Saving–investment correlations and the mobility of capital in the OECD countries: New evidence from cointegration breakdown tests

ABSTRACT

This study examines the long-run relationship between domestic saving and investment and undertakes an in-depth account of short-period breaks in the cointegrating vector for 24 OECD countries. The analysis is carried out in a time-series setting to take a country-by-country account of the evidence. The end-of-sample cointegration breakdown tests are performed on both FMOLS and FIML estimates of the model. The cointegrating relationship between domestic saving and investment prevails and the implied intertemporal budget constraint holds for most countries. The cointegration breaks down for some countries during the sub-sample periods. The results are generally consistent across various cointegration breakdown tests.

KEYWORDS:

• Saving
• investment
• capital mobility
• structural breaks
• cointegration breakdowns

Acknowledgments

This article is based on the research project undertaken under the Griffith University Research Grant (GURG), Griffith Business School, Griffith University, Australia. I am grateful to the Griffith Business School for the research grant for the project. I thank two anonymous referees of the journal for very useful comments and suggestions. I am, however, solely responsible for any errors and omissions that may remain in the article.

Disclosure statement

No potential conflict of interest was reported by the author.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

¹ The financial sector is vulnerable to a number of risks, including credit (default) risk, liquidity risk, exchange rate risk, operational risk, and market risk.

² The current account comprises (1) exports net of imports of goods and services, (2) net factor income (such as interest and dividends) from the rest of the world, and (3) net current transfers (such as foreign aid) from the rest of the world. The exports net of imports of goods and services are, however, normally a dominant component of current account.

³ International capital flows have increased dramatically since the 1980s, with much of the increase being due to trade in equity and bond markets. During the 1990s, the gross capital flows between industrial countries rose by 300%, while the trade flows increased by 63% and the real GDP by a comparatively modest 26% (Evans and Hnatkovska 2014).

⁴ The DF-GLS and DF-GLSu tests are also performed on the model estimated with a constant and a trend. The results obtained from the model estimated with a constant and a trend are not reported to conserve space, but are available from the author upon request.

⁵ The q test of Stock and Watson (1988) tests the null hypothesis of k unit roots (zero cointegrating vector) against the alternative hypothesis of m unit roots; where m < k. Alternatively stated, the null hypothesis in q test is k common stochastic trends (no cointegration), and the alternative hypothesis is m common stochastic trends. The alternative hypothesis thus corresponds to r = (k – m) number of cointegrating vectors. The rejection of the null hypothesis, H₀: q(k, m), implies the presence of r = (k – m) number of cointegrating vectors.

⁶ The end-of-sample cointegration breakdown tests in both FMOLS and FIML estimates are also performed to test the null hypothesis of cointegration for the full-sample period (1970 to 2006) against the alternative hypotheses of cointegration breakdowns during the sub-sample periods of (1) 1999 to 2006, m = 8; (2) 2000 to 2006, m = 7; and (3) 2004 to 2006, m = 3; where m denotes the number of observations in the sample breakdown periods. The results generally remain consistent and suggest that the cointegration prevails for most countries across different sub-sample periods. The results for these sub-sample periods are not reported to conserve space, but are available from the author upon request.

⁷ The maximum-likelihood (ML) estimation becomes a limited information maximum-likelihood (LIML) estimation when applied to a single equation and a full information maximum-likelihood (FIML) estimation when used for the system as a whole. The distinction between endogenous and exogenous variables becomes difficult when the variables in the system are mutually interdependent. Sims (1980) suggests that all variables in the system should be treated as endogenous when there is simultaneity among such variables. The VAR model abandons the distinction between endogenous and exogenous variables and treats all variables in the system as endogenous.

⁸ The FIML uses more information than the ordinary least squares (OLS), indirect least squares (ILS), two stage least squares (2SLS), and limited information maximum-likelihood (LIML) methods. Apart from all of the variables included in the system, the FIML method uses information relating to the mathematical form of the equations (that is, it takes into account the structure of all the equations of the system), as well as to the contemporaneous dependence of the disturbance variables of various equations.

Additional information

Funding

This research was supported by the Griffith Business School, Griffith University, Australia [Griffith University Research Grant (GURG)].