Several years ago, as the U.S. current account deficit was expanding to record levels, observers increasingly began to focus on the unsustainability of the U.S. external imbalances, as well as the possibility that the subsequent correction would be abrupt and disorderly.1 After peaking at 6.6 percent of GDP in the third quarter of 2006, however, the current account deficit has shrunk to about 5 percent in 2008 as a result of declines in the dollar, slower U.S. GDP growth, and expansion abroad. Concerns about a disorderly correction now appear to have become less prominent. This may in part reflect a growing conviction that a correction of the current account is likely to be orderly rather than disruptive. It may also reflect a view that, with the real multilateral dollar about 25 percent below its 2002 peak and the deficit reduced, no further correction of the U.S. current account may be necessary.
This paper addresses three simple questions: Is the U.S. current account now sustainable on a long-term basis? If not, how long might it take for measures of U.S. external indebtedness to expand beyond levels that global investors are willing to finance? And finally, if and when such levels are breached, how rapidly might a correction in asset prices and the current account ensue?
The first section starts by discussing the most common metric for assessing current account sustainability, the stability of an economy’s net debt as a share of GDP. Section II describes projections of U.S. external balance variables, based on simulations of a detailed model of the U.S. balance of payments, and measures them against the sustainability criterion described above. We find that, assuming the real value of the dollar remains flat at current levels, beyond the near term the current account deficit likely will begin widening again and U.S. external debt will rise steadily.
However, just because the current account is unsustainable in the long term does not mean that a correction is imminent. Theory provides no guidance as to how large the external debt must become before developments are triggered that would narrow the U.S. current account deficit. Our baseline projection suggests that the U.S. net external debt will grow from around 20 percent of GDP at present to around 60 percent of GDP by 2020. Is that a lot or a little? To answer this question, we look to the current pattern of external liabilities among industrial countries, and we find a number of countries whose external debt ratios currently are 60 percent or higher.
The net debt/GDP ratio, however, is not a perfect or unique indicator of the strength of a country’s international balance sheet. Accordingly, Section III considers a second metric of current account sustainability: the exposure of investors to U.S. assets in terms of the share of U.S. securities in foreign portfolios. If the financing of the current account deficit means that this exposure is rising without limit, that too would suggest that the current account is unsustainable. We find no evidence that, to date, the exposure of foreign investors to U.S. assets has been rising. Calculations based on our projections of the U.S. balance of payments (described above) suggest that this exposure will increase going forward, but not necessarily to a worrisome extent.
Finally, in the event that measures of U.S. external indebtedness were to breach any significant thresholds of creditworthiness and investor exposure, do we know how rapidly global investors might react by pulling back from U.S. assets, thereby engendering a correction of the current account deficit? To address this question, Section IV examines the extent to which, historically, higher levels of external debt or related imbalances have led to changes in an economy’s access to financing. Estimating panel regressions for a sample of industrial countries, we find that higher levels of debt push up interest rates in a country by only a very small extent, and exert no discernable effect on exchange rates.
Section V summarizes the findings and advances some tentative conclusions.
I. Assessing Current Account Sustainability
The NIIP/GDP Criterion
The net international investment position (NIIP) represents the sum of all claims by U.S. residents on foreign residents less the claims of foreigners on the United States. The NIIP is a key determinant (along with rates of return) of U.S. net investment income: the sum of receipts on foreign assets owned by U.S residents net of payments on foreign claims on U.S. residents. Therefore, most analysts underscore that a necessary condition for current account sustainability is that the NIIP/GDP ratio be stable (Mann, 1999 and 2002 and 2003; Mussa, 2004; Cline, 2005). Otherwise, if the (negative) NIIP/GDP ratio were to rise without limit, the ratio of net investment payments to GDP would rise as well, and would eventually exceed GDP.
Qualifications to the NIIP as a Measure of External Sustainability
The NIIP does not fully summarize the sustainability of the external position. First, as will be discussed further below, depending upon the rate of return, the same NIIP may be associated with very different net investment income flows.
Second, changes in the valuation of assets may affect the NIIP without affecting the economy’s underlying capacity to service the external position. For example, all else equal, a rise in U.S. stock prices will raise the value of foreign holdings of U.S. assets and thus cause the NIIP to become more negative, even though U.S. residents have become wealthier and better able to make payments to foreigners. An extreme example is Finland, where a substantial share of its external liabilities consists of foreign holdings of stock in Nokia. In 1999, the NIIP surged to nearly −170 percent of GDP in 1999, driven by a parallel surge in the price of Nokia stock; when that stock declined, so, too, did the size of Finland’s negative NIIP.
Third, and as a related point, one should not confuse the NIIP with the net wealth of an economy’s residents. Net wealth is comprised of total assets owned by residents, both domestic and foreign, less foreigners’ claims on those residents. This amounted to over $55 trillion in 2006, dwarfing the size of both gross claims of foreigners on the United States ($14.4 trillion) and the NIIP (–$2.2 trillion). Presumably, a country’s ability to repay external debt will depend not only on its GDP, but also on its total net wealth, just as a homeowner’s ability to repay his mortgage depends not just on his income, but on the value of his assets including, but not limited to, his house.
Finally, the aggregate NIIP may be only loosely related to the ability of U.S. residents to repay their external liabilities. For most international borrowers in an industrial economy, foreigners represent only a small portion of their creditor base, with most liabilities being to domestic residents. The diversity and distribution of creditworthiness across borrowers in a given economy is likely to influence the credit risks faced by foreign investors to a greater degree than the overall foreign indebtedness of the economy. (As discussed in more detail in Section II, the riskiness of U.S. external debt is reduced by the fact that the major debtor is the U.S. government.) This is reinforced by the fact that in most industrial countries, it is not possible to treat foreign creditors differently from domestic creditors.
II. Simulations of the U.S. Balance of Payments
With these qualifications in mind, we now consider projections of the U.S. balance of payments to determine whether the key criterion for current account sustainability—the stability of the NIIP/GDP ratio—is likely to be met. For this exercise, we use the Federal Reserve Board’s partial-equilibrium model of the balance of payments, which is described briefly below. (A more complete description is provided in the appendix to Bertaut, Kamin, and Thomas, 2008.)
Before proceeding, we address the desirability of using a partial equilibrium model—which assumes the key macroeconomic drivers of the current account to be exogenous—to assess current account sustainability. In principle, the new generation of forward-looking dynamic general equilibrium models might be better suited for longer run macroeconomic projections. However, in these models, trade deficits and external debt represent equilibrium responses to shocks, and asset prices and deficits start to correct long before economies reach any putative debt limits (see, for example, Erceg, Guerrieri, and Gust, 2006). Conversely, much of the debate over U.S. external sustainability assumes that U.S. indebtedness may expand until it reaches certain limits, after which correction may be triggered. In this context, it makes sense to use a partial equilibrium model, which does not assume spending adjusts endogenously in response to future financing constraints, to forecast the paths of external imbalances and debt under plausible assumptions about output growth and prices, and then to assess whether those paths are sustainable. Moreover, the U.S. international transactions (USIT) partial-equilibrium model described below treats the U.S. balance of payments in considerably more detail than any general equilibrium model currently in use, and such detail is essential to assessing current account sustainability
The U.S. International Transactions Model
The USIT model consists of 491 equations including 26 econometrically estimated behavioral equations, with the rest being identities and other computational equations. The model takes as exogenous projections for the central determinants of the U.S. external accounts, including: U.S. and foreign real GDP growth, U.S. and foreign inflation rates, U.S. interest rates, oil prices, and the foreign exchange value of the dollar. Based on these inputs, it then projects U.S. external balance variables in four broad categories: (1) trade flows, (2) nontrade components of the current account (especially investment income), (3) financing flows, and (4) the investment positions comprising the NIIP. Salient aspects of the modeling strategy and parameters are as follows:
(1) Trade sector
Import prices for most major categories are projected based on the level of the dollar, foreign consumer price indices (CPIs), and the U.S. CPI; the rate of pass-through from changes in the dollar to changes in merchandise import prices is quite low, about one-third. Export prices depend on measures of U.S. production costs and final prices.
Real imports for most major categories depend on both the prices of imports relative to U.S. prices—with an elasticity of about unity—and on U.S. GDP. Real exports for most major categories depend on the price of exports relative to exchange-rate-converted foreign CPIs—also with an elasticity of about unity—and on a trade-weighted aggregate of foreign GDPs. Importantly, these equations incorporate the Houthakker-Magee asymmetry in income elasticities: the elasticity of real imports with respect to U.S. GDP is about 1.75 on average, exceeding the elasticity of real exports with respect to foreign GDP, of about 1.25 on average.2
Both the quantity and price of oil imports are modeled separately, with the former based on trends in U.S. production and consumption of oil, and the latter based on current and prospective market developments.
(2) Nontrade components of the current account balance
Investment income is projected by applying income rates of return to different categories of U.S. external claims and liabilities.
Assumptions on U.S. interest rates are used to project income rates on portfolio (equity, bond and deposit) positions. Skipping ahead slightly to Figure 1, the rates of income on U.S. portfolio assets and liabilities have been roughly similar in recent years and are projected to remain so, at a level close to the projected U.S. short-term rate of interest, going forward.
The income rate of return on foreign direct investment in the United States depends on the U.S. output gap. The rate of return on U.S. direct investment abroad depends on the foreign output gap and the relative price of oil.
Historically, income rates on U.S. direct investment abroad have exceeded that on foreign direct investment in the United States. Although this gap has narrowed over the past decade, it remains large and we project it to remain large in the future.3
Transfers are projected exogenously, based on recent trends.
Figure 1.Extended Baseline Projection of the USIT Model
Sources: U.S. Bureau of Economic Analysis; Federal Reserve Board of Governors; and national authorities.
Note: Projection period is shaded area.
(3) Financing flows
Once the current account balance is projected, this pins down the amount of net financing flows into (out of) the U.S. economy. However, two additional facets of these flows must be specified. First, financial flows must be allocated among the various categories: direct investment and portfolio investment. Second, a given amount of net financing may be associated with any number of combinations of gross financing flows. For example, an $800 billion net inflow may be achieved by $800 billion in gross inflows from abroad, combined with zero outflows; or it could be achieved by $1,600 billion in gross inflows and $800 billion in gross outflows.
Gross direct investment flows to/from the United States depend on GDP growth in the recipient country.
Foreign flows into U.S. government assets and U.S. government flows into foreign assets are projected at their recent trends.
This leaves net private portfolio flows to balance the current account/financial account identity. The gross flows are constructed so as to produce growth rates in the stock of claims and liabilities that most closely match recent history while still having the implied net flows meet the net financing requirement.4
(4) Investment positions
For each category of investment (direct investment, private portfolio, government portfolio, etc.), the gross asset or liabilities position in a given year will be equal to the position in the preceding year plus (a) financial flows (positive or negative) in that category during the year, and (b) valuation changes in the position.
For all the simulation exercises, we report the direct investment positions and the NIIP using the current cost measure of direct investment.5
In the simulation projections presented here, we include valuation changes to the direct investment positions alone, not the portfolio positions. The valuation changes for the direct investment positions reflect changes in exchange rates and in domestic prices of the assets (land, machinery, structures, and so on) comprising the position.
Key Assumptions for the Projection
The most important assumptions underlying the baseline balance-of-payments projection are shown in Figure 1.
The real multilateral exchange value of the dollar is held constant at its level at the beginning of 2008.
For U.S. real GDP growth, rates in 2008 and 2009 are based on Organization for Economic Development (OECD) projections, while longer-term growth rates of 2.4 percent are based on the OECD assessment of U.S. potential GDP growth in 2009 (OECD, 2008). In between the near term and farther out, growth jumps temporarily to restore the actual level of GDP to potential, after which growth subsides to its potential rate and the output gap remains at zero.
GDP growth in the foreign industrial economies is projected in the same manner, based on OECD (2008). GDP growth in developing countries is based on the IMF’s World Economic Outlook projections for 2008, on the average growth rate from 1997 to 2007 for further out, and also incorporates a transitional period to restore levels of GDP to their potential levels.
For 2008 and 2009, U.S. long- and short-term interest rates are based on OECD (2008). Beyond a transitional period, long-term rates are set equal to the projected growth of nominal GDP: real growth of 2.4 percent plus inflation of 2 percent, based on the OECD 2009 projection. Short-term rates are set 1 percentage point below long-term rates.
The price of imported oil is set flat at its value at the beginning of this year. Note that imported oil is comprised of a mix of different grades, and its price is accordingly lower than that of West Texas Intermediate, whose price approached $100 per barrel around that time.
The Baseline Projection
As indicated in Figure 2, after some initial wiggles, real exports grow at a pace of nearly 5 percent over most of the projection period. Real imports, reflecting their higher elasticity with respect to GDP, grow a touch faster than 5 percent. In consequence, the trade balance widens gradually as a share of GDP, reaching about 5 percent by 2020. The nonoil trade deficit expands at a faster pace, as the volume of oil imports rises more slowly than other types of imports, consistent with trend declines in the oil-intensity of U.S. GDP.6 The current account balance also declines gradually as a share of GDP, reflecting not only the widening trade deficit, but also a (long-expected) shift in the balance on investment income from positive to negative as the NIIP gets more negative; net investment income and the current account deficit would deteriorate even faster were it not for the asymmetry in the rates of return on foreign direct investment. Finally, reflecting all of these developments, the NIIP/GDP ratio deteriorates steadily over the projection period, reaching over −60 percent of GDP by 2020.
Figure 2.Extended Baseline Projection, Continued
Sources: U.S. Bureau of Economic Analysis; Federal Reserve Board of Governors; and national authorities.
Note: Projection period is shaded area.
Because of their importance for net investment income and thus the current account balance, the bottom panels of Figure 1 provide more detail on the composition of projected gross investment positions and on the rates of return on these positions. The bottom right panel indicates that the high rates of return on direct investment claims that we are assuming are well in line with past history, while the rate of return on direct investment liabilities is, if anything, generous by historical standards. The similarity in our projection of rates of return on portfolio claims and liabilities is also supported by history. The bottom left panel shows that increases in U.S. direct investment claims, which have the potential to significantly improve the net investment income balance owing to their high rate of return, do not appear out of line with the evolution of other categories.
We draw four central conclusions from this projection. First, based on the standard criterion—stability of the NIIP/GDP ratio—the U.S. current account balance is not sustainable in the long term. The NIIP/GDP ratio deteriorates by more than 40 percentage points of GDP by 2020, or very roughly 4 percentage points per year. But, second, even if the current account is unsustainable, it is probably less unsustainable than would have been the case had we started the projection back in, say, 2000; at that time, as indicated at the top of Figure 1, the real value of the dollar was considerably higher, and U.S. and foreign growth seemed more similar. Moreover, and third, it is doubtful that, by 2020, the U.S. balance of payments will be entering any danger zone where external adjustment will be urgently required. Although the level of the net debt appears quite elevated (this will be discussed further below), net investment income payments still represent a paltry ½ percentage point of GDP. This debt burden implies only a minimal drag on spending, nor would it wave a red flag to investors concerned about the creditworthiness of the U.S. economy. Fourth and finally, even if investors took more signal from the NIIP/GDP ratio than from the net investment income balance, it would take quite a few years before the net debt rose above 60 percent of GDP.
Comparison with Other Projections
Compared with some previous exercises in projecting the U.S. balance of payments, our baseline projection pushes considerably farther into the future the date at which the U.S. external debt becomes a concern. For example, writing nearly a decade ago, Mann (1999) projected that with an unchanged exchange rate and standard growth assumptions, the U.S. current account deficit would reach 8 percent of GDP by 2010 and the NIIP/GDP ratio would reach −64 percent of GDP. With updated assumptions, Mann (2004) projected the current account deficit would reach roughly 13 percent of GDP by 2010. Cline (2005) projected that by 2010, the current account deficit would reach 7¼ percent of GDP and the NIIP/GDP ratio would reach 50 percent. Comparing model simulations is difficult, but several factors likely contribute to the more benign outlook in our projections compared with Mann (1999 and 2004) and Cline (2005): the dollar has fallen further from the levels assumed in their projections; a combination of valuation changes and data revisions have boosted the starting point for our projections of net investment income; and valuation changes and data revisions have boosted the starting point for our projections of the NIIP.
Two other projections might be mentioned, although they are more difficult to compare with ours. Higgins, Klitgaard, and Tille (2005) construct a scenario in which the NIIP/GDP reaches −65 percent of GDP in 2015 and −89 percent of GDP by 2025; this represents faster deterioration than in our projections, but it is based on the assumption that the current account deficit is fixed at 6 percent of GDP, a higher deficit than we project for most of the projection period. By contrast, Kitchen (2007) develops a projection in which the NIIP/GDP ratio reaches only −39 percent of GDP by 2015—compared with about −50 percent of GDP in our baseline—but this projection assumes dollar depreciation of over 1 percent annually. Notably, Kitchen’s analysis, like ours, assumes a persistent rate of return differential favoring U.S. direct investment abroad, and thus he also projects a very small deficit on net investment income, notwithstanding a still-substantial net external debt.
We believe the baseline projection described above to be plausible, but certainly the confidence interval around future projections must be very large indeed. Accordingly, in this section we present several alternative projections to illustrate the range of uncertainty, shown in Figure 3.
Figure 3.Alternative Projections of the USIT Model
Sources: U.S. Bureau of Economic Analysis; Federal Reserve Board of Governors; and national authorities.
Note: Projection period is shaded area.
In the first alternative projection, the rate of foreign GDP growth is increased by about ½ percentage point, so that it grows about 4 percent annually. As shown by the dot-dot-dashed lines, the trade and current account deficits flatten out and then start narrowing; the net investment income balance is much improved, as higher foreign growth leads to more high-earning U.S. direct investment abroad; and the NIIP/GDP ratio deteriorates more slowly. Hence, with this relatively small alternation of assumptions, the present configuration of asset prices, growth rates, and exchange rates would likely be sustainable in the long term.
In the second alternative projection, the rate of foreign GDP growth is lowered by about ½ percentage point and U.S. GDP growth is boosted about ½ percentage point, so that they both grow at about 3 percent annually. As denoted by the dashed lines, under this scenario, the current account deficit widens to more than 8 percent of GDP by 2020, the NIIP/GDP ratio deteriorates beyond −70 percent, while the balance on net investment income now declines to about −1 percent of GDP. These outcomes are less sustainable than those in the baseline projection, but nevertheless, the debt-service ratio remains benign.
In the third alternative projection, shown by the dotted lines, oil prices rise at 5 percent annually. In this scenario, the trade deficit widens to nearly 8 percent and the NIIP/GDP ratio also deteriorates beyond 70 percent of GDP by 2020. But surprisingly, the balance on net investment income is nearly zero, much better than in the baseline. Higher oil prices boost the profits of U.S.-based oil companies, leading to higher receipts on direct investment abroad. All told, therefore, the higher oil prices have mixed implications for sustainability.
Finally, we consider a scenario—the dot dashed lines—in which the dollar declines by 1 percent annually going forward and, as investors demand a higher return to compensate, U.S. interest rates rise by 1 percentage point as well. In this scenario, which might be regarded as quite gradual correction, the trade deficit is considerably reduced relative to baseline. This positive effect on external balance is partly offset by the higher payments required on U.S. portfolio liabilities, so that the net investment income deficit is larger than in the baseline. However, on balance the current account deficit is narrower than in the baseline projection and the NIIP/GDP ratio slightly less negative as well.
To sum up, the risks to our projection are both on the upside and the downside, and we believe the baseline represents a plausible modal scenario.
Ex Post Historical Simulation
Obviously, beyond uncertainty about the exogenous variables in our simulations, another source of error in our projections is parameter uncertainty. To at least partially address this concern, Figure 4 presents the results of a simulation of the model starting in the fourth quarter of 1994 and ending in the fourth quarter of 2007. The exogenous variables are set to their actual historical values, while the endogenous variables are simulated dynamically.
Figure 4.Ex Post Historical Simulation of the USIT Model
The model does a surprisingly good job of tracking movements in the trade and current account balances. The predicted ratio of the NIIP to GDP generally follows the actual path until 2003 or so, after which the predicted path continues to deteriorate while the actual NIIP/GDP ratio becomes less negative. However, much of the rise (to less negative values) of the actual NIIP/GDP ratio reflected revisions to the data which uncovered more U.S. assets abroad, and this could not be anticipated by the model. The predicted path of net investment income also follows the broad contours of the historical data except during 2001–04, when mispredictions of rates of return on investment income (not shown) and the noted revisions to the NIIP cause predicted net investment income to undershoot actual values.
All told, the ex post historical simulation provides some comfort that our model can give us useful insights into the outlook for the U.S. external balance.
Sensitivity to the Key Export Elasticity
The incorporation by our model of the Houthakker-Magee asymmetry in income elasticities leads to forecasts of larger current account deficits and net debt than if we assumed the income elasticities for imports and exports were equal. Although the USIT model’s trade equations have tracked history well, recent work by Thomas, Marquez, and Fahle (2008) suggests that mismeasurement of foreign prices may have introduced a downward bias to the estimated income elasticity for exports. If this is the case, then the Houthakker-Magee asymmetry embodied in the USIT model is unduly pessimistic for external adjustment.
The Thomas, Marquez, and Fahle measure of foreign prices is based on a weighted average relative price (WARP) which combines deviations from purchasing power parity (PPP) price levels with moving trade shares. Unlike conventional measures, the WARP treats a shift in trade shares from high-price countries to low-price countries as an effective decrease in foreign prices, even if all price levels remain unchanged. Since the trade shares of China and other low-price producers have increased markedly over the past 25 years, the WARP-based measure of foreign prices has risen less than standard measures, and thus the WARP-based measure of the relative price term for an export equation (price of U.S. exports/foreign prices in dollars) has increased relative to conventional measures.
We reestimated the USIT equation for goods exports (excluding high tech) using the WARP-based measure of foreign prices and found the income elasticity to be higher than in the standard formulation, while the estimated price elasticity was little changed.7 The WARP formulation boosts the overall income elasticity for total goods and services exports reported in Section II from 1.25 to 1.43, diminishing but not eliminating the Houthakker-Magee asymmetry.
Results from a simulation incorporating the higher export elasticity, with other assumptions at their baseline levels, are reported in Figure 5. Higher export growth produces a slight narrowing in the trade and current account balances, as a share of GDP, over the forecast horizon. However, the NIIP and net investment income still deteriorate as a share of GDP, albeit more slowly than in the baseline.
Figure 5.Sensitivity to Key Export Elasticity of the USIT Model
Note: Projection period is shaded area.
How Negative a NIIP Is Worrisome?
Our baseline projection of growing net indebtedness suggests that the U.S. current account is not sustainable in the long run. However, theory provides no guidance as to how large the negative NIIP/GDP ratio, or the ratio of net investment income payments to GDP, could become before triggering an adjustment that narrows the current account deficit. Accordingly, while the NIIP slated to reach roughly −60 percent of GDP by 2020 in the baseline scenario, we do not know whether, in reality, the current account deficit would be forced to adjust much before or much after it reached that point.
To shed some light on this issue, we look at the experience of other industrial countries. Figure 6a presents NIIP/GDP ratios for 19 industrial economies in 2006, the latest year for which these data are available for a broad array of countries. The exhibit makes clear that the current level of the NIIP/GDP ratio for the United States, about 20 percent, is well within normal bounds. More importantly, about five countries had negative NIIP/GDP ratios in the neighborhood of 60 percent or higher: New Zealand, Greece, Portugal, Australia, and Spain.
Figure 6.Net International Investment Positions and Net Investment Income Positions
Sources: (a) (c) (e) and (f): IMF, International Financial Statistics; (b) and (d): IMF, Balance of Payments Statistics.
Note: Data for Sweden in Panels (a) and (c) are from 2005. For Panels (e) and (f), adjustment years are drawn from Croke, Kamin, and Leduc (2006).
Figure 6b presents data on the ratio of net investment income to GDP for a similar group of industrial economies. A couple of countries—Ireland and New Zealand—have negative net investment income balances of 5 percent of GDP or more, while several more have balances in the −1½ to −2½ percent range. Under the baseline scenario, the U.S. net investment income balance does not reach –½ percent of GDP until 2019.
In considering sustainability, the NIIP and net investment income are compared with GDP because GDP is a measure of a country’s ability to service its debt. Panels (c) and (d) of Figure 6 present ratios of the NIIP and net investment income to an alternative measure of debt-repayment capacity, exports of goods and services. By these measures, the United States moves up a few rungs on the ladder of highly indebted countries, but remains below the five countries (listed above) with negative NIIP/GDP ratios exceeding 60 percent. According to the baseline projection described above, the U.S. NIIP/exports ratio expands to −440 percent by 2020, a little larger than the 2006 debt/export ratios for the most highly indebted countries shown in Panel (c) of Figure 6. However, in the baseline projection, the U.S. negative net investment income balance as a share of exports remains smaller than 5 percent, which is considerably below many of the ratios shown in Figure 6, Panel (d). Moreover, given that a country can reduce imports or expand exports as needed to repay external debt, we feel the NIIP/GDP and net investment income/GDP ratios shown in Panels (a) and (b) of Figures 6 are probably better proxies for debt-repayment capacity.
A more relevant set of comparisons may involve international investment positions of economies at the time that they begin to experience current account adjustment. Panel (e) of Figure 6 presents the NIIP/GDP ratio at the onset of current account adjustment; the adjustment years are the same ones identified in research by Croke, Kamin, and Leduc (2006), and data are available for 15 of the 23 episodes identified. There is a very wide range of NIIP/GDP ratios associated with current account adjustment, suggesting either that many adjustments were triggered well before NIIPs reached some notional limit, or that this limit varies across countries. Panel (f) of Figure 6 indicates a similarly wide dispersion of data on the ratio of net investment income to GDP during current account adjustment episodes. Notably, however, in 16 of the 22 episodes, the negative net investment income ratio exceeded 1 percent, a level that, in our baseline projection, the United States does not reach in the next 12 years.
Is the United States Special?
The above analysis suggests that, even as far away as 2020, the U.S. net external debt will still be no larger than it is today for five industrial economies, and the servicing burden on that debt will be relatively minor. But how relevant is the experience of other industrial economies for the United States?
Some arguments suggest that measures of external indebtedness could become even greater for the United States than for other industrial economies before adjustment was needed. First, an increasingly prevalent view holds that because the United States offers especially deep and liquid financial markets, global investors find U.S. assets unusually attractive and would be especially willing to finance the large deficit for a long period and on relatively cheap terms (Cooper, 2005; Hubbard, 2005).8
Second, among the different categories of U.S. debtors, a key international borrower is, of course, the federal government. At the end of 2007, foreign holdings of U.S. treasuries and agency debt amount to about $3.8 trillion, accounting for 22 percent of total U.S. foreign liabilities. Given the U.S. government’s commitment to the quality of its debt and its access to effective taxation, this likely makes U.S. external debt, overall, more creditworthy than if it had been issued primarily by U.S. households.
Third, as is often remarked, because the United States’ foreign-currency-denominated assets well exceed its foreign-currency-denominated liabilities, a decline in the dollar tends to reduce the net debt. This makes U.S. debt-servicing less vulnerable to dollar depreciation, and hence probably raises the level of sustainable debt.
A number of other considerations, however, suggest that the United States may have a diminished scope to issue external debt compared with other countries. Importantly, some of the highly indebted economies shown in Figure 6 may themselves represent special cases: Given their abundant resources, it may be sensible for Australia and New Zealand to import substantial capital and repay slowly over a long time horizon. Similarly, the large deficits and debts of Portugal, Spain, and Greece may be an artifact of their integration into the European Union and the euro area.
Moreover, put simply, the United States is the largest economy in the world, and heavy issuance of liabilities could saturate global demand. This consideration is addressed below.
III. Measures of Investor Exposure to U.S. Assets
The NIIP/GDP ratio is primarily a signal, albeit a highly imperfect one, of the weight of an economy’s debt service obligations; when the NIIP/GDP ratio reaches a certain size, investors may decide to limit their acquisition of the economy’s assets, fearing that larger NIIPs may not be serviceable. In addition to concerns about creditworthiness, however, investors may also seek to limit their exposure to an economy because that exposure threatens to breach a certain share of their portfolio.9 The NIIP/GDP ratio is not a very useful measure of this type of exposure, that is, of the weight of U.S. assets in foreign portfolios. In this section, we address several more direct measures of the exposure of investors to U.S. assets.
Recent Measures of Foreigners’ Exposure to U.S. Assets
Panel (a) of Figure 7 plots foreign holdings of all types of U.S. securities. Consistent with the increase in the U.S. net indebtedness position, holdings of all of these securities have risen in recent years. As panel (b) of Figure 7 illustrates, foreign holdings account for a noticeably growing share of the amounts outstanding of Treasury and agency securities, but there has been a less noticeable increase in the foreign share of U.S. corporate bonds and equities, as the total stocks of these securities have also increased rapidly.
Figure 7.Foreign Portfolio Holdings of U.S. Securities and Shares of U.S. Securities in Global Market Capitalization
Sources: (a) Authors’ estimates based on Treasury International Capital data; (b) authors’ estimates based on Treasury International Capital data and Flow of Funds accounts; (c) S&P, Global Markets Factbook 2007; (d) Bank for International Settlements data, adjusted for Brady bonds outstanding; (e)–(g): authors’ estimates based on IMF, Coordinated Portfolio Asset Surveys.
A more relevant point for understanding how foreign exposure to U.S. securities may have changed is to consider how rapidly global market capitalization has grown, and how large foreign holdings of U.S. securities are compared with other securities in their portfolios. Figures 7c and d plot the share of U.S. equities and bonds in the global market capitalization of those instruments. These measures gauge whether persistent U.S. current account deficits have been associated with more rapid issuance of equity and bond liabilities than is occurring in other economies. In fact, the data show little net rise since the late 1990s in the U.S. share of global market capitalization for these instruments, likely reflecting two factors. First, financial deepening is progressing rapidly abroad, and increased securities issuance has led to ratios of market capitalization to GDP abroad that approach those of the United States.10 Second, declines in stock prices, on balance, since 2000 and declines in the dollar since 2002 have also kept exposures to U.S. assets in terms of the shares held in check.
Panels (e) and (f) of Figure 7 address the question of how large the share of U.S. assets is in foreign portfolios, and compares, for the limited number of countries and years for which survey data are available, the change over time in (1) the aggregate share of holdings of U.S. equities and bonds in overall holdings by foreigners of these instruments (the dark shaded bars), and (2) the aggregate share in overall holdings by foreigners of these instruments of claims on residents outside their own countries (the light shaded bars).11 Note that overall holdings include not only a country’s cross-border holdings of equities or bonds, but also its holdings of its own domestic equities or bonds. A number of observations can be made. First, the share of U.S. equities and bonds in the aggregate portfolios of foreigners has not risen much since 1997, and most or all of that increase took place between 1997 and 2001.12 Second, compared with the share of U.S. assets in portfolios abroad, the share of all external (to them) assets in portfolios abroad has risen by as much or more. This relationship may be seen more easily in Panel (g) of Figure 7, which indicates the share of U.S. equities and bonds in the external portfolios of foreigners. (In these calculations, holdings of a given country’s domestic securities are subtracted from its overall holdings to arrive at its external holdings.) The shares of both U.S. equities and bonds in these portfolios rose between 1997 and 2001, but have declined or been little changed since 2001.
Panels (a) and (b) of Figure 8 assess the extent to which foreigners are underweight in U.S. assets and compares that to the extent that they are underweight in external assets more generally.13 Foreigners are considered to be appropriately weighted (in terms of a standard portfolio allocation model) in U.S. equities, for example, if the share of U.S. equities in their total equity portfolio—the dark shaded bars in Figure 7, Panel (e)—is equal to the share of U.S. equities outstanding in global equity capitalization—as shown in Figure 7, Panel (c). We thus compute foreigners’ relative portfolio weights in U.S. assets (plotted against the vertical axis) by dividing the share of U.S. securities in the total portfolio of foreigners by the size of the U.S. market relative to the world market:
Figure 8.Relative Foreign Portfolio Weights in U.S. Securities: 1997–2006, and Model Projections of Foreign Portfolio Shares and Weights
Sources: (a)–(b): authors’ estimates based on IMF, Coordinated Portfolio Asset Surveys; (c)–(f): authors’ calculations based on USIT model projections.
Note: For Panels (a) and (b), relative portfolio weights in U.S. securities are calculated as holdings of U.S. securities relative to holdings of all securities, divided by the size of U.S. market capitalization relative to global market capitalization. Relative portfolio weights in all external securities are calculated in a similar manner.
A relative portfolio weight of 1 implies appropriate weighting of U.S. assets in foreigners’ portfolios, while a relative weight less than 1 implies an underweighting of U.S. assets.
A similar calculation is undertaken for the foreigners’ relative portfolio weight in all external securities, where for any given country outside the United States, external refers to all countries external to that country (including the United States).14 A value of this calculation (plotted against the horizontal axis) less than 1 implies that foreigners are underweight in assets outside their own country and overweight in domestic securities—that is, they exhibit home bias.
Observations on the dashed 45 degree line in Panels (a) and (b) of Figure 8 would indicate that foreigners are equally underweight U.S. and other external assets. The data indicate that foreign investors are both underweight in U.S. assets and in external assets more generally—that is, they exhibit home bias. In 1997, the bias against U.S. assets appeared to be considerably greater than the bias against other countries’ assets. Since then, foreigners appear to have reduced their bias against U.S. and other countries’ equities about equally. The same appears true for foreign holdings of U.S. and other external bonds between 2001 and 2006; between 1997 and 2001, they appear to have reduced their bias against U.S. bonds by somewhat more. Although foreigners’ home bias has declined over this period, foreigners remain more underweight in U.S. securities than they do in external securities in general.
The bottom line from Figure 7 and Panels (a) and (b) of Figure 8 is that, in spite of the expansion of U.S. external liabilities since the mid-1990s, neither the shares nor the relative weights of U.S. assets in foreigners’ portfolios have increased to any meaningful extent. This somewhat surprising result is, in part, a reflection of the decline in the dollar since 2002, which has reduced the value of holdings of dollar-denominated instruments relative to those denominated in other currencies. However, this result also reflects the increases of asset holdings abroad more generally, of which increased holdings of U.S. assets are just a part.15 All told, there is no evidence that an overhang of excessive foreign exposure to U.S. assets is developing which would require further adjustments in the U.S. current account balance or in U.S. asset prices to correct.
Prospective Future Movements in Foreigners’ Exposure to U.S. Assets
Even if the exposure of foreigners to U.S. assets, relative to a number of benchmarks, does not appear to have increased much over the past decade, it is possible that the financing of continued current account deficits would push this exposure to more worrisome levels in the future. With the projected increase in the NIIP in our baseline projection described above, foreign investors will of necessity acquire many additional U.S. assets. How large a share of their portfolio foreign investors are willing to acquire is an open question. It depends importantly on factors beyond the evolution of the U.S. NIIP, including the growth of securities issuance in foreign economies and thus the share of U.S. securities in global market capitalization.
To devise an estimate of the potential magnitude of the effects of the projected increase in the NIIP on foreign portfolios, we perform the following exercise: We assume that both U.S. and foreign total market capitalization (reflecting equities and bonds combined) grow at their respective rates of nominal GDP, thus keeping their market cap/GDP ratios constant. As described in Section II, the USIT model simulations project gross portfolio flows into and out of the United States so that the resulting net financing is sufficient to meet the balance of payments requirements while also keeping the growth in the gross positions as close as possible to their recent historical rates.16 Thus, the total portfolio of foreign investors is projected to grow along with foreign market cap (minus the amount acquired by U.S. investors), plus estimated increased holdings of U.S. securities. Finally, we perform an alternative calculation in which the projections for financial flows into and out of the United States remain the same, but both U.S. and foreign market cap grow at the same rate (equal to the average of foreign and U.S. growth), so that U.S. market cap stays constant as a share of global market cap.
As illustrated in Panels (c)–(f) of Figure 8, the results of this projection exercise generally point to increases in the share and relative weight of U.S. assets in foreigners’ portfolios, but it is not clear those increases would be worrisome. Figure 8c shows projections for total U.S. market cap as a share of global market cap. (This panel is comparable to the share concept in Panels (c) and (d) of Figure 7, but combines equity and bond capitalization.) Under the baseline assumption, total U.S. market cap declines to about 32 percent of global market cap by 2020, reflecting the slower projected growth of U.S. nominal GDP and thus of U.S. market cap, compared with foreign GDP and market cap. In the alternative simulation, the U.S. share of market cap stays constant, by design.
Figure 8, Panel (d) shows that if foreigners acquire U.S. assets sufficient to accommodate the rise in the NIIP, the share of U.S. market cap held by foreigners rises from around 20 percent currently to near 40 percent in either simulation.
Figure 8e shows the evolution of U.S. securities as a share of the total portfolio of foreigners (comparable to the dark shaded bars in Panels (e) and (f) of Figure 7) and Figure 8, Panel (f) shows the corresponding evolution of the relative portfolio weights in U.S. securities (comparable to the relative portfolio weights in U.S. securities shown in Figure 8, Panels (a) and (b). Under either simulation, the share increases to around 20 percent by 2020, and the corresponding estimated relative weight of U.S. securities in foreigners’ portfolios increases to 55 to 60 percent.17 This increase in the relative weight is substantial, but even so, a relative weight of 0.55 implies that foreigners would remain underweight in U.S. assets. Moreover, although we cannot perform a similar projection of the relative weight of external securities in foreigners’ portfolios, presumably that weight would also be rising substantially, assuming the trends documented in Figure 8, Panels (a) and (b) continue.18
IV. How Responsive are Asset Prices to International Balance Sheet Indicators?
To make an assessment of how imminent is a correction in the current account balance, one must not only project the likely evolution of the external debt burden and the time at which it breaches some threshold of sustainability. One must also be able to predict, at such time as this breach occurs, how rapidly investors will pull back from a country’s assets, thus boosting interest rates, pushing down the currency, and inducing current account adjustment. As an initial rough cut at addressing this issue, we evaluated the sensitivity of interest rates and exchange rates to measures of international balance sheet positions.
Long-Term Interest Rates
In our analysis, we estimate panel regressions using annual data for 22 industrial countries over the period 1975 to 2006. The dependent variable is the nominal long-term (usually 10-year) yield on government benchmark bonds. Following Gruber and Kamin (2008) and Warnock and Warnock (2006), a set of control variables includes the overnight money market interest rate, the four-quarter rate of CPI inflation, the four-quarter rate of real GDP growth, the standard deviation of the quarterly change in the long-term nominal interest rates over the preceding 12 quarters, the ratio of the structural (full-employment) fiscal balance to GDP, and two annual lags of the dependent variable.
The first column of Table 1 presents the results of this equation, estimated including only the control variables. The results are, for the most part, consistent with expectations. Increases in money market interest rates, inflation, and real GDP growth all boost nominal long-term bond yields by a statistically significant extent. Also as one would expect, increases in the volatility of interest rates boost yields and increases in the fiscal balance reduce yields, although these effects are not statistically significant.
|Dependent Variable: 10-Year Nominal Government Bond Yields|
|10-year interest rate (–1)||0.426||0.578||0.587||0.564||0.559||0.596||0.395||0.596||0.371||0.573|
|10-year interest rate (–2)||0.157||0.064||0.056||0.057||0.060||0.069||0.176||0.071||0.167||0.060|
|Money market interest rate||0.246||0.198||0.194||0.204||0.211||0.195||0.238||0.170||0.265||0.206|
|Real GDP growth||0.077||0.060||0.064||0.071||0.066||0.049||0.089||0.045||0.110||0.057|
|Interest rate volatility||0.093||–0.486||–0.422||–0.392||–0.484||–0.483||0.166||–0.393||0.175||–0.436|
|External liability variable (–1)||0.020||–0.004||–0.001||0.006||0.013|
|Time fixed effects||No||Yes||Yes||Yes||Yes||Yes||No||Yes||No||Yes|
|Country fixed effects||No||Yes||Yes||Yes||Yes||Yes||No||No||Yes||Yes|
|No. of observations||541||541||532||518||511||474||438||438||438||438|
The next column of the table adds year and country fixed effects, as well as a dummy variable that becomes one in 1999, with the creation of the euro area. The coefficients on the control variables are, for the most part, little changed.
The next four columns add, separately, to this equation four different measures of external balance, all lagged one year: the NIIP/GDP ratio, the ratio of net investment income (NIINCOME) to GDP, the current account (CAB)/GDP ratio, and the share of a countries’ gross external liabilities in the gross external assets of foreigners (external liability). The remaining columns present results of equations that include all four of these external balance measures together, but with different combinations of year and country fixed effects.
By and large, there is little evidence that the external balance measures examined here are associated with a significant effect on long-term yields. To the extent that any of these variables has a consistent and nearly significant effect of the expected sign, it is the NIIP/GDP ratio. At its largest, the coefficient on this term is −0.005, implying that a 100 percent of GDP negative NIIP would be associated with an increase in the long-term nominal interest rate of 50 basis points—this is a discernable, but not especially large, increment. The current U.S. NIIP of about −20 percent of GDP implies a boost to the interest rate of a fifth of that, of only 10 basis points. (This is a much smaller effect than estimated by Lane and Milesi-Ferretti, 2001, using a more limited set of control variables.)
In interpreting the coefficients on the external balance variables, a prominent identification problem should be acknowledged. In principle, countries with highly developed financial systems and strong investor protections should attract foreign investors. These large capital inflows, in turn, ought to lower U.S. long-term yields. Accordingly, a priori, it is not clear whether large external debts should be associated with higher interest rates, because they raise concerns among investors, or lower interest rates, because the large debts reflect strong investor interest.
However, the estimates shown in columns 2–6 and 9–10 should, to a large extent, control for the simultaneity problem described above, as they include country fixed effects which should capture any special attractiveness of a country’s assets. In fact, in columns 7 and 8, where no country fixed effects are included, the coefficient on the NIIP/GDP ratio is much smaller than in columns 9 and 10, where country fixed effects have been included. This suggests that the country fixed effects are, indeed, helping to control for the simultaneity problem.
If investors react to high debt levels by pulling back from a country’s assets, we should also observe a depreciation of the exchange rate. To assess whether this is the case, we re-estimated the panel equations shown in Table 1, but substituted the real exchange rate in place of the nominal long-term interest rate. We use the CPI-deflated multilateral exchange rate published in the IMF’s International Financial Statistics.19 (An increase indicates appreciation.)
In Table 2, the dependent variable is the percent deviation of the real exchange rate from its country-specific sample mean. The coefficient on the money market interest rate is positive and significant, while that on the inflation rate is about the same magnitude but negative and significant; together, these results confirm our expectation that increases in the real interest rate should positively affect the real exchange rate. None of the other control variables have a significant coefficient of the expected sign—this is perhaps not surprising, as exchange rate models are notoriously hard to estimate. The measured effects of the external balance variables are a mixed bag as well. The NIIP and net investment income generally have a positive effect on the real exchange rate as expected, but the effect is inconsistent in terms of sign and significance. Higher current account balances lead to lower exchange rates and larger external liabilities lead to higher exchange rates, the opposite of what we’d expect.
|Dependent Variable: Real Effective Exchange Rate as Percent Deviation from Sample Country Mean (An increase indicates appreciation)|
|REER percent deviation (–1)||1.066||1.053||1.066||1.078||1.043||1.046||1.087||1.097||1.021||1.026|
|REER percent deviation (–2)||–0.333||–0.321||–0.331||–0.330||–0.313||–0.317||–0.334||–0.327||–0.326||–0.311|
|Money market interest rate||0.178||0.365||0.380||0.363||0.371||0.421||0.198||0.370||0.254||0.417|
|Real GDP growth||–0.173||–0.332||–0.344||–0.332||–0.337||–0.274||–0.111||–0.157||–0.167||–0.234|
|Interest rate volatility||–0.458||–0.937||–0.961||–0.557||–0.871||–1.314||–0.339||–0.244||–0.329||–0.692|
|External liability variable (–1)||0.271||0.001||0.012||0.509||0.486|
|Time fixed effects||No||Yes||Yes||Yes||Yes||Yes||No||Yes||No||Yes|
|Country fixed effects||No||Yes||Yes||Yes||Yes||Yes||No||No||Yes||Yes|
|No. of observations||497||497||488||480||473||434||404||404||404||404|
Finally, Table 3 reestimates these equations, with the dependent variable specified as the percent change in the real exchange rate from the previous year. Now, the current account balance appears to be positively associated with the real exchange rate, but the coefficients on the other variables remain insignificant or of the wrong sign.20
|Dependent Variable: Real Effective Exchange Rate Percent Change (An increase indicates appreciation)|
|REER percent change (–1)||0.185||0.185||0.205||0.207||0.165||0.206||0.201||0.214||0.180||0.179|
|REER percent change (–2)||–0.112||–0.131||–0.137||–0.138||–0.126||–0.162||–0.142||–0.150||–0.158||–0.172|
|Money market interest rate||0.162||0.356||0.387||0.356||0.443||0.398||0.206||0.345||0.285||0.476|
|Real GDP growth||0.024||–0.084||–0.069||–0.052||–0.151||0.022||0.063||0.051||0.038||0.004|
|Interest rate volatility||–1.201||–0.972||–0.995||–0.380||–1.416||–0.851||–0.795||–0.221||–1.164||–0.734|
|External liability variable (–1)||–0.065||0.012||0.023||0.178||0.186|
|Time fixed effects||No||Yes||Yes||Yes||Yes||Yes||No||Yes||No||Yes|
|Country fixed effects||No||Yes||Yes||Yes||Yes||Yes||No||No||Yes||Yes|
|No. of observations||488||488||479||471||464||427||397||397||397||397|
In this section, we found that greater external debt was associated with only small increases in interest rates and mixed effects on exchange rates. Accordingly, deteriorations of external balance positions do not seem to have been associated with significant pullbacks by global investors.21 Of course, modeling financial market prices is notoriously difficult, so it may not be surprising that we found little evidence of strong linkages among indebtedness, interest rates, and exchange rates. By the same token, however, the direst predictions that continued large U.S. current account deficits will lead to financial crisis remain unsubstantiated.
This paper has addressed three questions about the prospects for the large U.S. current account deficit. Is it sustainable in the long term? If not, how long will it take for measures of external debt and debt service to reach levels that could prompt some pullback by global investors? And if and when such levels are breached, how readily would asset prices respond and the current account start to narrow?
To address these questions, we started with projections of a detailed partial-equilibrium model of the U.S. balance of payments. Assuming plausible settings of macroeconomic indicators in the United States and abroad, as well as a flat real dollar, our projections indicate that the current account deficit will resume widening and the negative NIIP/GDP ratio will continue to expand. This suggests that the configuration of macroeconomic settings underlying the current account balance at present is not sustainable in the long term. Nevertheless, compared with earlier years when the dollar was much higher, the current account balance is likely somewhat less unsustainable at present.
Moreover, compared with other industrial economies, current levels of the U.S. debt and debt service are relatively modest, and foreign exposure to U.S. assets has been moving down rather than up. Our projections suggest that even by the year 2020, the negative NIIP/GDP ratio will be no higher than it is in five industrial economies today, and U.S. net investment income payments will remain below 1 percent of GDP. Absent changes in the dollar and consequent valuation adjustments, the share of U.S. claims in foreigners’ portfolios will likely rise over the next decade, but not to an obviously worrisome extent. All told, it seems likely it would take more than a decade for U.S. indebtedness to reach any limits of global investors’ willingness to extend financing.
Finally, we explored the historical responsiveness of asset prices and the current account in a wide range of industrial economies to increases in different measures of net indebtedness and external imbalances. We discerned only a very small effect of external indebtedness on interest rates, and no clear and concerted effects on exchange rates. Accordingly, it is not clear that, even were the U.S. net debt to approach limits to its sustainability, the developments needed to correct the current account—changes in growth rates, assets prices, exchange rates, and the like—would materialize all that rapidly.
We would emphasize that these findings do not imply that U.S. current account adjustment is necessarily many years away. Many factors could trigger such adjustment, including, inter alia, a surge in foreign growth, declines in U.S. growth, or intensified concerns about U.S. current account sustainability. In fact, judging by the developments of the last several years, we may already be in the middle of an adjustment episode. Our point is rather that international balance sheet considerations likely are not sufficient, by themselves, to require external adjustment any time soon.
Balakrishnan, Ravi, TaminBayoumi, and VolodymyrTulin, 2007, “Globalization, Gluts, Innovation, or Irrationality: What Explains the Easy Financing of the U.S. Current Account Deficit,” IMF Working Paper 07/160 (Washington, International Monetary Fund).
Bertaut, Carol, and WilliamGriever, 2004, “Recent Developments in Cross-Border Investment in Securities,” Board of Governors of the Federal Reserve System, Federal Reserve Bulletin (Winter).
Bertaut, Carol, and RalphTryon, 2006, “Understanding U.S. Cross-Border Securities Data,” Board of Governors of the Federal Reserve System, Federal Reserve Bulletin (May).
Bertaut, Carol, StevenKamin, and CharlesThomas, 2008, “How Long Can the Unsustainable U.S. Current Account Deficit Be Sustained?,” Board of Governors of the Federal Reserve System, International Finance Discussion Paper No. 935.
Chinn, Menzie David., and Hiro.Ito, 2007, “Current Account Balances, Financial Development, and Institutions: Assaying the World ‘Savings Glut’,” Journal of International Money and Finance, Vol. 26 (April), pp. 546–69.
Cline, William, 2005, The United States as a Debtor Nation (Washington, Institute for International Economics).
Cooper, Richard, 2005, “Living with Global Imbalances: A Contrarian View,” Institute for International Economics Policy Brief PB05-3 (November).
Croke, Hilary, StevenKamin, and SylvainLeduc, 2006, “An Assessment of the Disorderly Adjustment Hypothesis for Industrial Economies,” International Finance, Vol. 9, No. 1, pp. 37–61.
Curcuru, Stephanie, TomasDvorak, and FrancisWarnock, 2008, “Cross-Border Returns Differentials,” Quarterly Journal of Economics, Vol. 123, No. 4, pp. 495–530.
Erceg, Christopher, LucaGuerrieri, and ChristopherGust, 2006, “SIGMA: A New Open Economy Model for Policy Analysis,” International Journal of Central Banking, Vol. 2 (March), pp. 1–50.
Gagnon, Joseph., 1996, “Net Foreign Assets and Equilibrium Exchange Rates: Panel Evidence,” Board of Governors of the Federal Reserve System, International Finance Discussion Paper No. 574.
Gruber, Joseph, and StevenKamin, 2008, “Do Differences in Financial Development Explain the Global Pattern of Current Account Imbalances?” Board of Governors of the Federal Reserve System, International Finance Discussion Paper No. 923.
Higgins, Mathew, and ThomasKlitgaard, 2007, “Financial Globalization and the U.S. Current Account Deficit,” Federal Reserve Bank of New York, Current Issues in Economics and Finance (December).
Higgins, Mathew, and CedricTille, 2005, “The Income Implications of Rising International Liabilities,” Federal Reserve Bank of New York, Current Issues in Economics and Finance (December).
Hubbard, Robert Glenn, 2005, “A Paradox of Interest,” Wall Street Journal,June23.
Kitchen, John, 2007, “Sharecroppers or Shrewd Capitalists? Projections of the US Current Account, International Income Flows, and Net International Debt,” Review of International Economics, Vol. 15, No. 5, pp. 1036–61.
Lane, Philip Richard, and GianMaria Milesi-Ferretti, 2001, “Long-Term Capital Movements,” IMF Working Paper 01/107 (Washington, International Monetary Fund).
Lane, Philip Richard, and GianMaria Milesi-Ferretti, 2002, “External Wealth, the Trade Balance, and the Real Exchange Rate,” European Economic Review, Vol. 46 (6), pp. 1049–71.
Mann, Catherine, 1999, Is the U.S. Trade Deficit Sustainable? (Washington, Institute for International Economics).
Mann, Catherine, 2002, “Perspectives on the U.S. Current Account Deficit and Sustainability,” Journal of Economic Perspectives, Vol. 16, No. 3, pp. 131–52.
Mann, Catherine, 2003, “How Long the Strong Dollar?” in Dollar Overvaluation and the World Economy, ed. byC. FredBergsten and JohnWilliamson (Washington, Institute for International Economics).
Mann, Catherine, 2004, “Managing Exchange Rates: Achievement of Global Re-balancing or Evidence of Global Co-dependency,” Business Economics, Vol. 39, Issue, pp. 20. (July).
Mussa, Michael, 2004, “Exchange Rate Adjustments Needed to Reduce Global Payments Imbalances,” in Dollar Adjustment: How Far? Against Whom? ed. byC. FredBergsten and JohnWilliamson (Washington, Institute for International Economics).
Organization for Economic Cooperation and Development (OECD), 2008, OECD Economic Outlook No. 83, June (Paris, OECD).
Thomas, Charles Phipps, JaimeMarquez, and SeanFahle, 2008, “Measuring U.S. International Relative Prices: A WARP View of the World,” Board of Governors of the Federal Reserve System, International Finance Discussion Paper No. 917.
Warnock, Frank., and VeronicaCacdac Warnock, 2006, “International Capital Flows and U.S. Interest Rates,” NBER Working Paper 12560 (Cambridge, Massachusetts, National Bureau of Economic Research).
The authors are economists in the International Finance Division of the Federal Reserve Board. This paper has benefitted from comments by our editor Akito Matsumoto, an anonymous reviewer, Trevor Reeve, and participants at the Current Account Sustainability in Major Economies (II) conference at the University of Wisconsin, especially discussant Jeffrey Frankel. Jim Albertus, Sean Fahle and Dao Nguyan provided excellent research assistance.
The theoretical basis for the Houthakker-Magee asymmetry remains ambiguous and the subject of controversy among trade modelers. Even so, the estimated coefficients in our trade models continue to exhibit the Houthakker-Magee asymmetry for goods trade. For services trade, in contrast, the income elasticity for exports exceeds that for imports. However, as goods trade exceeds services trade, the income elasticity for total imports exceeds that of total exports.
A number of explanations have been advanced for the asymmetry of rates of return on direct investment, including greater efficiency of U.S. firms, better project selection by U.S. firms, younger and thus less mature investments for foreign firms in the United States, greater competitive pressures in the U.S. market, or differences in tax treatment (see Higgins, Klitgaard, and Tille, 2005). None of these factors seem likely to disappear in the near term.
This is accomplished by starting out with trend extrapolations of gross inflows and outflows. If more financing is needed, gross inflows are adjusted up and outflows are adjusted down symmetrically; the reverse occurs if less financing is needed.
This is the U.S. Bureau of Economic Analysis’s (BEA) preferred measure as it avoids many methodological issues associated with estimating the stock market value of nontraded equity positions. In addition, for our simulations, using the current cost measure means our estimates do not depend on our assumptions about future stock market movements.
The saw-toothed pattern of the trade balance is caused by a residual seasonal pattern (even after seasonal adjustment by the BEA) in oil imports.
In effect, the larger impetus to exports owing to the higher income elasticity is offset by the smaller impetus from prices owing to the less pronounced decline in relative export prices.
Mann (1999, 2002, and 2003) and Cline (2005), among others, also draw a distinction between creditworthiness- and exposure-based criteria for sustainability, and present calculations based on these concepts. Concerns about creditworthiness and exposure are not necessarily unrelated. In a portfolio balance model, investors allocate their wealth to different assets, based on expected returns and uncertainties about those returns. Therefore, the more creditworthy an asset is considered to be, the higher the share in wealth allocated to that asset.
Balakrishnan, Bayoumi, and Tulin (2007) find that declining home bias and financial deepening account for most of the financing of the recent large U.S. current account deficits, rather than increases in the share of U.S. assets in foreign portfolios.
Data on foreign holdings of U.S. and other external securities are derived from the IMF’s Coordinated Portfolio Investment Surveys (CPIS). Because the CPIS captures nonreserve holdings only, we impute an amount for holdings of both U.S. and other external securities held as reserves using data from the IMF SEFER and COFER surveys. Most industrial countries and a number of emerging-market countries now participate in the CPIS. In terms of major holders of U.S. securities, they exclude investments held in some major custodial centers, by most Middle East oil exporters, and by China. Holdings of U.S. securities accounted for by CPIS countries in 2006 represent about 70 percent of total U.S. securities held by foreign investors. See Bertaut, Griever, and Tryon (2006) for a discussion of the methodology for imputing reserve holdings, and for a more complete discussion of the comparability between holdings of U.S. securities as measured by the CPIS and by U.S. liability surveys. Data on holdings of domestic securities are derived from national source financial balance sheet accounts where available, and otherwise from estimates of domestic equity and bond market capitalization.
The 1997 and 2001 figures are not strictly comparable because more countries participated in the 2001 CPIS than in the 1997 CPIS. However, the difference in coverage is less critical for comparing relative shares than absolute holdings, and indeed the increase in shares held between the two years owes largely to the increases registered by countries that were participants in both years.
See Bertaut and Griever (2004), for a fuller elaboration of this approach.
In this analysis, we consider intra-euro area holdings of other euro area country securities as domestic securities.
Higgins and Klitgaard (2007) reach a similar conclusion that financial globalization has had the result that despite increases necessary to finance U.S. current account deficits, there has not been an unusual buildup of U.S. securities in foreign portfolios.
Specifically, we assume that as foreigners acquire additional nondirect investment claims on the United States, the share of these claims that are securities (as opposed to bank deposits, trade credits, and so on) will mirror their current share in nondirect investment claims. Similarly, as U.S. residents acquire additional nondirect investment claims on foreigners, the share of these claims that are in the form of securities will mirror their current share.
The starting figures for 2006—a 12 percent share of U.S. assets in total portfolios corresponding to a U.S. portfolio weight of about 0.28—are slightly larger than the shares and weights in Figure 7 (e and f) and Figure 8 (a and b) because for this exercise, we base total foreign holdings of U.S. securities on the more comprehensive liabilities estimates that underlie the NIIP calculations, and thus we are able to include all foreign holdings of U.S. securities, including those held by countries not participating in the CPIS surveys, notably international financial centers, Middle East oil exporters, and China. Note also that the average shares and weights will more closely resemble the bond shares and weights in the exhibits because the majority of foreign holdings are in the form of U.S. bonds.
We are not able to project how changes in foreigners’ shares and weights held in total external securities compare with the projected changes in shares and weights held in U.S. securities. Although we are able to forecast total foreign (non-U.S.) market cap held by foreigner investors, we have no way of allocating what fraction of that foreign market cap reflects foreigner investors’ home country securities and what fraction reflects holdings of other foreign securities.
These regressions are estimated for the same 22 industrial countries over the sample 1978–2006.
Gagnon (1996) finds evidence that net foreign assets scaled by trade flows are significantly associated with real exchange rates for a panel ending in 1995. In a multi-country probit study of industrial economies, Wright and Gagnon (2006) find that larger current account deficits are significantly associated with sharp real currency depreciations, but the magnitude of the effect is quite small. Other variables do not exert significant, robust effects on the probability of a sharp real depreciation.