Prepared May 2004, with thanks to Les Oxley.
Keywords: Growth & Innovation; Macroeconomics & Money;
This paper is an econometric successor to my 1997 book In Stormy Seas, and my 2004 lecture The Development of the New Zealand Economy. I pointed out there that post-war New Zealand GDP grew at broadly the same rate as the rest of the OECD except to the extent the export sector experienced changes in profitability indicated by the terms of trade and the real exchange rate.
This led to an account of New Zealand’s economic growth which was based upon the export sector dragging the rest of the economy along. There are various overseas versions of this account (Thirwall), but it is relatively unusual in New Zealand because it more demand side that supply side, because it pays attention to sectors rather than focussing on aggregate output, and because it sees prices – and underneath rewards to factors – as having a crucial role.
When I wrote >In Stormy Seas some experiences (particularly the substantial hike in the real exchange rate) were too recent to be able assess econometrically. A decade later, there is sufficient data to carry out an econometric estimation.
The Constancy of Relative RGDP Growth
The analysis begins by considering the ratio
(1) k = NZGDP/OECDGDP
where
NZGDP = the GDP of New Zealand in constant prices
OECDGDP = the GDP of all (or a goodly selection of) the OECD in constant prices
The empirical evidence for k is shown in the following graph.
It is similar to some earlier graphs I have used except the vertical axis is the indexed ratio of NZ GDP to OECD GDP (rather than the RGDPs by themselves). The other major change is that I know have four steps rather than three, the new one arising from the splitting of the middle step. I have been aware of this possibility for some time, but have been reluctant to complicate the story. However some statistical analysis I have done suggested it made more sense to do the split. (The explanation of the split is that there were two separate terms of trade falls, one in 1966 from the wool price collapse and one in 1974 (probably) from the oil price hike (but also possibly from the entry of Britain into the European Common Market). .
What the graph suggests is that the deterioration in New Zealand’s GDP relative to the rest of the OECD was not continuous, but due to three (or originally I had two) shocks which happened at very specific and relatively identifiable times. A historical search shows that there was the wool price collapse of late 1966, which coincides with the first step down an oil price hike in 1974 which coincides with the second step down, and a hike in the real exchange rate which led to the third step down.
(The standard variation of the actuals from the four steps is 2.4 percent. Aside from the transitions between steps, most of the deviations could probably be explained by times when the New Zealand business cycle was not synchronised with the OECD one. This hypothesis can be tested if a measure for the OECD business cycle can be found.)
Note how this analysis leads one to shift away from a measure of a single aggregate output to looking at the economy more sectorally. (I also mention that I conceived of this theory before the final step-down. Thus my original notion of the impact being via the terms of trade had to be modified, although not markedly because the essential notion was the terms of trade reflected the profitably of the external sector, and, of course, so does the real exchange rate.)
Perhaps something needs to be said about the OECD-wide implication of an equation such as
k = NZGDP/OECDGDP, where k is a constant (or, as we shall see, a function of a few variables). Familiarity with the record of the post-war OECD suggests this may not be a bad first approximation. Most (rich) OECD countries grow at roughly the rate of the rest of the OECD., which is perhaps not so surprising, since their access to technology, capital, labour (to a lesser degree) and education is broadly the same. In the competition for best economic policy most will not choose equally effective ones. (There have been a few comets which have grown rapidly from a relatively low base: Germany in the 1950s, Japan up to the 1980s, and Ireland in the 1990s, but once they have reached the cluster of top OECD countries (on a GDP per capita basis) they join the OECD average, orbiting at a similar rat to the rest.)
Varying the Constancy
We need a model which explains and generalises the equation k = NZGDP/OECDGDP, and is consistent (including econometrically) with the formal evidence. The theory I developed (although others have broadly done the same) amounts to
NZGDP grows at the same rate as OECDGDP (and so k is a constant) as long as the tradeable (hence the export) sector) prospers. A reduction in its profitability slows down the growth of NZGDP (relative to OECDGDP) and an increase in its profitability accelerate it.
Now we cannot measure export profitability directly (or not as precisely as we need). But we can measure the relative price of exports to the relative price of other commodities which will be a proxy for profitability. (Why not include factor inputs such as labour? Much of the labour is self-employed, so a wage rate is not always meaningful).
A mathematical formulation of the previous paragraph is given by
(2) Log(k) = b*(Log(Px) – a*Log(Pn) – (1-a)*Log(Pm)) + c
where Pm is the price of imports;
Pn is the price on non-tradeables;
Px is the price of exports.
We note that the ratio Px/((Pn^a)*(Pm^(1-a)) is a relative price of output to inputs where ‘a’ reflects their relative weights, and that b is an elasticity of responsiveness. (The constant ‘c’ is a scaling parameter.)
Equation (2) can be re-written as
(3) Log(k) =b* (1- a/2)*Log(TOT) – b*a*Log(REX) + c,
where TOT = Pe/Pm, the terms of trade;
REX = Pn/((Px*Pm)^(.5)), a measure of the real exchange rate.
I have made various attempts to calculate Pn, but have never got a really satisfactory measure. (For instance In Stormy Seas subtracted an estimate of the value of the tradeable sector from GDP which was divided by an estimate of the volume of GDP less the volume of tradeables. While theoretically this is correct, estimating the size of the tradeable sector (including importables) proved too unreliable.)
Suppose as a reasonable approximation that
(4) log(Pg) = .5*Log(Pn) +.25*Log(Pm) + .25*Log(Px)
Where Pg is the GDP deflator, and the weights, which reflect the size of the sectors, are about right (but in practice they shift over time).
Then equation (3) becomes
(5) Log(k) =b* (1- a/2)*Log(TOT) -2* b*a*Log(REX1) + c,
where REX1 = Pg/((Px*Pm)^(.5)).
We estimate equation (5) as
(6) Log(k) =α*Log(TOT) + β*Log(REX1) + γ,
although in practice we would expect some lagged behaviour as adjustment takes time.. So the α and β are long run parameters.
Estimating the Parameters
Les Oxley kindly estimated the α and β for me, using a data base for New Zealand for the 49 years from March year 1955 to March year 2003, all the variables coming from official statistics, The method involved a standard approach where the series were co-integrated to degree 1. His estimates (with the standard errors in brackets after, indicating the estimates are reasonably accurate) were
α = .6322 (0.0718)
and
β = -,3691 (0.0416).
This says a 10 percent increase in the terms of trade, all other things constant, will increase RGDP (measured on the production side) by 6.3 percent. (As we shall see other things do not remain constant since there can be a real exchange rate impact). A 10 percent rise in the real exchange rate will reduce RGDP by 3.7 percent.
We estimate a and b as
a = .254
b = .726.
Recall that ‘a’ is the weighting of non-tradeables in relative price equation, with 1-a as the weighting of imports. The estimate of .254 (say a quarter) is the right sign but the magnitude is lower than expected of a little over a half. The measurement of export prices before or after insurance and freight may be an issue, but that is likely to shift the value up only moderately. I shall have to think more about this coefficient.
The estimate of the coefficient ‘b’ is also the right sign, and the magnitude plausible: an increase (decrease) of 10 percent in the relative price of exports (say an increase (decrease) of in the price of exports while all other prices remain the same) increases (decreases) RGDP by 7.2 percent.
(There is an important caveat here. RGDI or Real Gross Disposable Income will increase by more when there is an increase in the terms of the trade. Given that exports are just over 30 percent of GDP, the increase in RGDI – the effective income of the economy – will be just over 10 percent for a 10 percent increase in the price of exports, all other things constant.
The Historical Outcomes
In a historical context we can report the following:
The following graph gives shows the paths of the two variables, and their 5 year moving averages. The latter is the basis for the assessments in the next two subsections, noting that moving averages can obscure turning points.
The Terms of Trade
Initially the terms of trade were near flat after 1955, but between 1966 and 1984 they fell 32 percent, reducing RGDP by about 27 percent.
After 1984 they rose 11 percent, which would have increased RGDP by about 7 percent.
In total, over the 50 year period the terms of trade fell by almost a quarter, depressing GDP by about 15 percent (and RGDI by about 23 percent).
The Real Exchange Rate
The real exchange rate (measured by REX1) increased by about 42 percent between March year 1955 and 1972. This would have reduced RGDP by about 14 percent.
It then fell about 20 percent between 1972 and 1982.. That would have increased RGDP by about 7 percent.
Between 1985 and 2000, the real exchange rate rose Around 70 percent, reducing RGDP by about 22 percent.
In summary between 1955 and 2000, the real exchange rate (on this measure) doubled, which probably depressed RGDP by near 30 percent. The effect of this depression was to reduce RGDP growth on average by 1.0 percent per annum.
Thus the real exchange rate did more to depress GDP in the period after the 1950s than the terms of trade. This represents a reversal of the balance in my past thinking. It focussed on shocks and ignored the significance of the year-on-year incremental rise in REX in the first 17 years of the period. Thus I gave more weight to the terms of trade shocks than the real exchange rate ones. (Moreover in the last decade the terms of trade shift has been favourable, while there has been a continued hike in the real exchange rate.)
A Concern
The above albeit – rough – calculations suggest that were the real exchange rate and terms of trade effects not to have happened, New Zealand would be well above the average OECD level, perhaps a third above, higher than where it was in the late 1950s . This seems optimistic. Some of the effects observed in In Stormy Seas – population growth and convergence – could reduce this a bit. There is more econometric work to be done before the model is complete.
Where Does Policy Belong?
It might seem that if the New Zealand economy is ruled by these macro-variables, there is little place for policy. However
The Terms of Trade
Policy has some influence over the terms of trade. A major factor in their depression has been international protection and it may be that the hike towards the end of the period reflects the success in trade rounds of reductions in trade barriers. Moreover, it is hypothesised– the research has to be done – the agricultural product subsidies of the early 1980s oversupplied (poor quality) products into world markets and drove the terms of trade against New Zealand. (In which case not all the lift afterwards can be attributed to the trade negotiations),
The Real Exchange Rate
Macroeconomic policy influences the real exchange rate. How else can we explain the dramatic increases form the mid 1980s, when the disinflation via monetary policy pushed up the nominal exchange rate, thereby lowering prices in the tradeable sector, and depressing internal prices. An inflation expectations. It is not only monetary policy which affects the exchange rate. So does fiscal policy, the wage path, and microeconomic policies which affect domestic price levels.
Microeconomic Policy
However, microeconomic policies also comes into the model in the analysis that generally k should be a constant among rich OECD countries. This is because they are all strenuously improving their microeconomic policies in responses to new circumstances and new knowledge. Because it is (implicitly) a competitive game, nobody gets a significant advantage, but a regime which stopped the strategy of improvement would lag behind.
A practical illustration of this principle is that the external diversification which took place after the 1966 terms of trade collapse was possible because a raft of policies had been introduced before 1966 – such as those which promoted forestry, horticulture and general manufacturing exports – which prepared the way for the diversification.
Thus the paper concludes that quality microeconomic management remains important for successful growth, but this history also tells us that good quality macroeconomic policy – especially the way it affects the economy’s engagement with the rest of the world – is important too. Nor should we forget external changes – such as changes in relative international prices – over which the economy has little control in assessing its growth performance.