The distribution sector plays a crucial and large role in most economies, (1) In the G-7 (United States, Japan, Germany, France, United Kingdom, Italy, and Canada), distribution's share of gross domestic product (GDP) ranges from 8% to 15%, and its share in employment ranges from 11% to 19% (Pilat 1997, table 2.1). Because almost all goods in an economy pass through this sector, distribution can heavily influence prices. For consumer goods, in particular, these price effects can be large. In Organisation for Economic Cooperation and Development (OECD) countries, consumer prices for final goods exceed the corresponding producer prices by 40% and more. In addition to--indeed, because of--its economic importance, distribution producers enjoy substantial political clout and have won many favorable regulations from the political process: see Kalirajan (2000) and Pilat (1997). These regulations may impose large welfare costs and may exert a large influence on trade flows. Nonetheless, relatively few publications rigorously analyze the potential overall welfare and trade effects of deregulating distribution. This study seeks to fill some of that void.
I first provide a brief overview of distribution regulations. Then I present new, internationally comparable distribution margins data for a sample of eight OECD countries. These data are used to infer the extent to which distribution regulations reduce efficiency. For four of these countries, I conduct an applied general equilibrium (AGE) analysis of the welfare effects of high margins and find that they impose large costs, rivaling those of protection. It appears that efforts to increase economic welfare should not ignore the potentially large gains from liberalizing distribution. This article also examines the relationship between restricted distribution and imports.
II. DISTRIBUTION REGULATIONS
Several studies have described a variety of distribution regulations in OECD countries. Nicoletti (2001), Boylaud (2000), Kalirajan (2000), and Pilat (1997) provide excellent overviews. These studies imply that, despite some liberalization, many impediments remain. The most burdensome regulations include: restrictions on large stores; onerous business set-up rules; restrictive zoning laws and other real estate regulations; limits on product ranges, store hours, and pricing; and monopoly distributors for certain products, such as pharmaceuticals, liquor, and tobacco. In addition, many local regulations work under the radar of studies such as these.
The limited literature on the subject implies that distribution regulations reduce welfare. Pellegrini (2000) estimates that deregulating retail trade would increase Italian GDP by more than 1%. Japan's Economic Planning Agency (1996) finds that loosening restrictions on large retailers has increased Japanese GDP by 1%. Pilat (1997) cites a European Commission study that finds that liberalizing store hours has greatly increased consumer welfare. Carree and Nijkamps (2001) find efficiency gains after entry barrier removal in the Netherlands. Nicoletti (2001) concludes: "Both simulation and econometric studies point unequivocally to potentially large welfare gains from the liberalization of ... retail trade." Also, Burda (2000) and studies cited in Nicoletti (2001) find that employment would increase, not decrease, with liberalization.
These studies provide valuable information, but none simulate the effect of a country's total package of distribution regulations on welfare. I do so by exploiting AGE modeling, which takes account of links among industries (and nations) to produce broad-brushed estimates of inflated margins' effects in multiple countries. Distribution's widespread influence on an economy makes an AGE approach especially important. (2)
III. MEASURING MARGINS
The Underlying Data
The distribution margins data grow out of protection data developed in Bradford (1998, 2003). He combines data on retail prices and distribution margins to derive producer price gaps and estimates of protection for 124 final demand categories: 103 household consumption goods and 21 capital goods. The data cover eight countries--Australia, Belgium, Canada, Germany, Japan, the Netherlands, the United Kingdom, and the United States--and include the years 1985, 1990, and 1993.
The underlying distribution data come from national input-output (IO) tables and are from 1990, except the United States (1992) and Japan (1995). Each IO table reports for detailed sectors the value of output in producer prices and in consumer prices. The ad valorem margin in each sector is the ratio of the latter to the former. These margins include wholesale trade, retail trade, and transportation. (3) The IO data have been concorded into the 124 categories used in this study. Table 1 has a list of categories.
New Estimates of Distribution Margins: Specific versus Ad Valorem
For ease of presentation, I have aggregated the ad valorem margins up to 28 categories and report these data in Table 2. Belgium, the Netherlands, and the United Kingdom have the lowest margins, and Japan and the United States have the highest. Other studies, such as Ito and Maruyama (1991) and Nishimura et al. (2000) have compared ad valorem distribution margins and have also concluded that Japan's margins roughly equal those of the United States. Ad valorem distribution margins, however, paint a misleading picture of distribution costs because these margins are percentages of the underlying producer prices. If other regulations, such as trade barriers, artificially inflate a country's producer prices, this will bias the ad valorem margins downward. To compare the actual costs of distribution, one needs the specific or absolute margins, that is, how much it costs to move each unit of a good from the factory to the store shelf, instead of the cost per dollar of output.
I derive specific margins by using the protection data mentioned. These data allow one to infer to what extent producer prices in these sectors and countries are inflated, which in turn enables one to adjust the margins so that they reflect costs per unit. This article thus calculates for the first time, to my knowledge, internationally comparable specific, as opposed to ad valorem, distribution margins for these countries.
The calculation goes as follows. In the data, protection is measured as the ratio of the producer price to the world price, and this ratio indicates how much the producer price is inflated. I adjust margins for producer price inflation by scaling up the ad valorem margin by the amount of the producer/world price ratio, for each product in each country. Such scaling gives the percentage margin that would apply if the producer price were equal to the world price and thus gives a set of margins data that is measured in terms of the same world price for each product in all of the countries.
Mathematically, because the ad valorem margin is the ratio of the consumer price to the producer price, whereas the specific margin is the difference, we have [m.sup.s.sub.ij] = [p.sup.c.sub.ij] - [p.sup.p.sub.ij] = [p.sup.p.sub.ij](1 + [m.sup.a.sub.ij]) - [p.sup.p.sub.ij] = [p.sup.p.sub.ij][m.sup.a.sub.ij] where [m.sup.s.sub.ij] is the specific margin for good i in country j, [p.sup.c.sub.ij] is the consumer price for good i in country j (relative to the world price), [p.sup.p.sub.ij] is the producer price for good i in country j (relative to the world price), and [m.sup.a.sub.ij] is the ad valorem margin for good i in country j. Thus, the specific margin is simply the product of the ad valorem margin and the ratio of the domestic producer price to the world price (which is the amount of protection). This specific margin tells how many cents it costs to move $1 worth of the good, valued at world prices, from the factory to the store shelf. So, a 25% ad valorem margin applied to a producer price that is twice the world price implies that 50 cents is spent on distribution for each dollar (in world prices) of output. On the other hand, the same 25% ad valorem margin applied to a producer price that equals the world price will show that 25 cents is spent on distribution for each dollar. Even though the ad valorem margins are the same in each case, the amount of resources devoted to distribution is twice as high in the former case.
Table 3 shows the specific margins for the sample. The five nations with the lowest margins are all bunched tightly in the 57-61 range. Australia's are somewhat higher. Germany and Japan have the highest margins, with Japan's exceeding all others by a wide margin. Most observers agree that Japan has a heavily regulated distribution sector, and these numbers support that view. These data also provide evidence for the claim that Japan's margins exceed those of the United States by a significant amount. Thus, correcting ad valorem margins for underlying producer prices tells a substantially different story concerning the costs of distribution in these countries.
These specific margins also provide summary indicators of the extent to which various regulations in these countries restrict the distribution sector. If distribution were perfectly tradable internationally, these numbers would be perfect indicators of distribution inefficiency: Any differences would reflect cost-raising regulations. In fact, distribution is internationally tradable to a greater degree than many believe, as shown by the worldwide march of such major retailers as Costco and Wal-mart. Also, the numbers in Table 3 indicate a large amount of convergence, even with various regulations. Nevertheless, removing all regulations probably would not cause the specific margins in these nations to become identical, because two factors work against such complete convergence: unavoidable differences in input costs and differences in the quality of distribution services provided. I briefly discuss each in turn.
Even if markets were completely free, land prices in Europe and Japan would probably exceed land prices in Australia, Canada, and the United States, causing the free market margins to differ. It turns out, though, that land only accounts for 5% of the distribution input costs in these countries, so that even large land cost differences would not drive large wedges between free market-specific margins. One may also wonder about differences in transportation costs across countries, but these costs only account for 3-5% of distribution's inputs. The most important distribution input is labor, which makes up about 50% of input costs. Unfortunately, internationally comparable data on hourly compensation in distribution do not seem to exist. Because these OECD countries all have workers with similar educational opportunities, free market wages probably do not differ much. To be sure, the U.S. distribution sector has many part-time workers with little or no benefits. So hourly costs for European workers may significantly exceed costs for U.S. workers. Labor in U.S. distribution, however, is the most productive in the world, so the hourly costs should be higher for the United States, not lower. If European labor costs are higher, this must reflect removable policies, not inevitable differentials. Thus, although input costs probably cannot be equated, any differences probably do not create large nonremovable gaps in the specific margins.
If quality differs across these countries, they could have equal input costs and be equally efficient at supplying distribution and yet have differing specific margins. For instance, shoppers in land-scarce countries such as Japan and the Netherlands may desire many different shops nearby, so they can easily make frequent trips and store less at home. On the other hand, shoppers in land-abundant nations such as Canada and the United States probably prefer stores with larger quantities and selection so that they can shop less frequently. The free market in these different countries may lead to different mixes of proximity and product selection and thus different specific margins. This introduces uncertainty into the connection between specific margins and restrictiveness. Such quality differences, though, may not bias the specific margins, because different quality mixes in different countries imply trade-offs among sets of desired attributes. Also, the internationalization of distribution implies that quality differences would not be very large without regulations. Thus, even though quality differences mean that specific margins measure restrictiveness with noise, this likely does not undermine the effort. (4)
For comparative purposes, Table 3 also shows two alternative measures of distribution restrictions for these countries: one from Kalirajan (2000) and the other from Nicoletti (2001), which draws on the OECD regulation database. Both studies use somewhat subjective methodologies, in which various types of regulations are converted into a single index using ad hoc weights, which cannot vary by country. Also, neither of these studies takes account of local regulations, which may be just as important as national ones. Nevertheless, these studies provide interesting information. Both put Belgium and Japan among the most restrictive and Australia and the United States among the least restrictive. Unlike my data, neither of them finds that Germany's distribution is unusually restrictive. Looking at these studies combined with this one, it appears that Japan's distribution probably has the most restrictions, the United States probably has the least, and the Canadian and Dutch regimes are relatively free.
IV. WELFARE FRAMEWORK
To provide perspective on the import of these distribution data, I simulate their welfare effects using the AGE framework of Harrison, Rutherford, and Tarr (HRT). (5) This section presents an overview of the model; section V describes modifications made for this article. See HRT (1997) for more details. This model's strengths are that it has more country and sectoral detail than most multicountry AGE models and that it allows for increasing returns to scale (IRS) and dynamic adjustment of the capital stock.
The model has 24 regions and 22 sectors, as shown in Table 4. The European Union nations are not broken out separately, meaning that the welfare analysis in this article will only cover Australia, Canada, Japan, and the United States. The underlying data come from the Global Trade Analysis Project (GTAP) database, version 2 (1992). Production uses intermediate goods and three factors--capital, labor, and land. Labor and land cannot move across national boundaries, but all factors can move freely across sectors. Value added in each sector has a Constant Elasticity of Substitution (CES) production function. I use HRT's values for these elasticities, which they estimated using U.S. data from 1947 to 1982 and using the same CES functional form as this AGE model has. The production function that combines intermediates and the value-added composite is Leontief. The results are robust to this assumption.
Some sectors have constant returns to scale. Other sectors are modeled with IRS and imperfect competition. (6) These sectors are assumed to have firm-level product differentiation, with output being a composite of varieties. Firms have fixed costs and constant marginal costs, meaning that fewer firms leads to rationalization gains. These firms compete using quantity conjectures, with entry and exit that drive profits to zero.
Dynamics are incorporated by allowing the capital stock to expand in response to temporary increases in the rate of return caused by liberalization. (7) Investment augments the capital stock until its return is driven back down to the long-run level, which is exogenously fixed. The results to be reported therefore reflect the model's predictions for what happens after the price of capital returns to its long-run value. The capital adjustment process is not modeled, (8) and the time horizon implied by these results depends on how long one thinks it takes capital to respond to interest rate differentials. The model ignores the consumption forgone by the increased investment, which overstates welfare gains. On the other hand, the model ignores any impact of productivity growth and innovation, which understates the gains.
On the demand side, each region has a representative consumer and a single government agent, each of whom has a nested CES utility function and practices multistage budgeting. At the top level, demand across the 22 sectors is Cobb-Douglas. At the second level, each of these 22 goods is a combination of domestic output and an import composite, with CES preferences across these two components. At the third level, the model uses the Armington assumption: Imports of the same good from different countries are imperfect substitutes. (9) Preferences across these different goods from different countries are also given by a CES utility function. Following HRT, I set the elasticity of substitution between the import composite and the domestic good, [[sigma].sub.DM], equal to 4 and the elasticity across import varieties, [[sigma].sub.MM], equal to 8. These elasticities do affect the magnitude of the results: in general, higher values of the parameters lead to higher welfare gains from liberalization. Even wide changes, though, do not change the signs of any of the main results or any qualitative conclusions.
The IRS sectors have yet another level of demand. The domestic good and each import good produced in each region, instead of being homogenous goods (albeit imperfect substitutes for each other), are themselves composites of different varieties produced by different firms. The elasticity of substitution across these varieties is set at 15. All results reported are robust to wide changes in this parameter.
V. MODIFYING THE MODEL TO ACCOUNT FOR DISTRIBUTION
The 1993 margins data are used to model distribution more accurately within the AGE framework. The HRT model, as with most AGE trade models, does not account for margins explicitly. All distribution services are lumped into the trade and transport sector and consumed as a separate good instead of being linked to the goods that actually use those services. This obscures the role of distribution in the economy and can skew the results of AGE analyses. For instance, simulations of price reductions in other sectors may imply a large substitution out of trade and transport services, even though actual consumption of these will probably increase to facilitate commodity flows. Also, not accounting for margins implies that consumers base choices on producer prices instead of the higher consumer prices that include margins.
This article attempts to address these problems by incorporating distribution explicitly into each final demand sector that has margins data. This is done by treating margins like taxes, because margins create wedges between consumer and producer prices. Thus, margin wedges were inserted into each of the sectors that has margins data. (10) Also, the value of the trade and transport sector was reduced by the total value of these margins. Finally, inputs into the trade and transport sector were reduced and redistributed across the final goods sectors in accordance the amount of distribution used in those sectors. (11)
VI. ESTIMATING THE WELFARE EFFECTS OF INFLATED MARGINS
The Margins Experiments: Convergence to the Minimum
I estimate the effects of bloated distribution by simulating the reduction of excessive margins to efficient levels. (12) In light of the discussion in section III and to keep the simulations as clean as possible, the minimum of the eight absolute margins is used as the estimate of the efficient margin for each product. This was done at the 124-sector level. These minimums were then aggregated to the categories of the model. Table 5 shows the measured margins and the "efficient" margins used in the model.
These simulations assume, therefore, that any margin higher than the minimum is needlessly inflated, due to protection from competitors, and that liberalizing distribution would lead to a convergence in margins, as best practices spread across countries and squeeze out inefficient technologies, just as trade opening leads to a convergence of producer prices. (13) As Nicoletti (2001) points out, distribution has been steadily internationalizing, with multinational distribution groups fostering increased competition, leading to technology diffusion and convergence. (14)
Still, one may wonder whether these are the best simulations. Is such convergence realistic? Might such experiments overstate the results? As mentioned, price differences for nontraded factors, such as labor and land, could prevent convergence, causing an overstatement of high margins' costs. Also, if some high margins stem from higher quality distribution, instead of inefficiency, then the results will be overestimates. Section III discusses why these factors probably do not significantly bias the results. In addition, two forces work in the opposite direction. First, with only eight countries in the sample, the lowest margin may not be efficient, causing the simulations to understate the gains from streamlining distribution. Second, the model does not account for productivity improvements in distribution. This has been occurring, and further opening would probably bring more progress. (15) The analysis shows the effects of converging to 1993's standard, even though 2003's may be much better. Thus, overall it seems that these estimates do not overstate things. Finally, no margin was reduced to zero. All final goods require a positive distribution margin, and reducing it to the minimum always leaves a substantial margin in place. In the end, if one were to conduct some other experiment with partial convergence, one would have to choose the levels to which each margin moves. Doing so would be speculative and would create opportunities for manipulating the results, introducing more uncertainty, not less.
As noted in section IV, the simulations cover the four non-European countries. I also simulate the reduction of margins in all four countries simultaneously, with a focus on overall welfare effects, as measured by the change in equivalent variation as a percentage of GDP. (16) I do not analyze more detailed effects, such as changes in real factor prices and changes in sectoral employment, because this model, in which margins are treated like tax wedges, does not adequately capture these effects. This is left to future work.
Rents versus Deadweight Loss
The extent to which distributors collect rents with inflated margins significantly influences the results. If distributors are efficient and enjoy high margins because of restrictions on competition, then excess margins produce rents, which are transfers, not deadweight loss. On the other hand, if distributors charge high margins because they are inefficient relative to what potential excluded competitors could do, those excess margins would reflect deadweight loss, instead of rents. Most likely, there is a combination of the two. Figure 1 illustrates, showing two extreme cases: pure rents and pure deadweight loss. Area 1 shows payments to distributors beyond what they would receive if the margins were at the efficient level. If the distributors' per unit costs are the same as the efficient margin, then Area 1 is pure rents. If the distributors' per unit costs equal the inflated margin (which implies being shielded from competition), then Area 1 reflects pure deadweight loss, the loss of inefficient firms using up scarce resources.
[FIGURE 1 OMITTED]
For the main results, it is assumed that excess margins are half rents and half deadweight loss. I also report the extreme cases of pure rents and pure deadweight loss to provide bounds on the estimates of the true welfare effects, though the central case will probably come closer to the actual effects. (17)
Most AGE researchers report point estimates and check robustness by varying influential parameters in systematic, but somewhat ad hoc ways. Abdelkhalek and Dufour (1998), however, present a procedure for developing well-grounded confidence intervals for AGE simulations. They point out that when simulation parameters are estimated, one can derive confidence intervals for endogenous variables. This involves using the standard errors for the estimated parameters to impose appropriate lower and upper bounds on those parameters and then minimizing and maximizing each endogenous variable of interest subject to those bounds. Let [[??].sub.i] be one of k estimated model parameters. Abdelkhalek and Dufour show that the lower bound of a 1 - [alpha] confidence interval for an endogenous variable can be found by minimizing that variable subject to having each of the estimated parameters bounded by a 1 - [alpha]/k confidence interval. Similarly, the upper bound of a 1 - a confidence interval for an endogenous variable is found by maximizing it subject to the same constraints on the estimated parameters. Thus, if [[??].sub.i] is normally distributed with standard error [s.sub.i], then each of the k parameters has a lower bound of [[??].sub.i] - [z.sub.[alpha]]/k[S.sub.i] (where z is a critical value from the standard normal) and an upper bound of [[??].sub.i] + [z.sub.[alpha]/k][s.sub.i]. These confidence intervals may be conservative, because this approach assumes that each parameter estimate is independent of each of the others. With knowledge of the covariance structure across the estimates, one could derive tighter intervals.
The only parameters that influence the results, besides [[sigma].sub.MM] and [[sigma].sub.DM], are the factor substitution elasticities. Because they have been estimated, their standard errors can be used to derive interval estimates for the welfare results. There are 11 factor substitution estimates, so a 95% interval estimate results from bounding these parameters with [[??].sub.i] [+ or -] [z.sub.0.05/11][s.sub.i] = [[??].sub.i] [+ or -] 2.61[s.sub.i].
These confidence intervals depend on the model and the other parameters (the ones not varied) being correct. Any confidence interval depends on the underlying model being correct, so this does not distinguish this article's confidence intervals from conventional ones. In addition, though, the confidence intervals in this article depend on the accuracy of influential fixed parameters (e.g., the domestic goods--import composite elasticity, [[sigma].sub.DM], and the elasticity of substitution across import varieties, [[sigma].sub.MM]). Varying these parameters widely (cutting them in half and doubling them) does not affect the magnitude of the welfare gains very much and does not affect whether any interval estimate lies entirely above zero. Thus, the main story told by the results reported next is robust to reasonable change in other parameters. As will be seen, the assumption concerning rents and deadweight loss has a much larger impact on the results than variations in any of the calibrated parameters.
VII. SIMULATION RESULTS
Table 6 reports the 95% interval estimates for overall welfare changes. This table reports the permanent annual effect of margins reductions on equivalent variation, as a percentage of GDP, once the capital stock has changed to its new equilibrium. In other words, this table reports the welfare costs, borne at home and abroad, of allowing inflated margins to persist in the four countries separately and as a group.
Distribution regulations appear to impose large costs. In the central case, each country would gain at least 2% of GDP from complete deregulation. Because these are permanent increases, the present discounted value is several times larger. To provide perspective, let us compare these results to trade opening simulations using the same model and data. Bradford (2003, table 7) shows that the 95% intervals for complete trade opening are Australia, [1.6, 5.7]; Canada, [2.4, 6.4]; Japan, [2.0, 2.5]; and the United States [0.0, 0.4]. For each country, the central case margin simulation results match or exceed these large trade opening gains.
Japan's results are striking. Its gains from opening distribution--around 5% of GDP--would be at least twice the substantial gains that trade opening would bring. Japan's distribution regulations appear to impose substantial drag on its economy. Some, as in Goldman (1992), might argue that these results overstate Japan's potential gains because its unique culture and economic structure would prevent margins convergence. Since the Large Scale Retail Law was relaxed in 1994, however, the number of large retailers has shot up, and mom-and-pop shops have steadily disappeared, implying that regulations, not culture, have been propping up small stores. (18)
The United States, too, appears to gain much from deregulation, if one assumes that half of the inflated margins result from inefficiency. Many observers, however, believe that the United States has the most efficient distribution system in the world. This would imply that most of the inflated margins in the United States simply transfer rents to distributors, which means that the welfare gains from reducing margins in the United States may be quite small.
Although many nations look to continued trade opening as a source of welfare gains, and rightfully so, these results imply that bringing more efficiency to the all-important distribution sector should produce just as many or more benefits. Of course, complete liberalization of distribution in these countries is not a policy option right now. These results show, however, that efforts in this direction can potentially produce large payoffs.
The results from margins reduction in all four countries simultaneously shown in Table 6 imply that the effects on individual countries differ little from unilateral deregulation. This indicates that there are few international spillovers connected with margins, at least in this model. This results partly from the fact that all sectors have zero profits in the model. So, while reduced margins do increase imports, and thus exports for other countries, these exports produce zero profits in this model and, thus, few net welfare gains.
Table 6 shows that the interval estimates in each of the three cases are quite tight. The underlying factor substitution elasticities do not influence the results a great deal. A far greater source of uncertainty stems from not knowing how much of the excess margins reflect rent transfers and how much reflect deadweight loss. Although the gains are positive no matter what is assumed, a fruitful area of further research would be to estimate these fractions. The next section provides a start.
Refining the Estimates Using Productivity Data
Pilat (1997, table 2.6) presents data on value added per worker in the distribution sector for most OECD countries. Although using total factor productivity would be better, these numbers provide some information. With the U.S. benchmarked at 100, the other three countries are Australia, 59.4; Canada, 58.4; and Japan, 60.3; implying that the United States is far more productive than the other three, which are bunched quite close together. Let us use these numbers to estimate what fraction of Area 1 in Figure 1 is deadweight loss for each sector in each country.
Because the United States appears to be the most productive, by far, I assume that inflated margins in the United States reflect pure rent transfers. Thus, for the United States, the estimate for the deadweight loss fraction in each sector is zero. This means that estimates of the gains from margin reductions in the United States should be taken from the "Pure Rents" column of Table 6: about 0.25% of GDP, a small number roughly equal to the estimated gains from trade liberalization. The productivity data imply that the other countries have costs that are about two-thirds higher than in the United States. We may infer from this that inflated margins in these countries that are within two-thirds of the U.S. level reflect inefficiency and deadweight loss. It is assumed, therefore, that when the margin exceeds the U.S. margin by more than 67%, the portion of the margin that is above that 67% threshold reflects pure rent transfers to distributors.
Referring to Table 5, Australia's margins are more than 67% higher than those of the United States for fishing-wool-wood-paper (71% higher) and for transport equipment (108% higher). For Canada, this only happens for processed rice (71%). Thus, for these two countries, the great bulk of high margins reflect deadweight loss rather than rents. Adding up over all sectors, 99.8% of Canada's margins are within 67% of the corresponding U.S. margin. For Australia, this ratio is 96.7%. (19) Japan's margins are much more likely to lie beyond the threshold. Table 5 shows that this happens in 10 out of 13 sectors. Aggregating over all sectors, 76.7% of Japan's margins are within the threshold.
So, letting [alpha] represent the fraction of excess margins that consist of deadweight loss, [[alpha].sup.A] = 96.7, [[alpha].sup.C] = 99.8, [[alpha].sup.J] = 76.7, and [[alpha].sup.US] = 0. As already discussed, the predicted welfare gains from distribution deregulation are a weighted average of the two extreme cases. These predictions are given, therefore, by (1 - [[alpha].sup.i])R + [[alpha].sup.i]DWL, where [[alpha].sup.i] is one of the fractions above, R is the gain under the pure rents assumption, and DWL is the gain under the pure deadweight loss assumption. Table 7 reports these results. They imply that liberalizing Japan's distribution would create a permanent 8% increase in GDP. Given the stagnant growth performance there since the early 1990s, such large potential payoffs from deregulation should not be ignored when considering policy actions to help rejuvenate the Japanese economy.
The Trade Impacts of Sluggish Distribution
The analysis implies that inefficient distribution imposes large domestic costs on at least three of these countries, while having little impact abroad. As already discussed, these small international effects stem from assuming zero profits. Policy makers and businessmen, though, often want to maximize exports for political reasons or because they see economic profit opportunities. Such people have expressed concern over the extent to which restricted distribution in potential export markets limits imports. Thus, it will prove useful to examine the impact that sluggish distribution has on imports. (20)
Table 8 reports percentage changes in imports by sector and country resulting from unilateral margins reductions. Only results for the goods sectors are shown, because service sector imports have significant measurement error. Also, the table does not report interval estimates, because they are extremely tight. Only the lower bounds are shown, which, rounded to one decimal point, usually equal the upper bounds.
The results for Japan, not surprisingly, stand out. The model predicts that their imports would increase by more than $18 billion, or 6.6%. The agricultural sectors would experience the largest jumps, in percentage and absolute terms. Wearing apparel and meat imports would also increase significantly. In the United States, imports would expand by $8 billion, or 1.4%. Nongrain crops, milk, and processed rice would increase the most in percentage terms, and the largest absolute increases would be in energy, machinery/ equipment/other and nongrain crops. Loosening distribution would increase Australian goods imports by 3.3% and Canadian imports by just 0.8%. In Australia, textile imports would expand the most in percentage terms, and the machinery/equipment/other, chemicals/ rubber/plastic, textiles, and transport sectors would increase the most in absolute terms. In Canada, processed rice and paddy rice would see the largest percentage surges, but these are small sectors. Energy, chemicals/rubber/ plastics, processed food, and nongrain crops would see the largest absolute increases. Imports in the machinery/equipment/other sector would actually decline. This results from substitution on the demand side: prices in other sectors decline by more than prices in this sector. (21)
These results show that distribution restrictions can have nontrivial impacts on imports. In percentage terms, such regulations have substantial effects in Australia and, especially Japan. The absolute effects in Japan are also quite large: Its inflated distribution margins restrict goods imports by about $18 billion. Most of these imports are in the food and agricultural sectors, a fact that should interest U.S. farmers.
Bandyopadhyay (1998) provides an interesting theoretical discussion of the connection between the distribution sector and international trade that complements these empirical simulations. As in this article, Bandyopadhyay stresses the importance of appropriately modeling the fact that traded goods and distribution services are consumed jointly. He incorporates a model of the distribution sector into a Ricardian trade model and shows that technological change in the distribution sector can influence trade through two important channels: changes in distribution costs and changes in consumer demand for distribution services. Allthough the modeling here stresses the former, Bandyopadhyay (1998) implies that the latter may be even more important.
The distribution sector, including wholesale and retail trade, forms an important part of any modern economy, and regulations in this sector can have significant welfare effects. Relatively little work, though, has attempted to estimate these effects, and there appears to be no other work that estimates overall welfare effects across a sample of countries. This article does so by presenting new data on distribution margins in eight OECD countries and using an AGE framework to assess the welfare impacts of inefficient distribution. The margins data make it possible to estimate the extent to which margins are inflated, due to regulations. This work has also compared margins across countries, and, in contrast to other studies, finds that Japan's margins are unusually high. The different results for Japan come from correcting the underlying producer prices for trade distortions.
The welfare estimates imply that distribution regulations impose substantial costs, especially in Japan, implying large potential payoffs from further efforts to streamline distribution in these countries. It is estimated that distribution restrictions reduce real consumption in Japan by 5-8% annually. In Canada and Australia, the losses are about 2-5%. Losses in the United States may be as high as 2.5% and as low 0.25%. A key implication of this study is that distribution regulations appear to cause just as much harm as trade barriers to the country imposing the regulations. I also find that distribution impediments can significantly reduce imports, especially in Japan.
Future work will focus on explicitly modeling the distribution sector, instead of using wedges between consumer and producer prices. One approach used by others is to model distribution services as inputs into production of the good that consumers buy, as in Bradford and Gohin (forthcoming). Such modeling would better capture the fact that distribution services are not purchased separately from the commodities that are distributed.
Other work will involve more recent data. I now have margins measures for 1996 and 1999. The latest version of the GTAP database makes it possible to perform separate margins simulations for Germany, the Netherlands, and the United Kingdom. Doing so would probably provide interesting insights. Finally, it would be instructive to see how trade liberalization and margins reductions simulations interact. Do inflated margins increase or decrease the gains from trade opening? Do trade barriers increase or decrease the gains from distribution deregulation? The modeling framework and data in this article would make it possible to provide answers to such questions.
AGE: Applied General Equilibrium
CES: Constant Elasticity of Substitution
GDP: Gross Domestic Product
GTAP: Global Trade Analysis Project
HRT: Harrison, Rutherford, and Tarr
IRS: Increasing Returns to Scale
OECD: Organisation for Economic Co-operation and Development
Abdelkhalek, T., and J.-M. Dufour. "Statistical Inference for Computable General Equilibrium Models, with Application to a Model of the Moroccan Economy." Review of Economics and Statistics, 80(4), 1998, 520-34.
Bandyopadhyay, U. "Distribution Costs and International Trade: A Ricardian Model." Review of International Economics, 6(1), 1998, 164-78.
Betancourt, R. R., and D.A. Gautschi. "The Economics of Retail Firms." Managerial and Decision Economics, 9, 1988, 133-44.
--. "Demand Complementarities, Household Production, and Retail Assortments." Marketing Science, 9(2), 1990, 146-61.
--. "The Outputs of Retail Activities: Concepts, Measurement, and Evidence from U.S. Census Data." Review of Economics and Statistics, 75(2), 1993, 294-301.
Bliss, C. "A Theory of Retail Pricing." Journal of Industrial Economics, 36(4), 1988, 375-91.
Boylaud, O. "Regulatory Reform in Road Freight and Retail Distribution." OECD Economics Department Working Paper No.255, 2000.
Bradford, S. C. A Theoretical and Empirical Analysis of Trade Protection in Industrialized Democracies. PhD dissertation, Harvard University, 1998.
--. "Paying the Price: Final Goods Protection in OECD Countries." Review of Economics and Statistics, 85(1), 2003, 24-37.
Bradford, S. C., and A. Gohin. "Modeling Distribution Services and Assessing Their Welfare Effects." Review of Development Economics, Forthcoming.
Burda, M. C. "Product Market Regulation and Labor Market Outcomes: How Can Deregulation Create Jobs?" CESifo Working Paper No. 230, 2000.
Carree, M. A., and J. Nijkamps. "Deregulation in Retailing: The Dutch Experience." Journal of Economics and Business, 53, 2001, 225-35.
Cotterill, R. W. "The Food Distribution System of the Future: Convergence Towards the US or UK Model?" Agribusiness, 13(2), 1997, 123-35.
Economic Planning Agency. "Provisional Estimates of the Economic Effects of Recent Deregulations." Tokyo, 1996.
Gohin, A. Modelisation du Complexe Agro-alimentaire Francais dans un Cadre D'equilibre General. PhD dissertation, Universite de Paris I, Pantheon-Sorbonne, 1998.
Goldman, A. "Evaluating the Performance of the Japanese Distribution System." Journal of Retailing, 68(1), 1992, 3-39.
Harrison, G. W., T. F. Rutherford, and D. Tarr. "Quantifying the Outcome of the Uruguay Round." Finance and Development, 32(4), 1995, 38-41.
--. "Quantifying the Uruguay Round," in The Uruguay Round and the Developing Countries, edited by W. Martin and L. A. Winters. New York: Cambridge University Press, 1996.
--. "Quantifying the Uruguay Round." Economic Journal, 107, 1997, 1405-30.
Ito, T., and M. Maruyama. "Is the Japanese Distribution System Really Inefficient?" in Trade with Japan: Has the Door Opened Wider?, edited by P. Krugman. Chicago: University of Chicago Press, 1991.
Itoh, M. "Competition in the Japanese Distribution Market and Market Access from Abroad," in Deregulation and Interdependence in the Asia-Pacific Region, edited by T. Ito and A. Krueger. Chicago: University of Chicago Press, 2000.
Kacker, M. "International Flow of Retailing Know-How: Bridging the Technology Gap in Distribution." Journal of Retailing, 61(1), 1988, 41 67.
Kalirajan, K. "Restrictions on Trade in Distribution Services." Productivity Commission Staff Research Paper. Canberra, 2000.
Knetter, M. M. "Why Are Retail Prices in Japan So High: Evidence from German Export Prices." International Journal of Industrial Organization, 15, 1997, 549-72.
Komen, M. H. C., and J. H. M. Peerlings. "WAGEM: An Applied General Equilibrium Model for Agricultural and Environmental Policy Analysis." Wageningen Economic Papers 4, 1996, Netherlands.
Nicoletti, G. "Regulation in Services: OECD Patterns and Economic Implications." OECD Economics Department Working Paper No. 287, 2001.
Nishimura, K. G., and L. F. Punzo. "The Distribution Structure in Three Continents: An Evolutionary Analysis of Italy, Japan, and the United States." Working Paper, 1998.
Nishimura, K. G., H. Tsubouchi, and T. Tachibana. "The Evolution of the Japanese Distribution Structure: An International and Institutional Perspective." Working Paper, 2000.
Noll, R. G. "Regulatory Reform and International Trade Policy," in Deregulation and Interdependence in the Asia-Pacific Region, edited by T. Ito and A. Krueger. Chicago: University of Chicago Press, 2000.
Pellegrini, L. "Regulations and the Retail Trade," in RegulatoryReform and Competitiveness in Europe, 2, Vertical Issues, edited by G. Galli and J. Pelkmans. Northampton, MA: Edward Elgar, 2000.
Pilat, D. "Regulation and Performance in the Distribution Sector." OECD Economics Department Working Paper No.180, 1997.
Riethmuller, P. "Reform of the Japanese Food Distribution System: Implications for Consumers." Journal of Consumer Policy, 19, 1996, 69-82.
Wrigley, N. "Exporting the British Model of Food Retailing to the US: Implications for EU-US Food Systems Convergence Debate." Agribusiness, 13(2), 1997, 123-35.
(1.) In this article, "distribution" includes retail trade, wholesale trade, transport, and storage.
(2.) The Pellergrini and Economic Planning Agency studies also use AGE modeling, but each considers only a single country.
(3.) Taxes collected by retailers are included in retail trade margins.
(4.) Because these are all OECD countries, the problems caused by quality differences will be smaller than if I was examining countries with larger differences in wealth. Quality's elusive, hard-to-measure nature means that attempts to correct for differences are speculative. The number of establishments per capita and the breadth of product assortment may provide some indication of quality. See Betancourt and Gaustschi (1993).
(5.) The model in this article is based on the computer code provided by Glenn Harrison, Thomas F. Rutherford, and David Tarr. Their code is available for public access online at http://dmsweb.moore.sc.edu/glenn/data/cge/ur and was used in their 1995, 1996, and 1997 articles.
(6.) Table 4 identifies the IRS sectors. The mark-ups used are available on request. Two different sets are available--one from GTAP and one from HRT. The results are robust to the choice of mark-ups.
(7.) The rate of return always increases in the simulations because streamlining margins always leads to a more efficient allocation of factors.
(8.) It could result from domestic accumulation or inward foreign investment or both.
(9.) The Armington assumption of imperfect substitutes may appear to be incompatible with my method for measuring protection and calculating margins, which depends on comparing goods that are equivalent. There is, in fact, no contradiction here, because I apply the Armington assumption to much more aggregated categories than were used to make the price comparisons.
(10.) See Gohin (1998) and Komen and Peerlings (1996) for other examples of modeling margins in this way within AGE models.
(11.) These modifications only apply to final goods. Due to lack of data, I do not modify the model to account for intermediate distribution. It turns out that these intermediate margins are quite a bit smaller than the margins for final goods.
(12.) As with the great majority of AGE work, this analysis varies the policies of focus (distribution margins) exogenously and does not allow them to adjust within the model. Doing would be interesting but lies beyond this article's scope. Betancourt and Gautschi (1988, 1990, 1993) and Bliss (1988) provide insightful analyses of what determines the size of distribution margins.
(13.) This approach implies either that firms do not have enough market power to maintain distribution margins above costs or that after convergence, mark-ups will be similar across countries. Betancourt and Gautschi (1993), in the most careful study of the issue I know, find little evidence of distributor price setting power to begin with. Also, an extensive body of literature draws on industrial organization theory to analyze the determinants of distribution margins. See Betancourt and Gautschi (1988, 1990, 1993) and Bliss (1988). Such work implies that completely unfettered distribution sectors in different countries could still have different margins if preferences or income levels (which affect the demand for distribution services) differ across countries. This article assumes that any such differences are not large enough to matter. Given the goal of surveying the forest instead of focusing on individual trees, relaxing this assumption lies beyond the scope of this article but would probably prove fruitful.
(14.) See Kacker (1988) for a detailed discussion. See also Cotterill (1997) and Wrigley (1997), who do not debate whether convergence will occur in food distribution but to which structure: that of the United States, in which manufacturers have large influence, or that of the United Kingdom, in which retailers are more influential.
(15.) As mentioned, internationalization has helped drive productivity improvements. Associated with this has been a trend toward more efficient scale, in which larger outlets replace inefficient small shops. See Boylaud (2000). Advances in information technology, especially the Internet, and better informed, more mobile, and more demanding consumers have also driven distributors to become more efficient. See Pilat (1997).
(16.) With CES preferences, the change in the value of consumption equals the change in equivalent variation.
(17.) Kalirajan (2000) provides evidence that high margins' deadweight loss outweighs their rents.
(18.) There is a fairly large set of literature on Japan's distribution system. In addition to Goldman (1992), see Itoh (2000), Nishimura and Punzo (1998), Knetter (1997), and Riethmuller (1996).
(19.) That the Australian and Canadian margins are usually close to, and sometimes below, the U.S. margins is not incompatible with evidence that the United States is more productive in distribution. If the U.S. margins in Table 5 include rent transfers, the true cost of distribution in the United States will be lower than the Table 5 numbers.
(20.) See Noll (2000) for an enlightening discussion of the connections between domestic regulation and international trade.
(21.) The import changes when all four countries liberalize simultaneously are very similar and are available on request.
SCOTT BRADFORD *
* I thank Alex Gohin, Glenn Harrison, and Tom Rutherford for helpful discussions. Special thanks go to Robert Lawrence. I am also grateful to an anonymous referee who provided very useful comments. I thank participants in the 2001 North American Summer Meetings of the Econometric Society and the Brigham Young University Raw Research ([R.sup.2]) Seminar.
Bradford: Assistant Professor, Brigham Young University, Provo, UT 84602. Phone 1-801-422-8358, Fax 1-801-422-0194, E-mail email@example.com
TABLE 1 Underlying Sample Product Categories Manufactured Food Products Household Goods Capital Goods Rice Men's clothing Structural metal products Flour and other Ladies' clothing Products of cereals boilermaking Bread Children's clothing Tools and finished metal goods Other bakery Infant's clothing Agricultural machinery products and tractors Pasta products Materials, yarns, Machine tools for metal accessories, etc. working Other cereal Men's footwear Equipment for mining, products metallurgy Fresh, frozen and Ladies' footwear Textile machinery chilled beef Fresh, frozen and Children's and Machinery for food, chilled veal infant's footwear chemicals, rubber Fresh, frozen and Furniture and Machinery for working chilled pork fixtures wood, paper Fresh, etc. lamb, Carpets and other Other machinery & mutton and goat floor coverings mechanical equipment Fresh, frozen and Household textiles, Office and data chilled poultry other furnishings processing machines Delicatessen Refrigerators and Precision instruments freezers Other meat Washing machines, Optical instruments, preparations, driers, dishwashers photographic equip. extracts Other fresh, frozen, Cookers, hobs and Electrical equipment chilled meat ovens including lamps Fresh, frozen or Heaters and Telecommunication & deep-frozen fish air-conditioners electrical equip. nec Dried, smoked or Vaccuum cleaners, Electronic equipment, salted fish polishers, etc. etc. Fresh, frozen, deep- Other major household Motor vehicles and frozen seafood appliances engines Preserved or Glassware and Boats, steamers, tugs, processed fish & tableware platforms, rigs seafood Fresh, pasteurised, Cutlery and Locomotives, vans, sterilized milk silverware wagons Condensed, powdered Motorless kitchen & Aircraft and other milk domestic utensils aeronautical equipment Other milk products Motorless garden Other transport excluding cheese appliances equipment Processed and Electric bulbs, unprocessed cheese wires, plugs, etc. Eggs and egg Cleaning and products maintenance products Butter Other nondurable household goods Margarine Drugs and medical preparations Edible oils Other medical supplies Other animal and Spectacle lenses and vegetable fats contact lenses Fresh fruit Orthopaedic and therapeutic appliances Dried fruit and nuts Passenger vehicles Frozen and preserved Motorcycles and fruit and juices bicycles Fresh vegetables Tyres, tubes, parts, accessories Dried vegetables Motor fuels, oils and greases Frozen vegetables Radio sets Preserved Television sets, vegetables, juices, video recorders, etc. soups Potatoes and other Record-players, tuber vegetables cassette recorders, etc. Potato products Cameras and photographic equipment Raw and refined Other durable sugar recreational goods Coffee and instant Records, tapes, coffee cassettes, etc. Tea and other Sports goods and infusions camping equipment Cocoa excluding Games, toys and cocoa preparations hobbies Jams, jellies, honey Films and and syrups photographic supplies Chocolate and cocoa Flowers, plants and preparations shrubs Confectionery Books Edible ice and Newspapers and other ice-cream printed matter Salt, spices, Durable toilet sauces, condiments articles and repairs Mineral water Nondurable toilet articles Other soft drinks Jewelery, watches and nec their repair Spirits and liqueurs Travel goods and baggage items Wine (not fortified Goods for babies, or sparkling) personal accessories Beer Writing & drawing equipment & supplies Other wines and alcoholic beverages Cigarettes Other tobacco products TABLE 2 Ad Valorem Margins: Distribution Costs as a Percentage of Output (with Output Valued at Producer Prices) ISIC2 Code Australia Belgium 1000 Agriculture, fisheries, 72.2 53.8 and forestry 3110/3120 Processed food 41.2 41.2 3130 Beverages 23.4 43.5 3140 Tobacco 19.1 13.3 3210 Textile 99.7 207.3 3220 Apparel 85.8 84.5 3230 Leather and products 100.9 57.4 3240 Footwear 103.6 55.9 3320 Furniture and fixtures 123.0 43.6 3410 Paper and products 101.6 52.4 3420 Printing and publishing 88.2 45.2 3522 Drugs and medicines 229.9 68.6 3529 Chemical products 80.6 76.2 3540 Petroleum and coal products 32.0 50.3 3550 Rubber products 62.6 19.8 3610 Pottery, china, etc 142.9 88.2 3810 Metal products 27.9 15.5 3825 Office and computing machinery 71.5 7.2 3829 Machinery and equipment, nec 41.5 3.2 3832 Radio, TV, and communication 42.8 19.9 equipment 3839 Electrical apparatus, nec 37.7 46.0 3841 Shipbuilding and repairing 0.2 0.8 3842 Railroad equipment 0.0 0.8 3843 Motor vhicles 47.4 16.8 3844 Motorcyles and bicycles 58.5 34.1 3845 Aircraft 1.7 0.8 3849 Transport equipment, nec 22.0 0.8 3850 Professional goods 90.5 33.7 3900 Other manufacturing, nec 90.3 56.9 Weighted geometric means 56.7 40.8 ISIC2 Code Canada Germany 1000 Agriculture, fisheries, 59.6 39.2 and forestry 3110/3120 Processed food 47.2 42.2 3130 Beverages 48.9 42.3 3140 Tobacco 45.6 42.3 3210 Textile 49.1 92.1 3220 Apparel 82.8 103.9 3230 Leather and products 23.2 103.9 3240 Footwear 74.6 103.9 3320 Furniture and fixtures 63.4 103.9 3410 Paper and products 66.9 103.9 3420 Printing and publishing 21.3 103.9 3522 Drugs and medicines 116.9 59.7 3529 Chemical products 79.8 73.5 3540 Petroleum and coal products 105.2 59.7 3550 Rubber products 98.4 37.4 3610 Pottery, china, etc 44.3 82.6 3810 Metal products 35.7 43.8 3825 Office and computing machinery 3.5 15.3 3829 Machinery and equipment, nec 18.0 15.3 3832 Radio, TV, and communication 23.6 32.3 equipment 3839 Electrical apparatus, nec 40.5 40.1 3841 Shipbuilding and repairing 4.8 3.5 3842 Railroad equipment 8.8 15.3 3843 Motor vhicles 22.0 30.2 3844 Motorcyles and bicycles 48.6 37.4 3845 Aircraft 1.5 15.3 3849 Transport equipment, nec 22.0 15.3 3850 Professional goods 73.9 72.3 3900 Other manufacturing, nec 60.2 66.1 Weighted geometric means 48.4 52.0 ISIC2 Code Japan Netherlands 1000 Agriculture, fisheries, 83.5 89.5 and forestry 3110/3120 Processed food 67.1 46.8 3130 Beverages 70.9 29.9 3140 Tobacco 39.8 23.0 3210 Textile 85.7 89.4 3220 Apparel 127.3 85.5 3230 Leather and products 107.1 67.0 3240 Footwear 92.6 19.8 3320 Furniture and fixtures 92.3 74.6 3410 Paper and products 78.6 70.6 3420 Printing and publishing 101.2 44.7 3522 Drugs and medicines 96.3 38.3 3529 Chemical products 93.7 87.9 3540 Petroleum and coal products 49.3 15.4 3550 Rubber products 100.7 25.0 3610 Pottery, china, etc 61.2 83.6 3810 Metal products 26.6 35.4 3825 Office and computing machinery 38.9 4.0 3829 Machinery and equipment, nec 26.1 4.9 3832 Radio, TV, and communication 31.8 31.5 equipment 3839 Electrical apparatus, nec 36.1 27.8 3841 Shipbuilding and repairing 5.0 0.0 3842 Railroad equipment 4.2 6.0 3843 Motor vhicles 51.7 14.6 3844 Motorcyles and bicycles 72.6 75.0 3845 Aircraft 16.8 0.0 3849 Transport equipment, nec 32.1 6.0 3850 Professional goods 74.4 58.5 3900 Other manufacturing, nec 101.0 79.3 Weighted geometric means 62.5 39.6 ISIC2 Code U.K. U.S. 1000 Agriculture, fisheries, 25.6 87.4 and forestry 3110/3120 Processed food 46.7 53.5 3130 Beverages 89.6 74.7 3140 Tobacco 12.9 57.8 3210 Textile 84.2 79.8 3220 Apparel 99.2 107.2 3230 Leather and products 100.1 85.0 3240 Footwear 94.5 120.4 3320 Furniture and fixtures 40.0 94.4 3410 Paper and products 96.3 145.7 3420 Printing and publishing 62.5 57.8 3522 Drugs and medicines 70.0 57.6 3529 Chemical products 79.5 69.3 3540 Petroleum and coal products 21.6 120.2 3550 Rubber products 21.6 134.1 3610 Pottery, china, etc 100.4 127.5 3810 Metal products 32.4 46.6 3825 Office and computing machinery 4.0 22.8 3829 Machinery and equipment, nec 4.2 26.1 3832 Radio, TV, and communication 22.3 28.2 equipment 3839 Electrical apparatus, nec 65.9 50.2 3841 Shipbuilding and repairing 0.0 5.5 3842 Railroad equipment 6.0 5.3 3843 Motor vhicles 16.5 20.8 3844 Motorcyles and bicycles 21.6 83.4 3845 Aircraft 0.0 1.4 3849 Transport equipment, nec 6.0 19.1 3850 Professional goods 61.7 83.2 3900 Other manufacturing, nec 73.3 104.8 Weighted geometric means 44.0 62.7 TABLE 3 Specific Margins: How Much It Costs to Move 100 Units of Output from the Factory to the Store Shelf (Where 1 Unit is $1 Worth of Factory Output Valued at World Prices) ISIC2 Code Australia Belgium 1000 Agriculture, fisheries, and 57.1 60.8 forestry 3110/3120 Processed food 41.1 55.6 3130 Beverages 33.9 47.9 3140 Tobacco 28.3 20.7 3210 Textile 110.6 196.2 3220 Apparel 99.6 132.7 3230 Leather and products 297.0 101.9 3240 Footwear 172.8 103.7 3320 Furniture and fixtures 159.8 85.4 3410 Paper and products 146.1 86.9 3420 Printing and publishing 87.6 59.3 3522 Drugs and medicines 139.3 116.0 3529 Chemical products 80.7 86.4 3540 Petroleum and coal products 68.0 169.9 3550 Rubber products 52.5 33.3 3610 Pottery, china, etc. 246.7 85.5 3810 Metal products 37.5 23.6 3825 Office and computing machinery 66.8 10.9 3829 Machinery and equipment, nec 55.0 5.7 3832 Radio, TV, and communication 49.5 31.3 equipment 3839 Electrical apparatus, nec 55.9 92.8 3841 Shipbuilding and repairing 0.3 1.0 3842 Railroad equipment 0.0 1.1 3843 Motor vehicles 52.2 23.3 3844 Motorcyles and bicycles 63.5 60.0 3845 Aircraft 2.0 1.0 3849 Transport equipment, nec 23.8 1.2 3850 Professional goods 95.0 56.7 3900 Other manufacturing, nec 112.2 105.0 Weighted geometric means 65.6 61.4 Other measures of restrictiveness Kalijaran (2000): 0-1 scale, 1 most restrictive 0.13 0.50 Nicoletti (2001): 1-4 scale, 4 most restrictive 1 3 ISIC2 Code Canada Germany 1000 Agriculture, fisheries, and 66.0 61.1 forestry 3110/3120 Processed food 53.7 61.5 3130 Beverages 68.9 53.4 3140 Tobacco 87.2 59.1 3210 Textile 65.6 125.4 3220 Apparel 92.9 152.3 3230 Leather and products 28.7 149.3 3240 Footwear 105.7 138.2 3320 Furniture and fixtures 98.8 144.7 3410 Paper and products 59.5 166.9 3420 Printing and publishing 27.3 106.4 3522 Drugs and medicines 313.2 157.7 3529 Chemical products 71.4 75.5 3540 Petroleum and coal products 138.9 169.9 3550 Rubber products 89.2 64.0 3610 Pottery, china, etc. 50.8 109.1 3810 Metal products 47.0 66.5 3825 Office and computing machinery 4.3 22.1 3829 Machinery and equipment, nec 18.0 20.5 3832 Radio, TV, and communication 28.9 41.8 equipment 3839 Electrical apparatus, nec 51.0 64.9 3841 Shipbuilding and repairing 5.3 5.2 3842 Railroad equipment 9.6 21.5 3843 Motor vehicles 26.3 39.1 3844 Motorcyles and bicycles 60.8 59.8 3845 Aircraft 1.7 19.6 3849 Transport equipment, nec 20.6 20.6 3850 Professional goods 78.5 102.7 3900 Other manufacturing, nec 70.4 121.8 Weighted geometric means 59.8 77.3 Other measures of restrictiveness Kalijaran (2000): 0-1 scale, 1 most restrictive 0.24 0.34 Nicoletti (2001): 1-4 scale, 4 most restrictive 2 2 ISIC2 Code Japan Netherlands 1000 Agriculture, fisheries, and 131.8 79.6 forestry 3110/3120 Processed food 139.8 60.8 3130 Beverages 107.8 38.2 3140 Tobacco 33.1 31.8 3210 Textile 143.0 132.5 3220 Apparel 175.4 109.5 3230 Leather and products 142.3 111.2 3240 Footwear 214.0 44.6 3320 Furniture and fixtures 249.8 109.8 3410 Paper and products 141.7 138.7 3420 Printing and publishing 117.1 57.5 3522 Drugs and medicines 117.2 128.4 3529 Chemical products 145.7 88.0 3540 Petroleum and coal products 165.6 66.7 3550 Rubber products 202.9 41.4 3610 Pottery, china, etc. 145.8 84.9 3810 Metal products 47.4 52.7 3825 Office and computing machinery 37.0 8.4 3829 Machinery and equipment, nec 40.8 7.8 3832 Radio, TV, and communication 36.4 43.6 equipment 3839 Electrical apparatus, nec 72.9 40.7 3841 Shipbuilding and repairing 5.9 0.0 3842 Railroad equipment 5.2 9.1 3843 Motor vehicles 45.2 24.4 3844 Motorcyles and bicycles 64.4 103.9 3845 Aircraft 17.2 0.0 3849 Transport equipment, nec 30.2 8.9 3850 Professional goods 74.2 79.1 3900 Other manufacturing, nec 248.4 129.2 Weighted geometric means 97.8 57.0 Other measures of restrictiveness Kalijaran (2000): 0-1 scale, 1 most restrictive 0.45 0.33 Nicoletti (2001): 1-4 scale, 4 most restrictive 4 2 ISIC2 Code U.K. U.S. 1000 Agriculture, fisheries, and 40.7 89.6 forestry 3110/3120 Processed food 53.4 52.7 3130 Beverages 97.3 69.4 3140 Tobacco 28.3 51.9 3210 Textile 131.7 78.4 3220 Apparel 97.9 90.1 3230 Leather and products 116.9 72.4 3240 Footwear 89.1 95.6 3320 Furniture and fixtures 86.9 76.6 3410 Paper and products 171.7 114.2 3420 Printing and publishing 56.7 45.7 3522 Drugs and medicines 129.1 178.8 3529 Chemical products 84.6 65.0 3540 Petroleum and coal products 87.8 95.9 3550 Rubber products 33.8 90.3 3610 Pottery, china, etc. 108.6 98.2 3810 Metal products 44.4 41.7 3825 Office and computing machinery 6.7 18.6 3829 Machinery and equipment, nec 6.1 26.5 3832 Radio, TV, and communication 27.4 24.4 equipment 3839 Electrical apparatus, nec 80.8 45.6 3841 Shipbuilding and repairing 0.0 5.2 3842 Railroad equipment 9.0 5.0 3843 Motor vehicles 27.9 22.3 3844 Motorcyles and bicycles 43.3 71.1 3845 Aircraft 0.0 1.3 3849 Transport equipment, nec 8.3 16.3 3850 Professional goods 93.1 76.0 3900 Other manufacturing, nec 90.9 79.8 Weighted geometric means 57.9 58.9 Other measures of restrictiveness Kalijaran (2000): 0-1 scale, 1 most restrictive 0.24 0.16 Nicoletti (2001): 1-4 scale, 4 most restrictive 3 1 TABLE 4 Sectors and Regions Used in the Applied General Equilibrium Model 22 Sectors 24 Regions Paddy rice Australia Wheat New Zealand Grains (other than rice and Canada wheat) Nongrain crops# United States Processed rice#* Japan Milk products# South Korea Meat products and livestock#* E.E.C. 12 Food, beverages, tobacco#* Indonesia Forestry, fishing, lumber, Malaysia wood paper wool#* Energy and energy products#* Philippines Minerals and mineral products* Singapore Textiles#* Thailand Wearing apparel#* China Chemicals, rubber, plastics#* Hong Kong Primary iron and steel* Taiwan Nonferrous metals* Argentina Fabricated metal#* Brazil Transport industry#* Mexico Machinery, equipment, other Rest of Latin America manufacturing#* Services and utilities Sub-Saharan Africa Trade and transport Middle East and North Africa Investment good Eastern Europe and Former Soviet Union South Asia Other European (EFTA, Switzerland, Turkey, SA) Notes: Sectors in boldface are the final goods sectors for which we inserted our margins measures. Underscored sectors are those that modeled with economies of scale and imperfect competition. Note: The final goods sectors for which we inserted our margins measures is indicated with #. Note: Those that modeled with economies of scale and imperfect competition is indicated with *. TABLE 5 Specific and "Efficient" Margins Australia Belgium Canada Nongrain crops 62.2 58.8 69.0 Processed rice 18.1 38.1 69.1 Milk products 21.5 53.4 43.9 Meat products and 42.8 82.0 52.5 livestock Food, beverages, tobacco 38.8 36.4 67.2 Fishing, wool, wood, paper 116.3 76.4 63.7 Energy products 68.0 169.9 138.9 Textiles 110.6 196.2 71.2 Wearing apparel 116.4 127.7 94.0 Chemical, rubber, plastic 109.3 89.8 127.0 Fabricated metal 37.5 23.6 47.0 Transport equipment 46.0 22.8 26.2 Machinery, equipment, 70.8 40.0 38.8 other manufacturing Germany Japan Netherlands Nongrain crops 65.7 178.5 87.9 Processed rice 69.1 84.2 44.5 Milk products 44.4 131.4 36.6 Meat products and 72.5 179.6 102.4 livestock Food, beverages, tobacco 56.9 121.6 43.8 Fishing, wool, wood, paper 127.8 126.0 84.7 Energy products 169.9 165.6 66.7 Textiles 125.4 143.0 132.5 Wearing apparel 150.1 177.3 98.3 Chemical, rubber, plastic 105.5 138.7 90.7 Fabricated metal 66.5 47.4 52.7 Transport equipment 37.6 42.2 24.1 Machinery, equipment, 56.1 62.6 39.1 other manufacturing Efficient U.K. U.S. Margin Nongrain crops 39.7 101.0 37.1 Processed rice 73.5 40.4 18.1 Milk products 60.0 48.0 20.7 Meat products and 55.3 51.5 40.1 livestock Food, beverages, tobacco 53.8 57.7 28.1 Fishing, wool, wood, paper 81.5 68.2 49.0 Energy products 87.8 95.9 66.7 Textiles 131.7 78.4 68.4 Wearing apparel 97.5 90.1 75.7 Chemical, rubber, plastic 78.8 105.3 74.9 Fabricated metal 44.4 41.7 17.0 Transport equipment 26.2 22.1 15.6 Machinery, equipment, 43.7 45.3 21.6 other manufacturing Notes: These numbers tell how much it costs to move one unit from the factory to the store shelf, where one unit is $100 worth of the product valued at world prices. Efficient margins are geometric means of the minimum margins at the detailed (124-sector) level. The simulations consist of setting all countries' margins for that product equal to the efficient margin. TABLE 6 95% Interval Estimates of Welfare Changes (Change in Equivalent Variation as a Percentage of GDP) Pure Rents * Lower Upper Impact of Australian reform on: Australia 0.14 0.24 Non-OECD countries 0.02 0.04 OECD countries 0.00 0.01 The world 0.01 0.01 Impact of Canadian reform on: Canada 0.31 0.35 Non-OECD countries 0.02 0.05 OECD countries 0.01 0.01 The world 0.01 0.02 Impact of Japanese reform on: Japan 1.14 1.17 Non-OECD countries 0.06 0.11 OECD countries 0.25 0.25 The world 0.21 0.22 Impact of U.S. reform on: United States 0.25 0.28 Non-OECD cCountries 0.06 0.11 OECD countries 0.08 0.09 The world 0.08 0.09 Impact of reform in all four countries on: Australia 0.26 0.31 Canada 0.46 0.48 Japan 1.14 1.17 United States 0.28 0.30 Non-OECD countries 0.09 0.18 OECD countries 0.35 0.36 The world 0.30 0.31 Central Case *** Lower Upper Impact of Australian reform on: Australia 2.44 2.54 Non-OECD countries 0.02 0.04 OECD countries 0.04 0.04 The world 0.04 0.04 Impact of Canadian reform on: Canada 2.71 2.75 Non-OECD countries 0.02 0.05 OECD countries 0.08 0.08 The world 0.07 0.07 Impact of Japanese reform on: Japan 5.66 5.69 Non-OECD countries 0.06 0.11 OECD countries 1.21 1.21 The world 0.97 0.98 Impact of U.S. reform on: United States 2.58 2.61 Non-OECD cCountries 0.06 0.11 OECD countries 0.82 0.83 The world 0.67 0.67 Impact of reform in all four countries on: Australia 2.56 2.61 Canada 2.86 2.88 Japan 5.66 5.69 United States 2.61 2.63 Non-OECD countries 0.09 0.18 OECD countries 2.16 2.16 The world 1.73 1.74 Pure DWL ** Lower Upper Impact of Australian reform on: Australia 4.74 4.84 Non-OECD countries 0.02 0.04 OECD countries 0.07 0.08 The world 0.07 0.07 Impact of Canadian reform on: Canada 5.11 5.15 Non-OECD countries 0.02 0.05 OECD countries 0.15 0.15 The world 0.12 0.13 Impact of Japanese reform on: Japan 10.17 10.20 Non-OECD countries 0.06 0.11 OECD countries 2.17 2.17 The world 1.73 1.74 Impact of U.S. reform on: United States 4.90 4.93 Non-OECD cCountries 0.06 0.11 OECD countries 1.56 1.57 The world 1.25 1.26 Impact of reform in all four countries on: Australia 4.86 4.91 Canada 5.26 5.28 Japan 10.17 10.20 United States 4.93 4.95 Non-OECD countries 0.09 0.18 OECD countries 3.96 3.97 The world 3.16 3.17 Note: These interval estimates assume that SIGMA MM = 8 and SIGMA DM = 4. * All of Area 1 in Figure 1 is assumed to be rent transfers for distributors. Thus, none of this rectangle is a welfare gain. ** All of Area 1 in Figure 1 is assumed to be deadweight loss. Thus, all of this rectangle is a welfare gain. *** This case assumes that half of Area 1 is rents and half is DWL. Other intermediate cases can be found by taking a weighted average of the two extreme cases. For instance, if one believes that 25% of Area 1 is rents, the welfare gains would be found by calculating 0.25 (PURE RENTS) + 0.75(PURE DWL). TABLE 7 Welfare Gain Estimates Using Distribution Productivity Measures from Pilat (1997) (Change in Equivalent Variation as a Percentage of GDP) 95% Interval Estimates Lower Upper Domestic impacts of reform Australia 4.59 4.69 Canada 5.10 5.14 Japan 8.07 8.10 United States 0.25 0.28 Impact of reform in all four countries on: Australia 4.71 4.76 Canada 5.25 5.27 Japan 8.07 8.10 United States 0.28 0.30 Note: These interval estimates assume that SIGMA MM = 8 and SIGMA DM = 4. TABLE 8 Import Changes in Nonservice Industries Caused by Distribution Liberalization (% and Millions of Dollars) Australia Canada Industry % Value % Value Food and agricultural products: Paddy rice -1.6 0 13.0 1 Wheat 0.5 0 3.6 0 Grains (other than rice and wheat) -0.3 0 4.4 5 Nongrain crops 6.7 22 8.5 269 Processed rice -3.0 -1 29.4 21 Milk products -3.7 -5 8.3 26 Meat products and livestock -2.9 -3 2.2 30 Other Processed Foods 0.3 5 10.2 421 Fishing-wool-wood-paper 0.1 3 -0.9 -72 Nonfood manufacturing: Energy and energy products -1.1 -23 10.0 595 Minerals and mineral products 0.0 0 0.0 0 Textiles 11.6 215 -2.2 -70 Wearing apparel 5.9 55 -2.3 -62 Chemicals, rubber, plastics 4.2 287 3.7 522 Primary iron and steel 1.0 8 0.2 5 Nonferrous metals 1.7 6 0.2 4 Fabricated metal 0.3 5 0.8 32 Transport industry 2.7 207 -0.4 -116 Machinery, equipment, other 4.2 803 -1.0 -493 manufacturing Total net change 3.3 1,583 0.8 1,119 Japan U.S. Industry % Value % Value Food and agricultural products: Paddy rice 8.0 0 3.7 0 Wheat 21.8 1,515 3.9 9 Grains (other than rice and wheat) 21.2 3,621 4.0 12 Nongrain crops 21.1 3,114 17.9 1,686 Processed rice 7.4 3 11.5 14 Milk products 28.4 738 13.0 134 Meat products and livestock 16.3 2,080 3.1 187 Other Processed Foods 18.3 2,467 8.9 1,388 Fishing-wool-wood-paper 4.4 1,269 0.8 256 Nonfood manufacturing: Energy and energy products 1.8 966 5.7 3,293 Minerals and mineral products -2.2 -299 0.5 73 Textiles 12.9 653 0.8 74 Wearing apparel 18.6 2,161 -3.4 -1,206 Chemicals, rubber, plastics 3.2 701 3.2 1,520 Primary iron and steel -1.0 -55 -0.3 -36 Nonferrous metals -0.6 -53 0.9 84 Fabricated metal -0.9 -26 1.7 197 Transport industry 0.7 90 -1.6 -1,463 Machinery, equipment, other -0.8 -389 0.8 1,975 manufacturing Total net change 6.6 18,557 1.4 8,196…