Cross-country variation in patience, persistent current account imbalances and the external wealth of nations

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Cross-country variation in patience, persistent current account imbalances and the external wealth of nations © 2021 The Author(s). Published by Elsevier Ltd. Published version Nieminen, Mika Nieminen, M. (2022). Cross-country variation in patience, persistent current account imbalances and the external wealth of nations. Journal of International Money and Finance, 121, Article 102517. https://doi.org/10.1016/j.jimonfin.2021.102517 2022 Cross-country variation in patience, persistent current account imbalances and the external wealth of nations q Mika Nieminen ⇑ University of Jyväskylä, Finland article info Article history: Available online xxxx JEL classification: F21 F32 Keywords: External wealth Current account External imbalances Patience Economic preferences abstract This paper is the first to utilize large-scale international surveys on economic preferences to examine the long-run relationships between patience, current accounts and external wealth. We find robust empirical evidence that countries with more patient individuals tend to run persistent current account surpluses, which in turn result in the accumulation of foreign assets. This theoretically plausible but empirically unexplored relationship holds true for euro area current account imbalances, global current account imbalances and net foreign asset positions worldwide. While the existing current account literature concentrates on proximate macroeconomic determinants, this paper’s extension of inferences from deep determinants (i.e., time preferences) to the distribution of external wealth of nations (i.e., net foreign asset positions) makes a unique contribution to the literature. Our finding showing that deep heterogeneities contribute to external imbalances suggests that the pattern of current account imbalances as well as the distribution of external wealth of nations might be very persistent. Ó2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). 1. Introduction Time preference is one of the most fundamental concepts in economics (see, e.g., Frederick et al., 2002 for a historical foundation of the discounted utility model). Standard economic theory proposes that if two countries with different time preference factors integrate, the country with less patient individuals becomes a net debtor, while the country with more patient individuals becomes a net creditor. However, due to the lack of a global dataset on economic preferences, this proposition has not been empirically tested before our paper. Falk et al. (2018) introduce the Global Preferences Survey (GPS) and provide comprehensive evidence that although within-country heterogeneity in economic preferences is larger than heterogeneity across countries, there is substantial between-country variation, for example, in patience. Additionally, a study by Wang et al. (2016), who performed the first large-scale international survey on time preferences, reveals large crosscountry variation in patience. https://doi.org/10.1016/j.jimonfin.2021.102517 0261-5606/Ó2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). q The author thanks guest editors Ansgar Belke, Joscha Beckmann and Daniel Gros as well as Bas van Aarle, Adam Gulan, Alok Kumar, Tuomas Malinen, Julia Niemeläinen and participants at the Jean Monnet Workshop on ‘‘Current Account (Im-)Balances of the Euro Area and within the Euro Area Member Countries – Determinants and Policy” in Brussels and at the XL Annual Meeting of the Finnish Economic Association in Turku for useful comments and suggestions. Juha Saarinen provided outstanding research assistance. The author acknowledges financial support from the OP Group Research Foundation. ⇑ Address: Jyväskylä University School of Business and Economics, Mattilanniemi 2, Agora Building, University of Jyväskylä, P.O. Box 35, FI-40014 Jyväskylä, Finland. E-mail address: [email protected] Journal of International Money and Finance xxx (xxxx) xxx Contents lists available at ScienceDirect Journal of International Money and Finance journal homepage: www.elsevier.com/locate/jimf Please cite this article as: M. Nieminen, Cross-country variation in patience, persistent current account imbalances and the external wealth of nations, Journal of International Money and Finance, https://doi.org/10.1016/j.jimonfin.2021.102517 This paper utilizes the first two large-scale international surveys on economic preferences mentioned above and provides a novel behavioral economics related answer to the following two research questions: 1) What are the determinants current account imbalances? 2) What are the determinants of the external wealth of nations? In this paper, we show that these external imbalances are explained, among other factors, by cross-country variation in patience. This new macro-level finding is in line with experimental individual-level studies that show that patience is positively related to saving (Sutter et al., 2013; Falk et al., 2018) and negatively related to indebtedness (Meier and Sprenger, 2010). In our analysis, we go deeper than the usual proximate macroeconomic determinants of current accounts and to take a step from net capital flows (i.e., current accounts) to net foreign asset positions. The fact that our results are based on two independent surveys on economic preferences increases the credibility of our findings. Until the 2009 euro crisis, the euro area as a whole had been in balance with the rest of the world. However, at the country level, several countries in the Economic and Monetary Union (EMU) experienced substantial current account imbalances (see Fig. A1 in the Appendix B). These imbalances had a tendency to increase after the adoption of the common currency in 1999. Most often, this pattern is explained by a catching-up process between rich Northern Europe and poor Southern Europe (see, e.g., Blanchard and Giavazzi, 2002; Schmitz and von Hagen, 2011) or by a divergence in cost competitiveness between the two regions (see, e.g., Arghyrou and Chortareas, 2008; Belke and Dreger, 2013). However, it is well known that there were large and persistent differences in the short-term interest rates among the EMU-12 countries 1 before they entered the third phase of the EMU (see Figs. A2-A3 in the Appendix B). According to economic theory, differences in autarky interest rates are linked with differences in time preferences (see Eq. (4) in Section 2 and Fig. A4 in the Appendix B). Countries that accumulated the largest current account deficits after the adoption of the euro were those that had the highest short-term interest rates before the EMU period (Fig. A5 in the Appendix B) and the lowest patience measures (Fig. A6 in the Appendix B). Many studies have explored the medium-term determinants of current accounts in a global context (see, e.g., Chinn and Prasad, 2003; Chinn and Ito, 2007; Gruber and Kamin, 2007; Ca’ Zorzi et al., 2012). The literature has concentrated on macroeconomic factors such as GDP per capita and government budget balance as well as institutional variables that measure differences in financial development and political stability. In this paper, we provide robust empirical evidence that patience (impatience) is associated with persistent current account surpluses (deficits). This holds true both for the euro area current account imbalances and the global current account imbalances. This empirical finding provides an unconventional yet theoretically plausible answer to the research question ‘‘What are the determinants of current account imbalances?” As preferences are fundamental determinants of economic development, our argumentation delves deeper than previously proposed hypotheses, which, for the most part, are based on proximate macroeconomic determinants. One can legitimately ask why current accounts matter. The main reason they matter is that over the long run, the cumulated current account tracks the net foreign asset position (Obstfeld, 2012). Hence, we extend our analysis to include the external wealth of nations and show that there is a positive relationship between patience and net foreign asset positions. This empirical result provides an unconventional yet theoretically plausible answer to the research question ‘‘What are the determinants of the external wealth of nations?” Overall, our findings showing that deep heterogeneities contribute to external imbalances suggest that the pattern of global current account imbalances as well as the distribution of external wealth of nations might be surprisingly persistent and less influenced by policy makers. We cannot rule out the possibility that financial integration among countries with heterogeneous time preferences results in unsustainable divergence in external positions. Hence, when assessing a country’s eligibility to join the EMU, considering similarities in time preferences across countries along with the Maastricht convergence criteria might be worthwhile. To produce any scientifically credible findings on the relationship between patience and net foreign asset positions, we must use economic theory as well as a rich dataset. Economic theory is needed to derive testable hypotheses and to explain empirical findings. An extensive set of control variables is needed to rule out any other plausible explanations. Hence, in Section 2, we derive testable hypotheses from economic theory, and in Section 3, we introduce our data and econometric specifications. In Section 4, we present our empirical results on the relationships between patience, current accounts and external wealth. Section 5 includes conclusions and a discussion. 2. Theoretical framework and testable hypotheses We present the simplest economic model from which we can derive testable hypotheses on time preference and current accounts. In the empirical analysis, we implicitly allow for a more rigorous theoretical model by controlling for several factors suggested by these alternative theories. Obstfeld and Rogoff (1996) show that if we generalize a time-separable utility function to an infinite-horizon setting, the representative agent maximizes U t ¼lim T!1 uC t ðÞþbuC tþ1 ðÞþb 2 uC tþ2 ðÞþ  ¼X 1 s¼t b st uC s ðÞ;ð1Þ 1 EMU-12 refers to the twelve countries that adopted the euro by 2001. M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 2 where u(C t ) is the period utility function, bis the time preference factor (also called the subjective discount factor), 2 and C t is consumption in period t. The intertemporal Euler equation is a necessary first-order condition for utility maximization. For every period s t, it must hold that u 0 C s ðÞ¼1þrðÞbu 0 C sþ1 ðÞ;ð2Þ where u 0 ðÞdenotes the first derivative of the utility function, and r is the interest rate. 3 By definition, consumption is constant in the steady state. Hence, if we assume that the time preference factor bdoes not depend on consumption, Eq. (2) implies that in the infinitely lived, representative agent, small country model, the following equality holds in the steady state: b¼1 1þr:ð3Þ (Obstfeld and Rogoff, 1996, pp. 60-72.) This can be written as follows: r A ¼1 b1;ð4Þ where r A is the autarky interest rate (i.e., interest rate equating saving and investment in a closed economy) and b2ð0;1Þ. According to Eq. (4), the autarky interest rate is determined by the time preference factor. In other words, we expect the following: Auxiliary Hypothesis 1A There is a negative relationship between patience and autarky interest rates. Let us assume that the world consists of two countries called Patient and Impatient. A representative agent of Patient has a high time preference factor, b, whereas a representative agent of Impatient has a low time preference factor. If the two countries become financially integrated, in equilibrium, the world interest rate equates global saving to global investment (Obstfeld and Rogoff, 1996, p. 31). As illustrated by a Metzler diagram, Patient becomes a net creditor, whereas Impatient becomes a net debtor (see Fig. A7 in the Appendix B;Obstfeld and Rogoff, 1996, pp. 31-34; Metzler, 1960). In other words, we expect the following: Auxiliary Hypothesis 1B After financial integration, there is a negative relationship between autarky interest rates and current account balances. On the other hand, Eq. (4) implies that if the world consists of two countries with different rates of time preferences, there cannot be a steady state with international mobility of financial capital. This point is made, for example, by Buiter (1981).To solve this problem, he proposes an overlapping-generations model in which individuals live for two periods and time preference determines the lifetime consumption profile. 4 In this case, the steady state is the sequence of momentary equilibria in which each generation’s lifetime consumption remains constant. In the steady state, first-period consumption does not need to equal second-period consumption, and thus, the interest rate does not need to equal the rate of time preference (Buiter, 1981). In other words, the difference in the rate of time preference has only a limited effect in an overlapping-generations model because the chain of planning for the future is blocked by the lack of intergenerational linkage (Fukao and Hamada, 1989, p. 12.) Another way to circumvent the problem is to assume varying time preferences (intertemporally nonadditive preferences) (see, e.g., Fukao and Hamada, 1989; Obstfeld, 1990). In addition to the problems posed by the concept of steady state, there is one practical issue: can the autarky interest rates be proxied by some actual rates, and when do countries become financially integrated? Despite these difficulties, standard economic theory suggests that if there is cross-country variation in patience, these differences are linked to the pattern of current account imbalances. All in all, we expect the following: Hypothesis 1. There is a positive relationship between patience and long-run current account balances. As stated, the main reason current accounts matter is that over the long run, the cumulated current account tracks the net foreign asset position (Obstfeld, 2012). If we do not consider valuation changes in gross foreign assets and liabilities, it holds that CA t ¼NFA t NFA t1 ;ð5Þ where CA t is the current account balance in period t, and NFA t is the net foreign assets position at the end of period t. Although valuation changes can be substantial (see, e.g., Gourinchas and Rey, 2014, Section 2.3) and the initial net foreign asset position plays a role, the balance of payments (e.g., Eq. (5)) implies the following: 2 A low bimplies impatience, and a high bimplies patience. 3 It is assumed that the period utility function u(C s ) is strictly increasing in consumption and strictly concave (u0Cs ðÞ>0 and u00 Cs ðÞ<0) and lim Cs!0u0Cs ðÞ¼1. 4 See also Ghironi et al. (2008), who build a general equilibrium model that generates non-zero steady-state net foreign asset positions by allowing for different discount factors across countries. M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 3 Auxiliary Hypothesis 2A Over long time spans, there is a close connection between cumulative current accounts and endof-period net foreign asset positions. If we combine the lessons from the two-country general equilibrium model and the balance of payments, we expect that in the long-run, cross-country variation in patience is linked to the dispersion of external wealth of nations. All in all, we expect the following: Hypothesis 2. There is a positive relationship between patience and net foreign asset positions. In sum, Hypothesis 1 relates to our first research question, ‘‘What are the determinants of current account imbalances?” and Hypothesis 2 relates to the second research question, ‘‘What are the determinants of the external wealth of nations?” 3. Data and econometric specifications 3.1. Survey-based data on patience Our data on patience are taken from two independent surveys, i.e., the GPS introduced by Falk et al. (2018) and survey data by Wang et al. (2016). Throughout the paper, all estimations are carried out using both of these data separately. At no stage do we mix these two data. By patience we refer to survey-based data on intertemporal choices between soonersmaller and later-larger rewards, which is our proxy for time preference (i.e., bin the time-separable utility function (Eq. (1))). Cohen et al., (2020) call these ‘‘money earlier or later” (MEL) experiments. The GPS introduced by Falk et al. (2018) is an experimentally validated survey dataset. The GPS data were collected within the framework of the 2012 Gallup World Poll. Consequently, the GPS is the first global dataset on economic preferences that is representative at the country level. It covers 76 countries and more than 80,000 participants worldwide. The measure for patience is derived from the combination of responses to two survey measures, one with a quantitative format (intertemporal choice sequence using the staircase method) and one with a qualitative format (‘‘How willing are you to give up something that is beneficial for you today in order to benefit more from that in the future?”) (Falk et al., 2018). The values are differences to the world mean in the standard deviation of patience. 5 Wang et al. (2016) conducted the first large-scale international survey on time preferences. The survey was part of a larger study called the International Test on Risk Attitudes (INTRA) conducted by the University of Zurich. In total, 6912 university students in 53 countries participated in the survey. One question was ‘‘Which offer would you prefer, a payment of 3400 US dollars this month, or a payment of 3800 US dollars next month?” The measure for patience was the share of the participants in each country who chose to wait for the 3800 US dollars next month. (Wang et al., 2016.) We call these two measures of patience ‘‘Patience (GPS)” and ‘‘Patience (Wang et al. 2016)”. Both of them are timeinvariant as Patience (GPS) was measured in 2012 and Patience (Wang et al. 2016) mainly in 2008 and 2009 (Rieger et al., 2015, p. 645). Having conducted a large field study, Meier and Sprenger (2015) provide empirical evidence that with regard to time preference, aggregate choice profiles and corresponding estimates of discount parameters are stable over time. Becker et al. (2018) utilize the GPS data and show that differences in preferences between populations are significantly increasing in the length of time elapsed since the ancestors of the respective groups drifted away from each other. In other words, ancient origins explain a large fraction of the global variation in economic preferences. Using different data, also Galor and Özak (2016) provide evidence for the historical origins of time preference. Based on these observations, we infer that it is reasonable to assume stability in our measures of patience. Yet, results on the relationship between patience and current accounts in 2015 are provided in Table A7 in the Appendix B. Overall, the relationship is insensitive to the time period of current accounts. The country-level correlation between the two measures of patience is 0.58 (see the scatter plot in Fig. A8 in the Appendix B). Descriptive statistics of these two variables are provided in Table 1. 6 One should remember that as shown by Falk et al. (2018,Table 3), in addition to between-country variation, there is substantial within-country variation in patience. Although the GPS is representative at the country level and covers a larger number of countries than Wang et al. (2016), the latter is utilized for the following two reasons: It enables us to perform a robustness check on our empirical results on patience, and it covers some additional countries that the GPS does not cover. As this is an empirical study on the relationship between patience and external imbalances, it is of great importance that our results are insensitive to which survey dataset is used to measure patience. 3.2. Testing Hypothesis 1 We test Hypothesis 1, ‘‘There is a positive relationship between patience and long-run current account balances”, in Section 4.1. We estimate the following cross-sectional regression model by the OLS estimator: 5 Negative values do not imply negative betas in equation (1). 6 A detailed description of the other independent variables and the data sources are provided in Table 2 in Section 3.4 and in Table A1 in the Appendix A. M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 4 CA  i ¼ a þdPatience i þx  0 i /þ e i ;ð9Þ where the dependent variable is a multi-year average of the current account balance, a is an intercept, Patience i is the patience measure for country i from the GPS or the share of participants choosing the ‘wait’ option in country i in Wang et al. (2016),x i is a vector of control variables, and e i is a residual. The set of control variables is derived from the current account literature (see the listing of variables in Section 3.4). 3.3. Step-by-step testing procedure for Hypothesis 2 A step-by-step testing procedure for Hypothesis 2 is based on the following reasoning: variation in patience across countries is related to cross-country variation in cumulative current account balances 7 , which in turn is related to cross-country variation in net foreign asset positions. In addition, we test the direct association between patience and net foreign asset positions. 3.3.1. Testing Auxiliary Hypothesis 2A In Section 4.2.1, we test how well net foreign asset positions can be tracked by cumulative current accounts over long time spans. This corresponds to testing Auxiliary Hypothesis 2A. We estimate the following cross-sectional regression model by the OLS estimator: NFA i;2014 ¼ a þd 1 X 2014 t¼1985 CA i;t þd 2 NFA i;1984 þ e i ;ð10Þ where the dependent variable is net foreign asset position (ratio to GDP) of country i at the end of 2014, a is an intercept, CA it is the current account balance (ratio to GDP) of country i in period t, and e i is a residual. Net foreign asset position is measured at the end of 2014 because this is the latest year for which we have data. 3.3.2. Testing Hypothesis 2 In Section 4.2.2, we test Hypothesis 2: ‘‘There is a positive relationship between patience and net foreign asset positions”. We estimate the following cross-sectional regression model by the OLS estimator: NFA i;2014 ¼ a þdPatience i þx  0 i /þ e i ;ð11Þ where the dependent variable is net foreign asset position (ratio to GDP) of country i at the end of 2014, a is an intercept, Patience is the patience measure on the GPS or the share of participants choosing the ‘wait’ option in Wang et al. (2016),x i is Table 1 Descriptive statistics for the measures of patience. Variable/Statistic Global country panel Patience (GPS) Min 0.613 Max 1.071 Mean 0.064 St. dev. 0.395 # Countries 54 Patience (Wang et al. 2016) Min 0.080 Max 0.890 Mean 0.654 St. dev. 0.173 # Countries 35 Notes: The statistics are calculated from a sample that includes only the countries for which we have comprehensive data on current accounts for the 1984–2015 period. 7 There is one-to-one relation between the cumulative current account balance over time period t and the long-run average of the current account balance over the same period. M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 5 a vector of control variables, and e i is a residual. Net foreign asset position is measured at the end of 2014 because this is the latest year for which we have data. 3.4. Descriptive statistics, model selection and different samples As already mentioned, to produce any scientifically credible findings on the relationship between patience and current accounts or on the relationship between patience and net foreign asset positions, we must use economic theory as well as a rich dataset. Economic theory is needed to derive testable hypotheses and to explain empirical findings. An extensive set of control variables is needed to rule out any other plausible explanations. We have exhaustive data on the possible determinants of current accounts. The set of control variables is derived from the current account literature (see, e.g., Chinn and Prasad, 2003; Chinn and Ito, 2007; Gruber and Kamin, 2007; Ca’ Zorzi et al., 2012). The descriptive statistics are provided in Table 2. 8 If we consider a linear regression model such as (9) or (11), there is uncertainty about which explanatory variables to include on the right-hand side of the equation. In order to control for this model uncertainty, we follow Fernández et al. (2001) by utilizing Bayesian model averaging with uniform model prior and some other reasonable assumptions. In Sections 4.1 and 4.2.2, we provide posterior inclusion probabilities for each of the explanatory variables in Eqs. (9) and (11) as well as posterior densities for the most significant explanatory variables. The results of posterior inclusion probabilities and posterior densities are based on all possible specifications of the explanatory variables, that is, in this case 65,535 different specifications. 9 To test the statistical significance of explanatory variables, we rely on the following model selection criteria when carrying out a regression analysis on current accounts (Eq. (9)) or net foreign asset positions (Eq. (11)): 10 Bayesian information criterion We rank all specifications in accordance with the Bayesian information criterion (BIC), which takes into account the statistical goodness of fit but also imposes a penalty for the number of explanatory variables. We run the ‘‘best” specification (i.e., the specification with the minimum BIC value), which includes Patience, and we report its ranking score. Akaike information criterion We rank all specifications in accordance with the Akaike information criterion (AIC), which takes into account the statistical goodness of fit but also imposes a penalty for the number of explanatory variables. We run the ‘‘best” specification (i.e., the specification with the minimum AIC value), which includes Patience, and we report its ranking score. In Sections 4.1 and 4.2.2, we present our results for the empirical specifications selected by the BIC and AIC, which are based on estimating Eqs. (9) and (11). When we analyze the long-run relationship between patience, current accounts and external wealth, we attempt to maximize the length of the sample period. However, we have the following two constraints: 1) data series measuring the quality of institutions begin in 1984, and 2) for some explanatory variables, we lack data for 2016 and thereafter. Thus, the sample period 1984–2015 in Section 4.1 results from a constrained optimization problem. For net foreign asset positions, the latest observation is from the end of 2014. Thus, in Sections 4.2.1–4.2.2, the sample period is 1984–2014. With the exception of net foreign asset positions, the numbers that we use are averages for the whole sample period. Thus, the issue of time coverage must be addressed. We follow the rule that if a country lacks no more than one annual observation on current accounts within the 1984–2015 period, it is included; otherwise, it is excluded. 11 When we apply the BIC or the AIC to select empirical specifications, we include all countries for which we have at least 16 annual observations on every explanatory variable within the 1984–2015 period. If we lack more than half of the annual observations on any explanatory variable, we consider the data coverage inadequate, and we exclude the country. The listing of countries in different samples is presented in Table A2 in the Appendix A. In total, there are 81 countries for which we have data on patience. For 60 of these 81 countries, we have comprehensive data on current accounts. This is the sample (n = 60) for which we estimate Eq. (10) in Section 4.2.1. The GPS covers 54 and Wang et al. (2016) 35 of these 60 countries. These are the two samples (n = 54 and n = 35) for which we estimate Eq. (9) in Section 4.1 and Eq. (11) in Section 4.2.2. If we do not follow the rule of current account coverage, we correspondingly have 70 countries (patience from the GPS) and 49 countries (patience from Wang et al. 2016). The results for these two samples, which cover 92% or 85% of the world GDP and 85% or 64% of the world population, are presented in the Appendix B. On the other hand, if we follow the 16/32 rule on all explanatory variables, we end up having 47 countries with the GPS and 32 countries with Wang et al. 2016. These are the two samples (n = 47 and n = 32) from which we calculate the BIC and AIC for all specifications of Eq. (9) in Section 4.1 and for all specifications of Eq. (11) in Section 4.2.2. 8 Descriptive statistics of Patience (GPS) and Patience (Wang et al. 2016) are provided in Table 1 in Section 3.1. 9 In order to save space, these are not provided for Patience (Wang et al. 2016). 10 Ca’ Zorzi et al. 2012 present a similar criterion on current accounts. 11 However, as a robustness check, we provide results in the Appendix B (Tables A4-A5, A9) for when this rule is not followed. M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 6 4. Empirical results 4.1. Patience and current account imbalances In Figs. 1–2, we utilize the Bayesian Model Averaging (BMA) method to assess the determinants of current account balances. In other words, the dependent variable is the average current account balance (ratio to GDP) during the 1984–2015 period in Figs. 1 and 2. The posterior inclusion probabilities of each explanatory variable is presented in Fig. 1. The high inclusion probability of patience (0.513) means that it significantly explains the cross-country variation in current accounts once the potential control variables have been taken into account. Posterior densities for the explanatory variables with the highest posterior inclusion probabilities are shown in Fig. 2. The posterior density of patience is centered on the positive side. This means that countries inhabited by patient individuals have a tendency to run current account surpluses, whereas countries inhabited by impatient individuals have a tendency to run current account deficits. The results of posterior inclusion probabilities and posterior densities (Figs. 1–2) are based on all possible specifications of the explanatory variables, that is, in this case 65,535 different specifications. Table 2 Descriptive statistics based on annual observations for 81 countries over the 1984–2015 period. Variable Min Max Mean St. dev. # Obs. Current account balance 0.429 0.532 0.011 0.072 2351 Net foreign asset position 10.742 4.301 0.229 0.727 2378 Macroeconomic factors Fuel exports 0.000 1.000 0.160 0.259 2218 GDP per capita 0.027 9.159 1.591 1.840 2469 GDP per capita growth 64.996 53.944 2.000 5.228 2475 Government budget balance 0.342 0.328 0.019 0.046 1877 Trade openness 0.085 4.200 0.603 0.414 2435 Financial openness 0.000 1.000 0.551 0.380 2429 Institutional factors Financial development 0.000 2.332 0.560 0.460 2389 Bureaucracy quality 0.000 4.000 2.470 1.127 2452 Corruption 0.000 6.000 3.220 1.411 2452 Democratic accountability 0.000 6.000 4.200 1.583 2452 Investment profile 1.000 12.000 7.662 2.466 2452 Law and order 0.000 6.000 3.884 1.492 2452 Demographic factors Old dependency ratio 0.009 0.427 0.134 0.076 2592 Child dependency ratio 0.149 1.064 0.476 0.230 2592 Notes: The statistics are calculated from a sample that includes only the countries for which we have data on patience. See Table A2 in the Appendix A for a listing of countries. Fig. 1. Inclusion probabilities for each of the explanatory variable of current account balance, 47 countries in 1984–2015. The data on patience are taken from the Global Preferences Survey (GPS). M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 7 In Tables 3–4, we present the results from estimating Eq. (9) for those countries for which we have comprehensive data for the 1984–2015 period. 12 Tables 3 and 4 differ with respect to the data source for patience. In Table 3, data on patience are taken from the GPS and in Table 4 from Wang et al. (2016). Our set of control variables is derived from the current account literature (see, e.g., Chinn and Prasad, 2003; Chinn and Ito, 2007; Gruber and Kamin, 2007; Ca’ Zorzi et al., 2012). In Table 3, we find that there is a strong positive linear relationship between current account balances and patience (specification (1)). 13 This result is not driven by an outlier (see Fig. A9 in the Appendix B). 14 In specifications (2)-(3), all countries with at least 16 annual observations on every explanatory variable within the 1984–2015 period are included. 15 The BIC does not include patience in the first best specification. Nevertheless, out of the 65,535 specifications, it is included in the secondbest statistical model (specification (2)). Based on the AIC, patience should be included in the statistical model, even if all other typical determinants are controlled for (specification (3)). However, the statistical significance of Patience (GPS) is slightly weaker in specification (3). This is because the correlation between Patience (GPS) and GDP per capita is as high as 0.79 in this particular sample (see Table A3 in the Appendix B). Falk et al. (2018) show that patience is the only preference measure that is robustly correlated with GDP per capita. The time preference factor is a deep determinant of both external balances and economic development. Hence, distinguishing the effect of patience on external balances from the effect of GDP per capita is econometrically challenging. Instead of fancy econometric identification, we follow the growth literature and categorize determinants as deep or proximate causes Fig. 2. Posterior densities for the most important explanatory variables of current accounts, 47 countries in 1984–2015. The data on patience are taken from the Global Preferences Survey (GPS). 12 All countries with at least 31 annual observations on current accounts within the 1984–2015 period are included (see Table A2 in the Appendix A for a listing of countries). See Tables A4-A5 in the Appendix B for the results when this rule was not followed. The results are identical. See Table A6 in the Appendix Bfor the results for the EMU-12 countries during the first 10 years after the adoption of the euro preceding the crisis period. Again, the results are identical. 13 The quadratic term of Patience is not statistically significant. 14 If, for some reason, one wishes to exclude Nicaragua (NIC), the positive relation between the average current account balance and patience remains statistically significant at the 1 % level in specification (1). 15 In addition to GDP per capita, government budget balance, trade openness, financial openness, corruption, child dependency ratio, and initial net foreign asset position, the set of control variables also included fuel exports, GDP per capita growth, old dependency ratio, financial development, bureaucracy quality, democratic accountability, law and order, and investment profile from the Political Risk Services’ International Country Risk Guide. M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 8 Table A1 Data sources and variable descriptions. Variable Description Frequency Source a Current account balance WDI: Current account balance (BoP, current US$) WDI: Current account balance (ratio to GDP) (global country panel) WEO: Current account balance (U.S. dollars) Annual WDI, WEO (United Arab E., Belgium until 2001) Net foreign asset position / Initial net foreign asset position ‘NFAY’ as a ratio / ‘NFAY’ in 1984 as a ratio Annual EWNII GDP GDP (current US$) Annual WDI Patience (GPS) Difference to world mean in standard deviation of patience Timeinvariant GPS Patience (Wang et al. 2016) Share of participants choosing the ‘wait’ option Timeinvariant Wang et al. 2016 Fuel exports Fuel exports (share of merchandise exports) Annual WDI GDP per capita GDP per capita (constant 2010 US$) in tens of thousands of dollars Annual WDI GDP per capita growth Percent change in GDP per capita (constant 2010 US$) Annual WDI Government budget balance Government budget balance (ratio to GDP) WDI: Net lending (+) / net borrowing () WEO: General government net lending/borrowing (CMR, CHN, HTI, HKG, SAU) Annual WDI, WEO Trade openness Merchandise trade as a ratio to GDP Annual WDI Financial openness Chinn-Ito index (ka_open). The index measures financial account openness. Scaled between 0 and 1. Annual CI Financial development Domestic credit to private sector (ratio to GDP) Annual WDI / GFD Bureaucracy quality Scaled between 0 and 4. Higher values imply better institutional quality. Annual PRS Corruption / Law and order / Democratic accountability Scaled between 0 and 6. Higher values imply better institutional quality. Annual PRS Investment profile Scaled between 0 and 12. Higher values imply better institutional quality. Annual PRS Old dependency ratio Number of people aged 65 or more divided by the number of people aged 15–64 Annual WDI Child dependency ratio Number of people aged 0–14 divided by the number of people aged 15–64 Annual WDI a CI: Financial openness index by Chinn and Ito <(http://web.pdx.edu/~ito/kaopen_2015.xls>; EWNII: External Wealth of Nations Mark II database by Lane and Milesi-Ferretti <https://www.imf.org/external/pubs/ft/wp/2006/data/update/wp0669.zip>; GFD: Global Financial Development Database; GPS: Global Preferences Survey; PRS: Political Risk Services’ International Country Risk Guide (Table 3B); Wang et al. 2016: Wang, M., Rieger, M. O., Hens, T. 2016. How time preferences differ: Evidence from 53 countries. Journal of Economic Psychology 52, 115–135, Table 2 (working paper for France); WDI: World Development Indicators (World Bank); WEO: World Economic Outlook Database, October 2017 (International Monetary Fund). Table A2 Listing of countries in different samples. Country Abbr. GPS Wang et al. 2016 CA 1984-2015 Control variables Algeria DZA X X Angola AGO X X Argentina ARG X X X X Australia AUS X X X X Austria AUT X X X X Azerbaijan AZE X Bangladesh BGD X X Belgium BEL X X Bolivia BOL X X X Botswana BWA X X X Brazil BRA X X X Cameroon CMR X X X Canada CAN X X X X Chile CHL X X X X China CHN X X X X Colombia COL X X X Costa Rica CRI X X X Croatia HRV X X Czech Rep. CZE X X Denmark DNK X X X Egypt EGY X X X Estonia EST X X (continued on next page) M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 15 Table A2 (continued) Country Abbr. GPS Wang et al. 2016 CA 1984-2015 Control variables Finland FIN X X X X France FRA X X X X Germany DEU X X X X Ghana GHA X X Greece GRC X X X X Guatemala GTM X X X Haiti HTI X Hong Kong HKG X X Hungary HUN X X X X India IND X X X X Indonesia IDN X X X Iran IRN X Iraq IRQ X Ireland IRL X X X Israel ISR X X X X Italy ITA X X X X Japan JPN X X X X Jordan JOR X X X Kazakhstan KAZ X Kenya KEN X X X Korea KOR X X X X Lebanon LBN X X Lithuania LTU X X Malawi MWI X X Malaysia MYS X X X Mexico MEX X X X X Moldova MDA X X Morocco MAR X X X Netherlands NLD X X X X New Zealand NLZ X X X Nicaragua NIC X X X Nigeria NGA X X X Norway NOR X X X Pakistan PAK X X X Peru PER X X X Philippines PHL X X X Poland POL X X X X Portugal PRT X X X X Romania ROU X X Russia RUS X X Saudi Arabia SAU X X X Slovenia SVN X South Africa ZAF X X X Spain ESP X X X X Sri Lanka LKA X X X Suriname SUR X X Sweden SWE X X X X Switzerland CHE X X X X Tanzania TZA X X Thailand THA X X X X Turkey TUR X X X X Uganda UGA X X X Ukraine UKR X United Arab E ARE X X UK GBR X X X X US USA X X X X Venezuela VEN X X X Vietnam VNM X X Zimbabwe ZWE X # Countries 70 49 60 56 M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 16 Appendix B. Supplementary material Supplementary data to this article can be found online at https://doi.org/10.1016/j.jimonfin.2021.102517. References Acemoglu, D., 2009. Introduction to Modern Economic Growth. Princeton University Press, Princeton (NJ). Arghyrou, M.G., Chortareas, G., 2008. Current account imbalances and real exchange rates in the euro area. Rev. Int. Econ. 9 (5), 747–764. https://doi.org/ 10.1111/j.1467-9396.2008.00773.x. Becker, A., Enke, B., Falk, A., 2018. Ancient origins of the global variation in economic preferences. NBER Working Paper No. 24291. https://www.nber. org/papers/w24291.pdf Belke, A., Dreger, C., 2013. Current account imbalances in the euro area: Does catching up explain the development? Rev. Int. Econ. 21 (1), 6–17. https://doi. org/10.1111/roie.2013.21.issue-110.1111/roie.12016. Blanchard, O., Giavazzi, F., 2002. Current account deficits in the euro area: the end of the Feldstein-Horioka puzzle? Brookings Papers Econ. Activity 2002 (2), 147–209 https://www.brookings.edu/wp-content/uploads/2002/06/2002b_bpea_blanchard.pdf. Buiter, W.H., 1981. Time preference and international lending and borrowing in an overlapping-generation model. J. Polit. Econ. 89 (4), 767–797 https:// www.journals.uchicago.edu/doi/pdfplus/10.1086/261002. Ca’ Zorzi, M., Chudik, A., Dieppe, A., 2012. Thousands of models, one story: Current account imbalances in the global economy. J. Int. Money Finance 31 (6), 1319–1338. https://doi.org/10.1016/j.jimonfin.2012.02.003. Chinn, M., Ito, H., 2007. Current account balances, financial development and institutions: assaying the world ‘‘saving glut”. J. Int. Money Finance 26 (4), 546–569. https://doi.org/10.1016/j.jimonfin.2007.03.006. Chinn, M., Prasad, E., 2003. Medium-term determinants of current accounts in industrial and developing countries: an empirical exploration. J. Int. Econ. 59 (1), 47–76. https://doi.org/10.1016/S0022-1996(02)00089-2. Cohen, J., Ericson, K.M., Laibson, D., Myles White, J., 2020. Measuring time preferences. J. Econ. Lit. 58 (2), 299–347. https://doi.org/10.1257/jel.20191074. Falk, A., Becker, A., Dohmen, T., Enke, B., Huffman, D., Sunde, U., 2018. Global evidence on economic preferences. Q. J. Econ. 133 (4), 1645–1692. https://doi. org/10.1093/qje/qjy013. Fernández, C., Ley, E., Steel, M.F.J., 2001. Model uncertainty in cross-country growth regressions. J. Appl. Econ. 16 (5), 563–576. https://doi.org/ 10.1002/jae.623. Fisher, I., 1930. The Theory of Interest. Macmillan, New York (NY). reprinted by Martino Publishing in 2012. Frederick, S., Loewenstein, G., O’Donoghue, T., 2002. Time discounting and time preferences: A critical review. J. Econ. Lit. 40 (2), 351–401. https://doi.org/ 10.1257/002205102320161311. Fukao, K., Hamada, K., 1989. The Fisherian time preference and the evolution of capital ownership patterns in a global economy. NBER Working Paper No. 3104. https://doi.org/10.3386/w3104 Galor, O., Özak, Ö., 2016. The agricultural origins of time preference. Am. Econ. Rev. 106 (10), 3064–3103. https://doi.org/10.1257/aer.20150020. Ghironi, F., _ Isßcan, T.B., Rebucci, A., 2008. Net foreign asset positions and consumption dynamics in the international economy. J. Int. Money Finance 27 (8), 1337–1359. https://doi.org/10.1016/j.jimonfin.2008.08.009. Gourinchas, P.-O., Rey, H., 2014. External adjustment, global imbalances, valuation effect. In: Gopinath, G., Helpman, E., Rogoff, K. (Eds.), Handbook of International Economics, Volume 4. North-Holland, pp. 585–645. Gruber, J.W., Kamin, S.B., 2007. Explaining the global pattern of current account imbalances. J. Int. Money Finance 26 (4), 500–522. https://doi.org/10.1016/j. jimonfin.2007.03.003. Meier, S., Sprenger, C., 2010. Present-biased preferences and credit card borrowing. Am. Econ. J.: Appl. Econ. 2 (1), 193–210. https://doi.org/10.1257/ app.2.1.193. Meier, S., Sprenger, C.D., 2015. Temporal stability of time preferences. Rev. Econ. Stat. 97 (2), 273–286. https://doi.org/10.1162/REST_a_00433. Metzler, L.A., 1960. The process of international adjustment under conditions of full employment: A Keynesian view. In: Caves, R.E., Johnson, H.G. (Eds.), Readings in International Economics. Richard D. Irwin, Homewood (IL), pp. 465–486. Obstfeld, M., 1990. Intertemporal dependence, impatience, and dynamics. J. Monet. Econ. 26 (1), 45–75. https://doi.org/10.1016/0304-3932(90)90031-X. Obstfeld, M., 2012. Does the current account still matter? Am. Econ. Rev.: Papers Proc. 102 (3), 1–23. https://doi.org/10.1257/aer.102.3.1. Obstfeld, M., Rogoff, K., 1996. Foundations of International Macroeconomics. The MIT Press, Cambridge (MA). Rieger, M.O., Wang, M., Hens, T., 2015. Risk preferences around the world. Manage Sci. 61 (3), 637–648. https://doi.org/10.1287/mnsc.2013.1869. Schmitz, B., von Hagen, J., 2011. Current account imbalances and financial integration in the euro area. J. Int. Money Finance 30 (8), 1676–1695. https://doi. org/10.1016/j.jimonfin.2011.08.003. Spolaore, E., Wacziarg, R., 2013. How deep are the roots of economic development? J. Econ. Lit. 51 (2), 325–369. https://doi.org/10.1257/jel.51.2.325. Sutter, M., Kocher, M.G., Glätzle-Rützler, D., Trautmann, S.T., 2013. Impatience and uncertainty: Experimental decisions predict adolescents’ field behavior. Am. Econ. Rev. 103 (1), 510–531. https://doi.org/10.1257/aer.103.1.510. Wang, M., Rieger, M.O., Hens, T., 2016. How time preferences differ: Evidence from 53 countries. J. Econ. Psychol. 52, 115–135. https://doi.org/10.1016/j. joep.2015.12.001. Weil, D., 2009. Economic Growth. Pearson, Boston (MA). M. Nieminen Journal of International Money and Finance xxx (xxxx) xxx 17