what adjusts?

Thoughts on International Finance, with Application to the US and China

(I wrote this back in 2020, and never posted it. The context is different now, but the substance still seems valid.)

Here is my mental model, for whatever it’s worth:

(1) The US-China trade balance is determined in the short to medium run by relative income growth in the two countries. In the medium to long run relative prices do play a role. But at least past the early stages of industrialization, the impact of exchange rates is thru producer entry/exit than thru expenditure switching. The impact of the overvalued dollar of the early 80s came mainly through e.g. the bankruptcy of US steel producers selling at world prices, rather than a loss of market share from selling at US prices.

(2) Chinese capital controls limit cross-border financial flows. This especially limits the acquisition of foreign assets (including real estate, consequentially for New York) by Chinese firms and households. This implies greater net inward financial flows than there would be in absence of controls. This is probably the most important Chinese policy with respect to cross-border flows — a broad liberalization would be more likely to push the renminbi down than up.

(3) The Chinese central bank passively accumulates/decumulates whatever level of reserves are implied by the combination of 1 and 2.

(4) The exchange rate is either chosen by one or both governments or determined in speculative markets. (In practice this means the Chinese government, but there’s no in-principle reason why this has to be so.) There is no meaningful link from the trade balance to the exchange rate, and at most a weak link from the exchange rate to the trade balance. Exchange rate interventions are not an independent factor in reserve changes.

(5) The interest rate on US Treasury debt is determined by some mix of Fed policy and self-confirming market expectations (convention). Chinese reserve purchases play zero role.

(6) US deficit spending is not constrained, required, or influenced in any way by foreign reserve accumulation. When desired foreign reserve accumulation departs from new Treasury issues, the gap is accommodated by net sales between foreign central banks and the private sector.

(7) If a mismatch between the supply of Treasury issues and the demand for reserve accumulation creates pressure anywhere, it will be on private assets that are close substitutes for Treasuries. In particular, it is plausible that insufficient federal borrowing in 1990s-2000s helped create the mortgage securitization market.

(8) Returning to exchange rates. The fact that import price elasticities are low, and that most trade is priced in dollars, means that exchange rates affect trade mainly via exporters’ profit margins. An appreciation can undermine exports, but this is a slow process of failure/exit by exporters, and thus strongly depends on financial capacity of exporters to operate with diminished margins. So for instance the large, roughly symmetrical movements in the dollar-yen exchange rate in the first and second halves of the 1980s affected the US tradable sector more than the Japanese, because Japan’s bank-based financial system plus the lack of shareholder pressure made it easier to sustain losses for extended period there than in the US.

In the textbooks, we get a picture of a tightly articulated system where a change in behavior in one place must lead to an exactly offsetting change somewhere else, mediated by price changes. Given a set of fundamental parameters, there is only one possible equilibrium. The considerations above suggest a different vision.

In the orthodox vision, international trade and financial flows are like a pool of water. If you drop a rock in, the whole surface of the pool rises by the same amount. Of course there are passing ripples. But knowing what level this part of the pool was at a while go doesn’t tell you anything about what level it is at now. One could, though, just as easily imagine a pile of rocks. When you move one rock, it normally affects only the rocks in the immediate vicinity. And the same rocks can be piled up in many different ways; where they are now depends on where they were before.

From where I’m sitting, there are three major sources of flexibility in the international system, all of which undermine any claim that shift in one flow must lead to equivalent shift in some other flow.

First is the existence of passive, accommodating positions that act as buffers. Central bank reserves can function this way; this is accepted in mainstream theory. But so can bank loans and deposits, and positions taken by fx specialists. In the short run, bank deposits are always accommodating buffers for any other flow.

Second is speculative price dynamics that make asset demand endogenous to current price. Concretely: If an asset is held largely in hope of capital gains, as opposed to yield or use in production, and if there are anchored expectations of normal or long-run price x, then any position that produces a price move away x implies capital gains for anyone who takes the other side of the position. In markets where these kinds of speculative dynamics operate – and I think they operate very widely – then even large changes in flows don’t have to lead to significant price adjustments. (Conversely, shifting expectations can lead to large price changes without any shift in flows.)

Third is the fact that trade adjustment happens mainly thru entry/exit rather than expenditure switching in product markets. This means in effect that the balance sheets of exporting firms act as shock absorbers. Let’s say that a country’s financial assets become more desirable to global wealth owners, causing a financial inflow and (plausibly though not necessarily) an appreciation of its currency. In the textbook story, this leads to an equal and immediate fall in net exports. But in reality, with exports priced in global markets, the immediate effect is a fall in the profitability of exporters. Only over time, as those firms go bankrupt or give up on export markets, will trade volumes change.

How to Think about the Balance of Payments: The US Position 2012-2013

In the previous post, I suggested that we should think of the various trade and financial flows in the balance of payments as evolving more or less independently, with imbalances between them normally accommodated by passive buffers rather than being closed by any kind of price adjustment. In that post I focused on the prewar gold standard. Here is a more recent example of what I’m talking about.

From 2012 to 2013 there was a general “risk on” shift in financial markets, with fears of a new crisis receding and investors focusing more on yield and less on safety and liquidity. In a risk-off environment investors prefer the safety of US assets even if yields are very low; in a risk-on environment, as we were moving toward in 2013, they prefer higher-yielding non-US assets.

Now, how was this shift in asset demand accommodated in the balance of payments? Orthodox theory suggests that there should be some offsetting change in interest rates and/or exchange rate expectations to keep demand for US and non-US assets balanced. But this didn’t happen — interest rate differentials didn’t close, and “risk-on” is associated with a falling rather than a rising dollar. And in fact, there was a large net outflow of portfolio investment: Net acquisition of foreign assets was $250 billion higher in 2013 than 2012, and net foreign acquisition of US assets by foreigners was $250 billion lower. Orthodox theory also says that if there is a net shift in investment flows, there should be an offsetting change in the current account. But the US current account shifted only $60 billion toward surplus, compared with the $500 billion net shift in portfolio flows. In a country with a fixed exchange rate, we would expect the remaining portfolio outflow to be accommodated by a fall in foreign exchange reserves. but of course the dollar floats, and the Fed does not hold significant reserves.

In fact, the entire shift was accommodated within the US banking system, most importantly by a rise in foreign-held deposits of $400 billion. Now this is an increase in US foreign liabilities, but it does not reflect a decision by anyone to borrow from abroad. It simply reflects the mechanics of international financial transactions. When an American spends money to purchase a foreign asset, the “money” they are using is a deposit at an American bank. When the asset is purchased, that deposit is transferred to the foreign asset-seller (or some intermediary), turning the deposit into a foreign liability of the bank. So the shift of portfolio investment out of the US does not require any change in prices (or incomes) to generate an offsetting flow into the US. The foreign liabilities that finance the purchase of foreign assets are generated mechanically in the course of the transaction itself.

Eventually, the effort to close out this residual long dollar position might produce downward pressure on the value of the dollar. And if the dollar does depreciate, that may increase demand for other US assets or for US exports sufficient to absorb the deposits. But there is no guarantee that either of these things will happen. And certainly they will not happen quickly. What we know for sure is that buffering within the banking system can offset quite large flows for substantial periods of time — in this case, a shift in portfolio flows of a couple percent of GDP sustained over a year. It might be that, with sufficient time, net sales of US assets might be large enough to push their price down, raising the yield enough to compensate for the lower safety premium. Or it might be that the downward pressure on the dollar will eventually lead to a big enough depreciation to raise US net exports enough to balance the portfolio outflow — but this will be a very long process, if it happens at all. It’s quite likely the portfolio will reverse for its own reasons (like a shift back toward “risk off”) before these adjustments even get started. Alternatively, liquidity constraints within the banking system may exhaust its buffering capacity before any other adjustment mechanism comes into play, requiring active intervention by the state or a catastrophic adjustment of the current account. (Presumably not in the case of the US, but often enough elsewhere.)

In practice, where we see payments balance maintained smoothly, it’s more likely because the underlying patterns of trade and investment are balanced and stable enough to not strain the buffering capacity of the banking system, rather than thanks to the operation of any adjustment mechanism.

How to Think about the Balance of Payments

There are many payments between countries — trade in goods and services, profits and interest paid to foreign capitalists, portfolio investment, FDI and bank lending, transactions between governments. All of these payments must balance out one way or another.

International-finance orthodoxy since David Hume has been about identifying an automatic mechanism that ensures that all these flows balance. This mechanism should take the form of a price adjustment, whether of the price level, the exchange rate, or the interest rate.

An alternative Keynesian approach is to make aggregate income the adjusting variable that maintains the balance of payments equality, just as it is in maintaining the domestic savings-investment balance. This is the idea behind balance of payments constrained growth.

Balance of payments constrained growth is certainly an improvement on the price adjustment mechanisms of orthodoxy. But I think it would be even better to consider both as items on a menu of things that may happen when a payments imbalance develops. The beginning of wisdom here is to recognize that there is no general mechanism that maintains payments balance. Changes in relative prices, exchange rates, interest rates or incomes may all play a role, depending on the timeframe we are considering and on the countries involved and the source of the imbalance.

Our theory of balance of payments adjustments should not begin with the universal logic of either orthodox or b.o.p.-constrained growth models, but with a concrete historical enumeration of the various sources of payments imbalance and the various kinds of adjustment in response to them.

We also need to consider other kinds of adjustment mechanisms, in particular, accommodation by buffers. This will always be the dominant mechanism if we are considering a short enough period. In the first instance, payments balance is maintained because there are some actors in the system who will passively take the other side of any open foreign exchange positions. The familiar example of this is a central bank that holds foreign exchange reserves: When it intervenes in the foreign exchange market, it passively allows its reserve position to adjust to accommodate whatever net demand there is for foreign currency. But there are also private buffers. In particular, there’s not nearly enough recognition of the special role of banks in the payments system, which requires them to take open foreign exchange positions when other units engage in cross-border transactions. An inflow of foreign investment, for instance, will in the first instance always result in a an increase in foreign assets in the banking system of the receiving country and foreign liabilities in the banking system of the investing country. How large are the imbalances that can be buffered in this way, and how long the banking system will passively maintain its open position without some other adjustment mechanism coming into play, are open questions. But there is no question that in the short run, the balance of payments is maintained through this sort of passive buffering, and not through any adjustment of either prices or incomes.

We also need to recognize the role of active policy in maintaining payments balance. We tend to think of policy “interventions” as modifications or “shocks” to an underlying structure of payments, but official actions may be an important adjustment mechanism by which that structure is maintained in the first place. This includes both bilateral or multilateral actions that generate offsetting official financial flows in the face of imbalances (important even in the19th century, in the form of central bank cooperation) and unilateral actions to limit outflows, including capital controls, import restrictions and so on.

The right starting point, I think, is to think of the various financial and trade flows as evolving essentially independently. If they happen to more or less balance, then the available buffers and whatever limited price adjustment is possible will be enough to maintain balance. If they don’t happen to balance, then the expected outcome is a crisis of some sort, ending with state intervention and/or a change in the “fundamental” parameters. There is no automatic mechanism that maintains balance. Where we see smooth payments balance over a long period time, it is probably because international payments are being actively managed by the authorities, or because productive capacities, import demands, asset preferences of foreign investors and so on have evolved to fit the existing pattern of payments, rather than vice versa.

The classic case is the London-centered gold standard system of the 19th century. Despite what someone like Barry Eichengreen will tell you, price flexibility was not an important element in the stability of this system. While prices and wages did rise and especially fall more freely before World War One, they almost always did so in parallel across trading partners, not in the opposite way that would offset trade imbalances. Instead, the system depended on the following institutionally specific features.

1. A large fraction of non-British savings, especially from Latin America and other less-developed countries, were held in London. This meant that many “international” payments simply involved a transfer from one British bank account to another, with no cross-border settlement required.

2. British foreign investment primarily funded purchases of British capital goods, so that financial outflows and exports naturally rose and fell together without the need for price adjustments.

3. The capital goods so purchased (for railroads especially) were largely used to produce exports to Britain, offsetting interest and divided payments back to London.

4. Slower growth in Britain was associated with lower interest rates there. So the slowdown in import payments abroad (due to lower incomes) was offset by an increase in foreign lending, which was quite interest-sensitive.

5. Within Europe central banks actively cooperated to offset any payments imbalances that did occur. On several occasions where there a net flow of gold from London to Paris seemed to be developing, the Bank of France made large loans to the Bank of England so that no actual gold had to move. In addition, the belief that gold convertibility would be maintained, or if suspended soon restored at the old parity, meant if a payments imbalance led to a deviation of the market exchange rate from the official parity, it would generate large speculative flows toward the depreciated currency.

6. Outside of Europe, crises and defaults were integral to the operation of the system. While interest-sensitive foreign lending meant that for England (and to some extent other European countries, and later the US), imports and financial outflows tended to move in opposite directions, higher interest rates could not reliably generate financial inflow for peripheral countries. Instead, the normal adjustment process for large imbalances was a catastrophic one in which large deficits periodically led to suspension of convertibility and default.

7. Over the longer run, the “fundamentals” in the periphery were shaped to produce payments balance at prevailing prices, rather than prices adjusting to fundamentals. Foreign investment financed development of export industries suiting the needs of the investing country, with higher-wage countries specializing in higher-value products. In settler colonies, migrant flows strengthened trade and financial links with the mother country.

Bottom line: there was no adjustment mechanism. Stability depended on the contingent fact that the prevailing “shocks” had roughly balanced effects on payment flows. Small imbalances were absorbed by buffers (which in the pre-WWI system included the cost of transporting gold). Large imbalances were actively managed or else led to the system breaking down, either locally, or globally as with the war.

For the gold standard era, I think the best statement of this perspective is Triffin’s “Myths and Realities of the So-Called ‘Gold Standard’.” Alec Ford’s The Gold Standard 1880-1914: Britain and Argentina is also very good (as is Barry Eichengreen’s discussion of it.) Peter Temin makes essentially this argument in his Lessons from the Great Depression — that the gold standard worked before World War I but broke down in the 1920s not because prices were more flexible before the war, but because in the prewar period it did not have to deal with big imbalances in trade and financial flows as developed after. Keynes makes the same larger point, as well as all seven of the specific points above, but at scattered places in his writing and correspondence rather than — as far as I know — in any single text. This perspective is in the same spirit as the “surplus recycling mechanism” that Varoufakis talks about in The Global Minotaur and elsewhere, the idea that there is no price mechanism that tends to bring about payments balance and so some specific institution is needed to offset persistent surpluses and deficits. (Though of course Varoufakis is focused on the more recent period.) The point that productive capacities are shaped by relative prices, rather than vice versa, was made by development economists like Arthur Lewis — it’s stated very clearly in his Evolution of the World Economic Order.

Obviously, the specifics will be different today. But I think the same basic perspective on the balance of payments still applies. Where payments balance exists, it is because of institutional factors that tend to generate offsetting disturbances to trade and financial flows, and because the international structure of production has evolved to generate balance at existing relative prices, rather than because prices have adjusted. And when imbalances do develop, they are accommodated first by passive buffers, and then either actively managed by authorities or else produce a breakdown in the system.

Note: I wrote most of this post in February 2015 and then for some reason never put it up. It really should have links, but given that it’s already sat around for over year I decided to just put it up as-is. Since the original post was very long, I’ve split it into two parts. The second half is here.

What Has Happened to Trade Balances in Europe?

It has gradually entered our awareness that the Greek trade account is now balanced. Greece no longer depends on financial markets (or official transfers, or remittances from workers abroad) to finance its imports. This is obviously important for negotiations with the “institutions,” or at least it ought to be.

I was wondering, how general is this shift toward a positive trade balance. In the FT last week, Martin Wolf pointed out that over the past five years, the Euro area as a whole has shifted from modest trade deficits to substantial trade surpluses, equal to 3 percent of euro-area GDP in 2013. He does not break it down by country, though. I decided to do that.

Euro area trade ratios, 2008 and 2013. The size of the dots is proportional to total 2008 trade.

Here, from Eurostat, are the export-import ratios for the euro countries in 2008 and 2013. Values greater than one on the horizontal axis represent a trade surplus in 2008; only a few northern European countries fall in that group. Meanwhile, in seven countries imports exceeded exports by 10 percent or more. By 2013, the large majority of the euro area is in surplus, while not a single country has an excess of imports over exports of more than 5 percent. The distance above the diagonal line indicates the improvement from 2008 to 2013; this is positive for every euro-area country except Austria, Finland and Luxembourg, and the biggest improvements are in the countries with the worst ratios in 2008. The surplus countries, apart from Finland, more or less maintained their surpluses; but the deficit countries all more or less eliminated their deficits.

So does this mean that austerity works? Yes and no. It is certainly true that Europe’s deficit countries have all achieved positive trade balances in the past few years, even including countries like Greece whose trade deficits long predated the euro. On the other hand, it’s also almost certainly true that this has more to do with the falls in domestic demand rather than any increase in competitiveness.

This is shown in the second figure, which gives the ratio of 2013 imports to 2008 exports on the vertical axis, and 2013 exports to 2008 imports on the horizontal axis. (This is in nominal euros.) Here a point on the diagonal line equals and equal growth rate of imports and exports. Most countries are clustered around 15% growth in imports and exports; these are the countries that had balanced trade or surpluses in 2008, and whose trade ratios have not changed much in the past five years. Only one country, Estonia, has export growth substantially above the European average. But all the former deficit countries have import growth much lower than average. (As indicated by their position to the left of the main cluster.) It’s evident from this diagram that the move toward balanced trade in the deficit countries is about throttling back imports, not boosting exports. This suggests that it has more to do with slow income growth than with lower costs.

Again, the sizes of the dots are proportional to 2008 trade volumes.

Still, the fact remains, trade deficits have almost been eliminated in the euro area. Liberal critics of the European establishment often say “not every country in Europe can be a net exporter” as if that were a truism. But it’s not even true, not in principle and evidently not in practice. It turns out it is quite possible for every country in the euro to run a trade surplus.

The next question is, with whom has the euro area’s trade balanced improved? Europe outside the euro, to begin with. The country with the biggest single increase in net imports from the euro zone is, surprisingly, Switzerland, whose deficit with the euro area has increased by close to 60 billion. Switzerland’s annual trade deficit with the euro area is now 75 billion, about a quarter of the area’s overall trade surplus. Norway and Turkey have increased their deficits by about 15 billion each. The rest of the increase in net exports are accounted for by increased surpluses with Africa (26 billion), the US (27 billion), and Latin America (35 billion, about half to Brazil), and a decreased deficit with Asia (135 billion, including a 55 billion smaller deficit with China, 30 billion smaller with Japan and 20 billion with Korea). Net exports to Australia have also increased by 10 billion.

Why do I bring this up? One, I haven’t seen it discussed much and it is interesting.

But more importantly, the lesson of the Europe-wide shift toward trade surpluses is that austerity can succeed on its own terms. I think there’s a tendency for liberal critics of austerity to assume that the people on the other side are just confused, or blinkered by ideology, and that there’s something incoherent or self-contradictory about competitiveness as a Europe-wide organizing principle. There’s a hope, I think, that economic logic will eventually compel policymakers to do what’s right for everyone. Personally, I don’t think that the masters of the euro care too much about the outcome of the struggle for competitiveness; it’s the struggle itself — and the constraints it imposes on public and private choices — that matters. But insofar as the test of the success of austerity is the trade balance, I suspect austerity can succeed indefinitely.

UPDATE: In comments Kostas Kalaveras points to a report from the European Commission that includes a similar breakdown of changes in trade balances across the euro area. There’s some useful data in there but the interpretation is that almost all the adjustment has been structural rather than cyclical. This is based on estimates of declining potential output in the periphery that I think are insane. But it’s interesting to see how official Europe thinks about this stuff.

Three Ways of Looking at alpha = r k

Piketty’s “first law of capitalism” is the accounting identity

α = r k

where α is the share of capital income in total output, r is the average return on capital, and k is the aggregate capital-output ratio.

As accounting, this is true by definition. As economics, what kind of economic behavior does it describe? There are three ways of looking at it.

In the standard version, the profit share is determined by a production function, which is given by technology. The profit rate r* required by capital owners is fixed by technology in combination with time preferences. In this closure, k is the endogenous, or adjusting, variable. Investment rises or falls whenever the realized profit rate differs from the required rate, thus keeping k at the level that satisfies the equation for r = r*.

In Piketty’s version, r is fixed (somehow; the mechanism is not clear) and k is determined by savings behavior and (exogenous) growth according to his “second law of capitalism”:

k = s/g

That leaves α to passively accommodate r and k. Capitalists get whatever the current capital stock and fixed profit rate entitle them to, and workers get whatever is left over; in effect, workers are the residual claimants in Piketty’s system. (This is the opposite of the classical view, in which wages are fixed and capitalists get the residual.)

In a third interpretation, we could say that α and r are set institutionally — α through some kind of bargaining process, or by the degree of monopoly, r perhaps by the interest rate set in the financial system. The value of the capital stock is then given by capitalizing the flow of profits α Y at the discount rate r. (Y is total output.) This interpretation is the natural one if we think of “capital” as a claim to a share of the surplus as opposed to physical means of production.

This interpretation clearly applies to pure land, or to the market value of a particular firm. What if it applied to capital in general? Since claims on the surplus — including claims exercised through nonproduced assets like land — are not created by reserving output from consumption, aggregate savings would be a meaningless accounting construct in this case. (Or we could adopt a Hicksian view of saving in which it equals the change in net wealth by definition.) Looking at things this way also puts r > g in a different light. Suppose we think of the capital stock as a whole as something like the stock of a firm, which entitles the owners to the flow of profits from that firm. If the profits today are α Y and output is expected to grow at a rate g, what is the value of the stock today? If we discount future profits at r, then it is the sum from t=0 to t=infinity of α Y (1 + g)^t / (1 + r)^t, which works out to α Y / (r – g). So if we can take the rate of return on capital as the discount rate on future profits, then r > g is implied by a finite value of the capital stock.

We shouldn’t ask what capital “really” is. It really is a quantity of money in a process of self-expansion, and it really is a mass of means of production, and it really is authority over the production process. But the particular historical questions Piketty is interested in may be better suited to thinking of capital as a claim on the social surplus than as a physical quantity of means of production. Seth Ackerman has some very interesting thoughts along these lines in his contribution to the Jacobin symposium on the book.

More teaching: We’re starting on the open economy now. Exchange rates, trade, international finance, the balance of payments. So one of the first things you have to explain, is the definition of real and nominal exchange rates:

e_R = e_N P*/P

where P and P* are the home and foreign price levels respectively, and the exchange rate e is defined as the price of foreign exchange (so an appreciation means that e falls and a depreciation means that it rises).

This is a useful definition to know — though of course it’s not as straightforward as it seems, since as we’ve discussed before there are various possibles Ps, and once we are dealing with more than two countries we have to decide how to weight them, with different statistical agencies using different weightings. But set all that aside. What I want to talk about now, is what a nice little example this equation offers of a structuralist perspective on the economy.

As given above, the equation is an accounting identity. It’s always exactly true, simply because that’s how we’ve defined the real exchange rate. As an accounting identity, it doesn’t in itself say anything about causation. But that doesn’t mean it’s vaacuous. After all, we picked this particular definition because we think it is associated with some causal story. [1] The question is, what story? And that’s where things get interesting.

Since we have one equation, we should have one endogenous (or dependent) variable. But which one, depends on the context.

If we are telling a story about exchange rate determination, we might think that the endogenous variable is e_N. If price levels are determined by the evolution of aggregate supply and demand (or the growth of the money stock, if you prefer) in each country, and if arbitrage in the goods market enforces something like Purchasing Power Parity (PPP), then the nominal exchange rate will have to adjust to keep the real price of a comparable basket of goods from diverging across countries.

On the other hand, we might not think PPP holds, at least in the short run, and we might think that the nominal exchange rate cannot adjust freely. (A fixed exchange rate is the obvious reason, but it’s also possible that the forex markets could push the nominal exchange rate to some arbitrary level.) In that case, it’s the real exchange rate that is endogenous, so we can see changes in the price of comparable goods in one country relative to another. This is implicitly the causal structure that people have in mind when they argue that China is pursuing a mercantilist strategy by pegging its nominal exchange rate, that devaluation would improve current account balances in the European periphery, or that the US could benefit from a lower (nominal) dollar. Here the causal story runs from e_N to e_R.

Alternatively, maybe the price level is endogenous. This is less intuitive, but there’s at least one important story where it’s the case. Anti-inflation programs in a number of countries, especially in Latin America, have made use of a fixed exchange rate as a “nominal anchor.” The idea here is that in a small open economy, especially where high inflation has led to widespread use of a foreign currency as the unit of account, the real exchange rate is effectively fixed. So if the nominal exchange rate can also be effectively fixed, then, like it or not, the domestic price level P will have to be fixed as well. Here’s Jeffrey Sachs on the Bolivian stabilization:

The sudden end of a 60,000 percent inflation seems almost miraculous… Thomas Sargent (1986) argued that such a dramatic change in price inflation results from a sudden and drastic change in the public’s expectations of future government policies… I suggest, in distinction to Sargent, that the Bolivian experience highlights a different and far simpler explanation of the very rapid end of hyperinflations. By August 1985,… prices were set either explicitly or implicitly in dollars, with transactions continuing to take place in peso notes, at prices determined by the dollar prices converted at the spot exchange rate. Therefore, by stabilizing the exchange rate, domestic inflation could be made to revert immediately to the US dollar inflation rate.

So here the causal story runs from e_N to P.

In the three cases so far, we implicitly assume that P* is fixed, or at least exogenous. This makes sense; since a single country is much smaller than the world as a whole, we don’t expect anything it does to affect the world price level much. So the last logical possibility, P* as the endogenous variable, might seem to lack a corresponding real world story. But an individual countries is not always so much smaller than the world as a whole, at least not if the individual country is the United States. It’s legitimate to ask whether a change in our price level or exchange rate might not show up as as inflation or deflation elsewhere. This is particularly likely if we are focusing on a bilateral relationship. For instance, it might well be that a devaluation of the dollar relative to the renminbi would simply (or mostly) produce corresponding deflation [2] in China, leaving the real exchange rate unchanged.

Here, of course, we have only one equation. But if we interpret it causally, that is already a model, and the question of “what adjusts?” can be rephrased as the choice between alternative model closures. With multiple-equation models, that choice gets trickier — and it can be tricky enough with one equation.

In my opinion, sensitivity to alternative model closures is at the heart of structuralist economics, and is the great methodological innovation of Keynes. The specific application that defines the General Theory is the model closure that endogenizes aggregate income — the interest rate, which was supposed to equilibrate savings and investment, is pinned down by the supply and demand of liquidity, so total income is what adjusts — but there’s a more general methodological principle. “Thinking like an economist,” that awful phrase, should mean being able to choose among different stories — different model closures — based on the historical context and your own interests. It should mean being able look at a complex social reality and judge which logical relationships represent the aspects of it you’re currently interested in, and which accounting identities are most relevant to the story you want to tell. Or as Keynes put it, economics should be thought of as

a branch of logic, a way of thinking … in terms of models, joined to the art of choosing models which are relevant to the contemporary world. … [The goal is] not to provide a machine, or method of blind manipulation, which will furnish an infallible answer, but to provide ourselves with an organised and orderly method of thinking out particular problems.

Much of mainstream macroeconomics assumes there is a “true” model of the world. Connected to this, there’s an insistence — shared even by a lot of heterodox macro — on regarding some variables as being strictly exogenous and others as strictly endogenous, so that in every story causality runs the same way. In the canonical story, tastes, technology and endowments (one can’t help hearing: by the Creator) are perfectly fixed, and everything else is perfectly adjustable. [3]

Better to follow Keynes, and think about models as more or less useful for clarifying the logic of particular stories.

EDIT: Of course not everyone who recognizes the methodological distinction I’m making here agrees that the eclecticism of structuralism is an advantage. Here is my teacher Peter Skott (with Ben Zipperer):

The `heterodox’ tradition in macroeconomics contains a wide range of models. Kaleckian models treat the utilization rate as an accommodating variable, both in the short and the long run. Goodwin’s celebrated formalization of Marx, by contrast, take the utilization rate as fixed and looks at the interaction between employment and distribution. Distribution is also central to Kaldorian and Robinsonian theories which, like Goodwin, endogenize the profit share and take the utilization rate as structurally determined in the long run but, like the Kaleckians, view short-run variations in utilization as an intrinsic part of the cycle. The differences in these and other areas are important, and this diversity of views on core issues is no cause for celebration.

EDIT 2: Trygve Haavelmo, quoted by Leijonhufvud:

There is no reason why the form of a realistic model (the form of its equations) should be the same under all values of its variables. We must face the fact that the form of the model may have to be regarded as a function of the values of the variables involved. This will usually be the case if the values of some of the variables affect the basic conditions of choice under which the behavior equations in the model are derived.

That’s what I’m talking about. There is no “true” model of the economy. The behavioral relationships change depending where we are in economic space.

Also, Bruce Wilder has a long and characteristically thoughtful comment below. I don’t agree with everything he says — it seems a little too hopeless about the possibility of useful formal analysis even in principle — but it’s very worth reading.

[1] “Accounting identities don’t tell causal stories” is a bit like “correlation doesn’t imply causation.”Both statements are true in principle, but the cases we’re interested in are precisely the cases where we have some reason to believe that it’s not true. And for both statements, the converse does not hold. A causal story that violates accounting identities, or for which there is no corresponding correlation, has a problem.

[2] Or lower real wages, the same thing in this context.

[3] Or you sometimes get a hierarchy of “fast” and “slow” variables, where the fast ones are supposed to fully adjust before the slow ones change at all.