This Selected Issues paper and Statistical Appendix focuses on two analytical approaches for judging whether the current account for Australia is sustainable. The paper implements the first approach, by asking how Australia’s net external liability position is likely to evolve over time, based on assumptions of future growth and interest rates. The paper implements the second approach by exploring the implications of a model of optimal external borrowing and lending. The main conclusions are also discussed in the paper.

Abstract

This Selected Issues paper and Statistical Appendix focuses on two analytical approaches for judging whether the current account for Australia is sustainable. The paper implements the first approach, by asking how Australia’s net external liability position is likely to evolve over time, based on assumptions of future growth and interest rates. The paper implements the second approach by exploring the implications of a model of optimal external borrowing and lending. The main conclusions are also discussed in the paper.

I. Australia’s Current Account: A Consumption-Smoothing Approach1

A. Introduction

1. Australia’s current account position has deteriorated since the mid-1980s. In the last 15 years, it has averaged about 4½ percent of GDP, compared with 2–3 percent that was the norm in the 1960s and 1970s. Because of the Asian crisis, Japan’s economic slowdown, and strong domestic demand, the current account position has recently widened significantly above its average level of the past 15 years. Moreover, net external liabilities—at 60 percent of GDP—are already the second highest among the advanced economies and are likely to continue to rise in relation to GDP over the next several years as a result of the recent widening of the current account deficit.

2. The Australian authorities clearly recognize the potential constraints associated with a low national saving rate, a sizable current account deficit, and high external debt. In the 1996/97 budget papers, for example, the authorities stated:

“Increasing dependence on foreign savings, as reflected in growing net foreign liabilities, exposes the economy to sudden shifts in market confidence, leads to higher borrowing costs for Australian business, and makes the economy more vulnerable to external shocks. Inevitably, the effect of these risks is to place an external “speed limit” on the pace at which economic growth can be sustained” (Section I, page 9).

3. As a general matter, the magnitude of a country’s external borrowing does not necessarily carry any normative implications. Australia’s relatively small current account deficits in the 1960s and 1970s, for example, may not have been desirable if they reflected the consequences of limited international capital mobility rather than optimizing behavior. Current account surpluses or deficits simply reflect intertemporal trade, which enables countries to smooth the path of aggregate consumption in the face of temporary shocks. If the public sector is in balance, then the current account is simply the difference between private investment and private saving. With profit opportunities determining investment and optimal consumption behavior determining saving, there is no implication for corrective policy measures.

4. However, a current account deficit may be a problem because not all the conditions of the frictionless intertemporal model exist. There may, for example, be an externality in borrowing, so that at a microeconomic level, agents take on the optimal level of debt, but the country’s total debt is sufficiently large to create a risk premium, which increases borrowing costs for all. This increased borrowing cost will inhibit investment and ultimately result in lower growth. In addition, financial markets may become less willing to finance large current account deficits on a sustained basis. A country with a large current account deficit, which is exposed to volatile export markets or has a small range of exports, will be more vulnerable to this type of reversal in sentiment.

5. There are several analytical approaches to judging whether the current account is sustainable, two of which are used in this paper. The first approach, one of the simplest and most commonly applied, is to examine the conditions under which the ratio of net external liabilities to GDP will remain constant. The disadvantage of this approach is that there are no objective criteria to judge whether the ratio, once it stabilizes, does so at an appropriate level. The second approach is to base judgments about the current account on the empirical estimation of a theoretical model of “optimal” external borrowing, which can yield a benchmark for the optimal current account deficit and level of external liabilities.

6. The remainder of this chapter is organized as follows. Section B implements the first approach, by asking how Australia’s net external liability position is likely to evolve over time, based on assumptions of future growth and interest rates. Section C implements the second approach, by exploring the implications of a model of optimal external borrowing and lending. Section D presents the main conclusions. An appendix provides details of the model and empirical results.

B. Sustainability of External Liabilities

7. A simple indicator of the sustainability of the current account is the stability of external liabilities in relation to GDP, which depends on such variables as the world real interest rate, the growth rate of the economy, and the ratio of the trade balance to GDP.2 If the implicit rate of interest on external liabilities exceeds the economy’s growth rate, then external liabilities will tend to grow faster than GDP, unless a country runs a trade surplus. Of course, the current account can be in permanent deficit and the liabilities to GDP ratio remain stable as long as the nominal growth rate of the economy is sufficiently positive.

8. The interest rate on U.S. debt instruments plus an estimate of the country risk premium is often used as a proxy for the implicit cost of capital. In the second half of the 1980s, there was a significant risk premium (as proxied by the interest rate differential, which includes both inflation risk and sovereign risk) on Australian bonds, which reached about 7 percentage points on two occasions (Figure I.1). However, in the 1990s, this risk premium has narrowed considerably, and in the past year it has virtually disappeared.

FIGURE 1.1
FIGURE 1.1

AUSTRALIA FOREIGN LIABILITIES, INTEREST RATES, AND CURRENT ACCOUNT BALANCE

Citation: IMF Staff Country Reports 1999, 001; 10.5089/9781451801958.002.A001

Sources: Australian Bureau of Statistics; Reserve Bank of Australia; IMF, International Financial Statistics; WEFA, INTLINE Database; and Fund staff estimates.1/ National accounts basis.2/ Excludes sale of gold in June and September 1997.

9. There are a number of factors, apart from the risk premium, which affect the implicit cost of capital. Particularly important in Australia’s case is the composition of external liabilities and assets. The returns on some external assets may be low or even negative, for example, where direct investment has not matured enough to become profitable. Similarly, investments in bonds denominated in strong currencies but carrying low yields, or in equities, often pay only small amounts in terms of current income.3 The implied rates of return on debt and equity investments are shown in Table I.1, which reveals that equity investments in Australia require a higher return than debt, presumably to compensate for increased risk.4 This higher income stream to external equity investors implies that Australia must run a higher trade surplus than if debt financing were used exclusively.

Table I.1.

Australia: Average Implied Rates of Return on External Assets and Liabilities, 1995–97

(In percent per annum)

article image
Source: Australian Bureau of Statistics, Balance of Payments and International Investment Position, June 1998.

10. An important feature of the effective cost of external capital relates to capital gains and losses.5 Profits and losses caused by exchange rate fluctuations can potentially be very large, even swamping underlying capital flows (Table I.2). For example, in 1997, net equity inflows were nearly offset by exchange rate and asset price changes, while the net debt position worsened considerably, largely because of exchange rate movements.

Table I.2.

Australia: Contributions to Net External Liabilities

(In percent of GDP)

article image
Source: Australian Bureau of Statistics, Balance of Payments and International Investment Position, June 1998.

11. To understand the implications of these dynamics for Australia, some simulations were conducted. The simulations were formulated to calculate the current account balance required under three different assumptions about the target ratio of net external liabilities to GDP, NEL: (i) stabilize NEL at 60 percent (the level at end-March 1998 as a percent of annual GDP); (ii) reduce NEL to 50 percent over the next five years; and (iii) reduce NEL to 50 percent by 2010. These calculations assume that interest rates and growth rates remain constant over time, but a range of values for these rates are used to gauge the sensitivity of the results.

12. Assuming that Australia’s long-run real potential growth rate is 3½ percent per year, inflation averages 2½ percent (the midpoint of the Reserve Bank’s target), and an interest rate of 6 percent (equal to the rate of nominal GDP growth), then to stabilize NEL at current levels, the trade account must be in approximate balance in the future. This results in a current account deficit of 3¾ percent of GDP, given the net income deficit. For each percentage point increase in the world interest rate or fall in the growth rate of the Australian economy, the required increase in the trade surplus to keep NEL constant is 0.6 percent of GDP.6

13. If the goal were to reduce NEL to 50 percent of GDP by the year 2003, then under the same growth and inflation assumptions as above, there would need to be a current account deficit of 1.7 percent of GDP. If there was a less ambitious goal to reduce NEL to 50 percent of GDP by the year 2010, the current account deficit target would be 2.7 percent of GDP.

C. The Consumption-Smoothing Approach to the Current Account

14. In the previous section, the question was to determine what level of the current account balance would be required to stabilize the net external liabilities ratio at its approximate current level, and also to consider some alternative scenarios to bring it down from the current level over different time horizons. No judgment was made as to the appropriate level of the current account, given the economic environment, or the appropriate level of external liabilities.

15. The question of whether a given current account position is inappropriate can really only be answered within the context of some model that yields predictions about the “optimal” path of external imbalances. The most common such model is the consumption-smoothing model of the current account, in which the current account is used to smooth consumption in the face of shocks to the economy.7

16. The model is narrowly focused on consumption and saving behavior and takes the supply side of the economy and the world real interest rate as given. In particular, the model assumes that output appears as stochastic returns to exogenously determined investment (which can be optimally chosen to maximize the net present value of income given the interest rate), and that the government has access to lump-sum taxation to finance its expenditure, choosing a spending and taxation path that results in intertemporal solvency. These assumptions are necessarily restrictive, and imply that the model omits certain factors that may impinge on the current account balance, including: terms of trade shocks; real interest rate shocks; changes in employment, investment, and productivity due to labor and product market reforms; and demographic factors.

17. At its most basic level, the consumption-smoothing model predicts that the current account will be in deficit when future changes in net output (output net of investment and government consumption) are expected to be positive, so that future net output is transferred to the present (by external borrowing) to smooth the path of consumption. For example, if due to a temporary adverse export shock (such as the Asian crisis) net output falls below trend, the optimal response of consumers would be to borrow against this increase in expected future net output to absorb the shock. In this case, both the actual and warranted current account balances would deteriorate, other things equal, though not necessarily by the same amount. In contrast, if the shock were expected to be permanent, then the optimal response would be to adjust consumption downward immediately; in this case, there would be no change in the warranted current account balance.

18. The actual current account and the warranted current account as generated by the model are shown in lower panel of Figure I.1 (see Appendix I.1 for details of the derivation of the warranted current account). The figure shows that the consumption-smoothing model fits quite well, particularly in the 1990s, and this is confirmed by the more formal tests of the model discussed in Appendix I.1. The model appears to be capturing economically and statistically significant elements of actual external borrowing behavior. This can be seen both by the close correlation between the two series as well as by the model’s ability, in a number of cases, to capture the timing of peaks and troughs in the actual pattern of external borrowing.

19. One question is how to interpret deviations between the actual and optimal current account generated by the model. Part of the reason for the difference between the series is simply sampling error. However, another reason may be related to too much external borrowing or lending. In fact, as can be seen in Figure I.1 (also see Appendix I.1), periods in which actual external borrowing exceeds the warranted level are not offset over the sample by periods in which actual borrowing falls short of the warranted level of borrowing. Indeed, on average over the entire sample, the actual deficit exceeds the warranted deficit by nearly ¾ percent per year, although the unwarranted accumulation of external liabilities was significantly less in the 1990s than in the 1980s (see Table I.1), perhaps reflecting the impact of structural reforms that improved economic performance and reduced distortions affecting saving and investment decisions.8 The cumulative impact over the entire sample is to have raised external liabilities above their warranted level—by nearly 10 percent of GDP—according to the benchmark provided by the model.

20. As seen in Figure I.1, the actual deficit is larger than the warranted deficit mainly in three distinct episodes: 1984–86, 1988–90, and 1994–95. The 1984–86 episode followed the liberalization of financial markets coupled with an income contraction from an adverse terms-of-trade shock.9 The financial liberalization reduced credit constraints allowing maintenance of consumption even if, over the long term, it was not optimal to do so. Adding to this problem, public savings deteriorated sharply in this period (as the underlying Commonwealth budget deficit widened to 3½ percent of GDP in 1983/84), reducing the pool of domestic savings available and increasing the reliance on external savings. The 1988–90 episode was associated with an investment boom as the interaction between the tax system and inflation created an incentive for highly leveraged investment in assets such as property and stocks rather than plant and equipment (Macfarlane, 1989, 1990). In the 1994–95 episode, strong growth in imports of investment goods coupled with a drought that reduced rural exports was a major cause of the rise in the current account deficit.

D. Summary and Conclusions

21. This chapter presented a simple sustainability analysis showing that for Australia to stabilize the net external liabilities ratio at current levels, the current account deficit would need to be brought down to about 3¾ percent of GDP. A higher external deficit would imply that net external liabilities would continue to rise in relation to GDP, other things equal. The chapter went on to examine the same issue in a more formal framework that provided a benchmark against which to evaluate whether actual current account imbalances have been too large or not. The framework was based on the permanent-income theory of consumption applied to a small open economy with access to international capital markets. In such a model, it is optimal for a country to use the current account to smooth consumption when faced with temporary disturbances to productivity, investment, or government spending.

22. The main finding from the consumption-smoothing model was that, since the early 1980s, the warranted level of the current account deficit has averaged about 3¾ percent of GDP. With the actual current account deficit having averaged 4½ percent of GDP, the model suggested that there was an unwarranted accumulation of external liabilities, on average over the sample, of about ¾ percent of GDP per year. This unwarranted accumulation, however, was significantly lower in the 1990s than in the 1980s and, during the 1990s, largely reflected the experience in 1994–95 (as discussed above). As a result of these higher-than-warranted current account deficits, Australia’s net external liabilities were found by the model to exceed—by about 10 percent of GDP—their warranted level.

23. The consumption-smoothing model developed in this chapter clearly ignores a range of other potentially important aspects of current account sustainability. Broader indicators of sustainability—for example, inflation, the fiscal consolidation program, structural reforms in factor and product markets, and the soundness of the financial system—are all clearly favorable in Australia’s case. Indeed, the strength of Australia’s underlying fundamentals has undoubtedly been a factor driving the reduction in long-term bond rates to international levels in the recent past. This said, the empirical results in this chapter may nevertheless underline the risks associated with the present widening of the current account deficit, given the legacy from the past accumulation of external liabilities.

APPENDIX I.1 The Consumption-Smoothing Model and Empirical Results

The consumption-smoothing model of the current account assumes a small open economy with access to world capital markets, in which the representative consumer maximizes

EtΣj=0βju(ct+j),0<β<1,(1)

where Et is the expectations operator, ct is private consumption at time t, u( ) is a separable utility function such that u'>0, u"<0, and p is the subjective discount factor. If β >1/(1+r), then a country will tilt consumption into the future, and vice-versa. The real world interest, r, is exogenously given. Letting bt denote the economy’s stock of net external liabilities at the beginning of period t, qt real output (GDP), it real investment, gt real government consumption, and Δ the first difference operator, the consumer’s budget constraint can be written as:

Δbt+1=rbt-(qt-ct-it-gt).(2)

For the purpose of the empirical implementation, a quadratic utility function is chosen. The optimal path of consumption in this case is given by

ct*=(r/Θ)[-bt+(1-r)-1Et(Σj=0(1+r)-jzt+j)](3)

where Θ = (β(1+r)2-1)/βr(1+r) measures the degree of consumption-tilting and zt = qt- it - gt is known as net output or national cash flow (GDP net of investment and government expenditure). When Θ < 1 (Θ>1), the country consumes more (less) than its permanent cash flow or tilts consumption toward the present (future). Defining the optimal consumption-smoothing current account balance as cat*=zt-Θct*-rbt we have

cat*=-Et[Σj=1(1+r)-jΔzt+j].(4)

An estimate of the optimal current account can be obtained by using current and lagged current accounts to form some proxy of the expected values in (4). This can formally be accomplished by estimating a bivariate autoregressive model of the current account balance and national cash flow of the form Wt = AWt-1 + ∈t, where Wt=(Δzt,cat*), ∈t is a 2×1 vector of disturbance terms, and A is a 2×2 matrix of coefficients. With the estimate of A from the VAR and using the fact that Et=[Wt+j] = Aj Wt, an estimate of the optimal consumption-smoothing component of the current account can be computed as

cat*=[-10][(1+r)-1A][I2-(1+r)-1A]-1WtΓWt(5)

For statistical purposes, it is necessary to remove any nonstationary components of the current account.1 This is achieved by defining the current account as the residuals of a cointegrating regression of consumption on national cash flow less interest payments. The slope coefficient from this regression yields an estimate of the consumption-tilting parameter.

To implement the consumption-smoothing model, national account data on GNP, GDP, private consumption, investment, and government consumption are required. The data source is the IMF’s International Financial Statistics. The nominal national accounts data are deflated by the implicit GDP price deflator in order to obtain all real magnitudes on a consistent basis. The model was estimated with quarterly data (seasonally adjusted, at annual rates from 1984:1 to 1998:1, with a break point estimated to be at 1990:4). The “consumption-tilting” component of the current account was removed from the data prior to estimation.

The next step in the empirical exercise is to evaluate the performance of the consumption-smoothing model. The easiest method is to compare the correlation of the actual and optimal current accounts. While this correlation was negative in the 1984–90 period it has risen to about 85 percent in the 1990s (first column of Table 1). A more formal (Wald) test of the model is also available. To pass this test, the actual current account (denoted by ca) must be equal to the warranted current account as predicted by the model (denoted by ca*). The result from this test indicates that the restriction is rejected for the earlier part of the sample but is not rejected by the data in the 1990s, (second column of Table 1). Furthermore, the average gap between the actual and warranted current account deficits in the 1990s is about half as large as the gap in the 1980s.

Table 1.

Australia. The Consumption-Smoothing Model: Statistical Results

article image
Source: Fund staff calculations.

A structural-break test was used to determine whether there was a change in behavior over the sample. The test indicated there was a break in 1990q4.

Rejected means that the Wald test of the null hypothesis that ca = ca* exceeds the critical value of the χ2(2) distribution at the 5 percent level.

In percent of GDP.

References

  • Bullock, M., S. Grenville, and G. Heenan, 1993, “The Exchange Rate and the Current Account,” in A. Blundell-Wignell (ed.), The Exchange Rate, International Trade and the Balance of Payments, Reserve Bank of Australia.

    • Search Google Scholar
    • Export Citation
  • Cashin, P., and C. J. McDermott, 1996, “Are Australia’s Current Account Deficits Excessive?,” IMF Working Paper 96/85 and Economic Record, forthcoming.

    • Search Google Scholar
    • Export Citation
  • Commonwealth of Australia, 1990, “Budget Statement 3,” in 1998-99 Budget Paper No. 1, (Canberra: Australian Government Publishing Service).

    • Search Google Scholar
    • Export Citation
  • Ghosh, A.R., 1995, “International Capital Mobility Amongst the Major Industrialized Countries: Too Little or Too Much?” Economic Journal, Vol. 105, pp. 10728.

    • Search Google Scholar
    • Export Citation
  • Ghosh, A.R., and J. D. Ostry, 1995, “The Current Account in Developing Countries: A Perspective from the Consumption Smoothing Approach,” World Bank Economic Review, Vol. 9, pp. 30533.

    • Search Google Scholar
    • Export Citation
  • Kent, C. J., 1997, “The Current Account, Consumption Smoothing, and Credit Constraints,” in C. J. Kent, Essays on the Current Account, Consumption Smoothing, and the Real Exchange Rate, Ph. D. Thesis, Department of Economics, Massachusetts Institute of Technology.

    • Search Google Scholar
    • Export Citation
  • Macfarlane, I. J., 1989, “Money, Credit and the Demand for Debt,” Reserve Bank of Australia, Bulletin (May).

  • Macfarlane, I. J., 1990, “Money, Credit and the Demand for Debt: Part II,” Reserve Bank of Australia, Bulletin (May).

  • Milbourne, R., and G. Otto, 1992, “Consumption Smoothing and the Current Account,” Australian Economic Papers, Vol. 31, pp. 36983.

    • Search Google Scholar
    • Export Citation
1

This chapter was prepared by John McDermott The chapter was presented at a seminar during the 1998 Article IV consultation mission to Australia, and the author is grateful to the participants and discussants for helpful comments and suggestions.

2

The dynamics are described by the equation ΔNELt+1=(r-γ1+γ)NELt-TBt

where Δ indicates the change from the previous period, NEL is the net external liabilities to GDP ratio, TB is the trade balance (goods and services excluding interest payments) to GDP ratio, r is the implicit cost of capital on net external liabilities (encompassing any valuation effects from exchange rate fluctuations and market price changes), and γ is the economy’s rate of nominal growth.

3

About 40 percent of gross external debt is denominated in Australian dollars: 32 percent in U.S. dollars, and the rest in other currencies.

4

Capital gains are excluded from the implicit returns and presumably they would be higher for equity.

5

According to the IMF’s Balance of Payments Manual, realized profits and losses are not reported in the current account balance but are recorded as a financial transaction. Unrealized profits and losses are not shown in the balance of payments.

6

If there are nominal capital gains on external liabilities, the surplus on the trade account required to stabilize net external liabilities would be higher.

7

Details of the model and a technical explanation of how the model was empirically implemented using Australian data can be found in Appendix I.1 and in Cashin and McDermott (1996).

8

The data were adjusted to exclude gold sales by the Reserve Bank in 1997 ($A 1.8 billion in the June quarter and $A 0.7 billion in September quarter) from the current account.

9

An attempt was made to extend the model to examine how changes in the terms of trade affect the current account. The empirical results, however, were poor and suggested that inclusion of the terms of trade does not improve the correlation of the warranted and actual current account deficits.