 Felly rhaid wyf hanfodm'n hoffa'n ddiddordeb, a ddim ddod ar gyfer i'w A-Reef ar y ddiddorol. Rhaid hwnnau i'r ddiddordeb Twitter, ac ymwneud y pryd yn ddim ei ddim yn ddiddordeb ac ar gyfer gyfer y cysylltu'r ddae ac y mae'r effeithd o licent. Ac mae'n rhai i wneud y cysylltu yma i bobl o licent a licent a licent, ac mae'n rhoi o ddiddordeb yn iddynt y sgol yn ddigon i economi ac mae'n ddigon i ddweithio, ac mae yn rhan o'r cymdeithas of Andrea's presentation as well. So, in terms of a kind of background, in terms of thinking about economic development and some of the context, make three points as a point of departure on introduction. First of all, I think it's, well we'll see if this is contentious or not, the traditional pathway to economic development, meaning industrialisation, expansion of manufacturing shares is, I would say becoming harder for developing countries, harder to start on and harder to sustain. And the argument of Kaplancki, Felipe and others is that as more and more countries enter similar export markets, the amount of manufacturing and jobs is getting spread across those countries. Secondly, and whether this is good or bad is an open question, I think you can have good and bad de-industrialisation. Many middle-income countries are actually de-industrialising. I mean a more accurate term might be tertiaryisation, I think, than de-industrialisation. Again, there's some contestation, there's a de-industrialisation or a peaking, perhaps not a decline yet, of manufacturing employment shares whilst where the value added is peaked or not depends on the countries you're looking at. And so, for many of the world's developing countries, the issue around economic development is actually one of the growing service sector and how that provides employment and what it means for inequality. So, these are the kind of the kind of contemporary or I guess this kind of stylised facts of a kind in terms of economic development. And we thought we'd take a look at what those mean for inequality and obviously indirectly there are consequences for poverty in terms of employment creation. Of course, Cresnets is the backdrop to all of these kind of things and so we come to that in due course. I'll skip that one. In terms of de-industrialisation, we focus on de-industrialisation principally because we look at the case study of Indonesia which are E4 when introduced in the moment. So, there's very much written on de-industrialisation but it's largely with reference to the rich countries or advanced countries, not only historically some very seminal papers but more recently too. The relevance of that literature to developing countries I think is more of an open question given that de-industrialisation is having an earlier stage for many countries than historically was the case for other countries. There's a small set of country case studies we've found so far of five or six countries and a relatively small but growing set of cross-country papers on these kind of issues around de-industrialisation. Now, I'm going to talk about one of the papers, more recent papers, I'm going to talk about this because I think it's a very interesting paper, I'm also going to talk about it because it's the next director of wider. Cunal is smiling very modusly. So, essentially, by most, take the Gronigan 10 sector database and look at the relationship between structural transformation and inequality using the standardised world income inequality database that's forthcoming not yet publicly available. What they find is the movement of employment towards manufacturing is unambiguously associated with decreasing income inequality and the movement workers into services has no discernible impact on inequality but in certain types of countries, what they refer to as structurally developing countries, that is countries where manufacturing is greater than agriculture in terms of sectoral shares and employment, you get an increase in inequality and in structurally developed countries you get a decrease in inequality. So, I think that's kind of interesting as the kind of global picture based on about 30 countries and it brings us back to a question I think in terms of theory about what Cusnets actually said. I know Ravi Campbell wrote a very interesting paper that brought to the fore something that had been on my mind for quite some time that the Cusnets is largely reduced to the curve but actually this seminal paper was much more interesting, thoughtful. It's absolutely stuffed with caveats and nuance and I think it raises an interesting question about the forces that are impacting on inequality as countries go through structural change. So, principally, I think of interest is the fact that Cusnets, using the dual economy model, talked about inequality rising in the early stages as growth tends to benefit those with capital and education, as people move out of the rural sector, rural wages rise and inequality falls. That means that essentially inequality is decomposible between three components, inequality in each sector, the mean income of each sector and the population shares in each sector. So that, although inequality may rise as a result of the movement between sectors, in fact Cusnets pointed that out, it may also be balanced by what happens within the sector and the shares of each sector overall and it's been some various attempts to look at developing theory from a number of writers writing in that kind of tradition. So, for us, we were thinking principally about these issues of economic development, relinking employment and inequality to economic development, a structural transformation and what we were interested in was how the sectoral composition of employment or changes in it affects inequality and principally because the case study is around Indonesia in a period of history in Indonesia where Indonesia was deindustrialising, at least reached manufacturing peaks, the impact of deindustrialisation on employment in terms of whether increases or reduces inequality. I will hand over to Arif next. Okay, thank you. Let me go. Now, why Indonesia so many reasons? My other is Indonesia in the past, particularly in the 1980s, has been generating rapid employment growth through industrialisation. However, that's quite halted in the decade of 2000s and then coincidentally during the same times during the 2000s Indonesia has been rising unprecedented increase in inequality. And for the purpose of this paper, actually, we use Indonesian district for several reasons. For example, one of them is the commonality because district represents a broader range of social landscape and also shares so many characteristics unlike cross-country studies. And that's not the most important thing. The most important thing is because we have access to the data. And then also, actually we found that income and inequality range of Indonesian districts represent quite a good range of cross-country data too as we all see here, for example, that the triangle that represent the 390 districts that we have and then we overlay the data with cross-country data and then you can find Indonesian district which is actually similar to Saharan African country but actually Jakarta city is equivalent to Spain and Italy in term of income and inequality also in term of income per capita. And then we have also quite good range of inequality. So this 15 years of 390 district is quite interesting to dig down, deeper into better analysis. So basically what we did is estimate the panel data models trying to explain the factor that determine inequality, various kind. Our main indicator is Gini but we also check other measures. And then controlling other factor, we also introduce the quadratic term in the share of employment by sectors and then as well as the quadratic term. So we want to test as well the linearity directly testing the QSNET curve. And we assemble, we managed to assemble that, I said, containing 390 districts for 15 years. Also we add, so from that data we can get employment share as well, other than inequality and various indicators as well as we combine it with other data value edit from other sources. And this is just to give you a look of the data when you scatterplot the income, mean income of the district and inequality. And you can see how it's delayed. And then to give you also an idea of how we group the sectors. So this is the sectors that we group. We divide into five and five plus sectors for a reason that I will tell you again later. But we divide into agriculture and then industry we divide into manufacturing and non-manufacturing. We divide market into two, services into two, market and non-market services. And in the five plus aggregations we separate finance and insurance and business related sectors. And also if you want to, if you're interested, I also plot the trend of for 2000, how it looks like as you can see that the agriculture decline, however the manufacturing in the decade 2000 stabilized. So and then it seems that the non-agriculture employment move from agriculture sector to non-market services. And then this is just to give you a view about the casual observations. So you can see that it's obvious, for example, that when people move from agriculture and inequality rise here, but to which sectors that labor move, actually it's correlations between inequality and dose share, even though seemingly to be more positive than negative, but it varies. So what we did is we use these fixed effect regressions. And then we found actually that if we use only non-agriculture, this non-agriculture is positive and significant. So it means that the more you move to the agriculture sector, the employment share, the higher the inequality. But when you disentangle, this aggregate, these non-agricultures, then you can see here that, for example, all industries, apparently in inconsistent with Cunal finding, for international case, that the industry, both non-manufacturer and manufacturer actually has a non-linear increase. So it's increasing inequality, however, with the turning point. So it's a kind of cusnet way, a cusnet point of view. And however, in services, the non-market services behave like cusnet, however, the market services not really. So that's why I aggregate again that market services into buyer services, buyer and financial and so on. And then here, so then for the finance and business sector, we again find the quadratic term. So that's what we found. It is interesting to note that Indonesia actually here more or less to be non-market services, if you see this picture, this is non-market services, where the action is here. So this is the highlight, but let's also depend on the turning point. So we found that actually the turning point, even if it's statistically significant, actually it's high. It's quite high turning point. So for example, in Indonesia, the proportion of those districts below the turning point is very high. So most districts actually below the turning point, so it's still far away. And if we compare to cross-country, so they are still high too, but not as high as if you are using Indonesian districts. So it means 88% of countries in the GGTC database actually fall below the turning point. In the case of Indonesia, let me not give you a small remark, because Indonesia in the last one and a half decade, the movement is from agriculture and non-market services here, you can see. So when you look at the result, it implies that that movement that happened in Indonesia during the decade of 2000 will tend to increase inequality. So from this point, then we will argue that, of course, whether or not the industrialization increase or reduce inequality, it depends on what is the initial positions of that shares, for example here or here, and the directions of the change. So that matters. So it's general. So it really depends on country specific factors, country specific initial positions. Also we add just to test, we also add value-added shares instead of employment shares. Actually we found that value-added shares in GDP does not have any strong correlations between them and inequality, unlike labour shares. The reason we argue that value-added shares and employment shares is actually correlated but actually weakly, particularly when we look into district data, because district data is the mobility of input, particularly capital between districts. I think that makes below for effect of return to factor, for example. So what happened in the economic activity that happened in certain districts is not necessarily translated into welfare of the citizen of the district because of this free mobility of input. So we have to be careful in using this value-added data in the districts. You may argue differently in the case of cross countries. So we also do robustness checks. So we use 10 different economic quality measures because we are using micro data to do this. And then the result is quite robust to those variations. And then we also use different kind of specification. And then also we change sample variation and more or less the result quite robust. So let me conclude. So Andy, at the beginning, asked questions and then ourselves and we also tried to answer. So how does the sectoral composition's employment or change in it affect inequality? Well we have three points of answering this. An equality rise when the employment share of industry rise. That's what we found with our data. However, an equality rise when the employment share of some services with high turning point. So it has high turning point. So with service as well, rise, inequality rise, but high turning point. Even though some services may have a little bit lower turning point. So in our view, this finding somewhat supports cost net hypothesis. Our note on whether or how the industrialisation of employment increases or reduces inequality. So again we argue that it depends on the initial positions of the employment share within each country, sectoral share within each countries and the direction toward where they are moving. But let me, let us give some, what is it, not projection but view about the current global situation. So we found that if agricultural employment share is generally declining and I think it happened in most countries. So if we combine that with the predictions that the industry and service employment share of most developing countries are below the turning point, for example in the GDC country 88% are below the turning point. So it is less likely that the industrialisation will reduce inequality. So it will either doesn't have any effect or rise in inequality like for example what happened in Indonesia. So this is the end of our presentation. Thank you.