What is it with economists and accounting identities?

It's very strange.  There seems to be something about accounting identities that causes otherwise reasonable economists -- pardon my bluntness -- to become either stupid or duplicitous.  The most famous example is probably Eugene Fama's painfully embarrassing blog post in which he seriously claimed that the national income accounting identity proved that stimulus spending could not possibly work.  (Aside 1)  But now Mark Thoma points us to what may be a new classic of the genre, written by Thomas Klitgaard and David Lucca  and published on the New York Fed's website.

The subject of Klitgaard and Lucca's piece is whether large-scale central bank asset purchase programs, like that announced by the European Central Bank this January, are likely to cause financial outflows.  This sounds like a worthwhile topic on which the New York Fed's Research and Statistics Group might have something valuable to contribute.

But a warning sign appears in the very first paragraph, where the authors allude to arguments which "ignore balance of payments accounting."  Uh oh.

There follows a brief tutorial on balance-of-payments accounting, explaining (correctly, of course) that a country's current account (its exports and imports of goods and services) is balanced by its financial account (its exports and imports of financial assets).  This is an accounting identity, true by definition.  So far, so good.

Then follows this paragraph:

Suppose for the moment that the ECB’s asset purchase program, by driving down interest rates, causes investors to consider investing abroad. Also assume that the euro area’s current account balance remains unaffected over the near term by the monetary policy shift. Then net financial outflows must be unchanged and any increase in the pace of domestic purchases of foreign assets can only be realized if foreign investors match that increase by buying more euro area assets. In other words, the desire to invest abroad may be there, but financial outflows are constrained by the current account and financial inflows. That means that the exchange rate and other asset prices need to move in response to the ECB policy change to keep financial outflows consistent with balance of payments identities.  (emphasis added)

Let's start with this:  "Also assume that the euro area's current account balance remains unaffected...." Now, the authors have just established that, by definition, a change in the balance of the financial account is a change in the balance of the current account.  They are linked not by some causal relationship, but by an accounting identity: they are the same thing.  Consequently, an assumption that the euro area's current account remains unchanged is exactly the same thing as an assumption that the financial account remains unchanged.

In other words, the authors have done nothing less than smuggle their conclusion in as an assumption. Did they fail to notice that they were assuming their conclusion?  Or did they simply rely on their readers' failure to notice the sleight-of-hand?

You might be inclined to guess "duplicity" rather than "stupidity," but the remainder of the paragraph is sufficiently muddled to suggest that the authors may actually not understand the accounting identity they're relying on.

For instance, they say this: "the desire to invest abroad may be there, but financial outflows are constrained by the current account and financial inflows."

Here's a rule of thumb for talking about accounting identities: Accounting identities do not constrain behavior; they constrain accounting.  If you find yourself saying or implying that an economic actor cannot do something they want to do because of an accounting identity, you have lost the thread.  Backtrack and rethink.

The authors also say: "exchange rate and other asset prices need to move in response to the ECB policy change to keep financial outflows consistent with balance of payments identities."

Here's a second rule of thumb for talking about accounting identities: Accounting identities are not enforced by mechanisms (like price changes); they are enforced by accounting.  If you find yourself saying or implying that accounting identities are preserved by some process of adjustment, you have lost the thread.  Backtrack and rethink.

So how does the accounting actually work in this case?

There are two key definitions to keep in mind.  First, exchanges of financial assets for other financial assets are not net financial flows.  They net to zero.  Second, money is a financial asset, just like any other (e.g. bonds and promissory notes).

Suppose, as Klitgaard and Lucca suggest, that I am in France, that I have a million euros handy, and that I wish to invest abroad (say, in U.S. Treasury bonds).  The first thing I do is spend my €1.0 million to buy (say) $1.2 million.  Next, I spend this $1.2 million to buy $1.2 million worth of Treasury bonds.  Notice that both of these transaction are simple exchanges of financial assets. Under the rules of the balance-of-payments accounting game, neither of these transactions constitutes a net financial flow.

So, I can invest my €1.0 million in U.S. Treasury bonds, and yet there is no "financial outflow" from France to the U.S.  How can this be?  Simple:  "financial outflow" in balance-of-payments accounting doesn't mean what you probably thought it did.  (Aside 2)

Notice also that, although I assumed that exchange rates and asset prices exist, the zero-net-flow outcome doesn't depend in any way on what they are or how they change.  That's how accounting identities work: they are preserved in all states of the world, simply by the rules for how individual transactions are written down.

What's really puzzling is that Klitgaard and Lucca's real point doesn't depend on accounting identities at all.  Their piece, read as a whole, seems to be making an unobjectionable argument:  currency markets move faster than bond markets, so exchange rates (rather than changes in portfolio composition) should be expected to do the lion's share of adjusting to the new central bank policy; and so it seems to have happened.  Why in the world do they get tangled up with accounting identities at all, and why do they go so completely off they rails when they do?

I can only guess about the reason for bringing up the accounting in the first place, but accusing one's rhetorical opponents of failing to "get" basic accounting identities and adding-up constraints does seem to be a pretty standard trope in international economics. (Aside 3)

The deeper cause of the problem, I suspect, has to do with economists' ingrained professional habit of "thinking in equilibria."  Accounting identities look a lot like equilibrium conditions, and one of the first thing an economist asks (or should ask!) about a posited equilibrium is, "How does the economy get there, and once it is there, what keeps it there?"  That is, the very idea that a system has an equilibrium implies that it is logically possible for the system to exist in some non-equilibrium state, and for the equilibrium to be meaningful, it must be accompanied by some dynamic mechanism that moves the system towards it equilibrium point.

But accounting identities are nothing like equilibria.  Each and every logically conceivable transaction preserves the accounting identities.  The "system" is never "out of equilibrium" and there is no "mechanism" for pushing it towards some "equilibrium" state.  As a result, reasoning about accounting identities with a mind pre-disposed to "think in equilibria" seems to make the cognitive equipment go a little haywire.







Aside 1:  I stopped reading Gene's blog (much to my own disappointment) at this point.  Why?  At the time, I assumed that he could not possibly be stupid enough to make such an elementary error, so I believed that he must have been using his blog as a platform for lending his professional prestige to politically useful "soundbite economics" in which he did not personally believe.  I was disappointed, because Gene is smart and interesting, and I had been looking forward to learning a lot from him.

Aside 2:  What is a financial outflow?  A financial outflow occurs when a country imports financial assets in exchange for exporting goods and/or services.  This is also known as selling stuff. Exporting financial assets in exchange for imports of goods and/or services (sometimes called buying stuff) is a financial inflow.  The connection between the financial account and the current account suddenly seems a lot less mystical, doesn't it?  The only really counter-intuitive bit is the way that "paying with money" and "paying with debt instruments" are the same thing for these purposes.

Aside 3: Paul Krugman, for example, employed the same trope twenty years ago (see the section on Investment and Trade Balance).  He also segues rather breezily from talking about the accounting identity to how things happen "in practice," by which he means, "forget about the accounting identity, let's talk about the path between two equilibria."