratios such as sales/purchases, earnings/sales, and the modal value (the most frequent value) that these indicators register for homogenous taxpayers.8 Basically, the abstract should include three groups of data.

(1) Data reported by the taxpayer in different tax periods, such as the amount of reported sales, the amount of reported purchases, credit carryforwards, and refunds requested. Here, calculation errors detected by the computer should be highlighted.

(2) Statistical data, such as modal value of the markup reported by taxpayers similar to the taxpayer to be audited.

(3) Data on the taxpayer’s economic transactions, collected from external sources. Details on value, date, and type of operation for transactions such as imports, sales made to the taxpayer by his suppliers, and purchases made by its customers should be included.

Sales to the Government

It is a good idea to process the purchases made by government companies and state agencies for establishing whether such purchases have been reported as sales by taxpayers who acted as suppliers. In some countries, tax evasion under this heading is frequent. Indeed, there are cases of sales made to tax administrations that have gone unreported by the suppliers.

Company Purchases and Sales

It is important to process the purchases of large and medium-sized companies for the same reason given in the preceding item, namely, to try to detect tax evasion on the part of suppliers. It is also important to process the data on sales made by large and medium-sized companies. In this case the objective is to detect those buyers who, over a considerable length of time, report sales figures lower than the amounts of their own purchases from large or medium-sized companies.

Cross-checking of data on purchases and sales should be selective. Only the data on large and medium-sized companies should be processed. In addition, the sales and purchases to be processed should be selected according to guidelines on such points as the minimum amount of operations to be looked at and the type of goods or services being transacted. These criteria should be defined bearing in mind the computer resources needed to enter the data in a timely fashion.

Some have argued that the cost-benefit ratio of these cross-checking systems is unfavorable. The argument is based on two premises: (1) that generating a magnetic-medium recording of the data on sales or purchases is very costly and (2) that the sales or purchases recorded would reflect “above-board” transactions, making this type of information useless in detecting operations “under the counter.”

As to the first argument, it would not be very costly to generate purchases or sales files if large and medium-sized companies are asked to supply these data to the tax administration on magnetic media. All that is needed is to ask companies to report, in connection with their sales and purchases, the date, the amount, and the TIN of the taxpayer buying or selling, as the case may be.9 Producing magnetic files with these data would be an insignificant cost to companies that have computerized invoicing and accounting systems, which is increasingly the case among large and medium-sized companies.

As for the second argument, experience indicates that large companies seldom make “under-the-counter” sales. Small companies, on the other hand, are far more likely to buy from large ones and make backroom sales. This activity could be detected with the proposed crosschecking system. Furthermore, the system could also pick up purchases deducted by large taxpayers that pertain to suppliers who are not registered with the tax administration or who, if registered, are not reporting their sales fully.


In order to detect the unreported sales or revenues of importers, it is important to process data on imports. By comparing the imports made by taxpayers with the sales or revenues they report, it is possible to deduce whether they are underreporting sales in connection with the VAT or revenues for income tax purposes. When dealing with the income tax of individuals, it is also useful to test the income reported by taxpayers against any imports of luxury articles they may have made.

IV. Determinants of Change in Tax Administration

Why do changes occur at a particular time? What are the circumstances that make it possible for a structural and permanent change to take place and significantly improve the effectiveness of a tax administration? We know very little about this. We do know that technology is necessary but not sufficient to ensure progress in tax administration. That it is not sufficient is something to be regretted, because it is, relatively speaking, the easiest part of the puzzle. Simply allocating in the budget of the tax administration a specific amount of money will guarantee the purchase of the most modern technology as far as the equipment is concerned. Many countries, however, have spent many millions on computer equipment without being able to improve the effectiveness of their tax administration—that is, without seeing any drop in the high ratios of tax evasion and delinquency.

Experience leads me to believe that the key ingredient in bringing about change is having a team of men and women with the will and the political authority needed to see the changes through. If significant success is to be achieved, it is essential to have a group of officials with an iron will and even obsessed by the idea of modernizing. This group of officials must have the courage to risk the unknown and must take all necessary precautions against failure.

The task of making a tax administration more effective is obviously neither a simple nor a short-term job, especially because it usually requires introducing reforms in many other areas directly connected with tax administration.

To begin with, it is very often essential to reform the tax laws to simplify the tax system. It is not news nowadays to say that tax policy and tax administration are intimately linked and that it is very difficult to draw the dividing line between these two areas. Nor is it a revelation that a tax system which on paper may seem fair and economically efficient may in fact be exceedingly unjust and inefficient because the tax administration is unable to enforce the law. Achieving substantial improvement in the effectiveness of tax administration also requires a bold simplification of the tax system, as the more successful examples of the 1980s have taught us.

Second, many administrative practices that have been followed for many years need to be changed. Simplification of tax returns, for instance, is likely to run into a great deal of opposition because there will always be ample reasons to justify asking for this or that piece of information or requiring this or that attachment. In response, tax administrators committed to change will be called upon to explain over and over again what history has shown, namely, that taxpayers are not awed by mere requests for information. The behavior of taxpayers will change only when the tax administration can count on timely and reliable information enabling it to detect noncompliance. For this, all that needs to be asked for are a few pieces of information because that is all that the tax administration can process with the degree of quality and timeliness required for effective administration of taxes.

V. Outlook

In my view of the foreseeable future and the prospects held out by technology in the area of tax administration, two predictable and desirable developments stand out: (1) the gradual disappearance of paper as the cornerstone of tax data, and (2) operational decentralization.

The gradual disappearance of paper as the cornerstone of tax data will bring about a profound structural transformation, which, perhaps, will be more important than the one that took place in the 1960s when computers were introduced in the area of tax administration. Even now, many administrations surrender at once before the majesty of paper. At present, in many countries, if the computer tells us that a particular taxpayer cannot be given a tax clearance certificate (paz y salvo) because he has not filed a tax return or has a tax debt outstanding, and if that taxpayer in turn proves by means of a copy of a form that he has filed a return or paid the tax, the tax administration gives up straight away and issues the certification applied for, without even checking whether the copy of the form made available by the taxpayer is authentic or a forgery. In tax administration, in principle, paper has the last word, and the information stored in the computer takes a back seat.

In the private sector, by contrast, just the opposite is true. What counts is the information stored in the computer, not the information on paper. Credit card operations, banking operations, and many other similar examples can be found in day-to-day activities where the information stored in the computer takes precedence, in principle, over the data on paper. This is so, of course, because the computers that support these activities produce timely and reliable information, which is often not true of tax administration.

What needs to change before tax administrations can catch up? The first issue to be resolved is the problem of data entry, so that within a short period of time the computer may record in an absolutely reliable way the data set out in tax returns and the payments made by taxpayers. For this reason, tax administrations in the more developed countries have as one of their basic objectives to encourage the transmission of data between the taxpayer and the tax administration without going through paper. This is known as “electronic filing.” The data filed by taxpayers in this way reaches the computers of the tax agency almost instantaneously and with a high degree of quality. Much remains to be done, to be sure, and many problems need to be resolved, particularly in the legal area, but the final objective is plain: to maximize the collection of data by electronic means or through magnetic devices in order to guarantee the timeliness and quality of the data processed by the tax administration.

The obstacles that remain now are more of a legal than a technical nature. For example, in the United States, where several million tax returns are already being filed electronically, the problem of the signature as the mark attesting to the intent of the taxpayer remains to be resolved. The taxpayer, besides sending in the data electronically, must also send in a signed paper confirming that he has filed his return. Curiously enough, in other areas of daily life the problem of the signature as validation of the intent of an individual has already been obviated. One may withdraw money from machines known as “automatic tellers,” check the balance in a bank account, play the stock market, and buy or sell millions in wheat, oil, or other commodities, merely by providing a secret code: the PIN (personal identification number). Why is it not possible to use a similar system in tax administration? Why is it not possible to assign a PIN to taxpayers who wish to file their tax returns electronically, thereby avoiding the duplication of mailing a signed statement to confirm the electronically transmitted data? Why should taxpayers have to also send in bulky paper records to confirm the data they have submitted in the form of a computer disk? These questions are for the legal profession to answer.

The second development mentioned as predictable and desirable within the near future is that of operational decentralization. More and more, technology is making it possible to use modern computer devices in small units at a reasonable cost. The economy-of-scale reasons that formerly justified the centralized processing of information have vanished. These days operational units located near the taxpayer and in locations where he carries on his economic activities are capable of processing the data and producing the information needed for a more effective control. Furthermore, one of the effects of decentralization is to make operational units increasingly responsible for producing information of the proper quality with the necessary speed.

ANNEX I Methodology for Estimating VAT Evasion

To estimate VAT evasion, taxable sales, purchases generating tax credits, and investments generating tax credits should be estimated. Using these figures the theoretical tax base can be estimated. Then the amount of potential receipts is calculated by applying the general rate to the theoretical tax base.10

Estimating Taxable Sales

Taxable sales are estimated from total supply. Total supply is equal to gross production plus imports. Taxable sales (which generate VAT debits) are calculated by deducting exemptions, exports, and inventory changes from total supply.

Estimating Net Purchases

Total purchases are defined as the intermediate consumption computed in national accounts. Net purchases (which generate VAT credits) are calculated by deducting from total purchases the purchases of exempted inputs (which generate no credits) and purchases of taxable inputs for the production of exempted goods and services (because these goods and services are not taxable and, therefore, the taxpayer cannot recover the VAT paid out for these inputs).

Estimating Net Investments

Net investments in fixed capital (which generate VAT credits) are estimated from the figures for gross formation of fixed capital in national accounts. To calculate net investments, from gross formation of fixed capital are deducted the exempted capital goods (because these goods are not taxable and, therefore, generate no VAT credits) and the taxable capital goods for the production of exempted goods and services (because these goods or services are not taxable and, therefore, the VAT content of these capital goods cannot be recovered).

Estimating the Tax Base

Deduction of net purchases and net investments from taxable sales yields the theoretical tax base plus the VAT collected. By deducting from this figure the VAT collected, the theoretical tax base (net of VAT) is calculated.

Calculating Theoretical Receipts

The potential or theoretical collection is calculated by applying the VAT rate to the theoretical tax base.

Calculating Collection Losses

Lost collection is calculated by deducting from potential receipts the VAT actually collected.

Outline of Methodology


Etcheberry Javier

As my main comment, I would like to state that the paper presented by Mr. Silvani is exceptionally thorough and covers all aspects of tax administration and tax legislation that should be examined to improve tax compliance in Latin America. The author’s classification of taxpayers according to their tax behavior is very useful and facilitates the formulation of enforcement policies. This classification and the comments, data, and specific strategies presented in the paper demonstrate the author’s experience—both theoretical and practical—in the field. His paper will help improve audit strategies and the effectiveness of tax administration in Latin America. Indeed, this seminar has demonstrated that the problems of tax administration and the economic and social situation of our countries have many common elements.

It should be noted that breaking down taxpayers as unregistered, stopfiling, evaders, and delinquent is not the only classification that can be used to formulate enforcement policies. Taxpayers can also be classified according to the economic activity they perform, the geographic area in which they are active, or perhaps by their size or legal status, and so forth.

Choosing one of the above-mentioned criteria for a basic taxpayer classification requires determining which criteria can most effectively be used to formulate an enforcement policy appropriate to the tax characteristics of a given country. These criteria may or may not coincide with the author’s recommendation. Once the basic classification is established, one of the other criteria can be used to further classify taxpayers in each of the main categories.

In Chile, current efforts are aimed at improving tax audit by stressing classification by sector, because the tax characteristics of the agricultural sector differ from those of the mining sector and these, in turn, differ from those of the industrial sector, and so on. For this reason, working groups have been set up to study audit policy for each of our economy’s major sectors.

As the author notes, these classifications are very useful for distributing among the various groups of taxpayers the scarce audit resources available to make tax administration more effective. It is also necessary to quantify the impact of audit on each of the groups within the classification used.

With regard to unregistered taxpayers, who frequently are part of the informal economy, I believe that it is not desirable to enact legislation that would reward these taxpayers with simplified tax requirements. If too many concessions are granted, the informal sector could, over time, tend to grow; moreover, formal sector taxpayers might increasingly use persons belonging to the informal sector for the production or marketing of their goods. It is preferable to have uniform tax legislation despite the fact that smaller taxpayers tend to be tax evaders, because as they become larger, it will be easier to monitor them and to enforce compliance.

Another comment refers to the author’s claim that tax evasion depends on the odds of being detected and the size of the penalty imposed. In my view, another element that should be included in the analysis is ethics. If people believe that tax evasion is unethical, tax compliance will be high despite low odds of being detected or low penalties. For this reason, governments, as well as social, religious, and political organizations, should take it upon themselves to ensure that tax evasion is considered unethical in our countries.

As an example, in Chile, during the military government of General Pinochet, the opposition never attempted to send the general population the message that not paying taxes was a way of defying the Government because, as in other Chilean organizations, there was relative consensus that contributing to undermine the Chilean taxpayer’s behavior would be unethical and hence undesirable.

Robert A. LeBaube

Carlos Silvani has had a long and impressive association with the subject of taxes. He has taken the opportunity to indicate those points that are of utmost importance to tax administrators in improving compliance with tax laws. I was further impressed with the connection he makes between effectiveness and efficiency in administration. Often governments fall into the attractive trap of efficiency and do not know whether their efforts are truly effective.

My comments fall into two categories, efforts to increase compliance and efforts to increase effectiveness.

Efforts to Increase Compliance

Mr. Silvani describes four tax gaps that must be narrowed in order to increase effectiveness and decrease noncompliance and they are all valid. A study done several years ago for the Internal Revenue Service (IRS) in the United States indicated a similar scenario. The study divided taxpayers into four groups:

(1) Those who understand and comply willingly with the law.

(2) Those who want to comply but do not understand.

(3) Those who understand but choose not to comply fully.

(4) Those who do not comply deliberately.

The trick is to move taxpayers from one group to another. There are many factors that affect this movement that are beyond the abilities of the tax administrators to influence. Things like confidence in the government, complexity of the law, perceptions of fairness of the tax law, fear of detection, and so on, can affect voluntary compliance in either direction.

I would offer two suggestions to increase compliance from a perspective based on my experience with IRS.

First, the use of informants. The IRS receives many thousands of pieces of correspondence about taxpayers alleging noncompliance with the law. In some cases rewards for such information are paid, but the informant must apply for a reward and payment procedures are complex. Much of the information has little or no value and is discarded quickly, but enough is valid to make such a system worthwhile. Public knowledge of such a system can be an effective deterrent.

The second suggestion is more positive in nature. A strong program involving taxpayer education, information, and access will help those who are disposed to comply to meet their obligations under the law. There are some dangers with such a program. Education must not include government propaganda in support of political issues. The education must be in the nature of why government needs taxes, how the laws are determined, and how a responsible citizen participates in the system.

Efforts to Increase Effectiveness

In the area of effectiveness I would offer the following: Carlos Silvani is quite correct when he warns that countries increasing their use of computers should be wary of the computer system controlling the tax administrator rather than the reverse. This has happened several times in the course of IRS experience over the past thirty years. As Malcolm Lane indicated in his earlier comments, go slowly and make sure that the system is giving you what you want and that you have the capacity to use what the system gives you. This is one area in which the IRS has not done well. For years the computer system has been able to give much more information than the personnel could use. The interesting point, however, is that the public thought IRS could do much more than it really was capable of.

The paper suggests a number of criteria to be examined in designing computer applications. They are excellent and should be followed.

I note with interest the desire to reduce the number of persons who participate in the tax system though the use of paper documents. The United States has embarked on electronic filing where there is no longer a paper return in the hands of IRS. However, there is a paper return in the hands of the taxpayer and the practitioner, if one is used. Most taxpayers, at least in the United States, want something in their hands that indicates they have fulfilled their obligation. A taxpayer may have little reason blindly to trust either a computer or a tax administrator. Even with a fairly sophisticated system and paper documents the U.S. system operates with approximately 15 percent processing errors, about half caused by the taxpayer and half caused by the IRS. With the advent of electronic filing that error rate, on electronic returns, has fallen to 3 percent. For 1990 returns, about 5 million were filed electronically and about 150,000 had errors. I would caution that mechanisms must be installed early to deal with errors quickly and courteously. Confidence and respect for the system can be increased through simple things such as apologies when the system is in error and sympathetic understanding when the taxpayer is correcting his error.

The paper discusses operational decentralization as one of the means to increase effectiveness. I agree fully on this point but caution that one must have the resources, particularly well-trained personnel, to effectively implement such a program. You must also have adequate security of the system to prevent improper access and manipulation of the system. Audit trails as described in the paper on Spain are essential for this purpose. But one must be committed to regular inspection of those trails to determine if anything improper has occurred.

The information and suggestions offered in the paper can serve administrators well in their efforts to maintain and improve taxpayer compliance.

The author wishes to thank Leif Mutén and Angel Boccia for their comments on an earlier version of this paper.

See Estela Calello, Susana Rovner, and Aldo Schlemenson, “Motivaciones de la evasión en el IVA” (Buenos Aires: Direccion General Impositiva. 1971).

See Alan A. Tait, ed., Value-Added Tax: Administrative and Policy Issues, IMF Occasional Paper No. 88 (Washington: International Monetary Fund, October 1991).

In some countries the tax administration is also able, if no tax return has been filed for a particular year, to assess the tax owed by reference to the tax reported in the preceding year.

L. Wanes and L. McArthur. El control de los impuestos en los paises en desarrollo (Santo Domingo, Dominican Republic, 1981).

Based on Milka Casanegra de Jantscher and Carlos Silvani, “Guidelines for Administering a VAT,” International VAT Monitor (December 1990), pp. 2-19.

Voluntary collection is revenue obtained from the amended returns (corrections) filed by taxpayers after they have been audited.

This section draws on Milka Casanegra de Jantscher and Carlos Silvani, “Guidelines for Administering a VAT,” International VAT Monitor (December 1990). pp. 2-19.

Taxpayers in a similar business activity and with a similar sales volume.

It would also be advisable to ask for the invoice number and a description of the goods transacted, as long as this does not entail an excessive cost to the taxpayer.

This methodology is for a VAT with a single rate. If there is more than one rate, the tax base estimate must be broken down for each rate.


The VAT collected is the net collection remaining after deducting refunds to exporters.

    Other Resources Citing This Publication