 Thank you. So we moved not only to a different country but also to a different kind of mobility. Here it's more geographical mobility because I will talk about the performance of migrant entrepreneurs, especially Indian entrepreneurs and entrepreneurs coming from different African countries in Uganda. So why is this question interesting? As we already heard in several other presentations, employment in micro and small enterprises accounts for 70 to 80 percent of total employment in many developing countries. And while most of this employment is typically self-employment, meaning that the employee person is the entrepreneur himself, hired family and non-family labor still makes up 30 to 40 percent of employment in the sector. So there's a huge potential for employment generation. And there's a stream of literature showing that there's large heterogeneity in the characteristics and in the performance of these micro and small enterprises. And one line among which this or along with this heterogeneity can be easily observed is ethnicity. So if you walk around the streets of Kampala or other East African cities, you usually notice a large percentage of Indian shops which seem to be performed very successful. So in Kampala, there is one percent of the population is of Indian origin. But they are over-represented in entrepreneurship, especially in the trade sector. And they seem to outperform their local counterparts. So the question is why are some groups more successful entrepreneurs than others? And I must say that in this paper, which is also preliminary work, we take a rather descriptive approach. So I'm not able to make any causal inference. So what does the literature say to this topic? Well, the reasons for the relative better performance of migrant or ethnic minority entrepreneurs, one reason is that especially entrepreneurs coming from Asian or European countries often have a better endowment with capital and also education. Then external environmental parameters are often named. What is meant by this is that the migrant entrepreneurs have a stronger motivation to succeed in business because they have been expelled from many other opportunities. For example, often they can't have land or they exclude from employment in the public sector in many countries they migrate to. Then by definition, migrants are more geographically mobile and this can be a valuable business resource because they get exposed to new contexts, new ideas, which they can use in their business. But the majority of the literature actually focuses on the effects of social networks. So it is argued that ethnic minority or migrant entrepreneurs have strong social ties and these provide them with advantages that the locals don't have. So through ethnicity-based information flows, they can ameliorate problems of incomplete information and they can use reputation as a contract enforcement mechanism. And through this, they have easier access to informal credit, they have more information to screen prospective employees, suppliers or credit recipients and this reputation can be used to substitute formal law enforcement. So the data we use, we use our own micro-firm-level survey data from 466 entrepreneurs in Kampala which was collected last year and we sampled across selected sectors in manufacturing, trade and services and we tried to stratify by migrant status. So the idea was to over-sample Indian and African migrant entrepreneurs, unfortunately that did not work out so well because it was extremely hard to motivate especially Indian entrepreneurs to participate in the survey. So the samples for the migrants are quite small. We only have 34 Indians and 40 African migrants in our sample. Here's just a picture of the diversity of businesses we have in our sample. So there's a typical market stall but also below you see a kind of shopping mall with printing and the electronic shops who have a rather large capital stock compared to the market store for example. Here are some descriptive statistics. At the left you see mean and median for the whole sample and then split for each ethnic group and already at first glance you can see that the Indian sample is significant and different in almost all respects compared to the Ugandan sample. So the stars denote significant difference in means compared to the Ugandan sample. And you can see that while the Ugandan and the African migrant sample are largely gender balanced there's a large male dominance in the Indian sample because there are also large differences in education. Maybe I can use a pointer. So if you look at the proportion of entrepreneurs who have completed university it's 74% for the Indians and also compared to only 9% of the Ugandan entrepreneurs for example. In contrast to what the literature suggests we do not find any big networks for the Indian sub-sample. So we have used a name generator to elicit information about the social network and on general 3.3 people were named but in the Indian sample they named basically no person at all. We are not sure if this is just a truth or if it's just a reporting error because as I said they were very reluctant to give out information so it might be that they just didn't name anybody but they still have big networks. We see that they are significantly larger they have about 6 employees on average compared to only one employee in the average Ugandan enterprise. And this is in part because they start out bigger and then they also grow faster in terms of employment growth. If we look at profits we see well first of all the values are highly skewed so median values are much smaller than the mean but we see the median profits for monthly profits for an Indian enterprise are still about 7800 international dollars compared to only 300 a little bit more than 300 dollars for the Ugandan enterprise so there's a huge difference and this is even more striking when we compare capital stocks where the capital stock where the average capital stock of an Indian enterprise is more than 50 times that of a Ugandan enterprise. If you look at the sectors we see that the Indians are largely involved in other wholesale and retail sectors so this involves trade in electronics and hardware and stationery. For the African migrants they are also in the other wholesale and retail and wholesale retail of fruit and beverages. So we estimate a profit function where Y is a locked self-reported profit then we have capital stock, locked capital stock measured as a replacement value of the total capital stock of the total business equipment, miners or excluding inventories. We have labor as the total number of hours worked in a month by the entrepreneur and all his employees. We have dummy variables for the Indian and the African migrant and a set of exogenous covariates so obviously there are a number of problems associated to this regression so the first one we can think of is omitted variable bias. There are probably characteristics of the entrepreneur such as entrepreneur ability for example which might be correlated to the profits and capital stock as well. Related to this there might be some form of selection bias that we can imagine that it's the most able Indians that migrate to become an entrepreneur. So that would bias the Indian dummy upward. There's also measurement error for sure in the profit and the capital stock and there might be also reverse causality so that people who have higher profits also accumulate faster the capital stock. Nonetheless here are the results and we see the Indian dummy is very high and significant across all three specifications. In the first one we just use the capital labor and the ethnicity dummy and this coefficient would if we interpret the coefficient it would mean that Indian profits on average have 485% higher profits than the Ugandan enterprise. If we include if we control for the sector and add other covariates and third specification we see that the coefficient shrinks in magnitude but still stays very big and highly significant so it does not seem to be the individual characteristics of the entrepreneur or the sector which explains this ethnicity gap in firm performance. If we do run the same regression but for each ethnic sample separately we basically see that the coefficients in capital and labor are different across all three samples and that apart from gender there's no significant coefficient in the Indian sample but it might also be due to the very small sample size. We do as a test also run the regression for the Ugandan sample where we only include those that are active in the same sectors as the Indians so that is other whole certain retail which is a reference category manufacturer of paper and printing and other services and we find that education seems to play no role in these sectors but otherwise the coefficients for the Ugandan sample stay the same more or less which also hints at that it's not the sector of activity which explains the ethnicity gap. We further try to decompose this ethnicity gap in firm performance using the Oaxaca blind-eye decomposition so this method decomposes the difference in predicted group means for the two groups in an endowment effect in a coefficient effect which is explained then in the coefficients of the profit function and an interaction effect of endowment and coefficients and what we find is so this tells us that predicted profits for the Indians are about three times higher than for the Ugandan enterprise and about a little more than half of this difference is explained by differences in endowments and here it's basically capital endowment that explains half of the difference in performance and the other major part is explained by differences in the product in the profit function so we assume of course this is not very satisfying we assume it's total factor productivity for example which is captured in this coefficient and the interaction term plays a natural role so to conclude we see that Indian entrepreneurs and in oh sorry Ugandan and Indian entrepreneurs differ significantly in their characteristics while we do not find such a large difference for African migrant enterprises we see that Indian entrepreneurs realize significantly higher profits than Ugandan or African migrant entrepreneurs and we see that this can be partly explained by the differences in capital endowment and but like a large part also remains more or less unexplained in the coefficient effect which might be capturing total factor productivity but also some of the bias in the regression