 So, this paper is during an exporting, importing and manufacturing firm performance in South Africa. It relates actually quite nicely to a couple of the sessions. The industrialization of smoke stacks and then the firms in Africa sessions that held earlier this morning and earlier this afternoon. So I'll try and link perhaps some of the insights or the outcomes from this paper to those sessions as well. So what the aim was is, I suppose, partly to showcase some work of the UNewider and this is really just, I was very fortunate to be involved in a UNewider project that was opened up access to administered data in South Africa. And this was a unique opportunity that really created a whole new space and opportunities to engage in firm level and worker specific research in South Africa that hadn't really been able in the same way as before. So it was a fantastic opportunity and this is part of some of the early work that arose from that particular project. So what am I going to do in particular? Well I'm going to focus really on the firm perspective in international trade and I'm going to focus on three key aspects. The first is just to present a background picture of heterogeneous firms looking at exporters, looking at importers and really just pulling out some of the descriptive pictures and descriptive story here. Try and compare this a little bit with what the international evidence is saying. I'm going to focus on manufacturing firms in particular. I'm then going to shift to looking at the complementary effects of imports on firm outcomes and I'm going to focus on two key outcomes. The one outcome is looking at firm productivity and the second outcome is on firm export performance. And this relates I think quite nicely to the earlier session because the earlier session Carol pointed out how exports and competitiveness is directly linked or strongly linked to productivity and labour costs. What this research will pull together is how traded input costs are also a fact that influence both firm export competitiveness but also how those link to firm productivity. So it brings in a different additional angle to that insight. It also relates quite nicely to the paper by John which spoke about capabilities and how thin many of the capabilities and particularly in the intermediate input sector are in Africa and in a way inputs and imported inputs are a substitute if you lack domestic input capabilities will access to imported intermediates provide some substitute. So I don't talk to those directly but there's an overlap between it. What I'm not going to talk about is the research that I'm currently working on but I'll close a little bit, close off on that as well. And finally I'll close off some implications. So let me establish the context a little bit. This is a picture I've just taken from the World Bank South African economic update but the broad context is that South African export performance hasn't really realised the expectations it had from the early 1990s when it liberalised. It hasn't met the expectations of the very strong export oriented growth plan in 1996, the gear macro policy and again this importance of exports in driving economic growth is also reflected in some of the new policy documents in South Africa. And this diagram captures the South African market share in exports. It's split into overall, it's split into minerals and metals and fuels and also split into non-commodity export products. And over the time what we can see is South African export growth is lagged out of other middle income economies. Even if it's lagged the economy is going to pull out to China as well. So it's share of world markets have fallen. Corresponding to this has actually been a decline in the share of the traded sector in the national economy. And this is a key point pointed out by Rodrik. And we see this in the decline, for example, the decline in the manufacturing share of GDP, which used to be 20% in 1990, it's now about 14%. The same trend with employment. So manufacturing and the traded sector as a share of the domestic economy has been declining. That's not unique to South Africa we know, but that decline has been relatively strong post 1990s. So it reflects in a way a lack of structural shift and adjustment towards the traded sector that was expected. A second structural problem that emerges behind this is the inability of the South African export bundle to diversify out of minerals. Minerals still accounts for about 50% of our export bundle and has done so for the entire period of all. So this is the broad context. Now the key problem with much of the analysis all the way up to this point, although this data does use transaction data, is that we had very little insight into the role that firms played in driving these dynamics because we didn't have firm data. We didn't have panel data. There were occasional firm surveys, but we couldn't follow firms over time and how firms traded over time. So the insights that we could obtain were very limited. And what we know from the literature is that firms are extremely heterogeneous and aggregating them into industries doesn't necessarily make sense. So come in the Saviors and that's the UNewider South African Revenue Services and National Treasury together come together to make available administered data. And they made three sets of data available. The first set was really the customs or transaction data. This is the individual export and import data that is collected by the customs and revenue authorities every time a good is exported or a good is brought into the economy. The second database was the company income tax data. So all firms that are registered for income tax, they submit their annual returns and they provide some information about production, profits, et cetera, cost of sales, et cetera. So that individual annual data was provided over the period as well. And finally, what was also provided was the employment tax certificate data. And this data varies by job. So individuals are repeated multiple times if they have multiple jobs for different companies. And on top of all this, we were able to combine all three databases. Very messy process. A lot of observations not fitting together. But effectively, we were able to combine a very detailed trade transaction database with an income tax database and a very detailed employment database as well. So a very rich panel of data over time. That has led to a lot of research. So what I'm going to focus on just is an overview first of the firms that trade. And try and get some perspective. Well, the first, I'm going to focus only on manufacturing firms for which we were able to calculate TFP. So it's a sub-sample of all the available firms. But the pictures effectively correspond when we use all firms as well, or we just look at the transaction database. But what we find is that on average, about 30% of South African firms engage in international trade. But when we look at those directly, these are directly import or directly export. When we look at amongst those traders, what we see is half of them both import and export. So exporting and importing are very strongly arrayed. Firms that trade tend to, at least half of them, tend to both export and directly import. So there's an association here that will exploit a little bit later. But what is also interesting is over time, we saw very few little increase in the total number of firms of the period. And we see very few, a very stable or stagnant number of firms engaging in international trade. So this is not a dynamic period in which we see new entrants entering on a stable basis, at least, into the international environment, either by importing or exporting directly. It's effectively a relatively stagnant period of engagement in international trade. That differs enormously from Ethiopia, differs from Denmark over this particular period as well. When we look at what we also notice, we do see, at the detail level, a lot of churning. So firms enter into trade and then they exit. And these predominantly are smaller firms. But very few firms trade over the entire period. Only 25% of the sample we looked at that traded in 2010 was still trading in 2014. So a lot of churning that is occurring. So partial exporting and importing is occurring. When we do look at the transition matrices, what we find is that very few non-traders enter into exporting or importing in the subsequent period, only 3%. And we see, if we look at the persistence, we see a high degree of persistence, at least amongst those firms that are exporting and importing. So it's really the firms that both export and import that are more persistent in engaging in international trade. 82% of them retain, stay imported and export, but only 2% of them exit into non-trading status in the subsequent period. So these are the realists. These firms, these exporters and importers, seem to be very different from exporters only and from importers only. Following the international literature, what we find again is that these firms that trade tend to be bigger, they employ more workers, they have a high value added per worker, they have more capital intensive, and they pay higher wages. Looking within these firms, we see that it's predominantly the exporter importers that are driving many of these premiums within the sample. So other insights though, these are not from my own work, but from some colleagues and other members of the research group. They also expanded and looked a little bit more at some of the heterogeneous exporters. And what they come across is some of the standardized facts that correspond with the international literature, is we see few firms are exporting, we've shown that, but actually when you look at the firms that do export, they export a very low proportion of their sales. So over 50% of the firms export less than 10% of their sales. So very few export only firms that predominantly export their goods. What we also see is that the productivity premium that we find that is generally corresponded, associated with export performance, that's very low for firms that export to Africa. In fact, in many cases, there's no significant difference in the productivity between firms that sell only to the domestic market, manufacturing firms that sell to the domestic market, and manufacturing firms that sell to the African market. And so that's a very interesting paradox, we're not too sure. And one argument could be is that these, that the large preferences that these firms have through the regional trade agreements are sufficient to offset some of the high input costs associated with accessing the regional market. But effectively, those firms that export to Africa tend to be very different to the firms that are exporting outside of Africa. And if we unpack that further, these are the firms that tend to export the non-commodity based products versus the resource based products. So that's an issue that we need to unpack a bit. Further, what we see is also high degrees of within firm wage inequality that very strong, particularly amongst exporters. So there are a couple of implications and I'm cautious about drawing up this relationship too strongly. But one key issue we're facing in this is that these firms, the exporting firms are not the firms that are going to create the jobs for the large, the large number of unemployed workers that we have in South Africa. The skill composition of the exporters. They tend to be relatively skill intense, they demand skilled labor, they tend to produce relatively sophisticated products. These are not the firms that are going to absorb the high, large number of unemployed workers within South Africa, unless new firms enter that absorb these. So that's a challenge that we need to face. It's not clear access to international market expands the scope and opportunity to exports, but it's not clear that these are the firms that are going to transition into the international market. They still seem to be far from the productivity frontier in order to access the international markets outside of Africa. I put a question mark on that because that hasn't been really studied. And then there are issues about whether growth in exports could actually contribute to the very high wage inequality as well. So if we transition into exporting those firms. Let me focus quickly on the last two components of what I want to talk about. The issue of importing and productivity was really the focus of our particular study. And what we were really interested in is to try to understand how imports affect firm outcomes. And why we took this particular angle is within the discourse and the policy discourse in South Africa, there's a strong anti-import sentiment. There's a sense that imports are crowding out. The competitive effects of imports tends to dominate much of the policy discussion. And what we felt was that this missed out some of the potential contribution of access to imported intermediate inputs as well. So we're really thinking of three particular channels through which imports could affect productivity. The one is the complementary territory of inputs. So firms have a large variety of inputs that they're able to source to the extent that inputs are complementary. This will enhance productivity as well. The technology transfer channel, we try and explore a little bit as well. This is about trying to unpack the extent to which technology embedded in imports affects firm productivity. And then we also know there's literature that looks at the issue of quality and whether quality of imports affects or inputs affects outcomes. We don't really look at that. But if we look at just the simple unconditional premium, what we see is that firms that tend to import are roughly about 40% more productive than firms that don't directly import. And we're only looking at direct imports as well. So we're ignoring the indirect imports that firms may access. And what we see in particular, the productivity premium is much higher for firms that both export and import as well. If we then try and play around trying to estimate some conditional relationships, and these I must say, I would be wary of inferring causality on these because we haven't really managed to grapple with the endogeneity concerns properly. So it's much better for us to just refer to these as associations. And so we run some simple regressions where we try and look at indicators of TFP and we regress these against various indicators of importing performance. And we restrict the sample again to only those where we can estimate TFP for manufacturing firms. And I'm going to break with the note and I'm gonna show you a table, but we don't have to worry too much about the diagram. The first story really is if we run our simple estimates where we look at the exporter premium, we see that exporters tend to have a large premium TFP premium relative to other firms. But when we add the importer term, what we find is quite a lot of that export premium is actually accounted for by the importer status. So it's both the combination, this premium that often we see for exporting, I think is in part a combination of the thresholds or the fixed costs of both accessing the export market, but also that needed to access the import market and the two factors play a role. And so I think some of that exporter premium is in fact in part relating to an importer premium in many of the estimates in the literature. And roughly we'd argue that the importer dummy, I haven't shown all the other controls, importing status effectively confers, is associated with a 3% higher TFP. We then play around and try and look at the complementarity of relationship. And we focus on looking at the variety of imports and the variety has defined in terms of the source, product-source combination. And we see again a very strong relationship between the varieties of firm as importing and its productivity relationship. We then, in terms of technology transfer, we split the varieties into whether they're sourced from high income countries as a proxy for embodied technology in imports, and we also split it into imports from emerging economies. And we find they're both significant, but there's no difference between the two. So it seems a variety is from all sources matter in terms of accessing input. So it's a strong support for that there's a complementarity relationship. When we look at exports, we then try and focus on two additional channels. I mean, we think that imports through its productivity channel will affect firm's ability to access the international market. But we also, there's also literature that would argue that access to imports, imported intermediate inputs, lowers, provides access to cheaper inputs, and that itself raises profitability enabling firms to access the international market as well. And we're interested in trying to unpack both whether this drives the value of exports of existing exporters, but also whether it enables firms to access new products or associated with firms accessing new products and new markets as well. If we again, just look at some of the basic statistics behind these firms, we see that exporter importers are very different from exporters only. And in particular, they sell more products per destination, 9.4 versus 7.6, they have more destinations per product, and they're much bigger. On average, they export 14.4 million around per year versus 2.2. So these are very different firms that export only. And that's just reflected again in this kernel density. We can see export importers tend to export higher values than just exporters only. So if we look within this data, if we try and in our simple estimates, our first estimate, we just try to understand if importing raises the probability of you exporting in the subsequent period, and we find some evidence that, yes, prior export participation is associated with a higher probability of exporting in the subsequent. This is just a linear probability model relationship here. We then interrogate a little bit more and try and understand whether the association of the export values. And we find a number of interesting relationships, a correlation linked to import value and export value, but no strong relationship between variety of imports and the value of exports. Where the variety kicks in is really in terms of the variety of exports. So there's a much stronger relationship between variety of imports that affirm imports and the variety of exports that it sells. So it seems that that relationship is stronger. And then we find some evidence to suggest that that relationship is stronger if you source varieties from high income economies versus emerging economies. But actually, those relations, that's not really significantly different. So in effect, where do we go next? I mean, what are the key insights? I think the number of insights, I think the first insight is really is that exporting and importing are very closely and strongly associated with in the South African manufacturing sector. And I think this corresponds internationally as well. It tells us that if we want to engage in raising exports, we also have to think about how to facilitate access to intermediate inputs. Because what we know is cheap intermediate inputs or technology embedded in intermediate inputs is a key factor that also helps explain firm export behavior. So that's more than just productivity directly and lab costs. I think we need to look at costs of inputs as well, costs of traded inputs as well. What the broader implications for South Africa? It seems our data is suggesting that we are on some stagnant, steady state on manufacturing behavior. We left alone, I can't see the data suggesting that we're gonna have no change in the situation. And that really does talk about interventions. And many of the interventions about lowering the cost of traded inputs, lowering the cost of non-traded inputs like transport, electricity, et cetera. My own general sentiment is that the export relationship reflects a general malaise in South African industry itself. And unless we get domestic investment, whether for domestic markets or international markets booming, we're not gonna break out of this system. So I think the problems are much bigger than just, we don't just need interventions for exporting firms alone. I think we need in general interventions to break the steady state, low growth trap that we stuck in. I won't talk about the new research directly in depth. We just look, trying to understand how this trade characteristics affect firms' export response. We see there's export exchange rate disconnect. Firms are not responding to the depreciation as we'd expect. And then also this link with allocative efficiency. Indicators of allocative efficiency in South African manufacturing are terrible. And it doesn't seem that trade, as we would expect, is driving improved allocative efficiency. But that's very provisional analysis. So that's to conclude. So thank you very much.