 Good morning. This is John Piper that I've done with my colleague, John Weiss at University of Bradford, looking at the special role of manufacturing. The paper is very much an empirical one, so we don't go into too much of a theoretical discussion and try to look at the role that manufacturing or other sectors within the economy could have, sectors or sub-sectors of overall growth that will come at then. So in terms of presentation, briefly look at the role of manufacturing, discuss briefly the methodology we've adapted in terms of establishing relationship between manufacturing and total growth, then data sources, which is similar to the one that my colleague just went through, data analysis and then conclusion and limitation of the paper. And there I would very much appreciate your comments on how we can improve the paper. In terms of role of manufacturing, there has been discussion that there are a special feature of manufacturing that impacts positively the overall growth and performance of an economy. The discussion goes back a long time. We go to Chennery et al. and the World Bank 60s, 70s, looking at the structure of change models. More recently, people discuss the role within a more dynamic sort of sense that manufacturing is very dynamic, is very innovative, relative to other sectors within the economy, and as a result, you'll have spillovers that are substantial learning by doing, learning by learning, and that fits through to the rest of the economy and to the overall growth and performance. It's also argued that manufacturing is an important source of technology and technical change. That's where you, within the economy, are building substantial embodied technology through machineries and the rest, that generates further changes within the economy, within the technology frontier. It's also argued that there are substantial externalities through the learning that I've just referred to, through forward and backward linkages with the rest of the economy, and that manufacturing sector also experiences increasing return to the trade, and as a result, it pushes off the level of growth within the economy. In terms of implication of all this discussion is that policy implication is that then by shifting resources into manufacturing sector, the economy is going to benefit. Growth will be higher and as a result, the level of economic activities will be higher than countries that would keep resources in less dynamic and innovative sector like agriculture. The result shift obviously is also, an old discussion goes back to Caldwell's 60s, 50s and 60s, and more recently I think Roderick also has looked into this. Just briefly discussed the recent view that complements to certain extent the discussion rather than contradicting, and that relates to the role of service sub sectors. And in our analysis here, we also find that certain sub sectors seems to have a more significant and statistically significant impact on growth relative to manufacturing sector that we just discussed. In terms of methodology that we adapt, our interest in this paper is to test for the extent of association between sector output and growth in the economy. That's the Caldwell talk discussion that we've just discussed. Looking to the sectoral productivity, catch up and convergence, and in this case we are looking into the beta convergence, unlike my colleague previously that was looking into sigma convergence. And also try to look at the presence of externalities between a sector output and the rest of the economy. For each of these we are adapting different methods of analysis. To test for one for example, this is the extent of association between sector output and growth. We're looking to the Caldwell's law and the one more recently adapted by Das Copta and Singhs in which the value growth of value added in an industry is positively related to the value added growth in manufacturing sector and negatively to value added, sorry, growth of labor in non-manufacturing sector. And in the data, in the paper, they use labor growth in agriculture as a sort of substitute for non-manufacturing sector because they are looking into developing countries' data sets. To test for the second issues that are raised, sectoral growth, sectoral productivity, catch-up and convergence, basically looking to the well-known model specified for convergence, cumulative growth within a particular sector J is negatively related to the initial level of activities in sector J and the role in terms of labor activities. Or alternatively, looking to the conditional convergence in a sense that we control for certain other factors. Z is a vector that catches up all the various factors that may impact convergence and for that, for the analysis that we conduct, we basically look at the institutional setup in order to see whether it makes much of a change in our analysis. For the third sort of test that we want to run in terms of presence of externalities and relationship between sectors and the rest of the economy, we adapt an augmented solar type model. This is Mankiew et al. and the rest that have followed Islam and the rest, in which case, gross, sorry, level of output within an economy, I at a time T is related to a set of capital and a set of labor. We apply the usual neoclassical assumptions, constant return to SK, specifically and simplifying the model we get to the equation one, gross of, gross within total gross within an economy, I is related to the rate of investment S that is made in different types of capital minus the initial level of output per labor. And the type of capital that we use in our analysis is the physical and human capital. The N that you see here, N relates to the rate of population gross. Sigma is the depreciation of capital, particular type of capital and gamma is the rate of technical change total factor productivity gross within an economy. We try to capture the impact of sectoral output through total factor productivity similar to the analysis that Temple and Johnson made in the 98th paper. A level of total factor productivity is related to output within sector J at time T as well as a set of other factors that affect labor, affect total productivity. We try to for example look into the impact that trade may have, impact that institutions may have, impact that policies, stability in the economy will have. And substituting this into our first model equation one, we simplify and get a relationship between the level of gross within an economy at time T that's related to initial level of total factor productivity plus two capital investment that goes on in an economy, initial level of labor productivity or sorry output and the rest of the variables that have just gone through. In terms of those data sources we try to, we make use of the Roderick and Macmillan's 2011 paper that is basically expanded data set that my colleague just went through but this is the 2005 version of it, up to 2005 version of it. The data set covers value added and employment for 38 countries. I think initially it was 30 and then Roderick and Macmillan expanded it included number of African countries and as well as eight sectors within the economy, agriculture, mining, manufacturing and a set of service of sectors. The data set includes a mixture of advanced economies, African Latin American and Asian economies and it covers data from 1995. We complement the data set from the word development indicators to trying to include other variables that are missing from the data set. Penword tables which are updated for 2011 and for institutional factors Kaufmann and Craig 2011 that produces six indicators. In terms of data analysis in order to as in literature in order to get rid of some business cycle and associated oscillation in data we convert data into five year averages so we have three sets of five year average data for analysis. In terms of data analysis we go into a bit of descriptive analysis basically correlation coefficients and then going to regressions that we apply ordinary to square for the issue of convergence and for externalities we make use of panel data. The data that we run various diagnostics and based on Hausmann's test we opt for fixed effect method of analysis and also the tests suggest that there are a lot of information in time and therefore we impose time dummy analysis. In terms of descriptive data as you can see from here this is the data the data set that includes all countries these are sort of initial indicators of convergence beta convergence again in this case and there's a mixed picture and as far as manufacturing is concerned doesn't seem to be convergence as expected this goes against the finding of Rodrik and associate that find a very strong beta convergence for manufacturers in terms of manufacturing sector they don't discuss much about the others however when we exclude Africa from the data set situation changes and then you find stronger beta convergence and manufacturing in particular becomes quite significant. We also have beta convergence for some of the sector-sub-sectors that will go through in terms of the externalities and the regression we run we look at the full data set and to look at potential impact and potential problems that we might come across there is substantial positive relationship for example between various indicators of institution that we use and therefore to try to catch the overall impact we use principal components and generate this PC country that well BC86 and that's the principal component of the old institutional indicators that we have used there are also potential problems of multicoloniality given a strong relationship between the dependent variable in our model so we try to avoid this when it comes to regression analysis instead in terms of Caldor's law we don't we find some relationship but not that very strong we try to use Caldor for both full data as well as data that excludes Africa we also make use of ordinary list score and a fixed panel result don't seem to change much we try to make use of the original Caldor that links productivity to growth of the economy to level of activities in manufacturing and then modified one that is applied by the sculptor and sing in more recent periods given the time that I'm running out of time so I'll push through the rest for the test of convergence as you can see here from the analysis here and specifically for the convergence as far as manufacturing CG man is cumulative growth in manufacturing sector doesn't seem to be convergence expected it is present in mining and some of the finance, some of the service sector the ones that you see here however when we look at the conditional convergence and include institutional proxy for into the model we find a strong convergence for quite a number of sectors included and when we exclude Africa from the data set again convergence is confirmed for number of sub sectors both conditional and unconditional are quite significant and as expected in terms of analysis when it comes to role of sector growth or output on overall GDP growth of the economy as you can see from the analysis here this one is log of physical capital or is the restricted version of the model that we've applied here this is proxy for human capital and that is secondary school education this is the GDP per capital for five years initial level of GDP for the country's concern the sectoral growth comes here this one is agriculture sector mining sector manufacturing sector and various other sub sectors that we include there is no evidence that there is a positive sectoral impact as far as manufacturing is concerned on growth and these are the time dummies that we use they are highly significant they may be catching up things that we are not catching up in our analysis in terms of sectoral growth on output using sorry I am using different indicators of sectors here this is dependent variable GDP per capita growth this is using sectoral growth for different sectors that we have used here we use sorry I have difficulty with my eyes I can't read well one we are using sectoral labour productivity and one we are using sectoral output okay this is the conclusion that we come to is that growth enhancing is the role of manufacturing if any is not unique to the manufacturing sector some service sub sectors also play a role there is evidence of convergence again beta convergence for manufacturing and some service sub sectors the convergence is stronger when Africa is excluded from the data set there is a role to argue that manufacturing has a special role given analysis however there are some limitations in our sort of results there are data limitations the data that my colleague used is much more expanded one that we have used and the data that Roderick Blackwell and used is also much more extensive the one that they used more recently there are issues in terms of modelling a strategy that we have adapted issues that we have not been really able to handle the return to scale ones effectively the externality issues need to be looked into more carefully whether they are costly externalities or not and the role of service sub sector why they are what the data set the data suggested are the analysis suggested there is positive in more positive interaction between some sub sectors but there are issues of indigeneity and causation we have not been able to handle in our analysis thank you very much