Revisiting Feldstein-horioka Puzzle - Econometric Evidences From Common Coefficient Mean Group Model


Abstract. The Feldstein-Horioka (1980) puzzle (FHP) is revisited by using Common Correlated Effect Mean Group (CCEMG) for a large group of countries over the period of 1980 to 2015. CCEMG methodology incorporates the issues of structural breaks and cross sectional dependence. Furthermore, we also investigate the role of several other macroeconomic factors, like judicial environment, governance and business environment, to improve the international capital mobility. There are two main findings of the paper. First, we confirm FHP. More exactly, there is a lack of international capital mobility among a larger group of countries. Second, this capital immobility can be declined through the improvement of globalization, judicial environment, governance and financial sector development.

Keywords: Feldstein-Horioka puzzle, Judicial environment, Governance, Business environment


    There is near to consensus that financial markets of the world are highly integrated. This view gets more strength in the presence of easily accessible information through the development of new communication technologies. Therefore, the financial economists have a common view that the capital mobility should be very high in the presence of integrated financial markets. Therefore, they test the degree of capital mobility through various indicators, methodologies and samples. However, the present study focuses on the measures and methodology of Feldstein and Horioka (1980) that is labeled as the mother of all puzzles by Obstfeld and Rogoff (2000).

    Feldstein and Horioka (1980) consider a relationship between domestic savings and domestic investment in open economy framework and document that if there is a high mobility of savings across the nations then the correlation between these two variables should be zero. This implies that domestic investment will be financed by foreign savings. On the other hand, the empirical analysis by Feldstein and Horioka (1980) shows a high correlation between domestic savings and domestic investment of OECD countries. This implies that most of the domestic investment is financed by the domestic saving. The high correlations between domestic savings and domestic investment can be termed as home bias instead of mobility.

    The seminal study of Feldstein and Horioka (1980) was revisited by Feldstein (1983) by extending the data for the OECD countries and confirmed the earlier findings. The FHP gets more strength. This strength of puzzle motivates a number of researchers to further test the correlations of savings and investments. Many of these researchers provide the support for FHP. However, some of the studies have a stance that the studies on the saving-investment relationship may not be informative in the context of mobility and financial market integrations. For example, the definition of capital mobility of Feldstein and Horioka (1980) is challenged by Sachs et al. (1981) and Ghosh (1995). Sachs et al. (1981) and Ghosh (1995) note that that current account volatility should be used as a proxy for capital mobility instead of savings and investment.

    Therefore, many of the empirical studies show that even if capital are mobile, saving and investment may be correlated because of the presence other macroeconomic factors. For example, the big economy effect, exchange rate regime, cost of investment, common causes and endogenous shocks (see, Murphy, 1984; Bayoumi, 1990; Coakley et al., 2004; Obstfeld and Rogoff, 2000; De Vita and Abbott, 2002; Corbin, 2004; Georgopoulos and Hejazi, 2009; Herwartz and Xu, 2010).

    Indeed, keeping this backdrop in view, the revisiting of FHP is not a unique idea. A plethora of research is available on the topic with different measures, samples, estimation methodologies and time span. Importantly, the more recent studies are augmenting FHP with various macroeconomic variables like size of economy, exchange rate regime, globalization, price of investment among many others. However, none of the studies incorporate the several issues of recent times. For example, the quality of institutions, the governance, the business environment and the impact of terrorism is almost completely ignored by almost all of the studies. The present study attempts to fill this gap.

    We accept the claim of Feldstein and Horioka (1980) and their followers that there exist home bias in the allocations of domestic saving due to various reasons that are mentioned in the literature. However, there arises an important and interesting question that whether the saving retention coefficient can be declined by including any variable in the Feldstein and Horioka (1980) regression. This article is an attempt to answer this question. Specifically, we investigate the question whether the saving retention coefficient can be declined over time by improving the situation of governance, doing business, quality of institutions and terrorism or not. If the saving retention coefficient declines by incorporating the mentioned factor then this will imply that these factors will play their role in the increasing of capital mobility across the world.

    Furthermore, most of the studies test FHP by using panel methods for the different sample from world. However, the researchers have not yet seriously explored the panel studies relating to FHP. Almost all of the studies are based on the single homogenous slope assumption. This implies that there is almost every country in the sample has the same saving retention coefficient. Obviously, this assumption is not very attractive. Furthermore, these studies do not take the structural break into account even taking a larger time span of data. Indeed, the empirical researchers should expect a number of structural breaks in the data of domestic saving and domestic investment due to various reforms in the financial sectors over the last two decades. Furthermore, the traditional panel data methods like fixed effects models and Generalized Method of Moments (GMM) models are based on the assumptions of cross sectional dependence.

    Importantly, the possibility of cross sectional independence cannot be denied in the presence of financial market integration. Keeping all these argument in view, FHP should be re-investigated by incorporating the slope heterogeneity and structural breaks assumptions robust to cross sectional dependence. This article is an attempt in this way.

    Therefore, the present study has the several contributions in the literature. First, this study reinvestigates FHP for a larger group of countries by using a long time series data. Second, we shall test the long run relationship between saving-investment by incorporating the structure breaks in a longer time series data. Finally, the article uses the second generation of the econometrics methodologies by incorporating the assumption of cross sectional dependence.

    The rest of the article is organized as follows. Section II will review the existing literature. Section III will explain the econometric specification and estimation Algorithms. The construction of the variables and data sources will be explained in Section IV. The details of empirical findings will be presented in Section V. Section VI will conclude the article.


    There is a plethora of research on FHP. However, even after discussion of last three decades, there is no consensus has been built. More specifically, the story is initiated by the seminal of paper of Feldstein and Horioka (1980). The paper finds that saving retention coefficient is andgt;zero that is interpreted as lack of capital mobility. Whereas this coefficient should be zero in the presence of financial market integrations and capital mobility. The difference between theory and empirical findings started a huge discussion in the financial literature. The researchers attempt to find the saving retention coefficient by using different measures, methodology and samples of the countries but find inconclusive results in this regard.

    Generally, the literature on the FHP can be divided into three strands of opinion. First, the saving retention coefficient is close to zero that implies that there exists perfect mobility of capital. Second, the saving retention coefficient is andgt;zero that implies the lack of capital mobility. Third, the saving retention coefficient can be declined to zero through some policy intervention. Only the third strand is an infant in the literature while the first two strands have good standing in the literature of finance.

    For example, Ketenci (2012) generally believe that there is no high correlation exist between the variables of domestic saving and domestic investment. This implies that the capital is highly mobile across nations. Similarly, Chu (2012) also shows through the experiment of Monte Carlo simulation that the FHP is upward biased and showing a spurious correlation between saving and investment. Ozdemir and Olgun (2009) also documents that FHP has very limited validity in the case of panel of country. Furthermore, the more recent studies like Singh (2013), Holmes and Otero (2014), Johnson and Lamdin (2014) and Kumar et al. (2014) note that the capital mobility is increasing in the recent times. On the other hand, the equally important studies like Penati and Dooley (1984) and Coakley et al. (2004) still confirm the findings of Feldstein and Horioka (1980) that the saving retention coefficient is much higher than zero.

    Similarly, Kumar et al. (2014), Chang and Smith (2014), Barros and Gil-Alana (2015), Chen and Shen (2015), and Konya (2015) also believe that the puzzle holds despite the many methodological and specification issues.

    The third strand postulates that there exist a high correlation between domestic saving and domestic investment. However, this is not because of the lack of capital mobility but it is attributes to some other macro-economic factors like the size of economy, exchange rate regime, governments spending, quality of institutions and globalizations...

To continue reading