 Thank you. First, let me begin by an apology, which is I wasn't actually supposed to be here. Originally, Maggie McMillan was supposed to be on this panel and present a paper, and she was unable to make it and taking advantage of the fact that I was sort of in the neighborhood by my usual standards. I've been spending a couple of months at Oxford. She asked me if I could do this, so I said I'd be happy. Once I looked at the program, it looked really fascinating. In fact, we just heard two very good presentations that, in comparison, mine is going to be extremely incoherent because, first, it's not based on a paper. Second, I'm trying to bring in ideas from three different lines of research that I've been engaged in, and I'm going to try to make all of this as relevant as possible to what you've already heard, and then we can discuss. Briefly, what I'm going to be talking about is a little bit about the economics of manufacturing and convergence, and then talk a little bit about the institutions and issues of institutional design and architecture of state business relations. Here, there will be lots of parallels and complementarities, what we've just heard, and then I'll just put on my amateur political scientist hat and make a couple of comments about the politics of industrial policy. Let me start first with the economics of manufacturing and the role that manufacturing plays in economic growth and convergence. One big question here is why is manufacturing special and different? One window on the role that manufacturing plays in economic growth comes from these results on the prevailing tendency of unconditional convergence in manufacturing around the world in terms of labor productivity. The striking result here is that once these manufacturing industries get established, there is an almost automatic tendency for convergence towards the productivity frontier, regardless of basically of the level of disaggregation that you're looking at. These pictures are two digit manufacturing industries. The panel on the left is really the full sample covers 118 countries. Each dot here is a two digit manufacturing industry in one country. The striking thing here is that when you look at this picture, I'm actually not controlling for any country-specific features because we normally think that convergence is highly conditional, depends on country characteristics, whether you've got your human capital right, whether you've got your institutions right, whether you've got your policies right, whether you've got your macro stability. But when you look at manufacturing, it turns out that regardless of that, as even if you don't control for country-level characteristics, once these industries have been set up, they tend to converge in labor productivity terms. And you can see this from the negative slope, which is that the lower the initial base of labor productivity, the more rapid the subsequent rate of growth of labor productivity over the next decade. Now, before I knew that this was also true for Africa, but before I came here I wanted to check again. So the right panel is what you see is the same sample, but now restricted to just the sub-Saharan African countries. There's about 20 sub-Saharan African countries here. And the relationship is, if anything, at least the slope is steeper in the sub-Saharan African case. Again, it's the same thing that every country enters multiple times because of different two-digit industries. Now there's a selection. There's an issue here in terms of how to interpret this, because the sample here in most cases are already registered formal firms. So you don't want to think about this manufacturing as a whole. You really want to think about it as organized manufacturing industries. In most of these African countries the sample covers firms that with at least 10 employees. So maybe not sort of the medium-sized firms that John Sutton was talking about, somewhat smaller firms as well, but the striking thing is that there is this pattern. As I said, also you can observe this at various levels of disaggregation. I won't go to the four-digit, which looks very similar. But even if you aggregate at the level of manufacturing, you get the same, which is, so this is now each observation is aggregate manufacturing in different countries. And for the sub-Saharan African sample, pretty much a negative relationship. Of course it's only in the latter case about 21 countries including South Africa. So it's much smaller samples. So again, to emphasize this, it doesn't mean to emphasize what this means. It doesn't mean that policies don't matter. It doesn't mean that if you don't have appropriate macroeconomic frameworks, you don't have good human capital and governance, that in fact the productivity increase in your manufacturing sector will be even higher. But the striking thing here is that you actually don't need that. That there is some inherent dynamic in manufacturing, or at least again, organized manufacturing that puts you on this escalator. So that's the sense in which organized manufacturing seems to be the escalator part of the economy. Where if you just basically get these industries up and going, that there is an inherent dynamic that's going to push you towards the productivity frontier. Now, if manufacturing exhibits such convergence, the question is why doesn't it not aggregate to the economy as a whole? Why doesn't it carry over to the entire economy? And here I think we begin to get a bit of a glimpse as to sort of how in fact successful countries differ from unsuccessful countries, because it is not by and large in terms of the internal dynamics of how well their manufacturing has done in terms of productivity, but in fact in terms of how successful they have been in increasing the size of their manufacturing industries. That is increasing the number of firms, increasing the size of the existing firms in these escalator industries. So if you just want to think about how this would aggregate over to the entire economy, think about the economy being made up of sort of the modern sector, this manufacturing part of the economy, which exhibits this escalator unconditional convergence behavior, and then think about the rest, think about the traditional part or what I've called here the non-manufacturing, but you may want to think about the rest of the economy that doesn't exhibit this convergence behavior. And then if you ask how does aggregate growth gets determined, well aggregate growth gets the benefit of this convergence property, where beta is the convergence rate, but the problem is that the part of the economy which is exhibiting this convergence is actually very, very small. It's the manufacturing, and that's alpha. So in an African country this might be 5%, no more. So even though part of the economy is getting this convergence kick, it's not really carrying over in the large to the economy simply because the part of it that gets this escalator treatment is such a tiny part. And actually a big differentiator across countries is really going to be whether in fact resources are moving in the right direction, that is the converging parts of the economy, the manufacturing or escalator parts of the industry are expanding because given there's going to be large differential productivity differential given this convergence behavior between manufacturing and non-manufacturing, you're going to get a large kick out of this. And that's to some extent the key difference here in terms of reallocation of resources, the movement into manufacturing, again something that John Sutton emphasized this morning, is a key dynamic. And now that can happen both because individual firms, existing firms are growing and in successful countries they're going to be growing much faster than in non-successful countries or because you get entry. More firms enter the registered organized parts of the economy. But the emphasis that this framework generates in terms of industrial policy and state business relations is going to be an emphasis that says, well perhaps you have to pay a little bit less attention to the productivity dynamic of existing firms, a little bit more attention to whether in fact they're growing because those are two different things. The productivity dynamic seems to be, there seems to be an inherent productivity dynamic there but what's really big difference across countries is in fact whether those escalator industries are able to draw resources from the rest of the economy, they're able to expand or not. So that provides an appropriate and important focus for state business relations and industrial policy, enterprise policy, whatever you want to call it, is simply to figure out what is preventing these firms from growing or what is preventing new firms from entering into these sectors as opposed to what things that we might think of in the advanced country context as productivity policies or R&D policies and other things that are converse that focus on this term. So let me now move on to sort of this, the second institutional part of my presentation, the state business relations. So here's a triangle that actually sort of looks a little bit like the one that Lindsay showed you in fact there's going to be lots of parallel. I like to distinguish when thinking about the architectural design of industrial policies as essentially thinking about the separate roles of the state and business but let's not forget that there's a society out there and one of the things that I think one of the problems of industrial policy in the Asian context is that the state business relations have been working very well, thank you, but there is some problem in terms of to what extent that actually serves society's well-being and to what extent it's actually accountable to society. So the first element which John already mentioned is the notion of embeddedness and the notion of embeddedness comes from the fact that really the government doesn't have enough information about where the obstacles are, where the opportunities are, sort of where do you, again, which wheels are broken, again to use John Sutton's term and that information is going to be diffused widely in society within across firms and so the state or the bureaucracy has got to be embedded in society so that there are mechanisms through which that information can be elicited. So in that setting actually the traditional way in which economists think about regulation which is in terms of principal agent models is very, very, I think it's the wrong way of thinking about the way that bureaucracy ought to relate to business in this context because the principal agent model begins by assuming that the state actually or the bureaucrats or the regulator knows what it wants to maximize, what it wants to optimize, what's the welfare function or what the maximum is. A lot of this is actually trying to figure out what is it that you're trying to do. So in that sense this embeddedness is very important. The notion, the term embeddedness I think was first used by Peter Evans and it's interesting that he used this in the context of the station countries like South Korea where he said yes, the state in South Korea has been autonomous. That's sort of been the traditional focus of why is it that South Korea had been successfully industrial policy. But he said autonomy isn't enough. It's also that it has been an embedded kind of an autonomy so that the bureaucrats, the state were actually able to on an ongoing basis elicit information from business. Now John's discussion of presidential councils advisory councils was very interesting because it is a deliberate attempt at creating some kind of embeddedness. But also suggests that it's embeddedness isn't something that you can have only at the highest level. So it's interesting that one of the findings as John stated is that these presidential councils weren't very good about identifying obstacles. Did I get that right? And that's exactly what you would expect because they're so high level. You have these 15 large firms and what is it that those large 15 firms can agree on? They can probably agree on taxes are too high and there's too much red tape and so they can only agree on that but there's going to be very little else that they're going to agree on whereas in fact the real obstacles, the real problems are going to be much more deeply in sort of at a much more fringrained level. So in some sense this embeddedness really has to run through deeper layers of society. Now of course I want to repeat what Lindsay said which is that you cannot expect everything that these structures aren't going to be identical but you can rely on those pockets of bureaucratic efficiency or some of the more capable institutions to perform that function as well as you can which is of course another reason why you don't want to be too orthodox about the instruments of industrial policy because the appropriate instruments may often depend on which agency actually has the capacity. So what you may want to do through credit instruments in one country may better be done through tax instruments in another simply because of the differential capacity of different agencies. The second is really sort of discipline, of course I need to say probably the least about this because obviously what you want, the state you want to be embedded in business but you don't want to be in bed with business and so I think the component of discipline which is that you need to provide the stick as well as the carrot and that is really ultimately going to rely on formulating clear objectives, including measurable targets, degree of monitoring, evaluation, and program review. So I think these are some of the obvious elements that go in there. And third, the issue of accountability because ultimately this is, industrial policy is not about a nice cozy working relationship between bureaucrats and business, it's really about serving society's well-being and so you will need mechanisms of accountability here. Now when you look at East Asia, it's actually quite interesting about the lack of accountability. I mean in Singapore there's absolutely no accountability of industrial policy. You have all these bureaucrats deciding on the fate of industries and firms behind completely closed doors and nobody really knows actually what they're doing and why they're doing it except that they're very good at what they do. And part of what that is perhaps that in Singapore that they're actually very, very well paid and you can afford to hire them. In China I think probably there's obviously a lot of corruption that goes on but by and large it served positive ends and you might say that it's been because of competition across provinces and municipalities that has kept bureaucrats relatively honest. So in the three minutes or so that remains I want to just turn to my last idea, last topic which is how do you get this thing? What is the politics that's going to actually allow you to set up such designs? And here I just want to complement what Lindsay has said in terms of the importance of political settlements, social contracts, contracts, you know what Jamal and Robinson often talk about in terms of critical junctures, these periods when you have sort of this realignment of interests that enable a new bargain to be enunciated to that creates the context and environment in which you can get these positive institutional changes going forward. Now the main contribution I want to make to that discussion is simply emphasize something that is very attractive to any academic but I think is also true which is the power of ideas. And we often think about and I think there's a tendency in the political science literature but also in the sort of economics and the economy literature to think about sort of how these conditions come about in too much power and vested interests terms. That is that you need to get the constellation of power and vested interests just right that you can set up these institutions. I would like to say that in fact that these outcomes are relatively under determined by interests because oftentimes you're able to tell stories or narratives about what states have got to do that are compatible with the same constellation of power and interests but can take you in many different directions. So think about two different objectives which the elites have. You know you could be that they want to enrich themselves or you could be that they are actually sort of bureaucratically minded or state elites that they want to aggrandize the power of the state for foreign policy or military reasons. Well you say how am I going to best pursue that end? Well you can tell yourself a story that the way to do it is by repress markets concentrate all the power in your hands or you can tell a story that in fact you're going to do that in the markets by globalizing in a controlled kind of way ex ante whether in fact this enriches you more or that enriches you more, I would argue is not fully determined by the initial constellation of interests or power and I think the same argument would go about the aggrandizement of the state. Often we talk about the South Koreans or the Taiwanese having turned towards that growth promoting industrial policies because they felt this external threat and they reacted to that external threat by saying we better be manufacturing or export superpower so we can stand up to that. Well you know many countries in the Middle East, Egypt for example, responded to the appearance of a similar external threat from Israel in a very different manner by in fact building up an inward oriented military industrial complex with very different consequences. So what I want to really emphasize here is that there's a role for framing, there's a role for ideas, there's a role for narratives about what you can achieve without necessarily powerful people losing their power but simply reinterpreting in fact what their interests are. So let me stop.