This paper was written in August 2004, for no particular purpose other than to clarify my own ideas.
Part I
Keywords: Growth & Innovation; History of Ideas, Methodology & Philosophy;
Either this kind of aggregate economics appeals or it doesn’t. Personally I belong to both schools. Robert Solow (1957)
PART II
The Double Step Chart
After 1997
Back to Econometric Estimation
Characterising Economic Growth
Standard Growth
Turbo-growth
The Effect of Shocks
Paradigm Conflict
PART I
To 1974: The Aggregate Supply-side Paradigm
The Crucial Experiment of 1974
1975 to 1981
1981 to 1986
The Grand Policy Break and Economic Modelling
The Intervention and Allocation Debate
Leaving the Institute
1986 to 1997
International Comparisons
Bryan Philpott
The Economy After 1985
Looking for the Recovery
This is an update on the original graph.
This is evident from the following graph. Since I cannot remember why I tried this presentational form, it is best if I just show it. While it has been updated since 1994 (and with the addition of extra OECD countries) the basic pattern has not changed. Note it is not too dependent on PPP measures.
The graph sets out the path of New Zealand volume GDP from March year 1955, where it is indexed to 1000. Over this New Zealand GDP path is superimposed three OECD GDP paths. The first, on the left of the chart, is set so that OECD GDP at the same 1000 in the March 1955 year. The middle path has the OECD GDP set at 820 in the March 1955 year, that is 18 percent lower than the first OECD path. The third path, on the right, has the OECD GDP set at 730 in the March 1955 year, or 11 percent lower than the middle path.
The graph does not challenge the conventional wisdom’s notion that the relative level of NZ GDP per capita (measured the usual way) fell over the post war period relative to the OECD as a whole. What it challenges is that this fall was continuous, incrementally every year. Rather, in two thirds of the years – perhaps more – the New Zealand economy grew at much the same rate as the rest of the OECD. The slowing down occurred in two transition periods: 1966-1977 and 1986-1993. The effect was like twice dropping a step on the ladder, rather than slithering down it. Moreover each step down is associated with a shock which we recognise as the terms of trade collapse of 1966, and the real exchange rate hike of 1985. With minor (cyclical) variations, this is a pattern of OECD growth, stagnation after shock, OECD growth, stagnation after shock, OECD growth.
What I want to do here is look at the question of how the ASSP might explain the graph. This is a very hard question to answer for someone who has a superior SOME paradigm (in the methodological sense of fully incorporating the inferior ASSP paradigm). It is like having chunks of one’s tool kit locked away, or explaining the path of a satellite using the Ptolemaic system..
One solution is to deny the graph, or ignore it. But confronted with the graph, how might the ASSP deal with it? The only useful way I can think of – there are not a lot of degrees of freedom in the ASSP – is that some or all of the labour, capital or technical progress stopped growing or contracted during the transition periods. If that is correct, then in principle they can measure how much of each, by looking at the five individual periods separately, rather than the whole period.
Of course having done that, the next question is why those quantities contracted. Any answer involves looking at the macroeconomic events of the economy at the time, which pushes one towards the SOME, just as it did to me decades ago.
In particular the graph makes plain the absurdity of many of the standard explanations which are used by the conventional wisdom to justify New Zealand’s slow growth rate. It hardly suggests New Zealand is too small or too distant. Did New Zealand get smaller and or more distant between 1966 and 1977 and 1985 and 1994? What about the theory, popular in the early years of the reforms, that it was excessive intervention ‘what done it’. Did intervention increase during the transition periods, and what about the 1982 to 1984 period when intervention did increase (the wage-price freeze) and there is no slower growth rate transition. (This reinforces my earlier observation that if intervention was the problem, how come New Zealand grew so spectacularly in the 1930s and 1940s when interventions were being increased.)
And if a reduction in corporate tax (or whatever is one’s pet policy solution) will accelerate economic growth, was corporate taxation (or whatever) raised at the beginning of the two transition periods and lowered at each’s end? It took me a long time to realise that a lot of the unattached theories which surround the ASSP are refuted – at least in the crude form they are generally presented – by this double step down graph. Already knowing they were inadequate I did not look for such a simple demonstration.
I used the graph twice in In Stormy Seas but with hindsight I did not make enough of it, perhaps because it was such a late addition to the book I had not fully grasped its significance. In any case I needed more years of data to understand it properly
I turned away from extending SOME after the publication of In Stormy Seas, and put my effort into other research areas. Perhaps there was an intellectual exhaustion, perhaps the economy was not doing anything interesting. Certainly the conventional wisdom was not, so there was little spur to improve the paradigm.
I monitored the economy, occasionally updating the double step graph and writing the story up. The one extension to SOME was the realisation that individual country comparisons were relevant, although even then this was not such an innovation since I had already done some sectoral comparisons in the supply-side chapter of In Stormy Seas . I had been interested in Australian-New Zealand comparisons since my visit there in 1987, and did a detailed policy comparison, published in early 1996, which suggested Australia’s less extremist policy reform resulted in a better economic performance (and showed how constitutional arrangements affected policy choice). The theme was taken up by other New Zealand economists somewhat later without any acknowledgement of this earlier work, which I thought discourteous to my co-author, Australian Rolf Gerritsen.
I added other quantitative comparisons of country performance, albeit without the same close policy consideration What they invariably showed was that New Zealand’s tradeable sector had done poorly in comparison to the others, which had better export performance and less imports (measured by import penetration), a fact which would be unintelligible to an aggregate commodity paradigm such as ASSP, but was predicted by SOME.
But I did not do a lot new for about six years.
I am not sure why I came out of hibernation. One factor may have been when Alan Bollard became its Secretary, to his credit, he insisted that the Treasury formalise their paradigm. Reading their research papers, led me to think about SOME again. Additionally, the Ministry of Economic Development asked me in early 2004 to provide a brief history of New Zealand’s economic growth, weighted towards recent events. I did not set out the formal SOME, but I used it in the discussion and had to think about it again as a part of the presentation.
Back to Econometric Estimation
I then noticed I had made an interesting methodological short-cut, understandable at the time, but no longer necessary. In order to gauge the effect of the terms of trade and the real exchange rate I had concentrated on two obvious major negative shocks. But there were minor shocks, and sometimes they were positive ones. Why not estimate the impact econometrically?
I had mucked around with some econometric estimation in about 1990. It is not reported in In Stormy Seas because I had ended up with a thoroughly confusing lag structure, and in any case the data series was relatively short and would not have had enough of the post 1985 shock to get a good bind. A decade-and-a-half later there was additional data, and an econometric technique called co-integration which cut through the lag problem. So with the help of Les Oxley, professor of economics at the University of Canterbury, I tried it.
Of course the econometrics simplified SOME. It assumed that the ratio of New Zealand GDP to OECD GDP was a function of the profitability of the export sector, which is algebraically equivalent to a combination of the terms of trade and the real exchange rate. What it says is that if these two variables remain constant New Zealand will grow at the same rate as the rest of the OECD (business cycles aside).
Blow me down, the econometrics worked. (Oh, me of little faith.) In summary a 1 percent long run fall in the terms of trade reduces (production side) GDP by .7 percent (and income side GDP by 1 percent) in the long run, while a 1 percent long run rise in the real exchange rate in reduces GDP (measured on either side) by .3 percent in the long run. (To my astonishment when reading my 1982 paper I observe that I had estimated the latter figure – i.e. .3 percent – all those years ago using a quite different method. It may be a coincidence but I need to go back and think about it.)
With these more precise estimates I could explore some aspects of the postwar economy which hitherto had not been possible. Among the new refinements were:
1. It is possible that there is a two step process in the transition from 1966 to 1977. You just can see it in the graph. I had tended to discount it, treating the 1972-1973 world commodity boom as a temporary shock. It does not materially change the model.
2. It appears part of the lift in the late 1990s and early 2000s may be due to a secular (i.e. long term) recovery in the terms of trade. Again you can see it in the graph. I have yet to study why that happened, but I should not be surprised if the Uruguay Round was a factor, which has implications for the effect of the much more favourable Doha Round on economic performance. Where do international trade agreements come into the ASSP? (Another factor may have been the falling price of computers, as a result of the spectacular rate of technical change. Again where does that come in the ASSP?)
3. Because I focussed on big shocks I completely failed to pick up that there had been a rising real exchange rate in the 1950s and 1960s which slowed down New Zealand’s growth rate. Again you can just see it in the graph, in that New Zealand was growing slightly slower than the OECD. (I was aware of this but the effect was small, and I attribute it in part to poor measurement of service growth, a hypothesis I have yet to abandon.) On the other hand the real exchange rate fell a little in the 1970s and that eased back the impact of the terms of trade fall. You cant see it in the graph because it is masked by the terms of trade transition. The reason I did not pick up the real exchange rate effects during the early part of the postwar era was partly because I was focusing on major shocks, but also because I graphed the series on a linear rather than log linear scale. Go on, look at it on page 87 of In Stormy Seas – botheration! I had discounted the real exchange rate’s importance before 1985, and I should not have.
This last finding led to an even more astonishing one. When I looked at the whole of the post-war era, I found that real exchange rate had a bigger overall impact than the terms of trade on of New Zealand’s GDP. Although the terms of trade impact is about double the real exchange rate for the same change, the real exchange rate deterioration over the period has been more than double the terms of trade deterioration.
Now this has an important, and for me not entirely expected, policy implication. Following the identification of the 1966 terms of trade shock, I had implicitly assumed that the deterioration in the New Zealand relativity was largely out of New Zealand’s hands for while we can have a little influence of the price on some commodities – by not pushing too much on the world market – generally our export and import prices are set overseas by factors outside New Zealand’s control.
However the real exchange rate is more subject to local influence, although it is not easy to control. It is not just a matter of setting the nominal exchange rate, since internal inflation may offset the change. Nevertheless the real exchange rate is more controllable than the terms of trade. The policy implication is that New Zealand’s poor growth performance was far more a matter of macroeconomic management than I had assumed. Had we paid more attention to it we could have done better. I can hear Bryan Philpott chuckling away at this conclusion.
Characterising Economic Growth
We can observe two sorts of growth patterns among rich OECD countries: standard growth and turbo-growth. But we need to take into responses to shocks into consideration..
One of the least remarked features of the modern growth patterns is that most of the top rich countries grow at roughly the same rate, with their per capita PPP adjusted GDPs moving along like a pack of runners bunched together. Some are at the front of the bunch, some at in the middle, some at the back. While there is some over-taking and falling behind, basically the bunch moves at the same broad rate.
We cannot even be sure of each runner’s positions in the bunch, because the statistical measures are not particularly accurate. We may have suspicions to explain some of the placings: this country may use more of its population in the workforce, that country’s workforce may work fewer hours than average, this seems to have more capital, that has more resources. When we try to measure these effects we end up in statistical difficulties and uncertainties.
How can we explain this bunching phenomenon? The ASSP says hardly anything, but let me go back to Joan Robinson’s model of technology as blueprints, an approach that I have had a revived interest in, following conversations with Ken Carlaw, at the University of Canterbury, who with Dick Lipsey and Carl Bekar is doing some really interesting thinking about technology in economic growth (which also provides a critique of the ASSP). I wont go into it all here in any detail – watch in the future how it changes my view – but here is my interpretation of a central idea. Suppose the technology blueprints are widely available so that all rich countries have access to them (at a low cost, for technology royalties are generally low compared to the national returns they generate – often because they lower prices). If ‘technology’ is the major determinant of economic growth, and allowing for the labour force and capital stock to be affected by migration and international capital flows – one would expect the bunching we observe among countries with well trained labour forces and favourable institutional arrangements.
There is a further feature of the Carlaw et al approach which attracts me. The ASSP sees technological progress mysteriously increasing over time, with no explanation as to how that happens. However Carlaw et al are more careful. They see the stock of technology blueprints increasing. This is also mysterious, but there is a substantial literature in industry economics which describes much of the process, so the mystery is shifted to a lower level, rather than the hand-waving of the ASSP.
Since these technologies are continually being applied, economic growth is really about numerous small increments, each lifting the economy to a slightly higher level of production. But as in quantum mechanics, the jerky increases at the micro-level appear smooth at the macro-level.
Carlaw et al go a step further with the notion of General Purpose Technologies (GPTs), that is very powerful technologies which dramatically open up the possibilities of numerous new blueprints which can be applied in a variety of situations. Examples are electricity and the silicon chip. They think there are a not of lot of GPTs over the entirety of the human history – although many of them have been in the last couple of centuries of economic growth. In the long run GPTs lift output by a significant step, but in the short run, because it takes time to explore all their implications – to create all the new blueprints – their sequential introduction gives the impression of smoothness. When the current GPTs have worked their way through and if there are no new ones, economic growth will largely come to a halt.
Now where GPTs come from – and what is the next one – is still mysterious. But the model pushes the mystery a bit further back, and it does have some policy implications which come out of the model rather than are arbitrarily attached. For instance: since most technology blueprints are generated overseas, a major policy issue is international technology transfer. (This is not incompatible with the ASSP, but because it is treated as a model of a closed economy, the point is overlooked, and the rhetoric – and, I’m afraid, policy – focuses on domestically produced technology (i.e. the creation of blueprints) and downplays the international dimension (their import and local utilisation).
While the pattern of bunching economies predominates there are a few shooting stars – economies that grow much faster than the typical OECD rich one. Examples include Germany in the 1950s, post-war Japan up to 1990, Ireland in the 1990s, and the Asian Tigers in the 1980s and 1990s. A feature of the growth is that the economy starts off well below the OECD rich bunch and the growth comes to an end when that economy joins them. There is no example of a shooting star which has overshot.
Turbo-growth involves economies which are technologically backward but have conditions which enable them to adopt the top-of-the line technologies used by the rich countries. Those conditions include appropriate institutional arrangements, a supply of underutilised labour (Asian agriculture or the Irish unemployed) or from migration (Germany and Ireland), which has the appropriate set of technical skills, and a supply of financial capital (from international capital markets or high domestic savings). Additionally there must be markets where they can sell the additional output. It is rare that they can rely upon their internal markets – that they can bootstrap – so exporting plays an important role especially if the economy is small or specialised. It is the application of top-of the-line technologies which are key and once they are fully utilised and the economy joins the rich bunch, the turbo-growth ceases and it grows at roughly the same rate as the bunch (although there may be difficulties adjusting to this slower growth as we saw with Japan in the 1990s).
Every one wants turbo-growth so there is much rhetoric about it. But the conditions for it are rather special and not available to the majority of rich countries because they already have the technologies, nor to poor countries because they lack the institutions and sufficient of the technologically adept labour force.
Is New Zealand capable of turbo-growth? It has no obvious reserves of labour. (It is not obvious that migration would be a source in the way it was for West Germany from displaced East Germans and Ireland from returning diaspora.) Perhaps a half-turbo could come from upgrading New Zealand’s technology to top-of the line, except where is it not? (Management?) It is also possible that New Zealand has the wrong industry mix, but we dont know which. After all, New Zealand may be behind the bunch because of measurement error rather than something inherently wrong. WE JUST DONT KNOW.
This reflection on turbo-growth raises the issue as to what growth enhancing policies are about? The conventional wisdom’s rhetoric says their purpose is to accelerate growth, but the last few pages suggest that is not particularly practical for a rich country.
There is a simpler explanation of the role of these policies. Their purpose is not to lift the growth rate, but to lift output – they are about progressing how far the car has gone, not making it faster. They sustain the current (or, more precisely, rich OECD) growth rate, rather than accelerate it. The policy actions enable the adoption of the new technologies that are becoming available or tackle the bottlenecks the growth generates where there is no market mechanism to resolve it optimally (i.e. transport congestion). While such an objective is less grand than the promise of faster growth, it is more realistic. Strip away policies (and advocated policies) which are primarily distributional, and the vast majority of new policies being implemented in a well functioning economy are dealing with new technologies or the consequences of growth which market responses cannot resolve.
The ACCP says that technical progress speeds the car along at some natural rate, and it is possible to take policy measures which make it go faster. Why the car goes at this natural speed at all is a mystery, and so largely is what the additional petrol which accelerates it.
On the other hand SOME (although it would be wrong to confine this account solely to the SOME paradigm) says the petrol that determines the speed is the newly available technologies (in the blueprint sense). For most cars that characterise the rich countries, there is a maximum amount of available petrol, so there is a maximum speed. The accelerator foot is already down on the floor and the car cant go any faster. However some cars have additional petrol because they have not used up all that has become available in the past (they have underutilised technologies), so they experience turbo-growth, until they use up their petrol reserves (the underutilised technologies). Since those without reserves all get the same amount of petrol, they all drive at roughly the same speed – which gives the bunching we observe.
Even if we omit technological change, every economy experiences an unremitting flow of shocks. Fortunately most are small, and they average out. Sometimes they are larger, but transient as when an important market goes through a cyclical fluctuation. (This paper’s account has largely ignored the business cycle although in In Stormy Seas it plays an integral role which could be explained in a longer paper.) However, occasionally an economy experiences a major external shock, of which New Zealand’s 1966 terms of trade collapse is a very good example.
The shocks change the relative profitability, which determines the choice of the blueprints which are applied. A fall in profitability usually means a switch of production techniques and change in the pattern of production. This involves a setback in the growth process as firms have to learn to use the new technologies. Moreover, they are doing so with constrained cash flow, which is not only used to fund new capital, but the upgrading of the skills of the work force and a host of other things such as market development which are integral to technology implementation.
In terms of the car analogy of the previous subsection, it is like getting a puncture. It slows you down and it takes time to change the tyre. However the economy does not really save any petrol (that is miss out any going technologies) from the experience, so when it gets going again it follows behind, unable to catch up with the bunch. But it continues to go at the same speed as the bunch.
When I first began developing this paradigm I assumed that there would be a permanent slowdown (which, incidentally, caused some of my confusion with the econometrics in the 1990s). However it is clear from the two-step graph (and the cointegration results) that once the transition to the new profitability structure is completed the economy seems to grow at broadly the same rate.
The first draft of this paper went on to discuss various ways in which the Carlaw et al approach could be incorporated into this transition. I’ve decided not to include it in the final draft, because my views are not adequately formed. One day I will go back and recall that perhaps the most important contribution of this paper to SOME was the material I have left out, because it began getting me to think about the problem. But at the moment it is just too early to tell.
The visiting economist from Mars would be very puzzled why the ASSP was so popular given its obvious limitations. He or she would acknowledge that it seemed seductively simple to non-economists who assume that the economists’ concepts that underpin it are deeper than they are. But why do economists hang on to it?
I, too, have puzzled over this and have come to the conclusion that the ASSP’s greatest merit is that because it is so vacuous it is does not threaten the policy agenda of the conventional wisdom. Because it has no macroeconomics, it does not challenge their macroeconomic policies; because it has no microeconomics it does not challenge their microeconomics; because it is empirically vague it does not force a consideration of whether past policies were less than optimal.
Any paradigm has to protect itself from challenge. Hence the neutralisation of Bryan Philpott, but this is an example of the wider phenomenon in which the conventional wisdom ignores alternative paradigms. Almost every bibliography of an ASSP paper is bereft of New Zealand generated research other than that which is in the ASSP tradition. The colonial cringe ensures the bibliography is stacked with irrelevant papers from overseas. But of critiques there is nary a reference.
Ignoring such research means the ASSP does not engage with it, and therefore never opens itself to the critical improvement which is central to scientific progress. That is why the ASSP is a stagnating paradigm, unable to generate any new insights or make new predictions about the real world. On the other hand, were it to engage with its critics – and the real world – it would have to be radically transformed.
I do not know whether the SOME model is ultimately correct. Actually I do because, as every Popperian researcher knows, it will be one day be replaced by a better paradigm. All one can say is that it addresses many of the issues which the ASSP ignores and – as this memoir has demonstrated – it has been an open one, evolving as it has engaged with other paradigms and with the real world. Hopefully it will continue to do so.