Skip to content

The Impact of Syrian Refugees on the Lebanese Labor Market

August 10, 2016

Editor’s note: This post is part of a series showcasing Barcelona GSE master projects by students in the Class of 2016. The project is a required component of every master program.

Nagham Abdel Ahad and Gleb Bychkov

Master’s Program:

Macroeconomic Policy and Financial Markets

Paper Abstract:

In the light of the Syrian crisis which erupted in March 2011 and which is still on-going, many outcomes were and are still being produced. In this paper, we show particular interest in the refugee crisis that developed after the fast-tracked evolution of events in Syrian. More precisely, we shed light on the case of Lebanon, the small country on the Mediterranean, having an estimated native population of 4.55 million and hosting, at present, an estimated 1.15 million Syrian refugees on its territories, that is, more than 25% of its original population.

Our study focuses more specifically on the negative spillovers of the Syrian refugee inflow on the Lebanese labor market. Our objective is to build a model which we use to determine both the steady state in the Lebanese labor market prior to the Syrian refugee crisis and the equilibrium in the Lebanese labor market post the escalation of the refugee crisis. We therefore approach the dynamics of the labor market observing its reaction to the positive labor supply shock generated by the refugee influx. After calibrating the model, we watch closely the changes in our main variables of interest, namely, unemployment rates and wage levels, before and after the crisis.

In addition, we compare the results and the values our model gives for our variables of interest with the actual figures and data published or predicted by international reputable institutions, such as The World Bank and the Food and Agriculture Organization of the United Nations, for these same variables. Accordingly, we evaluate our model showing how far it succeeds in reflecting the reality of the situation and thus in predicting and generating figures as close as possible to the actual and true ones.


Refugee inflows into host countries and communities can have significant impacts on these hosts on many levels. In our paper, we approach this issue from an economic perspective. More specifically, we focus on labor economics and labor market dynamics. In this context, we consider the case of the Lebanese labor market invaded by Syrian refugees who have fled to Lebanon because of the on-going war in Syria. We build a model, we calibrate it, we get the results, and we discuss them and use them to evaluate the performance of our model.

While The World Bank estimates a 20 percent unemployment rate in Lebanon post-crisis (almost double the rate pre-crisis), our model estimates an approximate 6.68 percent.

We proceed afterwards with a thorough discussion centered around these contradictory observations and we also go over a couple of limitations our model has, all of which might be able to account to some extent to the inconsistency in figures. This idea is interesting as it opens horizons and broadens the scopes for this work as one might expand the model so as to include an informal sector or additional distinctions between Lebanese native workers and Syrian migrant workers. We did not engage in doing this activity given the time constraints that we had. However, such expansions of the model can add great value to this work and can pave the way for further understanding and more successful outcomes.

We view our paper as a first step towards developing a flexible quantitative model that integrates opposing forces and that allows for a proper welfare analysis. More analysis is clearly welcome.

Re-examining the Global Liquidity-Asset Prices Linkage: Case of G7

July 27, 2016

Editor’s note: This post is part of a series showcasing Barcelona GSE master projects by students in the Class of 2016. The project is a required component of every master program.

Ryan Jacildo and Ekaterina Rezepina

Master’s Program:

Macroeconomic Policy and Financial Markets

Paper Abstract:

Research concerning the linkage between global liquidity and domestic economic affairs is hardly new. Interestingly, however, it never gets old mainly because of its policy significance. The welfare impact of shocks to capital flows (may it be short or long-term) is by and large the bottom line of all the discussions. Inquiries about other pertinent issues such as global financial imbalance and asset price bubbles, international financial stability and global financial safety nets, and economic early warning systems are in one way or the other broadly tied with global liquidity. Indeed, the impact of shocks to global liquidity can be systemically disruptive. And because international financial landscape constantly changes (i.e. degree of linkages estimated in one period may not hold in the subsequent periods), regular spot checks are important. The aftershocks of the global financial crisis (GFC) in 2007/2008, for instance, re-emphasize the significance of understanding the consequences of fluxes in capital movement and the extent of these consequences in various settings and time periods.


The main motivation behind this study is to contribute to empirical literature on cross-border liquidity spillover effects on asset prices in light of broadening global economic integration. We decided to focus on the case of the Group of 7 (G7) economies (e.g. Canada, France, Germany, Italy, Japan, United Kingdom, and United States) and follow closely the earlier work of Darius and Radde (2010) – henceforth D&R. For the same set of economies, D&R looked at the relationship of global liquidity and asset prices before and after the “Great Moderation” period. In an attempt to provide an account of what happened after 2007, this study examines the behavior of the same variables until end of 2015 and checks whether there are significant changes to the magnitude of the pass though effects of global liquidity, particularly on the equity and property prices in recent years.


In light of the developments in the past decade that led central banks to flood the international financial system with liquidity, we deem it relevant to empirically re-examine the linkage between global liquidity and asset prices in large economies. To do the econometric analysis, we used available data from 1984q1 to 2015q4 and employed a VAR model following the specification suggested by D&R.

In the global analysis, we found that global liquidity has a positive significant impact on commodity prices using the sample from 1984q1 and 2015q4 but insignificant impact on equity prices. However, in our subsample analysis using the data from 1984q1 to 2007q4, our results showed that the impulse responses of both the CRB and the MSCI were positively significant and persistent while the impulse response function of house prices remained insignificant. Interestingly, D&R, which also used 1984q1 to 2007q4 as its Great Moderation subsample, found that the responses of commodity and equity prices to a liquidity shock were insignificant. In terms of the house prices, the results we obtained differed from those of D&R for periods from 1984q1 to 2007q4 in a sense that we found significantly negatively response to global liquidity albeit with a substantial lag of 16 quarters.


In our spillover analysis, we extended the model of D&R by adding the local stock prices to the model. For each economy, we ran the regression using data from 1984q1 to 2015q4 as well as from 1984q1 to 2007q4 (to serve as our pre-GFC subsample). The results of this exercise convey that the positive effect of a global liquidity shock on house prices in Japan obtained using data from 1984q1 to 2015q period disappears when only pre-GFC period is considered. In the full sample analysis both global and domestic liquidity did not affect stock prices in Japan, whereas the effect of global liquidity turned out to be positive for the pre-GFC period.

Notwithstanding the sample used (may it be full or pre-GFC), the effect of global liquidity on house prices in France stays significant and negative, while the negative impact of global liquidity on stock prices obtained using full sample disappears in pre-GFC subsample. In the case of the latter, the stock prices turned out to be significantly positively affected by local liquidity, while the inclusion of post crisis years made this response negatively significant with a substantial lag.

Lastly, the result of the pre-GFC subsample analysis involving the UK reveals that the effect of local liquidity shock on stock prices is not significant as opposed to full sample estimation when the effect was positive. Moreover, the variance decomposition dictates that global GDP growth rate explains the largest proportion of the volatility of stock prices in the UK.

Moving forward, one way to get a better understanding of the results would be to properly assess the country-level intertemporal idiosyncratic factors just like in the global analysis. Certainly, the nature and timing of these structural shifts can vary from one country to another. We likewise suggest trying different proxies for the global liquidity or run the model for the monthly data without house prices that are available only quarterly and the monthly proxy for GDP. Using monthly data would allow a closer analysis of dynamics in the post-GFC period. It would also be interesting to extend the scope of this exercise to emerging economies.

The Italian Productivity Slump: A Tale of Zombies

July 21, 2016

Editor’s note: This post is part of a series showcasing Barcelona GSE master projects by students in the Class of 2016. The project is a required component of every master program.

Andrea Fabiani, Enrico Frezzini, Federico MorescalchiWilly Scherrieble, and Ugur Yesilbayraktar

Master’s Program:


Paper Abstract:

During periods of low productivity and stagnation, banks develop incentives to renew loans at subsidized rates to firms that would otherwise be insolvent. Proliferation of these firms, which we label as Zombies, have ramifications on the economy. Theoretical models predict that industries with a large share of zombies should experience productivity slowdowns in conjunction with lackluster employment and investment trends.

In this paper, we try to document whether this form of capital misallocation prevails between 2007-2014, by analyzing balance sheet data for more than 19,000 non-financial companies from the AMADEUS database. In fact, Italy’s productivity growth has been ailing since the early 2000’s and the volume of non-performing loans increased dramatically after the 2008 financial crisis. Moreover, prevalence of relationship banking and lack of loan loss provisioning in the country created a breeding ground for zombie lending. We identify zombies by comparing the yearly interest rate payments for the companies in our sample with those implied by a weighted average of Italian sovereign yields.

Our main contribution to the literature is to extend the seminal work by Caballero et al. (2008) to Italy, so to quantify the statistical association between zombie lending and several indicators of economic performance. Up to our knowledge, we are the first to carry a similar exercise for an economy different from Japan, whose experience in the 1990’s gave rise to the literature on zombie lending.

The descriptive analysis of our data reveals the increasing trend in the number of zombies in the aftermath of the financial crisis of 2008; this phenomenon appears to be widespread across different sectors of economic activity.


Figure 1: Pre and post-crisis average fraction of zombies (Overall economy)



Figure 2: Pre and post-crisis average fraction of zombies (Asset weighted – Industry-level)

Furthermore, by means of OLS regressions, we show that TFP is lower and more unequally distributed in sectors with a relatively higher fraction of zombie firms, in accordance with theoretical predictions. However, the same exercise hints to a positive partial correlation between employment in healthy firms and zombie-lending, at odds with the theory.

We call for future research to explain this finding, either by enriching the underlying theoretical model so to account for realistic features of labor market or by testing the same hypothesis for other countries.


Brexit: BGSE Community Analysis

July 7, 2016


We invite all Barcelona GSE students and alumni to share their early reflections on the potential economic consequences of the UK’s recent vote to leave the EU. Did you focus on a related topic in your master project? Are you working at a think tank, central bank, or consulting firm where your projects will be impacted by this decision? Have you seen any articles or links that you found useful for understanding what lies ahead?

Here are a couple of pieces we’ve found to get the discussion going:

After Brexit: What next for the EMU, EU and UK?
(ADEMU webinar)

The BGSE participates in A Dynamic Economic and Monetary Union (ADEMU), a project of the EU Horizon 2020 Program. Last week, ADEMU researchers held a webinar to discuss the Brexit.


Europe has grown out of its crises when reason and solidarity have prevailed, but it has also been devastated by its crises when fear and nationalism have taken the lead. Brexit, in the aftermath of the euro crisis, brings this dichotomy back to the foreground. Since 2010 there have been important advances in the development of the Economic and Monetary Union (EMU) and flexible forms of participation have allowed other EU countries, reluctant to join the euro, to share the basic principles that define the EU and have a common presence in the interdependent global world.

According to the panelists, Brexit raises 3 crucial questions:

  1. Should the EMU be accelerated to become a centre of gravity within the EU, or slowed down to avoid a centrifugal diaspora? If accelerated, how?
  2. Should an ‘exit’ country be allowed free entry to the single market and other EU public goods without accepting freedom of movement?
  3. Should the EU remain as it is, or increase its capacity to offer common public services (Banking Union, border security, research funding, environment, etc.), or limit its scope of activity to the EU single and integrated market?

Webinar Panel:
– Joaquín Almunia (Former Vice-President of the European Commission, honorary president of the Barcelona GSE)
– Ramon Marimon (European University Institute and UPF – Barcelona GSE; ADEMU)
– Gorgio Monti (European University Institute; ADEMU)
– Morten Ravn (University College London; ADEMU)

Annika Zorn (European University Institute; Florence School of Banking & Finance)

From Brexit to the Future
(Joseph Stiglitz)

Nobel Laureate and Barcelona GSE Scientific Council member Joseph Stiglitz shares some reflections in the wake of the Brexit decision

What are you thoughts on Brexit?

We want to know what the BGSE community is thinking and reading about the Brexit. Please share your ideas, favorite sources for analysis, or observations from economists you respect in the comments below.

The rise of Voting Advice Applications – and the Spanish case

June 21, 2016


With over 700,000 users, data from the app suggests how VAAs could represent a whole new way of surveying the general public before an election and collecting data on the political position of the population.

The creator of the app is BGSE alum Hugo Ferradáns ’15, graduate from the Economics of Public Policy Program. Follow him on Twitter @Hferradans.

The rise of the internet era opened a door for innovative ways to help voters be informed about their political choices prior to casting their ballot. During the past 2015 Spanish General Election, new tools such as ( in English), an app that matches users’ policy preferences with parties’ proposed policies, became an easy and straightforward alternative for users to explore their political position and compare it to that of the biggest parties. Its success, with over 800,000 users and more than 30 million responses, suggests how technology and the social sciences can work successfully together to create a more informed and accountable electorate, especially in a multiparty political system such as the Spanish one.

But encouraging are more informed electorate is not the only benefit of Voting Advice Applications. In fact, the large amount of data that is generated from online applications such as can be a source of analysis and study regarding why people make their choices1, as well as a way to estimate what users care most about in a real-time basis before an election. This article, thus, will try to shed light on the usefulness of Voting Advice Applications to gather data on the political positioning of users. I will show some of the results that were acquired from, both on the policy preferences of users and on their most politically-aligned parties.

But first things first- What is exactly is what is called in the field of political economy research a “Voting Advice Application” (VAA). VAAs are essentially an online test that matches users to parties depending on individual responses to policy-related statements. The user can either disagree or agree with the statements, as well as indicate whether that specific policy is important to him or her. After replying to several questions, the VAA gives the user a summary of what parties the user disagrees and agrees most with, mainly in the form of a ranking or a political map.


Even though there some VAAs more sophisticated than others2, all VAAs acquire essentially the same data:

  1. the position of the user regarding a specific question (in a scale of completely agree to completely disagree with the statement in question),
  2. whether that user gives importance to that question and
  3. after answering all questions, the ranking of most preferred parties for each user. was able to gather information on 756,908 people, after dropping all users that did not complete at least level 1 (that is, replied to 31 questions).

What did users get as an advice from

If we look at what party was the most first-ranked among users, we see that the centre-right Ciudadanos was the most preferred party throughout the whole period for roughly 33% of users. However, interestingly enough, the overall amount of people that voted for parties that are more leaned towards the left (Podemos,PSOE, United Left and Nós, representing 62.8% of votes)  is much higher than those in line with liberal and conservative policies (Ciudadanos, PP, PNV and DiL, being 37.2% of users’ first choices), indicating that users from are consistently left-wing.


It is particularly noticeable the different layout that the results present when compared to the results from General Elections. For example, the conservative Partido Popular, which was ranked first in the elections with roughly 25% of votes, appeared last almost throughout the whole period for It is clear that this might certainly come from the fact that VAA users are consistently younger and more left-wing than the average citizen, but it also poses a question that would be interesting to explore: do people vote in line with their policy preferences or are there other factors that are influencing voters’ decisions in the field of electoral politics?

How do people position themselves about certain issues and what they think are most important?

Unsurprisingly, the topics related to corruption were the ones users gave most importance to, with almost 10.67% of respondents (that is, 80,410 individuals) giving importance to the question “Politicians accused of corruption should resign and be illegitimated to run for office”, of which almost 93% of people responded that they agree or completely agree with the statement.

The second and third place of most-given-importance questions are related to the presence of religion in the political sphere (second place) and the presence of religion in the education curriculum (third place), for which both find a strong rejection towards religion.  Furthermore, social policy is an area of much importance to individuals as well, surely very much related to Spain’s current economic woes. Indeed, Spanish law related to mass evictions over the past years3 takes fourth place in most-given importance question (8.06% of total questions replied), followed by a statement on the education budget (7,46%), for which most people agree that increasing the budget is a top priority within government policy. These results are roughly constant throughout time, although the amount of users that gave importance to questions declined (graph 2).


In terms of the most controversial topics out of all questions, where there are large amounts of people agreeing and disagreeing with the statement, we find the prohibition of bullfighting, the abolition of escuelas concertadas4 and the law regarding underage abortion5, having all of them a rather high rate of importance-responses as well.

Regarding what users are not interested on, that is, the questions that were least given importance to, it is seen that the four topics that are least important to users (starting from the least important) are the deficit and the ceiling of government expenditure, the legalization of prostitution, the regulation the financial sector, and the financing of the Autonomous Communities (the different regions of Spain).

What is the political position of the average user?

In order to give users the most interactive experience when analyzing their results, we created a map of their political position using eight different axis, as the Swiss VAA smartvote6 did. Using an algorithm, each response that a user gives contributes to create its “political map”, which can be later compared to the political map of the parties. Thus, using the responses from each user, we computed the political map for the average user, creating the image below.


As it can be seen, the average user is very much in favor of strong democratic institutions that condemn corruption at all levels, as it presents a rather high value for the axis related to democratic regeneration. Furthermore, it also presents a high value for welfare state and liberal society, and quite a low value for those questions supporting a liberal economy and a restrictive fiscal policy, which goes in line with the results mentioned above that users are more prone to identify themselves with left-wing policies.

Also, it can be seen that the average user rejects all statements related to regional nationalism, and favors those regarding state centralization. This changes, however, when comparing the average users from different regions, as people from Autonomous Communities such as Catalonia and the Basque Country strongly reject state centralization and favor regional nationalist policies.

What is left to be done from VAAs like

Although VAAs can give academics a rich database, there are a number of methodological challenges that need to be overcome7, mainly regarding the representativeness of the sample. Indeed, if we want to make inferences on the positioning of the whole Spanish population, it is crucial that we acquire good quality data on the characteristics of users; something that has been proved difficult for online surveys. From, we are working to improve the process of data collection, providing users with the option to sign into an account where they can store their information and reply to surveys at any time. Nevertheless, we believe that more attention from Universities and governments should be given to these tools so that institutions and VAA organizations collectively work to make VAAs a better tool both for users and for the academia. Hopefully, that is what will happen in the next years to come.

For more information on the effect of VAAs on voting behavior, please check my article on Politikon

For inquiries on, please send an email to

  1. El auge de las aplicaciones de orientación del voto y su efecto en el comportamiento electoral, Politikon, June 2016.
  2. A review of the top Voter Advice Applications for the 2015 General Election, LSE British Politics and Policy Blog, April 2015.
  3. “If a citizen cannot pay his/her mortgage, giving the house to the bank should cancel his/her debt”
  4. An “escuela concertada” is a semi-private school that receives money from the government and at the same time charges fees to each student. It is unique to Spain.
  5. Whether underage girls should have permission from their parents by law to be able to have an abortion.
  7. Pianzola (2014), Selection biases in Voting Advice Application research.

Industrial game over: can low-income countries grow through services rather than industry?

June 8, 2016

alumniMarco Antonielli ’12 (International Trade, Finance, and Development) is a consultant with Nathan Associates in London. Prior to this he was a consultant at the OECD in Paris and a research assistant at the Bruegel think tank in Brussels. The following piece by Marco originally appeared on Nathan’s website. (All opinion and analysis are only those of the author.)

Follow Marco on Twitter @AntonielliM and read his blog.

In a global economy with fewer opportunities to industrialize, low-income countries will need to embed the service sector in their vision for inclusive growth.

Amid a gloomy global economic outlook and crashing commodity prices, low-income countries ended 2015 with the slowest growth since 2009, and remain in serious need of new sources of inclusive growth. One major challenge to achieving higher living standards stems from the vast income and productivity gaps within these countries and in relation to the rest of the world.

Large-scale industrialization has traditionally been viewed as the main solution for bridging these gaps, as well as a strategic objective to create jobs and support future growth. Yet latecomers to development may have embarked on a path on which manufacturing—arguably the most promising sector—is expanding slowly in absolute terms, and often shrinking in relation to GDP. The questions are then: why do low-income countries struggle to industrialize? And could alternative sectors such as services replace manufacturing as engines of inclusive growth?

Growing out of the Traditional Economy

Let’s take a step back. While all economies are characterized by varying degrees of productivity and dynamism among sectors and businesses, the low-income countries feature tremendous structural gaps within their economies. Most of the workforce is employed in informal and traditional agricultural businesses, while manufacturing is limited and not fully organized and the dynamic services are largely confined to the cities. Also the modern and formal agricultural businesses are not as widespread as they could be.

To escape poverty, millions of workers need to move from low-productivity sectors and businesses, mainly agriculture, to high-productivity ones, where they will find better and more secure jobs. The reallocation of resources to modern and dynamic sectors can generate positive transformation and help low-income countries achieve inclusive growth.

However, economic transformation can lead to labor and capital being reallocated to more inefficient activities. Recent studies have found that from a macroeconomic perspective, structural transformation (i.e., intersectoral movement of resources) can be a drag on growth for long periods of time, and this is part of the reason why the growth dynamics of low- and middle-income countries have been so diverse. Such a pattern is illustrated in figure 1. Observing the breakdown (“decomposition”) of aggregate productivity growth in the sum of sectoral components and a component accounting for cross-sectoral labor reallocation, it can be noted that between the 1990s and the 2010s Asian and Eastern European countries benefited from the structural transformation of their economies, while Latin American and Sub-Saharan African countries had the opposite experience. Developing countries are therefore not necessarily transforming well over their growth paths.

Figure 1—Decomposition of aggregate productivity growth, 1990–2008

Figure 1

Source: Dabla-Norris et al. (2014)

CESEE: Central, Eastern and Southeastern Europe; CIS: Commonwealth of Independent States; LAC: Latin America and the Caribbean; MENA: Middle East and North Africa; SSA: Sub-Saharan Africa

Organized and modern manufacturing is commonly understood as the business where workers in informal or more traditional forms of agriculture should be reemployed. This is because, while manufacturing is not necessarily the most efficient sector in the economy, it can be a growth accelerator and engine of inclusive growth for at least three reasons. First, manufacturers in emerging economies can benefit from manufacturing technologies developed in more advanced countries, and can achieve fast productivity growth. Second, manufacturing can absorb unskilled labor—thus providing improved employment opportunities for agricultural workers in low-income countries. Finally, manufacturers can export their products, so their growth will not be confined by limited domestic demand. Tradability is key, because high productivity growth can quickly lead producers to lower their prices and shed labor and capital if they cannot scale up their sales in bigger markets.

Is Industrialization a Broken Engine?

Virtually all successful emerging economies in the past 30 years have industrialized by leveraging this potential. Manufacturing offers opportunities to diversify away from agricultural and other traditional products, and helps the country pull itself out of poverty. But is this growth trajectory still feasible for today’s developing countries?

In most countries, the share of jobs and GDP arising from manufacturing expands in the early stages of development, then peaks and starts shrinking as relative prices decline and the economy matures. As Dani Rodrik and others have recently argued, latecomers to development in Africa and Latin America are hitting the peak earlier in the process, and are starting to deindustrialize when manufacturing has exploited only part of its potential. Ghani and O’Connell, for example, explore this inverted-U relationship between the level of economic development and the industry’s share of total employment, in a panel of 100 countries. They show how, in recent times, jobs in industry have grown more slowly and shrunk earlier in the development process (figure 2). The engine of industrialization seems to be running out of steam.

According to Rodrik, this manufacturing decline is mainly due to the adverse effects of trade and globalization on low- and middle-income countries in Africa and Latin America in two respects. First, these countries struggle in the international goods market because of a decline in the relative price of manufacturing in advanced economies, where technological progress has pushed up efficiency and reduced the need for expensive labor. Second, low transport costs and low trade barriers expose them to hyper-cheap production from East Asia, effectively reducing the scope for “import substitution” to expand the boost in manufacturing exports to the wider economy. This would suggest that today’s low-income countries will need to wait until East Asia becomes expensive before they industrialize.

A competing theory is that the low-income countries have subscribed to a trade system that is altogether unfavorable to them. On the one hand, to get access to international markets they are required to forgo protectionist policies that foster import substitution and screen nascent industries from foreign competition during their early development (see e.g., Ha-Joon Chang). On the other hand, trade barriers to advanced markets like the EU are set low for raw materials such as coffee beans and cocoa pods but high for the products obtained from processing of materials—in these examples, roasted coffee and chocolate. This means that the entry points to industrialization of commodity-dependent countries are essentially shut down.

Figure 2—Is Industrialization Running out of Steam?

Figure 2

Source: Ghani and O’Connell (2014) with World Bank data

Help Services

Both theories offer plausible explanations of why low-income countries struggle to industrialize. While more evidence on the causes of the problem is needed, it is increasingly clear that vast-scale industrialization has not featured in the development of most low-income countries. In contrast, the service sector has grown rapidly and absorbed lots of labor. Looking at Sub-Saharan Africa, for example, in the 15 years of this century . This pattern does not adequately represent how low-income countries grow and expand their productive capabilities, at least in that it does not capture the role of the variety and complexity of the products menu offered by these countries. Yet it can raise the question of how services can replace manufacturing as an engine of inclusive development. At least three routes can be identified.

First, there is a fringe of dynamic and tradable services that can boost the economy just as manufacturing does. Banking, customer services, and communications are examples of services which the ICT revolution has opened up to trade, and which can take low-income countries on a growth escalator, as the Indian boom has demonstrated. Crucially, investments in infrastructure, education, and human capital need to be made to facilitate development in these services. An alternative service attracting foreign demand with decent labor-absorption capacity is tourism.

Second, services are crucial inputs to manufacturing and there is evidence that their importance is growing. Hence cheap and efficient services such as transport and telecommunications can translate into stronger competitiveness of the tradable sector—both manufacturing and services.

Finally, the fact that manufacturing and services are becoming increasingly “blurred,” with services activities making up a higher share of manufacturing output, means that low-income countries could exploit a competitive edge on relevant service tasks. Moreover, these tasks can often be unbundled from merchandise production and traded along the global value chain. Logistics, marketing and post-sales services have been on the rise, not only in developed economies but also in developing ones. Furthermore, this trend could lead to a misinterpretation of statistics based on obsolete sector categories, effectively misleading our understanding of structural change.

In sum, the service sector offers new and interesting opportunities for growth, both through tradable services that plug directly into the global economy and through services that support competitiveness of manufacturing. In a global economy with fewer opportunities to industrialize, low-income countries will need to embed the service sector in their vision of inclusive growth and focus on the conditions that enable these opportunities.

Many thanks to my colleagues Joe Holden and Ignacio Fiestas for their helpful comments. This blog first appeared at: 

Can computers see?

June 6, 2016

image recognition

Computers today have the ability to process information from images notably thanks to object recognition. This ability improved greatly over the past few years and reached human levels on complex tasks in 2015. The algorithm allowing them to do such thing is called Convolutional Neural Network (CNN). This method also enabled Google’s AI to beat one of the best GO players in the world and build self-driving cars.

This week, the Renyi Hour invited Dario Garcia-Gasulla from the Barcelona Supercomputing Center who introduced the Data Science students to this method.

You can find the presentation slides here:


Repost from Barcelona GSE Data Scientists blog