Factors affecting the US current account: an historical decomposition.

Barrell, R. ; Hollan, D. ; Hurst, I. 等

The US current account deficit was around 6V2 per cent of GDP in 2006, having risen by 2 per cent of GDP since 2002 when the dollar reached its peak and started to decline. Between the peak in 2002 and the last quarter of 2006 the dollar fell by 13.4 per cent in effective terms using 2003 trade weights, and it might have been expected that the current account would have improved. However, over the same period the household savings rate fell by more than 4 per cent of personal incomes and private sector investment rose by 0.6 per cent of GDP. At the same time, the public sector deficit improved by around 1 per cent of GDP, but this was not enough to offset the major decline in net savings in the private sector. These domestic imbalances have been partly responsible for the deterioration in the current account, and appear to have more than offset the impacts of the decline in the dollar. In addition, since 2002 the oil price has risen and this has led to a significant deterioration in the US current account.

We can decompose the current account effects of the rise in the oil price and the fall in the exchange rate using our model NiGEM. It is usual to presume that agents in the foreign exchange markets look forward, and form expectations about interest rates and other events that may affect the evolution of the currency. In this counterfactual analysis we adopt this standard approach. The arbitrage equation for the bilateral exchange rate [e.sub.t], in terms of domestic currency per unit of foreign currency, may be written as

[e.sub.t] = [e.sub.t+1]((1 + [rf.sub.t]))(1 + [rp.sub.t]) (1)

where [rh.sub.t] is the interest rate at home, [rf.sub.t] is the interest rate in the partner country and [rp.sub.t] is a risk premium. Exchange rates change because one of these factors changes. A rise in domestic interest rates (now or expected in the future) will cause the exchange rate to strengthen, whilst the same change abroad will cause it to weaken. We presume that interest rates are set by the monetary authority in response to a target for inflation and deviations from that target that are expected to happen in the future. For instance a rise in the oil price will induce a rise in interest rates, and in our model the forward-looking long-term interest rate would rise. A rise in the risk premium will cause the exchange rate to fall and there is evidence that there is a systematic element to the risk premium, for instance in Al Eyd, Barrell and Holland (2006) who present evidence of an asset related risk premium on the US exchange rate.

The current exchange rate should fully discount the future path of asset holdings and hence the exchange rate we observe may be sustainable even with growing debts. However, if 'news' arrives that suggests that the nation's debt will rise, then the perceived risk premium could increase and hence the exchange rate would jump down now. Each time the exchange rate fell we might expect an initial worsening of the current account for a year as prices change in advance of quantities (the J curve effect of the first year textbook). Hence we might have expected no sustained improvement until at least a year after the last downward step towards the end of 2004. (1)

[FIGURE 1 OMITTED]

It is possible to model this history by taking off each of the major steps down in the currency, starting with the last, and evaluating what would have happened if the fall had not taken place. The new 'history with a higher exchange rate' is then used as the baseline against which we remove another fall in the exchange rate. As we can see from figure 1, we repeat these steps until we have an approximately constant exchange rate from 2003. This is a similar procedure to that used in Barrell et al. (2007a), who analyse the possibility of an orderly risk premium driven adjustment of the US exchange rate and of US imbalances.

The exchange rate changes are assumed to be driven by small changes in the risk premium, and as discussed in Barrell et al. (2007b) this has real effects in the longer term, as it causes a wedge to develop between real interest rates in the US and those elsewhere, and hence changes relative domestic absorption as well as causing expenditure switching because of the change in the real exchange rate. Figure 2 plots the impacts on the US current account and these experiments suggest that, if the exchange rate had not fallen, the US current account would have been approximately 2 per cent of GDP worse than it now is.

[FIGURE 2 OMITTED]

The US is a major importer of oil and other forms of energy, although the share of energy in imports fell consistently from 35 per cent in 1980 to 9 per cent in 1997, partly as oil prices fell in real terms, but also because volumes of other imports rose rapidly. Between 1997 and 2005, the oil price, in US dollars, rose approximately three fold, and energy imports as a share of the total rose from 9 per cent in 1997 to 18 per cent in 2005. The overall decline in the energy balance, which was mainly driven by prices, contributed around a third of the deterioration in the US trade balance since 1997 and 60 per cent of the deterioration since 2002. The overall decline comes from the effects of the change in price and the decline in US production as well as the rise in usage driven by strong domestic growth. We look at the effects of the change in price, assuming oil production in the US would not have varied with the oil price, and that oil consumption changes 'mirror' GDP changes.

The current account effects of a rise in the oil price are interesting, as they may not be obvious. If the owner of an exhaustible resource receives a windfall gain in revenue of K for one year, then they should spend rK in that year, where r is the rate of return, and use the remainder of the revenue to invest in non-exhaustible assets.2 Hence the balance of payments of the resource producer improves by (1 - r)K in that year and the balance of payments of the rest of the world worsens by the same amount. Over the past five years we have seen the price of an exhaustible resource rise noticeably. Even if the price rise is permanent, the resource producers should invest much of the revenue produced in foreign assets. If the price rise is perceived as temporary then more would be saved. Oil producers such as Norway have for some time had oil funds that accumulate foreign assets, and such funds are becoming more common, for instance in Russia. All oil producers should have been accumulating assets and running current account surpluses in the past few years, and they have been doing so. The rest of the world must run a structural current account deficit for the same period. It is not surprising that much of this structural deficit might turn up in the US (and the UK) as these are perhaps the economies where it is easiest to invest. Hence it is possible that the effects of the oil price rise on the US current account deficit have been greater than our analysis suggests, but it would be hard to quantify this effect.

Figure 3 plots the oil price profile in a sequence of simulations that remove the major increases in prices seen in the past few years, but does not take account of every gyration in the price. NiGEM has been run in standard forward mode each time3 and each part of the stack uses the resulting baseline constructed in the previous part of the stack. These stacked simulations allow expectations to emerge so that individuals in 2002 are not aware that the oil price will rise by $40 per barrel. The effects of this counterfactual analysis on the US current balance can be seen in figure 4, which shows the path the US current account would have otherwise taken, given that policy reactions would have been in place and that oil producers, who receive less revenue, would have spent less. Overall the rise in the oil price seen since 2002 has worsened the US current account by around 1 per cent of GDP.

[FIGURES 3-4 OMITTED]

This analysis suggests that the effects of the fall in the dollar over the five years to the end of 2006 have already come through to the current account. In addition, even a large and sustained fall in oil prices will only have a limited impact on the US deficit. Unless there is a major change in absorption in the US, it is unlikely that there will be any significant improvement in the US current account. As we argue in Barrell, Holland and Hurst (2007a) we would then expect a risk premium to develop slowly on US assets, resulting in a weaker real exchange rate and higher real interest rates. The former would induce expenditure switching whilst the latter would induce a reduction in absorption. If the US dollar were to drift down a further 18-20 per cent driven by a rise in risk premia on US assets, the US current account would improve by 3 1/2 to 4 per cent of GDp.

REFERENCES

Al-Eyd, A., Barrell, R. and Holland, D. (2006), 'A portfolio balance explanation of the euro dollar rate', in Cobham, D. (ed.), The Travails of the Eurozone, Palgrave.

Barrell, R., Holland, D. and Hurst, A.I. (2007a), 'Correcting US imbalances', National Institute Discussion Paper no. 290.

--(2007b), 'Sustainable adjustment of global imbalances', in Aslund, A. and Dbrowsk, M. (eds), Challenges of Globalization: Macroeconomic Imbalances and Development Models, Peterson Institute of International Economics.

Sefton, J.A. and Weale, M.R. (1996), 'The net national product and exhaustible resources: the effects of foreign trade', Journal of Public Economics, 61 (1), July.

NOTES

(1) The appreciation of the dollar was relatively recent in 2002, and the increase over the previous four years may not have had much impact on the US current account.

(2) This is Hartwick's rule. See Sefton and Weale (1996), for a relevant discussion.

(3) Financial and labour markets are assumed to be forward looking, as is price setting and the investment decision. Although consumers observe and react to 'jumps' in their wealth, they are not assumed to change their perception of their permanent income.