Friday, 3 February 2017

Global Saving and Current Account Imbalances



I've read various stuff recently on the role of savings behaviour in the causes and effects of current account imbalances, some of which I'd say is rather confused.

It has to be said that this is quite a tricky topic, because it has a number of moving parts, various complicating factors and the common problem of lack of clarity over what is being assumed unchanged when we invoke the ceteris paribus assumption.

Here's one way I think about this question.

We start from the sectoral balances identity that says that private sector net acquisition of financial assets (NAFA) is equal to the public sector borrowing requirement (PSBR) plus the current account surplus (CAS).  The PSBR we will assume is a negative function of GDP (Y) and the current account surplus we will assume is a negative function of GDP, a positive function of the GDP of the rest of the world (Y*) and a negative function of the real exchange rate (e)[1].  Thus we have for each national economy:

NAFA = PSBR ( Y ) + CAS ( Y, Y*, e )

So, the first question is what happens if the private sector in country A decides to increase its net saving, i.e. if NAFA rises?  Well, the answer is that it depends.  And what it depends on is a portfolio decision[2].  The portfolio decision here is whether the private sector wants to accumulate domestic public sector assets or foreign assets.

In the first instance, assume that the private sector wants to accumulate domestic government bonds, so that PSBR increases to match the rise in NAFA and the CAS stays unchanged.  In order for the PSBR to rise (within our limited framework here), GDP has to fall.  And for the CAS to stay constant with falling GDP (and constant GDP in other countries), the real exchange rate has to rise.

In the alternative case, the private sector wants to accumulate foreign assets, so the CAS must rise to match the increase in NAFA whilst the PSBR remains unchanged.  As the PSBR is unchanged, GDP is unaffected and the real exchange rate falls to facilitate the rise in the current account balance.  

At this point, we need to turn to the implications of this for other national economies.  To do this, let's assume there are only two countries, so for the other country - country B - the current account surplus is simply the negative of country A's, and the real exchange rate is the inverse of country A's.  Country B has the same sectoral balances equation and we will further assume that country B's NAFA does not change.  Y* for country B is Y for country A and visa versa.

In the first instance, where country A's private sector accumulates only domestic public sector debt, country B's CAS is unchanged, because country A's CAS is unchanged.  Therefore country B's PSBR is also unchanged and so its GDP is unaffected.  Country B experiences a fall in its own exchange rate, but any impact on exports and imports is offset by the change in Country A's GDP.

In the second scenario, country A's GDP stays the same but country B sees a rise in its own exchange rate and a fall (i.e. becoming more negative) in its own current account surplus.  As its own NAFA is assumed unchanged, the fall in the CAS implies a rise in the PSBR which implies that country B's GDP must fall.  This fall in GDP feeds back into the function for the CAS, but given our assumptions about country A, this only impacts on the exchange rate not the actual level of the CAS.

So, the overall effects are as follows.

1. If country A's private sector wants to save more, the impact depends on whether it wants to accumulate domestic or foreign assets. 

2. If it wants to accumulate domestic assets, this will hit domestic GDP and not foreign GDP. 

3. If it wants to accumulate foreign assets, this will hit foreign GDP and not domestic GDP.

4. Only in the latter case does a current account imbalances arise.

Obviously, I've had to make many simplifications here.  Two of particular importance are worth noting.

First, I have assumed that the portfolio decisions do not depend on the exchange rate.  In practice, capital movements are sufficiently fluid in response to exchange rate deviations that they can dominate the real exchange rate in the short term.

Secondly, I have ignored any kind of inflation mechanism in the relationship between the real exchange rate and the level of GDP.  If, in fact, certain levels of real exchange rate create unmanageable inflation pressures, then domestic policy is likely to respond by influencing the NAFA or the PSBR and this will have knock-on effects for the analysis.

Notwithstanding these points, I think the general conclusions still apply.  This would mean, for example, that the ability of the US to run a persistent current account deficit is not so much about absolute savings behaviour in the rest of the world as the attraction of the dollar as a vehicle for saving.


[1] The real exchange rate here is expressed as the number of foreign currency units that can be purchased for one unit of the domestic currency.
[2] In my opinion, the role of portfolio decisions is one of the most important themes of post-Keynesian economics.  It underlies liquidity preference and it goes to the heart of why the role of banks matters to macro-economic outcomes.

6 comments:

  1. Nick,

    While obviously there are effects, it needn't be one particular way.

    So:

    "In the first instance, assume that the private sector wants to accumulate domestic government bonds, so that PSBR increases to match the rise in NAFA and the CAS stays unchanged."

    The private sector's preference to increase its share of government bonds can be because of rise in its preference for government bonds as compared to money or other securities. In that case the adjustment happens via price not GDP.

    Also in your scenario, you have GDP falling, so that reduces income, so the current account balance has to change since imports depend on income.

    Another case:

    " If it wants to accumulate foreign assets, this will hit foreign GDP and not domestic GDP."

    It's better to describe all this as Keynes' "two step process" rather than combining all as one. But even without doing it, take an example:

    India and UK, exchange rate fixed. A rise in the UK private sector's preference to hold Indian assets can be adjusted with the Indian central bank accumulating UK reserves.

    "This would mean, for example, that the ability of the US to run a persistent current account deficit is not so much about absolute savings behaviour in the rest of the world as the attraction of the dollar as a vehicle for saving."

    That isn't true. Suppose the Chinese government creates a large fiscal stimulus. The US current account deficit will fall. So obviously that's contradictory to what you say.

    You are correct about the ceteris paribus assumption, but my point is that there's a more fundamental way to do it.

    Also current account balance behaviour also depends on competitiveness (price and non-price) so those things are also important: what they are, why are they different and so on.

    ReplyDelete
    Replies
    1. "The private sector's preference to increase its share of government bonds can be because of rise in its preference for government bonds as compared to money or other securities."

      Maybe I was being a bit lax with terminology, but what I meant here was by domestic government bonds was any public sector liability, including money. The distinction that is relevant here is who is on the liability side of the asset, not what the asset is.

      "Also in your scenario, you have GDP falling, so that reduces income, so the current account balance has to change since imports depend on income."

      No. By assumption the current account balance does not change because there is no change in the desire to hold foreign assets. So the real exchange rate has to change to compensate for the fall in GDP. Now, I appreciate that this is a case where we might want to question the assumption that portfolio preference does not depend on the exchange rate. In practice it does, but how it does depends on some complicated expectations of the future, including critically how people expect the fiscal and monetary authority might respond to changes in the exchange rate and what this implies for inflation. We might conclude that such things do indeed change the portfolio preference, but I don't think that invalidates the conclusion here.

      "India and UK, exchange rate fixed. A rise in the UK private sector's preference to hold Indian assets can be adjusted with the Indian central bank accumulating UK reserves."

      Agreed. I was thinking of floating exchange rates and perhaps should have made this clearer. I agree that the public sector can neutralise any private portfolio preference to the extent it can take the opposite position.

      "Suppose the Chinese government creates a large fiscal stimulus. The US current account deficit will fall."

      I wouldn't say that necessarily follows and this is to some extent the whole purpose of this post. Suppose that there is zero domestic demand for additional Chinese government bonds. A fiscal stimulus implies that the PSBR is higher for any given level of GDP. As we have assumed that the PSBR cannot rise, because there is no demand for additional bonds, the fiscal stimulus must raise GDP (or potentially inflate away the real value of existing bondholdings creating space for new holdings - but that's beyond the scope of this analysis). If NAFA is unchanged and the PSBR is unchanged, the CAS must be unchanged. However, if GDP rises, it seems plausible that NAFA will also rise in which case, the Chinese current account surplus will rise, not fall as the Chinese private sector acquires dollar assets. Given what is happening with GDP, this would require significant weakening of the RMB.

      Of course, it is not a realistic assumption to suppose that there would be no new demand for Chinese government debt, but the point here is that the response to a Chinese fiscal stimulus is very strongly depend on portfolio preference. If raised incomes in China, in response to such a stimulus, was to lead to increased saving into US dollar assets, then the impact on imbalances could be a lot weaker than you might expect (subject of course to things like China's currency controls).

      Lastly, yes, general competiveness matters. In this framework, it impacts on the relationship between GDP, the exchange rate and the current account balance. Subject to my closing caveats, changes in non-price competitiveness would simply lead to changes in the real exchange rate. However, in practice, the authorities are likely to react to avoid the inflation implications of major changes in the real exchange rate by adjusting policy and this will impact on the NAFA and PSBR functions. The upshot is that changes in general competitiveness will affect imbalances.

      btw, I mentioned to you that I was thinking of writing a post related to your reference to Cripps' piece. This post wasn't it.

      Delete
    2. Yeah understood this post wasn't about it.

      About your point:

      "No. By assumption the current account balance does not change because there is no change in the desire to hold foreign assets."

      But you are assuming a direction of causality from the financial account to the current account. The causality is in both directions.

      Sure portfolio preferences affect trade balances but them not changing doesn't mean trade balances are unaffected.

      If GDP of a country falls, it will obviously affect the current account balance since imports are income dependent and GDP and income are related. It's unlikely that a fall in GDP not affecting the trade balance, whatever portfolio preferences are doing.


      Delete
    3. The important point is to note that it depends on what the exchange rate does. A movement in the exchange rate could eliminate any change in the current account balance that is due to a rise in GDP. You have to ask what causes the change in GDP and what do those same factors imply for the exchange rate. And portfolio preferences play a major role in determining the exchange rate, because the financial flows are so large nowadays. In theory, the impact through the exchange rate could even outweigh the GDP impact, so the current account actually improved. I'd say this was unlikely, but it's not inconceivable.

      I should stress that I'm talking about the long term position here as well. Generally, the current account balance responds fairly slowly to exchange rate changes.

      You are right though that I am ignoring some important mechanisms in assuming that portfolio preference is fixed rather than recognising the two way causation. I did acknowledge this at the end of the post. I think I can justify this though on the basis that I think the general conclusion about the importance of portfolio preference still holds even when you add in suitable feedback effects.

      Delete
  2. I think of the systems more as financial dealer networks in each country. So there would be an aggregate bank sector, central bank, federal Treasury in each country. The floating exchange rate is driven by pressures to swap one currency for another on the books of the respective aggregate bank sector. A central bank can enact policies that do not impact FX rate, increase the direction of change in the FX rate, or decrease the change in the FX rate, at any point in time. The float of net financial assets provided by a fiat sovereign government depends on the historic path of tax, spend, and credit policy so demand for savings is not a factor in the size of this float. The allocation of this float between domestic and foreign owners of financial assets will be the result of net trade flows and cross-border investment positions. I agree it is a complex system. The key is to figure out which decisions are taken independently, such as tax, spend, and credit policy, trade deficit, and foreign investment allocation, and see how the financial sector adjusts balance sheets and FX rates under those complex forces.

    ReplyDelete
    Replies
    1. Well, I think all of those things are important, and you could build a complicated model to try and map all that, but what I was trying to do here is set out a fairly simple, but still mathematically consistent, analysis. The downside of making it simple of course is that I have to leave a lot out.

      Delete