Bank Debit Taxes in Latin America
An Analysis of Recent Trends

Six Latin American countries have levied taxes on withdrawals from bank accounts, which have been viewed as a convenient tax handle during a difficult fiscal period. The paper reviews the arguments for and against this type of taxation, describes the taxes, and surveys their revenue performance and economic impact. It concludes that the recently implemented taxes have been successful in raising revenue in the short term, but that adverse allocational impacts have likely been significant. The tax may work better in times of fiscal crisis, when financial intermediation is deep, and when the tax rate is modest.


Six Latin American countries have levied taxes on withdrawals from bank accounts, which have been viewed as a convenient tax handle during a difficult fiscal period. The paper reviews the arguments for and against this type of taxation, describes the taxes, and surveys their revenue performance and economic impact. It concludes that the recently implemented taxes have been successful in raising revenue in the short term, but that adverse allocational impacts have likely been significant. The tax may work better in times of fiscal crisis, when financial intermediation is deep, and when the tax rate is modest.

I. Introduction

In the last 13 years, a number of Latin American countries have resorted to the use of new financial transactions taxes in order to raise revenue. These taxes have been imposed on banking transactions, mainly on debits. First introduced in Argentina in 1983, this type of tax was reintroduced there in 1988 and 2001, and implemented in Peru (1989), Brazil (1994 and 1997), Venezuela (1994 and 1998), Colombia (1998), and Ecuador (1999) (Figure 1). These taxes can be seen in the context of a larger spectrum of “financial transactions taxes,” the provenance of which stems back hundreds of years to documentary stamp duties in Europe, and includes more modern securities transfer taxes as well as theoretical “Tobin” type taxes on currency exchanges.2 This paper will not, however, look at financial transactions taxes in general,3 but will focus more narrowly on domestic, revenue-oriented, bank debit taxes in the context of Latin America.4

Figure 1.
Figure 1.

Rates of Taxes on Bank Debits, by Country

Citation: IMF Working Papers 2001, 067; 10.5089/9781451848908.001.A001

Source: Country documents.1/Beginning in March 1999, the tax paid by individual taxpayers is creditable against the personal income tax. The precise amounts are unknown.

Like the original stamp taxes (described in Appendix II), these new bank debit taxes have been imposed because the transactions on which they fall were viewed as a convenient and effective tax handle, against a background of weak tax administration and, typically, in the face of a difficult fiscal/revenue situation. As the authorities were aware of the potential allocational problems associated with these taxes, they were generally introduced with a specific, short, intended life span. The objective has been to produce a burst of revenue until such time as more desirable taxes can be designed and enforced. Nonetheless, the spread of the new taxes, however well intentioned, raises tax policy concerns, particularly as regards their impact on resource allocation.

This paper attempts to address several questions.5 Given the stated revenue objective of the recently introduced debit taxes, what have been the actual revenue effects? What changes in economic behavior can be identified as a result of the introduction of the new taxes?

What have past experiences with such taxes revealed? The inquiry proceeds by reviewing the arguments for and against bank debit type taxes in general. The next section then describes the introduction and characteristics of bank debit taxes in the three Latin American countries—Argentina, Peru, and Venezuela—which introduced and subsequently revoked such taxes, and the three where they are currently in effect—Brazil, Colombia, and Ecuador.6 Finally, the paper attempts to evaluate the performance of the bank debit tax, and to indicate the lessons that may reasonably be drawn.

II. Bank Debit Taxes: Theoretical Considerations

What are the principal theoretical considerations that bear on the usefulness of bank debit taxes as revenue raisers? Tax systems raise revenue while balancing the competing considerations of minimizing allocational effects, taking account of equity concerns and administrative abilities. Box 1 briefly summarizes how the general principles of optimal tax theory may influence the design of tax systems.

Consider first the allocational effects. The base of bank debit taxes differs from that of other financial transactions taxes imposed upon equity trading, foreign currency flows, or other transactions. In the case of bank debit taxes, the base could, in the limit, be considered equivalent to that of a proportional income tax plus a turnover tax on the entire economy.7

Generally, taxes based simply on turnover are especially distortionary, in part because the tax burden depends solely upon the structure of transactions. The adverse implications of taxes with respect to financial intermediation may be even more serious than those with respect to turnover taxes on other types of transactions, and go beyond the static allocational effects described in Box 1, in that they target the main means of payment, the smooth functioning of which is critical to economic development.8 Financial intermediation generates significant positive externalities; in particular, the capital market development associated with financial deepening will facilitate enhanced mobilization of savings and a more efficient allocation of investment, and hence support higher growth. Moreover, the reality is that many countries have adopted or considered adopting the bank debits tax at times when their banking systems could be judged to be undercapitalized, with additional investment in such banking systems being a goal in some cases.9

An Optimal Tax Perspective on Bank Debit Taxes

Optimal tax theory typically views the tax problem as one of meeting a revenue target while balancing the often-competing considerations of efficiency (in the loose sense of raising revenue in a way that does not harm aggregate real income) and equity. Two results are particularly relevant in the present context. First, when some relatively stringent restrictions on consumer preferences hold, then optimality (indeed, Pareto efficiency) involves using only a nonlinear tax on income—intuitively, earned income contains as much information as one can hope for on the underlying unobservable differences in ability to pay, upon which, in an ideal world, taxation would be based (Edwards and others, 1994). Indirect taxes are then entirely unnecessary. Second, under competitive conditions, and assuming that pure profits are fully taxed and that the government is unconstrained in its ability to deploy distorting taxes, then production efficiency is desirable: that is, taxes should not distort the way in which commodities are produced; the intuition being that if they did, it would be possible to find a tax reform that increased output of all goods and so could make everyone better off.

In the real world, even though the conditions calling for production efficiency are unlikely to be fully satisfied, it is generally still seen as a desirable feature of a modern tax system—as evidenced by the common desire to ensure that all investors face “a level playing field.” In contrast, heavy reliance on income taxes is not evident in developing countries. This is in large part because the optimal tax framework abstracts from tax administration considerations, with income taxes being particularly difficult to administer. As a result, many developing countries often rely on indirect taxes for much of their revenue. For its part, the IMF has supported consumption-based value-added taxes (VAT) as a reasonable compromise between efficiency, equity, and administration considerations. The consumption base ensures that the VAT is consistent with production efficiency. Related to this, VAT was developed partly as an antidote to the cascading problems associated with the precursor turnover taxes. (Cascading refers to the “tax on tax” that arises when tax is charged both on an input into some process and on the output of that process; when taxes cascade through the chain of production, production efficiency will likely be violated.)

Where do bank debit taxes stand in light of this? Taking account of the theoretical results above, this type of transaction tax would appear to be considerably inferior to the income tax wherever that is feasible and, more relevantly, to the VAT. Not only do bank debit taxes have the potential to cascade but, by taxing inter alia the transactions medium of an economy, they are levied primarily on intermediate rather than final goods, further compromising production efficiency. They are, in effect, a particularly opaque form of turnover tax.

To elaborate on how the tax affects economic activity as regards firms, each taxable transaction contributes to profit and, as long as that contribution exceeds the tax, the transaction will take place. The size of the surplus or deficit of this contribution relative to the tax will be market determined and will likely vary systematically across sectors and industries. The extent to which businesses take actions to avoid the tax when it is imposed or increased will depend upon the size of the surplus relative to the increased cost of completing the transaction in an alternative manner (e.g., offshore). Transaction intensive businesses, such as retailing and trading in financial instruments, would likely be most impacted by a bank debit tax, though financial trading businesses will likely have greater access to avoidance mechanisms at lower costs. Conversely, just as turnover taxes cascade and create an incentive for vertical integration, so bank debit taxes are less onerous for vertically integrated operations that internalize financial transactions. Households in principle face the same calculations as firms, though the opportunities for avoiding taxable transactions may be more limited and/or more costly than is the case for businesses.

The degree to which bank debit taxes alter behavior will also depend on the type of transaction being taxed in addition to who is being taxed. The impact of these taxes on different transactions—such as between demand deposits and loans of varying maturities withdrawn from the banking sector—is potentially quite different.10 11

Finally, the systematic alteration of behavior relating to financial intermediation likely takes time and is not without cost. Thus, the adverse impact of bank debit taxes may not occur immediately. Further, where the bank debit tax is credibly introduced at a low level and for a predetermined short period, such behavioral alterations may be modest. Conversely, however (and this is precisely what makes use of these taxes so risky), hysteresis likely does exist—once financial activities have been moved offshore, for example, both transactions costs and the realization of additional benefits (e.g., avoidance of other taxes) may make it relatively unlikely that the shifts in transactions mechanisms will be reversed when the bank debit tax is eliminated.

As a practical matter, it is challenging to design financial transactions taxes to prevent taxpayers from circumventing them. This may be done by conducting transactions in cash, moving trades outside the reach of the country imposing the tax on its own markets, and/or by the use of derivative financial instruments, trading in which achieves the equivalent effect as trades executed in the underlying taxable transaction. (This is a distinct issue from whether these taxes are easy to administer in the narrow sense of ease of collection on designated transactions. In the limit, of course, if perfect substitutes for the taxed transactions exist, then no revenue would in fact be raised, even if, in principle, it is easy to collect.) This type of avoidance behavior can be expected to take place with bank debit taxes, albeit possibly somewhat more slowly in countries where there is not already a tradition of offshore banking, as suggested by the Brazilian experience discussed below. Note that, to the extent there is a shift to cash-based and offshore transactions, this potentially erodes further the bases for the VAT and income taxes.

To this point, the focus has been on the allocational implications of bank debit taxes. What is the incidence of the tax (an issue which bears on the equity objectives of the authorities)? One of the reasons that these taxes were a popular revenue source, at least initially, may be that it was believed they would be perceived as “victimless.” The reality could be quite different. While a comprehensive assessment of the true incidence would be beyond the scope of this paper, some qualitative observations can be made. In particular, consider the implications of these taxes for the owners of bank capital, suppliers of savings, and users of savings. In the short run, the spread between effective borrowing and lending rates will tend to increase to accommodate the tax, with the actual net-of-tax deposit rate being essentially determined by the relative short-run elasticities of demand for and supply of bank deposits. As an example, if the banks invest most of their holdings in government securities, and if those same banks are a major source of finance for the budget, the government could find itself paying most of the tax in the form of higher interest rates.

Over the longer term, the elasticities of demand for, and supply of, bank deposits will change as individuals and banks find new ways to avoid the tax—for example, by moving accounts offshore. The incidence of the tax can then be expected to change at the expense of domestic banks, with owners of existing bank capital suffering a capital loss as bank profitability would tend to be adversely affected—new investments in banking will be made elsewhere, including overseas. This would, again, tend to raise the costs of doing business as a result of the increased disintermediation. Finally, the incidence effects would also be influenced by the base of the tax, depending, for example, on whether interbank transactions, clearing operations, etc., are subject to the tax.

A broader approach to incidence analysis is to consider how the bank debit tax will ultimately affect income distribution. The poorest in society will likely be little affected directly since they typically are not heavy users of the banking system, although this conclusion would need to be modified when secondary effects are considered: bank debit taxes could result in higher prices for basic goods purchased in transactions-intensive retailing outlets. Otherwise bank debit taxes would tend to be regressive to the extent that the richest are in the best position to use offshore accounts and the like that facilitate avoidance of the taxes.

Finally, the collection costs of a bank debit tax are likely less than those of many other taxes by dint of the fact that the banks, which are in all cases the collection agents for the government, are few in number, typically have highly automated accounting systems, and keep track of all financial transactions as part of doing business. It is precisely the simplicity and clarity of these administrative arrangements that has contributed to the recent popularity of the bank debit tax in Latin America.

III. Recent Experiences with Bank Debit Taxes: The Cases of Argentina, Brazil, Colombia, Ecuador, Peru, and Venezuela

Details regarding the taxes of the individual Latin American countries that have used the bank debit tax are provided in Appendix III.12 This section highlights generalizations that can be drawn from these taxes with respect to the circumstances of their introduction and their broad characteristics.

A. Circumstances of Introduction of the Taxes

In all cases, with the exception of Brazil, the bank debit tax was introduced at a time of, and in response to, general economic crisis, as an emergency means of raising government revenue. When the Argentinean tax was introduced in 1988, tax revenue was declining dramatically owing to hyperinflation, increased evasion, and depressed economic activity. In 1989, when the Impuesto a los Débitos Bancarios y Financieros was introduced as a temporary and extraordinary revenue measure, Peru was immersed in a deep economic crisis. Real GDP fell by 20 percent in 1988–89 and prices rose by 1,700 percent in 1988 and 2,770 percent in 1989. Central government revenues fell from 14.9 percent of GDP in 1985 to 6.1 percent of GDP in 1989. In Colombia, when the Impuesto a las Transacciones Financieras (commonly known as ”dos por mil”)13 was adopted in November 1998, the health of the financial sector had already deteriorated markedly and the government had declared an economic emergency. In Ecuador, the tax was introduced in 1999, at a time when the economy plunged into a major economic and financial crisis. In view of Venezuela’s pressing economic problems, the Congress in 1994 authorized the Caldera government to issue tax measures (dubbed the “Sosa package”) by decree.14 The tax was then adopted as a temporary expedient, applying only in the period May 9–December 31, 1994, and, in fact, the tax ceased to be collected at the end of 1994. In 1999, faced with a difficult economic situation and a severe reduction in oil revenue, an Enabling Law was passed which allowed the President to issue legislation on economic and financial matters. One of the measures immediately decreed reintroduced the debit tax effective for 12 months beginning in May 1999.

B. Broad Characteristics of the Tax

In all cases, the tax was explicitly introduced on a temporary basis, though in some cases it was then extended. Rates have ranged between 0.2 and 2.0 percent, with lower rates sometimes applied to specified transactions such as those with the central bank in some countries, and with most rates falling between 0.25 and 1.0 percent. As regards the base of the taxes, they are generally imposed upon debits to (withdrawals from) checking, savings, and term accounts in banks and other financial institutions, and loan withdrawals (e.g., this is the case in Brazil and Colombia—in addition, the Colombian tax is imposed on credits of bank interest to accounts and on repos). In Ecuador, the base of the tax is somewhat different, with the tax being imposed not only on check cashing, but, notably, also on financial institution credits to checking, savings, term, loan, and other accounts, as well as on remittances abroad and on payments abroad by exporters and importers. As a result, in that case, both deposits to and withdrawals from the same accounts would be subject to the tax.15 Most of the countries have provided exemptions for transactions by certain types of institutions such as government agencies and charitable organizations, in some instances, inter-institution transfers and other transactions, including, for example, repos and transactions with the central bank.

In two countries—Argentina (through 1992) and Ecuador—the bank debit tax has been creditable against the income tax or the VAT.16 The authorities in both Colombia and Brazil have discussed adopting such a provision in the future as well as making the tax permanent. In December 2000, the Colombian tax was raised from 0.2 percent to 0.3 percent and made permanent, while the Brazilian government has indicated that the phasing out of the tax in 2002 will be possible only if substitute sources of revenue are found.

Finally, experiences with the bank debits tax a decade ago have informed certain design features in the round of adoptions beginning late in the 1990s. In particular, anti-avoidance measures such as restrictions on the use of cash for settlements and prohibition of re-endorsements of checks, or application of the tax to all but the first endorsement upon final settlement, have been implemented, apparently with some success in stemming disintermediation in the short run.

IV. Evaluation of the Latin American Bank Debit Taxes

As already noted, raising revenue in an “optimal” manner means balancing often-competing considerations. At different times and in different circumstances, these competing considerations may receive different weights in the policymaker’s calculus. In the case of bank debit taxes, the urgency of raising substantial revenue in the short term when the existing traditional instruments had proven incapable of providing the needed increases in effect meant that ease of collection received a very heavy weight at the expense of efficiency and (uncertain) equity considerations. This section considers how successful the taxes have been in achieving their primary objective of raising revenue as well as the costs of those taxes in terms of allocational distortions.

A. Revenue Productivity

As can be seen from Table 1, the short-term revenue performance of the taxes, particularly in Brazil, Colombia, and Ecuador, has been quite strong (on a gross basis; in the case of Ecuador, crediting of the tax is allowed for individual income taxpayers), underscoring that these are taxes that can be collected with relative ease. In particular, the taxes in Brazil and Colombia have produced revenues in the range of 0.6 to 1.3 percentage points of GDP for ad valorem tax rates in the range of 0.2 to 0.3 percent. This indeed would seem to confirm that these transactions taxes do serve, at least in the short term, their intended function. The performance of the tax in Ecuador in its first year must be judged, however, taking account of the fact that its tax has a broader base, applying to debits and credits, and that part of the gross revenues reported are creditable as well. It is interesting that the taxes imposed in Argentina and especially Peru were significantly less productive (gauged by the ratio of revenues as a percent of GDP to the average statutory rate). However, the Argentine economy was in crisis at the time the tax was introduced. The situation was by far the worst in Peru, where the economy and the financial system were in a state of collapse due to hyperinflation at the time the tax was in place.

Table 1.

Gross Revenue from Bank Debit Taxes

article image
Source: Country documents; and staff estimates.

Revenue in percent of GDP divided by average statutory rate.

Average of rates, adjusted for the period during which the tax was in effect.

Not adjusted for the fact that the tax base included both debits and credits.

The tax was reintroduced in April 3, 2001.

Adjusted for the period during which the tax was in effect.

In the case of Colombia and Ecuador, there was a decline in the yield of the tax relative to the rate in 2000 compared to 1999, indicating that economic agents may have found ways to economize in the use of taxed transactions.

The more recent taxes apparently have been more productive than those originally introduced. In the case of Brazil, in particular, a high revenue yield has been sustained over several years. However, the data in Table 1 reveals that in the case of Colombia, Venezuela, and Ecuador during 1999, monthly real revenues from the tax have been on a declining trend (Figure 2). In Venezuela, revenues held up through end-1999 from the tax’s introduction earlier in that year, but in 2000 have declined rapidly. The revenue from the Ecuadorian Impuesto a la Circulación de Capitales (ICC) has been high but declining—in 1999 it reached 3.5 percent of GDP with a rate of 1 percent; in 2000 it reached 2.3 percent of GDP with a rate of 0.8 percent. That is, the decline in revenue as percentage of GDP (1/3) was much more pronounced than the relative decline in the tax rate (1/5).

Figure 2.
Figure 2.

Revenue from the Bank Debit Tax in Selected Countries

(In real terms; month following the introduction of the tax = 100)

Citation: IMF Working Papers 2001, 067; 10.5089/9781451848908.001.A001

Source: Country data; and IFS database.1/Rate 0.38 percent through 6/16/00 and 0.30 percent afterwards.2/Rate 1 percent through 12/31/99 and 0.8 percent afterwards.

Leaving aside the trend decline, why have these newer taxes been more productive than the original ones? One answer may be that changes were made in the design of the taxes—notably with respect to prohibition or taxation of multiple check endorsements—that mitigated avoidance and disintermediation. It should also be noted that revenue productivity appears to decline with higher tax rates. For example, recognizing that the base of the tax in Ecuador was much broader than in the other cases, revenue productivity in that country was in fact considerably lower than in Brazil and in Colombia, where the tax rates are lower.

What can explain the relatively strong and sustained revenue performance of the tax in Brazil? The performance is consistent with the proposition that the tax may work better when it is implemented in times of other than extreme crisis when banking systems are often under considerable stress. Further, Brazil has already achieved a significant degree of financial sophistication; and, possibly most importantly, it does not have an established tradition of moving financial assets offshore. In addition, the fact that the tax rate in Brazil is relatively modest may also be facilitating the sustained revenue performance. It should be noted, however, that the bulk of the revenue appears to be coming from basic financial transactions—checks paid for consumption or business purposes—rather than from investment transactions and complex financial operations (see further discussion in the “Brazil” section below).

B. Allocational Effects

Despite their theoretical adverse allocational effects, the recent bank debit taxes may provide a better tax handle than other permutations of financial transactions taxes, involving as they do only withholding from resident institutions (of which, in some of the relevant countries, there are relatively few). It also appears that certain fairly basic design features can reduce the potential for avoidance, at least in the short run. For example, as already noted, in contrast to the earlier episodes, Brazil and Ecuador imposed the rule that only one endorsement per check is permitted. In Ecuador, bearer checks are banned; in Venezuela, as regards the most recent use of the tax, second and subsequent check endorsements are subject to the debit tax at time of settlement. The new taxes also commonly exempt certain debits, typically including: transfers between accounts of the same person in the same financial institution; accounts of public sector agencies and, sometimes, of charitable institutions; inter-institution accounts and transactions; accounts and transactions with the central bank; and interbank clearing operations (e.g., credit card settlements, and interbank transfers related to the operation of ATMs).17 Such exemptions of intermediate financial transactions reduce the potential for cascading.

Two factors complicate the search for empirical evidence documenting the adverse allocational impact of bank debit taxes. First, since in most recent instances the tax was introduced in times of economic stress, the impact of bank debit taxes on the financial sector is hard to distinguish from coincidental macroeconomic effects. Second, to the extent that the tax creates an incentive to move transactions into the informal economy and/or offshore—both of which are difficult to observe directly—it often becomes necessary to infer the impact of the tax from some assumptions concerning counterfactuals had the tax not been implemented. Although one needs to take account of the impact of simultaneous macroeconomic developments, some of the more dramatic examples of disintermediation are those that one would have expected a bank debit tax to have precipitated, supporting the hypothesis that these taxes at least contributed to disintermediation.

Figure 3 presents some summary data demonstrating that, in all four countries, which have used the tax in the last two years, the change in the ratio of currency outside banks to narrow money has concomitant macroeconomic developments.18 Nonetheless, there is some data that supports the proposition that the bank debit taxes may be exacerbating the shift to currency. In particular, Figure 4 indicates that, in the case of Colombia, there was a striking reduction in the number of checks cleared following the introduction of the tax.

Figure 3.
Figure 3.

Currency/Money Ratio for Selected Countries 1/

Citation: IMF Working Papers 2001, 067; 10.5089/9781451848908.001.A001

Source: Country data; and IFS database.1/ Currency outside deposit money banks as percentage of narrow money (Ml)
Figure 4.
Figure 4.

Colombia: Volume of Cleared Checks, 1998-99

Citation: IMF Working Papers 2001, 067; 10.5089/9781451848908.001.A001

Source: Banco de la República.

Beyond some summary statistics, much of the available evidence is, of necessity, anecdotal. Nonetheless, such evidence as is available to date does suggest that the imposition of these taxes has contributed to significant disintermediation of the banking system in almost every case.

  • Argentina—It was very easy for Argentineans to open bank accounts in Uruguay.19 To minimize tax payments, agents avoided depositing checks by endorsing them and passing them along to creditors. Checks (especially bearer checks) circulated repeatedly, without ever being presented to the bank for settlement. Multiple re-endorsements of checks were the rule. Fund staff reported in 1992 that this tax was eliminated in order to remove distortions in the financial system. The government stated when the tax was revoked ahead of schedule that it was inhibiting financial intermediation and fostering the informal economy.20 21

  • Peru—The government stated that the tax was inhibiting financial intermediation and fostering the informal economy. A 1992 analysis by IMF staff concluded that both the real revenues of the tax and the real level of current account deposits were declining. Among the practices induced by the tax, the study notes the clearance of transactions between enterprises transpiring directly without debiting bank current accounts, and a more frequent endorsement of checks. Evasion was also facilitated by the exemption of savings and housing lenders’ associations from the tax. In response, these associations created instruments that substituted for the checks of a regular banking system.22

  • Venezuela—The 1994 tax encouraged the use of cash and the movement of banking offshore. To avoid the tax, and to prevent losing balances in banks perceived as fragile, the public made large withdrawals from the banking system. Holdings of currency by the public doubled between January and December 1994, corresponding to an increase of 20 percent in real terms, while M2 fell by 8 percent in real terms during the same period. Major players in the stock exchange appear to have moved their trading abroad. In 1999, the year the tax was reintroduced, 75 percent of the traded value of equities in 13 major companies occurred in New York through depository receipt operations (ADRs).23 In February 2000, bank debits related to stock market transactions were exempted from the tax with the purpose of fostering private capital markets and the secondary market for government bonds.24


Given its strong and sustained performance, an assessment of the allocational effects of the bank debit tax (the CPMF) in Brazil is of particular interest. The evidence is consistent with the CPMF having altered financial and investment behavior, especially in the wake of its introduction at the end of January 1997. Between January and February 1997, demand deposits increased by almost 40 percent as the introduction of the CPMF reduced the opportunity cost of holding funds in noninterest-bearing demand deposits.

The evidence can be broken down between the impact of the tax on the markets for fixed interest instruments and its impact on other securities. In the case of the former, with the agreement of the regulatory authorities, financial institutions and investors have redesigned their investment strategies in ways to minimize the impact of the tax on fixed income markets by introducing new financial products. This implies that, in these instances, much of the impact of the tax has been mitigated, though there are still residual allocational effects since the market has ended up with a set of financial instruments that are somewhat different from those in place before the tax was introduced. Specific examples include:25

  • Term deposit certificates were disadvantaged on account of the CPMF being levied each time the certificates are renewed. In response, banks created fixed income funds, whereby the investor pays the CPMF only on the initial investment. In like manner, investors were discouraged by the CPMF from directly buying government bonds whose maturities are relatively short in Brazil. Again, the market created funds to provide this type of investment, thereby avoiding the CPMF.26 To take advantage of the fact that money transfers between financial institutions are tax exempt, fund managers created and regulators allowed ”exclusive” investment funds with widely diversified portfolios. Within these funds—which often have “free from CPMF” in their names—large investors can change their portfolio composition between stocks, bonds, foreign exchange, interest rate futures, and other instruments without the CPMF charge.

  • Commercial banks created privatization funds in which workers can invest by transferring resources from their accounts held with the Fundo de Garantia de Tempo de Serviço-FGTS (a severance fund), and moving the principal and interest back to their FGTS account after one year without paying CPMF.

  • Currently, some companies are seeking assurances that new special purpose companies would be exempt from CPMF. Such companies would issue debentures on behalf of another company to securitize receivables, particularly banking receivables. There are also proposals for the securitization of real estate based receivables with CPMF exemption.

While the impact of the CPMF on fixed-income markets has been significantly mitigated by the creation of new financial instruments,27 the CPMF appears to be having a more lasting impact on securities markets, notably its alleged role in exacerbating the migration of business from BOVESPA (the Sao Paulo Stock Exchange) to the New York Stock Exchange, NASDAQ, and other foreign equity markets. As the data in Table 2 indicate, the increase in the stock of shares set aside for depository receipt operations (ADR) abroad has been impressive. The volume of ADRs traded abroad increased by well over 50 percent in the two years from May 1998 to May 2000.28

Table 2.

Brazil: Stock of Shares Set Aside for Depository Receipt Operations Abroad

article image
Source: CVM.

It is difficult to be definitive about the reasons for these shifts, and several factors other than the CPMF are also likely involved. These factors include: uncertainty about the rights of minority stock holders, of which the related regulating law is being revised by Congress; the increased integration of Brazilian firms in international capital markets;29 and access to new sources of liquidity for Brazilian firms. However, the impact of bank debit taxes in other countries such as Venezuela (as noted above) and Sweden;30 the economic argument that the impact of a bank debit tax would likely be most discernible in markets such as equity markets that are cost sensitive; and the observations of local observers31—all suggest that the CPMF has contributed to the growth in ADR operations.

Finally, as an additional example of how officials responded to the tax, the National Monetary Council simplified the rules for foreign investment in the stock market. Under the new rules, a smaller number of foreign exchange transactions are subjected to the CPMF. The Council also allowed nonresident individual investors to buy stocks directly, instead of only through mutual funds, thus reducing transactions costs and partially offseting the burden of the CPMF.

On balance, the anecdotal evidence and summary data are consistent with the view that the impact of the CPMF is most readily apparent precisely where it would be expected—namely, in financial markets, where the margins in individual transactions would likely be modest. This indicates that, if a bank debit tax is to be used, the precise definition of the base of that tax is important—exempting securities transactions from the tax would work to reduce the overall adverse allocational implications of the tax, assuming that influencing these markets is not an objective of economic policy.


When the tax was introduced in mid-November 1998, a preferential rate of 0.012 percent—sinstead of the 0.2 percent standard rate—applied to interbank market transactions, in order to minimize damage to that market. However, in March 1999, the constitutional court struck down the preferential rate, causing the interbank market to quickly disappear. It only reemerged in August 1999, when interbank transactions in domestic currency were exempted from the tax (Figure 5). During the intervening period, interbank money needs were addressed by a surge in bilateral transactions (repos and reverse repos) with the central bank, which were not taxed.32

Figure 5.
Figure 5.

Colombia: Interbank Market Weekly Turnover, 1998-May 2000

Citation: IMF Working Papers 2001, 067; 10.5089/9781451848908.001.A001

Source: Banco de la República.

During the first week the tax was introduced, the daily volume on the foreign exchange market fell from US$170 million to US$30 million, the treasury bill market volume fell to less than 10 percent of normal, and the treasury bill interest rate rose by 2 percentage points. Simulations made by the private financial sector in 2000 indicate that most interest paid on bank deposits is absorbed by the tax, with cash in circulation growing at an annual rate of 29.5 percent, compared with 16 percent the year before, while inflation was actually substantially reduced between 1998 and 1999.33 34 Following the introduction of the tax, there was a significant decline in the volume of transactions settled through checks cleared in the first half of 1998 (before the tax) and in the first half of 1999 (after the tax) (Figure 4). The number of checks declined by a full 47 percent.

To avoid the tax, economic agents developed new, nontaxed instruments, including: multiple endorsements of checks (which are not prohibited or taxed in Colombia); and extension of the average term of fixed-term deposits. When taking office in July 2000, the Minister of Finance, Mr. Juan Manuel Santos, declared that the tax is a necessary evil and that he would like to replace it with alternative revenue sources. At the end of 2000, the tax was made permanent with a 0.3 percent rate. Financial sector observers reported an intensification in the use of cash and a lower turnover in bank accounts.


An interesting instance of tax avoidance was the development of a “popular” offshore facility. In Aguas Verdes, a small city in the most southern tip of the country facing the Peruvian city of Huaquillas, banking activity developed to facilitate avoidance of the Ecuadorian ICC. Moreover, as implied by the revenue data presented earlier, the tax suffered significant evasion. There was, for example, a high profile criminal investigation against a former director of Banco del Progreso S.A. accused of having failed to transfer to the tax administration US$12 million in ICC revenue collected from clients in February–March 1999. The principal defendant has been jailed.

C. Crediting of the Tax Against Other Taxes

None of the countries at present levying a bank debits tax (Argentina, Brazil, Colombia) permits crediting against other taxes. While in effect, the Ecuadorian tax could be offset against the personal income tax, and the Argentine law empowers the government to introduce crediting against the VAT, the income tax, and the single tax on small traders. What are the implications of crediting?

In an extreme case, where the tax is completely creditable, no net revenue will be raised. If there is no net additional tax paid by a given actor, there will be no distortions of behavior, and no tendency for disintermediation. In effect, there would then be no separate tax since the bank debit tax would serve simply as a collection device for existing taxes. If, however, such a mechanism is designed to permit only partial crediting, or functions that way in effect, at the margin the bank debit tax will be paid, and, at the margin, will lead to distortions. The behavioral response of businesses, if crediting were permitted against the corporate income tax, would depend upon whether they had credits in excess of their profit tax liability. This factor itself may lead to distortions, for example, by causing companies to reduce debt where interest deductions are no longer useful at the margin.35

In some cases, the authorities’ interest in bank debit taxes extends beyond their revenue potential. Thus, even a total offset of revenue through crediting may not mitigate the desirability of the tax because, in their view, regardless of revenue, the tax supports the enforcement of other taxes by permitting, inter alia, the collection of information that is otherwise unavailable, for practical or legal reasons.36 While assistance in legitimate tax enforcement is sorely needed in countries with less highly developed tax administrations, this second-best approach may distract from the first-best solution of directly strengthening tax administration. It has also been argued that the tax on bank debits reaches, indirectly, income that escapes the traditional forms of taxation. For example, in Colombia, where security considerations make impractical the use of large quantities of cash, a study estimated that about 21 percent of the revenue from the debit tax is generated by such illegal activities as smuggling, money laundering, and extortion.37

V. Conclusions

This paper is narrowly focused on the bank debit taxes that have been implemented in a number of Latin American countries. While the recent implementation of most of these taxes necessarily limits the information available, sufficient evidence has been uncovered to suggest the following conclusions and lessons:

  • The recently implemented bank debit taxes have been successful in raising revenue in the short term.

  • The market responses to the tax in various countries indicate that there have likely been adverse allocational impacts, including, in particular, significant financial disintermediation.

How, then, do these taxes ultimately fare from a policy standpoint? Bank debit taxes are not in principle an efficient source of revenue. Moreover, use of bank debit taxes as a long-term tool risks significant domestic disintermediation, which could be difficult to reverse even if and when the bank debit tax is revoked. In addition, the true cost of their introduction may be underestimated, particularly regarding their adverse impact on the yield of other major taxes if economic activity is driven underground. Finally, the incidence of these taxes is not transparent. The overall policy conclusion is that the use of these taxes should be avoided.

If they are to be used, however, such evidence as is available suggests that it should only be at low rates, and as a temporary expedient in the face of fiscal pressures and a deterioration in the performance of more appropriate tax instruments due to weak tax administration. The objective should be to replace the tax as soon as feasible with revenue from taxes such as the VAT and income taxes. Moreover, the evidence indicates that the tax can be strengthened by ensuring that the obvious avoidance measures, such as multiple check endorsements, are ruled out and by defining the base so as to exclude some of the most cost-sensitive transactions, particularly those in securities markets.


The Australian Debit Tax

In April 1983, Australia introduced a tax at the national level on withdrawals from checking accounts in banks.38 In 1990, this tax was transferred to the state level and put on a voluntary basis, although the central government continued to be responsible for its collection. If a state chooses not to impose the tax, grants to that state are reduced by an amount equal to the foregone revenue. Therefore, the states have a strong incentive to adopt the tax.

At present, this state levy, which is called a “debits tax,” is applied by all but one state, and has a schedule of rates starting with A$0.30 for debits under A$100 and reaching A$4 for debits of A$10,000 or more.39 Exemptions apply to accounts held by domestic and foreign government agencies, charitable and religious institutions, public or nonprofit hospitals, educational institutions, welfare recipients, and savings accounts in general. In fiscal year 1988–89, this tax constituted 3.2 percent of total tax revenues in New South Wales, 3.3 percent in Western Australia, and 4.5 percent in Victoria.

Several states have acknowledged in their budget papers the deficiencies of the bank debit taxes, particularly from the point of view of efficiency and equity. On the latter point, it is said that the taxes burden the less well off more than those who are wealthier (on the grounds that the Australian taxes are not completely ad valorem, and that large firms can better “bundle” their transactions). Some firms and even consumers have overseas accounts, or accounts in banks in Queensland, the one state where the tax is not imposed. Further, the tax is not imposed on accounts without checking facilities (the tax has induced the overuse of savings accounts, even by enterprises), nor on credit unions, which allows avoidance of the tax. The tax is said to impede development of the financial system. It is believed to be among Australia’s most inefficient.40

The Debits Tax was slated for removal with the major tax reform that will launch the General Sales Tax (a tax on value added) on July 1, 2000. However, due to concern about potential revenue losses, elimination of the Debits Tax was deferred to July 1, 2005 pending a review by a panel comprising representatives of the central and state governments.


Traditional “Stamp Duties

A number of countries levy taxes on transactions such as the issuance or transfer of securities, the transfer of immovable property, the conclusion of contracts, and other transactions that require legal documents. These taxes go by different names, but a traditional name is stamp duty, derived from the fact that historically the taxes have been paid by affixing stamps to documents.

The modern trend has been to narrow the scope of the application of stamp duties. The stamp duty was first introduced in the British Empire in 1671 as “an imposition on proceedings at law,” derived from Dutch practice (Sinclair, 1803). It was basically a tax on contracts, and over time acquired an extensive list of taxable objects (mostly deeds and conveyances). In the United Kingdom, the present stamp duty is part of the “modern” system of documentary stamp taxes dating from 1891, and, with respect to corporate securities, was originally applicable both to their issuance and transfer. The tax on original issuance of corporate securities was repealed in 1988. The tax on transfers is still in place, at a rate of 0.5 percent, though this tax is currently under attack from institutions in London on the grounds that it is taking business out of the city in favor of jurisdictions without such taxes (BBC News, May 19, 2000). Besides securities transfers, the most important element of the current stamp duty in the United Kingdom is transfers of property (in practice, it is mostly transfers of immovable property that are reached by the tax) (Quinlan, 1998).

Most countries with stamp duties currently levy this tax primarily on two types of transactions: (a) the transfer of immovable property or the registration of mortgages, and (b) the issuance or transfer of stock shares. Some countries still impose stamp duties on a broad range of documents (e.g., Portugal). Tax rates on transfers of immovable property can be substantial (the top rate in the United Kingdom for transactions in excess of £500,000 is now 4 percent, and Belgium imposes a 12.5 percent tax on transfers of immovable property) (European Commission, 1996).

Traditionally, stamp duties worked well because of their self-enforcing nature—in order to be legally binding, documents had to be stamped by the government in accordance with the tax. The imposition of stamp duties, however, invariably encourages economic agents to conclude transactions without using documents subject to duty. For example, taxpayers may seek to avoid duty on the transfer of immovable property by using long-term leases. As a result, long-term leases have typically also been subjected to tax. Often, duty can be avoided by executing documents offshore. In the case of securities transfers, a legislative response has been to impose a higher rate of tax on transfers into a system, which allows subsequent transfers to be made without tax (Quinlan, 1998).


Summary of Taxes on Financial Transactions

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

The taxation of bank debits in Argentina goes back to 1976 when a 0.2 percent emergency tax was introduced but repealed after three months. In October 1983, the tax was reintroduced with a rate of 0.1 percent, which was increased to 0.2 percent in August 1985.

If the bank decides to support the tax liability instead of charging the client, the tax base is grossed up.

The initial period (through end-1992) was curtailed by six months. Provinces introduced a tax of fixed amount per check, which was collected until 1996.

The Supreme Court suspended the tax, adopted in July 93 for collection in late August 93, from 9/15/93 through 12/31/93. The amounts received in 1993 (0.07 percent of GDP) were refunded.

Rates vary from year to year and are not uniform across states.


  • Edwards, Jeremy, Michael Keen, and Matti Tuomala, 1994, “Income Tax, Commodity Tax and Public Good Provision: A Brief Guide,” FinanzArchiv, Vol. 51, No. 4, pp. 472-87.

    • Search Google Scholar
    • Export Citation
  • Shome, Parthasarathi and Janet Stotsky, 1996, “Financial Transactions Taxes,” Tax Notes International, January, pp. 47-56.

  • Sinclair, John, 1803, The History of the Public Revenue of the British Empire, 3 rd ed., Vol. II (London: T. Cadell and W. Davies).

  • Spahn, Paul B., 1995, “International Financial Flows and Transactions Taxes: Survey and Options,” IMF Working Paper 95/60 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Umlauf, Steven, R., 1993, “Transactions Taxes and the Behavior of the Swedish Stock Market,” Journal of Financial Economics, Vol. 33, Issue 2, pp. 227-40.

    • Search Google Scholar
    • Export Citation

The authors would like to thank their many IMF colleagues who provided useful comments and information.


The term “financial transactions tax” as used here means a tax levied on each instance of specified banking, equity, currency, securities, or other financial dealings between a broad base of market actors. “Bank debit taxes” are a subset of financial transactions taxes, levied on withdrawals from or other debits to bank accounts, and generally including the clearance of checks, cash withdrawals, payment of loan proceeds, withdrawals through ATMs, and possibly charges to bank issued (or other domestic) credit cards. The bases of the bank debit taxes in Latin America are detailed in Appendix III.


In contrast to the Tobin-type taxes, which have the objective of altering behavior to minimize “undesirable” market effects, the bank debit taxes aim at raising revenue in a manner that minimizes the impact on economic behavior.


Little published research exists evaluating the recently introduced bank debit taxes, nor is much actual data available. This study accordingly relies in part on facts and assessments regarding these taxes obtained from IMF staff reports and provided by expert officials in the countries in question. Future work in this area would usefully include econometric analyses to obtain a finer discrimination between effects associated with bank debit taxes and those associated with underlying economic developments.


The reintroduction of the tax in Argentina in April 2001 is too recent to warrant analysis.


This assumes that all transactions flow through the banking system, that the taxes are applied only to one way transactions, and that mechanisms exist to eliminate from the base transactions between accounts within and between financial institutions on behalf of the same taxpayer (e.g., no tax on the settlement of a credit card bill where the charge to the credit card has already been subjected to the tax).


The distortionary nature of these taxes has been recognized for some time in Australia where similar taxes are used at the state level (see Appendix I).


Including from abroad—the bank debit taxes as currently practiced apply equally to branches of foreign banks and to nonresidents.


Taking the example of Tobin taxes, the annual interest rate that would be required to match a 4 percent rate on domestic currency deposits given a Tobin-type tax at a 0.5 percent rate would range from 4.6 percent, for a foreign currency investment with five-year maturity, to 90.7 percent, for one with a three-day maturity (Spahn, 1995). Of course, this is one of the reasons Tobin taxes have been proposed.


As with other transactions taxes, the excess burden of the bank debit tax depends in part on the elasticity of demand for the taxed service. Demand for checking accounts may be quite inelastic at very low tax rates; however, the elasticity would increase as the rates rise, rendering the tax more inefficient.


Although not discussed due to the Latin American focus of the current paper, Australia introduced a similar bank transactions tax at the federal level in 1983. This became a state level tax in 1990. The tax is now slated to be removed by 2005. See Appendix I.


Decree 2331/98.


Tax Notes International, April 25, 1994.


The taxation of both debits and credits is also a feature of the tax recently introduced in Argentina.


The tax law adopted in April 2001 by Argentina allows for credit against the VAT, income tax, and the small traders tax; however, the government preferred to adopt a rate (0.25 percent) much lower than the maximum (0.60 percent) instead of enabling the credit mechanism.


In addition, the tax base in Ecuador exempts withdrawals from savings accounts and from ATMs. Since the Ecuadorian tax applies to credits as well as debits, however, these latter exemptions in effect simply make the tax in those cases a one-way transactions fee.


Is the rising trend only evident in those Latin American countries that rely on the bank debit tax? A review of the behavior of the cash to narrow-money ratio in some Latin American countries that did not have the tax in the same period of time presents a mixed picture. Specifically, a rising trend is evident in Mexico whereas the ratio declines in Uruguay and is trendless for Argentina and Mexico.


Remarks by Guillermo Calvo, Carlos Rodriguez, Orlando Ferreres, and Juán Alemann in Ambito Financiero, Buenos Aires, May 11, 2000.


Foreword of Decree No. 20-92-PCM, March 17, 1992.


At present, there are strict limitations to the number of check endorsements.


Superintendencia Nacional de Administración Tributaria (SUNAT).


La Republica, Caracas, January 15, 2000. The issue arises as to whether the increased popularity of ADRs reflects the tax or just globalization trends in Latin America in general. The data we have is only partial—in particular, it concerns issuance of stocks rather than trading volume. The former, however, may be a more useful indicator of the interest in ADRs as a result of globalization. New York Stock Exchange data indicate that, in the cases of both Argentina and Mexico, the peak period of new companies issuing ADRs was 1992–1994, which is consistent with the fact of the initial surge in ADR activity being early in the decade.


La Republica, Caracas, February 10, 1999.


The facts in this section have been gleaned from numerous articles in various Brazilian and international business publications.


Intrafund transfers are not taxed—neither purchases nor redemptions at maturity of government bonds are taxed when made by the funds. In the case of individual investors, both (plus the rollover of the proceeds) are taxed.


If the new instruments are perfect substitutes for those they replaced, the distortionary impact of the tax would be eliminated.


Trade in derivatives, mostly conducted through the Commodities and Futures Exchange (BM&F) and money market operations, did not decline much with the advent of the CPMF because most transactions in these markets have financial institutions at both ends, and are thus exempt from the tax.


As noted in footnote 25 above, the popularity of ADRs in other Latin American countries appears to date to an earlier period. In the Brazilian case, however, and focusing again on the New York Stock Exchange, only one Brazilian company floated ADRs in the period 1991–96, whereas 12 companies did so in 1997/98, a period when the tax was in place.


When a transactions tax of 2 percent (1 percent on both purchases and sales) was introduced in Sweden in 1986, 30 percent of all trading in the Stockholm Stock Exchange migrated to London. By 1990, the proportion of Swedish share volumes trading in London had grown to over 50 percent (Umlauf, 1993).


According to the president of the Securities and Exchange Commission (CVM), “The main impact of the CPMF is the movement of business to the New York markets, where business people can buy the same stocks at lower cost.” (Gazeta Mercantil InvestNews, May 29, 2000). In addition, reportedly, Central Bank Governor Fraga recently declared that “the CPMF is the great enemy for the formation of capital and savings in Brazil.”


Ignacio Lozano E. and Jorge Ramos F., “Análisis sobre la incidencia del impuesto del 2x1000 a las transacciones financeiras,” Borradores de Economía, Series No. 143 (Santa Fé de Bogotá: Banco de la República, 2000).


“2x1000: Para Siempre,” Carta Financiera, March 2000, pp. 54–58 (Bogota: ANIF).


IMF staff estimate (December on December).


The countries that do not permit crediting generally do allow the debit tax as a deduction from taxable corporate income, leading to similar implications for the business sector.


In Brazil, a law was passed in January 2001 allowing the use of information collected in connection with the debits tax to control compliance with other taxes. In Ecuador, the data collected through the transactions tax purportedly permitted the authorities to circumvent banking secrecy requirements.


From a study prepared by the Dirección de Impuestos Nacionales y Aduanas-DIAN, cited by the President of the Third Commission of the Chamber of Representatives (La República, Bogotá, May 25, 2000). Of course, it is also possible that disintermediation caused in the first instance by the tax could, itself, indirectly foster or facilitate underground activity.


The tax was originally called the “Bank Account Debits Tax” and became known as the BAD Tax. The tax also applies to payment order accounts with nonbanking financial institutions.


The unweighted average of these rates, taken at midpoints (except for the top rate, which is taken at the lower limit), corresponds approximately to an ad valorem rate of 0.2 percent.


“Tax Reform: Not a New Tax, A New Tax System” (Canberra: Treasurer of the Commonwealth, 1998).

Bank Debit Taxes in Latin America: An Analysis of Recent Trends
Author: Mr. Isaias Coelho, Ms. Victoria J Perry, and Mr. Liam P. Ebrill