class: center, middle, inverse, title-slide # Productivity Differences and Convergence Clubs in Latin America ### Carlos Mendez<br /><a href="https://carlos-mendez.rbind.io" class="uri">https://carlos-mendez.rbind.io</a><br /><br />Associate Professor<br />Graduate School of International Development<br />Nagoya University<br /> JAPAN ### Prepared for the 2020 Bolivian Conference of Development Economics <br /><br />[ Slides, paper, and a video presentation are available at: <a href="http://bit.ly/bcde2020" class="uri">http://bit.ly/bcde2020</a> ] --- class: center, middle <style type="text/css"> .highlight-last-item > ul > li, .highlight-last-item > ol > li { opacity: 0.5; } .highlight-last-item > ul > li:last-of-type, .highlight-last-item > ol > li:last-of-type { opacity: 1; } </style> This paper is **a supplementary resource** of the forthcoming book:  --- class: middle,center # A summary of the paper in 2 slides... --- class: highlight-last-item ## Motivation: - Inconclusive literature about Latin America: **Convergence vs Divergence vs Convergence Clubs** (Galvao and Reis-Gomes, 2007; Barrios et. al, 2018; Martin and Vazquez, 2015) -- - Development potential of Latin America constrained by **low productivity** (Daude and Fernndez-Arias, 2010; Pages 2010; Restuccia, 2013) -- ## Research Objective: - (Re)evaluate the **convergence hypothesis across economies** in Latin America with particular **emphasis on productivity** differences and the formation of **multiple convergence clubs**. -- ## Methods: - Nonlinear dynamic factor model (Phillips and Sul, 2007, 2009) -- - Clustering algorithm for panel data (Phillips and Sul, 2007, 2009) -- ## Data: - Labor productivity and total factor productivity (Fernandez-Arias, 2017) -- - 20 Latin American countries over the 1980-2014 period -- --- class: middle, highlight-last-item ## Main Results: 1. **Lack of overall(global) convergence** in both labor productivity and total factor productivity -- 2. **Multiple local convergence clubs:** above and below the average -- 3. **Convergence clubs characteristics:** - Labor productivity: **Four clubs** of countries - Total factor productivity: **Three clubs** of countries - Clubs show non-parallel trends: crossings, limited stability, and separating trends - The lowest-productivity club (Honduras and Nicaragua) is diverging from the rest **at the highest speed**. --- class: middle # Outline of this presentation 1. Some stylized facts - Large and heterogeneous productivity differences across Latin America 2. Convergence framework - Global convergence test (intuition) - Local convergence clubs (intuition) 3. Main results of the paper - Lack of overall convergence - Multiple convergence clubs above and bellow the average - Convergence clubs characteristics <br /> <br /> [ Slides, paper, and video presentation are available at: http://bit.ly/bcde2020 ] --- class: center, middle # (1) Some stylized facts Large and heterogeneous productivity differences across Latin America --- class: middle,center ## Large and heterogeneous productivity differences across Latin America  Note: Labor productivity is computed as the long-run trend of (log) GDP per worker. The Hodrick-Prescott filter with a smoothing parameter of 6.25 is applied to obtain the long-run trends. --- class: middle,center ## Large and heterogeneous productivity differences across Latin America  Note: Total factor productivity is computed by dividing GDP per worker by an aggregate index of physical capital and human capital. The Hodrick-Prescott filter with a smoothing parameter of 6.25 is applied to obtain the long-run trends. --- class: center, middle # (2) Convergence framework Global convergence test (intuition) Local convergence clubs (intuition) --- class: middle # Convergence framework (brief overview) - First, define **a relative transition parameter** as `$$h_{it}=\frac{y_{it}}{\frac{1}{N}\sum_{i=1}^{N}y_{it}}$$` - Second, **convergence is defined as** `$$H_{t}=\frac{1}{N}\sum_{i=1}^{N}\left(h_{it}-1\right)^{2}\rightarrow 0$$` In other words, when the relative transition parameter converges to unity, `\(h_{it}\rightarrow1\)`, the cross-sectional variance converges to zero, `\(H_{t}\rightarrow0\)`. - Third, evaluate convergence using **the model of Phillips and Sul (2007)** `$$log\left(\frac{H_{1}}{H_{t}} \right)-2log\left\{ log\left(t\right)\right\} = a+b\:log\left(t\right)+\epsilon_{t}$$` --- class: middle, center # Convergence test (intuition)  --- class: middle, center # Convergence clubs (intuition)  --- class: middle, center # (3) Main results Lack of overall convergence Multiple convergence clubs above and below the average Convergence clubs characteristics --- class: middle, center ## Lack of overall convergence  --- class: middle, center ## Multiple convergence clubs  --- class: middle, center ## Multiple convergence clubs: Above and below the average  --- class: middle, center ## Convergence clubs characteristics: Labor productivity  --- class: middle, center ## Convergence clubs characteristics: Total factor productivity  --- class: highlight-last-item # Concluding Remarks - **Reject (overall) convergence** for both labor productivity and total factor productivity -- - Multiple **convergence clubs below and above the mean** -- - The clubs show **different convergence speeds** and **separating tendencies** - The poor economic performance of Honduras and Nicaragua is driving the separation of clubs over time. -- ## Implications and further research -- - Convergence clubs may help us **identify economies facing similar challenges** -- - Call for **better coordination and cooperation policies** both within and between clubs - International technology transfer initiatives. -- - Masked behind the low productivity of Latin America, there is **still a high degree of heterogeneity** that is worth exploring - Next extension: Re-evaluate the composition of **convergence clubs using subnational information** from satellite nightlight data. --- class: center, middle # Thank you very much for your attention https://carlos-mendez.rbind.io Slides and working paper available at: http://bit.ly/bcde2020  **Quantitative Regional and Computational Science lab** https://quarcs-lab.rbind.io *** This research project was supported by JSPS KAKENHI Grant Number 19K13669