Work in progress
Barriers to mobility or sorting? Sources and aggregate implications of income gaps across sectors in Indonesia (with T. Święcki)
Revision requested at American Economic Journal: Macroeconomics
The existence of large income gaps between agricultural and non-agricultural workers in developing countries is well known. However, the source of these gaps is still being debated and the two main hypotheses – barriers to labor mobility and sorting of workers based on unobserved productivity – have opposing implications for aggregate efficiency. We use a panel of Indonesian workers to move beyond the cross-sectional gaps and document that workers moving out of agriculture see income gains of over 20% while those moving into agriculture see similar income losses, with large flows of workers in both directions. To interpret these findings, we structurally estimate a model featuring both sorting and barriers to sectoral mobility. Our estimates indicate that while self-selection is important, there is more misallocation than is suggested in the recent literature. Removing mobility barriers would lead one third of workers to reallocate and would increase aggregate output by as much as 21%.
Intra- and inter-industry misallocation and comparative advantage
Micro-level resource misallocation, both within and across industries, can affect the relative unit costs of production across sectors, distorting comparative advantage. After presenting evidence on how changes in factor misallocation of a particular country (Colombia) relate to the dynamics of its revealed comparative advantage, I use a mis-allocation model with international trade to evaluate how its specialization patterns would change if resources were used efficiently. The new specialization would allow Colombia to raise its ratio of exports to manufacturing GDP by 18 pp. This industrial composition effect is absent in the workhorse models of misallocation under closed economies.
Misallocation of the immigrant workforce: Aggregate productivity effects for the host country (with A. Varón)
Mass migrations can impact the amount of labor misallocation in the host country if immigrants, relative to natives, face more frictions that prevent them from working in their preferred occupations. The resulting misallocation would imply an aggregate productivity loss in the short run while migration occurs, but an subsequent lapse of productivity growth when the immigrants start to be assimilated by the labor market. We study the case of Colombia during 2015-2019, a period when the country received a massive inflow of migrants from Venezuela. Through the lens of a Roy model of occupational choice with two types of frictions – discrimination and barriers preventing workers from choosing their preferred occupations – we quantify the extent of occupational misallocation for immigrants, and its implications for Colombian aggregate labor productivity. Our estimates indicate that both type of frictions significatively misallocate Venezuelan immigrants. Removing those frictions would lead around one third of immigrants to reallocate, permanently increasing Colombian aggregate productivity by 0.9%.
Effects of the Covid-19 pandemic on the Colombian labor market: Disentangling the effect of sector-specific mobility restrictions (with L. Morales, L. Bonilla., L. Flórez, D. Hermida, K. Pulido and F. Lasso)
We assess the effect of the Covid-19 pandemic and the lockdown of some economic sectors on the Colombian labor market. We exploit the variation between excluded and non-excluded sectors from the lockdown, as well as the timing of the restriction policies, to identify the effect of sector-specific restriction policies. These restrictions had negative effects on employment, accounting for approximately a quarter of the total job loss between February and April of 2020. Therefore, we should expect important employment losses even in the absence of such restrictions. The regional patterns of the disease spread and other epidemiological and economic factors affecting the country during this period, account for the remaining three quarters of the job losses. In contrast, we find no significant effect of sector-specific restrictions on average worked hours or wages, indicating that most of the adjustment of the labor market took place in the extensive margin. Moreover, sector-specific restrictions only affect salaried workers, while self-employment is more responsive to the disease spread. These findings suggest that labor market rigidities may amplify the impact of mobility restrictions.
Welfare effects from financial innovations: The case of banking through networks of retail agents (with F. Arias)
Financial innovations in developing countries are increasingly reducing the cost of serving customers in remote locations where they otherwise would be financially excluded. We study the welfare implications of the business correspondent model in Colombia, a system in which financial intermediaries provide banking services through third-party non-financial commercial establishments. To characterize the system we use a model of multi-product heterogeneous banks with binary technologies that serve multiple locations with different demand attributes. We structurally estimate the model using a unique dataset that records monthly transactions by financial product at the bank-municipality level over 8 years. A counterfactual exercise in which the low-cost technology is suppressed allow us to quantify the welfare impact of the presence of banking correspondents. Our findings suggest that networks of correspondents deepens bank penetration and financial inclusion, reduces the cost of financial services, boosts banks revenues and increases total welfare by 4%.
Alfonso, V., Arango, L., Arias, F. Cangrejo, L. and Pulido, J. (2013) “Business cycles in Colombia: 1975-2011″. Lecturas de Economía, No. 78, pp. 115-149, Universidad de Antioquia.
We propose a chronology for the business cycles in Colombia following the NBER classic notion; that is, dates of peaks and troughs of economic activity are estimated without decomposing the series used in their transitory and permanent components. The estimated chronology suggests that the four complete cycles that occurred between 1975 and 2011 are asymmetric and have an approximate duration of 6.8 years. Expansions lasted, on average, 5.4 years while contractions took about 1.3 years. These results are derived from the application of a cumulative diffusion index to 41 series of economic activity.
Kamil, H., Pulido, J. and Torres, J. (2010) “IMACO”: a monthly leading indicator of economic activity in Colombia. Monetaria, Vol. XXXIII, No. 4, pp. 495-598, CEMLA – Center for Latin American Monetary Studies.
This paper describes the construction of a new monthly leading indicator of economic activity in Colombia (IMACO). The procedure is based on a search algorithm that selects an optimal group of seven leading variables so that the composite indicator anticipates GDP movements five months in advance and with a 93 % correlation. Also, the IMACO has others desirables predictive properties: it anticipates the break points in the Colombian business cycle without giving any wrong signals, and minimizes the forecast errors on GDP growth. Due to its simplicity and low computational cost, the IMACO indicator offers a tool for the continuous monitoring of the economic activity and economic policy design, and can be replicated for other macroeconomic aggregates in Colombia and also applied in other Latin-American countries.
Cristiano, D., Hernandez, D. and Pulido J. “Nowcasting Colombian economic activity”. Borradores de Economía, No. 724, Banco de la República.
Policy makers usually require estimates of the performance of economic activity in real time. However, the information used is only available at the level of hard indicators and opinion polls, which usually have different frequencies and publication lags, in addition to idiosyncratic shocks. In this paper, we adapt Camacho and Perez-Quiros (2009-2010) forecasting schemes that produce estimates of GDP growth in real time for the Colombian economy. The adapted dynamic factors model involves economic series of different frequency, availability and origin, used with the information available at the time of each publication. The forecast evaluation suggests that the model presents a better performance against other reference schemes, and that the accuracy of the forecasts increases when we include the flow of information in real time of the activity indicators.
Alonso, G., Hernandez, N., Pulido, J. and Villa, M. (2009) “Alternative measures of real exchange rate for Colombia”. Borradores de Economía, No. 514, Banco de la República.
To be added.