مدل ارزش فعلی در حساب جاری مردود گشته: جمع آوری مظنونهای همیشگی

Tests of the present-value model (PVM) of the current account are frequently rejected by data. Standard explanations rely on the “usual suspects” of non-separable preferences, fiscal policy and world real interest rate shocks, external imperfect international capital mobility, and an internalized risk premium. We confirm these rejections on post-war Canadian data, then investigate their source by calibrating and simulating alternative versions of a small open economy, real business cycle model (RBC). Bayesian Monte Carlo experiments reveal that a “canonical” RBC model is close to the data, but far from the PVM predictions. Although each suspect matters in some way, none improve the fit to the data. However, the PVM restrictions are reproduced when the internalized risk premium is introduced into the canonical model. By adding the exogenous world real interest rate shock to this version of the model, it matches the data better and is moved closer to the PVM predictions. This suggests that there is an important common world component to current account fluctuations, which points to additional underlying macroeconomic factors that drive the current account.

Current account fluctuations resist easy explanations. Large current account deficits have persisted in the U.S. through periods of large government budget deficits and surpluses, large and persistent real appreciations and depreciations of the dollar, and all phases of the business cycle. In Canada, the expansion of the 1980s coincided with large current account deficits, but an expansion in the 1990s witnessed current account surpluses. Economists increasingly employ the intertemporal approach to study the current account. The intertemporal approach views the current account as a tool domestic residents use to smooth consumption by borrowing from or lending to the rest of the world. For example, if future income is expected to rise, say due to a technology shock, domestic agents try to smooth consumption by borrowing abroad prior to the high-income years, thereby running a current account deficit. As such, the intertemporal approach relies on permanent income fluctuations (driven by technology shocks) to explain current account movements. Compared to traditional Keynesian views, the intertemporal approach reduces emphasis on the economy's intratemporal competitiveness measured by the real exchange rate. The intertemporal approach to the current account is encompassed by several classes of small open economy models. The most basic is the present-value model (PVM) of the current account. Sheffrin and Woo (1990), Otto (1992), Ghosh (1995), and Bergin and Sheffrin (2000) test the PVM and find it routinely rejected by the data.1 Despite rejections of the PVM cross-equation restrictions, it is argued that abandoning the underlying scheme is unwarranted. Adherents point out that the (in-sample) current account forecast of the most unadorned PVM often closely tracks the actual current account (e.g., Obstfeld and Rogoff, 1996, pp. 92–94). Thus, the PVM, which is rejected, is seen in the literature as “useful” overall. Although appropriate, this conclusion is unsatisfactory because it fails to say which parts of the intertemporal model are most responsible for the poor empirical performance of the PVM. This paper studies a set of “usual suspects”–factors other than technology shocks that theory teaches can matter for the current account–as potential sources of empirical rejections of the PVM. These factors are non-separable preferences, country-specific fiscal and world real interest rate shocks, imperfect international capital mobility, and an internalized risk premium. We place the suspects in a “canonical” small open economy-real business cycle (RBC) model that nests the PVM and serves as our benchmark intertemporal model of the current account. Our “testing” strategy compares moments of synthetic data produced by the RBC model to those of actual data. Rather than focus on the usual variances and covariances, the “moments” we study are the cross-equation restrictions of the PVM. The actual data we use is from post-war Canada, a proto-type small open economy for which rejections of the PVM are found uniformly in the literature.2 We evaluate the empirical and theoretical distributions implied by the cross-equation restrictions using Bayesian Monte Carlo methods, which measure the fit of the RBC model to the actual data. The canonical RBC model includes no suspects. This economy features perfect international capital mobility and a permanent, country-specific technology shock, but lacks transitory fiscal policy or world real interest rate shocks. Thus, the canonical model nests the PVM. Next, we add suspects to the canonical model one at a time to create alternative specifications. With each alternative, we generate artificial data, test the cross-equation restrictions, and compare synthetic to sample moments. These exercises give us evidence about the culpability of each of the suspects. Our choice of suspects is guided by related work on the intertemporal approach. Non-separable preferences can affect the current account in several ways. Ghosh and Ostry (1997) find that incorporating precautionary saving into the PVM helps to explain current account volatility, but interestingly not for Canada. Bergin and Sheffrin (2000) improve the fit of the PVM by adding non-separable utility between tradable and non-tradable goods and a stochastic world real interest rate. Striking evidence of the importance of fiscal shocks, especially large ones, for intertemporal external borrowing decisions has been presented by Ahmed (1986) and Ahmed and Rogers (1995). Hercowitz (1986) and Blankenau et al. (2001) report that world real interest rate shocks help to explain aggregate fluctuations in small open economies. Cole and Obstfeld (1991) argue that small barriers to international capital mobility negate the benefits of consumption smoothing. Barriers to international capital mobility are modeled by Mendoza, 1991 and Mendoza, 2002, Valderrama (2002), and Schmitt-Grohé and Uribe (2003). Mendoza (1991) puts exogenous capital controls into a RBC model, Mendoza (2002) and Valderrama place bounds on a household's debt–income ratio that play the role of a risk premium in the aggregate optimality conditions, while Schmitt-Grohé and Uribe study a time-varying risk premium in a RBC model. We identify the time-varying risk premium with the debt–output ratio, to create an internalized risk premium, which becomes a suspect. In the next section, we describe the testable predictions of the PVM of the current account and confirm rejections of those predictions on Canadian data. Section 3 presents our small open economy-RBC model, derives its optimality conditions, and explains the numerical methods employed to solve the model. We report Monte Carlo experiment results in Section 4. Conclusions appear in Section 5.

We study the importance of various explanations for the poor empirical performance of a basic intertemporal model of the current account, the present value model (PVM). First, we confirm the results of existing papers that reject the cross-equation restrictions and orthogonality conditions of the PVM, in our case using a sample of post-war Canadian data. To understand these rejections, we construct a small open economy-real business cycle model. We show that a “canonical” version of the model is consistent with the actual data, but not the theoretical PVM predictions. The usual suspects we study, non-separable preferences, shocks to fiscal policy and the real interest rate, imperfect international capital mobility, and an internalized risk premium, are portrayed in the literature as potential explanations of rejections of the PVM. We conduct Bayesian Monte Carlo experiments to generate evidence about the culpability of our suspects. Although each matters in some way, we find the more parsimonious canonical model is close to the data, but far from the PVM restrictions. This is explained with labor responses to permanent income shocks in the small open economy. We show that the portfolio decisions of the internalized risk premium-RBC model negate the labor market response to permanent income movements, which allows the model to replicate the PVM predictions. Since this model is far from the data in some ways, we add each of the remaining suspects one-by-one to the internalized risk premium-RBC model. It is the addition of the world real interest rate shock that moves this model closest to the actual data. Our results indicate that the attention paid to transitory movements in domestic fiscal policy to explain the current account while appropriate may have missed other important factors. The internalized risk premium and exogenous world real interest rate shocks are two of the leading factors that we identify. Thus, future research should look for additional underlying macroeconomic factors that drive the current account. Finally, our results suggest that current account movements have a larger common, cross-country element than is perhaps usually suspected, at least in the case of Canada. The intuition for our results rests with households hedging against country-specific permanent income shocks through the current account. Any transitory shock to consumption generates current account fluctuations independent of movements in permanent income and hence could produce rejections of the PVM. Shocks to the world real interest rate, for example, produce these sorts of current account fluctuations, whether they come from an internalized country-specific component, an exogenous world shock, or an endogenous common world component. This is especially important because ever since Cole and Obstfeld (1991) pointed out that even small imperfections in international capital markets can wipe out the (consumption smoothing) benefits of international portfolio diversification, the sources and causes of such imperfections have eluded researchers. We hope this paper invigorates this research agenda.