 Good morning, everyone. My name is Neru Saleem Siba from the University of Gothenburg. I'll be presenting a work on learning by exporting in the case of Ethiopian manufacturing sector. And this is a joint work with Mulu Gabrielsus, who is working in UNU Merit in Netherlands. So the development strategy of Ethiopian government is called agricultural development late industrialization, where the government focuses on improving productivity of smallholder agriculture in the hope that industrialization will follow afterwards once this productivity improvements trickle down to the industrial sector in the way of increased demand to industrial goods. The industrial development strategy of the Ethiopian government also focuses on private sector development as engine of growth with particular focus on labor intensive and export oriented sectors. And the motivation of the selection of these sectors is motivated by mainly comparative advantage as well as exports are important source of foreign currency and dependable markets for value added agricultural goods. It's also a way of setting standard in terms of having internationally competitive industrial sector in Ethiopia. There are some studies showing that exporting is associated with output growth and also domestic industrial development. For these reasons, the government has preferential treatment for export oriented sectors such as textile and garment, leather and agro processing and construction and lastly micro and small enterprises in Ethiopia. And the government has intensified export promotion efforts, especially since 2005 and the nature of the export promotion involves providing direct support to export oriented sectors in terms of providing economic incentives such as cheaper credits and easier access to land and tax exemption of these sectors. Promoting export oriented cluster development as well as building capacity building in the supply of skilled manpower to these sectors. During this period, the number of firms has increased by about 70% as well as the number of exporters has increased by 40% since 2005 although only 4% of the firms export in Ethiopia and the total export is mainly concentrated in six major sectors that you see highlighted here. Parallel to this, we see also a shift in export destination of Ethiopian manufactured goods from developing countries towards developed world. So there is a reduction in the share of export to the developed world and its questionable whether firms, the scope of learning for firms is limited due to this shift in export destination. So the objective of this study is to investigate the effect of increased export engagement on enterprise performance in Ethiopia using a data set firm level panel data set collected by the Central Statistical Agency of Ethiopia and the data set includes all formal manufacturing firms employing at least ten employees. The contribution of the study is twofold we think. The first one is because we have longer panel data set we will be able to capture any long term effects of entering into the export market and capturing any long term production adjustment required to benefit from exporting. We also since our data set cover also the time period after the increased export promotion we will be able to capture any effect on the enterprise performance after increased export promotion by the government. So the previous two presentations has shown that there are many studies indicating a positive relationship between exporting and enterprise performance but the nature of causality is debated between two competing hypothesis which is the learning by exporting and self-selection into export market. So the working mechanisms of the learning by exporting is through increased access to better practices in terms of technology and management, increased competition as well as economies of scale to access to larger markets whereas the self-selection into exporting presses that there is high entry cost to export market mainly due to product updating and the search cost for new markets and new networks and we also discussed that there is a mixed result in support of either of the hypothesis. So we follow standard approach where firms export status is more or less a function of lagged productivity as well as previous export history and other firm characteristics. As Carol discussed earlier as well a positive coefficient on productivity is considered as an evidence of self-selection into export market. For the learning by exporting hypothesis we use firms output as a function of inputs used as well as its exports history and firm characteristics. For the current version of this paper we are using a value of output but I would discuss ways to improve this indexing price and productivity effect later on and a positive coefficient on the export history is considered as an evidence by learning from exporting. There are estimation issues discussed already involving firm heterogeneity as well as heterogeneity introduced due to lagged output as well as export status. We try to deal with this by way of differencing and using standard GMM approach as well as combining matching with GMM so that we compare similar firms with similar likelihood of entering into the export market. Preliminary descriptive statistics indicate that exporting firms do have larger productivity, labor productivity as well as employ larger inputs and they are large in size but we need to establish causality of productivity and exporting further on. The first result involves the selection into exporting where we use a simple profit regression explaining the likelihood of firms entering into the export market. We see that there is high persistence in exporting and also there is a positive effect of lagged labor productivity into likelihood of exporting and once the results get improved when we control for also the number of years the firms are exporting or the export experiences of firms once they enter the export market as well as some size category and as well as other control variables and results are also robust when we focus only on the six export oriented sectors. So there is some evidence of selection into more productive firms selecting themselves into the export market. When it comes to the learning by exporting hypothesis, so we start with a baseline oil less regression where we do not control for farm heterogeneity or endogeneity of export status. Once we control for full set of control variables, we find a positive relationship between export history and productivity of labor output of the firm. However, oil less is biased due to firm specific characteristics which we try to control for using the system GMM approach treating the firms export history as well as lagged output as an endogenous variable in our regression. So we also find controlling for all sets of firm specific characteristics. We find a positive relationship between export history and firms productivity as well. So focusing on the matched sample where we try to compare similar firms with similar firm characteristics and with equal likelihood of entering into the export market. We see that there is a positive relationship between lagged exporting and firms output and we also find that there is the number of years in exporting also matters for productivity. This is also robust when focusing on the six most export oriented sectors in our dataset. So to summarize my results, using a firm-level panel dataset, we find exporters are large in size, use more inputs and have a higher productivity. And this positive relationship between exporting status and productivity is both because of selection into exporting where more productive firms self-select into exporting and also because of learning by exporting. IE firms learn once they enter even if most productive firms are likely to export but once they enter into the export market they tend to learn and we see that by an increase in the level of productivity they score. So as I said earlier this is a preliminary work and ways we are planning to improve the work include differentiating between markups and productivity of firms and also to consider the extent of participation in the export market. Currently we just focus on whether they are exporting or not or how long they have been exporting. So maybe the volume of export also matters for the extent of learning. I think we can improve on that. We do not have information on so far or information on destination of exports but this is one way of extending the work that we had the data where the scope of learning depends on whether the country is exporting to a more developed country with advanced technology versus with similar to countries with similar level of advancement, technological advancement.