Costa Rica: Selected Issues and Analytical Notes

International Monetary Fund. Western Hemisphere Dept.

doi:10.5089/9781475527025.002.A001

Jump to Content Jump to Main Navigation

Costa Rica: Selected Issues and Analytical Notes

Author: International Monetary Fund. Western Hemisphere Dept.

Publication Date:: 04 Feb 2015

eISBN:: 9781475527025

Language:: English

Keywords:: ISCR; CR; Gdp; staff analysis; error terms; mean reversion; Intel exit; Phillips curve; Potential output; Output gap; Labor force; Central America

Download PDF (598.4 KB)

In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.

Abstract

I. Selected Real Sector Issues¹

This note examines several real sector issues, including estimates of potential output, the effect of Intel’s withdrawal on GDP, labor market and inequality and electricity prices in Costa Rica. Estimates suggest that potential GDP growth is about 4.3 percent, the output gap is broadly closed and Intel’s withdrawal will lower real GDP growth in about ½ percentage point. Significant wage premia are identified across public versus private sectors and some evidence of intergenerational inequality is also presented. Electricity tariffs are found to be regionally competitive albeit with inefficiencies in their determination.

A. Potential Output Estimates

1. Staff analysis suggests that on average Costa Rica’s potential output growth is about 4.3 percent and the output gap is closed. Results are relatively robust across different methodologies which include a production function approach, several well-known univariate time series filters, univariate Kalman filters and multivariate filters which take into account inflation and financial variables. For the period 1999–2008, before the financial crisis, Costa Rica’s potential output grew at an average rate of 4.7 percent. The analogous estimate for the post-crisis period is, as expected, lower, at 4.1 percent but recovering moderately in recent years. The output gap was estimated to have broadly closed in 2013 (-0.3 percent) but it somewhat widened in 2014 to -0.6 percent.

2. The production function approach shows that the main drivers of fluctuations in GDP growth are TFP and labor supply. Results show that potential output grew at an average rate of 4.5 percent in 1999–2013, one of the highest in Central America. While contributions from capital remained relatively stable since 1991, most of the changes in GDP growth are driven by changes in productivity (TFP) and human capital weighted labor supply. Contrary to other countries in Central America, productivity growth in Costa Rica has been large and positive across several years with labor generating significant positive contributions during downturns when TFP contribution was negative.

3. These results on TFP, however, should be interpreted with caution. The TFP measure is by definition a residual—the difference between output growth and the growth in the quantity (and quality) of inputs. Thus, any measurement errors in the labor and capital series are automatically attributed to TFP. For instance, employment shifts from the formal to the informal sector, migration of skilled labor, changes in the quality of the capital and labor stocks which are not correctly accounted for, and changes in the level of capital utilization and the use of land would be reflected in TFP.

Table 1.

Potential Output Growth and Output Gap Estimates

Source: Fund staff estimates.

^{^*/}Estimate for 2013.

4. Estimates with cycle extraction filters and univariate Kalman filters suggest that potential output growth is 4.2 percent and 4.7, respectively. Although cycle extraction filters have several shortcomings such as the inability to capture structure changes in the economy, and should be taken with caution, it is nevertheless reassuring that most of the methods point to an estimate for potential growth of about 4.1 percent in 2014. The CF filter, which allows for phase shifts (and is thus more flexible), deviates the most (4.5 percent). A similar effect takes place with the univariate Kalman filters where the model which allows for mean reversion (and is thus more flexible) estimates a higher potential output (4.9 versus 4.5 percent)

5. Estimates using multivariate filters that include inflation and financial variables suggest that potential output growth is about 4.2 percent. These multivariate filters consider the information in inflation (through a Phillips curve, for example), financial variables (e.g. stock prices, interest rates, exchange rates, credit levels) and government deficit to identify whether growth is above potential.² For both the periods before and after the crisis, the results of these multivariate filters are within the range of the estimates examined above and in line with the overall average of all methods.

B. Estimating the Impact of Intel Exit

6. Two different approaches were used to estimate the impact of Intel’s exit on real GDP. This exercise presented some difficulties due to the fixed and outdated base year of the National Accounts (1991) used to derive GDP and its components. This meant that 1991 constant price weights are used to derive Intel’s contribution to real GDP growth in 2014, greatly overestimating it. This approach would have implied a negative contribution to growth of 6.4 percent due to Intel’s relocation, which is clearly overestimated in light of the significant drop in silicon chip prices in recent years and the fact that Intel’s value added relative to GDP was estimated to be around only 0.5 percent in 2013. Staff instead used two other approaches to obtain more meaningful estimates.

7. In the first approach, annualized short-term indicators and chain-linked volume measures generated an estimated Intel contribution to growth of 0.64 percent. Using a large number of high frequency price indicators for all the industries (mostly monthly) staff derived a set of consistent annual price indices for 2013–14 in order to compute 2014 GDP at constant 2013 prices.³ These price series, combined with BCCR’s estimates for GDP components and Intel’s value-added in 2013 were then used to obtain corresponding forecasted values for each GDP component (and Intel) in 2014.

8. The second approach employing the 2011 input-output table compiled by the BCCR produced a slightly higher estimate. Using the input-output matrix, a Leontief demand model was used to estimate the effect of the change on final demand (through lower Intel exports) on total output of all industries, and thus GDP.⁴ Using this model, real GDP is estimated to be 0.74 percent lower after Intel’s complete manufacturing withdrawal from Costa Rica. Thus, the two methods give similar estimates for the impact of Intel’s relocation of its manufacturing activity on Costa Rica’s real GDP. While the first method’s accuracy depends on the number of estimates needed to complete monthly prices data for year 2014, the second method’s accuracy depends on the proximity of the reference year of the input-output table.

C. Labor Markets and Inequality

9. Although Costa Rica’s labor force is very well educated, there is significant labor market inequality along gender, age and education divides. Participation rates among women are low, unemployment disproportionately affects the uneducated and the young, and a there are significant gender and skill wage gaps. Furthermore, there is also evidence of a public sector wage premium, even after controlling for education and gender.

10. While labor force participation (LFP) rates for men are on par with LAC, they lag severely behind for women. With a female LFP rate of only 46 percent, Costa Rica lags behind not only the LAC average of 54 percent but it is the country in the region with the second lowest female LFP rate, ahead of only Honduras. In survey responses, nearly one-third of the women out of the labor force report child-care and elderly-care responsibilities as the main driver for their choice. According to ILO, women’s work is a key poverty-reducing factor in developing economies. Higher female LFP can not only mitigate the impact of population aging, but has also been shown to contribute to overall development through improved intergenerational gender equity (e.g. higher school enrollment for girls).

11. Overall unemployment rate is relatively high at 7.7 percent, although there are differences across gender and age. Costa Rica has the third highest unemployment rate in the region after Dominican Republic (14.7 percent) and Nicaragua (8.0 percent). However, one should take care when interpreting unemployment statistics in the region, since countries like El Salvador, Guatemala and Honduras have large informal sectors (nearly two thirds of the labor force) which distort statistics. The informal sector is relatively small in Costa Rica, and is estimated to be about 30 percent, affecting mostly rural areas where 62 percent of those working informally live. Unemployment rates are also higher for women (similar to most of the region) and mostly affect the young and the less educated: over 60 percent of those unemployed are under the age of 30 while nearly 90 percent of those unemployed have less than 13 years of education.

12. Costa Rica’s unemployment rate is estimated to increase in the future. Two key determinants of the evolution of unemployment rates are (i) the elasticity of employment to GDP growth and (ii) population growth. Using a simple panel regression, staff estimated an employment to GDP elasticity of 0.46 for Costa Rica (similar to estimates for overall Latin America found in the literature) which although it is not as high as that of Dominican Republic (0.98) or Nicaragua (1.2), it is much higher than that of for example, El Salvador (0.13). Intuitively, this relatively high elasticity is in line with the production function decomposition in GDP growth which showed a relatively large contribution of labor to GDP growth. Using this estimated elasticity and under baseline growth projections, unemployment is estimated to reach 8.8 percent by 2019. Under an alternative scenario where growth would reach 6.3 percent in 2019, the unemployment rate would drop to 5.3 percent.

13. The labor force is comparatively well educated. About 60 percent of Costa Rica’s labor force has at least secondary education, compared to the LAC average of 40 percent. Nearly one quarter of Costa Rica’s labor force has tertiary education, making it the only country in the region which exceeds the LAC average of 20 percent. Furthermore, education levels appear to have been increasing across cohorts. For example, in 2012, amongst those aged 36–40, only 30 percent had completed 10 years of education, compared to 53 percent of those in the cohort aged 21–25.

14. However, wage inequality along schooling and gender lines is also large. Costa Rica has a significant “skill premium”: the average labor income for those with more than 13 years of education is nearly double of those who have less than 13 years of schooling.⁵ This statistic is the same for both genders and in both the private and public sectors. This is relatively high compared to the skill premium observed, for example, in the US of about 80 percent. On the other hand, the ratio of female to male income in Costa Rica is similar to the one observed in advanced economies (0.66). This “gender gap” is much smaller in the public sector—where the ratio of women’s to men’s income is about 0.80—than the private, where the ratio is 0.64 after controlling for education.

15. The public sector wage premium is positive and very high in Costa Rica. For those with less than 9 years of schooling, a man can earn about 60 percent more in the public sector (compared to private) while his female counterpart can earn most than twice as much in the public sector. For higher levels of education, these premia drop to about 50 percent for men and 80 percent for women, which are nevertheless very high. As a comparison, at the global level, public sector wages are about 20 percent lower than those of the private sector.

16. There are also signs of persistent intergenerational inequality. Children from wealthier households and with better educated fathers tend to have much higher levels of schooling. Examining across income levels, for example, while 80 percent of children from households in the top quintile of income complete at least 9^th grade, only 52 percent of those in the lowest quintile household do. Furthermore, the schooling completion rates of children from the third quintile are closer to those of the bottom quintile, indicating a very different behavior at the top of the distribution. Across education levels of the household head, the difference is also striking: while nearly 90 percent of children of fathers with more than 11 years of schooling complete at least 9^th grade, only 40 percent of those from households with less than 6 years of schooling do.

D. Electricity sector

17. Compared to the region, Costa Rica has a relatively well performing electricity system. The service quality is high (nearly complete coverage, limited interruptions, low technical losses), prices are relatively competitive (industrial tariffs are the lowest in the region), and it has an adequate regulatory and policy framework. The generation matrix is the greenest in the region, with 66 percent coming from hydro, 12 percent from geothermal and wind and the rest coming from thermal energy. Although Costa Rica’s hydro potential remains significant, it is estimated that, by 2040, its growth potential will be severely limited. There is also an immense potential for solar energy, which currently remains untapped in part owing to its significant variability.

18. The Costa Rican Electricity Institute (ICE), a state-owned enterprise, largely controls generation, transmission and distribution. ICE has a monopoly on electricity transmission and controls 79 percent of distribution, with co-ops and other small state-owned companies distributing the rest. Until 1990, ICE also had a monopoly in generation. Since then, the share of private sector participation in generation has been growing; private sector generation now stands at around 17 percent, and is expected to continue increasing as ICE faces capacity expansion constraints. The President has also declared that he is open to higher private sector participation if it leads to lower tariffs.

19. Electricity tariffs, in real terms, have been steady over the last decade. However, an increase of over 30 percent in 2013 (after 3 years with no increases) generated an outcry over electricity prices. This increase was due to a need to “catch-up” combined with other factors notably (i) increased demand, (ii) higher distribution costs and (iii) increased reliance on costly thermal generation due to seasonal fluctuations in rainfall. Owing to its large reliance on hydro, Costa Rica needs either significant excess capacity or more cost-efficient alternatives to thermal in order to safeguard generation during dry periods.⁶ ICE has recently made large capital investments (in particular in thermal generation) in order to ameliorate this situation but these have drawn criticism about excessive use of debt financing on unfavorable terms. ⁷ In May 2013, ICE issued USD500 million of debt with 30 years maturity in order to improve its debt profile. Currently, only about one third of its debt comes due in the next 5 years.

20. Although prices are regionally competitive, they are high relative to other competitors, and exhibit inefficiencies in their determination. Malaysia, China, Japan, Korea and OECD countries all have more efficient systems with fewer technical losses. In terms of industrial tariffs, Malaysia, Korea, China and OECD countries also have lower rates owing to fiscal subsidies, which ICE does not receive. In addition, Costa Rica has relatively similar average residential and industrial rates (although this is common in the region and reflects social equity concerns in Costa Rica), while OECD countries tend to have industrial rates which are lower than residential ones. Current tariff setting procedures also imply that fluctuations in costs may not be fully transferred to consumer prices.

Box 1.

Methodologies for Potential Output Estimates

In the production function approach, potential output is modeled as a Cobb-Douglas function of labor and capital inputs, and TFP:

Y_{t} = A_{t} K_{t}^{α} L_{t}^{1 - α}

where Y_t is output, K_t and L_t are capital and labor inputs, and A_t is the contribution of technology or TFP. Output elasticities sum up to one and α is set at 0.35. Labor force data up to 2010 comes from Penn World Table 7.1 (PWT) and is assumed to grow at the 2000-10 average annual rate thereafter. The capital stock series is constructed using a perpetual inventory method:

K_{t} = (1 - δ) K_{t - 1} + I_{t}

where the depreciation rate δ is set as 0.05, while the initial capital stock is computed as K₀ = I^*/(g + δ). I^* is the benchmark investment (average share of investment in GDP) and g is the average economic growth over 1991-2013. Finally, TFP is estimated as a residual, $A_{t} = Y_{t} / (K_{t}^{α} L_{t}^{1 - α})$ .

All univariate filters are based on separating a time series into trend and cyclical components. Standard parameters are used for most of the filters but the restriction parameter for the HP filter merits discussion. This parameter trades off goodness of fit with smoothness and it is set at 6.25 for annual data, which is equivalent to 1600 for quarterly data (the value proposed by the authors).

Two univariate Kalman filters are used with increasing flexibility across specifications. The first one envisages a deterministic drift:

\begin{matrix} y_{t} = y_{t}^{p} + \hat{y_{t}} \\ \begin{matrix} \begin{matrix} y_{t}^{p} = \bar{μ} + y_{t - 1}^{p} \end{matrix} \\ \begin{matrix} \begin{matrix} \hat{y_{t}} = ρ_{1} \hat{y_{t - 1}} + ρ_{2} \hat{y_{t - 2}} + \in_{t} \end{matrix} \\ \in_{t} \sim N (0, σ^{2}) \end{matrix} \end{matrix} \end{matrix}

where y_t is output, $y_{t}^{p}$ potential output, $\hat{y_{t}}$ the output gap, $\bar{μ}$ is the long-term steady state growth rate, and ε_t is a normally distributed error term. In this specification, potential output follows a random walk with deterministic drift (or trend) and the output gap is given by an AR(2).

The second specification allows for mean reversion in the drift with an adjustment coefficient β ∈ (0,1). Intuitively, β measures the persistence of shocks to the potential output growth rate. The second equation thus becomes $y_{t}^{p} = μ_{t} + y_{t - 1}^{p}$ , where $μ_{t} = (1 - β) \bar{μ} + β μ_{t - 1}$ .

Two multivariate filters based on inflation only and inflation together with financial and fiscal variables are also employed. The first specification broadly follows Laxton et al. (2010) and is given by the following decomposition for GDP:

\begin{matrix} \log ({G D P}_{t}) = \bar{Y_{t}} + Y_{t} \\ \begin{matrix} \begin{matrix} \bar{Y_{t}} = \bar{Y_{t - 1}} + G + \in_{t}^{\bar{Y}} \end{matrix} \\ \begin{matrix} \begin{matrix} G_{t} = θ G^{s s} + (1 - θ) G_{t - 1} + \in_{t}^{G} \end{matrix} \\ Y_{t} = φ Y_{t - 1} + \in_{t}^{Y} \end{matrix} \end{matrix} \end{matrix}

which is augmented by a Phillips curve:

π_{t} = {λ π}_{t + 1} + (1 - λ) π_{t - 1} + β y_{t} + \in_{t}^{π}

and inflation and growth expectations data, modeled as

\begin{array}{l} π_{t + j}^{c} = π_{t + j} + \in_{t + j}^{π^{c}}, & j = 0, 1 \\ G_{t + j}^{C} = G_{t + j} + \in_{t + j}^{G^{c}}, & j = 0, 1 \end{array}

The model is then estimated using Bayesian maximum likelihood with informative priors for some parameters. For shocks, priors which reflect more volatility in the cycle component are used – this is in line with what is observed for advanced economies.

The second specification is given by a measurement equation of GDP, y_t = c_t + τ_t + ε_t with a cycle (c_t) and trend components (τ_t), together with a state equation:

[\begin{matrix} \begin{matrix} c_{t} \\ τ_{t} \end{matrix} \\ τ_{t - 1} \end{matrix}] = [\begin{matrix} θ & 0 & 0 \\ 0 & 2 & - 1 \\ 0 & 1 & 0 \end{matrix}] (\begin{matrix} \begin{matrix} c_{t - 1} \\ τ_{t - 1} \end{matrix} \\ τ_{t - 2} \end{matrix}) + [\begin{matrix} γ_{1} & γ_{2} & γ_{3} & γ_{4} & γ_{5} & γ_{6} \\ 0 & 0 & 0 & 0 & 0 & 0 \\ 0 & 0 & 0 & 0 & 0 & 0 \end{matrix}] w_{t} + [\begin{matrix} \begin{matrix} ɛ_{t}^{c} \\ ɛ_{t}^{τ} \end{matrix} \\ 0 \end{matrix}]

where w_t = [r, rer, cred, fdef, stockpr, π_core]. This can be thought of as an augmented HP filter where observed GDP is decomposed between trend and cycle using the variables in w_t, namely: real interest rate and exchange rate, log of credit to consumption, housing, services, and retail, the annual change in the central government’s primary deficit, log of stock prices, and core inflation. All error terms are Gaussian and a ratio of variances between the error terms $ɛ_{t}^{c}, ɛ_{t}^{τ}$ of 6 is also imposed.

References

Aguirre, D., L. Hoteit, C. Rupp, and K. Sabbagh, 2012, “Empowering the Third Billion. Women and the World of Work in 2012,” Booz and Company.
- Search Google Scholar
- Export Citation
Hamilton, J., 1989, “A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle,” Econometrica, 57 (2), pp. 357–84.
- Search Google Scholar
- Export Citation
Hamilton, J., 1990, “Analysis of Time Series Subject to Changes in Regime,” Journal of Econometrics, Vol. 45, pp. 39–70.
- Search Google Scholar
- Export Citation
Hamilton, J., 1991, “A Quasi-Bayesian Approach to Estimating Parameters for Mixtures of Normal Distributions”, Journal of Business and Economic Statistics, Vol. 9, pp. 27–39.
- Search Google Scholar
- Export Citation
Hamilton, J., 1993, “Estimation, Inference, and Forecasting of Time Series Subject to Changes in Regime,” in Eds. G. D. Maddala, C. R. Rao, and H. D. Vinod, Handbook of Statistics, Vol. 11, pp. 231–59, New York.
- Search Google Scholar
- Export Citation
Hamilton, J., 1994, Time Series Analysis. Princeton University Press.
- Search Google Scholar
- Export Citation
Heintz, J., 2006, “Globalization, Economic Policy and Employment: Poverty and Gender Implications,” International Labour Organization, Geneva.
- Search Google Scholar
- Export Citation
Johnson, C., 2013, “Potential Output and Output Gap in Central America, Panama and Dominican Republic,” IMF Working Paper 13/145 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation
Kapsos, S., 2005, “The Employment Intensity of Growth: Trends and Macroeconomic Determinants”, International Labour Organization, Geneva.
- Search Google Scholar
- Export Citation
Laxton, D. and R. Tetlow, 1992, “A Simple Multivariate Filter for the Measurement of Potential Output,” Technical Report No. 59, Ottawa, Bank of Canada.
- Search Google Scholar
- Export Citation
Laxton, D., J. Benes, K. Clinton, R. Garcia-Saltos, M. Johnson, P. Manchev, and T. Matheson, 2010, “Estimating Potential Output with a Multivariate Filter,” IMF Working Paper 10/285 (Washington: International Monetary Fund).
- Search Google Scholar
- Export Citation

Prepared by Patrick Blagrave, Jorge Restrepo, Jose Pablo Valdes and Joyce Wong.

While both of these methods estimate a level of “sustainable” output, they differ somewhat in their definitions of sustainability. In the first approach using only inflation, the level of potential output is defined as the level which does not trigger inflation (Okun’s definition) and thus the most directly pertinent for the conduct of conventional inflation-targeting monetary policy. On the other hand, the second measure takes into account financial sector imbalances, making it more relevant for macro-prudential policy.

These were obtained from the CBRC’s website and from the National Statistics Institute’s website.

Mathematically, the input-output matrix represents a linear system of n unknowns and n equations, where n is the number of industries in the economy: : X_i = z_i1 + z_i2 + z_i3 + … + z_in + Y_i, where i = 1, …, n and X_i represents the total output of industry i, z_ij represents industry j’s intermediate consumption of industry i’s output, and Y_i represent the total final demand of industry i’s output. The Leontief demand model then consists of expressing z_ij in terms of X_i such that z_ij = a_ij X_j, ∀ i, j = 1, …, n where a_ij are the Leontief coefficients. Using matrix notation, the model can be written as X = (I – A)^–1Y where A is the n x n matrix with the Leontief coefficients and I is the identity matrix. The matrix (I – A)^-1 is the Leontief inverse.

Given the lack of data for hours worked, the comparison is done using total labor income and not hourly wages. Thus, some of this difference could be driven by the differences in hours worked which, if they are higher for those with higher education, could lead to an overestimation of the skill premium.

Maximum demand is around 1600 MW while installed capacity is 2731 MW, including 600 MW of thermal plants.

Most of ICE’s recent investments have been financed through debt, and according to Fitch, the leverage ratio of ICE is relatively high and could deteriorate if projects like Reventazón do not begin operations in the short term or if tariffs are not further adjusted upwards.

Other Publishers

Save
Cite
Email this content

Share Link

Copy this link, or click below to email it to a friend
Email this content
or copy the link directly:

https://www.elibrary.imf.org/view/journals/002/2015/030/article-A001-en.xml
The link was not copied. Your current browser may not support copying via this button.

Link copied successfully

Costa Rica: Selected Issues and Analytical Notes

Author: International Monetary Fund. Western Hemisphere Dept.

Volume/Issue:: Volume 2015: Issue 030

Publisher:: International Monetary Fund

ISBN:: 9781475527025

ISSN:: 1934-7685

Pages:: 100

DOI:: https://doi.org/10.5089/9781475527025.002

Keywords
I. Selected Real Sector Issues1
References

View raw image

Contributions to GDP Growth

(Average of time period)

View raw image

Labor Force Participation Rates

(Percent of 15+ population, 2012)

View raw image

Unemployment Rates

(Percent of labor force, 2012)

View raw image

Unemployment Distribution by Age

(Frequency in percent)

View raw image

Unemployment Rate Projections

(Percent)

View raw image

Labor Force with Secondary and Tertiary School

(Percent of labor force, 2012)

View raw image

Cumulative Percentage People in Each Cohort who Have at Least X Years of Schooling

View raw image

Cumulative percentage of children who have at least x years of schooling

(conditional on quintile of household income)

View raw image

Cumulative percentage of children who have at least x years of schooling

(conditional on household head’s years of schooling)

View raw image

Average Electricity Tariffs as 2011

(Residential vs. industrial, in USD cents/KwH)

View raw image

Technical Losses as 2011

(Percent of total output)

	*Potential GDP growth rate*		*Output Gap*
	1999-2008	2009-13	2014	2013	2014
Production Function Approach	4.8	3.8	4.1	0.1	−0.2
*Cycle Extraction Filters*	*Potential GDP growth rate*		*Output Gap*
*Cycle Extraction Filters*	1999-2008	2009-13	2014	2013	2014
Hodrick-Prescott	4.8	3.8	4.1	0.1	−0.2
Butterworth	4.8	3.7	4.1	0.2	−0.1
Christiano-Fitzgerald	4.9	3.3	4.5	1.2	0.5
Baxter-King	4.9	3.8	4.1	−0.1	−0.4
*Univariate Kalman Filters*	*Potential GDP growth rate*		*Output Gap*
*Univariate Kalman Filters*	1999-2008	2009-13	2014	2013	2014
Deterministic Drift	4.5	4.5	4.5	1.9	1.1
Mean Reversion	4.9	4.9	4.9	−4.3	−4.6
*Multivariate Filters*	*Potential GDP growth rate*		*Output Gap*
*Multivariate Filters*	2001-08	2009-13	2014	2013	2014
With inflation	4.5	4.2	4.2	−0.2	−1.0
With inflation and financial vars.	4.8	3.7	4.1^*	−2.0	—
*Average of All Models*	4.7	4.0	4.3	-0.3	-0.6

	*Potential GDP growth rate*		*Output Gap*
	1999-2008	2009-13	2014	2013	2014
Production Function Approach	4.8	3.8	4.1	0.1	−0.2
*Cycle Extraction Filters*	*Potential GDP growth rate*		*Output Gap*
*Cycle Extraction Filters*	1999-2008	2009-13	2014	2013	2014
Hodrick-Prescott	4.8	3.8	4.1	0.1	−0.2
Butterworth	4.8	3.7	4.1	0.2	−0.1
Christiano-Fitzgerald	4.9	3.3	4.5	1.2	0.5
Baxter-King	4.9	3.8	4.1	−0.1	−0.4
*Univariate Kalman Filters*	*Potential GDP growth rate*		*Output Gap*
*Univariate Kalman Filters*	1999-2008	2009-13	2014	2013	2014
Deterministic Drift	4.5	4.5	4.5	1.9	1.1
Mean Reversion	4.9	4.9	4.9	−4.3	−4.6
*Multivariate Filters*	*Potential GDP growth rate*		*Output Gap*
*Multivariate Filters*	2001-08	2009-13	2014	2013	2014
With inflation	4.5	4.2	4.2	−0.2	−1.0
With inflation and financial vars.	4.8	3.7	4.1^*	−2.0	—
*Average of All Models*	4.7	4.0	4.3	-0.3	-0.6

	*Potential GDP growth rate*		*Output Gap*
	1999-2008	2009-13	2014	2013	2014
Production Function Approach	4.8	3.8	4.1	0.1	−0.2
*Cycle Extraction Filters*	*Potential GDP growth rate*		*Output Gap*
*Cycle Extraction Filters*	1999-2008	2009-13	2014	2013	2014
Hodrick-Prescott	4.8	3.8	4.1	0.1	−0.2
Butterworth	4.8	3.7	4.1	0.2	−0.1
Christiano-Fitzgerald	4.9	3.3	4.5	1.2	0.5
Baxter-King	4.9	3.8	4.1	−0.1	−0.4
*Univariate Kalman Filters*	*Potential GDP growth rate*		*Output Gap*
*Univariate Kalman Filters*	1999-2008	2009-13	2014	2013	2014
Deterministic Drift	4.5	4.5	4.5	1.9	1.1
Mean Reversion	4.9	4.9	4.9	−4.3	−4.6
*Multivariate Filters*	*Potential GDP growth rate*		*Output Gap*
*Multivariate Filters*	2001-08	2009-13	2014	2013	2014
With inflation	4.5	4.2	4.2	−0.2	−1.0
With inflation and financial vars.	4.8	3.7	4.1^*	−2.0	—
*Average of All Models*	4.7	4.0	4.3	-0.3	-0.6

Topics

Countries

Series

About

Resources

Abstract

I. Selected Real Sector Issues1

A. Potential Output Estimates

B. Estimating the Impact of Intel Exit

C. Labor Markets and Inequality

D. Electricity sector

References

Same Series

Other IMF Content

Other Publishers

Asian Development Bank

Inter-American Development Bank

The World Bank

Share Link

Table of Contents

Headings

Figures

Metrics

Foresee

I. Selected Real Sector Issues¹