Currency Misalignment: A Reprise

As the Congress debates currency manipulation [1], it occurs to me useful to reprise my earlier primer on currency misalignment (first published in March 2010), where misalignment is one component of some definitions of currency manipulation.

Currency misalignment can be determined on the basis of the following criteria or models:

  • Relative purchasing power parity (PPP)
  • Absolute purchasing power parity
  • The “Penn Effect”
  • The behavioral equilibrium exchange rate (BEER) approach
  • The macroeconomic balance effect

  • The basic flows approach
  • An equilibrium approach

I have discussed several of these approaches in the past [0] [1], but a review of the approaches bear repeating, if only because there so much confusion regarding what constitutes currency misalignment.

Relative PPP

Relative PPP can be expressed as:

s = μ + p – p*

where lowercase letters denote log values, s is the price of foreign currency, p is the price index, and * denotes a foreign variable, and μ is a constant arising from the fact that p and p* are indices.

Using this criteria, a currency is misaligned if s deviates from the μ + p – p*. One difficulty is that μ has to be estimated. Typically, estimates of μ can vary drastically with sample period. Oftentimes, there is a time trend in q ( which equals s – μ – p + p*), which means that relative PPP cannot literally hold. Then, one might have to allow for a (ad hoc) time trend, which itself has to be estimated. In Chinn (Emerging Markets Review, 2000), I apply this approach to the East Asian currencies, pre-crisis.

In the case of China, presented below is the (log) trade weighted real effective exchange rate of China, (q), using the latest data spliced to an older series incorporating the swap rates pre-1994 per discussion in Chinn, Dooley, Shrestha (1999). Upwards denotes depreciation.


Figure 1: Log trade weighted real effective exchange rate, CPI deflated (blue); and linear trend estimated 1980-2009. Source:
IMF, International Financial Statistics, various issues; and author’s calculations.

Using a simple linear trend, one obtains the counter-intuitive result that the RMB is overvalued. This suggests caution.

For more on effective exchange rates, see this survey paper.

Absolute PPP

Given the drawback of relative PPP, it seems like one could get around the problem of estimating μ by using actual prices of identical bundles of goods across countries, rather than price indices.

s = p – p* or equivalently p = s + p*

where lowercase letters denote log values, s is the price of foreign currency, p is the price level, and * denotes a foreign variable.

The problem is that prices of identical bundles of goods are not collected. One thing that comes close is the Big Mac — hence the MacParity measure. But the problem is that prices of Big Macs (when expressed in a common currency) are systematically lower in lower income countries, and systematically higher in higher income countries. This is true when using Big Mac prices [2] [latest estimate] Parsley and Wei (2005) and when using the “price levels” in the Penn World Tables. This is so much a stylized fact that it is sometimes called “The Penn Effect”.

The Penn Effect

Instead of viewing the “Penn Effect” as a problem, one can exploit this stylized fact. Define r ≡ p – s – p*. Then one can exploit the relationship:

r = α 0 + α 1 (y-n)

Where y-n is log per capita income. In Cheung et al. (2008), we exploited this relationship (following Frankel (2005), in a panel regression setting. In that study, we found an approximate 40% RMB misalignment (in log terms). Using updated data, namely the 2008 vintage of the World Development Indicators, we found something closer to 10% misalignment. The scatterplot of data, the regression line, and the RMB’s path are depicted in Figure 2, originally discussed in this post (paper here).


Figure 2: Price level-per capita income in PPP terms relationship (blue line), +/- 1 std error band (long dashed lines), +/- 2 std error band (short dashed lines); OLS estimates for 1980-2006 period. Source: authors’ calculations.

The Behavioral Equilibrium Exchange Rate (and related) Approach

Yet another approach is to use some theoretically and empirically motivated equations to estimate an exchange rate relationship. The variables can include productivity variables, or relative price of nontradables, or fiscal variables like the deficit, or interest differentials. Typically the variables can be motivated by some model of the exchange rate; which ones are included are motivated by goodness of fit. Goldman Sachs and JP Morgan had models of this sort. Zhang (China Economic Review, 2001) and Wang (2004) in Prasad (IMF, 2004) are some models in the public domain (see this 2007 Teasury working paper). Chinn (2000) implements a specific type of BEER (or productivity based) model for East Asian exchange rates.

The Macroeconomic Balance approach

The Macroeconomic Balance approach takes the perspective from saving and investment rates (see Peter Isard’s survey). Recall:

CA ≡ (T-G) + (S-I)

In other words, the current account is, by an accounting identity, equal to the budget balance and the private saving-investment gap. This is a tautology, unless one imposes some structure and causality. One can do this by taking the budget balance as exogenous (or use the cyclically adjusted budget balance), and then include the determinants of investment and saving. Then one obtains “norms” for the current account. Chinn and Prasad (2003) is one example of this approach.

Then, using trade elasticities, one can back out the real exchange rate that would yield that current account. If that exchange rate is stronger than the actually observed exchange, then that currency would be considered “undervalued”.

The closely-related Fundamental Equilibrium Exchange Rate (FEER) determines the current account norm on a more judgmental basis (in other words, the current account norm is not estimated econometrically, just imposed per the analysts priors).

[update 3/24, 8am Pacific] I should have mentioned Paul Krugman’s take on this issue. My interpretation is that it fits into the Macroeconomic Balance approach, except he sidesteps the step of comparing the counterfactual exchange rate that hits the CA norm against the actual, and just compares the CA against the CA norm. In Chinn and Ito (2008) (Figure 4), we found — at least in the 2001-04 period — that China’s CA was verging on being statistically significantly different from the norm. My guess is that the CA would be significantly different from the CA norm in the 2004-08 period.

The Basic Balance approach

One could take a more ad hoc approach, asking what is the “normal” level of stable inflows — for instance looking at the sum of the current account and foreign direct investment, and see whether that value “made sense”. Or one could look at the sum of the current account and private capital inflows. If either of the flows are “too large”, then the currency would be considered undervalued (since a stronger currency would imply a smaller current account balance).

It is interesting to make two observations. First, note the need for many non-model based judgments. To see this point, recall the balance of payments accounting definition:

CA + KA + ORT ≡ 0

Where CA is current account, KA is private capital inflows, and ORT is official reserves transactions (+ is a reduction in forex reserves).

Saying CA + KA is too big is the same, then, as saying ORT is too small, i.e., reserves are rising “too fast”. Morris Goldstein and Nick Lardy are among the most prominent exponents of this approach (see e.g., this this small volume).

Alternatively, running surpluses that are “too large” for “too long” will lead to foreign exchange reserves that are “too large”. Obviously, a lot of judgment is necessary here.

Second, one aspect of this judgment is that it is conditional on the constellation of all other macro policies, including monetary, fiscal and regulatory, in place. If the CA+KA is adjudged to be “too large”, one could say the exchange rate is “too weak”, but one could say with equal validity that the fiscal policy is “insufficiently expansionary”.

A (newer) Equilibrium View

The final approach would be to step back and think in terms of “equilibrium” exchange rates. One might argue that the exchange rate is undervalued if, in the absence of central bank intervention, the exchange rate would be stronger. Of course, this means that whenever any central bank pegs an exchange rate, then the exchange rate is definitionally misaligned. I suspect when people use this particular definition, it is usually conjoined with some sort of threshold. One problem in my mind with operationalizing this definition is figuring out what that the “threshold” is.

A more fundamental conception of what an “equilibrium” exchange rate is is laid out by my colleague Charles Engel. As noted in this post, the equilibrium exchange rate is the one that minimizes the distortion from sticky prices and other rigidities. That is not necessarily the exchange rate delivered by a free float, even in the absence of capital controls. Indeed, it might be best delivered by some type of monetary policy (see the paper).

This approach departs from the “equilibrium” approach embodied in the real models of the late Alan Stockman, as summarized here, in that those models assumed away nominal rigidities. With complete markets, the free floating exchange rate is almost irrelevant since the real rate will always adjust to the right levels.

Review and Summing Up

Two last observations. First, each of these approaches has advantages and disadvantages. PPP and variants are easy to implement, but are more akin to parity conditions, so it’s not clear over what horizon they apply to. The macroeconomic balance and FEER approaches are most appropriate to the medium term, and hence perhaps more relevant to policy questions. But they require more judgment on the parameters of the models. Moving to the basic balance approach, the time horizon is the shortest, and of perhaps most interest to policy analysts, and yet requires the greatest subjective judgment, since one needs to take a strong stand on what constitutes the appropriate values for critical variables. (There are also data considerations as well — the PPP criteria is most popular in part because of the limited data requirements.)

Second, in some ways, it’s not necessarily correct to think of these approaches as all inconsistent (in some instances they might be). Better to think that some approaches are more appropriate at one horizon versus another horizon. (This echoes Richard Cooper’s description of how elasticities, absorption and monetarist interpretations were all consistent over time, in thinking about the effects of devaluations.) That is why a finding that the RMB is only 10% in value below that predicted by the Penn Effect is not definitive when talking about misalignment in the short run (and it’s why, despite findings of a fairly small misalignment in Cheung et al. (2009), I think a revaluation of the RMB combined with expansionary fiscal policy, and altered regulatory policy, is called for in the context of an IS-LM-BP model. [3])

(Interesting aside: Fan Gang and Yiping Huang have cited Cheung et al. in favor of the no-undervaluation thesis. I thank them for their citation (who doesn’t like publicity?!), but I still think that a finding of small misalignment along one dimension is not conclusive. Other criteria do suggest undervaluation, although I do not think a RMB revaluation is in and of itself sufficient to remedy the imbalance (see Prasad (2010)). See the estimates of trade elasticities cited here.)

I summarize the typology of studies in this table (which is adapted from Cheung et al. (2006).


Table 1 (modified): from Cheung, Chinn and Fujii (2006)).

For a volume on the topic of currency misalignment in developing countries, see Hinkle and Montiel (1999).

A recent survey of currency issues within the TPP debate by Fergusson, McMinimy, and Williams of the Congressional Research Service (March 2015) on pages 50-51. Some more recent discussion of currency misalignment here. PPP applied here.

A survey of TPP economic issues by Williams of the CRS (2013) here, by Petri and Plummer here.

9 thoughts on “Currency Misalignment: A Reprise

  1. don

    “With complete markets, the free floating exchange rate is almost irrelevant since the real rate will always adjust to the right levels.”

    I assume “with complete markets” means the absence of things like capital controls. Japan is again intervening actively in currency markets to depress the value of the yen, so even in fairly open markets, there must be enough friction to allow traction from such policies, or those instituting them are foolishly engaged in a sisyphus labor.
    Currency intervention is a discretionary action. One can argue that reserves are accumulated to prevent future shortfalls, but who believes that is a valid explanation for China’s almost $4 trillion in currency reserves? It seems obvious that the policy was used as a strategy to promote export-led growth – the same strategy used by Japan decades earlier. The only question is the extent to which private capital flows reacted to offset the effects of the official interventions. If capital controls are effective, then the foreign currency purchases are matched with an equal increase in the current account balance. Simple as that.

          1. don

            Effective capital controls? If they are completely effective, the current account will be moved by currency intervention, dollar-for-dollar, ne?

  2. BC

    Nice work.

    Were China to ever permit convertibility and free flow of capital, we would very quickly see just how “overvalued” the Yuan is (but not for long), as US and Japanese firms and Chinese nationals would flee en masse (rather than quietly through the back door via Hong Kong and Singapore as has been occurring for years now), resulting in the great sucking sound of the financial black hole that China has become.

    One of the primary reasons western firms and politicians want Yuan convertibility is so they can repatriate dollars directly out of China where they are no longer profitable and before the system implodes, rather than by way of swaps with the Fed custodial accounts and the PBoC via TBTE banks. Of course, the Chinese know this and are not keen to play along and invite disaster.

    Just as the US and Japan are reducing FDI in China, Europeans and the oil emirates are increasing FDI much too late in the day.

    China is a four-letter word: SELL (if you can get your money out, that is).

  3. PeakTrader

    Large current account surpluses are expected when private saving surpluses are much greater than budget deficits.

    China’s consumer spending declined from 45% of GDP in 2000 to 36% of GDP in 2010 (compared to about 70% in the U.S. and about 60% in Western Europe). However, it seems to have roughly stabilized, since 2010, while China’s GDP continued to grow, although at a slowing pace.

    And, “the average Chinese household saves as much as 40 percent of its income” (compared to around 5% in the U.S.).

    It seems, China’s budget deficit is around 2% of GDP (it’s questionable if all of China’s statistics are accurate).

  4. Tom

    It’s nice to see you taking on a topic for a purpose other than scoring points for the Ds. But I think you need to spend more time clarifying your thinking in prose before you march into the math.

    The starting point with this topic should always be to clearly define your terms. Valuation is inherently subjective, so anybody can define overvaluation and undervaluation by any standard they choose. To understand any argument about mis-valuation of currency, I need to know exactly how you’re defining it. You do that but in a way that’s aimed towards a narrow professional audience not lay blog readers. You could be much more relevant to a much wider audience if you wrote more and clearer prose. After all, the arguments that you’re up against mostly fall apart at the stage where they should define their terms and don’t.

    Most lay people have no idea how purchasing-power parity is calculated. A well-written explanation of how the process works and the issues and controversies involved could be very helpful and interesting. The relationship between relative income levels across countries and relative prices of similar goods can be looked at from many angles. Most obviously as local productivity rises, so with it does the cost of land and even the least productive locally available labor – but depending of course on supply and such factors as immigration policy. So at least usually and roughly does the tax burden and the size of government relative to GDP, which absorbs some labor supply. So there are lots of reasons for the “Penn effect,” which I think boils down to an axiom that real productivity improvements in part of the economy relative to the rest of the world also increase relative prices in parts of the economy where there has been no real productivity improvement relative to the rest of the world. A lot of PPP turns out to be measuring the extent to which relative real productivity between two countries is amplified by the Penn effect in their relative nominal prices.

    But that’s not the only reason for large price differentials between low-income and high-income countries. There is also the matter of the quality of governance. GDP data necessarily values government at cost. Most people would laugh if they knew how PPP methodology measures the relative real productivity of governments.
    Buut even libertarians such as Tyler Cowen recognize that quality government has great real value, and it would be pretty hard to argue that higher income, higher price countries aren’t so at least partly because they’re governed better. When you buy a Big Mac in Switzerland or Sweden you don’t just pay for a Big Mac, you pay to have it served to you in the political environment of Switzerland or Sweden.

    The inherent difficulty of comparing prices across nations is another rich topic. Goods across nations are never exactly the same, and objective comparison is especially problematic among advanced economies where most value is in services. Take again the Big Mac index. Many forget that McDonalds is not a grocery store selling merchandise, it’s a restaurant selling food service. Part of the value in its products is the restaurant experience, albeit a bare-bones one. Outside North America, McDonalds primarily serves tourists, long-haul drivers and suburban commuters with relatively attractive, well-maintained restaurants located in major city centers and along busy roads. In North America such types of restaurants are a small minority of locations, and there are many times more not so well maintained restaurants that cater primarily to the low income and others who choose their restaurants mainly by lowest price.

  5. Tom

    PS People who speak of “currency manipulation” also need to be challenged to define the term. I personally can’t see how it would be possible for any state NOT to manipulate its currency unless it did away with monetary policy, and I suppose even then its private banks would be accused.

    Obviously two of the main determinants of relative currency pricing are relative interest rates and money supplies, the core targeted parameters of monetary policy. It’s impossible to conduct monetary policy without also manipulating currency valuation.

Comments are closed.