Too much focus on the yuan?

Alan Auerbach, Maurice Obstfeld

23 October 2010



China’s trading partners have long criticised the country’s policy of maintaining a weak renminbi. These complaints have intensified in the face of continuing high unemployment and slow recovery in high-income industrial countries (see recent contributions on this site by Caballero 2010 and Huang 2010). Threats of trade sanctions by the US Congress have resulted in periods of measured and limited renminbi appreciation – most recently, a 2.3% rise against the dollar between early September and mid-October 2010. But such gestures by the Chinese authorities fall far short of the 20%-or-better, maxi-revaluation demanded by China’s critics in the US and elsewhere.

International tensions have intensified as a number of other emerging market economies, following China, are likewise resisting currency appreciation. But promoting domestic economic growth at the expense of other countries through currency devaluation raises the fear that a “currency war” escalates into a war of administrative trade barriers that would severely harm the international trading system.

Why are pro-aggregate demand devaluations so popular?

Why is currency devaluation perennially so attractive as a way to stimulate aggregate demand? One reason is the perception that its cost – in terms of higher prices paid for imports relative to those charged for exports – is moderate and spread quite broadly among consumers, so that the cost per household is small. But other policies to stimulate demand, such as fiscal policies, also have costs that would be shared quite broadly among households. If the goal of policy in the current slowdown is to achieve a given stimulus at the lowest aggregate cost, then it is not obvious that pressuring China over its exchange-rate intervention policies is the most efficient way to go.

This is not to deny that other important issues, such as the need to reduce global imbalances or the risks of political conflict with China, will ultimately affect the calculus. But we believe that an essential starting point for any more comprehensive analysis is to calculate the balance between immediate and quantifiable benefits and costs of any policy under consideration. The resulting benefit-cost ratios provide an initial means for comparing possible policy interventions.

Thus, although pushing for a Chinese revaluation makes more sense in a high-unemployment period than in normal times, its attractiveness should be compared to other policies we might consider, and in this note we consider one obvious and relevant alternative, a fiscal expansion via government purchases, which should also be more attractive in recession, again because we can hope to increase output.

When thinking of policies aimed at stimulating short-run aggregate economic activity, it is natural to measure benefits by increases in output, and costs by real income losses incurred to achieve such output increases. For a yuan appreciation, the cost is measured in the lost purchasing power due to more expensive imports. For an increase in government spending, meanwhile, the cost is the future taxes (or spending reductions) that must pay for the spending, net of any value placed on that new spending.

To simplify our analysis, we argue that we can ignore multiplier effects, including the revenue feedback associated with higher output. This is because we expect the relative attractiveness of the two policies not to change according to whether we have high unemployment or not.

Cost-benefit of fiscal stimulus

For fiscal policy, the benefit-cost calculation is straightforward. Increased output per dollar of government purchases is 1.0 minus the fall in private consumption due to higher expected taxes (if tax increases are anticipated) or future government spending cuts. For taxes, this impact on consumption will vary from 0 (myopia/liquidity constraints) to that based on a present-value calculation. But assuming that we are considering a temporary increase in spending, the upper bound on the private consumption response is small, say 0.05 given a reasonable consumption-wealth ratio. (One would not anticipate a response any larger if future spending cuts were expected.) So the impact benefit per dollar of increased output is close to 1. As to the net cost (assuming again that it is borne through tax increases), there are two components: the cost of extra revenue, including the excess burden of taxation, less the value society places on the new public spending; this value would be zero for the classic Keynesian policy of digging ditches and filling them in, but strictly between 0 and 1 + Dead-weight-loss (DWL) per dollar of spending for policies that have value but under normal conditions would not be worth adopting. Recent estimates of DWL are scarce, so one frequently cited is Ballard et al. (1985), who estimated the DWL for the US tax system of around 0.33, with a range as high as 0.5. So let us say that the benefit per dollar of spending is .95, and the cost is, say, 1.4 - v, where v is the value to consumers of the dollar of public spending. This leads to a benefit-cost ratio of 0.95/(1.4 - v).

Cost-benefit of exchange-rate policy

For exchange-rate policy, let total national expenditure be A. From the definition A = D + pM, where p is the relative price of imports in terms of domestic goods and D is spending on domestic goods, we get the standard result that the percent expenditure change equivalent to a percent rise in relative import prices is , where the ^ symbol represents a percentage change. Only under balanced trade is expenditure A equal to GDP Y. The effect on output of a fall in the terms of trade (ignoring the multiplier effect, as discussed above) is , where the s are the (absolute) export and import elasticities with respect to price. This makes the benefit-cost ratio, as we have defined it for fiscal policy (output per income loss) equal to


For balanced trade, pM = X.1 In 2009 U.S. data, the ratio of exports of goods and services to imports is about 0.8. Putting both s in the range [0.5, 2.0] would bracket most existing estimates of aggregate trade elasticities for the United States. The resulting range for the benefit-cost ratio is [-0.1, 2.6]. For the lower end of this ratio, fiscal policy is more attractive for all non-negative values of v. For the top end of this ratio, fiscal policy is more attractive if v = 1.04, which is still below 1.4. So, the elasticity range goes from fiscal policy definitely being preferred to exchange-rate policy possibly but not definitely being preferred.

But very few economists would argue that both the aggregate import and export elasticities are as high as 2. For more conventional estimates of both elasticities being equal to 1, we get that v must exceed 0.09, which seems highly likely. So, looking only at the short-run effects on output relative to the real-income losses, fiscal policy seems far more attractive.

Concluding remarks

In the preceding analysis, we implicitly assume, as do many other commentators, that in calculating the trade balance response and the real income loss alike, the appropriate measure of the change in the relative price of imports is simply the percent change in the dollar-yuan exchange-rate scaled by China’s share in overall US trade (see for example Cline 2010). We also assume that the trade-balance response to yuan revaluation would be captured by the overall terms-of-trade change, measured in this way, and the aggregate trade elasticities estimated with respect to historical average terms-of-trade changes.

In reality, however, the trade response to a given change in the overall effective (or trade-weighted average) terms of trade can differ greatly depending on the particular bilateral changes generating the effective change. For example, China’s foreign customers are much more likely to switch to Mexican rather than US exports as a result of a yuan revaluation; in that case, the primary benefit to US exports would come from higher Mexican income, which would lead Mexico to demand more US goods.

Moreover, in considering the exchange-rate question, we have not discussed alternative policies we might seek from China, in particular having it pursue its own domestic output expansion. This would be preferred from the US perspective, for it would increase export demand without necessitating a dollar depreciation. On the other hand, imposing tariffs on Chinese goods would be preferable to yuan appreciation against the dollar on strict US welfare terms – this is basically the optimal tariff argument – but China would suffer more and the collateral damage to the world trading system may well be immense.

Yet while these calculations are admittedly oversimplified in many dimensions, we believe they are illustrative of the orders of magnitude involved. Our numbers suggest that a large Chinese revaluation, whether forced or voluntary, is far from a free lunch for the United States, and that further fiscal stimulus should not be dismissed as an alternative if the goal is to create jobs at lowest cost in terms of national income.


Ballard Charles L, John B. Shoven, and John Whalley (1985), “General Equilibrium Computations of the Marginal Welfare Costs of Taxes in the United States,” American Economic Review, 75(1):128-138.

Cabellero, Ricardo (2010),”Feasible global rebalancing: A case for monitored and temporary dual exchange rates”,, 19 October.

Cline, William R (2010), “Renminbi Undervaluation, China’s Surplus, and the US Trade Deficit”, Policy Brief 10-20, Peterson Institute for International Economics, p 3

Huang, Yipping (2010),”A currency war the US cannot win”,, 19 October.

1 In the case of balanced trade, the sum of the elasticities must exceed 2 for a devaluation to have a positive effect on real national income.



Topics:  Global economy International trade

Tags:  US, global imbalances, China, fiscal stimulus, exchange-rate policy

Professor of Economics and Law at the University of California, Berkeley

Economic Counsellor and Director of Research, IMF