Abstract
This paper compares the performance of three competing corridors serving landlocked SADC countries (Beira, Dar es Salaam, and Durban) based on total economic cost from the perspective of transporters, retailers, and manufacturers. The motivation for the research is the paradox that, while Beira is closest to the hinterland served by these corridors, it attracts the least cargo. Historical research compares corridors in terms of both direct costs and time delays, but without translating time delays and variability in time delays into the economic costs experienced by corridor users. Unpredictable time delays reduce the competitiveness of cargo owners forming part of global just-in-time value chains. Our novel TEC model includes direct costs and the cost impact of delays and variability in delays and quantifies the relative contributions of ports, border posts, and road travel. The Port’s efficiency proved to be the biggest differentiator between these corridors, followed by border posts and road links. We found that while the Beira corridor has the lowest cost if only average travel time is considered, the Durban corridor proves to be the most competitive when variability in time delays is also considered, explaining why Durban enjoys the largest share of cargo transported to the landlocked hinterland.
1. Introduction
The landlocked countries of the South African Development Community (SADC), comprising the Eastern Democratic Republic of Congo (DRC), Malawi, Zambia, and Zimbabwe, are served by three intermodal corridors, the Beira Corridor, Dar es Salaam Corridor, or North-South Corridor (with Durban as port), displayed in Figure 1. When choosing between routes users consider time efficiency and cost-effectiveness of the corridor as of paramount importance. According to Hanaoka et al. [1], many landlocked developing countries (LLDCs) encounter difficulties importing and exporting goods because cargo must pass through land borders and lengthy roads or railways to access seaports located in transit countries (TCs), forcing LLDCs to endure higher logistics costs, and longer transit times compared with TCs, making them less competitive in the global context.
Various studies have reviewed the development of multinational transport corridors and assessed their performance [1, 2]. Barriers to cross-border movement exist because of lengthy customs procedures, while inefficiencies in the transport system have an adverse impact on the economic development of LLDCs [3]. From an economic cost perspective, it would be expected that the proximity of competing ports to the hinterland should determine the share of cargo moving on each corridor between the hinterland and the seaboard. This is, however, not true for this case study, while Beira Port is by far the closest to the primary economic hubs of the landlocked SADC countries, it also enjoys the smallest share of imports and exports. We investigate this paradox by developing a novel Total Economic Cost (TEC) model to explain the rational choices made by cargo owners and logistics service providers when selecting the most appropriate route to move cargo between the coast and the hinterland.
Hanaoka et al. [1] evaluated corridor performance from the perspective of infrastructure owners (ports, roads, and railways), considering the cost of infrastructure and direct transport costs, but without converting time delay and variability of time delay into their respective economic impacts on corridor users. We found that this approach could not explain the previous paradox, as both the cost required to provide and operate infrastructure, as well as the direct transportation cost is the lowest for Beira Corridor. Our study conducts the evaluation from the perspective of retailers and manufacturers importing goods as part of business operations. The three corridors are evaluated in terms of the cost impact of both total transportation time and variability in time delays. We, therefore, proceeded to quantify the hidden costs from the perspective of cargo owners, mostly resulting from variability in time delays, to explain the reasons for the observed share of cargo that each corridor enjoys.
The use of the TEC model provided evidence that the choices of the corridor made by cargo owners and their logistics service providers are indeed rational and based on maximising economic value for corridor users. The lack of providing for the impact of variability in time delays is thus proven to be a severe limitation in previous work that compared corridor performance. It is furthermore important to notice that the TEC model incorporates all the factors impacting corridor performance, including policy, regulations, infrastructure, and direct and indirect costs, since the total delays experienced by cargo transported along a corridor reflect the impact of all these factors.
The rest of the paper is structured as follows: in Section 2, we review the literature on intermodal transport corridor performance assessment. This is followed by the method for corridor assessment used for this paper in Section 3 and an outline of data collection in Section 4. Section 5 describes the theory behind the TEC model that forms the core of the method to compare different corridors from the perspective of corridor users. The quantified results are reported in Section 6, while conclusions and policy recommendations for the improvement of respective corridors are addressed in Section 7.
2. Literature Review
Transport corridors can provide an answer to LLDCs’ poor accessibility to resources and markets via maritime transport. According to Rodrigue, et al. [4] a transport corridor is a linear orientation of one or more transport routes and flows connecting important locations that act as origins, destinations, or points of transshipment. Reference [5] defines a corridor as an international intermodal transport route that can expedite the movements of goods and people across international borders by connecting key freight transport points in different countries. International cooperation is essential to provide transit access and efficient transportation systems for landlocked countries [6].
The level of development of the logistics industry in a country as reflected by the Logistics Performance Index (LPI) has much bearing on the overall efficiency of transport processes. According to the 2018 LPI scores, among the four SADC LLCs along the case study corridors, Malawi (2.59) and Zambia (2.53) have higher LPI compared to their two counterparts, DRC (2.43) and Zimbabwe (2.12). Among the countries operating the respective ports, South Africa has the highest LPI of 3.38, with Tanzania at 2.91 and Mozambique at 2.24. This provides an indication of the condition of infrastructure, procedural impediments, and intermodal transport operation and management along the respective corridors.
Several studies have cited problems related to customs clearance and delays at border posts as a major bottleneck in the transport process along most corridors. The World Bank [7] found that on the Almaty-Europe corridor, more than 50% of transit time is lost in downtime at borders. In a similar study, Walker et al. [8] identified bottlenecks restricting the use of intermodal freight transport linking Western Europe with Central and Eastern Europe and analysed various policies, and prioritised those for reducing bottlenecks. Reference [9] claimed that the increase in logistics cost and transit time and border-crossing problems are some of the reasons for the trade imbalance and low trade volumes between Central Asia and Europe. Islam et al., [10] assessed impediments to the development of efficient multimodal transport in Bangladesh and recommended the reform of customs procedures to improve operational efficiency on corridors.
Lehtinen and Bask [11] found that while theoretical studies indicated a particular corridor as being favoured by importers, this was not the case in practice. According to Regmi and Hanaoka [6], the time-cost-distance approach is extensively used by the Economic and Social Commission for Asia and the Pacific and Asian Development Bank for the assessment of transport operations. They also noted the contributions of emerging ICT technology for efficient intermodal transport operations and freight transport. A decision support tool, including a cost model, was developed to assist logistics service providers to select optimum multimodal routes, in the process optimising transportation routing within GMS countries [12].
Several other studies, especially Athukorala and Narayanan [13], Yang et al. [5], Hanaoka, et al. [1], and Goldmann and Wessel [2], combined the evaluation of international transport processes with trade and transport facilitation measures. Jiang (2019) used a simple time-cost-distance approach to evaluate intermodal transport corridors in North-East and Central Asia and identified time and cost-related barriers. Yang et al. [5] built a performance evaluation model to understand the relative performance of the Traditional Sea-Land Line (TSLL) alongside the two emerging container routes forming part of the Belt and Road (B & R) initiative between China and Europe. Rodrigue [14] recommends the use of ICT, Radio Frequency Identification (RFID), and satellite positioning systems to reduce cargo processing and border clearance time as well as transportation cost.
The previous literature provides clear evidence that the competitiveness of a corridor is not only determined by its physical length, but is closely linked to the time delays for cargo that travel along the corridor. Such time delays can be attributed to three main causes such as ports, road links, and border posts.
The reasons for long port delays were the subject of earlier research [15] which found that it results from several factors as follows:(i)Customs authorities tend to subject 100% of import cargo to intrusive inspections as well as X-ray scanning; this results in delays of up to two weeks before cargo leave the port(ii)Ports either do not have electronic systems to issue invoices to and receive payments from importers, or they only allow local freight agents to access the electronic system, thus causing long delays for transit cargo handled by freight agents operating from LLCs(iii)Congestion in ports resulting from limited infrastructure capacity and suboptimal maintenance practices cause long delays for trucks visiting the ports
Long and highly variable delays related to road travel time can be attributed to the following causes:(i)The bad state of the road network, specifically along the Beira corridor, results in average truck travel speeds of as low as 10 km/h in some sections(ii)A large number of police stops along the corridors, specifically in Zimbabwe, Mozambique, and Tanzania results in unnecessary delays for trucks(iii)Weighbridges along the same route are not linked to each other; as a result, a truck that is legally loaded is often weighed several times during the same trip, resulting in unnecessary queueing times, in line with previous findings from other research [16]
Border post delays are mostly attributed to the high level of physical customs inspections that cause some trucks to spend more than 2 weeks at a border. The reason for the high rate of customs inspections can be linked to the lack of intelligent customs risk engines. Work in this field [17, 18] found that intelligent risk profiling may reduce physical inspections to a much smaller fraction of cargo without incurring higher compliance risks.
3. Methodology
The economic impact of logistics time delay performance on corridor users was the topic of previous research [3]. Hoffman [19] demonstrated that, in addition to service fees charged by transport operators, the time delay and variability in the time delay of transport operations result in various additional costs to the overall operations of the cargo owner. To objectively compare the Beira, Dar es Salaam, and North-South corridors we use a definition of TEC that includes direct transport and logistics cost, interest paid on cargo, stock shrinkage, cost of stockholding of cargo, and losses due to out-of-stock situations in case cargo is delivered late [20, 21]. This approach incorporates the impact of indirect cost factors such as infrastructure and regulations, as it quantifies the consequences of time delays and variability in time delays both on transport service providers as well as cargo owners, regardless of the underlying causes of such time delays.
Our methodology consists of the following steps, which are also explained in Figure 2:(1)We obtained figures of the average number of trucks crossing borders per annum for each of the corridors, using border posts that connect LLCs to the respective ports. This allowed us to determine the current market share of each corridor to service the needs of SADC LLCs. To simplify the comparison, we used Lusaka, Zambia as a proxy for all SADC LLCs, to determine the preference of commercial cargo owners that have to select a corridor based on economic considerations, as Lusaka is the biggest economic hub that is located approximately the same distance from Durban to the south and Dar es Salaam to the north.(2)We obtained actual figures of direct transport and logistics charges for a standard 40-foot container to be transported from either of the 3 competing ports to Lusaka. This provided an indication of the comparative costs that cargo owners are prepared to pay to have their goods transported along each of the competing corridors. Transport operators using corridors that are in a less competitive situation would have to charge less compared to transport operators using corridors that are more competitive in terms of nontariff-related factors. Should cargo owners be prepared to pay a higher direct charge for one corridor compared to the other, it would indicate that there are other considerations, not reflected by direct costs, that influence their decisions.(3)We then, constructed a model of direct transport costs based on distance and time, to determine what the charges for a 40 ft container from each port to Lusaka should be if the competitive playing field is equal for transporters operating along all corridors. We calculated direct transport costs by considering the cost to operate a truck as an economic asset over its economic life, the cost to employ the driver, and fuel costs, using the same approach as [3]. We assumed values for the different cost parameters as displayed in Table 1 to calculate the monthly cost to own and operate a truck. Using the total round trip time delay for each corridor we converted this into an ownership cost per trip. We added to this the fuel cost, using the distances covered per corridor and fuel economy figures as obtained from industry data [22]. Typical monthly driver salary was converted to driver cost per trip by using the average trip turnaround time.(4)To determine the TEC from the perspective of cargo owners that must select a specific corridor, we constructed a model that incorporates direct charges from port to destination, interest costs on investment in cargo in transit, shrinkage costs, stock holding costs, and losses suffered by cargo owners should cargo not arrive by the time that buffer stocks have been depleted. The calculation of the latter cost required us to quantify the variability in delivery times as caused by variations in port, customs, and road transport delays.(5)To allow us to quantify the time variability of moving cargo from port to destination and vice versa, we extracted cargo turnaround time from actual port data and GPS tracking data collected from fleets of trucks using the respective corridors to transport cargo from each port to Lusaka. The time delays were divided into ports, border posts, and road links between these waypoints. As we had access to trip level data, we could calculate the statistics of delay times per individual waypoint and road link, as well as for the total corridor. This allowed us to calculate not only the contribution of the total corridor but also of each corridor element, more specifically ports, border posts, and road links. These results enabled us to identify which of the factors forming part of corridor performance have the biggest impact on costs as experienced by cargo owners.(6)The costs to the cargo owner resulting from variability in delivery times are impacted by the buffer stock policy implemented by the cargo owner. To obtain realistic cost figures, we assumed that the cargo owner is rational in terms of minimising total costs and that buffer stock levels are optimised to minimize total costs. Higher buffer stock levels will result in higher interest charges and higher stock holding costs but will reduce losses due to disrupted operations, and vice versa for lower buffer stock levels. We, therefore, calculated total costs for a wide spectrum of buffer stock periods (the period that buffer stock will last before it is depleted should there be no new stock deliveries) and used the buffer stock periods that coincide with the lowest total costs. Higher variability in transport and logistics delays will result in policies that use higher buffer stock levels, and therefore also higher total costs, compared to a case where the variability in time delays is smaller. As the optimal buffer stock level is in each case determined based on actual historical corridor time delays, we can accept that the calculated costs reflect the true impact of corridor operations on costs and the resulting decisions of rational corridor users that try to minimize their costs.
4. Data Sets
The following datasets were used to support the methodology:(1)Truck traffic volumes per border as measured as part of border traffic surveys conducted during the period 2016 to 2018 [23]. This information included verification of whether each truck delivered its cargo in the country of import (e.g., cargo entering Zimbabwe from Mozambique and remaining in Zimbabwe) or whether it carried transit goods destined for a different country (e.g., cargo entering Zimbabwe from Mozambique but destined for Zambia). These surveys were conducted according to the WCO methodology for border surveys over a period of 7 days in each case, counting all traffic moving through the respective border in both directions.(2)Transport charges for each corridor were collected from a representative set of freight forwarders carrying goods along the 3 corridors to the hinterland [24].(3)Ports time delay data collected from the respective corridors [24–26]. These delays were partly obtained from port annual reports and partly from actual port data extending over periods of one to two years.(4)Truck GPS traffic data collected for representative sets of trucks travelling across the respective corridors during the period 2016–2019 [24, 26]. The truck fleets used in the study contained between 400 and 1,600 trucks per corridor.(5)Estimated costs to the operations of retailers and manufacturers resulting from logistics time delays [27]. Estimates of cost parameters were made based on surveys conducted within the customer base of several SADC-based freight forwarders.
It was assumed that the available data sets are representative of corridor behaviour in general.
5. Total Economic Cost Model
The equations appearing in this section indicate how the various cost parameters were used in the calculation of TEC.
5.1. Direct Costs
Tables 1 and 2 display the cost parameters that were used in the cost calculations; these figures were obtained through discussions with industry practitioners and reflect costs that are typical for the SADC region.
The following equations describe how direct costs were calculated as follows:where is the round trip delay in days for the i-th corridor and is the trip delay from origin to destination. Four days are added to the travel time to provide for the loading and offloading of cargo, refuelling, and maintenance of vehicles.where is the driver cost per trip for the i-th corridor and is the monthly cost to employ a driver.where : fuel cost for corridor i, : distance from the origin to destination for corridor i, is the cost of fuel per litre, and is the fuel economy in km/litre.where is the direct trip costs, CBE is cross-border expenses (e.g., road taxes), and OC is the other costs (e.g., subsistence for the driver).where is the number of trips per month for the i-th corridor.where is the total cost per month per truck for the i-th corridor, and is the monthly installment per truck.where is the total cost per trip for the i-th corridor.
The transport cost for each corridor was then calculated as a fraction of the cost of cargo for both retail and manufacturing.
5.2. Costs Resulting from Variable Time Delays
While some time-dependent costs, e.g., interest paid on a stock in transit, can be calculated using only the average value of cargo time delays, that is not the case for other time-dependent costs, such as loss in sales or production due to depletion of buffer stocks. By implementing a specific buffer stock level policy, the cargo owner will try to minimize the overall cost of his operation. As this policy remains constant after optimal buffer stock levels have been determined, it may still happen for each cargo delivery that an above average delay time will lead to a stock-out situation at the retailer or manufacturer, which will result in economic losses. To determine the actual TEC for either a retailer or manufacturer, it is, therefore, necessary to calculate the costs for the total spectrum of possible time delays, taking into consideration the likelihood of each possible time delay. Losses will be suffered only for those cases where the cargo time delays exceed the buffer stock period. The use of the standard deviation or coefficient of variation will therefore not allow the accurate calculation of the relationship between time delays and resulting losses.
To allow us to accurately calculate such costs only for cases where a cost will indeed be incurred, we calculated the percentiles of time delays for each corridor and then calculated the expected TEC for each percentile. By integrating these contributions, the total expected cost for all cargo deliveries can be calculated based on the measured spread of time delays.
The following indirect costs were identified from the perspective of retailers and manufacturers importing cargo as part of their operations, using the same approach as [19]:(1)Impact of varying time delays: time-dependent costs increase with the increase in the period that stock is in transit. As not all cargo takes the same time to move along a corridor, the total cost should be calculated by integrating all possible time delays as follows: where is the cost incurred for time delay t, and is the probability distribution for all possible time delays. As however, we do not know the true probability distribution of time delays, the best we can do is to take the average over all percentiles for stock time delays as follows: where is the cost incurred corresponding to the i-th percentile of time delays. This was applied to all time-dependent costs calculated. For simplicity, the summation is not explicitly shown in each case.(2)Interest paid on investments in stock-in-transit from origin to points of consumption, calculated as a fraction of the value of cargo: as the importer must pay for goods once it is shipped on board at the port of origin, the importer must invest in stock-in-transit for the time duration as from cargo being shipped until final delivery to a retail distribution centre or manufacturing plant. This cost can be expressed as a fraction of the value of the goods as follows: Therefore, where(3)Shrinkage of stock in transit: as shrinkage losses increase with the increase in the period that stock is in transit, the total shrinkage was calculated by summing over all percentiles for stock time delays as follows:where is the total shrinkage for all stock, including all possible time delays, is the shrinkage for the i-th percentile, is the daily shrinkage fraction, and is the time delay in days for the i-th percentile. The equation above takes into account that after each day in transit, there is less stock left that is still exposed to further shrinkage.(4)Losses in sales or production, should an out-of-stock situation occur, as a fraction of the value of cargo: these losses will occur if the actual delivery is delayed beyond the normally expected delivery time. If a buffer stock is maintained to prevent these losses, then an actual loss will only occur if the unexpected delay is longer than the period covered by the buffer stock [19]. Manufacturing operations will be differently impacted by cargo delays compared to retail operations for various reasons. First, each manufactured product consists of many parts, each of which may experience different delays. For manufacturing operations, we, therefore, need to calculate the probability of a delay should any of the parts be delayed. Second, for retail operations, the cargo transported along the corridor represents the entire product, whereas for manufacturing operations there is further value addition by the manufacturer. If any part is delayed, it will impact a larger value that may be lost due to delays than just the value that is transported along the corridor. As a result, the ratio between the total cost of delays and the cost of value being transported requires a different formula, as follows:(i)Retail: for a retail operation it is assumed that losses in sales will occur when the buffer stock is depleted before the next delivery is made as follows: where where(ii)Manufacturing: for a manufacturing operation it is assumed that losses in production will occur when the buffer stock in any component used in production is depleted before the next delivery is made. We calculate the fraction of uninterrupted production runs, which are those production runs for which no component experienced an out-of-stock situation. Annual consumption is taken as the total annual value of components imported for use in manufacturing. As the value of gross margin on the total manufactured product is lost if there is a production loss, we also need to use the fraction that components represent the total product value. where n is the number of components used in the final product(5)Storage costs paid for buffer stocks, as a fraction of the value of cargo. Therefore, where
Equations (1)–(26) allow the total cost resulting from transport and other logistics delays to be expressed as a fraction of the total value of goods purchased as follows:where
We stated before that TEC will be calculated for the optimal value of the buffer stock period BSP where TEC reaches its minimum value. As the effect of BSP on TEC must be obtained by integrating over all possible values of actual time delays, and as we do not have an analytical expression for the probability distribution of actual time delays, it is not possible to find an analytical expression for the optimal value of BSP. Instead, we determine this value through numerical means, by firstly averaging TEC over the measured percentile values for total time delays, and by then repeating this calculation over a sufficient range of possible BSP values, allowing us to find the optimal value for BSP.
5.3. Sensitivity of TEC with Respect to Cost Parameters
The previous equations show that the TEC from the perspective of the cargo owner depends on several cost parameters. To investigate this dependence, we calculated the TEC while varying the values of these parameters over a range of values as indicated in Table 3. As components for products requiring a large number of parts are often ordered in batches including many parts supplied by the same manufacturer, we limited the maximum number of components to a reasonable amount.
6. Results
6.1. Division of Cargo between Competing Corridors
In Table 4, we show the average daily truck volumes that cross various border posts that link the 3 ports under consideration to Lusaka. As not all trucks crossing a border are destined for the same destination, we separately counted total and transit trucks. Transit trucks are destined to leave the country of entry with the same cargo.
In the case of the Dar es Salaam corridor, we assume that trucks crossing the Tunduma/Nakonde border post (from Tanzania to Zambia) and that are not transit trucks (to the DRC) are destined for Lusaka. In the case of the other corridors, trucks destined for Lusaka move through Zimbabwe, using the border posts of Forbes/Machipanda (from Beira) and Beitbridge (from Durban). We obtained the number of trucks moving through each of these borders to Lusaka by counting the number of transit trucks in each case. Using these numbers, we divided the number of trucks crossing the Chirundu border post from Zimbabwe to Zambia between the Beira and Durban corridors.
The results of this approximate analysis, based on daily truck traffic volumes as displayed in Table 4, are explained as follows:(i)400 trucks cross the Tunduma/Nakonde border between Tanzania and Zambia, of which 50 are in transit to or from the DRC; as a result, 350 travel on the Dar es Salaam—Lusaka corridor(ii)423 trucks cross the Forbes/Machipanda border between Mozambique and Zimbabwe, of which 208 are in transit to or from Zambia, and thus travel on the Beira—Lusaka corridor.(iii)1000 trucks cross the Beitbridge border between South Africa and Zimbabwe, of which 492 are in transit to Zambia, and thus travel on the Durban—Lusaka corridor(iv)700 trucks cross the Chirundu border between Zimbabwe and Zambia, of which 208 travel on the Beira and 492 on the Durban corridor(v)In total 1050 trucks thus travel to or from Lusaka, of which 208 (or 19.8%) travel on the Beira, 492 (or 46.8%) on the Durban, and 350 (or 33.3%) on the Dar es Salaam corridor
We, therefore, find that the Durban-Lusaka route has the largest share of Lusaka traffic, despite being the largest distance from Lusaka. There must be specific reasons why most commercial traders give preference to Durban port which is more than twice the distance from Lusaka compared to Beira. The rest of our analysis will try to explain this observation in terms of TEC to traders.
6.2. Direct Transport Costs
Table 5 displays the distance from each port to Lusaka, the direct transport cost based on the cost model for each corridor, the average fraction that direct transport cost represents of the value of the cargo itself, as well as the actual transport charges obtained from various freight agents using these corridors. Beira is approximately only 50% of the distance from Lusaka compared to the other ports and should, therefore, be in a very strong competitive situation if all other aspects of corridor performance are identical. This is confirmed by the fact that the direct transport cost for Beira represents a lower fraction of the average cargo value compared to the other ports.
We observe that freight agents operating from Durban charge their customers at relatively higher levels for the same service when compared to either direct cost or actual charges from Beira or Dar es Salaam. This suggests that the Durban-Lusaka corridor is more competitive in terms of the other factors incorporated into the TEC model. Application of the TEC model to actual time delay data will indicate if this can indeed explain the higher rates charged by operators on the Durban–Lusaka corridor.
6.3. TEC Model
We first calculated the average, standard deviation, and coefficient of variation (standard deviation divided by average) for each corridor. These results are displayed in Table 6. The high coefficient of variation of the Dar es Salaam corridor is caused by some very long delays experienced by a small fraction of cargo.
By combining the average time delay with corridor length, we could calculate the average speed of movement of cargo. The average time delays are not in proportion to the length of each corridor, with traffic moving much faster along the Durban corridor compared to the others. The low average speed of movement of cargo is partly due to long delays in ports and at border posts.
We then calculated the percentiles for time delays for the three corridors, and for each corridor, we separately calculated these percentiles for ports, road segments, and border posts. This allowed us to quantify the contribution of each of these elements towards the total costs as experienced by cargo owners. When quantifying the contribution of a specific corridor element to TEC, we assumed that only that corridor element showed variable time delays, while using average time delays for all the other corridor elements.
The percentiles are displayed in Figure 3. These graphs indicate how time delay gradually increases for higher percentile values, starting with the fastest moving cargo corresponding to the 1st percentile, up to the slowest moving cargo corresponding to the 100th percentile, with the median time delay corresponding to the 50th percentile.
It can be observed that for Beira and Dar es Salaam the port operations are the biggest contributor to delays, while for Durban, which is the furthest from Lusaka, road transportation represents the biggest delay. It is furthermore clear that there is a very wide spread of time delays experienced by different cargo consignments, from a few days up to more than 60 days. This confirms the importance of calculating the impact of variability in time delays on TECs to the cargo owner.
Figure 4 displays how different contributions towards logistics cost vary as a function of buffer stock period for both retail and manufacturing operations. As the buffer stock period increases the cost of interest and storage costs increase but the cost of lost sales and lost production is reduced.
Figure 5 displays how total logistics cost varies as a function of buffer stock. It can be observed that, as expected, the relationship between the buffer stock period and relative logistics costs is significantly different for manufacturers compared to retailers. There is a slower increase in cost with an increase in buffer stock period, as parts form only a part of the total cost of the product in the case of manufacturing, while purchased goods represent the complete product for retail.
The curves display an optimal value for the buffer stock period where the total cost reaches a minimum value. The rest of the TEC evaluations were performed by using this optimal buffer stock value for each corridor and for retail or manufacturing operations, respectively.
Table 7 displays the different elements contributing towards TEC from the perspective of retailers and manufacturers for all three corridors.
While the Beira corridor has the lowest direct transport cost, the cost of the total variability of corridor time delays is much higher for Beira compared to Durban. As a result, the TEC of transport and logistics between port and destination is higher for Beira than for Durban. This at least partly explains why Durban enjoys the highest share of this market despite being the longest distance from Lusaka.
When the contributions towards costs resulting from the variability of delay times are broken down, we observe that it is mainly the high variability in port delay times that causes both Beira and Dar es Salaam to suffer from higher TEC compared to Durban. While border posts and road links also play a role, it is not of the same magnitude as port delays. This result is similar for both retailers and manufacturers. The fact that Dar es Salaam port enjoys a higher fraction of trade than the Beira corridor, despite having a higher TEC, can possibly be attributed to the fact that it has more regular visits from vessels compared to Beira port.
In practice, the cost parameters displayed in Table 2 will not be the same for all retail and manufacturing operations. To test the sensitivity of our results for these parameter values, we repeated the cost calculations for a spread of values for each of these parameters. Some of these results are displayed in Figures 6 (varying component cost as a fraction of total cost) and 7 (varying inventory costs). While these parameters have a significant impact on total logistics cost, the behaviour for each corridor is similar across the range of values, and the ranking of relative cost remains the same for all three corridors over the entire range of values. Similar results were obtained for all the other cost parameters but are not displayed here due to lack of space. We can, therefore, state with certainty that the comparative results for the corridors will not be different if a different set of cost parameter values were used.
7. Conclusions and Recommendations
The objective of this work was to assess the ability of three corridors to compete effectively in serving LLCs. We addressed this by developing an integrated model for TEC as experienced by retailers and manufacturers, including both direct costs and indirect costs resulting from time delays and variability in time delays experienced by cargo moving through the corridors. We used Lusaka as a proxy for SADC LLCs that depend on the Beira, Dar es Salaam, and Durban ports for imports and exports.
The TEC model provides clear evidence that variability in time delays contributes much more to the total cost as experienced by cargo owners than direct transport costs. In the case of ineffective corridors, such as the Beira corridor, the costs resulting from variability in time delays can be as high as 80% of total corridor costs. This may cause corridor costs to exceed 25% of the cost of the goods being transported, which is much higher than the global average of less than 10%.
In line with this finding our results show that, although Beira port is the closest to Lusaka, it enjoys the smallest share of cargo traffic of the three competing corridors. This can be largely explained by the fact that the TEC for the Beira corridor is higher than that of the corridor from Durban. The primary contributor to this higher cost is the variability of delay times experienced through the port of Beira, compared to Durban port. Road link and border post delays also contribute towards the high TEC, although to a lesser degree.
Based on the previous findings we make the following policy recommendations to improve the competitiveness of the corridors serving SADC LLCs, and in the process the global competitiveness of the entire region:(i)Ports should improve the quality of their ICT systems to improve the efficiency of terminal operations and streamline the exchange of information with port stakeholders. Such systems should be online and accessible to freight agents operating from LLCs.(ii)Regulations for road transport should be harmonized for all countries along a corridor, to prevent avoidable delays at borders where trucks that were legally loaded on one side of the border become illegal once it crosses the border.(iii)Customs authorities should rely on intelligent customs risk engines to drastically reduce the rate of physical inspections without increasing the risk of reduced levels of compliance.(iv)The quality of the road networks should be improved to allow trucks to travel at reasonable speeds along the main road networks.(v)The network of weighbridges along the entire corridor should be linked to avoid the unnecessary repeated weighing of vehicles that are legally loaded.(vi)A single institution should be established to manage each transport corridor from end-to-end. Such an institution should have sufficient authority to enforce regulations and procedures that have been optimised from the perspective of the corridor and thus the entire economic region.
Future research should include the impact of vessel schedules on TEC to refine the model from the perspective of specific import and export industries. Such a refined model can be used to determine which of the competing corridors is optimal for each industry sector. It will also allow corridor performance to be optimised from the perspective of the primary industries served by that corridor.
Data Availability
The data used in this research is made available as a supplementary file to this submission.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
A shortened version of the material covered in this manuscript was presented at the Atiner Conference in Athens, Greece, in April 2022: see https://www.atiner.gr/abstracts/2022ABST-TRA.pdf. The material in this paper also formed part of the PhD thesis of Crynos Mutendera with Alwyn Hoffman as study leader: see https://repository.nwu.ac.za/bitstream/handle/10394/39668/CMutendera_Final%20examination_PhD_Thesis_accepted_Printing.pdf?isAllowed=y&sequence=1. Cross-Border Road Transport Agency through Memorandum of Agreement with North-West University.
Supplementary Materials
A data file containing the percentiles of time delays for each corridor that are the subject of this paper is included as Supplementary Material in this paper. (Supplementary Materials)