Tuesday, January 28, 2014

The Mystery of Economic Identities

Dean Baker has a takedown of Robert Z. Lawrence on his website today.  RZL said that reduced US fossil fuel exports can’t have an effect on the current account, since the current account is determined by net national savings, and changes in energy sources don’t affect the determinants of net savings.  Dean showed how, step by step, changes in the destination of US purchases can alter savings.  He also showed how the value of the dollar doesn’t have to adjust to maintain a constant current account position.

I hate to take issue with Dean, a much nicer guy than his public pit bull profile would suggest, but both sides of this dispute are mired in confusion.  At issue is one version of the macroeconomic net savings and current account identity:

(S – I) + (T – G) ≡ CA

where S is a country’s total savings, I is investment, T tax revenue, G government purchases, and CA is the current account balance.  (I prefer the financial balances version of this, but they are equivalent.)

What RZL did was to say that changes in spending decisions can’t change the CA because it’s determined by decisions to save, invest, spend public money and tax, and these haven’t changed.  What Dean did was to say that, au contraire, changes in spending decisions that redirect money to domestic producers can create more domestic income, which alters S and T, thereby altering CA.

What I object to is treating this identity like an equation.  What’s the difference between = and ≡?  The thing on the left side of = is presented as equal in measure to the thing on the right side.  The left can determine the right or the right the left, or perhaps the equation fails and they end up not being equal this time.  When you see an ≡, however, the thing on the left is the same thing as the thing on the right.  One does not cause they other; they are two different ways of expressing one identical entity.

The ultimate basis for macroeconomic identities lies in individual identities: a purchase is identically a sale, non-consumption of income is identically saving, and a credit asset is identically a debt liability.  One doesn’t cause they other, and they can’t add up to different quantities, ever, even for a really, really short period of time.

The problem with Dean’s answer, then, is that it is couched in causation and sequence.  A happens, which then causes B and C.  This opens the door to needless disputes over what causes what and what happens first.  Of course, there are causal mechanisms at work, but they are operating in all directions and simultaneously.  And measured net saving is the consequence of all of them taken together, as is the current account balance—which is simply net saving using a different set of measurement categories.

Meanwhile, regarding the possibly offsetting role of exchange rates in response to expenditure switching, RZL is pledging allegiance to the venerable specie flow mechanism, according to which changes in the trade balance automatically engender a change in relative prices that restores the initial trade balance.  It didn’t work under the gold standard, however, and it doesn’t work under floating exchange rates either.  This is an empirical reality, universally agreed to by those who study such things.  One often gets the impression with RZL that the policy conclusions constitute the fixed point, and arguments are adduced as needed to support them.

UPDATE: OK, I was flip in the last paragraph.  Species flow is an internal adjustment, and offsetting exchange rate movements are an external adjustment.  You could believe in one and not the other.  The fact that they are both pollyana-ish is not material in analytical terms.  Nevertheless, it remains the case that there isn't an empirical tendency for exchange rate adjustments to automatically offset CA imbalances.

1 comment:

Anonymous said...

We have the identity,

(S – I) + (T – G) ≡ CA

At time T0 we can write that as

(S0 – I0) + (T0 – G0) ≡ CA0

and at time T1 we can write

(S1 – I1) + (T1 – G1) ≡ CA1

We can also write

(S1 – I1) + (T1 – G1) - (S0 – I0) + (T0 – G0) ≡ CA1 - CA0

or

(ΔS – ΔI) + (ΔT – ΔG) ≡ ΔCA

Now, apparently Lawrence claims first that the domestic US oil "boom" will have little effect on S and I. That is, ΔS ≃ 0, ΔI ≃ 0, and ΔS - ΔI ≃ 0. Second, he claims that therefore ΔCA ≃ 0.

Even granting his first claim, his second claim is silly.

Dean Baker disputes his first claim. As for invoking causality to show how S, I, T and G may change, Baker does not use the identity to do so.