{ _ WP?s l A3 Policy Research WORKING PAPERS Trade Policy Policy Pesearch Department The World Bank May 1993 WPS 1143 Tariff Rates, Tariff Revenue, and Tariff Reform Some New Facts Lant Pritchett and Geeta Sethi Tariff rates on specific products and the ratio of tariff revenue to import value are only tenuously related. Above a 50 percent rate, collected rates do not increase at all despite increases in official rates. Policy Rescarch Woiing Papers disseminate the findings of work in progress and encourage the exchange of idcas among Bank staffand allothecs intermsted in developmentissues.Thescpapers, distributed by the RescarchAdvisory Staff, carry thenames of the authors, reflect only theirviews. and should be used and cited accordingly. Tbe findings, interpretations, and conclusions are the authors' own.They should not be attributed to the World Bank. its Board of Directors, its management, or any of its member countries. Policy Research *~~~~~~~ Trade Policy WPS 1143 This paper - a product of the Trade Policy DiN ision, Policy Research Department - is part of a larger effort in the department to assist in the design of tariff reforms. The study was funded by the Bank's Research Support Budget under research project "Design of Tariff Reform: Theory, Evidence, and Implication"(RPO 676-77). Copiesofthe paperare available free from the World Bank, 1818 H StreetNW, Washington, DC 20433. Please contact Maria Femandez, room S9-035, extension 33766 (May 1993,23 pages). The ad valorem tariff rates on specific products Collection rates appear to level off at roughly 50 and the ratio of tariff revenue to import value, percent. (In Kenya, collected rates are lower for the collected rate, are only tenuously related, high-tariff than for moderate-tariff items. contend Pritcheut and Sethi. Assigning lower rates for high-tariff items would actually increase revenue on those items.) Using tariff and revenue data (at the tariff code line level of detail) for three developing The implications of these findings are countries, Pritchett and Sethi cnmpare the twofold for calculating general revenue: statutory ad valorem tariff rates (official rates) with the ratio of tariff revenues to import values * Rates are not the critical determinant of (collected rates). They document four facts: revenues. The revenue implications of large rate changes can be offset by modest changes in the T* he collected rate or any given ^.em of the system of exemptions, for example. T he benefit tariff code has almost no relationship to the of eliminating exemptions is primarily official rate for that item. transpan.ncy. The costs of programs that provide import exemptions for, say, regional promotion, * The variation of collected rates around the are often hidden ir customs statistics. Official rate increases as the level of the official rate increases. ' If pressures that cause collected rates not to increase one-for-one with tariff rates will * The collected rates increase much less, on continue to be present in any tariff regime, then average, than one-for-one with the official rates. these must be factored into tariff reform design. * Above a certain level, collected rates do not increase at all despite increases in official rates. The Policy Research Working PaperSeries disseminates the findings of work under way in theBank. Anobjectiveofthe series is to get these findings out quickly, even if presentations are less than fully polished. The findings, interpretations. and conclusions in these papers do not necessarily represent official Bank policy. Produced by the Policy Research Dissemination Center Tariff Rates, Tariff Revenue and Tariff Reform: Some New Facts Lant Pritchet. and Geeta Sethi Tariff rates. Tariff revenue and Tariff reform: Some New Facts Tariff reforms are one aspect of the package of trade policy reform supported by World Bank adjustment lending. Tariff reforms aim to rationalize the tariff code, reduce t!2e dispersion of rates, and lower the average tariff rate. However, since many tariff reforms are undertaken during periods of stabilization and fiscal austerity, the potential loss of tax revenues from lowering and unifying tariff rates has been perceived as an important constraint on tariff reform (Rajaram 1992, Mitra, 1992). This paper uses data from three developing countries at the tariff code line level of detail to examine the relationship between tariff revenues and tariff rates. We construct for each item of the tariff code the ratio of import tax revenues to import value, the "collected rate" of tariffs. Comparing these collected rates to the official statutory rates of the tariff code we demonstrate four facts about the relationship between tariff rates and tariff revenues'. D Collected and official tariff rates are almost completely unrelated. In pre-reform countries, differences of official rates of tariff explain only a quarter of the variation in collected rates. * The variance of collected rates increases strongly with the level of the statutory rate. * The collected rate increases much less than one for one with increases in the statutory rate. 1 The words "official" or "statutory" rate of tariff are used interchangeably to mean the ad valorem rate recorded as "the' rate for each tariff code item. Of course, in many cases, different rates are specified within the tariff code for the same item, with exceptions granted (or changes added) depending on country of origin or end use. For simplicity, we refer to these as exceptions from the "official" rate. -2- * Above a certain level of the official tariff rate, further increases in the official tariff rate produce no increase (and there is some evidence of a decrease) in the collected rate. One way to frame the four facts is that one might suppose if all tariff revenue were collected, that, in principle, a relationship would exist such that for the il item: Collected ratei = ct + o * official rate ' + ei, the explanatory power of this relationship is high, , = 1, and the relationship is linear. But our research shows otherwise: the explanatory power is low, a does not equal 1, the relationship is not linear, and, incidentally, e' does not have a constant variance. We argue that these facts represent general features of pre-reform tariff codes in developing countries, with two immediate implications. In refcrms of the entire system of tariffs and tariff revenue collection, the change in official rates of tariff, especially at the high levels, is likely to be the least important element for revenue. Changes in procedures for assessment of import value, granting of exemptidns, and collections are likely to be much more ~important. Second, simulations that assume that tariff revenues fall one for one with rates (such as those supported by the Bank's SINTIA software) overstate the impact of rate reductions on revenues by assuming constant collection rates. The impact on revenue needs to carefully assess the exemption status of imports. This paper has six sections and a conclusion. The first section describes the data and the tariff regime in Pakistan, Kenya, and Jamaica, the countries for which tariff line level data on revenue is available. The next four sections present the evidence for each of the four stylized facts listed above. The sixth section discusses evidence from other countries suggesting these four features are general to pre-reform tariff systems in LDCs. The conclusion summarizes and 3- discusses the implications for the analysis of the relationship between revenues and tariff reform. I) Tariff regimes and data on tariffs. collections. and value As part of ongoing research on tariff reform we have collected data on tariffs at the tariff code line level for a large number of countries (see Pritchett and Sethi, 1992). For three of these countries, Pakistan, Kenya, and Jamaica, we have been able to obtain official rates of duty, iTnport values and import duty collected. Table I summarizes some characteristics of the tariff code(s) for these three countries. The second column gives the number of separate items distinguished in the tariff nomenclature of the countries. These tariff codes are quite detailed and distinguish between 3,000 and 5,000 items2. Table 1: EBasic facts on the tariff code Number of Number of Average rate of the: tariff items official rates official collected with imports l Pakistan 4317 15 42 32 Kenya 3392 33 22 13 Jamaica 3303 10 16 12 Notes: The average of the official rates is import weighted. Items with zero import value are excluded. The third column indicates the total number of rates of import duty for each country. In addition to import duty, various fees (e.g. customs processing fees) and additional taxes (e.g. 2 Tariff nomenclatures for individual countries are based either on the Brussels or the new Harmonized System. Countries then further differentiate products to the six, eight or ten digit level. The number of items distinguished ill the tariff code varies across countries, from as few as 2000 to over 12,000 (see Pritchett and Sethi, 1992). -4- excise, sales, or luxury) may be levied and collected on imports at the port of entry. We focus only on the component of import duties as the others are either small, do not vary across items or are not imp,_rt taxes. Our collection numbers correspond as closely as possible to these import duties3. The last two columns present the import value weighted mean of the statutory tariff rate and the average collected rate4. The import weighted tariff rate gives the hypothetical revenue, that is the revenue from the tariff code if all import taxes were cc lected at the official rate. As with other countries, actual revenues are far below this hypothetical as the average official rate is almost twice as high as the average collection rate. The difference between hypothetical and actual revenue is due to exemptions5. Since both revenues and import values are taken directly from customs statistics, tariff losses from smuggling, the underdeclaration of import values (underinvoic.ig) and the misdeclaration of items are omitted. To the extent that these are important, we understate the gap between actual revenues and those that would be collected if all imports were paid on the official tariff. The differences we examine are entirely due to recorded exemptions from the tariff code. The three countries were at various stages of their trade reforms in the tariff codes we use. Jamaica had already had several rounds of tariff reform, with reductions in the average rate. Kenya's trade reform has mainly focused on import licensing to date, including some 3 For Jamaica, import revenue is from Common External Tariff (CET). For both Kenya and Pakistan, the revenues are based on collection from just customs duty. I Tariff code items with zero import value were excluded from the all calculations are the collected rate could no be constructed. 5 At this stage we'll just define exemptions this way. What these exemptions constitute will be discussed below. -5 tariffication that has raised the unweighted average. Paldstan's tariff code has already undergone substantial rationalization as part of their adjustment efforts. II Collected rates and statutory rates: not even cousin Official rates have almost no influence on tariff code items at collected rates across tariff line items. For a few selected levels of the official tariff rate, Table 2 presents the number of items at this tariff rate, the mean collected rate of items at this rate, the &tandard deviation and coefficient of variation and the 25th and 75th percentile. These results heuristically preview a number of the results to be presented statistically later in the paper, the interesting feature in the present context is tie huge variation of collected rates for items with the same official rate. For example, in Paldstan the mean collected rate for the 899 items with a tariff rate of 80% is 51%, but the standard deviation is 31. One quarter of the 80% rate items paid duty less than 21% while one quarter paid a duty abo e 78%. For the 495 items with a 100% tariff, one quarter paid less than 7% while one quarter paid more than 94%. Similarly for Kenya, among the 429 items with a tariff rate of 50% the mean collected rate is only 26%, one quarter paid less than 6%, one quarter paid more than 26%, and the standard deviation is 27. -6- Table 2: Summary Statistics of Collected Rates By Level of Official Rate Official Number of Mean Standard 25 75 1h rate items Collected Deviation percentile percentile .______ ________ _ (C.V.) _ _._ . Pakistan 20 547 15 14(92) 5 20 40 789 32 18 (56) i.46 40 60 267 40 21 (53) 23 58 80 899 51 31 (61) 21 78 100 495 52 40 (76) 7 94 125 ! 605 54 47 (88) 3 95 Kenya 20 331 5 10 (194) 0 3 30 950 9 14 (156) 0 22 40 435 20 20 (101) 0 24 50 429 26 27 (106) 6 26 60 206 43 39 (92) 26 56 80 306 31 30 (95) 26 31 100 108 36 37 (102) 4 34 Jamaica 10 1375 7 7(91) 3.18 10 30 517 16 12 (76) 5 26 45 1 662 24 17 (72) 7 44 The fact that the variation of collected rates even for the same official rate is so large suggests that there is very little systematic relationship between the statutory tariff levels and actual collection rates. Table 3 illustrates that statement statistically. An analysis of variance -7- (ANOVA) decomposed the variation of the collected rates of individual items about the overall mean into a component associated with variations in official rates and the fraction of the variation of collected rates that actually occurs for items with the same official rate6. For Pakistan only 23% of the collected rate variance is explained by official rates; for Kenya this is only 27 % and for Jamaica, after several rounds of tariff reform, the number is up to 31 %. Typically less than one quarter of the variation of the tariff revenues collected across products is related to variations in the official rates. This low explanatory power is especially surprising giveni the enormous variation in the official rates.. Table 3: Explanatory power of official rate for collected rate [Functional Pakistan Kenya Jamaica Form: _ ANOVIA .23 .27 .31 Regression R-squared Linear .16 .21 .29 Log-Log .18 .24 .30 Linear, with .22 .22 .29 spline I _ Quadratic .20 .21 .29 Notes: Functional forms of regressions of collected rate (CR) on the official rate (OR) are: Linear, CR-= a + g *OR; Log-Log, ln(CR) a + j *ln(OR); Linear with spline, CR= a + a *OR+ j *(OR-t), where t is the level of official tariff at which the slope changes; Quadratic, CR= c, + 0 *OR + a l*OR2. 6 Since the official rate is a discrete variable (the number of different levels is given in table 1) it can be treated as a "treatment" variable in a traditional analysis of variance. An alternative procedure is to assume some relationship between the collected rate and the official rate of the forn: Collected rate = f (official rate). If the collected rate were primarily a function of the tariff rate, one would expect a relationship of this type to have high explanatory power. The second tc, fifth lines of Table 3 report the R- squared of a regression of collected rates on statutory ratus for a variety of functional forms for each country. The number is very low7, confirming the results of the analysis of variance. For Kenya and Pakistan the R2 is undiformly below 25% while forJamaica it is roughly 30%. A low R-squared across the various forms reveals that little of the variation in collections is related to variations in official rates. III) Increasing variation of the collected rate The second fact is that the variation of the collected rate of items with the same official rate increases with the level of the official rate. In Table 2 we can see that for each country the standard deviation increases substantially. For instance, in Kenya, when the tariff is 20% the standard deviation is 10, increasing to 20 at 40% and 39 at 60%. Of course, since the standard deviation is not scale invariant this is not altogether surprising. The coefficient of variation (in parenthesis in column 4 of Table 2) divides the standard deviation by ihe mean to achieve a scale invariant number. The coefficient of variation indicates that not only do higher statutory rates have higher absolute deviations, but, except for the lowest rates, they are generally larger as a fraction of the mean value as well. I Since the regression formulations impose additional structural restrictions on the relationship, the R2 will be lower than the ANOVA. -9- Table 4 reports statistical tests of increasing variance in several forms. Ir. the first row are the results of an ANOVA with the levels of the official rate as the treatment variable and the variance of collection rates as the dependent variable (as opposed to the level in Table 3). We find that the level of the official tariff is at least as successful in explaining the variation of collected rates around the official rate as it was in explaining actual collections, as the ANOVA "fraction explained" of .41, .13 and .42 are substantially higher than those in Table 3 for Paldstan and Jamaica. We can predict with more confidence that increasing the tariff will increase the variance that it will raise mean collections. In the following rows, the absolute value (or squares) of the residuals of the collection rate regressed on the statutory level of the tariff are themselves regressed on the level of the statutory rate8. The explanatory power of the log-log specification shows that the percentage variation of the residual increases along wiJh the absolute level. Table 4: v ane of Collecl i Rates and the Level of Official Tariff Pakistan Kenya Jamaica ANOVA 0.41 0.13 0.42 Regression form R2: e I (linear) 0.38 0.08 0.41 e'I (log) 0.41 0.10 0.42 (eI)A2/sA2 0.21 0.04 0.24 Notes: Row 2 regresses the absolute values of the residuals of the linear regression on the official rate. Row 3 regresses the absolute values of the log-log regression on the log official rate. Row 4 regresses the normalized squared residuals of the linear regression on the official rate. 8 This is essentially a test of heteroskedasticity of the error term, which in thus case is not so interesting in its own right. - 10- L)YLHigher rates. not higher:revenues There are a number of good reasons to believe that tariff revenues will not increase one for one with increases in tariff rates. To the extent that the good itself has some elasticity with respect to the tariff rate the value of the import in a category will decrease with an increase in the ta.-ff rate (ceteris paribus, of course). A second revenue elasticity is relevant as the value of imports reported for tariff purposes will decline, even for a given level of actual imports, because underinvoicing and smuggling as the tariff rises. This study documents a third channel. Even for a given value of imports declared to customs, the ratio of imports coming in with exemption will increase as the tariff rate increases. This will happen for at least three reasons. First, even under a fixed set of exemptions (say a scheme that provides exemptions to exporters), as the tariff is raised, the value of imports coming in under non-exemptions will decrease (with the magnitude depending on the price elasticity of the non-exempt goods) while the value coming in under exemption will remain constant, and this will increase the fraction under ex2mption. Second, the incentive to lobby for exemptions will increase with the level of tariffs. Exemptions either for specific types of goods, for specific importers, or just plain discretionary are endemic in tariff systems. In a situation in which the lobbying for exemptions and the degree of exemptions granted is endogenous, an increase in the rate will increase the fraction of imports coming in under exemption as exemptions increase. Finally the temptations for abuse of any system of exemptions will increase with the level of the tariff. For instance, exemptions are commonly granted to exporters, or diplomatic missions and charitable activities, or to returning residents. The higher - 11 - the tariff, the larger the incentive for false diversion of imports under any scheme into other channels. This third channel for low revenue elasticity to tariff changes is potentially quite important, as seen in Table 2. For Pakistan, the mean of the collected rate for items with a 60 percent tariff is 40 percent. For items with an 80 percent tariff, the mean of the collected rate increases only to 51 percent and then is roughly the same (52 and 54 percent) for items with official rates of 100 and 125 percent. In Kenya, the mean of the collection rate decreases from 43 percent at 60 percent official to 31 percent at 80 percent and 36 percent at 100 percent. If the official rate were collected on all imports, the collections for the ith tariff item would be: Collected ratei = a + * (official rate') + ei, with a = 0 and 0 = 1. We start with this simple model in the first line of Table 6. For each country, the slope (the response of collected rate to a change in the official rate) is both substantively and statistically much less than one9. An increase of ten percentage points of the official rates produces an increase of only 3.3 percentage points in the collected rate for Pakistan, 4.9 for Kenya and 4.7 for Jamaica. While this may seem obvious and important, many simulations of revenue loss from tariff reforms have relied essentially on the simple model above with c = 0, , = 1. The subsequent rows of Table 6 verify that this result is robust to econometric variations. 9 Given the small standard errors of the coefficient estimates, the t-tests reject the null hypothesis that , = 1 at any reasonable significance level. - 12 - Since at the 5,000 item level many import categories are quite small in value terms"0, we weight the regression by import value to be sure the small coefficient estimates are not being driven by the odd behavior in the data for the smaller import categories. The weighted OLS results in row 2 show coefficients that are in every case lower than the unweighted results. In row 3 we report another weighted regression, this time with the statutory rate as the weighte1. The coefficients are nearly the same, except for Pakistan, which is much lower. Table 6: Regression of Collected Rate on the Official Rat Country: Pakistan Kenya Jamaica Functional form of Coeff. R2 Coeff. R2 Coeff. R regression: (Std. (Std. (Std. .! ____________ Err) Err) Err) Linear 0.33 0.16 0.49 0.21 0.47 0.30 (.01) (.01) (.01) ___ Weighted by import 0.29 0.14 0.10 0.01 0.28 0.16 values (.01) (.02) (.01) l Weighted by official rate 0.12 0.02 0.47 0.02 0.50 0.20 (.02) (.01) (.02) Linear, excluding 0.38 0.20 0.31 0.12 0.50 0.31 observations where (.01) (.01) (.01) CR=0 1 As noted above, a number of the import categories have zero vs'ue. The division of the tariff code leads to code items of intrinsically very different sizes. The ratio of the value of imports in the item of the 75d pe.centile to those in the 25' percentile is 27 (Jamaica), 75 (Pakistan) and over 1,000 (Kenya). 11 After all, line 2 of table 4 is a Bruesch-Pagan test for heteroskedasticity iinear in the statutory rate. The problem of inconsistent standard error estimates is less serious. The White heteroskedasticity consistent standard estimates produce all the same results, because -the standard errors are so small relative to the coefficients in any case. - 13 - The fourth row excludes those items for which the collection rate is zero even though recorded import values are positive. These may be tariff items that have a positive official rate reported but have been uniformly excluded from tariff collections for some reason. These items would perhaps be more accurately counted as items with a zero rate and this might create a downward bias. This procedure of excluding all items with a zero rate will, however, also exclude items for which a tariff actually was in force but all imports received an individual exemption. This will produce an upward bias in the slope. The fourth row shows that the coefficients are increased slightly, but not dramatically, for Palistan and Jamaica and are lower for Kenya. The regressions in log-log form produce no better overall fit and in percentage form the conclusions are roughly the same. O) Might rates be over the Laffer Curve? A reduction in tariff levels could actually increase revenue if increases in the tariff rate above a certain level reduce the collection rate12. This is one way of producing a 'Laffer Effect."3 Our results provide some evidence that substantial parts of the tariff code for Pakistan are on the declining revenue portion of the Laffer curve: in general, collection rates fall 12 This is a sufficient, but by no means necessary, condition. Since tariff revenue (CM) is equal to the ad valorem collected rate times the import value (MV), TR=CR*MV then the total differential is dTR/dt = CR*dMV/dt + MV*dCR/dt. Usual analysis focuses on the dMV/dt term, which depends on the elasticity of tariff value w.r.t tariff changes. 13 The Laffer curve shows the relationship of tax revenues to the tax rate. For example, for an income tax by assumption zero revenue is collected at 0 percent tax rate and, plausibly, zero revenue is collected at 100 percent rate, which implies that tariff revenues are at a maximum for some rate between zero and hundred percent taxation. - 14 - as the official rate rises. The first column for each country in Table 7 allows the introduction of a spline'4, which allows the slope of the relationship to be different above and below a given level of the tariff. Algebraically if t is the turning point: CRk = af + #*(ORi) + , l*(ORi - t) + ei, where CR is the collected rate and OR is the official rate. The slope of the relationship between collected rates and official rates below the rate t' is 0, while above the rate t' the slope is + j3 1. For all three countries we find strong evidence that the slope falls as the official rate rise. For Palistan below a tariff rate of 80% the estimated slope is .6 while above that point the slope is actually negative, -.02'5. In Jamaica the slope falls from .43 to .11 above 40%. For Kenya below the kink point tariff of 60 the slope is .58 while above the slope is only .25. 14 A spline regression allows the regression to have a different slope above and below a certain point. The function is continuous but without continuous first derivatives as the slope jumps at the ldnk point. 15 As shown by the standard error, in each of the cases the differences of slope are statistically significant, but the smaller slope may or may not be statistically different from zero. - 15 - Table 7: Non-linear Terms in the Collected Rate - Official Rate Regression Pakistan Kenya Jamaica Spline Quadratic Spline Quadratic Spline Quadratic Official .60 .60 .58 .51 .43 .39 rate (.02) (.02) (.025) (.05) (.02) (.06) OR above -.02 .25 .11 t* (.03) _ (.04) (.05) Quadratic -.0015 -.00028 .000 term (.0001) (.00043) (.000) t* or 80 200 60 1000 40 N/A turning point 2 .2 R' .22 .20 .21 .21 .29 .29 The second column for each country shows the result of the quadratic regressions, which allow the slope to change over time. Again there is some evidence of non-linearity16. The rate at which the collection rate rises as the tariff rate increases falls, and ultimately becomes negative for Pakistan and Kenya17. The estimated turning point for Pakistan is 220. The data for Kenya are unusual. In Table 2 we see that the collected rate actually falls from 43 percent for items with an official rate of 60 percent to 31 for 80 percent items and 36 for 100 percent items. This would suggest a sharp downward slope. However, there are also 22 (of 3,400) items with a tariff of 135 percent for which the collection rate is 117 percent. These observations weaken the statistical findings of non-linearity. 16 Another indication of non-linearity is that the Durbin-Watson statistic is quite low, when the data are ordered prior to estimation by the official rate. 17 If the model is CR = a + fl *OR + j3 l*ORA2, then the slope is dCR/DOR = + 2,3 1*OR. The turning point of the regression is - , /2, (1. - 16 - The data suggest that the relationship is non-linear, such that the increase in the collection rate for a 10 percentage point increase in official rates is much smaller for higher rates than lower rates. We did not find strong evidence of Laffer effects in collected rates alone (a very weakly sufficient condition) but non-linearity adds an additional channel for Laffer effects to appear. In any case, this non-linearity of collected rates and official rates has important implications for the revenue consequences of "concertina" type tariff reforms that push rates down from the top (and up from the bottom). VI) Are these countries secial? The three countries for which analysis is presented are simply those for which we were able to acquire data on tariffs, import values and revenue at the detailed level necessary. We suspect that these countries are typical and that the wide divergence between individual items' official rates and the collection of revenue is a common feature of pre-reform tariffs. We present two pieces of evidence for this view: comparisons of aggregate revenues versus hypothetical revenue across countries, and studies of exemptions in a number of countries. The import weighted average tariff rate gives the ratio of hypothetical revenue to imports, that is the ratio of total tariff revenue to import values that would result if import duty were fully collected at the single official rate on all imports. It has been widely noted that the import weighted average and the aggregate ratio of revenue to collections diverge widely. In earlier work, we (Pritchett and Sethi, 1992) find that for the eight countries for which we can calculate - 17 - both the trade weighted average and the ratio of import duty to imports'8, the excess of hypothetical revenue over actual varies from quite small (55% vs 51 % for India) to immense (32% vs 7% for Brazil). Nogues and Gulati (1992) give the ratio of the weighted average tariff rate to collection ratios in three Latin American countries (in a pre-reform year) as Argentina (1988) 39 percent vs. 17 percent, Costa Rica (1988) 16 percent vs. 7 percent, Peru (1989) 45 percent vs. 16 percent19. Kostecld and Tymowski (198 ) review import weighted average tariff and the collected rate (they call the 'ad valorem incidence") and find substantial divergence between the two, with coliected rates generally half or less the trade weighted average. The magnitude was quite different across countries: in 1977, the trade weighted average ir Venezuela was around 27 percent while collections were 4.7 percent whereas the trade weighted average in Colombia was around 22 percent, but collections were 12.3 percent. Erzan, et al, (1989), in a review of the structure of protection for a large number of LDCs as of 1985, give the average import-weighted level of tariffs20 at 30 percent, ranging from 66 percent in Central America to 5 percent in West Asia. A comparison with figures from the IMF's Government Finance Statistics2' for 18 In this case the import duty information is coming from the IMF's GFS, as we don't have actual collections data. l9 This combines information from Nogues and Gulati in table 3.10 and tables 1.1, 1.3 and 1.5. 20 This includes what they refer to as "para-tariffs," non-import duty charges levied on imports such as; customs charges, stamp taxes, taxes on foreign exchange transactions, etc., but excluding sales or excise taxes. 21 This compares the ratio of GFS line 6.1 Revenue from Import Duties to IMF IFS line 71, Imports CIF in local currency. - 18 - similar years finds average collection rates generally between 10 percent and 20 percent (except for a few high collection countries like India). Direct evidence on the amount of tariff revenue lost to exemptions is difficult to come by. Exemptions, including preferential lower tariff rates which are partial exemptions, are granted for a wide variety of reasons: country of origin (e.g. customs unions, or the GSP), end- use (e.g. imports for exports), type of importer (e.g. charities, parastatals), type of financing (e.g. donor financed imports, foreign investment), or simply discretionary exemptions granted for worthy causes. Data show that in 1988, Brazilian parastatals paid almost no tariffs (.5% of import value) while private firms pay a substantial tariff (13.4%), although this is still a small fraction of the import weighted tariff (39.8%). Foroutan (1990) examines import exemptions in Argentina and finds a large amount of revenue lost to exemptions of various types. A study in 1988 in Tanzania found that one third of all exemptions were made under the discretionary authority granted to the Minister of Finance. We have data from Cote d'lvoire on the tariff rate, import values and the value of imports fully exempt from tariffs, although no information on revenues. The proportion of import value exempt increases significantly with the level of the official tariff (see Table 8). For India we have data on import tariff levels and collected rates, but only aggregated to the two digit level. Here again we firA that the collected rate increases at a much less than one for one level with increases in the official tariff as well'. 22 There must be some bias as a result of aggregation, but we haven't figured out which way it goes. - 19 - Table 8: Regressions for Countries without Detailed Revenue Data on the Official Rate [Country: Cote D'Ivoire India l Dependent Variable Percent of fully exempt Average ratio of collections imports in total import value to imports at two digit Official Rate .26 .24 (standard error) (.042) | (.08) R_2 .016 .09 Number observations 2507 92 The available evidence suggests that exemptions are a prominent feature of pre-reform tariff systems in many, if not most, countries. Implications and Conclusions As usual, correctly interpreting the implications of our four facts is more difficult than documenting the facts themselves. These results have implications in two areas of tariff analysis, revenue and protection. The distinction between changes in tariff rates ceteris paribus (that is, holding all other aspects of the tariff regime constant) and a change in tariff rates simultaneous with other changes (e.g. changes in procedures for receiving exemptions) is important. An easy conclusion is that a calculation of the revenue implications of tariff rate change that calculates hypothetical revenue under the existing tariff and hypothetical revenue under the new tariff rates is probably worse than useless3. If the change in rates is in fact ceteris paribus then this calculation will vastly 2 This kind of calculation was quite common and the SINTIA software distributed by the Bank facilitates this calculation, although more correct calculations are possible by downgrading imports by exemption status. - 20 - overstate the revenue lost. For instance, in Table 2 we see that for Pakistan the revenue foregone from a concertina reform that lowered al! -ates down to 80 would in fact be minimal as the collection rates essentially level off at 50% anyway. The "hypothetical revenue lost" under the standard assumption would be the total import value of those items above 80% times the difference between the previous tariff and 80%, which is a substantial number. In Kenya, given that collected rates are lower for high tariff than moderate tariff items, moving those items to lower rates would actually increase revenue24 rather than the massive loss a "hypothetical revenue" calculation would suggest. Of course tariff reform generally includes broader treatment of the tariff regime than just a change in tariff rates. If that is so, the relationship between collection rates and official rates might change if, for example, exemptions were curtailed. The implications of our results !or more general revenue calculations are twofold. First, rates are not the critical determinant of revenues. The revenue implications of large changes in rates can be offset by modest changes elsewhere in the system of exemptions, and areas such as smuggling or underinvoicing and false customs declarations have not even been discussed. Second, the pressures that lead collected rates not to increase one-for-one with tariff rates will continue to be present in any tariff regime. Many exemptions are actually "tax expenditures" and a collection of the tax and allocated expenditures elsewhere would be a wash in the budget. For instance, an exemption from import duties for exporters will be roughly equivalent to a duty drawback that collects the revenue and then pays it out again to exporters. A tariff exemption to government agencies essentially 24 If those new items then had the same collection performance as other items at the same rate. This may not be true if a high rate and exemptions were acting as a discretionary quota (see the discussion below on the protection implications). - 21 - allocates income to them from the foregone import duties and a collection of the duty with offsetting increases in their budget would produce the same effect. In these cases the benefit of eliminating exemptions is primarily transparency, as the costs of programs that provide import exemptions for say, regional promotion, are often hidden in the customs statistics. The last area is the implication of the wide divergence between collected rates and official rates for analysis of protection provided by the tariff. An initial inclination is to use the average ratio of collections to import value as the "effective" tariff for an item. Howvever whether or not this is correct depends on why revenues are less than the official tariff. Three examples will hopefully clarify the point. A) Say the tariff is 50 percent but parastatals are exempt and are responsible for 50 percent of the imports. The ratio of import duty to import value would be 25 percent but the marginal price an importer would have to pay (if the parastatals are forbidden to resell) would be the full tariff of 50 percent and the fact that any imports came in at 50 percent at all would indicate that the difference between the domestic price and the border price is at least 50 percent. This is just another way of saying the obvious: that it is the marginal, not the average, price that matters for incentive calculations. B) Say the tariff rate is 50 percent but a preferential arrangement stipulates that imports from country X receive a rate of 25 percent. If country X can fully satisfy the demand for the item then the marginal and average tariff is the 25 percent preferential rate and using the ratio of import duty to value would be correct. C) Say the tariff was 50 percent but a complete exemption was granted to some group (say charitable organizations) but that resale was not effectivelv prohibited. Then the - 22 - marginal and average rate would be 0. However, if some group (say foreign firms) were forbidden to take advantage of the secondary market for exempt imports and forced to pay the full rate and this disadvantaged group accounted for 20 percent of the import value, then the official rate would be 50 percent, the average collected rate would be 10 percent while the marginal rate would be 0. This adds another reason (to an already long list) why analysis of nominal tariff rates is unlikely to provide much information on protection. - 23 - Bibliography Rajaram, Anand, 1991, "Tariff and Tax Reform: Do World Bank Recommendations Adequately Integrate Revenue and Protection Objectives," mimeo. Erzan, Refik, Hiroaki Kuwahara, Serafinao Marchese and Rene Vossenaar, 1989, "The Profile of Protection in Developing Countries," UNCTAD Rpvigg, vol 1, #1, pp 29-49. Foroutan, Faezeh, 1990, "The Distortionary Effect of Exemptions in Argentina," World Bank PRE Working Paper #388. Kostecki, M. M. and M. J. Tymowski, 1985, "Customs Duties versus Other Import Charges in the Developing Countries," Joumaal of World Trade Law, Vol. 19 #13, pp. 269-286. Mitra, Pradeep, 1992, "The Coordinated Reform of Tariffs and Indirect Taxes," World Bank Research Observer, vol 7 (2), pp. 195-218. Nogues, Julio and Sunil Gulati, "Economic Polic. ts and Performance under Alternative Trade Regimes: Latin America during the 1980s," LAC Technical department. Regional Studies Program, #16. Pritchett, Lant and Geeta Sethi, 1992, "The Structure and Reform of Tariffs," mimeo. Policy Research Working Paper Series Contact Ttls Au.hor Date for paper WPS1115 Looking at the Facts: What We Know Ross Levine March 1993 D. Evans about Policy and Growth from Cross- Sara Zervos 38526 Country Analysis WPSI 116 Implications of Agricultural Trade Antonio Salazar Brandao March 1993 D. Gustafson Liberalization for the Developing Will Martin 33714 Countries WPS1 117 Portfolio Investment Flows to Sudarshan Gooptu March 1993 R. Vo Emerging Markets 31047 WPS1 118 Trends in Retirement Systems and Olivia S. Mitchell March 1993 ESP Lessons for Reform 33680 WPS1119 The North American Free Trade Raed Safadi March 1993 J. Jacobson Agreement: Its Effect on South Asia Alexander Yeats 33710 WPS1 120 Policies for Coping with Price Donald F. Larson March 1993 D. Gustafson Uncertainty for Mexican Maize 33714 WPS1 121 Measuring Capital Flight: A Case Harald Eggerstedt March 1993 H. Abbey Study of Mexico Rebecca Brideau Hall 80512 Sweder van Wijnbergen WPS1 122 Fiscal Decentralization in Transitional Richard Bird March 1993 B. Pacheco Economies: Toward a Systemic Christine Wallich 37033 Analysis WPS1 123 Social Development is Economic Nancy Birdsall April 1993 S. Rothschild Development 37460 WPS1 124 A New Database on Human Capital Vikram Nehru April 1993 M. Coleridge- Stock: Sources, Methodology, and Eric Swanson Taylor Results Ashutosh Dubey 33704 WPS1 125 Industrial Development and the Adriaan Ten Kate April 1993 C. Jones Environment in Mexico 37699 WPSI 126 The Costs and Benefits of Slovenian Milan Cvikl April 1993 S. Moussa Independence Evan Kraft 39019 Milan Vodopivec WPS1 127 How International Economic I inks Vikram Nehru April 1993 M. Coleridge- Affect East Asia Taylor 33704 WPS1128 The International Ocean Transport Hans JOrgen Peters April 1993 J. Lucas- Industry in Crisis: Assessing the Walker Reasons and Outlook 31078 Pollcy Research Working Paper Series Contact Title Author Date for paper WPS1 129 How Pollcy Changes Affected Cocoa Jonathan R. Coleman April 1993 G. llogon Sectors in Sub-Saharan African Takamasa Akiyama 33732 Countries Panos N. Varangis WPSI 130 Poverty and Policy Michael Lipton April 1993 P. Cook Martin Ravallion 33902 WPS1 131 Prices and Protocols in Public Jeffrey S. Hammer April 1993 J. S. Yang Health Care 81418 WPS1 132 An Analysis of Repressed Inflation Andrew Feltenstein April 1993 E. Zamora in Three Transitional Economies Jiming Ha 33706 WPS1 133 Macroeconomic Framework for an Ibrahim Elbadawi April 1993 A. Maranon Oil-Based Economy: The Case of Nader Majd 31450 Bahrain WPS1 134 Managing a Nonrenewable Resource: Ibrahim A. Elbadawi April 1993 A. Maranon Savings and Exchange-Rate Policies Nader Maid 31450 in Bahrain WPS1 135 Inflation in Czechoslovakia, 1985-91 Zdenek Drabek May 1993 E. Zamora Kamil Janacek 33706 Zdenek Tuma WPS1 136 The Dynamic Behavior of Quota Kala Krishna May 1993 D. Gustafson License Prices: Theory and Evidence Ling Hui Tan 33714 from the Hong Kong Apparel Quotas WPS1 137 Railway Reform in the oentral and Philip W. Blackshaw May 1993 TW1UTD Eastern European Economies Louis S. Thompson 31005 WPS1 138 The Economic Impact of Military Daniel Landau May 1993 C. Jones Expenditures 37699 WPS1 139 Should Sub-Saharan Africa Expand Jonathan R. Coleman May 1993 G. liogon Conon Exports? M. Elton Thigpen 33732 WPS1 140 How Retail Food Markets Responded Bruce Gardner May 1993 C. Spooner to Price Liberalization in Russia Karen M. Brooks 32116 after January 1992 WPS1 141 Foreign Direct Investment in a Maxwell J. Fry May 1993 R. Vo Macroeconomic Framework: Finance, 31047 Efficiency, Incentives, and Distortions WPS1 142 Rent-Seeking Trade Policy: A Time- Martin Rama May 1993 D. Ballantyne Series Approach 37947 WPS1 143 Tariff Rates, Tariff Revenue, and Lant Pritchett May 1993 M. Fernandez Tariff Reform: Some New Facts Geeta Sethi 33766