Abstract
Studies show that the relationship between openness and output volatility is theoretically ambiguous, but most of these studies provide an empirical estimation for this relationship. This paper investigates the impact of trade openness on output volatility, and how this impact may be affected by the country’s level of development. We use a panel dataset for 33 countries for the years of 1980 through 2009. A standard deviation of quarterly real GDP over a 5-year span is used as the dependent variable. Controlling for the country and period-specific effects, the main results are as follows: trade openness increases the output volatility. And, the output volatility of countries with a higher level of development is less affected by trade openness.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
Output volatility affects the aggregate economic growth, the stock market, and economic forecasting. This paper examines the impact of trade openness on output volatility. In addition, the paper investigates whether this impact is asymmetric or not. Recent empirical papers on the impact of trade openness on output volatility argue that this relationship is theoretically ambiguous, but empirically testable.
The literature that is relevant to discussing the relationship between openness and business cycle volatility is rapidly growing. However, many studies failed to use the right methodology, which can lead to an incorrect outcome. Karras (2006) finds that trade openness has a significant negative impact on output volatility. Karras calculates the average of trade openness and the standard deviation of output from 1951 to 1998, a time span that is too wide to yield accurate results. In contrast, this analysis uses a panel data for 33 countries. Time is divided into 5-year periods, as this was seen as a more appropriate measure for business cycle volatility.
Calderón and Schmidt-Hebbel (2008) employ an annual panel data for 82 countries from 1975 to 2005. In relying on annual data and by accounting for year- and country-fixed effects, the authors find that trade openness reduces the standard deviation of real GDP growth rate. However, unlike Calderón and Schmidt-Hebbel and other papers in the literature that use annual data on output, our research uses quarterly data on log of real GDP, as this allows for a more precise estimation for output variability.
Studies such as those conducted by Calderón and Schmidt-Hebbel (2008) and Bekaert et al. (2006) focus on the standard deviation of output or consumption over a 5-year period. Splitting the series into 5-year periods is better suited for this type of analysis for two reasons: First, it gives more distance for output variability. Second, it is more consistent with the NBER recessions.
Giovanni and Levchenko (2009) argue that sectors more open to international trade are more volatile. While Yanikkaya (2003) argues that trade barriers and regulations affect the openness–economic growth relationship. Examples of these barriers include: taxes, methods of payment, and tariffs.
The paper contributes to the existing literature the following: We use quarterly data on real GDP to account for the output volatility, in which more observations are included in each of the 5-year periods. In addition, this paper includes country- and time-fixed effect, and control variables such as size of the economy, financial openness, real exchange rate, inflation rate, and nominal short-term interest rate. Finally, and most importantly, this research shows how the country’s level of development determines the degree of the impact of trade openness on output volatility. The remainder of this paper is organized as follows: Sect. 2 describes the data. Section 3 provides empirical models; and Sect. 4 presents the results and the conclusion of our research.
2 Data
Our data are a panel for 33 countries from 1980 to 2009. Data on real GDP, real exchange rate, inflation rate, and nominal short-term interest rate is quarterly. While data on trade openness, financial openness, country size, and human development index are annual, the following variables are extracted from Smith and Galesi (2011): real GDP, real exchange rate, inflation rate, and nominal short-term interest rate. Trade openness and country size variables are taken from Penn World Table 7.1, while financial openness is taken from Chinn-Ito Financial Openness Index (2011) and the human development index is drawn from the United Nations Development Programme.Footnote 1 Table 1 in the “Appendix” provides the list of countries used in the sample. And Table 2 provides the data source.
3 Empirical model
The output volatility is represented by the standard deviation of log(RGDP),Footnote 2 and the 5-year standard deviation of output is defined as follows:
The variable \(\sigma _{ yit}\) is the standard deviation of the log of real GDP for country \(i\) in period \(t\), where \(i=\) 1,2,3,...33 and \(t=\) 1,2,...6. Because there are 30 years, time is divided into six periods, with each being 5 years in length. We divide the sum of squared deviations by \(\frac{1}{20}\) because we have 20 quarters in a period of 5 years. \(y_{ qit}\) is the output for quarter \(q\) in country \(i\) within the period \(t\). And \(\bar{y}_{it}\) is the average of 5 year output for that particular country \(i\).
The trade openness is the sum of exports, EX, and imports, IM, as a fraction of GDP:
The country size is defined as the aggregate Gross Domestic Product of a country \(i\) at time \(t\) relative to the USA.Footnote 3
Baseline regression
where \(\sigma _{ yit}\) is the standard deviation of the log of real GDP for country \(i\) in period \(t\), \(\mu _i\), and \(\tau _t\) are country- and time-fixed effects, respectively, and \(\hbox {OPEN}_{ it}\) is the log of openness as defined in (2).Footnote 4 \(\hbox {Z}_{it}\) is a vector of control variables: country size, financial openness, real exchange rate, inflation rate, and nominal short-term interest rate. \(\varepsilon _{ it}\) is an error term.
Equation (5) has the same variables as Eq. (4), except we add a development variable and an interaction term. The variable \(\hbox {HDI}_{it}\) refers to the human development index, and the interaction term \((\hbox {HDI}\times \hbox {OPEN})_{ it}\) counts for investigating if there is asymmetric impact of trade openness on output volatility.Footnote 5
4 Results and conclusion
Table 3 reports the results of the baseline regression. These results show a positive impact of log of trade openness on output volatility, with a coefficient of 0.029814 and significant at a 1 % level. These results differ from Calderón and Schmidt-Hebbel (2008) and Karras (2006), where both studies find a negative significant impact of trade openness on output variability.
Our results are robust because of the inclusion of controls and country characteristics. In Karras’s paper (2006), the results change dramatically when country size is added to the simple linear regression. This alters the sign of the openness coefficient from insignificant positive to significant negative.
Table 3 shows that the larger the size of the economy, the less the output volatility. The coefficient equals \(-\)0.03331 and significant at a 1 % level. None of the other control variables are significant. Additionally, the adjusted R-squared is sufficiently large: 0.708. Taking into consideration the country- and time-fixed effects, our main result as indicated in Table 3 shows that trade openness increases the output volatility.
Table 4 includes the human development index and a continuous interaction term between this variable and the log of trade openness. The coefficient of the log of trade openness is positive and significant. The developed countries experience higher jumps and drops in output than developing countries, but this coefficient is insignificant. However, the coefficients of the inflation rate and the nominal short-term interest rate are both significant.
The coefficient of the interaction term between human development and trade openness is significant negative: \(-\)0.084158. This indicates that output volatility in more developed countries is less affected by trade openness than countries that are less developed. Table 4 demonstrates that the output volatility coming from trade openness is weaker in developed countries than that in developing countries. These results are reasonable. Because developed countries have more complicated economic structures and different output volatility sources, while developing countries depend more on international trade.
Notes
See UNDP 2014 Human Development Report.
\(\hbox {Log}(\hbox {RGDP}_{ it})=\hbox {Log}(\frac{\hbox {GDP}_{ it}}{\hbox {CPI}_{ it}})\), where \(\hbox {CPI}_{it}\) is the Consumer Price Index for country \(i\) in time \(t\).
Total PPP Converted GDP, G-K method, at current prices (in millions $).
Since \(t\) represents a period of 5 years and the openness is given by annual data, first, we take the logs of openness, and then, we calculate the average of logs over 5 years, that is, the average of the logs, not the log of the averages.
Note that the interaction term here is a continuous variable.
References
Bekaert G, Harvey C, Lundblad C (2006) Growth volatility and financial liberalization. J Int Money Finance 25(3):370–403
Calderón C, Schmidt-Hebbel K (2008) Openness and growth volatility. Working Paper No. 483, Central Bank of Chile
Giovanni J, Levchenko A (2009) Trade openness and volatility. Rev Econ Stat 91(3):558–585
Karras G (2006) Trade openness, economic size, and macroeconomic volatility: theory and empirical evidence. J Econ Integr 21(2):254–272
Smith LV, Galesi A (2011) GVAR Toolbox 1.1. http://www.cfap.jbs.cam.ac.uk/research/gvartoolbox
Yanikkaya H (2003) Trade openness and economic growth: a cross-country empirical investigation. J Dev Econ 72(1):57–89
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Abubaker, R. The asymmetric impact of trade openness on output volatility. Empir Econ 49, 881–887 (2015). https://doi.org/10.1007/s00181-014-0899-2
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00181-014-0899-2