Determinants of the negative impact of being landlocked on trade: an empirical investigation through the Central Asian case.
The fall of the USSR led to the creation of 15 new countries, of which nine are landlocked. By geographic definition, a landlocked country is one that does not have open access to the sea. (2) At present, more than one out of five countries in the world is landlocked.
If we exclude micro- or mini-European States, (3) only two out of 38 landlocked countries--both being European States as well--are classified as high-income economies, while 46 out of 207 coastal states are classified as high-income economies. The economic definition of being landlocked is much more complex than the geographic explanation. Countries are economically landlocked in the sense that their economic development is constrained when several factors are present like remoteness from major markets, poor infrastructure and border crossing difficulties, which imply high transportation costs. It is worth noting that most of these factors are present in island developing country economies as well.
Until now, most of the papers concerning trade in land-locked countries have focused primarily on the link between being landlocked, transport costs and infrastructure or being landlocked and growth (eg Radelet and Sachs, 1998; Stone, 2001; MacKellar, 2000). The only exception worth noting is Limao and Venables (2001), who completed a study of the impact of being landlocked on trade.
Difficult access to port facilities should not seriously impede trade and economic growth on developing countries. Nevertheless, countries in development, which are deprived of sea access, register poor trade growth. Europe is the only exception to this phenomenon.
Today, land-locked Central Asian countries accumulate geographic, economic and institutional obstacles, while Central European countries gather geographic, economic and institutional strengths. Central Asian countries (4) have been neglected in the surveys concerning land-lockedness, but are gravely affected by geographic obstacles. Land-locked regions have been studied since 1991 to explain the growing gap between Russia's southern and western flanks. First, we will state reasons linked to the impact of being landlocked in this region. Then we will give a more general analysis of land-locked countries. However, the main objective of this paper is to quantify the impact of being landlocked on trade using a gravity approach and to determine the most important factors of this impact.
BEING LANDLOCKED, TRANSPORT COSTS, TRADE AND GROWTH
Some geographers argue that the new economic geography neglects "real places'. Geographers have studied the issue of being landlocked extensively, and to get a better understanding of how these problems affect economic conditions in Central Asia, it is crucial to acknowledge their contribution.
According to Doumenge (1986), independence for land-locked states has generally been granted in exchange for acceptance of territories with inadequate climates and economic difficulties. Most of these states are quasi-deserts, deserts or mountainous areas. Most European land-locked states are on the Alpine Arc, most of those in Africa lie in the Sahel region or on the continental ridge and the Central Asian land-locked states are at the heart of the world's largest endoreic (5) basin, which is mostly semi-desert. Land-locked countries are generally small countries. (6)
Geographers like Debrie and Steck (2001) claim, however, that the impact of being landlocked is relative. Indeed, throughout history a centrally located country could be viewed as having ah advantage, whereby several decades later, perhaps after a period of economic decline, it could be considered a disadvantage to be landlocked. For instance, from the third century until the 16th century, Central Asia was at the crossroad of the trade between the East and West. However at the beginning of the 17th century, for different reasons, caravan trade between Asia and Western Europe declined. By using this example, a country could be categorised as being either centrally located or landlocked, even though the geographic characteristics do not change.
Gallup et al. (1999) highlight the fact that land-locked countries may be at a disadvantage for three reasons: cross-border migration of labour is more difficult than internal migration, infrastructure development across national borders is much more difficult to arrange than similar investments within a country, and coastal economies may have military or economic incentives to impose costs on landlocked countries. MacKellar et al. (2000) analysed the impact on growth. From a sample of 92 low-income and middle-income countries, over the period between 1960 and 1992, they estimated that states that had no direct access to sea outlets had a 1.5% lower growth rate. They advocate for the development of alternative transportation routes, rather than a focus on trade policies because it is 'fundamentally a transport problem'. Finally, they argue that regional trade arrangements significantly mitigate the negative effects of being landlocked on trade.
Some economists have also studied the transport costs issue. Radelet and Sachs (1998), using cif/fob data from IMF for 97 developing countries (of which 17 were landlocked), have estimated that transport and insurance costs are twice as high for landlocked countries as coastal countries. They demonstrate that there is a link between transport costs and economic growth, notably for developing countries. For these states, exports are crucial for foreign currency earning, and are necessary for purchasing the imported capital goods needed to obtain long-run growth. Higher shipping costs dramatically reduce the rents earned and increase the cost of expensive intermediate goods imports. In conclusion, Radelet and Sachs state that 'geographic isolation and higher shipping costs may make it much more difficult it not impossible for relatively isolated developing countries to succeed in promoting manufactured exports'. Limao and Venables (2001) obtained the same results concerning transport costs. They also highlight the positive impact of a developed transport infrastructure. Using a gravity model, they estimated that the median land-locked country only has 30% of the trade volume of the median coastal economy. For Sub-Saharan Africa, they find that poor infrastructure is a major determinant of the low bilateral trade levels. Stone (2001), using IMF estimations of 'freight payments as a percentage of imports', found a comparable transport cost burden. Out of 64 possible comparisons between land-locked countries and transit countries, the transport costs were lower in transit states than in land-locked countries in 75% of the cases. Overall, 18 out of the 30 land-locked countries studied have transport costs higher than 10% of the value of imports, while only 11 of the 32 coastal transit countries have rates as high. In Africa, for instance, 13 out of 15 land-locked countries have burdens of at least 10%, with seven exceeding 20%.
Determinants of transport costs
In the case of Central Asian countries (cf. Table 1), the assumption of a high transport cost burden is confirmed. Indeed, per kilometre, it is two and a half times more expensive to send a container from the USA to Bishkek than Ankara.
Recently, there have been several studies that have attempted to find the major determinants of transport costs. (7) These studies find that there are several crucial factors affecting transport costs in land-locked countries: the commodity structure of trade, the relative remoteness from major markets, the infrastructure level, the percentage of overland transport and the possibility of alternative transport routes.
An important factor in determining overall trade is the inverse relation between transport costs and the total value of goods transported. (8) The problem of a high transport cost ratio is particularly acute in Central Asia, where raw materials are a substantial proportion of exports. According to the COMTRADE database, at the three- of four-digit levels, in 2000, the five main export products represented, respectively, 90% of the total exports in Turkmenistan, 90% in Tajikistan, 70% in Kazakhstan, 60% in Kyrgyzstan and 60% in Uzbekistan. Among all these flows, other than aircraft parts in Tajikistan, textile fabrics in Turkmenistan and flat rolled iron in Kazakhstan the other exports are completely dependent on raw materials.
The commodity structure of trade has an indirect impact on transportation in the region. Indeed, in Central Asia, transport is almost exclusive]y achieved by land (road, rail or pipeline). It was a low priority of Soviet planners to develop air freight. As aircraft is the most expensive mode of transport, low value-added goods are not transported by air. Consequently, air freight, which could break the effects of being landlocked, is negligible in the region. According to World Bank data in 2000, air freight (estimated in millions of tons-kilometre) was 12 for Kazakhstan, four for Kyrgyzstan, three for Tajikistan, 12 for Turkmenistan and 75 for Uzbekistan. Comparable figures were 30,131 for the USA, 1,937 for Switzerland and 1,041 for Russia. Although the commodity structure of trade is important for landlocked countries, their problem is not unique. Developing coastal countries face the same challenge of the structure of trade as land-locked developing countries, but do not have the same transport costs burden.
The theory, "distance melts', underlines this point indirectly by referring to large ports with good infrastructure. Sea access is crucial nowadays. Maritime trade has continuously expanded in the last three centuries. The decline of real costs of shipping explain its development. States deprived of sea coast could consequently be victims of this trend because inland haulage transport cost has not diminished at the same pace. Limao and Venables (2001) find that it is more than seven times more expensive to transport goods by land than by sea.
Why is overland transport more expensive than maritime transport? A plausible explanation concerns the border-crossing obstacle. Unlike coastal states, land-locked states have no maritime borders, but only land borders. UNCTAD emphasises the real cost of border-crossing for land-locked countries. Beilock et al. (1996) calculated that each border-crossing within the FSU region increases truckload freight rates by 400 USD.
All the freight forwarders in Kazakhstan (local and international) (9) found that the cost of border-crossings is the main impediment to trade in this region. Corruption, excessive documentation, inadequate border infrastructure and excessive delays all increase the costs finally charged to customers.
Moreover, within national borders there are 'internal borders' that increase the costs for transport companies. The director of the main foreign freight forwarder company in Kyrgyzstan (10) explained that any crossing of oblast borders in Kazakhstan requires a payment between 50 and 100 USD for any Kyrgyz truck in transit towards Russia.
While it is difficult to estimate the total cost for land-locked countries of border-crossing, they are a severe impediment to trade. For instance, freight in transit through Kazakhstan has dropped by more than 90% (in comparison with 1992) (cf. Figure 1). The production collapse is one explanation, but transport professionals attribute the decline to the dramatic increase of transport costs (eg new tariffs, more corruption and infrastructure decay).
[FIGURE 1 OMITTED]
In any case, the Central Asian example demonstrates that land-locked states are heavily dependent on border-crossing. Hence, multiplying border-crossings can be a factor that explains the impact of being landlocked on trade, especially among developing countries of those in transition.
DETERMINANTS OF THE IMPACT OF BEING LANDLOCKED ON TRADE
In economics, the main tool to evaluate trade levels for land-locked countries is gravity equations. The first estimations date back to the early 1960s. Tinbergen (1962) estimated the potential bilateral trade level between 42 countries, based on the function of their 'size' and 'distance'. In order to make these estimates, he used GDP and bilateral distances as proxies. Linnemann (1966) proposed to explain more theoretically the robustness of this relation between trade and distance. Today, the estimation of gravity equations is at the centre of applied economics regarding trade issues.
Consequently, striving to estimate the impact of being landlocked on trade and the determinants of these effects, we chose to test the following benchmark equation:
(1) ln IM[P.sub.ijt] = [alpha] + [delta] ln [Gdp.sub.it] + [phi] ln [Gdp.sub.jt] + [lambda] ln [Dist.sub.ij] + [xi] ln [Instit.sub.it] + [psi] ln [Instit.sub.jt] [tau] ln [Access.sub.i] + [kappa] [Landlock.sub.ij] + [omega] ln [Infra.sub.i] + [iota] ln [Infra.sub.j] + [rho] ln [Tariff.sub.it] + [[epsilon].sub.ijt]
Definition of the variables
In order to have accurate estimations of the impact of being landlocked on trade, we used control variables representing trade openness, institutional development and infrastructure level.
Al-Atrash and Havrylyshym (1998) point out the positive link between openness, reform and trade in the case of economies in transition. Consequently, we strived to introduce an appropriate variable for trade openness. Unfortunately, the Trade Openness Index calculated by Lawson, Gwartney and Skipton (www.freetheworld.com) cannot be used because of missing variables (approximately half of the sample countries were not included in the study). Consequently, the only variable reflecting trade openness is tariff duties. [Tariff.sub.i] is a measure of the mean tariff duties on imports in percentage (unweighted). Overall, it is difficult to collect consistent protection data on Central Asian countries and countries in transition, but the most important trade impediment in this regard in Central Asia is most probably non-tariff duties or non-conventional protection.
It is important to avoid Summers, called the 'tyranny of geography'. He warned against concluding that 'the economic failures of isolated, tropical nations with poor soil, an erratic climate and vulnerability to infectious disease can be traced simply to the failure of governments to put in place the right enabling environment'. (11) As Anderson and Marcouiller (1999) argue, it is crucial to introduce a variable that reflects institutional capability as well as geographic factors to understand correctly the negative impact of being landlocked. The Keefer and Knack (1997) indicators would have been useful to analyse, but unfortunately their samples were limited. We have, therefore, chosen to use the Coface indicators (see Appendix B for details).
Recently, several articles have clearly demonstrated the positive impact of transport infrastructure on trade (cf. Bougheas et al., 1999; Limao and Venables, 2001). A composite index has been calculated for all sample countries that examine the importance of paved roads and railway networks. This variable is called [infra.sub.i] or [infra.sub.j] (for country i and country j).
We also included another geographical variable: access to a large market. The remoteness or location index has been used in several studies in the past few years (cf. Polak, 1996; Helliwell, 1997; Smarczyinska, 2001). Following Helliwell (1997), Fujita et al. (1999), we use the mean distance weighted by partners' GDP. In this paper, following Gallup et al. (1999), a variable representing the market access disadvantage is introduced. Our variable, [Access.sub.i], is the shortest distance (in kilometres) (12) between country i and one of the three major markets: European Union, Japan of the US.
The major interest of this paper is to measure the impact of being landlocked on trade using four measures of being landlocked.
Successive estimations have been hypothesized. Firstly, a dummy variable has been introduced for trade relations between two land-locked countries. (the variable equals 1 if both countries are landlocked; 0 otherwise.)
The second variable is [ShortD.sub.i] or [ShortD.sub.j], the shortest distance between a country and the nearest major port facility. One factor that could help explain the differences of trade between land-locked countries in Europe and Asia is the distance of a country (measured from the economic capital) from a port. Indeed, Switzerland or the Czech Republic is 500-600km away from ports like Hamburg or Genoa, whereas Almaty (Kazakhstan) of Bishkek (Kyrgyz Republic) is more than 3,000 km by road from Saint Petersburg of Karachi. This variable is the shortest distance (in kilometers, by road) ffom the main economic city to the closest major port facility. Using the distance to port facilities, Doumenge (1986, pp. 21-22) has developed a system of classification, which comprises four classes ranging from countries where "landlockedness is bearable' (distance from a port of less than 500 km and access to two major port facilities) to 'excessively land-locked countries' (distance more than 2,000 km). Even among land-locked countries, especially large land-locked countries like Kazakhstan, the distance variable can affect regions differently. For example, the Mangistau region is more than 1,000 km closer to Russian port facilities than is the Almaty region. It is difficult to illustrate the results using this database (based on country data), but this variable emphasised the importance of the real geographical distance.
A third measure analyses the ability of a land-locked country to bargain. [Coastnb.sub.i] of [Coastnb.sub.j] represents the number of borders with coastal countries. It is critical for a transit country to attract transit trade flows from land-locked states. If a monopoly exists, the landlocked country has no bargaining power, which has a negative impact on trade. For example, Turkmenistan can only rely on Iran to develop trade outside CIS space.
The fourth measure is the number of national borders crossed to trade goods. This factor has not been estimated in other gravity-type papers, but the cost of crossing borders can explain why overland transport is more expensive than maritime transport. In fact, a major difference between maritime and overland transport is the multiple border-crossings.
Statistical and econometric approaches
Unlike most studies, all the CIS countries have been included in our sample, notably all the CIS landlocked states (see Appendix A). The sample consists of 46 countries, of which 18 are landlocked. African landlocked countries have been excluded because all the existing surveys of this issue have been dedicated to these countries (UNCTAD, World Bank) (this phrase is unclear to me), with the exception of Asian countries (except Nepal, Laos and Mongolia). Nevertheless, the disintegration of the USSR, followed by the Czechoslovakia split and the end of the Yugoslav Federation produced 11 additional land-locked countries, which increased their population in the world by one-quarter. Estimations have only been conducted over a period of 5 years (1995-1999) due to serious data inconsistencies before 1995 for CIS countries. The total number of potential observations is 10,350. Owing to statistical problems with tariff data, the number of observations has been reduced to 5,522 when [Tariff.sub.i] is introduced.
As Egger (2000) suggested, most studies related to gravity-type estimations have been cross-section, (13) as employed by Cheng and Wall (1999) and Wall (2000).
The estimation is carried out in two steps:
(1) A gravity model with fixed effects is estimated.
(2) ln IM[P.sub.ijt] = [alpha] + [delta] + ln [Gdp.sub.it] + [phi] ln [Gdp.sub.jt] + [xi] ln [Instit.sub.it] + [psi] ln [Instit.sub.jt] + [rho] ln [Tariff.sub.it] + [[epsilon].sub.ijt]
(2) OLS is used with the estimated fixed effects regressed on the geographic variables.
(3) [[lambda].sub.ij] [equivalent to] a + d ln [Dist.sub.ij] + b ln [Access.sub.i] + c [Landlock.sub.ij] + e ln [Infra.sub.i] + f ln [Infra.sub.j] + [[epsilon].sub.ij]
This procedure makes it possible to estimate trading-pair effects and gives results for the time-invariant variables, which is not the case for the fixed-effects models.
In the first step, two estimations were conducted: with and without tariff data. Owing to statistical problems, the estimations in stage 2 were carried out without [Tariff.sub.i]. The second step tests the four measures of being landlocked; and the last estimation is a combination of these measures.
As shown in Table 2, the four measures of being landlocked, which include institutional and economic factors are statistically significant, which means that being landlocked has a high negative impact on trade. According to the estimation with a dummy variable, land-lockedness reduces trade by more than 80%. (14) This result is close to that of Limao and Venables (2001). The distance between the importing country and major port facilities is also a significant factor affecting trade-level differences among land-locked countries Cf. estimation II). This factor contributes to the explanation of why Central Asian land-locked countries are far more crippled by their geographic location than European land-locked countries. With respect to measures of being landlocked, database analysis confirms the importance of the bargaining power of land-locked states. Indeed, estimation III confirms that the greater the number of a land-locked country's options, the more the landlocked country imports. For example, Switzerland, Austria, the Czech or Slovak Republic in Central Europe have more bargaining power than Central Asian land-locked countries. European countries have opportunities to bargain for reduced transit costs. Estimation IV illustrates that the number of border-crossings for trading goods has a negative impact on trade.
Estimation V combines the four measures in order to define the main determinants of the negative impact of being landlocked. From the two regressions, it appears that the border-crossing issue is the most robust measure. Indeed, when the four measures are included, only [landlock.sub.ij] and [bordercros.sub.ij] estimations remain statistically significant (with the expected sign). Moreover, [R.sup.2] is the highest when [bordercros.sub.ij]) is introduced in the estimated equation (it is going from 0.3 to 0.42). The database analysis shows that multiplying the number of border-crossings is equally as destructive on trade as is the distance factor. This result helps us understand why landlocked developing countries trade little in comparison to coastal developing countries.
The last geographic variable, the remoteness of land-locked countries from major markets, is important (around -0.4), statistically significant and characterised by a negative sign, which means that the further you are located from major markets, the lower the trade flow will be. It confirms that a state that is landlocked at the core of Europe is not equivalent in terms of trade level to a land-locked state in the heart of Central Asia due to the fact that Czech Republic, for instance, is a part of the European Union market, which is obviously not the case of Central Asian economies. In order to explain bilateral trade level, it is consequently advantageous to examine the transport costs component thoroughly.
Moreover, the infrastructure variable has a positive impact on trade, thus confirming the Limao and Venables (2001) results. It highlights the mitigating effect of good transport infrastructure for land-locked developing countries.
Finally, the estimations confirm the existence of gravity-type relationships (with distance and gross domestic product as variables). In most of the cases, the institutional variable is not statistically significant. The irrelevance of the variable is most probably due to data inconsistencies, but as demonstrated in Appendix B, it is extremely difficult to obtain data for all the countries of this sample regarding this issue. Moreover, [bordercros.sub.ij] can capture a part of the institutional deficiencies.
The conclusion is that the four tested measures, which take into account institutional and economic factors on land-locked countries, are robust, which means that being landlocked has a major negative impact on trade. The purpose of this paper is to demonstrate that the effect is closely linked to geographical location, distance from major markets, main trade flows and main hub [airports or ports) facilities measured by additional transport cost and the number of border-crossings. It is obvious that the institutional quality has an impact on the border-crossing issue. Nevertheless, it is also related to the geographical location. Indeed, from the heart of Central Asia to Western Europe, a truck has to cross more than five borders. With regard to the Central Asian case, studies before 1992 had not taken into account the economic disadvantage of the geographic location of land-locked countries in the Central Asian region. Countries, which were part of the Soviet Union, were not affected by the transport cost burden. In the Soviet economic system, transportation was never considered to be a cost factor, but was accepted as a by-product of the politically determined production and consumption structure. Hence, resource allocation decisions ignored transport costs. With the end of the Soviet system, however, Central Asian states were again at an economic disadvantage: they lacked infrastructure and proximity from major economic markets. After the disintegration of the USSR, Central Asia faced the consequences of being landlocked and felt the negative impact on trade and future development.
(1) Land-locked countries (18)
Europe: Armenia, Azerbaijan, Moldova, Belarus, Switzerland, Austria, Luxembourg, Hungary, Czech Republic, Slovak Republic.
Asia: Kyrgyz Republic, Uzbekistan, Kazakhstan, Tajikistan, Turkmenistan, Afghanistan, Nepal, Mongolia.
(2) Island countries (10)
Fiji, Vanuatu, Papua New Guinea, Tonga, Samoa, Solomon, Maldives, Sri Lanka, Mauritius, Indonesia.
(3) Partner countries (18)
Europe: European Union, United States, Russia, Georgia, Ukraine.
Asia: Turkey, China, Iran, Pakistan, India, Japan, South Korea, Singapore, Thailand, Malaysia, United Arab Emirates.
Oceania: Australia, New Zealand.
Import ([IMP.sub.ijt]): IMF direction of Trade Statistics. Baldwin (1994) explains that in the case of zeros in the trade data, three approaches to the problem have been used: (1) discard all zero flows, (2) substitute small values, (3) use estimating techniques like Tobit. The third solution is the right approach, but, with panel data the Tobit procedure is extremely difficult to use and it has been demonstrated that 'the resulting estimates are not substantially affected by the choice of approach' (see Baldwin, 1994, p. 85). Consequently, in the case of zero import data, export figures from partner countries have been used. When the statistic was still missing, an arbitrary 0.1 was integrated in the database. (A figure less than 1 is generally used in the gravity literature.) More than missing data, we did face an absence of trade between, for instance, some Central Asian countries and East European or South Pacific States because their export base is very weak.
Gross domestic product ([Gdp.sub.i] and [Gdp.sub.j]): World development indicators for gross domestic product (at purchase power parity).
Distance ([Dist.sub.ij]): This is always controversial. With a large sample spanning three continents, a simple distance measure is more acceptable. This is what has been used in the current study. Bilateral distances are extracted from the website, at www.indo.com/distance, using polar coordinates for unreported countries (Samoa, Tonga and Mongolia). Capital cities have been selected, except when the economic strength of the capital city was economically negligible vis-a-vis the main city (for instance, New York and Washington in the USA, Sydney and Canberra in Australia, Karachi and Islamabad in Pakistan, and Istanbul and Ankara in Turkey). In case of the European Union, Brussels was chosen as its economic centre.
Tariff ([Tariff.sub.i]): Mean tariff duties vis-a-vis imports as a percentage (unweighted). Data have been collected from different international sources (World Bank, WTO, OECD), but many observations remain incomplete (half of the sample).
Institutions ([Instit.sub.i] and [Instit.sub.j]): The selected proxy is the Coface indicator (@rating, for the last ratings, available at www.cofacerating.fr). For this indicator, country risk is measured on a scale from Al to D, by the public credit insurance credit office. It was previously measured on a scale from 1 to 6. The former scale has been selected. Foreign currency shortcoming risk, a state's capability to respect its commitments vis-a-vis foreign creditors, and indicators of payment problems are included in this indicator along with political factors. Hence, it is more global than a simple institutional indicator. Its main interest to this survey resides in the fact that it is available for all the sample countries.
Remoteness ([Access.sub.i]): The shortest distance (in kilometres) between country i and one of the three major markets: European Union, Japan or the US. For the source, see distance data.
Infrastructure ([Infra.sub.i] and [Infra.sub.j]): A composite index that has been calculated for all sample countries. It combines the importance of paved roads and railway networks. The first two categories come from the CIA World Fact Book (available at www.cia.gov/cia/publications/factbook). The total infrastructure stock is defined as the sum of the two categories. In order to take into account the difference between small and large countries, this sum has been divided by the total population. This index is very close to the one calculated by Limao and Venables (2001).
Landlockedness ([Landlock.sub.ij]): It is a dummy variable capturing trade relations between two land-locked countries; equal 1 if both countries are landlocked, 0 otherwise. It is a traditional variable used in the literature to capture the effects of being landlocked.
Shortest distance from a port facility ([ShortD.sub.i] and [ShortD.sub.j]): It is the shortest distance (by road, in kilometres) between a land-locked country and the nearest major port facility. Data are from the Encarta. A factor of zero has been arbitrarily assigned to coastal and island countries, assuming that the main economic centre is located on the coast, which is the case for most of the countries sampled.
Number of coastal neighbouring countries ([Coastnb.sub.i] and [Coastnb.sub.j]): This variable ranges from 0 (Uzbekistan) to 5 (Belarus and Hungary). Considering that coastal states have a very high bargaining power because they have an open access to sea, they have been given a value of 5.
Number of national borders crossed ([Bordercros.sub.ij]): It is the number of national borders crossed from one importing country from the supplier country. In the case of trade between two coastal countries, it has been considered as the crossing of one border. For trade flows between European and Central Asian countries, routes usually used by trucks and trains have been selected, that is, through Russia, Belarus, Poland, etc.
Table 1: Comparative costs of shipping a container from the USA to Central Asia Country of City of Port of entry Cost in destination destination USD Kazakhstan Almaty St. Petersburg 12,000 Kyrgyzstan Bishkek St. Petersburg 12,000 Russia Moscow St. Petersburg 6,000 Turkey Ankara Izmir 4,000 Country of Distance Price per km destination (in km) (in USD) Kazakhstan 10,490 1.14 Kyrgyzstan 10,478 1.15 Russia 7,828 0.77 Turkey 8,733 0.46 Source: Stone (2001, pp. 36-37) Table 2: Two-stage regression results Stage 1: gravity equation with trading-pair intercepts [Gdp.sub.j] 0.56 *** 0.25 ** (3) (2.44) [Gdp.sub.j] 0.51 *** 0.52 *** (3.14) (5.06) [Tariff.sub.i] -0.6 *** (-2.75) [Instit.sub.i] -0.13 ** -0.06 (-2.02) (-1.23) [Instit.sub.j] -0.25 *** -0.20 *** (-3.47) (-4.12) [R.sup.-2] 0.58 0.54 Observations 5522 10259 Stage 2: fixed effects regressed on geographic variables I II III [Dist.sub.ij] -1.28 *** -1.44 *** -1.13 *** (-44.18) (-50.19) (-41.72) [Access.sub.i] -0.40 *** -0.39 *** -0.39 *** (-34.35) (-34.82) (-33.63) [Infra.sub.i] 0.1 *** 0.11 *** 0.09 *** (10.04) (11.08) (9.08) [Infra.sub.j] 0.21 *** 0.23 *** 0.20 *** (23.37) (26.23) (22.24) [Landlock.sub.ij] -1.85 *** (-26) [ShortD.sub.i] -0.16 *** (-21.4) [ShortD.sub.j] -0.25 *** (-32.88) [Coastnb.sub.i] 0.53 *** (15.4) [Coastnb.sub.j] 0.76 *** (22.3) [Bordercros.sub.ij] Const 14.87 *** 17.59 *** 11.61 *** (55.6) (62.41) (47.44) Observations 10259 10259 10259 [R.sup.-] 0.3 0.34 0.3 IV V [Dist.sub.ij] -1.05 *** -1.14 *** (-43.23) (-40.38) [Access.sub.i] -0.37 *** -0.37 *** (-34.7) (-35.36) [Infra.sub.i] 0.13 *** 0.12 *** (14.46) (13.75) [Infra.sub.j] 0.23 *** 0.24 *** (27.93) (29.02) [Landlock.sub.ij] [ShortD.sub.i] 0 (-0.52) [ShortD.sub.j] -0.09 *** (9.72) [Coastnb.sub.i] -0.04 (-1.21) [Coastnb.sub.j] -0.02 (-0.61) [Bordercros.sub.ij] -2.21 *** -2.02 *** (-53.19) (-35.87) Const 13.73 *** 14.88 *** (60.89) (52.19) Observations 10259 10259 [R.sup.-] 0.41 0.42 I: with Londlock, II: with ShortD, III: with Coastnb, IV: with Bordercros, V: with ShortD, Coastnb, Bordercros. Students ratio is given in Parantheses.
(1) I am grateful for the comments and suggestions from Gerard Duchene, Mathilde Maurel, Yelena Kalyuzhnova, Jeff Miller and Thierry Mayer. This paper was presented in May 2002 during the Second Spring school in economic geography in the University of Pau (France) and during the Second joint workshop ROSES-Centre for Euro Asian studies (University of Reading) in Reading (UK). I am also grateful to the French Ministry of Research for a grant, which enabled me to conduct a ground study in Kazakhstan and Kyrgyzstan in October-November 2002.
(2) Open access to the sea does not mean absence of seacoast. This distinction is particularly salient for Central Asian countries. Indeed, the Caspian is a salted sea. The canal Volga-Don links the Caspian Sea to the Black Sea. Nevertheless, one cannot consider this canal as breaking the landlockedness of Central Asian countries. This system of canals does not break economic isolation because Russian authorities are overcharging ships crossing the Volga-Don canal (13,300 USD), the Volga-Baltic (10000 USD), thus drastically limiting exports via inland waterways from Central Asian countries (cf. National Ports and Waterways Institute, 1996).
(3) Andorra, Liechtenstein, Luxemburg, San Marino.
(4) In this paper Central Asia is defined as post-Soviet Central Asia, which includes five countries: The Kyrgyz Republic, Kazakhstan, Tajikistan, Turkmenistan and Uzbekistan.
(5) A region whose surface drainage waters do not reach the oceans.
(6) Indeed, the average land area of high-income economies is around 720,000 square kilometres, 730,000 for middle-income economies, 520,000 for low-income economies and only 410,000 for land-locked countries. Author calculations are from World Bank data, calculated for 43 high-income, 91 middle-income, 64 low-income and 41 land-locked independent countries.
(7) See notably Micco and Perez (2002) and Martinez-Zarsoso et al. (2003).
(8) Moneta (1959) estimated that the transport ratio (transport costs divided by the value of the goods transported) was higher than 15% for cereals, coal and oil, and less than 1% for machinery and equipment.
(9) Opinions gathered from the main transport professionals during a ground study carried out in Kazakhstan and Kyrgyzstan in October-November 2002.
(10) Cf. meeting with Abdel Benyagoub, director Globalink Kyrgyzstan on 4th November 2002.
(11) Quoted in Hausmann (2001).
(12) The authors explain that they experimented a certain number of distance measures, all of which produced similar outcomes. Consequently, they chose the simplest. We make the same selection. However unlike these authors, it has not been weighted by population figures.
(13) For the panel use in case of gravity approach, we refer to Baldwin (1994), Polak (1996) and Matyas (1997).
(14) Impact = 1-(exp (-1.85)) = 0.84.
Al-Atrash, H and Havrylyshym, O. 1998: Opening up and geographic diversification of trade in transition economies. IMF Working paper no. 98/22, IMF: Washington.
Anderson, JE and Marcouiller, D. 1999: Trade, insecurity, and home bias: An empirical investigation. NBER working paper no. 7000, NBER: Cambridge, MA.
Baldwin, R. 1994: Towards an integrated Europe. CEPR: London.
Beilock, R, Boneva, P, Jostova, G, Kostadinova, K and Vassileva, D. 1996: Road conditions border crossings and freight rates in Europe and Western Asia. Transportation Quarterly 50(1): 79-90.
Bougheas, S, Demetriades, PO and Morgenroth, ELW. 1999: Infrastructure, transport costs and trade. Journal of International Economics 47(1): 169-189.
Cheng, I-H and Wall, HJ. 1999: Controlling for heterogeneity in gravity models of trade. Federal Reserve Bank of St. Louis working paper no. 99-010A, St. Louis: Federal Reserve Bank of St. Louis (http://www.stls. frb.org/research/wp/99-010.hmtl).
Debrie J, and Steck, B. 2001: L'enclavement, reevaluation theorique et application a l'Afrique de l'Ouest. L'espace geographique 30(1): 26-36.
Doumenge, F. 1986: Enclavement et developpement. ACCT: Paris.
Egger, P. 2000: A note on the proper econometric specification of the gravity equation. Economics Letters 66(1): 25-31.
Fujita, M, Krugman, P and Venables, AJ. 1999: The spatial economy. MIT Press: Cambridge, MA.
Gallup, JL, Mellinger, AD, and Sachs JD. 1999: Geography and economic development. International Regional Science Review 22 (2): 179-232.
Hausmann, R. 2001: Prisoners of geography. Foreign Policy, January/February, 45-53.
Helliwell, JF. 1997: National borders, trade and migration. Pacific Economic Review 2(3): 165-185.
Keefer, P, and Knack, S. 1997: Why don't poor countries catch up? A cross-national test of an institutional explanation. Economic Inquiry 35(3): 590-601.
Limao, N, and Venables AJ. 2001: Infrastructure, geographical disadvantage and traosport costs. World Bank Economic Review 15(3): 451-479.
Linnemann, H. 1966: An econometric study of international flows. North-Holland: Amsterdam.
MacKellar, L, Worgotter, A and Worz, J. 2000: Economic development problems of landlocked countries. IHS working paper no. 14 Institut fur Hohere Studien: Vienna.
Martinez-Zarsoso, I, Garcia-Menendez, L, and Burguet Suarez, C. 2003: Impact of transport costs on international trade: the case of Spanish ceramic exports. Maritime Economics and Logistics 5(2): 179-198.
Matyas, L., 1997: Proper econometric specification of the gravity model. The World Economy 20(3): 363-368.
Micco A, and Perez N. 2002: Determinants of maritime transport costs. Working paper no. 441, Inter-American development bank: Washington.
Moneta, C. 1959: The estimation of transport costs in international trade. Journal of Political Economy 67(1): 41-58.
National Ports and Waterways Institute. 1996: Feasibility study, Turkmenbashi port development. mimeo.
Polak, JJ. 1996: Is APEC a natural regional trading bloc? A critique of the 'gravity model' of international trade. The World Economy 19(5): 533-543.
Radelet, S and Sachs, JD. 1998: Shipping costs, manufactured exports, and Economic Growth. Mimeo.
Smarzynska, B. 2001: Does relative location matter for bilateral trade flows? An extension of the gravity model Journal of Economic Integration 16(3): 379-398.
Stone, JI. 2001: Infrastructure development in landlocked and transit developing countries: Foreign aid private investment and the transport cost burden of landlocked developing countries (UNCTAD/LDC/II2). UNCTAD: Geneva.
Tinbergen, J. 1962: Shaping the world economy. New York: XXth Century fund.
Wall, HJ. 2000: Gravity model specification and the effects of the Canada-U.S. border. Reserve Bank of St. Louis working paper no. 2002-024A, St. Louis: Federal Reserve Bank of St. Louis (http:// research.stlouisfed.org/wp/2000/2000-024.pdf).
GAEL RABALLAND (1)
ROSES, Sorbonne University, 106-112 Boulevard de l'Hospital, 75013 Paris, France. E-mail: firstname.lastname@example.org
|Printer friendly Cite/link Email Feedback|
|Publication:||Comparative Economic Studies|
|Date:||Dec 1, 2003|
|Previous Article:||Kyrgyz labour market in the late 1990s: the challenge of formal job creation.|
|Next Article:||Former socialist economies and the undergraduate curriculum.|