Printer Friendly

Estimating sustainable output growth in emerging market economies.

INTRODUCTION

The concepts of potential growth and the output gap play a key role in the formulation and implementation of macroeconomic policies. Monetary, fiscal and macroprudential policies take into account these estimates in order to adapt the policy stance to reduce possible macroeconomic imbalances and dampen aggregate fluctuations. The relevance and usefulness of these concepts depend on how accurately the potential growth estimate reflects the sustainable path of economic development and the output gap and serves to summarize the imbalances of the economy.

In this regard, the estimation of potential output growth in emerging markets has recently been a challenging task. Estimates obtained by using conventional univariate statistical filters (eg the Hodrick-Prescott (HP) filter) generally failed to detect imbalances prior to the onset of the crisis in late 2008. Moreover, these filters were not always helpful in decomposing the post-crisis slowdown in output growth into its cyclical and trend components. For example, Borio et al. (2013, 2014) analyse the real-time performance of the HP filter and show that in the US, the UK and Spain, univariate filter estimates had large upward bias before the recent crisis, which was revealed only after the crisis. Traditional approaches also overestimated potential output growth in the euro area before the crisis (ECB, 2011; Marcellino and Musso, 2011). This effect is even more pronounced in Central and Eastern European countries (Bernhofer et al, 2014).

In these circumstances, it seems to be appropriate to rely on additional macroeconomic indicators to diagnose the state of the business cycle. It is generally accepted that inflationary pressure builds when output is above potential and subsides when output falls below potential. As such, inflation in particular is viewed as a key symptom of unsustainability. The same applies to another conventional theory that links fluctuations in unemployment with the output gap (Okun's Law).

As discussed in Bernhofer et al. (2014), this consensus in macroeconomics was severely challenged by the global financial crisis. It is becoming increasingly clear that certain cyclical activities are not captured by this approach, such as unsustainable developments in the financial sector. For example, asset price bubbles can generate huge business cycles without creating any inflation as reflected by the average household consumer basket, which is the common notion of inflation. The global financial crisis is a case in point. Hume and Sentance (2009) propose two explanations for the decoupling of asset and output inflation. First, the financial upturn of the 2000s had a relatively limited impact on effective demand. Second, in cases where the demand effect was larger, inflation pressure was dampened by a deterioration of external balances instead of reaching domestic capacity constraints. Borio et al. (2013) discuss four additional reasons why output inflation could remain low and stable against the backdrop of soaring asset price inflation, namely (i) financial booms that coincide with positive supply shocks, (ii) increases in potential output in prolonged economic upturns (as measured by conventional approaches), (iii) capital inflows leading to currency appreciation and (iv) the existence of sectoral misallocation rather than 'aggregate' capacity constraints. Financial indicators are therefore essential for the balanced assessment of output growth, and the aim of this paper is to incorporate the information contained in them into the estimation of sustainable output growth in emerging market economies.

Our work is related to the recent literature on the link between business cycles and financial cycles (Alessi and Detken, 2011; Claessens et al, 2012; Schularick and Taylor, 2012). We concentrate here on developments in emerging markets, which means that data limitations will effectively restrict our analysis to the latest boom/bust episode. This closely links our work with the literature on the main factors explaining output fluctuations during the crisis of 2008 (Frankel and Saravelos, 2010; Lane and Milesi-Ferretti, 2011; Cecchetti et al, 2011; Feldkircher, 2014). Our main contribution to these strands of research is that we follow Alberola et al (2013), Borio et al (2013,2014) and Bernhofer et al (2014) in employing an empirical model that enables us to decompose output fluctuations into cycle and trend components based on the empirical relationships with various measures of imbalances. The resulting indicators may be interpreted, in an economic sense, as metrics of sustainable (ie not associated with the build-up of imbalances) output and the output gap.

The remainder of this paper is structured as follows. The next section discusses the set-up of the model. The subsequent section presents the data set. The section following it reports the empirical results. The penultimate section discusses the output gap estimates for the cross-section of emerging markets in general and provides more detailed results for Russia. The final section concludes.

MODEL SET-UP

We follow Borio et al. (2013, 2014) and employ a multivariate HP (MVHP) filter in a state-space form: (1)

[DELTA][y*.sub.it] = [DELTA][y*.sub.it-1] + [[epsilon].sub.it] (1)

[y.sub.it] - [y*.sub.it] = [gamma]' [X.sub.it-s] + [[zeta].sub.it] (2)

[epsilon].sub.it] ~ N(0,[[sigma].sup.2.sub.1]) (3)

[[zeta].sub.it] ~ N(0, [[sigma].sup.2.sub.2]) (4)

[[sigma].sup.2.sub.2]/[[sigma].sup.2.sub.1] = 1,600 (5)

where [y.sub.it] is the log of real GDP and [y*.sub.it] is its unobserved trend component. The residuals of state equation 1 and signal equation 2 are assumed to be a normally and independently distributed error with mean zero and variance [[sigma].sup.2.sub.1] and [[sigma].sup.2.sub.2]. The so-called signal-to-noise ratio ([[sigma].sup.2.sub.2]/ [[sigma].sup.2.sub.1]) determines the relative variability of the estimated potential output series. We set this ratio at 1,600, which corresponds to the smoothing parameter [lambda] = 1,600 in a conventional univariate HP filter. [x.sub.it] represents the imbalances indicators with lag order s. (2)

Our analysis lies at the nexus of traditional potential output modelling, which is usually (3) country-specific, and research on financial imbalances, which is conducted almost exclusively based on panel data. We choose the latter approach. We believe that this may be appropriate given that owing to data limitations our time series in most cases start in the 2000s. This means that in each individual case we are effectively limited to the analysis of just one episode of large output fluctuations (ie. before and after 2008), which is obviously not enough for the empirical validation of the model. (4) It may therefore be appropriate to pool the information about the boom/bust episodes in other countries, albeit observed in one wave.

Instead of relying on country-specific analysis, we conduct a pooled estimation for the cross-section of emerging market economies. Technically, this means that we specify a state-space model consisting of blocks comprising equations 1 and 2 attributed to individual countries in the cross-section. We thus allow for country-specific trend GDP ([y*.sub.it]) but assume common coefficients that link its developments with the imbalances indicators ([gamma]). We use a Kalman filter to obtain the maximum likelihood estimates of these parameters and the unobserved trend GDP.

DATA

The aim of our methodology is to obtain estimates of sustainable growth rates. The sustainable growth rate is defined as the output growth that does not generate or widen macroeconomic imbalances, which are identified through a wide set of indicators. Conveniently, in recent years there has been a significant number of contributions to the literature on such imbalances indicators. In fact, several international organizations have developed various frameworks for the evaluation and early detection of macroeconomic imbalances (see Alberola et al. (2013) for a review). We follow these studies (most notably, Alessi and Detken (2011) and Frankel and Saravelos (2010)) in our choice of imbalances indicators, using those that have produced robust results under a variety of specifications of the model. (5) We use the credit/GDP ([C.sub.t]) and broad money/GDP ([M.sup.t]) ratios, as well as stock market capitalization ([S.sub.t]) (all in logs), as proxies for financial imbalances. We also use the share of gross fixed capital formation in GDP ([INV.sub.t]). These are combined with traditional imbalances indicators: annual CPI growth ([[pi].sub.t]) and the unemployment rate ([U.sub.t]). All the data are standardized and seasonally adjusted and de-trended by means of the HP filter ([lambda] = 100,000). (6)

Our main data source is the IMF 1FS database, except for gross capital formation shares and stock market capitalization data, which are from the World Bank WDI database (see Appendix). We use quarterly data, and where only annual data are available, we interpolate by using cubic splines.

Based on these data sources, we were able to compile the cross-section of 28 emerging market economies (Table 1). The model was estimated over an (unbalanced (7)) time sample from 2000Q1 to 2012Q4. All available data were used for the preliminary de-trending of the imbalances indicators.

EMPIRICAL RESULTS

We begin by estimating the bivariate versions of the model, which include the imbalances indicators individually, and then proceed by including all the indicators jointly (Table 2). With the exception (8) of the broad money/GDP variable, all variables have the expected signs and high statistical significance when included in the model together with inflation and unemployment. These results generally confirm the idea that developments in financial variables are associated with cyclical fluctuations in output and importantly can provide information about the state of the business cycle beyond that contained in conventional indicators (ie inflation and unemployment).

We consider the resulting parameterization to provide a benchmark model, even though arguably one may reasonably use a homogeneous cross-section that includes only relevantly similar economies (eg from one region). The caveat here is that it is also desirable to have a data set that is balanced as regards the presence of boom/bust occurrences. For example, if our data set only included European countries, most of which experienced dramatic output fluctuations, we would be unable to test the performance of the model in a more tranquil environment. Nevertheless, in order to check the robustness of the results, we also report the estimates obtained for the subsamples. First, we split our cross-section into regional groups: Asia, Central and Eastern Europe (CEE), the former Soviet Union (FSU) and Latin America. As might be expected, the estimates (Table 3) were most ambiguous (only the inflation and stock prices variables had significant coefficients with the correct signs) for the Asian subgroup, where the boom/bust episode was less pronounced than in the other regions. On the contrary, all parameters were significant for European countries and, most notably, the credit variable's coefficient was much larger than that in the benchmark model. The results obtained for the Latin American region were similar to those for the benchmark model, albeit the credit and inflation variables had low statistical significance.

As pointed out by Bemhofer et al. (2014), the relationship between financial imbalances and output growth might be quite heterogeneous among emerging markets, as these economies have been on a convergence path during the past decade and are at highly different stages of economic development. In other words, the relevance of the financial variables may vary depending on the size of the financial sector. We proxy financial depth by using the credit/GDP ratio (averaged over 2000Q1-2012Q4) and divide our cross-section into quartiles: the first group containing the countries with the lowest credit/GDP ratios and the last group those with the highest. We find no distinct pattern in the results (Table 4).

For example, the credit variable performs best in the first and last quartiles, while the stock prices variable is highly significant for all subsamples. We conclude that the relevance of the financial indicators for the cyclical output fluctuations variables is not conditioned by the size of the financial sector, at least as measured by using the credit/GDP ratio.

OUTPUT GAP ESTIMATES

General results

By using the benchmark parameterization reported in Table 2, we compute the trend and cycle components of GDP for all the countries in our cross-section. We can then compare the ranges of the output gap estimates obtained with the univariate and multivariate versions of the HP filter (Figure 1). Several distinct differences can be identified between the two ranges. Prior to 2006, the standard versions of the output gap were fluctuating close to zero, while the MVHP versions were mostly negative. At their peak in late 2007, the MVHP versions of output gaps were higher and after the crisis lower than those of the standard HP versions. The variability and magnitude of fluctuations in the MVHP versions were also generally larger.

The underlying reason for the difference between these two sets of estimates can be illustrated by plotting the growth rates of trend GDP (Figure 2). The growth rate estimates based on the univariate HP filter display notable variability, decreasing by about 4 p.p. in the second part of the time sample compared with the pre-crisis level. This may, of course, be a true reflection of the severe damage caused by the crisis to potential economic growth. However, more likely, given the relatively short time sample and magnitude of output fluctuation during the crisis, the univariate filter 'overfits' the data by introducing excessive variability into the trend and weakens its interpretation as the sustainable level of output. As regards the MVHP version, some of the actual GDP fluctuations are explained by the imbalances indicators, ensuring more stable trend GDP growth.

We also apply the concept of cumulated potential GDP losses to illustrate the difference between the two versions of the filter. To this end, for the period starting from 2008, we estimate the difference between actual trend GDP and the extrapolated (9) trend values. Interestingly, we can compare our results with the estimates of typical output losses over previous recessions reported by existing studies (10) such as Abiad et al. (2009), Furceri and Mourougane (2012),

Haltmaier (2012) and Howard et al. (2011). The results obtained with the MVHP filters are generally in line with these estimates, while those derived from the standard HP filters indicate that the output losses after the recent crisis were notably larger than on average in the historical cases (Figure 3).

Country-specific results: the case of Russia

We expand on our findings by providing more detailed results from our model's application to GDP fluctuations in Russia. Similar to the general results for the whole cross-section, the MVHP output gap estimate is wider at its peak and narrower after the crisis compared with the HP version. This finding implies that the annual growth trend decreased from about 5% to 3% after the crisis (from 6% to 2% in the case of univariate filtering) (Figure 4).

The recursive estimates (Figure 5) show that the application of the univariate HP filter to Russian GDP was not sufficiently informative. Multivariate filters may improve the real-time performance of the analysis. We conduct quasi (11) recursive estimates by using the MVHP version of the filter (Figure 6). These estimates are much more stable over time. The caveat is that this is no longer the case when the imbalances indicators are also de-trended recursively. Apparently, the model is quite sensitive to the end-point problem that arises in the preliminary step.

Finally, we examine the contributions of different indicators to the explained part of the output gap (ie [gamma]'[x.sub.it-s]) (Figure 7). The results show that, prior to the crisis, all the imbalance variables unanimously indicated that GDP was above the sustainable level. For the post-crisis period, the results are ambiguous. Stock price developments are the most important indicators for explaining output gap fluctuations, followed by gross capital formation and unemployment. Credit developments and CPI inflation do not play an important role, at least under this parameterization. Admittedly, our model

is purely empirical and does not provide a structural interpretation. It is therefore impossible, based on our results, to say that stock market developments as such were the underlying factor behind the output gap formation in Russia. Instead, we may argue that stock prices could be regarded as a good summary indicator for financial conditions in the economy (perhaps serving as a proxy indicator for eg asset price developments, capital inflows, risk perception) and that, based on the observed asset price boom on the Russian stock market, one can analyse the deviation in output growth from the sustainable path.

CONCLUSIONS

During the recent financial crisis, doubts were expressed as to the relevance and usefulness of conventional approaches to estimating the output gap. Thinking of potential output only as non-inflationary output or output not associated with a reduced unemployment rate had proven to be too restrictive. Recent history has demonstrated that other imbalances, notably in the financial sector and in asset markets, can emerge as inflation and unemployment remain stable. In our paper, we present a model that helps incorporate the information contained in financial indicators into the estimation of sustainable output growth in emerging market economies.

We specify a state-space model representing a multivariate HP filter that links cyclical fluctuation in GDP with several indicators of macroeconomic imbalances. The latter include financial variables as well as conventional CPI inflation and the unemployment rate. We obtain the parameterization of the model by estimating it jointly for a cross-section of emerging market economies. The results indicate that the imbalances indicators are statistically significant in explaining output gap fluctuations, particularly in the case of European countries, meaning that they contain information beyond that contained in the inflation and unemployment variables. A stock price indicator seems to perform especially well. As the model has no structural interpretation, this does not mean that stock market developments as such were the main factor behind these output gap fluctuations. Nevertheless, one might well claim that a rise in stock prices is an important symptom that could help distinguish between trend and cyclical output growth acceleration.

The output gaps obtained based on the estimated model differ substantially from those calculated with the univariate version of the HP filter. Most notably, trend output growth rates are more stable and therefore more consistent with the notion of sustainable output. Cumulative output losses after the recession in 2008, estimated on the basis of a multivariate filter, are, unlike those estimated by using the univariate version comparable with typical episodes reported in the literature. Employing the multivariate filter may thus help improve the real-time robustness of the model, although our approach is still quite sensitive to the end-point problem associated with the transformation of the imbalance variables.

APPENDIX

Table A1: Data sources

Variables   Description                                  Data source

y           Real GDP Index (2005 = 100)                  IMF elibrary
[pi]        Quarterly CPI growth rate                    IMF eLibrary
u           Quarterly unemployment rate                  World Bank WDI
INV         Gross fixed capital formation-to-GDP ratio   World Bank WDI
S           Stock market capitalization-to-GDP ratio     World Bank WDI
C           Claims on private sector-to-GDP ratio        IMF eLibrary
M           Broad money-to-GDP ratio                     IMF eLibrary


REFERENCES

Abiad, A, Balakrishnan, R, Brooks, PK, Leigh, D and Tytell, 1.2009: What's the damage? Medium-term output dynamics after banking crises. IMF Working Paper 09/245.

Alberola, E, Estrada, A and Santabarbara, D. 2013: Growth beyond imbalances: Sustainable growth rates and output gap reassessment. Banco de Espana Documentos de Trabajo N1313.

Alessi, L and Detken, C. 2011: Quasi real time early warning indicators for costly asset price boom/ bust cycles: A role for global liquidity. European Journal of Political Economy 27(3): 520-533.

Bernhofer, D, Fernandez-Amador, O, Gachter, M and Sindermann, F. 2014: Finance, potential output and the business cycle: Empirical evidence from selected advanced and CESEE economies. Focus on European Economic Integration Q2/14: 52-75.

Borio, C, Disyatat, P and Juselius, M. 2013: Rethinking potential output: Embedding information about the financial cycle. BIS Working Papers 404.

Borio, C, Disyatat, P and Juselius, M. 2014: A parsimonious approach to incorporating economic information in measures of potential output. BIS Working Papers 442.

Cecchetti, SG, King, MR and Yetman, J. 2011: Weathering the financial crisis: Good policy or good luck? BIS Working Papers 351.

Claessens, S, Rose, MA and Terrones, ME. 2012: How do business and financial cycles interact? Journal of International Economics 87(1): 178-190.

Duffy, J and Papageorgiou, C. 2000: A cross-country empirical investigation of the aggregate production function specification. Journal of Economic Growth 5(1): 87-120.

ECB. 2011: Recent evidence on the uncertainty surrounding real-time estimates of the Euro area output gap. Monthly Bulletin November. Box 5.

Feldkircher, M. 2014: The determinants of vulnerability to the global financial crisis 2008 to 2009: Credit growth and other sources of risk. Journal of International Money and Finance 43: 19-49.

Frankel, JA and Saravelos, G. 2010: Are leading indicators of financial crises useful for assessing country vulnerability? Evidence from the 2008-09 global crisis. NBER Working Papers 16047.

Furceri, D and Mourougane, A. 2012: The effect of financial crises on potential output: New empirical evidence from OECD countries. Journal of Macroeconomics 34(3): 822-832.

Haltmaier, J. 2012: Do recessions affect potential output? IFDP 1066.

Howard, G, Martin, R and Wilson, BA. 2011: Are recoveries from banking and financial crises really so different? IFDP 1037.

Hume, M and Sentance, A. 2009: The global credit boom: Challenges for macroeconomics and policy. Journal of International Money and Finance 28(98): 1426-1461.

Lane, PR and Milesi-Ferretti, GM. 2011: The cross-country incidence of the global crisis. IMF Economic Review 59(1): 77-110.

Marcellino, M and Musso, A. 2011: The reliability of real-time estimates of the Euro area output gap. Economic Modelling 28(4): 1842-1856.

Schularick, M and Taylor, A. 2012: Credit booms gone bust: Monetary policy, leverage cycles, and financial crises, 1870-2008. American Economic Review 102(2): 1029-1061.

Senhadji, A. 1999: Sources of economic growth: An extensive accounting exercise. IMF Working Paper WP/99/77.

(1) Unlike Borio et al. (2013, 2014}, we do not use a dynamic version of the HP filter, which involves the addition of a lagged output gap term on the right-hand side of (2). The unrestricted estimation of this term's coefficient yields a value close to unity, which is economically implausible. Arguably, this may be due to insufficient variability in the output gap for the relatively short time sample in emerging markets. In addition, as shown in Borio et al. (2014), when using the dynamic HP filter, the smoothing parameter X should be recalibrated for each specific case in order to make the results comparable with the static version. That would seriously complicate our analysis, which is based on pooled estimation.

(2) We tested s from 0 to 4 and found that in most cases s=l yields the best results.

(3) Although not necessarily. See for example Senhadji (1999) and Duffy and Papageorgiou (2000) for panel estimates of the production function.

(4) In particular, we found it extremely challenging to obtain satisfactory results with country-specific models that included more than one or two explanatory variables, as the filter tended to favour only few indicators with the highest correlation with output growth. In the pooled estimates, we obtained more balanced results, as the performance of the explanatory variables was averaged across countries. Admittedly, this approach may be misleading if one expects to find substantial systematic differences in the relationship between the imbalances indicators and output in different countries. The reported results should therefore be regarded as evidence of the general relevance of the imbalances indicators for output gap diagnostics rather than the optimal parameterization of the model.

(5) We tested a broad range of indicators before making this selection. Most notably, indicators of external imbalances (trade balance, external debt, real effective exchange rate), although not included in the final model, worked well in other specifications. In addition, admittedly, the availability of financial indicators for emerging markets is severely limited, making their compilation for the whole cross-section quite difficult. We therefore were unable to test some indicators that could be useful (eg housing prices).

(6) This transformation is different from that of Borio et al. (2014), who use de-meaned growth rates. Such a transformation seems less applicable to emerging markets for which sample means are rarely associated with equilibrium values (eg CPI mean growth in the case of gradual disinflation). Admittedly, de-trending the data exacerbates the end-point problem and thus worsens the real-time performance of the model. We experimented with de-trending the imbalances variables jointly with GDP; however, while the model became computationally heavier, the results were not significantly different.

(7) Conducting the estimates on the shorter balanced time sample did not change the results dramatically.

(8) For the models presented in Tables 2-4, we removed those variables with the 'wrong' signs.

(9) For extrapolation, we used the average growth rate of trend GDP in 2005-2007.

(10) We report the resultant output evolution values after the banking crises estimated in Abiad et al. (2009) and output as a percentage of the pre-crisis trend after deep and long recessions in emerging economies reported in Howard et al. (2011), the estimated impact of severe financial crisis reported in Furceri and Mourougane (2012) and the average cumulative level change in potential output in emerging markets after stand-alone recessions reported in Haltmaier (2012).

(11) We assume that parameterization [gamma] is known and that it does not conduct recursive estimates for the pooled data set. We are not able to fully replicate the real-time analysis because in our sample most of the information on boom/bust occurrence comes in one batch.

ANNA KRUPKINA [1,2], ELENA DERYUGINA [1] & ALEXEY PONOMARENKO [1]

[1] Bank of Russia, 107016, Neglinnaya 12, Moscow, Russia. E-mails: PonomarenkoAA@cbr.ru; DeryuginaEB@cbr.ru

[2] NRU-HSE, Myasnitskaya 20, 101000, Moscow, Russia. E-mail: KrupkinaAS@cbr.ru

Table 1: Countries in the cross-section

Argentina
Armenia
Brazil
Bulgaria
Chile
China
Croatia
Czech Republic
Ecuador
Estonia
Georgia
Hungary
Indonesia
Kazakhstan
Korea
Latvia
Lithuania
Macedonia
Malaysia
Mexico
Peru
Poland
Romania
Russia
Slovakia
Slovenia
Thailand
Ukraine

Table 2: Estimates of parameter [gamma]
(z-statistics in parentheses)

[[pi].sub.t]     [U.sub.t-1]    [INV.sub.t-1]    [S.sub.t-1]

Parameter Estimates for Bivariate Regressions
0.08 (20.5)          --              --              --
--              -0.19 (-41.6)        --              --
--                   --          0.14 (41.4)         --
--                   --              --          0.16 (51.3)
--                   --              --              --
--                   --              --              --
Parameter Estimate for Regression Using All Variables
0.03 (5.1)      -0.11 (-11.2)    0.08 (7.7)      0.16 (26.4)

[[pi].sub.t]     [C.sub.t-1]     [M.sub.t-1]

Parameter Estimates for Bivariate Regressions
0.08 (20.5)          --              --
--                   --              --
--                   --              --
--                   --              --
--               0.08 (4.9)          --
--                   --          0.01 (2.5)
Parameter Estimate for Regression Using All Variables
0.03 (5.1)       0.05 (3.2)          --

Table 3: Estimates of parameter [gamma] for the
regional subsamples (z-statistics in parentheses)

Subsample       [[pi].sub.t]     [U.sub.t-1]    [INV.sub.t-1]

Asia             0.02 (2.6)     -0.01 (-1.0)         --
CEE              0.04 (2.9)     -0.06 (-1.9)     0.08 (3.4)
FSU              0.04 (3.1)     -0.26 (-10.2)    0.07 (2.9)
Latin America    0.01 (0.3)     -0.09 (-3.9)     0.06 (2.8)

Subsample        [S.sub.t-1]     [C.sub.t-1]

Asia             0.07 (10.8)         --
CEE              0.2 (13.3)      0.19 (4.2)
FSU              0.17 (8.0)      0.13 (2.6)
Latin America    0.16 (11.3)     0.03 (1.4)

Asia: China, Indonesia, Korea, Malaysia, Thailand. CEE: Bulgaria,
Croatia, Czech Republic, Hungary, Macedonia, Slovakia, Slovenia,
Poland, Romania. FSU: Armenia, Estonia, Kazakhstan, Georgia,
Latvia, Lithuania, Russia, Ukraine. Latin America: Argentina,
Brazil, Chile, Ecuador, Mexico, Peru.

Table 4: Estimates of parameter y for the subsamples
by financial sector size (z-statistics in parentheses)

Subsample       [[pi].sub.t]     [U.sub.t-1]    [INV.sub.t-1]

1st quartile         --         -0.14 (-6.7)     0.09 (2.9)
2nd quartile     0.05 (3.2)     -0.12 (-3.5)     0.06 (2.4)
3rd quartile     0.04 (2.9)      -0.1 (-3.7)     0.11 (4.3)
4th quartile     0.04 (3.3)     -0.09 (-3.7)     0.07 (3.0)

Subsample        [S.sub.t-1]     [C.sub.t-1]

1st quartile     0.14 (3.5)      0.06 (2.1)
2nd quartile     0.18 (9.2)          --
3rd quartile     0.22 (11.4)     0.04 (1.2)
4th quartile     0.12 (11.1)     0.04 (1.6)

1st quartile: Argentina, Armenia, Ecuador, Georgia, Mexico, Peru,
Romania. 2nd quartile: Indonesia, Kazakhstan, Macedonia, Poland,
Russia, Slovenia, Ukraine. 3rd quartile: Brazil, Bulgaria, Croatia,
Czech Republic, Hungary, Lithuania. 4th quartile: China, Estonia,
Korea, Latvia, Malaysia, Slovakia, Thailand.
COPYRIGHT 2015 Association for Comparative Economic Studies
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2015 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Krupkina, Anna; Deryugina, Elena; Ponomarenko, Alexey
Publication:Comparative Economic Studies
Article Type:Report
Geographic Code:4EUSP
Date:Mar 1, 2015
Words:4690
Previous Article:Technology adoption and growth in sub-Saharan African countries.
Next Article:Strength in diversity: a spatial dynamic panel analysis of Mexican regional industrial convergence, 1960-2003.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |