This paper was written in august 2004, for no particular purpose other than to clarify my own ideas.
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)
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
The Economy After 1985
Looking for the Recovery
Paradigms of New Zealand Economic Growth: A Memoir II
The Double Step Chart
Back to Econometric Estimation
Characterising Economic Growth
The Effect of Shocks
It is now thirty years since when in 1974 I stumbled on the ‘crucial experiment’ which has led to a story of New Zealand’s economic growth, which the conventional wisdom still has yet to engage properly, as it clings to the ‘aggregate supply-side paradigm’ (ASSP).
I was almost at the beginning of the modern study of growth economics in New Zealand. It was not practical before the 1950s, because data bases were inadequate. Indeed some critical ones were not available until the 1980s, and in some ways they are still not adequate. The other crucial requirement was the theoretical paradigm, which might be said to have been first set out in 1957 in Robert Solow’s seminal paper ‘Technical Change and the Aggregate Production Function’.
The paper has two key elements. First it enables some quantification of the growth process, and second it shows that economic growth is not simply a matter of additional labour and capital. There is something else which drives increasing output per worker which Solow called ‘technical progress’. The quantification says that this ‘something else’ is perhaps four times more important than extra capital per worker.
Over the following fifty years, it has proved very difficult to explain the ‘something else’ with any precision. In 1962 Tommy Balogh and Paul Streeten called it a ‘coefficient of ignorance’, the unexplained residual, and that is the way it largely remains forty and more years on, with surprisingly little evidence of the causal links which generate its magnitude. Even so it is hard to avoid using the notion, as Part V of my In Stormy Seas does.
However, underpinning Solow’s theory is a notion which has dominated much thinking about growth ever since. It is there in the paper’s title: ‘the aggregate production function’. This means that one does not think of total output of an economy simply as the sum of all the (net) outputs of the various businesses in the economy. Rather, one treats the total output as if it were a single good produced by a single business using two separate inputs of labour and capital (plus the mysterious ‘technical progress’).
This aggregation to one commodity and two inputs is an enormous simplification for if it is true – or near enough to true – it means that an economist does not have to worry about individual businesses or sectors, in order to explain economic growth. It is such a seductive idea that it has been one of the workhorses of economics ever since, and it underpins the New Zealand conventional wisdom (and that of many other economists). Hence the paradigm’s entitlement to a name – the ‘aggregate supply-side paradigm’ – and an acronym – ‘ASSP’.
The ‘supply-side’ implies there is little attention to demand effects. Prices are almost irrelevant to the paradigm. In the old fashioned discourse one would call ‘Stalinists’ those who only use ASSP for research purposes, that is they are concerned only with the real side of the economy, like a Stalinist planner, ignoring the role of prices in economic development.
At first I was a committed ASSP adherent too. Indeed I was fortunate, because the director, Conrad Blyth of the NZIER where I worked from late 1963 to 1966, had completed a PhD in capital theory in the late 1950s, and cottoned onto the ASSP paradigm very quickly (as some of the research papers he published then demonstrate).
As Solow’s quotation which begins this paper indicates, virtually all users of ASSP have a split personality, because they apply much more disaggregated accounts of the economy when it suits them. It is not uncommon even today to listen to a paper whose research is based upon ASSP, and then hear the presenter jumps to policy conclusions which have nothing to do with it but come from a different – more disaggregated – paradigm.
(Sometimes it is as if economists have two tool kits, or their brain has been separated into two uncoordinated parts. I was intrigued recently by Stephen Pinker’s description of humans whose corpus callosums had been severed so their brains were literally two separate hemispheres. If one ‘brain’ was given an instruction unbeknownst to the other, the second ‘brain’ would make up an explanation of what happened which was plausible but wrong. ‘For instance, if an experimenter flashed the command “WALK” to the right hemisphere (by keeping it in the part of the visual field that only the right hemisphere can see), the person will … walk out of the room. But when the … person’s left hemisphere is asked why he just got up he will say in all sincerity, “To get a coke” rather than “I dont really know”.’ (The Blank Slate, p.43). Sometimes that seems to happen as economists switch between two paradigms.)
The two (and more) paradigm user certainly applies to me. I think of each as a part of one’s tool-kit, each paradigm being a tool to be used for a particular purpose. The skilled craftsman not only loves each instrument-paradigm, but knows its strengths and weaknesses. Choosing the right tool is a part of the skill.
For our purposes, the central point is that, even though it may derive its policy conclusions from other paradigms, the conventional wisdom uses the ASSP to analyse New Zealand’s economic growth. I no longer do. Why I changed my mind probably began in 1974, with what in the methodology of science may be called the ‘critical experiment’.
(The Solow paradigm was extended about 20 years ago by ‘endogenous growth theory’. This has not really been taken up by the New Zealand conventional wisdom yet, so I will not refer further to it. It is interesting, but like the Solow based paradigm so aggregate it does not provide many insights. I add that I like the notion of technology embodied in labour and capital: you cant have technology without the other two, but this is a much older idea.)
Returning to New Zealand in late 1970, I had decided to focus my research activity on the income distribution, which eventually culminated in Income Distribution in New Zealand. The reasons for the choice need not concern us here, although one was I wanted to look at the New Zealand economy (and political economy) through a different prism from that which was usual.
By 1974 I had constructed two series of the return on wealth, each of which in principle reflected the same underlying notion, although given the independent data bases from which they constructed I did not expect the series to correspond exactly. But to my dismay, I found that one increased through time and the other decreased. After all, one of the traditional questions of economics was whether there is a falling rate of profit or not. Between them, the series answered ‘yes’ and ‘no’. Botheration.
I recall that one night – it must have been Saturday – musing over this, it occurred to me that the two series did not cover exactly the same period. The different trends could be explained by the underlying series rising peaking and then falling. That required a statistical calculation. In those days there were not personal computers, and certainly one did not keep the old mechanical calculators at home. So somewhat agitated – to use a nineteenth century expression – I biked out to the Ilam campus the following day – Sunday – and ran a quadratic (a curved trend) through both sets of data. Glory be, there was the curvature with the peak in the mid-to-late 1960s.
Often solving one research problem creates another. The crucial experiment resolved the inconsistency but why had the peak occurred? I was out of New Zealand from late 1966 to late 1970, so I did not have direct experience of the period. So I had to read about what happened, and analyse the data.
Eventually I tracked down the critical change. There are a number of candidates – including the devaluation of 1967, and the Nil Wage Order of 1968, although those two ought to have boosted the return on wealth, not depressed it. The evidence settled on the spectacular collapse in the wool price in December 1966, which was the earliest candidate; it generally coincided with the peaks I had estimates, and it explained the other candidates. In those days wool made up about forty percent of all exports by value, and with sheep meats, the contribution of sheep farmers and processors came to about 65 percent of all export receipts. So a wool price fall of about 40 percent – permanent, except for a spike in 1972 and 1973 – was a nasty shock to the export economy.
The wool price appears in the terms of trade, the price of exports divided by the price of imports. That they tended to fall – sometimes dramatically – had long been a concern of New Zealand economists given the way they had plunged in the early 1930s. (The official series only started in 1926. Later I was to construct a series which went back to the 1850s.) Although born a decade after the Great Depression, I was aware of this concern, particularly through Bill Sutch and also in the international development literature. It argued there was an inevitable tendency for the price of primary commodities (such as wool, meat and dairy products) to fall relative to the price of manufactures, and so the terms of trade of primary producing countries (such as New Zealand) would decline in the long time, making economic growth more difficult for third world countries (and New Zealand).
I have done much research and reading on this over the years, and I do not now think – I may have in the past – that it is inevitable that the terms of trade for primary producers will decline. The evidence is that different primary products have different price patterns. Resource constrained ones – oil and fish – may result in a tendency for their relative price to rise, perhaps too for minerals and timber. The price experience of meat and dairy products – and no doubt other ones I have not investigated (cotton? sugar?) – in part reflects dumping on to world markets by rich countries who subsidise their farm producers to over -production . This is not an inevitable policy response, as the Doha Round shows. Some products – such as wool, red meat, and butter – have been squeezed by substitutes, by health fashions and saturated consumption in key (affluent) markets. But while there is no inevitability of decline, the prudent country would want a degree of diversification by export commodity and country.
I began collecting data series which showed a change in trend following 1966. As well as the terms of trade and the return on wealth, they included the profitably of farming, agricultural output, growth of output, unemployment and inflation. (When I wrote a Listener column about this in May 1979, a reader suggested I look at installed telephones, and. sure enough, there was a break about then too.) It is not hard to write a story about how these are inter-related, as an the external shock of a significant fall in the terms of trade impacted on some important sectors of the economy, which fed through to the rest of the economy, including slowing the growth of the economy.
As I recall, I had largely got this broad framework together by the mid 1970s, well before the long term consequences of the entry of Britain into the (now) European Union in 1973 or the oil price hike of 1974 would be evident in the data. Moreover the structural break was before 1973. Many people focus on the 1973 break with Britain: undoubtedly an emotional break for them, especially if they have a colonial mentality. The oil price shock (which also depressed the terms of trade) is a better candidate for a later date. I have never completely ruled it out, for it is difficult to separate the oil price effect from the short-term world commodity boom which preceded it. My work has tended to suggest that the oil price hike took New Zealand export prices back to where they were tracking before the boom. (While the price of imports lifted, so did the price of some exports – meat and dairy products – sold to the now more affluent oil producers.) Some very recent work – reported below – suggests that there may have been an additional effect from the oil price hike – or even possibly the British entry to the EU. Even so the 1966 effect appears larger and is, of course, earlier.
About this time, probably in 1975, I hit a major methodological crisis: the ‘capital reversing debate’. We dont need to go through it here, except to say it raises serious questions about the practical possibility of aggregating to a single production function in a meaningful way. It may not affect the ASSP because it is related to anomalous behaviour in prices (factor returns), and the ASSP is not very dependent upon these prices. Working between macro-economic aggregate and sectoral disaggregation the phenomenon was much more distressing, and I even contemplated giving up economics. Today I try to work at as a disaggregated level as I can, and always check when working at high levels of aggregation that capital reversing would not upset the story.
While the capital reversing debate was painful, it enabled me to have an understanding of what practically ‘technology’ might mean and how it might be fitted into an account of growth, without the handwaving of a technical progress index. Joan Robinson suggested we think of technology as blueprints, descriptions of how one might combine resources commodities and capital into output. I describe some of the insights of this micro-approach below.
As my publications list shows, I proceeded on a wide front after this discovery, continuing to work on distributional economics, but also getting sucked into the growth and macroeconomic implications of the findings, including investigating earlier periods of New Zealand economic history to see if the evolving theory worked.
The work for this period culminates in my inaugural address as director of the NZIER in 1982: External Impact and Internal Response: The New Zealand Economy in the 1970s and 1980s. It follows the 1979 Listener column, albeit in more detail, describing how the terms of trade fall from 1966 impacted on the profitability of the farm sector, then onto farm output, and the economy as a whole, again emphasising that numerous data series show a structural break in the 1966 to 1968 period.
The paper also sets out a formal model, from orthodox international trade theory, of the impact of the terms of trade on the production possibilities and hence output and growth. I recall one of the brighter young economists of the day saying after the lecture he thought I had got it wrong, and he would come back to me. He never did, and to this day – as a respected economist – he continues to use the ASSP without any awareness of the consequence of sector and price shifts. The ASSP assumes that economic output can be characterised by a single commodity. It makes no sense to export and import this same commodity. So the ASSP, ignoring the external sector all together and focussing on domestic production, is not very useful for economies where international trade is important. Now just suppose that the poor growth performance had something to do with the external sector. Those who relied solely upon the ASSP would have no way of investigating what happened.
Reflecting all these years later, I realise that I was doing something unorthodox in terms of the economics of the day. The tool kit – the various bit of economic theory I applied – was unquestionably orthodox, but I used it imaginatively, combining different tools for particular purposes. My approach has been eclectic. Faced by a problem, I draw from the tool kit any tools that seem useful. Having been alerted to a structural problem in the post-1966 era, I did not go back to a particular paradigm, but looked around for the tools which seemed relevant, synthesising them to what proves to be a different paradigm, which is treated as unorthodox by the conventional wisdom, even though every element of it is orthodox. Of course, that is the way intellectual progress occurs, with the conventional wisdom typically lagging behind.
In particular, once I had identified that the rupture was due to a price shock (a change in relative prices) I had to go outside the ASSP, which has no prices, and look for other parts of economic theory which did. Naturally I fell upon the theory of international trade and general equilibrium analysis, which both contain numerous commodities and hence have relative prices between them. So sectors became important, since different prices impacted differently on different sectors, as I illustrated earlier with the farm sector.
I called the paradigm the ‘Small Open Multisectoral Economy’ or SOME. Small because New Zealand is a small player in the world economy and so is largely a price-taker for exports, imports and foreign capital. Open because international engagement is central to the New Zealand economy. Multisectoral because other than in the short run, relative changes in sectors (and the prices which influence these changes) are crucial in the understanding of the economy’s behaviour. The three concepts are closely related, in that the most important production sectors – the absolute minimum one can get away with – are the exportable sector (which provides exports and some domestic expenditure), the importable sector (which competes with imports) and non-tradeables which are not directly influenced by international trade.
Note that three sectors means three prices and two price relatives. Initially I only looked at the terms of trade – the ratio of the price of exportables to importables – but by the early 1980s I was also concerned about the second price relative which is the real exchange rate – the ratio of non-tradeables over tradeables (although there are other ways of measuring it). I’ll be coming back to that later, but in passing I observe that some New Zealand economist confuse the real exchange rate and the terms of trade, perhaps because they have difficulty getting from a one commodity model to a multi-commodity one. There is an instance in a New Zealand textbook, and only this year a recent graduate came up to me and insisted that his teachers had not made the distinction. (Sigh)
Once having crossed the threshold from a single commodity to a two or three sector one, there is potentially no stopping, so the exact number of sectors depends upon the precise question that is being answered. In In Stormy Seas I said that I frequently think of a six sector model: agriculture; other exportables; hydrocarbons; other importables; market non-tradeables; government services. But of course I will go down to a greater level of disaggregation if necessary. I have long thought – certainly since 1982 – that one of the greatest achievements of the New Zealand economy in the 1970s was the diversification of exports by commodity and country, although I did not know then that it was the greatest in the OECD according to some work done by John Gould. But how can the ASSP acknowledge that, given there are no exports, let alone different kinds of exports, in the paradigm?
The next five years were spent as director of the NZIER. I had come on the basis that it was a research institution, but increasingly its work was dominated by having to raise money from consultancy. Now consultancy can enhance research by providing a wider perspective and unusual experiences. Sometimes it makes a surplus which can be invested in research. But it became more difficult over the years, especially as I was also trying to run a sophisticated macro-economic forecasting unit, rather than the mechanically auto-regressive ones that became common in later years, and that required funding for research too.
A summary of the research program over the five years will be found in my 1986 valedictory address The Exchange Rate Since 1981: Performance and Policy. By now terms of trade shocks were not important, and the focus was on the real exchange rate, which had risen sharply in 1985 following the floating of the exchange rate. The path of the economy was exactly what SOME would predict. A higher exchange rate, like a lower terms of trade, reduces the profitability of the tradeable sector (i.e exportables and importables), their expansion slows down, and so does economic growth. Once again the ASSR could make no prediction, lacking exports and prices.
This led to the grand policy break I had with the conventional wisdom. It was apparent from Economic Management, the Treasury 1984 post-election briefing – not to mention the odd conversation with and other published paper by officials – that there were some gaps in the Treasury thinking, as well as some very casual use of statistical data. Through much of the next two years, I assumed that they knew what they were doing, and scoured official papers to identify the rigorous underpinning model which they were using. In the end I decided that there was not one.
I was trained – extremely well by Derek Lawden at the University of Canterbury – as an applied mathematician, which involves describing the world by models characterised by a set of equations. Analytical economics does this too, so it is a matter of habit that when someone is talking about the economy I try to work out her or his formal consistent economic model. In a surprising number of occasions one becomes aware they have none. The kindest thing is to suggest their implicit model is incomplete, although some seem to have no more than a few relationships held together by an ideology.
It is difficult to explain this briefly to someone untrained in modelling, so I am going to give a very simple example from recent times. Suppose one says that the reason for New Zealand’s poor economic performance is its distance from significant markets. I will challenge that proposition with evidence which contradicts the assertion (in one of its forms) later, but there are those who say it despite the evidence, reflecting a bad habit in New Zealand public discussion to think that any two facts are necessarily connected if there is a vague feeling they ought to be.
Now it is possible to write a equation connecting distance and income in two fundamentally different ways. I wont here, but one says that the level of income is depressed, because New Zealand has to pay a lot to ship its products overseas (and so on) so we dont get as good a return for our exports (after these costs) as we would if we were closer. There is no problem formulating the relationship this way. The second implicit equation is that the (proportional) growth of income is depressed by distance from markets. Now that is a really interesting proposition (even if it is complicated by effective distance getting shorter). But what mechanism have the advocates got in mind? I listened to them really carefully, and soon recognised that they explain the second equation about growth by referring to the first equation about levels. It is a confusion, like saying to a traffic officer that you had not exceed the 50kph speed limit because you only drove 40kms.
Many people – including economists – get confused between growth and accelerating growth, between levels and changes in levels. Perhaps the following (imperfect) metaphor might help. Suppose we treat the level of per capita GDP as the distance a car has gone. The rate of growth of GDP is the speed of the car. Very often the two are confused as in the last sentences of the previous paragraph. Because people dont model well, the confusion is endemic.
As it became clear that the advocates of the reforms did not have a complete account of the economic model they were using, I looked for the ‘missing equation(s)’. I dont want to get too technical here but, as every modeller knows, an economy with an external sector (i.e. exports and imports) requires one additional equation to ‘close’ the model, that is to enable all the equations to be solved. There are a number of possible candidates including full employment, a zero deficit in the balance of payments, no inflation, a particular level of economic growth, and so on. Experience shows that the choice of that final equation leads to quite different outcomes. So which one did the advocates of the policies of the late 1980s think applied? The answer proved to be, as far as I could make out, they were not even aware that there was a problem. At a certain point they just put their hands together and prayed that the economy would ‘close’ itself.
In practice the closure came via the Reserve Bank’s monetary policy, although those administering it seemed unaware what they were doing. Their missing equation was the determination of the exchange rate. The conventional wisdom seemed to think that the market would set its own level, but it was unclear how. In principle the economy could settle around any given nominal exchange rate, although that had implications for output, employment, inflation and growth. Their analysis overlooked the fact that the fiscal and monetary actions of the government would impact on the exchange rate, or perhaps it assumed – wrongly – that the impacts were small and could be ignored. So they ran a fiscal position which would drive up the exchange rate, while the RBNZ ‘closure’ involved a high exchange rate too, with the outcome of slower growth and unemployment (albeit with disinflation). Which is exactly what happened.
Now you may think this has little to do with the paradigmatic clash. But the RBNZ model of money in the economy was closely related to the ASSP, because it assumed that there is a single commodity. If it did not, the RBNZ would have to think about multiple prices in an economy, and different rates of inflation, and that would undermine much of the rhetoric that surrounds monetary policy. (Nowadays the RBNZ looks at a number of price indexes, even if the policy targets agreement is about only one of them, the Consumer Price Index.)
Another part of the conventional wisdom’s brain talked about the allocative gains from microeconomic reform. Now much of my life as an economist has been about the tension between the success of the market mechanism and the failure of the market mechanism. The world is dominated by extremists who mention only one of theses features, and ignore the complexity. But the good Lord never gave me such simple beliefs, and in any case the balance has to be found in the particularities of the empirical evidence.
So when in the late 1970s, there arose the argument that New Zealand’s poor growth rate was a consequence of the high degree of interference with the market, I began to look around for empirical evidence. By this time there was a well established paradigm which calculated the losses due to misallocation of resources from market intervention, especially border protection. So I trawled through the research (much of which was done by Bryan Philpott, see below). None of the research showed a gain from the total elimination of border protection of more than 1 percent of GDP. (Actually one did, but it proved to be that a decimal point had been misplaced so a .7 percent increase became a 7 percent increase.) Given such a minuscule change – I was looking to explain a 20 plus percent loss over the first 20 years of the post-war era – I concluded that even were we to add in allocative gains from internal interventions, inappropriate intervention would explain only a very small proportion of the decline.
The smallness of the gains from eliminating protection is not surprising. Were there big gains in inefficiency from an intervention, a shrewd economic manager could work out how to reap them, while compensating losers. If the gains are small it is much less possible to do this. Disputes over protection are heated, not because the gains to the nation of eliminating it are large, but because they are small. However protection changes the incomes between various groups in the economy very substantially, so that if it is removed some people make large gains and some people make almost as large losses. But a dispute about distribution is too naked, so as with taxes, it has to be dressed up as a dispute about allocative gains and efficiency, just as the tax debate is presented today.
Because those who argued against border protection were bereft of evidence, they changed the rules, and focussed on effective rates of protection. If they are big they are bad in efficiency terms but you dont have to know a lot of economics to realise a big effective rate of protection (I am avoiding telling you how to measure them) means a big loss of efficiency. However the anti-protectionists made the equation even if they had no analytic reason for doing so.
Things became even more peculiar when they were calculated by different ways for manufacturing from agriculture, but then presented as if they were the same. You dont have to be a rocket scientist to know if one length is measure with a foot rule and the other with a theodolite, one takes care in any comparison. But so passionate was the debate that an obvious point was ignored.
Its get worse. The calculation – I am avoiding the technical details – requires that the cost excess in each industry be estimated. The cost excesses were calculated in the case of manufacturing but they were assumed (yes, ‘assumed’) in the case of farming. Initially they were assumed to be 10 or 20 or 30 percent, that is farmers had to pay an extra 10 or 20 or 30 percent, for all their inputs because of border protection, which would be quite a penalty. Eventually the debate forgot these figures were assumptions and settled on the mid rate as the true rate. So everyone wandered around talking about the 20 percent cost excess farmers faced. Farmers were most surprised that when we stripped out border protection they did not have a spectacular fall in their farm costs, and spent the 1990s hunting around for the gains they had been promised but which did not occur.
They should not have been surprised. In 1986 Bryan Philpott provided estimates of the cost excesses for farming and found they were 3 percent for current inputs and 5 percent for capital inputs, tiny by comparison with the assumed rates. The 20 percent gains never turned up, simply because they were not there. (Some farmers were sold down the line by their own organisations as a result, because they gave up significant support in exchange for almost non-existent concessions.)
When the interventions began being stripped away from 1984 I knew the research evidence did not support the extravagant promises by the ideologues. Of course, I could have been wrong, but I am not at all surprised that the research projections out-performed the extremist promises. There were no great gains from abandoning allocation, either on the level of output or its growth rate.
There is an irony in all this. If the theory of misallocation from intervention were correct, the ASSP would work even more clumsily. Again we have the case of one side of the brain saying one thing and the other saying the other.
Perhaps they were arguing that the costs of intervention, while mysterious, were still significant, and that removal of intervention would accelerate economic growth. However our economic history contradicted this, although I was not able to quantify it exactly until 1990. In the 1980s I could point out that in the early 1950s New Zealand had one of highest per capita GDPs in the world. Yet in the previous decades prior to 1950, intervention had increased the New Zealand economy. So where the correlation? In 1990 I provided an estimate of New Zealand’s volume GDP which showed that 1935 to 1945 was the period when New Zealand had its faster growth rate ever. And it was the period that interventions were imposed thickest and fastest.
Now lest I be misunderstood – not by you dear reader but you would be surprised at the speed at which some members of my profession turn correlation into causation while assuming that if one disagrees with their analysis one must support totally opposite policies – I dont believe that returning to the intervention levels of the 1940s would accelerate economic growth, and certainly the post 1935 experience does not prove that. On the whole I support low levels of intervention because the market gives consumers more choice and producers more flexibility, and because a government has so many important things to do it should not muck around with interventions which at best are of marginal benefit. However as my extensive writings show, there are places where I support intervening, and not only to get a fairer society.
It was not an easy decision to leave the NZIER when my five year term ended, despite being offered a renewal. In a way it has permanently broken my heart. But I concluded that the NZIER was being steadily pushed away from research into consulting, which has happened despite, a marvellous rear guard action by Alan Bollard who took over as director at the end of the year, maintaining a really interesting industrial research program (although, alas, the macroeconomic one fell away). And in any case, I was by now in deep conflict with the conventional wisdom of the Treasury and the Reserve Bank: while nominally the NZIER is ‘independent’, the Wellington environment is more complicated than that.
I am proud of my valedictory speech, not only because it reflects the quality research program I had managed to run over the five years (on the smell of an oily rag), but also because with hindsight I got it broadly correct. In particular it was gloomy about the prospects of the New Zealand economy given the high exchange rate strategy, concluding ‘some paths are self destructive. I fear we could be on one at the moment, for the long term consequences of an over-valued exchange rate is both an increase in total overseas debt, with a reduction in investment in the tradeable sector, and hence in the nation’s ability to service the debt.’
Saying the exchange rate presents a problem to economic performance is no surprise today, but it was then for the conventional wisdom assumed that the market rate was the appropriate one for economic growth. Alas, the rate set by fiscal and monetary policy would generate stagnation. And even nowadays, the lesson is ignored in ASSP circles. I recently spent a couple of days at an officially sponsored seminar on productivity where the ASSP so dominant that I dont think the exchange rate was mentioned at all.
All this was before the six year stagnation from 1987. I went out as a consultant, doing lots of interesting work. However, rather than take the high returns of consultancy by working long hours (which could have compromised my independence), or taking time off, I chose a modest income by consultancy standards and used the spare time for further research and writing, culminating in the publication of In Stormy Seas: The Post-war New Zealand Economy| in 1997. (For various reasons it was a couple of years later than planned. I should also add that on occasions I got a little research funding to support the work, although funding for research was not nearly as generous then as it is today.) There are three salient events to report over this period.
The first began in early 1987 when I was at the university of Melbourne as a visiting professor (the Downing Fellow), and became aware of the intense Australian interest in growth economics. Earlier in the decade the University of Pennsylvania had produced estimates of countries’ GDP (i.e. output) in the same prices, which made comparisons between the size of the economies more valid. GDP per capita at purchasing power parity (PPP) prices are somewhat more reliable than the previous estimates based on exchange rates – I was scathing about those in my 1986 lecture. I extended my work to use this database. Over a decade later they are now used in the New Zealand ASSP, and in the rhetoric of public debate.
Unfortunately the ASSP and the rhetoric uses the data uncritically, paying little attention to their accuracy. I spent a lot of time working on the accuracy of New Zealand’s GDP data (see Appendix 2 of In Stormy Seas and the related working papers) and am confident that on occasions the reported data has been misleading. I could not check the international comparisons with the same intensity, although the OECD 1987 figures gives an indication of the problems.
For instance, its international cost of housing was based on the cost of building a representative two-up-two-down brick terraced house. When asked for the cost, the New Zealand statisticians went to builders who gulped, having never built one, and gave a price based on what they would tender, without allowing for the savings they would make after having the experience of building hundreds. The cost was inappropriately high, with the result that the estimates 1987 have New Zealand investment too low. This was remedied in the 1990 round although getting international quality comparisons of investment goods still remains problematic, while when it comes to comparing the government sectors, the practice is to assume all civil servants with the same qualifications are equally productive. Oops there I go again, thinking about sectors. Tut-tut.
Even after intense effort to improve data quality, the changes between estimates at each point in time 1990, 1993, 1996, 1999, and soon 2002, do not connect with the GDP changes between them estimated by the official statisticians. Sometimes the differences are huge, certainly more than can be explained by errors in the measurement of GDP. Nobody is sure why. The sensible thing is to use the data cautiously.
It may seem odd that the local ASSP is so immune to the problems of data quality. It reflects, I think, that it does not push research frontiers, but works well within them. In one sense the data does not matter, as the other half of the brain gets on with policies which are only tenuously connected with the ASSP.
Very recently – I am writing this in 2004 and to my knowledge the issue only became apparent in late 2003 – there has been a realisation that the measured described as ‘GDP per capita at purchasing power parity’ does not measure production at all. It measures income. They are exactly equal at market prices, but they need not be equal at any other prices – and generally are not at PPP ones. One source of difference for New Zealand is that the practice of some rich countries of dumping products which compete with New Zealand exports not only lowers New Zealand’s foreign receipts, but lowers its income relative to its production at PPP prices. We dont know how big that is, but one study found a gap of around 9 percent in the estimate for Portugal relative to the US. New Zealand’s gap is likely to be more since it more greatly suffers from dumping.
Thus the international GDP comparisons do not compare productive power but expenditure power. The latter is useful for assessing material standard of living – if that be important – but it is of much less use for international productivity comparisons, and hence for the ASSP – if it really matters.
My approach has been to use the international GDP comparisons for illustration, but not to rely on them too much for research (or, for that matter, policy) conclusions. The conventional wisdom has gone the opposite direction, placing greater weight on the data than it can possibly bear. Perhaps it is a matter of rhetoric rather than analysis, a conclusion forced upon one by the oddity of the way the ASSP research has progressed.
Their emphasis has been on ‘rankings’ – that is the number of OECD countries which are above New Zealand on the (flawed) measure of GDP per capita – rather than ‘relativity’ – how New Zealand’s level compares with the average.
Any competent quantitative social science would be alerted to the peculiarity of this choice. Rankings involve ‘ordinal’ measures, while relativities involve ‘cardinals’. A child can work with ordinals, that is count, but their limitations soon become apparent and the child moves on to adding, that is using numbers as cardinals. Ordinals are a much more primitive means of handling numbers, and yet involve much clumsiness. For that reason their statistical properties are less understood. (Technically, any statistician can instantly tell you about the properties of the mean of a sample, a cardinal measure, but would fumble over the same characteristics of the equivalent ordinal measure – the median.)
I rejected the use of rankings when I first looked at the international data, a matter so obvious to me that I argued the case only briefly in In Stormy Seas . Yet a decade later they were fashionable, although I detect some progress in the last year. Perhaps the analysis is entering the quantitative primers.
There is one further complication, explained below. But to foreshadow it: the aggregates, even if accurate, do not mean a lot. One needs to know about what is happening by sector.
Returning from the University of Melbourne, I settled into a routine of doing enough contract work to make a living, using the spare time to do public interest research. I began to see Bryan Philpott who had just retired from the McCarthy Chair of Economics at the nearby Victoria University of Wellington, weekly when we were both there, and we had grand old chats, sharing analysis, discussing recent developments, exchanging research findings, reviewing policies. It was the very collegial relationship which should go on within any university economics department. But that belongs to another memoir.
The University’s treatment of Bryan, the most distinguished holder of the chair, after retirement was a puzzle. They gave him an office plus storage, and access to facilities like library and computing, and left him alone to continue his research program. There was, apparently, no recognition of the inherent value of the research program, nor of the enormous data base which Bryan had built up. They could have appointed staff who would work with him, or help preserve the data base which he continued to extend. It was almost as if they did not care about research and it was no surprise that when the performance based research assessments were reported in 2004, the university’s economics department was not among the top ones.
Bryan had been an earlier user of the ASSP, but by the late 1960s he had moved on working with ‘Computational General Equilibrium’ (CGE) Models which were empirically based sectoral representations of the economy, and therefore had inter-sectoral prices. For this memoir’s purposes this is all we need to know about them.
What probably drove Bryan away from a crude ASSP position, in addition to his general intelligence and enthusiasm, was trying to model the economy for the 1968 National Development Conference, following the 1967 devaluation precipitated by the 1966 wool price collapse. I was not there, but I read the reports in the 1970s, and was struck by how those involved, Bryan included, were struggling to get their head around what happens when a number of major prices in the economy change dramatically. The answers are obvious enough now to me – and no doubt now to them – but the resolution of the challenge required moving away from a single commodity view of the world to a more disaggregated one.
Not only did the university fail to support the research program – which, alas, died with Bryan in 2000 – but neither did officialdom. One might say that they no longer needed the research and transferred the funding to other priorities, although they left an impression that they abandoned Bryan because his research contradicted their policy premises. ( In fact the predictions based on his conclusions were rather closer to actual outcome that the optimistic forecasts of the conventional wisdom.)
Moreover, at a later stage overseas consultants were hired to carry out the sort of modelling that Bryan and his students were expert at. You will have to ask the hirers why they did so, but while the contractors were competent, their lack of knowledge of the New Zealand economy often left much to be desired.
I recall Bryan coming back from a university seminar on some Business Roundtable commissioned research – I was not invited – remarking that the researchers had found that the burden of taxation had risen following the reforms of the 1980s. Now one does not have to be a Business Roundtable extremist to know that just has to be a nonsense, because one of the successes of the period was the substantial improvement in the efficiency of the tax system. (What happened to equity is more contentious.) This judgement is based on orthodox economics, and of course the foreign researchers were unquestionably orthodox. Bryan went on that the research assumed leisure as a response to tax rates, so it treated the rise in unemployment in the late 1980s as overtaxed workers choosing leisure (rather than due to economic stagnation arising from the over-valued exchange rate). When I looked at the paper, he proved to be quite right, a timely reminder that this man in his late 70s was still perceptive relative to overseas contractors, who knew little about the New Zealand economy.
One of the features of New Zealand is that once inside, people get a second chance following a cockup. So the overseas contractors were contracted by the Treasury to measure Solow’s index of aggregate technical progress, again an expertise of Bryan (who also did it by sector by sector, a much more revealing exercise). Again it was Bryan who first drew my attention to their use of an invalid capital series, a result of their not understanding the New Zealand data. It raised serious doubts of the validity of the conclusion, since ultimately the calculation is based on just three series – output, labour and capital – so if any is of poor quality the result is likely to be undermined. (Later work by me showed the output series was flawed and by Simon Chapple that the labour series was also flawed. Even so, the work is still quoted.)
I have read most of the research reports by Bryan and his students – and learned much. I have already mentioned the closure problem and the sensitivity of the outcome to the way it was resolved. The other obvious lesson was the practical importance of sectoral analysis. Some CGE modelling runs resulted in behaviour close enough to an ASSP single commodity outcome, but in other runs the outcomes were very different from a prediction of what was happening based on a single commodity economy.
Suppose New Zealand’s production per worker (properly measured) is less than the OECD average. So what? GDP is an aggregate of numerous businesses which are grouped into sectors. At the very least we want to know which sectors – as a proxy for the akin businesses – are below the OECD average, and which are above. Looking at only the aggregate implicitly assumes all business must have equally poor productivity. That surely cannot be true. We know it is not true in the case of the government sector, because it is measured so that all OECD government sectors are equally productive after adjusting for the measured quality of the workforce (a faute de mieux I hasten to add). And we would be astonished if the New Zealand farming sector had a lower productivity than the OECD average. So which sectors are letting us down and why?
International sectoral comparisons are a relatively new area of research, so one can but conjecture answers to this question, as we ponder each sector. It may not be manufacturing – the usual suspect – once we adjust for its composition, for subsector by subsector New Zealand manufacturing may be relatively productive. In any case, would lower productivity manufacturing be sufficient to explain the entire 15 percent or so reduction? (The answer is no, for given its size it would have to have zero productivity if all other sectors were average.)
I shant go through the other sectors, even though I once tried to do international comparisons. At the end of the day, we just dont know. My guess is there are some sectors which are relatively weak, in some cases because of distribution costs in a low density country with a rugged terrain and a strait in the middle, that for various reasons our industry composition may be unfavourable, and that the international income estimate substantially underestimates true New Zealand output (relative to the OECD average). Suppose I am right. What does one make of the policy proposals – such as cutting corporation tax – which pop out of the ASSP, albeit with no logical connection? Should not my musings be at least explored?
This of course, does not explain the relatively low GDP growth rate over the period. That is addressed in the next subsection.
When I left the NZIER in mid 1986, the economy was also struggling. It had a brief expansion fuelled by a fiscal stimulus, monetary slackness, and a wild unsustainable financial boom, consequential on the abandonment of the old regulatory framework and the failure to impose another one. Ian Cross once told me that he had reviewed the financial pages in early 1987 and that only two writers said that the sharemarket prices were then too high. The other was Muldoon – the strange company one keeps. Any claim to fame I have here was due to my interpreting some others’ research which suggested the P/E ratios of shares – the price of shares relative to the earnings they were generating – were wildly above their long term average. I confess to I naively assuming that the reported earnings per share was honest. In fact they were seriously overestimated, so my forebodings of overpriced shares were on the conservative side.
Underpinning my scepticism of the P/E ratios was the recognition that analyses promising a far better future economic performance were a nonsense. They were based on a faulty understanding of what had happened in the past (ignoring a major reason for the slower growth was the terms of trade fall); they over-estimated the gains from liberalisation (I had to be cautious here, because the literature and data I looked at might have been wrong, although events subsequently confirmed them); the theoretical underpinnings of macroeconomic policy was a nonsense.
My one serious forecast failure was that I did not believe that the policy would be so quickly successful at squeezing out inflation. This was because I did not foresee the breaking of past income and price relativities under the depressed conditions which were to dominate the economy in the late 1980s and early 1990s, together with a ruthless ignoring of equity in the policy framework (something which has still not been fully reversed). Even so, this outcome was consistent with the theoretical model I used in my income distribution work. What I failed to predict was the change in institutional relations.
One important institutional change was the destruction of the traditional public sector to private sector pay relativities in the late 1980s. They had been undermined by the 1982-1984 wage freeze, restored shortly thereafter, and undermined again by the 1988 State Sector Act, reinforced by the 1989 Public Finance Act. Another change, crucial for the disinflation, was the high real exchange rate. It impacted directly by lowering the prices of imports (and also exportables which were consumed locally) thereby directly reducing the consumer price index which was a key linkage in the inflationary spiral mechanism, and indirectly because cheaper imports forced down the price of local producers who competed with them causing unemployment which restrained wage demands
Inevitably the economy was to grow more slowly, although I confess to having been being surprised at the severity of the stagnation which followed the 1987 sharemarket crash – for six successive years, per capita GDP was to fall. Now of course this was not directly predicted from my pre-1981 version of SOME, since there was no major terms of trade fall in this period (they were broadly constant). However, as we saw in the previous section, I had begun to generalise the model. The impact of the terms of trade fall was it affected the profitability of the tradeable sector. Anything else which similarly impacted on export profitability – say a widespread outbreak of foot and mouth disease – would have a similar effect on economic growth. In the late 1980s the shock was the hike in the real exchange rate.
(I was not adverse to its removal of border protection, which also reduces the profitability of import substitution, albeit I wanted a slower pace than occurred and with more support to those restructured. However it was gross stupidity to remove protection while hiking the real exchange rate.)
Faced by a the high exchange rate impacting on profitability and cash flow, businesses in the tradeable sector closed down or slowed down investment (including market development). So export growth slowed down too, while imports rose. Slower economic growth followed.
This did not come to me as an instantaneous insight so much as the SOME paradigm evolved as the economy stagnated. In real time there were alarms and excursions over particular policy issues which can be traced through my various publications, not all of what were irrelevant to the model’s development.
One of the most curious was the mother-of-all-budgets in 1991. The available evidence suggests the economy was moving into the recovery phase of the cycle at the end of 1990. The November 1990 elected National government had instituted a package in December, which unnerved business to the extent that it held over its investment plans aborting the weak recovery and continuing the stagnation. Undismayed the government promised further major cuts of expenditure in its 1991 Budget. It seemed to be an overreaction, a conclusion supported by the graph below which suggested the economy was pushed below its GDP growth track of the 1990s.
I , and a handful of other economists, predicted the economy would contract. It did, but not nearly as much as I had expected. By 1992 the answer was becoming evident. The government had lost fiscal control, with tax revenue falling faster than government spending so the budget deficit increased when it had been expected to decrease. I know, I know, I should have thought of that possibility, but I was not involved in any forecasting team at that time and I relied on the consensus forecast that the deficit would narrow. So the mother-of-all budgets turned into the mother-of-all deficits, saving the economy from an even further serious contraction. (It also enabled the government to claim it had met it promise of halving the unemployment rate. To do this they had to first double the unemployment rate.)
Probably – I have never seen a systematic study – these cuts together with the fiscal measures of the late 1980s had enabled the government accounts to swing into surplus when economic growth recommenced a couple of years later. The surplus has largely continued (it being mainly used to finance government investment). But the cuts impacted on some people harshly. Mike Moore use to say that the government deficit had been replaced by a social deficit.
The SOME paradigm argues that a government surplus may be necessary as a part of keeping the real exchange rate down and exporting profitable. It is related to the two-gap theory in which the internal savings deficit is offset by the external current account deficit. While the theory is not perfect, the sucking of foreign savings through the external account to offset the domestic savings deficiency tends to drive up the exchange rate.
I had conceived of this by the mid 1980s and used to worry the life out of one economist as I mulled over its details. He was the best macro-economist I knew in New Zealand, but sadly his university did not value him, so he returned to Australia where his achievements continued, including becoming the head of department of one of Australia’s most prestigious economic departments. Later my sounding board was Bryan Philpott, and I was cheered when he began to accept the logic of the theory. Like many older economists he had grown up with the traditional Keynesian policy prescription that the government had to run an internal deficit. Keynes largely wrote in the context of a closed economy, whereas SOME is open to international trade. In which case, the Keynesian prescription becomes ‘manage the fiscal stance’ – rather than the traditional monetarist position that the deficit/surplus should be set at zero.
The question of the fiscal stance is not a sidetrack. True, the ASSP does not have a fiscal or monetary component. That it is why it is supply-side. But the alternative – SOME – does, and with that demand-side it is a much more comprehensive account of the economy.
Sometimes one may be aware of the crucial experiment almost at the time it occurs, as I was over the 1974 identification of a change in the economy in the mid to late 1960s, even if I did not know where it would lead. On the other hand, the experiment is not at all understood. For instance the ‘Ruth Cohen’ curiosum was identified in the early 1950s, and hung around for a decade after which it was seen to be an early example of the capital reversing which had so disturbed me in the mid-1970s. So it takes time for a curiosum to be recognised as the ‘crucial experiment’. So while I apologise, one need not be surprised that I cannot recall when the one I am about to relate first happened.
I do know I published it a Listener column in December 1994. I usually let any research findings stew for some time before I write a column. Especially since this was no ‘eureka’ event it probably stewed for a while, although I doubt it was as early as 1993. The most likely identification was when I was completing the final draft of the version of In Stormy Seas which went to the publisher.
I was greatly touched by an historian who read my columns as a means of getting a feel for the events of the day: perhaps others will read through them to trace the development of my thinking (after allowing for the lags discussed in the previous paragraph). However, one must not overestimate their influence. A loud economist said he never read my columns, so the journalist he told reported to me, and then proceeded to explain where I was wrong. On the basis of the account he gave of my views, he was absolutely correct: he had obviously not read my columns.
Even so, it is not unusual to get feedback from economists admiring something I have written, perhaps engaging with the analysis, while on occasions some idea I wrote up – which I thought original, but may not have been – would shortly after enter the economic debate, albeit without attribution. On the other hand some ideas are so revolutionary they get constantly overlooked, in this case for almost a decade, by which time we had more data.
Since this is a memoir, I might be permitted a diversion. In 1985, when I was Director of the NZIER and just before the floating of the exchange rate, I was visited by some of the OECD delegation on their annual or biennial review of the economy. They asked me what I thought was going on. As we interacted I got the impression they were worried about something– in later I thought it might be that they thought the officials they have talked to were too ideological, but this is only speculation. At the end of the hour they rang for a taxi, and in the small conversation while we waited, I said I was puzzled about what was happening and asked if there was another economy’s experience, that I could look to. There was a silence – I felt that I had made some unpardonable gaffe of international protocol: one does not ask visiting officials questions, they ask you. And after what seemed an interminable period they said but one word ‘Chile’.
So I went and look at the economic literature on Chile. It was useful, but then most international comparisons were. The one thing I remember was it took a darned long time for an economy subject to rapid liberalisation to get back to normal – seven years in the case of Chile.
So in late 1992 I began looking for the signs of a return to growth. And there was the evidence of a cyclic upturn! As I recall I was one of the first economists to mention it in public, using a standard economic term that we were in ‘the recovery stage of the business cycle’, that is the swingup from the bottom or trough. The term ‘recovery’ was seized on by politicians to mean that New Zealand was now in a phase of strong sustainable growth.
Admittedly it looked a long term recovery if you did not think about it too closely since economic growth was strong in 1994 and 1995. I could not see that such growth was sustainable, and took the view that much was a catchup from the contraction caused by National’s 1990 and 1991 fiscal stance. And so it proved to be, slowing down in 1996, followed by a contraction in 1997 in part because the Reserve Bank over-tightened its monetary stance in a response to the Asian crisis.
And so the glee that at last the rogernomics and ruthanasia reforms were working was replaced by the realisation that New Zealand was still subject to cyclical fluctuations and was not growing markedly faster than the rest of the world. But optimism springs eternal, and in the early 2000s the growth spurt was again being attributed to the reforms.
In a way they may be right. The characteristic of the Labour-coalition governments of the early 2000s was that they stuck to the moderate reforms and reversed, where they could, the extreme ones, and did not get too ideological. Even so, New Zealand was at a much lower GDP level relative to the rich world, than it had been when the reforms started.
This is an update on the original graph.
Go to Part II which explains this graph.