Introduction

Already at the turn of the millennium, trade within multinational corporations made up 60 percent of world trade, and among the 100 largest economic entities in the world in 2005, there were 54 states and 47 corporations. By manipulating the prices of transactions within corporations, multinational corporations can shift profits from high-tax countries to low-tax countries. Such profit shifting can have a large impact on the tax base for corporate taxation in host countries. In Norway, about 10-15 percent of the tax base for corporate taxes derives from foreign-owned corporations, and 20 percent from Norwegian-owned companies with foreign affiliates. This appendix briefly sums up the knowledge produced by academic research on the extent of the problem of transfer pricing by foreign corporations. The appendix is based on a larger report, Balsvik, Jensen, Møen and Tropina (2009), written for the Government Commission on Capital Flight from Poor Countries and The Norwegian Agency for Development Cooperation (NORAD).

No empirical analysis is needed to establish that multinational corporations manipulate transfer pricing to reduce their total tax burden. Many instances have been uncovered by tax authorities, and have been discussed in newspapers articles both nationally and internationally. Another indicator to the problem is found in the economic literature. Textbooks on international finance will typically discuss transfer pricing in detail, and leave no doubt that problems of taxation are central. However, such sources cannot tell us how much profit is withheld from taxation.

Empirical research into transfer pricing in multinational corporations seeks to quantify the profits that are withheld from taxation in different countries, and to establish which mechanisms are particularly important. Empirical research in fields that border on economic crime, however, is very difficult. If precise data were available, the tax authorities would quickly come to grips with the problem without assistance from researchers.

There are very few previous empirical studies based on Norwegian data. The literature consists of one scholarly article by Langli and Saudagaran in European Accounting Review in 2004, and a few Master’s dissertations. The central question of these studies is whether multinational corporations report lower taxable profits in Norway than other companies – all else being equal. In Balsvik et al. (2009) we update and expand on the analysis of Langli and Saudagaran along several dimensions. Langli and Saudagaran analysed the difference in profitability between Norwegian and foreign-owned companies based on accounting statistics from the mid 1990s. We have available to us a further nine years of accounting statistics, and use modern panel data techniques. Furthermore, we distinguish Norwegian multinational enterprises as a separate group and include more industries. In an innovative complementary analysis, we test a model by Jensen and Schjelderup (2009) of how aggregated flows of goods and services between foreign affiliates and their Norwegian parent companies should vary with the tax differential to Norway if they engage in tax motivated transfer pricing.

International research on transfer pricing and tax evasion has mainly been conducted on large corporations and in countries with high corporate tax rates, like the USA and Germany. Analyses of Norwegian data are interesting because Norwegian corporate tax rates are not particularly high, and because the structure of our economy includes many small companies. It is therefore not obvious that results reported in the international literature are relevant for Norway – or for developing countries, which along these dimensions are more like Norway than like the USA or Germany. A priori, it is not obvious whether the net profit shifting goes to Norway or from Norway.

Data

The analyses in Balsvik et al. are based on the linking of three different databases for the years 1992-2005:

• Annual accounts for all Norwegian enterprises with an obligation to report accounts to the Register of company accounts at Brønnøysund

• The SIFON-register of Statistics Norway, which includes an overview of direct and indirect foreign ownership in enterprises registered in Norway

• The census of foreign assets and liabilities from The Directorate of Taxes, which gives a survey of the foreign activity of enterprises registered in Norway (Utenlandsoppgaven)

Future research on multinational corporations would benefit greatly from improved data.

First, the classification of Norwegian-owned enterprises into Norwegian domestic and multinational enterprises is not exhaustive in our analysis. Among the Norwegian-owned enterprises with no activities abroad there will be affiliates of Norwegian multinational enterprises. In order to identify these as parts of multinational corporations, we need to know the complete corporate structure of all Norwegian enterprises, but historical information on the corporate structure of Norwegian companies is not easily available. Further work should be done to survey this.

Second, we do not know the corporate structure of the corporations that the foreign-owned enterprises in Norway belong to. When we do not know in which countries these corporations are active, we do not know their incentives for shifting profits from their Norwegian affiliates. Moreover, we have no information on transactions between Norwegian-owned enterprises and other foreign enterprises in the same group that are not affiliates owned from Norway. An extension of the obligation to file statements on transactions with closely related companies, as proposed in Ot.prp. nr. 62, 2006 – 2007, On the law on changes in tax legislation (transfer pricing), would remedy this.

Third, Norwegian customs data do not contain information on whether registered trade takes place with an affiliated company. American companies are obliged to give information on this when declaring imports and exports. This is because American authorities have, for a number of years, been concerned with the problem of transfer pricing. A simple improvement like this in the Norwegian data would make it possible to compare prices directly, and will probably also facilitate the control functions of the Norwegian Tax Administration.

The presence and economic importance of multinational corporations in Norway

The number of enterprises in Norway with foreign majority owners has increased gradually from a little more than 2000 in 1993 to nearly 5000 in 2005. If we include enterprises whose indirect foreign ownership share is more than 50 percent, there were more than 7000 enterprises with foreign majority ownership in 2005. Most foreign multinational enterprises are found in the trade sector. About 35 percent of the foreign-owned enterprises are in this sector. Nearly 20 percent of the foreign enterprises are in the knowledge-intensive part of the service sector, and close to 15 percent in the manufacturing and construction sectors.

In 2005, 1200 enterprises in Norway had direct foreign investments. They had a total of 4800 establishments abroad. The value of these investments was about 600 billion NOK. In 2005, the 10 enterprises with most capital invested abroad held as much as 52 percent of the total. Between 1990 and 1998, this share was more than 70 percent. This indicates that the strong increase in the number of enterprises registered with foreign investments between 1998 and 2001 is driven by a series of relatively small engagements. It is not clear whether the increase between 1998 and 2001 is real or is primarily caused by the fact that the work on the Census of foreign assets and liabilities was transferred from the Central Bank of Norway to Statistics Norway in 1998, and that the registration of ownership interests abroad was somewhat expanded in that connection. Manufacturing accounts for the greatest part of Norwegian enterprises with ownership interests abroad in 2005. A little more than 26 percent of Norwegian enterprises with ownership interests abroad are in the manufacturing industry.

About two thirds of the Norwegian-owned companies abroad are in the OECD-area, but the number of investments in Asia and Eastern Europe has increased since 1997. Based on country information in the Census of foreign assets and liabilities, it does not seem that Norwegian enterprises particularly often establish affiliates in tax havens, but there may be a significant under-registration of such companies. In 2005, 138 establishments were registered in tax havens. The shipping industry owned more than 40 percent of the establishments located in tax havens in the period from 1990 to 2005.

The count of the number of Norwegian enterprises that are a part of multinational corporations clearly shows that the Norwegian economy has seen an increase in multinational presence, and consequently an increased globalization. The importance of foreign multinational enterprises measured by their share of combined operating income, wage costs, and net capital has grown from about 10 percent in 1992 to somewhere between 20 and 30 percent in 2005. Unlike foreign multinational enterprises, Norwegian multinational enterprises have seen a slight reduction in their share of the economic activity in Norway in this period. The activity of Norwegian multinational corporations is concentrated to a few large corporations.

We find that enterprises that are a part of a Norwegian or foreign multinational corporation have a 10 to 15 percentage points higher probability of not being in a tax position than purely Norwegian enterprises in the same sector with a comparable size and debt/equity ratio.

The shifting of profits from Norway to low-tax countries by multinational corporations

The most convincing studies of transfer pricing compare the prices of goods traded within multinational corporations to the prices at which the same goods are traded in a market where the parties are independent, and discrepancies are seen in relation to the tax incentives of the companies. Although there are other management rationales for allowing transfer prices to diverge from market prices, such divergence is not allowed by OECD guidelines, which are based on the principle of “arms’ length”. Prices should be set as if the transaction took place between independent parties.

Researchers are seldom allowed access to suitable price data. For this reason, most of the literature on transfer pricing uses “indirect” methods. Since the point of manipulating transfer prices is to influence taxable profit, the most common research method is to compare the profits of national and multinational corporations by regression analysis. One advantage of the indirect method is that it can account for the effects of manipulated transfer prices on very company-specific goods and services, such as semi-finished products, royalties and head-office functions. The problem of transfer pricing is particularly important for such goods and services, precisely because a direct comparison of prices is not possible.

One main objection against the indirect method is that one can never “prove” that the observed differences are caused by the manipulation of transfer prices. In principle, other unobserved characteristics of multinational enterprises can be the cause of the observed difference. However, the suspicion that transfer prices are manipulated is strengthened if one finds that the differences between various types of companies vary according to the ease or difficulty with which transfer prices can be controlled, and according to the size of the tax differentials faced.

Langli and Saudagaran (2004) compare the profitability of Norwegian-owned and foreign-owned companies in manufacturing and trade in the years 1993 to 1996. They find that foreign-owned enterprises have a profit margin 2.6 percentage points lower than Norwegian-owned enterprises. This is consistent with a net shifting of profits out of Norway by foreign-owned enterprises. Langli and Saudagaran thus show that the problem of profit shifting is not limited to large enterprises and enterprises in countries with particularly high corporate taxes.

However, there is likely to exist long lasting differences in profitability between enterprises related to unobservable characteristics. Such effects can be caused, for instance by technology, market power, quality of management, location, or discrepancies between the real value of capital and its book value. It cannot be ruled out that such unobservable differences are correlated with foreign ownership. On the contrary, economic theory suggests that foreign-owned companies – or at least their parent companies – should have a better quality of management and better technology. This may have led previous comparative studies of profitability to misjudge the extent of profit shifting.

The problem of long lasting – fixed – unobservable effects can in principle be solved using methods that compare the change in profitability for companies which have been bought up (or bought home) with the change in profitability for companies whose ownership has not changed, and whose possibility for shifting profits has consequently not changed. However, such methods will underestimate the extent of profit shifting if some companies are wrongly classified. In that case, we would mix enterprises that have the possibility to shift profits with enterprises that do not. Another possible source of error is that “shocks” in profitability systematically influence the probability that ownership of enterprises shift between Norwegian and foreign owners. If foreign owners tend to buy enterprises that perform badly in order to restructure them, we will overestimate the extent of profit shifting out of Norway, and if foreign owners tend to buy up growth companies, we will underestimate the extent of profit shifting out of Norway. Earlier findings indicate that foreign take-overs in Norway are most frequently directed at growth enterprises.

An analysis of profit shifting based on a comparison of the profitability of multinational corporations in Norway to the profitability of similar Norwegian domestic corporations

The main analysis in Balsvik et al. explicitly takes Langli and Saudagaran (2004) as its point of departure, because it is the only published work based on Norwegian data and because it uses a recognized method. Thus, we can compare our results to theirs to provide a control for the quality and plausibility of our results.

Our selection includes only enterprises with limited liability with more than 1 million NOK in balance. Observations that lack central variables, or whose values diverge greatly, are also omitted. In the first part of the analysis, we use only enterprises in the sectors of manufacturing and trade, as did Langli and Saudagaran. The proportion of foreign-owned enterprises is about 3.6 percent in manufacturing, 1.1 percent in retail trade and 15.3 percent in wholesale trade. The foreign-owned enterprises are about four times as big as the Norwegian-owned ones. Our primary dependent variable is the profit margin, measured as results before tax as a proportion of sales. Results before tax are adjusted for changes in deferred tax costs and tax benefits. The average profit margin is 4.78 percent for Norwegian-owned enterprises and 3.10 for foreign-owned enterprises. The unconditional difference is thus 1.68 – or 35 percent.

In the regression analyses, we control for age, size, leverage, branch of industry, proportion of real capital and accounting year. We only partly succeed in replicating the results of Langli and Saudagaran. For the years 1993 to 1996, we find a dependent difference in profit margins between Norwegian-owned and foreign-owned enterprises in manufacturing and trade of 1.56 percentage points. The foreign-owned enterprises are the least profitable. The corresponding difference in Langli and Saudagaran is 2.57 percentage points. Qualitatively, however, the two analyses correspond well.

When we expand the sample to all the years from 1993 to 2005 and compare profitability within a detailed industrial classification, we estimate a dependent difference in profitability of 2.52 percentage points. The estimated difference is fairly stable from year to year, and there is no clear trend towards greater or lesser difference in profitability between Norwegian-owned and foreign-owned enterprises. If we control for unobservable, lasting, enterprise-specific fixed effects, the estimate is reduced to 1.64 percentage points.

The difference in profitability between Norwegian-owned and foreign-owned enterprises is greater for small enterprises than for large ones. We also find that the difference between Norwegian-owned and foreign-owned enterprises is particularly great for enterprises with poor profitability. Among the enterprises that are most profitable, for given characteristics, it seems that foreign-owned enterprises are somewhat more profitable than Norwegian-owned ones. The most obvious interpretation of this finding is that the tax authorities should focus particularly on foreign-owned enterprises that are substantially less profitable than expected. However, such a conclusion may be too hasty, as these findings are based on estimates that do not control for unobservable fixed effects. Enterprises with a high fixed effect, i.e., enterprises that are consistently more profitable then one would expect given their observable characteristics, will have a strong incentive to reduce taxes through manipulating transfer prices. They would also have a low risk of being discovered, since the resulting profits after the manipulation of transfer prices will seem quite normal. The tax authorities should therefore also verify transactions in enterprises with normal profitability.

Norwegian-owned multinational enterprises are more profitable than both foreign-owned multinational enterprises and Norwegian-owned enterprises with no activity abroad. This is reasonable. We would expect that the best domestic companies would be the ones to expand internationally. When we include a enterprise specific fixed effect in the regression, we analyse the change in profitability when enterprises change status. We find that when Norwegian-owned enterprises with no foreign activity establish affiliates abroad, their profit margins fall by 1.14 percentage points. This is consistent with a hypothesis that enterprises begin to shift profits out of Norway when they establish affiliates abroad. This effect is identified by those enterprises that change status in this respect during the period of observation. We probably misclassify several changes in multinational status, as the number of enterprises that submit a statement of foreign assets and liabilities varies conspicuously over time. As mentioned earlier, this will tend to make us underestimate the difference in profitability. For enterprises that are bought up by foreign owners (or bought home), we find that profit margins are 1.70 percentage points lower in years where enterprises are classified as multinational because of foreign ownership.

In the final part of our analysis, we include all industries in the private sector, except oil extraction and mining. We wish to exclude this sector from our analysis because oil companies are substantially larger than other companies, and are subject to a special tax regime. We find that foreign multinational enterprises have a profit margin 3.93 percentage points lower than Norwegian domestic enterprises. When we control for unobservable, enterprise-specific, fixed effects, the estimate falls to 2.38. This should be regarded as average values for the Norwegian mainland economy over the past decade. If we look at large, single industries, we find that the result is especially marked in real estate, renting and business activities. Here, the difference in profitability is estimated to 7.42 percentage points, 3.93 when we control for lasting effects. The results are also clear for construction and wholesale trade. We find that Norwegian multinational enterprises have a profit margin 1.69 percentage points lower than Norwegian domestic enterprises when we control for unobservable, enterprise-specific, fixed effects. It is thus a general trait that Norwegian-owned enterprises, too, become less profitable when they become multinational.

If we assume that the estimated differences in profit margins between Norwegian domestic and multinational enterprises is caused by the manipulation of transfer prices, we can use our estimates for a counterfactual analysis to give a rough estimate of what the tax revenue would have been if transfer prices were correct from the point of view of taxation. Our best guess is that between 25 and 40 percent of the potential tax take from foreign multinational enterprises in Norway is lost because of profit shifting. For the companies in our selection, this could amount to 15-25 billion NOK. Our selection represents around 90 percent of the turnover of all foreign-owned joint stock companies in Norway outside of oil extraction and mining. As a comparison, the Norwegian Tax Administration last year uncovered about 50 cases of what they believe is juggling with prices and invoices between closely related companies, involving all told 6.6 billion NOK.

An analysis of profit shifting based on the internal data of corporations on export and import

The analyses summarised above estimate net profit shifting. Many multinational corporations have affiliates in several countries, and in such cases, the incentive to manipulate transfer prices will vary according to the tax differential between Norway and the host country. This variation is lost if one analyses the effect of transfer pricing only on profits in Norway, since it is aggregated across the total engagement of the enterprise. There is every reason to believe that profits are moved out of Norway to countries with lower corporate taxes, and into Norway from countries with higher corporate taxes.

In the final part of our report, the analysis of the effect of tax incentives is based directly on book values of trade within corporations. The effect on corporate profits must necessarily derive from the effect on the value of trade within the corporation, so this approach may be regarded as somewhat more “direct” than the comparison of profits. Our analysis uses data from the Census of foreign assets and liabilities for the aggregated flow of goods and services between Norwegian parent companies and their foreign affiliates. Note that this set of data only involves a minority of the enterprises used in the profit comparisons.

Based on Jensen and Schjelderup (2009), Balsvik et al . set up a stylized model for trade between a parent company and an affiliate in another country. The model shows that changes in tax rates will have an effect on both price and quantity. We have data for the value – price multiplied by quantity – of the goods and services traded by parent companies in Norway with their affiliates abroad. From the model, we therefore derive predictions for the value of imports and exports in Norwegian parent companies – and for net export, which is the value of exports minus the value of imports. The theoretical analysis shows that the following relations should apply:

1. When the tax in Norway is higher than in the foreign country, and the tax level relative to the foreign country increases, the value of imports to the parent company from the foreign affiliate will increase. In this case, the extent of profit shifting from Norway to the foreign affiliate will increase. When the tax level relative to the foreign country falls, the effect will be the opposite. The effect on the value of the parent company’s imports cannot be unequivocally established in cases where the tax in Norway is lower.

2. When the tax in Norway is lower than in the foreign country, and the tax level relative to the foreign country increases, the value of the parent company’s exports to the foreign affiliate will decrease. In this case, less profit is shifted to Norway from the foreign affiliate. When the tax level relative to the foreign county falls, the effect will be the opposite. The effect on the value of the parent company’s exports cannot be unequivocally established in cases where the tax in Norway is higher.

3. When the tax level in Norway relative to the foreign country increases, the value of the parent company’s net export to the foreign affiliate will decrease. If the tax in Norway is lower at the outset – so that the change in taxation leads the tax levels of the two countries to converge – less profit will be shifted to Norway. If taxes in Norway are higher at the outset, more profit will be shifted from Norway.

We test these relationships on foreign affiliates, located in the OECD-area, of enterprises registered in Norway. Regression analyses that seek to explain the data from the Census of foreign assets and liabilities on the value of exports, imports, and net export with the tax differential between Norway and the respective host countries, support hypotheses 2 and 3. Hypothesis 1 on the effect of tax differentials on the shifting of profits out of Norway through manipulating the book value of imports to the Norwegian parent companies, does not find support. The reason for this can be that the prediction here refers to cases where the tax in Norway is higher than in the host country, and there are not many host countries in the OECD area where this is the case. There may also be questions in connection with offshoring of production that are not captured by the model.

Finally, we perform some calculations that illustrate the magnitude of the estimated effects. We find that if the tax rate in Norway increases from 28 to 30 percent, the value of exports from Norway to foreign affiliates in countries with higher taxes than Norway will be reduced by 7 to 14 percent. The imports to Norway from foreign affiliates in countries with lower tax than Norway will increase by 1 to 2 percent. The asymmetry between the effect on import and export values is primarily caused by the fact that the selection includes more parent company/affiliate-relationships where the host country has a higher tax rate than Norway.

Conclusion

We have performed extensive analyses of data for Norwegian enterprises, and have uncovered interdependencies that are consistent with profit shifting through the manipulation of transfer prices. The analyses are documented in Balsvik, Jensen, Møen and Tropina (2009). We find that multinational corporations shift profits both out of Norway and into Norway. We assess that the net flow is out of Norway, and that the loss in tax revenue can be in the order of 30 percent of the potential tax revenue from foreign multinational enterprises. We find that multinational enterprises in Norway have a profit margin of 1.5 to 4 percentage points lower than comparable domestic enterprises. This is consistent with the findings of Langli and Saudagaran (2004).

In the empirical analyses, we have had to make a number of discretionary choices about specific definitions and how to limit the selection. It would have been desirable to perform more robustness analyses than what has been possible within the project’s time frame. The results must therefore be regarded as indicative rather than finished and fully quality controlled. The estimate for the loss in tax revenue is particularly uncertain, and we will continue working on questions connected to this problem within the framework of other projects. However, the analyses we have summarised in this appendix back up the conclusion that, potentially , multinational corporations can withhold large amounts of money from taxation by shifting profits out of the country. Empirical research on multinational corporations and tax must therefore be characterized as a neglected field in Norway.

References

Balsvik, Ragnhild, Sissel Jensen, Jarle Møen and Julia Tropina (2009): “Kunnskapsstatus for hva økonomisk forskning har avdekket om flernasjonale selskapers internprising i Norge”, Forthcoming report from the Institute for Research in Economics and Business Administration in Bergen. The report will be available at www.snf.no.

Jensen, Sissel and Guttorm Schjelderup (2009): “Multinationals and Tax Evasion: Estimating a Direct Channel for Income Shifting”, mimeo, The Norwegian school of Economics and Business Adminstration

Langli J.C. and S.M. Saudagaran (2004): “Taxable Income Differences Between Foreign and Domestic Controlled Corporations in Norway”, European Accounting Review , 13(4), 713-741