Economic Research Seminar 2019
Monday, December 2nd, 2019
Inflation expectations embedded the Costa Rican sovereign debt market: are they anchored? [PDF] (Spanish)
(Carlos Segura Rodríguez)
The Central Bank requires measurements to assess its credibility, from which one of the most relevant is inflation expectations of economic agents. This study uses a linear model of interest rates on sovereign bond transactions issued by the Ministry of Finance and the Central Bank, as alternative and complementary measures of inflation expectations. These measurements are shown to be better forecasters on inflation than the expectations collected from the Expectations Survey of the Central Bank of Costa Rica (BCCR). The proposed methodology is further extended to obtain and evaluate expectations on exchange rate variations at different horizons.
Finally, this study evidences that during recent years, long-term inflation expectations (5 years) are anchored – that is, expectations do not react to conditions considered as transitory. However, the anchored value is slightly below the inflation target defined by the BCCR.
Forecasting inflation in Costa Rica using machine learning methods [PDF] (Spanish)
(Adolfo Rodríguez Vargas)
Machine learning methods have increased in popularity as a forecasting tool due to the increasing availability of large databases, computing power, and access to specialized software. Their use is more widespread in discrete choice variable classification problems, such as prediction of default on loans or consumer purchase decisions, in which they have proven to be of great help compared to traditional methods. However, machine learning methods can also be adapted to predict continuous temporal variables, such as inflation or real output growth. This study carries out a first comprehensive evaluation of the performance of machine learning methods in forecasting inflation for Costa Rica. To do so, it verifies that its forecasts comply with the properties of optimal forecasts under quadratic loss, and if its performance is superior to alternative methods currently in use at the BCCR.
Forecasts are calculated using two variants of K-Nearest Neighbors (KNN), random forests and extreme gradient boosting, and a long short-term memory (LSTM) model. The results were promising: It was found that the forecasts with the best performance are those resulting from applying LSTM, univariate KNN and random forests. Additionally, a combination of these forecasts improves performance against each individual forecast at all horizons and outperforms the average of the univariate forecasts.
Endogenous asymmetric pass-through from exchange rate to prices
(Manfred Esquivel Monge and José Fabio Gómez Rodríguez)
A multivariate endogenous regime change model is proposed to study the Exchange-rate pass-through (ERPT) in the Costa Rican economy. By proposing a multivariate approach, the study extends the theoretical framework proposed by Chang, Choi, and Park (2017). A vector autoregression (VAR) model of changing regimes is estimated where the probability of change between regimes is endogenously generated.
It was found that the ERPT is not constant over time, which is consistent with empirical literature that uses different econometric methods and other data samples. Furthermore, there is evidence that the non-linearity of the ERPT is associated with both shocks and differences in the joint volatility of inflation, exchange rate (nominal and real), and level of dollarization of the economy.
Methodologies for estimating the equilibrium real exchange rate [PDF] (Spanish)
(Cristian Álvarez Corrales, Juan Diego Chavarría Mejía, Jorge León Murillo and Adolfo Rodríguez Vargas)
The real exchange rate is a crucial relative price in the economy, as it shows the relationship between the prices of tradable and non-tradable goods; it serves as guide for the allocation of productive resources between both sectors. The equilibrium real exchange rate (ERER) is the level of the real exchange rate that prevails when the economy is in internal equilibrium, with production at its potential level and the absence of inflationary or deflationary pressures, and in external equilibrium, when the current account of the balance of payments is financed with long-term external capital flows.
The estimate of the ERER allows for the explanation of its movements based on the evolution of its determinants (fundamental macroeconomic variables) in the medium and long run. The BCCR has used a diverse set of methodologies to approximate a zone of coherence with equilibrium for the RER. This presentation describes in detail these methodologies, which comprise two versions of Behavioral Equilibrium Exchange Rate (BEER): the External Balance Assessment (EBA) of the International Monetary Fund, the Fundamental Equilibrium Exchange Rate (FEER) and Desired Equilibrium Exchange Rate (DEER).
Cost-benefit analysis of public investment in general equilibrium [PDF] (Spanish)
(Onil Banerjee and Martín Cicowiz)
This work shows how IEEM (Integrated Economic - Environmental Modeling) can be used to carry out a cost-benefit analysis of public investment. This paper analyzes hypothetical investments in different types of tourism in Costa Rica (e.g., health tourism, education, leisure, etc.), in order to capture the direct and indirect effects of shocks on this sector.
Energy substitution in transportation in Costa Rica: macroeconomic and environmental impacts [PDF] (Spanish)
(Irene Alvarado Quesada and Mónica Rodríguez Zúñiga)
Given the commitment acquired by Costa Rica within the Paris Agreement, the country undertook the pledge to reduce greenhouse gas (GHG) emissions in 2030 by 25% when compared to 2012. Given that the largest proportion of emissions and use of energy in the country are associated with the consumption of fossil fuels for transportation, the proposed policies are oriented towards the substitution of clean energy sources, as well as energy efficient equipment and machinery.
The purpose of this research is to study the impact of an incentive for the use of electric vehicles in transport for the period 2012-2030. In the model, this impact is reflected by two simultaneous shocks: a substitution of the use of electricity for that of fossil fuels in transportation, and an improvement in energy efficiency.
Macroeconomic effects of a reduction in the single/flat fuel tax [PDF] (Spanish)
(Diego Agüero Morera)
In Costa Rica, the Tax Simplification and Efficiency Law establishes a flat tax for each type of fuel; the rates of this tax are applied to the sales per liter of the different fuels.
Costa Rica has the highest fuel prices in Central America, which is mainly by the high tax levied on these oil derivatives, which are also proportionally the highest in the region, in some cases making the tax the largest component of the final sale price.
The objective of this investigation is to determine the possible macroeconomic effects of a reduction in the flat fuel tax, both on main macroeconomic variables as well as on government tax collection and CO2 emissions, and to also determine if such a tax reduction would have the ability to stimulate the economy without generating additional pressure on the government budget.
Modeling the effects of rice market liberalization: preliminary results [PDF] (Spanish)
(Jorge Leon Murillo)
In Costa Rica, the government has the power to regulate prices under exceptional conditions and on a temporary basis. But, rice is an exception, as it is the only good whose price is regulated by the State in the entire agricultural chain. Economic theory suggests that the liberalization of prices in a market allows the forces of supply and demand to define the price of the good, thus contributing to a more efficient allocation of the economy’s resources, leading to an improvement in the general well-being of society.
The research carried out makes it possible to measure the effect that liberalizing the rice market would have on the main macroeconomic variables, as well as the effects of such liberalization on consumption and employment. The methodology makes it possible to calculate the effects on the rice sector at the regional and canton level.
The results presented are preliminary in nature, and subject to revision and updating.
Tuesday, December 3nd, 2019
Effects of becoming a supplier to a multinational corporation: new evidence of linkages between companies [PDF] (Spanish)
(Alonso Alfaro Ureña, Isabela Manelici and José Pablo Vásquez Carvajal)
Many countries around the world compete to attract Direct Investment (DI), usually in the form of affiliates, subsidiaries or headquarters of Multinational Corporations (MNCs). To assist in this competition, public policies including tax exemptions and subsidies for industrial infrastructure are established, aimed at increasing the attractiveness of the country as a destination for this type of investment.
One of the motivations for creating these policies lies on the assumption that the entry of highly competitive companies will generate a positive impact on the host economy as it improves the performance and productivity of domestic firms. Direct supply chain linkage is one of the channels that allow technology and knowledge transfer between MNCs and national companies.
Application of Markovian models and the Kalman filter to identify the economic cycle in Costa Rica [PDF] (Spanish)
(Manfred Esquivel Monge)
Two methods are implemented to identify the economic cycle in Costa Rica, and based on its identification, an appropriate smoothing parameter is approximated to apply the Hodrick-Prescott (HP) Filter to the economic activity indicator series in Costa Rica.
The first method, originally proposed by Stock and Watson (1991), consists of inferring an unobservable time series whose data generating process is stationary and autoregressive (the cycle), which determines the behavior of observable series indicating economic activity. The other method, which is based on the proposal of Hamilton (1989), consists of estimating the probability that the economy is in an expansionary or recessive phase using a Markovian model of regime change.
Both methods are characterized by not having ad-hoc assumptions about the way in which the moments of the distributions of the cyclical and trend components of the series are related. This makes them attractive as appropriate methods for independently estimating smoothing parameters for the HP Filter in applications using production data from Costa Rica.
The evolution of labor income and inequality in Costa Rica: evidence with micro data [PDF] (Spanish)
(Alonso Alfaro Ureña, Isabela Manelici, Luis Alfredo Mendoza Fernández and José Pablo Vásquez Carvajal)
In this investigation, an employer-employee linked micro-database was used to study wage inequality in the formal Costa Rican private sector during the period 2006-2017. Evidence was found that real income of all percentiles of the wage distribution had increased by at least 14%, while during the same period its distribution changed shape; the larger relative increases were found among those with the highest wages.
There is also evidence that the augmentation in the share of employment by multinational companies in total employment was one of the variables that could explain the observed behavior, as was the structural change towards a service economy that the Costa Rican economy has experienced. Finally, a variance decomposition is presented that shows how the effect of workers (as opposed to that of companies) contributes about 60% of the change in inequality. Part of this is explained by the entry of multinationals into Costa Rica.
Forecasting economic activity in real time [PDF] (Spanish)
(Kerry Loaiza Marín)
The relevance of real-time forecasting of the economy arises because statistics on key variables are generally presented with a lag. In Costa Rica, for example, the process of generating quarterly GDP data makes it available one quarter after the reference period. This is a significant limitation for economic authorities who need to assess the state of the economy in real time for decision making.
The principle of real-time forecasting has become widespread in the literature as a solution to this problem. It is based on the intensive use of information published in advance, which also has a higher temporal frequency than the variable of interest (for example, the Monthly Index of Economic Activity). This approach seeks to obtain early estimates before the release of the official data. Rodríguez (2014) proposed a methodology that was used as the starting point for the present study.
Design of a leading indicator for economic activity [PDF] (Spanish)
(Silvia González Brenes and Carlos Monge Badilla)
The importance of leading indicators has been recognized for several decades, because anticipating the direction of economic activity provides relevant information for policy makers and private agents. They can also be used to complement short-term production projections.
In the construction of compound cyclical indicators, two types of methodologies are particularly notable: i) the classical method and ii) factorial models. The classical method follows a series of steps that include variable preselection, series filtering (seasonal adjustment), turning point determination, variable selection, and aggregation. For its part, the factorial method assumes that there are common relationships between variables because they are expressions of factors that are not directly “observable.” This is similar to the classical method except for the filtering the series, since when using this methodology the composite indicator is constructed from the common and idiosyncratic components of the series with greater commonality.
Dynamics of the labor market in Costa Rica [PDF] (Spanish)
(Alonso Alfaro Ureña, Santiago Campos Rodríguez and Valerie Lankester Campos)
Stylized facts of formal employment are presented using firm-level data. The results suggest that the neoclassical theory of employment meets in a heterogeneous manner between economic sectors in Costa Rica; for example, wages are a stronger determinant of labor demand in manufacturing and construction. Furthermore, the size of a firm is a relevant variable. For larger firms, the level of employment is more persistent, and they adjust their payroll to a greater extent to changes in wages and income. Compared to similar economies, Costa Rican labor demand is more sensitive to changes in wages but less to changes in production.
Wednesday, December 4th
Dynamic effects of fiscal policy in Costa Rica [PDF] (Spanish)
(Valerie Lankester Campos and Kerry Loaiza Marín)
The debate among economists about the impact of fiscal policy on economic growth was revived after the financial crisis of 2008. Different governments decided to implement stimulus programs even though there is no consensus in the empirical literature on the effectiveness of fiscal policy to boost growth. The discussion is clearly framed between two lines of argument of economic thought: neoclassical and neo-Keynesian.
This research aims to provide evidence on this debate by characterizing the effects of fiscal policy in an emerging economy such as Costa Rica by estimating its fiscal multipliers. Knowing the magnitude, direction and conditional asymmetry in the economic cycle from an impact of an exogenous shock in government spending or income on determines the level of effectiveness of fiscal policy, which is essential for the discussion of public policies.
Analysis of fiscal sustainability for Costa Rica [PDF] (Spanish)
(Valerie Lankester Campos, Kerry Loaiza Marín and Carlos Monge Badilla)
The political discussion for the approval of the fiscal reform in Costa Rica became more relevant after extensive fiscal changes were part of the measures adopted in response to the financial crisis of 2008. However, it took almost a decade more for a law to be passed. What is the impact of this law in terms of debt sustainability?
The analysis in this investigation methodologically begins with the inter-temporal budget restriction, which is complemented by the estimation of the fiscal reaction function and evaluates different risk scenarios with the fan charts method. Using data from 1974 to 2018, the estimated reaction function makes it possible to define the reference value for observed short- and long-term sustainability and performing a medium-term (2020-2023) analysis of relevant variables for fiscal sustainability.
Crypto assets: analysis and implications from the perspective of the BCCR [PDF] (Spanish)
(Alonso Alfaro Ureña and Evelyn Muñoz Salas)
Although the beginnings of the assets called crypto currencies lie in the search for a mechanism to carry out transactions that would allow financial system entities to be dispensed with, their development and global expansion has been facilitated by a growing interest in new ways of accumulating value. This has been combined with advances in cryptography research, more widespread internet access, and the increased processing capacity of computers.
Some characteristics of these assets allow their use as a means of exchange and storage of value, uses that are traditionally associated with money issued by central banks. This can be a source of confusion for the end user about their potential management and the risks that their adoption implies for personal finances. This technical note therefore seeks to facilitate the understanding of the technological processes involved in the creation of these assets, and describes the possible implications associated with their use.