Export Promotion Agencies

Transcription

1 Daniel Lederman World Bank Marcelo Olarreaga University of Geneva and CEPR Lucy Payton Boston Consulting WTO and World Bank Workshop Conference on Export Diversification, Geneva May 20, 2009

2 Objectives Rest of the talk Justification for EPAs Market and coordination failures External economies of scale and scope in production Duplication of costs when collecting foreign market information Pioneer exporters and under provision of discoveries (Rodrik and Hausmann, 2003) Under provision of country image and reputation

3 Objectives Rest of the talk Case against EPAs Inefficiencies of the public sector (Hogan, Keesing and Singer, 1991) firms do not know about programs and do not respond to needs agencies are too burocratic: public sector incentives underfunded simple transfer from government to exporters they cannot beat the large anti-trade bias of the 80s

4 Objectives Rest of the talk Old debate Market failure versus government failure It s an empirical question Based on firm surveys and case study evidence in the early 1990s the balanced tilted towards government failures ITC kept on working on helping them work better Since then, very little evidence...and that s where we come in Exception is Rose (2007) on embassies

5 Objectives Rest of the talk What we are doing Survey EPAs behavior, activities, strategies Systematically explore whether they help promote exports What type of institutional design, activity and exporter targeting works better? What we are not doing Checking whether the intervention is socially desirable Or whether the return is sufficiently high to justify investing in EPAs rather than schooling, roads, etc. Relative modest objective but where the literature on externalities often stops

6 Objectives Rest of the talk Rest of the talk Empirical Model

7 Design Descriptive statistics Design Contacted 149 countries, 116 had an EPA and we sent our survey by 94 EPAs answered, 88 positively All survey data was collected over a 4 month period All responses were followed up with phone calls to double check and clarify answers Survey contained 18 questions over structure, financing, mission, activities and expenditure

8 Design Descriptive statistics Example of survey question

9 Design Descriptive statistics Descriptive statistics The objective of 60% of agencies is to promote aggregate exports. In 20% an important objective are non-traditional exports, and in another 20% they have sector objectives. Funding: nothing changed? Half of the agencies get more than 75% of their budget from public funds Only two get more than 75% of their budget from selling their services Very little donor support (25% of developing countries and less than 10% of total budget) But agencies did respond to earlier criticism

10 Design Descriptive statistics Share of executive board seats in the hands of the private sector

11 Design Descriptive statistics Do they work?

12 The empirical model Sample bias Endogeneity The empirical model The basic export equation is: ln(exp/pop) c = β 0 + β 1 lnbud/pop c + β 2 lngdp/pop c + β 3 lnt c + β 4 lnma c + β 5 lnvol c + β 6 lnreg c + β 7 lnf &R c + Dummies R + e c (1) And we are interested in β 1 We also introduce in (1) EPA s institutional design, and budget allocation to see what works best

13 Sample bias Motivation The empirical model Sample bias Endogeneity Probability of answering the survey was smaller in poor and small countries Thus we used the size of the country and Aid per capita in the selection equation as exclusion restrictions from the export equation We correct using a two-stage Heckman model, but also report OLS results which can be more efficient if: exclusion restrictions are highly correlated with exports per capita or if the sample bias is not too large (76% response rate).

14 The empirical model Sample bias Endogeneity Endogeneity Number of years since EPA was created (and the squared). Recently created are likely to be better funded than old EPAs, but also experience may be a factor (so squared term) Number of years to next election (and the squared). Close to election time, governments are likely to let spending go. But also if elections are a long time from now (so squared term).

15 Correcting for sample bias Correcting for sample bias Correcting for endogeneity Robustness checks OLS Heckman Export Eq. Selection Log of Budget per capita (ln Bud/pop) a ( ) ( ) Log of GDP per capita (ln GDP/pop) ( ) ( ) ( ) Log of Trade restrictiveness (ln T ) ( ) ( ) ( ) Log of Trade restrictiveness in ROW (ln MA) ( ) ( ) ( ) Log of Forex volatility (ln Vol) ( ) ( ) ( ) Days to comply with export regulation (ln Reg) ( ) ( ) ( ) Log of geo-trade/gdp ratio (ln F &R) ( ) ( ) ( ) Log of GDP (ln GDP) ( ) Log of Aid per capita (ln Aidc) ( ) Regional dummies b Yes Yes Yes λ d NA ( 0.24 ) ( 0.24 )

16 Correcting for endogeneity Correcting for sample bias Correcting for endogeneity Robustness checks 2SLS 2SLS No Heckman Heckman Log of Budget per capita (ln Bud/pop) ( ) ( ) Log of GDP per capita (ln GDP/pop) ( ) ( ) Log of Trade restrictiveness (ln T ) ( ) ( ) Log of Trade restrictiveness in ROW (ln MA) ( ) ( ) Log of Forex volatility (ln Vol) ( ) ( ) Days to comply with export regulation (ln Reg) ( ) ( ) Log of geo-trade/gdp ratio (ln F &R) ( ) ( ) λ NA ( )

17 Correcting for sample bias Correcting for endogeneity Robustness checks Robustness checks Controlling for exports/gdp at the time of EPA creation yields results that are not statistically different Interaction of budget with log of trade restrictiveness in ROW yields positive and statistically significant coefficient Interaction of budget with a dummy for homogenous goods yields a negative and statistically significant coefficient. EPAs seem to be particularly efficient when helping overcome barriers faced by exporters...

18 Correcting for sample bias Correcting for endogeneity Robustness checks Different impact for homogenous goods? Conservative definition Liberal definition Interaction homogenous with Budget ( ) ( ) Country dummies Yes Yes Product dummies Yes Yes R # of observations

19 Concluding remarks EPAs do seem to help exporters overcome barriers faced by exporters Some work better than others: private executive board publicly funded sectoral or non traditional export focus Strong and centralized agency Focus on non established exporters which are not necessarily small (or established)