Presentation on theme: "Evaluating Trade Logistics Reforms Ana M. Fernandes, Russell Hillberry, Aaditya Mattoo DIME Workshop Paris France November 15, 2012."— Presentation transcript:
Evaluating Trade Logistics Reforms Ana M. Fernandes, Russell Hillberry, Aaditya Mattoo DIME Workshop Paris France November 15, 2012
Outline Why are we doing impact evaluation in trade facilitation? What trade facilitation measures may be undertaken? What are the measurement challenges? What can be done?
The (possible) benefits of trade facilitation Higher living standards Faster economic growth More trade Lower transportation costs Faster shipping times Trade facilitation measures 1)Introduce Single Window 2)Create Trade Information Portal 3)Introduce/improve risk management 4)Reduce documentation 5)Implement performance measurement for customs officials etc.
Central questions How big are the effects of specific trade facilitation reforms: – on shipping costs? – on shipping times? – on trade flows? Which reforms are most effective? Does the joint implementation of reforms have a more than proportionate impact? Does policy coordination across borders add even more? The goal of this project should be to answer one or more of these questions carefully.
Why is it important to answer these questions carefully and well? We would like to be able to give advice about which trade facilitation reforms are most effective and what outcomes each policy produces (lower costs, faster times, etc) Policymakers can then make better judgments about investments at the border, compared to other investments in logistics But also more broadly, policymakers can make better judgments about the merits of trade facilitation investments, relative to other uses of public funds (health, education, etc.)
What do we already know, and how do we know it? Higher living standards Faster economic growth More trade Lower transportation costs Faster shipping times Trade facilitation measures 1)Introduce Single Window 2)Create Trade Information Portal 3)Introduce Risk management 4)Reduce documentation 5)Implement performance measurement for customs officials etc….. ? ? ?
Trade facilitation and trade: evidence from cross-country studies Much of what we know about trade and logistics comes from big picture, cross-country studies. – Hoekman and Nicita (2008) Focus on logistics performance indices (LPI) and show that countries with better LPI trade more. – Estimates suggest that a 10 percent improvement in the LPI raises trade by 4.8 percent. – Freund and Rocha (2010) Focus on effects on African exports of delays in domestic transit, documentation, and customs and ports. Biggest effects appear to be in domestic transit times – A one day reduction in transit increases exports by 7 percent. Transit delays have a greater effect on reducing trade than delays in other areas
Limits of our knowledge from cross- country studies While useful, these studies do not tell us the direct impact of implementing a particular customs reform (or doing so in a particular place) – The estimates rely on having a global (cross-country) statistical model to control for other factors. This model may be wrong in some way. The variables that control for other factors may be incomplete. The effect of logistics on trade may differ across countries The effect of specific logistics reforms is unclear. The model estimates the effect of an improvement in logistics performance but does not say what reforms would improve logistics performance. It is useful therefore to turn to impact evaluation methods to isolate the impact of specific reforms.
Impact evaluation examples in logistics/customs: Yang (2008) The Philippine government staggered the introduction of pre- shipment inspections (PSI) on imports so that, for a period of time, PSIs were applied to imports from some countries and not to imports from other countries During the period where PSI affected only some imports, the Philippine government reduced minimum value thresholds in an attempt to reduce duty-avoidance activities – One can evaluate the effectiveness of the policy change by comparing the change in low-value imports from countries where PSI took place to the change in low-value imports from countries where it did not Results indicated that the change in policy shifted import activity from regular channels to export processing zones (EPZs), where imports get less scrutiny. – The implication is that the existence of EPZs gives another option to importers trying to avoid duties, and thus impedes the effectiveness of the reforms.
Impact evaluation examples in logistics/customs: Sequeira (2011) Assess impact of railway/port investments in Mozambique on export behavior – Three railway corridors in Mozambique, each initially laid in late 1800s Mozambique received WB funds to develop these corridors – Normally one worries that transport investments depend on the nature of activity at either end, but the pre-existence of these rail links assuages this worry. – Survey 600 Mozambican firms and 600 South African firms South African study – NE parts of South Africa are closer to the Mozambican port of Maputo than to South African ports – Use firms from other parts of South Africa as the control group, those from NE South Africa as the treated group. – Measure firms’ proximity to rail station. If new Mozambican rail line important, the impact of proximity to a rail station on exports (as well as other measures of firm performance) should have risen more among firms in NE than among other firms. This study is ongoing. Results not yet in.
Evaluation built into program design Evaluation not built into program design POSSIBLY TARGETED Improved risk management, Reduce documentation Experimental methods, i.e. randomized control trials, are usually technically feasible so long as some products, firms, can be left untreated. Use selective implementation or staggered implementation. Alternative: quasi experimental methods Quasi-experimental methods: Matching DID Instrumental variable Regression discontinuity design NON-TARGETED Port improvements, single window, trade information portal Randomization is difficult because it is difficult to treat some firms and not others. Quasi-experimental methods usually more suited. Non-experimental methods: detailed before-after comparisons are feasible. Difficult but may be possible to ex- post reconstruct a counterfactual. Targeting and built-in program design
Data availability DECTI has detailed data on exports coming from ASYCUDA or related systems. – E.g. for Albanian exports we have: Firm id, destination country, year, hs6, destination, export value, quantity – This kind of detail is useful but…. Need geographic detail in order to usefully study impacts on exporters Some other countries, Uruguay, for example, include port of exit, total time in customs, what channel the goods transited (red, green, etc) For important questions, it would be ideal to be able to link this with firm level data on employment, total sales, measures of investment, etc. The types of reforms considered here are mostly on the import side – Presumably there is ASYCUDA data on imports that is at least as good as what exists for Uruguayan exporters. – Value, quantity, port of entry, time spent in customs, channel(red green etc). – It seems possible (to me at least) that there might also be quite detailed data on the share of imports that followed each of the paths indicated on the following slide…..
Lessons for Western Balkans project design Need to identify necessary reforms ideally being able to implement them partially or in stages – Good candidates: Reducing documentation, implementing risk management, as noted in plan for Albania and Serbia – We need some more information about the status quo Need to identify treatment and control groups – Firms? Products? Border posts? Ports? – What data is available? Can we link detailed trade data with other data such as the geography of firm location and other firm characteristics.? Identify outcomes of interest – Value of trade, shipping times, measures of trade costs, productivity/employment of firms importing intermediates, etc. Consider whether we can implement more than one reform, in order to investigate their separate impact as well as their joint impact. – Possible candidate: Reforms at Durres for Albanian/Kosovo imports If Kosovo also implements reforms on these imports we have differential treatments as well as controls (imports through other ports and/or Kosovo imports through other partners
What links can we credibly make? Other micro-level outcomes: Productivity, employment for firms Prices, available varieties for consumers Trade outcomes: quantity, prices Proximate outcomes: costs, time Policy changes (various)