 上传我的文档
 下载
 收藏
熟悉机械原理及结构；能熟练操作测量仪器和工具；熟悉各种安全规范；了解质量管理体系及管理方法
 下载此文档
正在努力加载中...
how costly are borrowing costs an analysis of alternative
下载积分：2500
内容提示：how costly are borrowing costs an analysis of alternative
文档格式：PDF|
浏览次数：0|
上传日期： 07:56:28|
文档星级：
全文阅读已结束，如果下载本文需要使用
 2500 积分
下载此文档
该用户还上传了这些文档
how costly are borrowing costs an analysis of alte
官方公共微信Merchanting and Current Account Balances - Beusch - 2015 - The World Economy - Wiley Online Library
Advertisement
Merchanting is goods trade that does not cross the border of the firm's country of residence. Merchanting grew strongly in the last decade in several European economies and has become an important determinant of these countries& current account. Because merchanting firms reinvest their earnings abroad to expand their international activities, this practice raises national savings in the home country without increasing domestic investment. This paper examines the empirical linkages between merchanting and the current account balance. Using a sample of 53 countries during , it shows that merchanting activity is a determinant of the medium-term current account balance.1&IntroductionMerchant trade is an offshore business in services for homogenous goods such as commodities, microprocessors and pharmaceuticals. A merchanting firm purchases goods from a supplier abroad and sells these to a buyer abroad without the goods entering or leaving the firm's country of residence. These goods do not undergo any transformation in processing between purchase and resale. Thus, a merchanting firm acts as an intermediary between companies located abroad that produce a good and companies located abroad that demand the good by providing storage and transportation services. The difference between revenues from the sale of the merchanter's goods and the purchase of the goods together with the incurred expenses to finance, insure, store and transport them is recorded as (net) merchanting. The asymmetrical accounting of merchanting is special in that the service is recorded as a positive entry in the balance of payments (BoP).Merchanting has recently become an important export in several European countries. Figure& plots merchanting as a share of GDP since 1990 for the four largest merchanting countries: Finland, Ireland, Sweden and Switzerland. The dynamics for Ireland are particularly striking: the merchanting-to-GDP ratio grew rapidly from 1.7 per cent in 2004 to 4.7 per cent in 2010. Similarly, Switzerland's merchanting-to-GDP ratio of 3.8 per cent in 2011 is not trivial. The same figure also shows that merchanting grew steadily after 2000 and continued to expand even during the financial crisis. Part of this expansion is explained by the observation that many merchanting firms continued to relocate to merchanting countries during the global economic downturn. In particular, Beusch and D&beli () show that the number of merchanting firms in Switzerland continued to increase during the financial crisis and this phenomenon contributed to the continued growth of merchanting receipts while trade in goods collapsed globally. At a time when the Swiss franc appreciated strongly, firm relocation cushioned adjustments in the current account (CA). Such a development makes it difficult to identify merchanting's effects for current account adjustment even in the face of large global shocks. Further, we know very little about the global scale of merchanting and its potential impact on a country's external balances.Figure&1. Merchanting/GDP Source: IMF BOPS, national institutions, WDI.This paper's objective is to show that merchanting is coincident with movements in the current account balance across a large set of countries over a long period of time. Presently, we leave out the development of a theoretical model instead, we focus on an empirical medium-term current account model. Based on our observations in the data, we argue that the linkages between merchanting and current account balances may occur through two channels. First, the relocation of merchanting firms may contribute positively or negatively to the current account. There is limited evidence that merchanting firms specialise in specific sectors and tend to cluster in particular countries. This channel is motivated in particular by the Swiss case during the global financial crisis. The second channel lies in the nature of merchanting's international activities. To expand their international activities in logistic and storage services, merchanting firms have to undertake capital-intensive investments abroad. Assuming that a portion of merchanting's retained earnings contribute to national savings in the home country without increasing domestic investment, this increase in the savings&investment gap impacts the current account. This channel is motivated by the fact that many merchanting firms are in the ownership of relatively few shareholders, are self-financed and are engaged in a capital-intensive business.To show the empirical linkages between merchanting and the current account, the savings impact of merchanting countries is estimated in empirical models of the medium-term current account. In regression models that control for a range of fundamental variables, merchanting is shown to impact positively the current account-to-GDP and the savings-to-GDP ratio but not the investment-to-GDP ratio.These new empirical results for merchanting contribute to the open economy literature on external adjustment. We expand the list of determinants that explain medium-term current account behaviour. The empirical model used in this study is closely related to empirical models in Chinn and Prasad (), Gruber and Kamin (, ), Chinn and Ito (, , ), Lee et&al. (), and Gagnon (). Using pooled regressions with cluster-corrected standard errors for a sample of 53 countries from 1980 to 2011, we show that merchanting as a share of GDP increases the medium-term current account balance-to-GDP ratio on average by 3 percentage points.The paper is organised as follows. Section& defines merchanting. The same section provides statistical evidence that shows merchanting's high level of persistence even during the financial crisis. Section& describes the empirical methodology and the data used in the panel regressions. Sections
presents the empirical results. Section& concludes.2&Merchant Trade: Definitions and DataThis descriptive section has three subsections. The first subsection addresses definitional issues. The second subsection presents information of the main characteristics for merchanting firms. Summary statistics used to motivate the empirical analysis in Section
are presented in the third subsection.2.1&DefinitionsThe IMF defines merchanting as the purchase of goods by a resident of the compiling economy from a non-resident combined with the subsequent resale of the same goods to another non-resident without the goods being present in the compiling economy (see IMF, ). The amount recorded under merchanting is the amount received by the domestic resident from the foreign customer that is less than the amount paid by the domestic resident to the foreign goods provider. The net profit resulting from these two transactions is recorded as a positive export value of business services, while any net loss is recorded as a negative export value of business services. Hereafter, we refer to net merchanting simply as merchantingMerchanting can arise from different sources and for any homogeneous tradeable good.A traditional form of merchanting is commodity trading where the merchanting firm buys and sells the goods from third parties and trades at world prices. A different case is multinational coordination activity, where merchanting is the result of the international fragmentation of production processes within a firm and reflects the organisational plans of a global multinational that locates merchanting services in one country, while the underlying production and ultimate distribution is elsewhere. In this case, the income from merchanting ultimately accrues to the foreign multinational and transfer pricing strategies might generate a measured current account surplus.2.2&Characteristics of Merchanting FirmsMerchanting's growth benefited strongly from the expansion of the global supply chain. The fragmentation and relocation of production processes have played a crucial part in merchanting's development. Although we are unaware of any empirical study that examines the microstructural features of merchanting firms, several observations can be offered.A first feature is firm mobility. Beusch and D&beli (), for example, show that the number of merchanting companies operating in Switzerland has doubled between 1990 and 2011. The vast majority of them were relocated from abroad and were not newly founded firms. The arrival of new merchant firms to Switzerland is the main reason why Swiss merchant trade continued to expand even during the financial crisis. In the face of global contraction, the continuous arrival of new merchant firms can partly be explained by the favourable domestic infrastructure (i.e. tax environment, liberal regulation, high concentration of merchant firms and support services to merchant firms such as law firms). Figure& shows that the income of merchant firms already domiciled in Switzerland before 2008 (marked in dark grey) decreased by 15.9 per cent between 2008 and 2009, which is equivalent to 1.3 per cent of GDP. This reduction in income shows that established merchanting firms were strongly affected by the financial crisis.Figure&2. Income of Merchant Trade of Domiciled and New Entrants to Switzerland, 2001&11 in CHF Billion Notes: A new entrant to Switzerland is defined as a merchanting firm that provides data to the Swiss National Bank for the first time in 2009 or later and entered the register of companies two to three years earlier. Before 2009 there were also new merchanting entrants to Switzerland but for the purpose of this illustration of relocation during the financial crisis only merchanting firms from 2009 and thereafter are considered.The active recruitment of merchanting firms by several countries underscores their mobility. Singapore, for example, created the Global Trading Programme in 2001. The intention is to attract mobile commodity trading houses with low taxes and light regulation. Malaysia has a similar relocation programme designed to attract 20 commodity trading firms by 2017. Firm relocation also explains why merchanting activities are concentrated in several cities: Dublin, Geneva, Hong Kong, Houston, London and Singapore. This agglomeration can partially be&explained by the tax environment of the merchanting countries. The average corporate tax&rate is lower in countries where merchanting is prevalent compared to countries where merchanting is absent.A further feature of merchanting firms is the importance of international communication networks between buyers and sellers. This point is illustrated by Swiss merchanting. A large share of Swiss merchanting activity is in commodities. The possibility that a seafaring activity is conducted in a landlocked country stems from the fact that factors beyond transport costs matter: well-established communication networks, the proximity of financial services, a non-restrictive regulatory environment and a flexible labour market are all equally important for Swiss merchanting.Another feature of merchanting is that it is not concentrated in a specific sector across the globe. For example, in Switzerland, merchanting is concentrated in commodities, chemicals and pharmaceuticals. In Finland, electronics and computers are the main merchanting activities, whereas in Ireland, publishing and chemical processing are important.2.3&Properties of the DataThe number of economies reporting merchanting has increased in recent years. Nevertheless, only 67 economies reported merchanting data to the IMF for 2010. Some of the largest economies (including China, the United Kingdom and the United States) and some likely important merchanting economies (including Hong Kong and Singapore) did not report merchanting data to the IMF.Following the existing literature on the determinants of the medium-term current account, such as Lee et&al. (), we choose our sample consisting of 53 countries that covers about 90 per cent of the world's GDP. The countries in our sample are listed in Appendix . Within our sample, merchanting data are available only for 27 countries from the IMF BOPS in 2010. Therefore, we extend and expand the available data on merchanting using the data provided by central banks and/or national statistical offices. Thus, our data set has merchanting data for 35 countries in 2010, and overall, there are merchanting data for 38 countries. Still, merchanting data for some important countries are missing in our sample.Figure& plots the merchanting-to-GDP ratio on the horizontal axis and the current account-to-GDP ratio in 2010 for 35 countries with available merchanting data. The figure shows that merchanting is relatively small for most countries. However, in most countries with a sizeable level of merchanting activity, that is greater than 0.5 per cent of GDP, merchanting contributes positively to the current account. We observe that the positive merchanting positions are large and concentrated among many small open economies with current account surpluses. By contrast, the negative merchanting positions tend to be small and are dispersed across many countries without a clear relationship with the current account. Based on the BoP accounting definition and the observations from Figure&, our empirical analysis focuses on countries with a positive value in merchanting.Figure&3. Merchanting/GDP and CA/GDP Balances in 2010 Source: IMF BOPS, national institutions, WDI.Table& provides statistical information for the 13 countries that we call merchanting countries in the empirical Section . These countries have a merchanting-to-GDP ratio of 0.5 per cent or higher for at least one year in our sample from 1980 to 2011. All measures in Table& are expressed as a share of GDP. The country rank for merchanting, shown in the first column, is based on the average merchanting value from 2008 to 2011. Next, the table presents mean, standard deviation, minimum and maximum values for merchanting based on annual data from 2000 to 2011. Four countries have an average merchanting income &1 per cent of GDP. These are Finland, Ireland, Sweden and Switzerland. The standard deviations are small and do not show large discrepancies. The minimum values show that merchanting countries, except for Malaysia, register positive merchanting from 2000 to 2011.Table&1.&Statistics for Merchanting/GDP in Countries with at Least One Observation Crossing the Threshold of 0.5 Per Cent1Ireland3.641.151.654.703.9023.33&5.950.072Switzerland1.711.240.303.713.710.714.1011.653Finland2.720.681.203.842.03&0.57&1.57&0.704Sweden1.390.440.832.191.702.412.177.115Hungary0.910.580.011.411.284.021.941.476Austria0.820.230.511.281.28&2.313.341.917Denmark0.870.190.621.111.092.931.606.678Belgium0.770.360.421.360.66&1.960.13&0.739Germany0.340.170.080.660.666.01&1.515.7210France0.320.120.160.550.55&3.690.31&2.1711Netherlands0.210.010.200.230.226.991.369.1312Malaysia0.160.28&0.120.520.3720.870.3016.48&Luxembourg0.870.780.001.530.00&10.4238.1811.99The last four columns of Table& show data on merchanting/GDP, the trade in goods balance/GDP, the trade in services balance (excluding merchanting)/GDP and the current account/GDP for these countries. All figures are net balances for the year 2011. For Belgium and Finland, merchanting is larger than the goods and the service balance (excluding merchanting). The last column shows that almost all merchanting countries have positive current account/GDP ratios.To motivate the empirical analysis in Section , Figure& offers a descriptive observation as to whether merchanting mitigates adjustments in the trade balance in merchanting countries. It is generally believed that the trade imbalance can be corrected through external demand or exchange rate adjustments. To determine whether merchanting behaves similarly to other trade components, Figure& plots merchanting, trade in goods and trade in services (excluding merchanting) for the last decade. Each series is expressed in terms of net balances and as a share of GDP. For each country, merchanting is less volatile than the other two series. A striking feature of the three time series is that merchanting was hardly affected by the financial crisis (post 2007) or by the great trade collapse (2008&09), whereas the other two series reveal temporary or even structural shifts. Figure& also shows that merchanting has been increasing or stable over time for all merchanting countries. Furthermore, it also shows a slow compositional shift from trade in goods towards merchanting in countries like Switzerland and Sweden.Figure&4. Components of the Current Account in Merchanting Countries (% of GDP) Source: IMF BOPS, national institutions, WDI.The analysis on positive trade balances in merchanting is also motivated by the linkages between export volatility and external savings. For example, it is argued that volatile oil exports lead to an increase in precautionary savings, which results in a positive external balance. By contrast, a visual inspection of Figure& suggests that merchanting firms& revenues exhibit a high level of persistence on the aggregate country level. Merchanting firms invest their earnings abroad to expand their logistic activities in storage and transportation (see Pirrong ). This structural feature of merchanting firms increases the savings&investment gap and thereby increases the current account surplus.A simple regression analysis supports the view that merchanting is highly persistent. Table& presents panel AR(1) regressions for those 38 countries for which we have merchanting data. The coefficient for the lagged variable is considerably higher for merchanting (i.e.&0.82) than for trade in goods (0.72), but it is similar to trade in services excluding merchanting (0.79). It is also important to note that the crises dummies for the years 2008 and 2009 are not significant for merchanting. This says that merchanting was not heavily influenced by the financial crisis. This is not true for the other trade components.Table&2.&AR(1) RegressionOwn lag0.83***0.79***0.72***[0.02][0.03][0.03]Dummy for 2008&0.00&0.00*&0.01*[0.00][0.00][0.00]Dummy for 2009&0.00&0.00**0.01***[0.00][0.00][0.00]Time variable0.00***0.00**0.00[0.00][0.00][0.00]Constant&0.00&0.00&0.00[0.00][0.00][0.00]Observations585585585Countries383838
0.7680.6320.525 overall0.9240.9680.925 between0.9660.9980.994To highlight the smoothness of merchanting over the financial crisis, variances of the residuals from the AR(1) regression in Table& are presented in Table&. In the samples considered, the variance of the residuals for merchanting is negligible compared to that of trade in services and that of trade in goods. In particular, during the post-crisis sample the variance of the residuals for trade in services and trade in goods increased, while merchanting's variance for the post-crisis period is similar to the pre-crisis sample.Table&3.&Variance of Residuals from AR(1) RegressionsMerchanting0.00000.00000.0000Trade in services0.00010.00060.0001Trade in goods0.00040.00280.0005The properties of increasing size and high persistence mean that merchanting does not behave like other components in the trade balance. These properties also imply that the current account balance of merchanting countries becomes more sticky. In other words, larger adjustments in either the exchange rate or external demand are needed to correct imbalances in the merchanting countries. These issues are analysed more formally in the following sections.3&Empirical MethodologyThe empirical framework used to estimate the medium-term determinants, that is four-year averages, on current account balances follows Lee et&al. (). In this model, the pooled regression is specified as follows:(1)where
is the current account balance of country i expressed as a share of GDP for period t (i.e. four-year average). Similarly,
is a vector of macroeconomic and demographic variables,
captures institutional or structural features through dummy variables, and , the error term, is assumed to be independent of the explanatory variables and normally distributed. In our set-up, equation
is extended to include merchanting:(2)where
is a dummy and captures the merchanting, , of country i. The merchanting dummy, , is +1 if
& 0.5 per cent in any year for otherwise 0. A threshold of 0.5 per cent is used to capture merchanting effects of a certain volume. Merchanting's impact is expected to act positively on a county's current account. The sample of 13 merchanting countries is given in Table&.The use of the merchanting dummy is motivated by the poor quality of the merchanting data. We suspect that merchanting is underreported (or enters elsewhere as an export in the BoP). Also, to test the robustness of merchanting, the dummy variable can be expanded for those countries that are believed to be active in merchanting but do not report it. As an alternative to the merchanting dummy, we also report OLS and IV regressions using the actual data. Apart from measurement issues, the IV regressions are also motivated by the fact that merchanting enters endogenously (i.e. relocation due to tax changes, networking effects, etc.) in equation .The selection of the remaining variables follows Lee et&al. () and covers 53 countries for the sample from 1980 to 2011. The panel is unbalanced, meaning that for some variables the length of the series varies by country due to missing data. Appendix Table
lists the data sources and offers brief comments.The macroeconomic and demographic variables, , in equation
are standard in the literature and are briefly discussed next. These variables include the fiscal balance, demographic determinants, net foreign assets (NFA) and economic growth. For the fiscal balance, it is assumed that a higher government budget balance raises national saving. This, in turn, increases the current account balance. The fiscal balance in equation
is defined as the ratio of the general government budget balance to GDP in deviation from the average budget balance of trading partners: if the government budget balance improves in all countries, there would be a worldwide macroeconomic effect but little expected effect on the current account balance of each country.The demographic determinants assume that a higher share of the economically inactive dependent population reduces national saving and decreases the current account balance. To proxy for this, Lee et&al. () include an old-age dependency ratio as well as population growth. The intention of the latter variable is to capture the share of economically dependent young people. Both demographic variables are measured in deviation from trading partner averages and are expected to decrease the current account balance.Net foreign assets enters as a determinant in equation . The assumption is that economies with a high NFA benefit from higher net foreign income flows, which tend to create a positive association between NFA and current account balances. The initial NFA position is used in equation
to avoid capturing a reverse link from the current account balance to NFA.Economic growth is included for two reasons. If economies in the early stages of development have a greater need for investment, this is often financed through external borrowing. As developing economies grow and approach the income levels of advanced economies, their current account balances should improve. Alternatively, if countries are at a similar stage of development, the stronger economic growth relative to its trading partners should lower the current account balance.Equation
includes two measures of growth. The first variable is the ratio of GDP per&capita in purchasing power parity terms to the US level, which Lee et&al. () define as relative income. This variable is assumed to measure the relative stage of economic development. The second growth variable is the deviation of the real per capita GDP growth rate from its trading partner average. This variable is used to capture relative economic growth. In equation , the current account balance is expected to increase with relative income but to decrease with relative growth.Equation
also includes countries& oil balance. Higher oil prices increase the current account balance of oil-exporting countries and decrease the balance of oil-importing countries. In equation , Norway is treated as a separate oil country because of its high level of intergenerational savings.Several dummy variables, , are included in equation
to capture country or industry-specific features. A dummy that controls for small open economies with large financial centres is included among others. The evidence in Lee et&al. () shows that financial centres tend to run substantial current account surpluses. This effect is captured with a dummy for the following countries: Belgium, Hong Kong, Luxembourg, the Netherlands, Singapore and Switzerland.Empirical evidence shows that crisis dummies have an impact even after controlling for a range of macroeconomic factors. Chinn and Ito (), Lee et&al. () and Gagnon () show that economic crises tend to unleash strong current account adjustments as a by-product of macroeconomic contraction because of the reduced availability of international financing. Two sets of crises dummies are considered. The first dummy controls for the Asian crisis. Aizenman () and others argue that Asian countries increased their precautionary savings after the Asian crisis to insure themselves against future crisis. This dummy acts as a levels shift. The second dummy captures episodes of banking crises. We use the Laeven and Valencia () measure of international banking crises. The motivation is to capture temporary output losses that are linked to banking crises.A last set of dummy variables control for aging societies and the introduction of the euro. These dummies have not been used extensively in the literature but do enter the Lee et&al. () set-up. The aging dummy is +1 for Germany, Italy, Japan and Switzerland and 0 for the rest. This dummy treats the four aging societies as outliers. Further, we introduce a first euro dummy that takes value +1 for Germany starting in 1999 and 0 for all other countries, and a second euro dummy that takes value +1 for Greece, Portugal and Spain and 0 for all other countries. The intention here is to capture potential extreme countries within the currency union. In the specification where we replicate Lee et&al. (), there is only one euro dummy with value +1 for Germany, &1 for Greece, Portugal and Spain and 0 for the rest.4&The Empirical Impact of Merchanting in Medium-term CA ModelsThis section presents the empirical results. All regressions are with clustered standard errors. The first set of results presented in subsection 4a shows that merchanting is a robust determinant of the current account. Merchanting's impact of 3 per cent in the baseline specification of equation
is sizable. Further robustness checks are presented in subsection 4b. The regressions show that the results from 4a are not sensitive to different specifications of the merchanting dummy. Subsection 4c presents IV regressions to account for endogeneity and measurement errors.4.1&Merchanting Countries in Medium-term Current Account ModelsOur baseline regressions of equation
are presented in Table&. Column 1 shows regression estimates for the medium-term model as specified by Lee et&al. () without dummy variables. All the estimated coefficients
however, the demographic and growth variables are statistically insignificant. Column 2 shows the same regression but now adds the merchanting dummy. This variable has a coefficient of 0.03 and is statistically significant at the 1&per cent level. This result says that the current account is increased by 3 per cent for those countries that have merchanting exports &0.5 per cent of GDP. In other words, merchanting is coincident with an over-proportional increase in the current account. Because the unweighted average size of the merchanting-to-GDP ratio when the merchanting dummy is +1 is 0.96 per cent, this implies that the estimate of merchanting's impact is greater than the benchmark of unity (i.e. one dollar of merchanting raises the current account by one dollar).Table&4.&Baseline CA Regressions ()Fiscal balance0.27***0.26***0.26***0.24***0.22***0.21**[0.09][0.09][0.09][0.09][0.08][0.08]Old-age dependency ratio&0.10&0.10&0.05&0.050.020.01[0.07][0.07][0.07][0.06][0.07][0.06]Population growth0.070.140.170.250.490.55[0.75][0.79][0.72][0.75][0.72][0.76]Initial NFA0.06***0.06***0.05***0.05***0.03**0.03***[0.01][0.01][0.01][0.01][0.01][0.01]Oil balance, Norway0.18**0.24***0.17**0.24***0.27***0.32***[0.08][0.08][0.08][0.08][0.08][0.08]Oil balance, rest0.21***0.21***0.27***0.27***0.30***0.30***[0.04][0.04][0.04][0.04][0.04][0.04]Output growth0.090.100.110.130.130.14[0.17][0.16][0.15][0.15][0.15][0.15]Relative income0.020.010.03**0.020.020.01[0.01][0.01][0.01][0.01][0.01][0.01]Banking crisis dummy&&&&&0.00&0.01&&&&[0.01][0.00]Asian crisis dummy&&0.05***0.05***0.05***0.05***&&[0.01][0.01][0.01][0.01]Financial centre dummy&&&&0.03**0.03&&&&[0.01][0.02]Euro introduction: Germany&&&&0.02*0.01&&&&[0.01][0.01]Euro introduction: Periphery&&&&&0.04***&0.04***&&&&[0.01][0.01]Aging society dummy&&&&0.010.02&&&&[0.01][0.01]Merchanting/GDP&0.5% dummy& 0.03*** & 0.04*** & 0.03*** & [0.01] & [0.01] & [0.01] Constant&0.000.00&0.01&0.01&0.01&0.01[0.01][0.01][0.01][0.01][0.01][0.01]Observations287287287287287287
0.5660.5990.6240.6580.6680.696One interpretation of the large coefficient is that the dummy variable is capturing related activities to merchanting. For commodities, for example, merchanting entails storage and transportation but our dummy is also possibly capturing an additional transformation of processing. An alternative interpretation is a compositional effect: merchanting is underreported, but the missing activity is incorrectly attributed to another component in the BoP. In this case, merchanting is understated but the current account is not. In subsection 4c, we present IV regressions based on actual merchanting to overcome the potential problem of measurement errors as an alternative to the dummy variable estimates.Next, the Asian crisis dummy is added to the specification. The regressions with and without the merchanting dummy are shown in columns 3 and 4. The regression estimates in column 4 show that merchanting unleashes almost the same level of external savings as the Asian crisis (i.e. 5 per cent for the Asian crisis versus 4 per cent for the merchanting dummy). Both dummies are highly statistically significant.A further step to determine the robustness of our estimate is to examine whether merchanting holds up with other dummies that have been argued to be important. The regressions in columns 5 and 6 include the small financial centres dummy, the banking crisis dummy, the euro dummy and the aging dummy. The estimated coefficient for the merchanting dummy remains stable at 3 per cent. The regression in column 6 shows that the strength of these dummy variables is weakened once merchanting is introduced. For example, the dummy for small financial centres is no longer significant in column 6. The significance of the euro dummy is only significant at the 10 per cent level when merchanting is introduced. Similarly, the banking and the aging dummies never figure prominently with or without merchanting.Another simple check is to compare the results in Table& with the estimates from Lee et&al. (). For this exercise, we shorten our sample from 1980 to 2007 and consider the alternative specification in Lee et&al. () based on the lagged current account. These results are given in Table&. Columns 2 and 5 show that the coefficient on the merchanting dummy remains stable at 3 per cent and is significant in the shortened sample for the NFA. The same is true for the lagged capital a however, the estimated coefficient is lower. Our estimates for the NFA specification in column 1 are close to the estimates of Lee et&al. () shown in column 3 under the IMF heading. The main difference in the coefficients is for population growth. In Lee et&al. (), this coefficient is about six times smaller. Instead for the lagged current account specification shown in columns 4 to 6, there is the additional difference for the coefficient on the fiscal balance. Our estimates show that this coefficient is five times smaller and statistically insignificant compared to the estimates by Lee et&al. (), which are reproduced in column 6.Table&5.&Comparative CA Regressions ()Fiscal balance0.21**0.21**0.20***0.040.040.19***[0.08][0.08]&[0.07][0.08]&Old-age dependency ratio&0.05&0.06&0.14**0.010.01&0.12**[0.07][0.07]&[0.05][0.05]&Population growth&0.26&0.18&1.21***&0.09&0.05&1.03[0.71][0.75]&[0.62][0.65]&Initial NFA0.04***0.04***0.02***&&&[0.01][0.01]&&&&Lagged CA&&&0.67***0.64***0.37***&&&[0.06][0.06]&Oil balance, Norway0.28***0.33***&0.25***0.28***&[0.08][0.08]&[0.06][0.06]&Oil balance, rest0.33***0.33***0.23***0.19***0.20***0.17***[0.05][0.04]&[0.04][0.04]&Output growth0.070.09&0.21**&0.11&0.08&0.16*[0.14][0.14]&[0.13][0.12]&Relative income0.010.000.02*&0.00&0.000.02*[0.01][0.01]&[0.01][0.01]&Banking crisis dummy&0.01*&0.010.01*&0.01*&0.010.01[0.01][0.01]&[0.00][0.00]&Asian crisis dummy0.04***0.04***0.06***0.03***0.03***0.04***[0.01][0.01]&[0.01][0.01]&Financial centre dummy0.03**0.03*0.03***0.04***&&0.03***&&&&&&[0.01][0.02]&[0.01][0.01]&Euro introduction dummy0.01*0.01&0.02**0.01**&[0.01][0.01]&[0.01][0.01]&Aging society dummy0.010.02&0.010.01&[0.01][0.01]&[0.01][0.01]&Merchanting/GDP&0.5% dummy& 0.03** && 0.01** && [0.01] && [0.01] &Constant&0.01&0.01&&0.00&0.00&[0.01][0.01]&[0.01][0.01]&Observations234234NA220220NA
0.6490.6770.520.7370.7420.56Next, we show that merchanting's impact on the current account operates through an increase in the savings&investment gap. To do this, Table& presents regressions for the current account-to-GDP, investment&savings-to-GDP, investment-to-GDP and the savings-to-GDP ratio. The specification follows the baseline model (i.e. column 6 in Table&). Table& presents the coefficient estimates for the merchanting dummy. The results show that the coefficient for merchanting dummy is positive and statistically significant with the investment&savings gap and savings, but not with investment.Table&6.&Merchanting and the Savings&Investment GapMerchanting/GDP (if & 0.5%) dummy3.01***3.73***&1.232.51**&[1.05][1.54][0.94][1.22]Observations296306306306
0.6820.5890.3920.6564.2&Merchanting Countries: Robustness ChecksIn this subsection, alternative definitions of merchanting countries are shown to be robust in equation . The previous regressions were based on a single definition for the merchanting dummy, that is whether merchanting in a year for period t is &0.5 per cent with respect to GDP. The regression results with alternative definitions of merchanting countries are presented in Table&.Table&7.&Robustness of Merchanting CountriesFiscal balance0.22***0.23***0.23***0.21**0.22***0.22***0.21**[0.08][0.08][0.08][0.08][0.08][0.08][0.08]Old-age dependency ratio0.020.020.020.010.020.020.01[0.07][0.07][0.07][0.06][0.06][0.07][0.06]Population growth0.490.530.530.550.540.430.53[0.73][0.72][0.72][0.76][0.74][0.74][0.76]Initial NFA0.03**0.03**0.03**0.03***0.03**0.03**0.03***[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Oil balance, Norway0.27***0.27***0.27***0.32***0.30***0.29***0.32***[0.08][0.08][0.08][0.08][0.08][0.08][0.08]Oil balance, rest0.30***0.30***0.30***0.30***0.30***0.30***0.30***[0.05][0.04][0.05][0.04][0.04][0.04][0.04]Output growth0.130.130.130.140.130.130.14[0.15][0.15][0.15][0.15][0.15][0.16][0.15]Relative income0.020.020.020.010.010.010.01[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Banking crisis dummy&0.00&0.00&0.00&0.01&0.01&0.00&0.01[0.01][0.01][0.01][0.00][0.01][0.01][0.00]Asian crisis dummy0.05***0.05***0.05***0.05***0.05***0.05***0.05***[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Financial centre dummy0.03**0.03**0.03**0.030.03*0.03**0.03[0.01][0.01][0.01][0.02][0.02][0.01][0.02]Euro introduction: Germany0.02*0.02*0.02*0.010.02*0.02**0.02[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Euro introduction: periphery&0.04***&0.04***&0.04***&0.04***&0.04***&0.04***&0.04***[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Aging society dummy0.010.020.010.020.020.010.02[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Merchanting&&&0 dummy0.00&0.00&&&&[0.01]&[0.01]&&&&Merchanting&&&0 dummy&0.000.00&&&&&[0.01][0.01]&&&&Merchanting&&&0.5% dummy&&& 0.03*** &&&&&& [0.01] &&&Merchanting&&&1% dummy&&&& 0.03*** &&&&&& [0.01] &&0.5%&&&Merchanting&&&2% dummy&&&&&& 0.03** &&&&&& [0.01] Merchanting&&&2% dummy&&&&& 0.04**
0.04** &&&&& [0.01]
[0.02] Constant&0.01&0.01&0.01&0.01&0.01&0.01&0.01[0.01][0.01][0.01][0.01][0.01][0.01][0.01]Observations287287287287287287287
0.6690.6690.6690.6960.6860.6810.697The regressions show that the volume of merchanting activity is important for its impact on the current account. The regressions in the first three columns define a dummy variable equal to one if a country reports positive or negative merchanting values at least once during the four-year average. In each of these regressions, merchanting is not statistically significant. The regression in column 4 uses the definition from Table& with a threshold of 0.5 per cent. It is reproduced for completeness. Next, in the regression shown in column 5, the threshold for the merchanting-to-GDP ratio is increased from 0.5 to 1.0 per cent. This change in the threshold has no impact on the regression estimates. There is no difference in the regression estimates shown in columns 4 and 5. Similarly, the regression in column 6 augments the threshold to 2.0 per cent with no change in the coefficient and in statistical significance. These results show that the definition used in Tables
is robust to higher threshold levels.A reasonable suspicion based on these results is that the size of the merchanting dummy is driven by the few observations with high merchanting-to-GDP ratios. Column 7 shows that this is not the case. The main result remains unchanged if two separate merchanting dummy variables are used. The first dummy takes the value one for observations with merchanting-to-GDP ratios between 0.5 and 2 per cent, and the second dummy for those above 2 per cent (identical to the dummy in column 6). The coefficients of the two dummies are 0.03 and 0.04 and are both highly significant. From this evidence, we conclude that the merchanting result is not driven by the largest merchanting countries.The robustness of the merchanting countries is further examined in two ways: expanding the number of merchanting countries in which no reported information is available and by reducing the country sample from 53 to 38. The first exercise expands the merchanting dummy for the United States, Hong Kong and Singapore for the last two four-year averages. The second exercise reduces the sample to 38 countries for which we have merchanting data. Table& presents the regressions for these two exercises. The first column shows that the merchanting dummy remains significant even if we consider additional countries for which we have no data. The second column again shows that the merchanting dummy remains statistically significant even when the country sample is reduced. As expected in the reduced cross-country sample, the coefficient estimates differ sharply from the full country sample.Table&8.&Expanding the Number of Merchanting Countries and Reducing the Country SampleFiscal balance0.20**0.08[0.08][0.07]Old-age dependency ratio&0.06&0.14[0.07][0.08]Population growth&0.09&1.54***[0.70][0.56]Initial NFA0.03**0.03**[0.01][0.01]Oil balance, Norway0.35***0.00[0.07][0.01]Oil balance, rest0.31***0.36***[0.04][0.05]Output growth0.10&0.24[0.15][0.17]Relative income0.010.02[0.01][0.02]Banking crisis dummy&0.01&0.00[0.01][0.00]Asian crisis dummy0.05***0.05***[0.01][0.02]Financial centre dummy0.03*0.03[0.02][0.02]Euro introduction dummy: Germany0.01*0.02*[0.01][0.01]Euro introduction dummy: periphery0.010.02*[0.01][0.01]Aging society dummy0.02*0.01[0.01][0.01]Merchanting/GDP & 0.5% dummy 0.03***
[0.01] Constant&0.01&0.01[0.01][0.01]Observations296218
0.6490.647To demonstrate that our merchanting variable is a special activity missing in standard medium-term models of the current account and not merely a random subcomponent of the current account, a counterfactual exercise is performed with financial services. Financial services may be defined as an upstream industry that is also important for merchanting. The objective is to show that a financial services dummy defined in a similar manner as the merchanting dummy with a threshold of 0.5 per cent does not have the same coefficient of 0.03 or even better is not statistically significant.Table& presents OLS regressions with financial services. The results show that financial services have a negative coefficient and its statistical significance is not robust. Column 1 shows the baseline regression of column 5 in Table& without merchanting and without financial services. It serves as a reference for the regressions presented in columns 2 to 5. The OLS regression in column 2 is the baseline regression with merchanting with a coefficient of 0.03. Column 3 presents the same regression with financial services. In this regression, the coefficient on financial services is &0.02 and is statistically significant at the 5 per cent level. In other words, a positive trade balance in financial services is associated with a lower current account. This odd result is partially explained by the offsetting increase in the financial centre dummy. This financial centre dummy has a stronger effect with the introduction of financial services. To determine the strength of financial services on its own, we next drop the financial centre dummy. This regression, presented in column 4, shows that the coefficient for financial services is &0.01 and is statistically insignificant if the financial centre dummy is dropped. Next, the regression in column 5 presents the estimates from the full model with merchanting and financial services. The robustness of merchanting's coefficient estimate of 0.03 and its statistical significance holds, while financial services& coefficient is negative and is not statistically significant. From this evidence, we conclude that merchanting is a special activity that has not been captured in medium-term current account models.Table&9.&Financial ServicesFiscal balance0.22***0.21**0.23***0.25***0.22**[0.08][0.08][0.08][0.09][0.08]Old-age dependency ratio0.020.010.01&0.020.00[0.07][0.06][0.07][0.07][0.06]Population growth0.490.550.580.650.64[0.72][0.76][0.72][0.74][0.75]Initial NFA0.03**0.03***0.03**0.04***0.03***[0.01][0.01][0.01][0.01][0.01]Oil balance, Norway0.27***0.32***0.24***0.17*0.29***[0.08][0.08][0.08][0.09][0.08]Oil balance, rest0.30***0.30***0.29***0.27***0.29***[0.04][0.04][0.04][0.04][0.04]Output growth0.130.140.130.110.14[0.15][0.15][0.15][0.15][0.14]Relative income0.020.010.020.03*0.01[0.01][0.01][0.01][0.02][0.01]Banking crisis dummy&0.00&0.01&0.00&0.00&0.00[0.01][0.00][0.01][0.01][0.01]Asian crisis dummy0.05***0.05***0.05***0.05***0.05***[0.01][0.01][0.01][0.01][0.01]Financial centre dummy0.03**0.030.04***&0.03*[0.01][0.02][0.01]&[0.02]Euro introduction dummy: Germany0.02*0.010.02*0.010.01[0.01][0.01][0.01][0.01][0.01]Euro introduction dummy: periphery&0.04***&0.04***&0.04***&0.04***&0.04***[0.01][0.01][0.01][0.01][0.01]Aging society dummy0.010.020.020.02*0.02[0.01][0.01][0.01][0.01][0.01]Merchanting/GDP&0.5% dummy& 0.03*** && 0.03*** & [0.01] && [0.01] Financial services/GDP&0.5% dummy&&&0.02**&0.01&0.02&& [0.01]
[0.01] Constant&0.01&0.01&0.01&0.01&0.01[0.01][0.01][0.01][0.01][0.01]Observations287287287287287
0.6680.6960.6720.6570.7004.3&Interpreting Merchanting's Impact with IV EstimatesThis section offers IV regression estimates of merchanting's impact on the current account based on the actual merchanting-to-GDP ratio. Our intention is to interpret the previous estimates using an alternative estimation strategy that corrects for endogeneity and potential measurement errors. The IV strategy instruments for the merchanting-to-GDP ratio with the merchanting(rest of the world)-to-GDP(rest of the world) ratio. For this exercise, the merchanting variable is defined as
are the four-year averages for the merchanting-to-GDP ratio. The rest of the world instrument, , first defines the set of j merchanting countries, when
for a particular year in period t. Next, if i is in the set of the j merchanting countries, the nominator of
sums merchanting in the j countries except for country i, otherwise 0. For the denominator of , it is the sum of
except for .The motivation for our instrument is based on information and communications technology used in merchanting. As noted in Section , merchanting in a country tends to be sector specific. Thus, merchanting from the rest of the world should not be correlated with the current account of country i. However, what is common about merchanting across countries is that it is a logistic and network service for homogeneous goods. Logistic and network services have expanded because of recent advances in information and communications technology. This technology enables intermediary parties to operate in a country that is independent from the final buyer and original seller.Tables
present estimates from the two-stage IV regressions. The evidence is consistent with the view that our instrumentation strategy is valid and that the estimated impact of merchanting remains above unity when controlling for endogeneity and measurement problems. Further, the differences in the coefficient estimates between the OLS and IV regressions are small.Table&10.&Robustness Check: IV Estimations, First StageRest of world&&&&merchanting/GDP (if & 0.5%)&1.20***&&&&[0.25]&&merchanting/GDP (if &1%)&&0.96***&&&&[0.20]&financial services/GDP (if & 0.5%)&&&0.71&&&&[0.46]Observations&287287287
&0.5770.6740.420F-stat&620.3577.20.730Kleiberger-Paap F-stat&22.3123.932.313Table&11.&Robustness Check: IV Estimations, Second StageFiscal balance0.25***0.22***0.22***0.20**[0.08][0.08][0.08][0.09]Old-age dependency ratio0.060.020.02&0.05[0.06][0.06][0.06][0.14]Population growth0.550.500.500.62[0.74][0.72][0.72][0.70]Initial NFA0.03***0.03***0.03***0.03***[0.01][0.01][0.01][0.01]Oil balance, Norway0.31***0.32***0.31***0.20[0.08][0.08][0.08][0.13]Oil balance, rest0.29***0.30***0.30***0.29***[0.04][0.04][0.04][0.04]Output growth0.120.130.130.19[0.14][0.15][0.15][0.20]Relative income&0.000.010.010.04[0.01][0.01][0.01][0.05]Asian crisis dummy0.05***0.05***0.05***0.05***[0.01][0.01][0.01][0.01]Financial centre dummy0.03**0.03**0.03**0.03***[0.02][0.02][0.01][0.01]Euro introduction dummy: Germany0.02*0.02*0.02*0.02**[0.01][0.01][0.01][0.01]Euro introduction dummy: periphery&0.04***&0.04***&0.04***&0.04***[0.01][0.01][0.01][0.01]Aging society dummy0.010.020.020.01[0.01][0.01][0.01][0.01]Merchanting/GDP (if & 0.5%) 1.55**
2.02*** && [0.60]
[0.78] &&Merchanting/GDP (if & 1%)&& 1.73*** &&& [0.64] &Financial services/GDP (if & 0.5%)&&&&0.33&&& [0.55] Observations287287287287
0.6870.6800.6820.629F-stat&292528482236Table& presents only the coefficient and the standard errors of the instrument, , from the first-stage regression. The coefficient of the instrument is positive and highly significant for different threshold levels: 0.5 and 1.0 per cent. In each specification, the instrument passes tests of weak identification. The Kleibergen and Paap () statistic as well as the F-statistic from the first-stage regressions reveals that the criticism of weak instruments is not an issue. A further test of our instrument shows that the instrumentation strategy does not work for financial services. The coefficient for financial services (rest of the world), shown in column 4, is insignificant.The second-stage IV regressions are presented in Table&. As a means for comparison, the OLS regression is presented in column 1. Here, merchanting has a coefficient of 1.6 in the OLS regression. This says if the merchanting-to-GDP ratio is &0.5 per cent, then on average the CA-to-GDP ratio will increase by an amount of 1.6 as large as the merchanting-to-GDP ratio. The second column shows the second-stage IV regression with the merchanting variable with a threshold of 0.5 per cent. Merchanting has a significant coefficient of 2. The coefficient differences in the OLS and IV models are small. Column 3 performs the same regression as in column 2 but with the threshold set to 1.0 per cent instead of 0.5 per cent. In this IV regression, the coefficient for merchanting decreases slightly from 2 to 1.7. As a counterfactual exercise, column 4 shows that the instrumentation strategy for merchanting does not work for financial services.5&ConclusionsThis paper presents evidence for the macroeconomic relationship between merchanting and the current account. Merchanting is an export service (i.e. logistic and storage services) of goods that do not undergo any form of processing and are bought and sold without crossing the national borders of the residing merchanting firm. In countries with high levels of merchanting activity, the current account increases. This mechanism is explained by the fact that merchanting increases the savings&investment gap. Unlike many other exporting firms with domestic production, merchanting firms invest their earnings abroad to expand their logistic and storage facilities. The empirical results show that merchanting is associated with an over-proportional increase in the current account. This result is robust to different model specifications and different variable definitions for merchanting. The estimated impact greater than unity is explained by&the fact that merchanting tends to be underreported and that the timing of merchanting (i.e. logistic and storage services) is closely linked to other exports (i.e. processing).The importance of merchanting in the medium-term current account models also has implications for the adjustment debate on global imbalances. The size and persistence of merchanting have changed the dynamics of a country's current account. Because merchanting is difficult to predict (i.e. poor data quality and firm relocation), this introduces a further source of uncertainty in studies by Cline and Williamson () and Lee et&al. (), and others that make exchange rate assessments based on medium-term current account models.The empirical evidence for merchanting supports several directions for future research. One avenue would be to develop a theoretical model that shows why merchanting improves the current account balance. A starting point would be to assume that large merchanting activity reflects temporary monopoly power in an intertemporal smoothing model. Another avenue that merits greater analysis is estimating merchanting's sensitivity to exchange rate movements. Our conjecture is that merchanting is less sensitive to real exchange rate movements, than say is trade in goods. While several studies highlight differences in exchange rate elasticities between goods and services, elasticities for merchanting across sectors have not been estimated.Notes1The users of merchanting services record their transactions as either an imported or exported good. Other components of the BoP are symmetric in that positive and negative entries are possible. Negative&entries for merchanting in the BoP however, they are in general small and non-persistent (i.e. a duration of negative profits for aggregated merchanting services is unlikely).2These private firms do not publish financial statements. Thus, their earned earnings and dividend policy is not publicly available. See, for example, Swiss National Bank () and Pirrong () for more information on merchanting.3The merchanting channel for the current account is different from the behaviour of intertemporal consumption smoothing in commodity exporting countries, see Van der Ploeg and Venables (). In commodity exporting countries, a sovereign wealth fund is often created on precautionary savings grounds to insure future generations and current investment decisions from price swings in commodities. Instead, as noted in Pirrong (), merchanting in commodities trading is sensitive to swings in volumes and less so in prices.4In this paper, the analysis for merchanting uses the IMF BPM5 classification, which treats merchanting as a component of trade in services. In the newly introduced IMF BPM6 classification, merchanting of goods is reclassified from services to goods. The value of net exports of goods under merchanting under BPM6 is of the same value as merchanting services under BPM5.5This includes, for example, hard commodities such as crude oil, soft commodities such
as well as computer chips, books or chemical raw materials.6The same may hold for commodity traders who have vertically integrated production and distribution. The locational choice for merchanting services is often driven by tax optimisation strategies. See Swiss National Bank () for more information on merchanting.7Beusch and D&beli () also show that in the period from 1990 to 2011, most of the merchanting firms are clustered in Geneva and Zug, five merchanting firms in Switzerland had been liquidated, four merchanting firms merged, and no merchanting firm resettled abroad.8See, for example, the information under -corporate-tax-rates as well as &Singapore's low taxes lure Trafigura&, Financial Times, 22 May 2012.9Furthermore, merchanting is often highly concentrated among large firms. In the Irish case, the top 10 companies account for approximately 70 to 80 per cent of overall merchant trade in 2010 (Private correspondence with the Irish Central Bank.) Similar to Ireland, the eight largest merchanting firms are responsible for 70 per cent of Switzerland's merchanting activity (see Beusch and D&beli, ).10See Table
in Appendix .11This information is based on email exchanges with national authorities.12Close to 200 countries reported BoP data to the IMF for that year. Among the 67 countries with merchanting data in 2010, 15 entries registered no merchanting activity.13Some countries provided data in the past but no longer do. The Netherlands is a case in point. The problem of missing observations for the non-reporting countries is compounded by an underreporting bias for those countries that do report. First, there is the problem of lagged reporting when new firms are identified to be engaged in merchanting activities. Second, not all merchanting firms are identified in the country BoP surveys. We do not attempt to correct for these problems, but note that these biases understate results presented in section .14However, merchanting is equal to zero for two countries.15Hong Kong's Census and Statistics Department publishes data on the &gross margin involved in merchanting& as part of Hong Kong's offshore trade statistics. We choose, however, not to include this data because of the high value (11 per cent of GDP in 2011), and Hong Kong's trade links with China suggest that merchanting in Hong Kong is not comparable with the IMF definition in the BoP Manual. Our sample captures a merchanting activity of US$97 billion in 2011: a fivefold increase from 2000.16The IMF BoP Statistics includes values for Ireland for 2000 and 2001 (&1.1 per cent and 0). Based on discussion with the Central Statistics Office of Ireland, they only started to collect merchanting data in 2004. We thus ignore the earlier IMF values. It should also be noted that the Belgian time series includes breaks due to methodological changes from 2006 to 2007 and 2009 to 2010.17The three observations for Luxembourg are not displayed but follow similar patterns as described further in the text.18There is a large literature that examines the links between export income volatility and external savings. Recent examples include Bems and de Carvalho Filho () and Cherif and Hasanov ().19As a consequence, volatile export revenues of oil-producing countries are often filtered out of empirical models of the medium-term current account. See in particular Chinn and Prasad (), Chinn and Ito (, , ), Gruber and Kamin (, ), Lee et al. () and Gagnon ().20A further consideration for external adjustment, not pursued in this paper, is a firm's sensitivity to exchange rate movements. Wren-Lewis and Driver (), Crane et al. () and Bosworth and Collins (), highlight the observation that external adjustment through trade in services is slower than through trade in goods. The common view is that an exchange rate appreciation facilitates external adjustment to correct a trade surplus. Because a large share of merchanting activity brings together buyers and sellers of standardised products (i.e. commodities, microchips) traded outside of the national borders, the volume of this service is heavily dependent on global demand and less on domestic currency movements. This means that merchanting should be less sensitive to exchange rate movements than say trade in goods.21The model by Lee et&al. () has been updated by Phillips et&al. () and expanded by Sastre and Viani (). These models focus on the cyclical nature of the current account and consider the normative contribution of policy variables. Because we are interested in the structural nature of merchanting, the strategy developed by Lee et&al. () is used.22The endogeneity of merchanting may arise from an omitted variable such as low tax rates that we do not control for in the panel. Tax rates may be linked with other elements of the CA such as income flows from multinationals.23See Appendix
for a list of the countries.24See also Appendix 2.1 in Lee et&al. () for further discussion of the data set.25Only in the case of Ricardian equivalence, where private saving fully offsets changes in public saving, is the link broken between government budget balances and current account balances.26In the four-year averages, the dummy is set to +1 if it takes value +1 in one of the four years.27The Lee et&al. () sample is from 1973 to 2004. Hence, we are unable to fully replicate their results.28In an alternative specification, three dummies were included: the first dummy is +1 if
for observations between 0.5 and 1 per cent, otherwise 0; a second dummy is +1 if
for observations between 1 and 2 per cent, otherwise 0; and a third dummy is +1 if
for observations above 2 per cent, otherwise 0. All merchanting dummy variables were again statistically significant and have similar coefficients as the merchanting dummy in the baseline regression.29Obviously, this does not exhaust
however, financial services do figure prominently with the importance of financial centres in current account regressions. Note also there are no obvious candidates in which the BoP entry is strictly positive as in the case of merchanting.30Another strategy uses the dynamic methods of Arellano and Bond (). Experimentation with this method revealed mixed results largely because the medium-term framework of Lee et&al. () has many variables and is not suitable for dynamic instrumentation.31The clustering of a particular merchanting service for a particular country arises because important specialised activities in finance, legal services and insurance, which support merchanting, are developed at the local level.32At the 2.0 per cent threshold, there are too few observations.33The banking crisis dummy was dropped in the regression presented in Table&. This was done because the banking crisis dummy was always insignificant in the previous regressions and was correlated with our instrument.Appendix&A:&Data DescriptionSampleAlgeria, Argentina, Australia, Austria, Belgium, Brazil, Canada*, Chile*, China*, Columbia, Croatia*, Czech Republic, Denmark, Egypt*, Finland, France, Germany, Greece, Hong Kong*, Hungary, Indonesia*, India, Ireland, Israel*, Italy, Japan, Korea, Luxembourg, Malaysia, Mexico, Morocco, New Zealand, the Netherlands, Norway*, Pakistan, Peru*, Philippines, Poland, Portugal, Russia, Singapore*, Slovakia, Slovenia, South Africa*, Sweden, Switzerland, Spain, Thailand*, Tunisia, Turkey, Venezuela*, United Kingdom, United States*.Countries denoted with a * have no available merchanting data for the sample .Merchanting DummyAustria (7), Belgium (3), Denmark (2), Finland (3), France (1), Germany (1), Hungary (2), Ireland (2), Luxembourg (1), Malaysia (1), the Netherlands (1), Sweden (4), Switzerland (3).Note: The number of observations where the dummy is +1, that is when merchanting/GDP & 0.5 per cent is given in the parentheses. Numbers in bold denote a sequence of +1 dummy values that terminate with the final observation. For example, Denmark (2) denotes a dummy of +1 for the observation 2004&07 and 2008&11; otherwise, the dummy is zero.Table&A1.&Data and their SourcesCurrent accountIMF BOPSMeasured as ratio to GDPGDPWDIReal GPD in 2000 US$Fiscal balanceWEOGeneral government net lendingOld-age dependency ratioUNOld-age dependency ratio (population between 30 and 64 as ratio to population & 65)Population growthUNPopulation growthGDP growthWDIGDP growth (per capita, real LCU)Initial net foreign assets (NFA)IFS, LMWhen NFA is missing in the lane and Milesi-Feretti data, it is substituted with IFS dataOil balance, NorwayWDI&Oil balance, othersWDI&Relative incomeCGERRelative income (ratio of per capita PPP GDP to US level, 2000 US$)MerchantingBOPS, otherMissing BOPS data is replaced by central bank and statistical offices& data whenever possible Trade data: Nonoil and oil tradeDOTS, WDITotal exports/imports from DOTS, fuel exports/imports from WDIGoods and services tradeUN&Weights for global consistency calculationDOTSOwn calculationWeights for deviation from trading partnerUNOwn calculation Dummy variables: Banking crisisLVLaeven and Valencia () class. Borderline crises are not taken into accountAsian crisisLee et&al. ()Asian crisis =emerging Asia countries as classified by IMF; 0=all other. See Lee et&al. ()Financial centreLee et&al. ()1=Switzerland, Luxembourg, Hong Kong, the Netherlands, Singapore, B 0=all other. See Lee et&al. ()Euro introductionLee et&al. ()1=Germany, &1=Portugal, Spain, G 0=all otherEuro introduction : Germanyown1=G 0=all otherEuro introduction: Peripheryown1=Portugal, Spain, G 0=all otherAging populationLee et&al. ()1=Germany, Switzerland, Japan, I 0=all otherAppendix&B:&Tax Rates in Merchanting CountriesAverage corporate tax rates have been lower in countries where merchanting has been prevalent compared to countries where merchanting has been absent. Between 2000 and 2010, the average corporate tax rate in the 13 merchanting countries in our sample was 27 per cent, whereas for the nonmerchanting countries, it was 30 per cent. Table
lists average and latest tax rates for merchanting versus nonmerchanting countries. The (unweighted) average measure based on World Bank corporate tax rates understates the true difference between merchanting and nonmerchanting countries. Several countries, such as Hong Kong, Singapore and the United States, which we define as nonmerchanting country in the empirical analysis, also have low corporate taxes. Further, for some countries, taxes for merchanting activity are considerably lower than the national corporate tax rates.Table&B1.&Tax Rates for Merchanting Countries (%)Profit tax&&13.4217.32Income, profit and capital gains tax27.0430.0426.0629.60Total tax rate44.2249.0742.0246.56
Advertisement}