Talk to VCTA Teachers Day, Melbourne, Monday, August 11, 2014
Productivity isn’t everything, but in the long run it is almost everything. A country’s ability to improve its standard of living over time depends almost entirely on its ability to raise its output per worker.
Paul Krugman, The Age of Diminishing Expectations (1994)
There are few economic topics more important than productivity improvement, or more often in the news. The economic managers frequently express concern about our seemingly poor performance in recent years and about the slower rate of growth in income per person we are likely to experience following in the deterioration in our terms of trade, and the decline in labour force participation caused by population ageing, unless we greatly improve our productivity performance.
And yet productivity is a tricky topic. A lot of people who use the term don’t actually understand what it means. Some people think it’s just a fancy word for production, or that when economists say we need higher productivity they mean we should be working harder. A lot of business people think productivity and profitability are pretty much the same thing. And also that productivity improvement is something governments do for businesses by making changes that make business life easier, rather than something that emerges from the efforts of their own business to survive and prosper. And even a lot of economists view productivity through the prism of their economic model, advocating certain ways of improving it, but not others. So much of what businesspeople say about productivity is rent-seeking and much of what economists say is propaganda.
Productivity is tricky in another respect: like a lot of basic concepts in economics - such as the NAIRU or the cyclically-adjusted budget balance - it’s conceptually simple but very hard to measure in practice. And once you’ve arrived at a measure you have to be very careful how you interpret it. There are circumstances where an improvement in productivity can have bad causes and a deterioration can have good causes. There are other times when a deterioration may be illusory or nothing to worry about. And, because of limitations in both measurement and our understanding of the phenomenon, it’s common for us never to be sure exactly why it’s improved or deteriorated.
I want to make sure you’re up to speed on all the ins and outs of productivity. I’ll go into the topic more deeply than you would need to with your students, but it’s never a bad thing for you to know more about the topic than they do.
Definition
Productivity is the ratio of output produced to inputs used, or output per unit in input. Colloquially, it’s what you get out compared with what you put in, or it’s bang for your buck. It’s a measure of productive (or technical) efficiency, necessary but not sufficient to achieve allocative (economic) efficiency.
The simplest, most commonly used - and probably least inaccurately measured - measure is the productivity of labour. You take the quantity of goods and services produce during a period (real GDP) and divide it by the number of hours of labour required to achieve that production. But you can improve the productivity of labour simply by giving workers more machines to work with. And this tells us nothing about the efficiency with which the economy’s physical capital is being used. So in recent years it has become fashionable to focus on a more sophisticated measure called ‘multi-factor productivity’.
MFP (sometimes called total-factor productivity) is the growth in real GDP (output) that can’t be explained by any increase in inputs of both labour and physical capital. So, in principle, multi-factor productivity should arise primarily from ‘technological progress’ - the invention of better physical technology and the discovery of better ways to organise the production of goods and services. It’s technological advance that does most to raise material living standards.
But though MFP is often regarded as a proxy for technological advance, in truth it measures much more than this. In this era of the knowledge economy, it will also reflect increases in human capital (the skill, education and training of the workforce), which could be significant. As well, it will reflect economies of scale and scope, gains from specialisation, changes in capacity utilisation, changes in the composition of industry (if, for example, low-productivity industries are growing faster than high-productivity industries), and even weather events and water availability.
Importance
If you accept the prevalent view among economists that we need as much economic growth as we can get to achieve an ever-increasing material standard of living - as simply measured by real income per person - then, as Paul Krugman has said in his famous quote, in the long run, productivity is almost everything. It’s productivity improvement that does most to cause real income to grow faster than the population. (For both social and environmental reasons, I don’t share this prevalent view. For me, quality of life is a more important objective than standard of living.)
Over the past four decades, growth in the productivity of labour accounted for about 80 pc of the growth in Australians’ real income per person. The main way this was achieved was by giving workers more machines (physical capital) to work with. But 36 percentage points of the 80 came from MFP improvement.
Measurement
The first problem with measuring productivity is that it’s cyclical, varying with the economy’s degree of capacity utilisation. Productivity is likely to rise strongly as the economy recovers from a recession, but then slow as the economy reaches full employment. In the downswing, productivity is likely to worsen as firms lose sales but hang on to their workforce and use their capital equipment to produce fewer products. Then, when the economy begins to recover, and sales and production increase without any more labour and capital being used, productivity shoots up again.
Combine this cyclicality with the fact that the national accounts are so heavily revised and you see that little significance should be attached to quarter-to-quarter changes in productivity. Even annual changes should not be taken too literally. This explains why the Bureau of Statistics prefers to measure productivity as the average over a ‘productivity cycle’ between years when the degree of capacity utilisation is similar. The length of these cycles differ, but have averaged about five years. Because it take so long before another cycle is completed, the econocrats just rely on using averages over the past few years.
The second problem with measuring productivity is that when an industry’s output isn’t sold in a market it’s not possible to measure its productivity. When the activities of governments and government-dominated sectors are included in GDP the value of their output is assumed to be equal to the cost of their inputs. That is, their productivity never changes. For this reason the bureau also calculates productivity in the ‘market sector’ which is 16 of the 19 industry categories, excluding education, health and public administration. This market sector accounts for about 83 pc of GDP. The bureau also calculates figures for a 12-industry market sector - accounting for two-thirds of GDP - and this narrower definition is the one used by the Productivity Commission.
While GDP and total hours worked can be measured with reasonable accuracy, thus making labour productivity figures reasonably reliable, the same can’t be said for MFP. Measuring the volume of capital inputs - ‘units of capital services’ - is particularly difficult. In truth, it is based on a lot of assumptions, which may or may not hold, making the figures pretty ropey. The other major problem with MFP is that it’s measured as a residual, so can be distorted by mismeasurement of any or all of the other factors. One important cause of mismeasurement is changes in the quality of either inputs or outputs. Combine this measurement problem with all the many possible components of MFP and it’s hard to draw many confident conclusions from its ups and downs.
Scorecard
According to figures calculated by the Reserve Bank in a speech by Glenn Stevens in July 2014, the productivity of labour for the whole economy improved at the trend rate of 2.1 pc a year over the 14 years to 2004, but slowed to 0.9 pc a year in the six years to 2010, before accelerating to 2.0 pc a year over the three and a quarter years to March 2014. This implies that, whatever the problem was in the second half of the noughties, it seems to have gone away.
But the figures for MFP in the 12-industry market sector presented in the Productivity Commission’s latest annual productivity update show a much less reassuring picture. The long-term rate of MFP improvement between 1973-74 and 2012-13 has averaged 0.7 pc a year. Between 2003-04 and 2007-08, however, the improvement deteriorated to minus 0.1 pc a year, essentially, four years of zero improvement. And over the five years to 2012-13 MFP actually worsened by 0.6 pc a year.
This is a very different story to that told by the labour productivity figures, and the Productivity Commission regards it as very concerning. But it’s also very puzzling. It’s highly unusual for labour productivity to be improving while MFP deteriorates, and for this to continue for a decade. With something so dramatic, you’d expect the problem to be glaringly apparent. But nothing sticks out. The formerly common claim that the problem was the alleged ‘reregulation’ of the labour market under Labor’s Fair Work, simply doesn’t fit. The rot began years before the Fair Work changes took effect from the beginning of 2010, which was about when labour productivity started performing normally. In any case, you’d hardly expect a supposed worsening in industrial relations to fit with an improvement in labour productivity but a worsening in MFP. About the only factor big enough to be a suspect in explaining this strange behaviour is the resources boom.
Explaining the deterioration in MFP
The Productivity Commission has analysed the figures by industry and identified three industry sectors as doing most to explain the deterioration: mining, utilities (electricity, gas and water) and manufacturing. It has closely examined each of these industries and identified many factors that, between them, do much to explain the apparent deterioration. But almost all of them fall under the headings of temporary factors that will right themselves in due course, factors that have worsened productivity without themselves being bad, factors beyond our control, or the productivity cost of pursuing other desirable policy goals.
The primary reason for mining’s poor productivity performance, which does much to explain the poor performance overall, is well known: the industry has poured a lot of inputs into building new mines and natural gas facilities which have yet to come on line and start contributing to output. When they do, the industry’s performance will improve markedly. However, the high prices for coal and iron ore made it profitable for miners to begin mining lower-grade or hard-to-reach deposits that were formerly sub-economic. The mining of these marginal deposits will permanently lower the industry’s level of productivity (because more inputs are need to win a given quantity of output), but it is good, not bad, that higher prices have allowed us to exploit more of our natural endowment.
In the utilities sector, the productivity of water has been hit by drought (reduced capacity utilisation), by the decision to build and then mothball desalination plants in every state capital, and by the lasting effect of water-use efficiency measures in reducing demand for water (eg piping to prevent evaporation in irrigation; better shower heads). The productivity of electricity has been reduced by greatly increased investment in the natural monopoly distribution network (‘poles and wires’) to meet peak levels of demand and excessively levels of reliability, at a time when demand for electricity has been falling for various, mainly good reasons.
The three industries within manufacturing that do most to explain its poor productivity performance are petroleum and chemicals, food and beverages, and metal products. Petroleum and chemicals’ poor performance is explained mainly by increased investment by petroleum refineries to meet new environmental standards - ie an unmeasured quality improvement. Metal products’ poor performance is explained partly by an expansion in alumina refining capacity which has yet to come on line. Food and beverages’ poor performance is partly explained by a change in consumers’ preferences in favour of products made in smaller-scale, more labour-intensive bakeries. Both petroleum and chemicals and metal products are suffering from reduced capacity utilisation rising from their loss of competitiveness caused by the high dollar.
The point of all this is that the startling MFP figures seem to have been produce by multitude of factors, many of which will correct themselves or are no cause for concern and some of which are beyond our control. That is, the commission’s analysis could identify ‘no overarching systemic reason for the decline’ and thus no problem or problems the government should be fixing, bar one: it’s clear defective price regulation in electricity is encouraging excessive investment in the distribution network.
In the light of all this, and remembering how prone MFP is to mismeasurement, I’m not persuaded we need to hit panic stations.
Bad productivity improvements and good productivity declines
Productivity is just a means to an end - the end of greater economic wellbeing - so it’s important not to treat it as an infallible indicator. Increases aren’t always good and decreases aren’t always bad; you need to examine the reasons for them. At the margin, labour and capital are substitutes in the production process. If wage rates get too high, firms may decide to invest in more labour-saving equipment. For the particular firm involved, this means fewer workers but the same or more output, leading to higher labour productivity. But is this a good thing if it has been caused by excessive wage rates? It’s the position that obtained during the Whitlam and Fraser years. After the Hawke government came to power in 1983 it used its accord with the ACTU to achieve a significant fall in real wages. The result was strong growth in employment and rapidly falling unemployment. But this also meant much weaker labour productivity improvement. Was this a bad thing?
Similarly, labour productivity is likely to deteriorate as an economy approaches full employment because employers are obliged to hire the remaining, less-qualified, less-productive workers. Does that make full employment a bad thing? And, as we’ve already seen, rising commodity prices encourage miners to exploit less attractive, harder-to-win mineral deposits. Is this bad?
Productivity, profits and wages
It’s important to remember that productivity is a ‘real’ concept. That is, it compares quantities, not prices or values - the quantity of output with the quantity of inputs. This is why productivity and profitability are quite different concepts. Because of this, it’s likely most firms don’t collect the data that would allow them to calculate changes in their productivity. And it’s arguable they don’t need to bother: their role is to pursue profits; what effect this has on the nation’s productivity is of no direct concern to them. All we know is that over the more than 200 years since the industrial revolution, firms’ pursuit of profits has led to almost continuous productivity improvement.
It’s less common these days, but you still hear employers arguing that wage rises can only be justified by the equivalent improvement in labour productivity. Wages settlements in excess of productivity growth will be inflationary. This is why it’s important to remember productivity is a ‘real’ concept. The employers are conveniently forgetting this. It’s only necessary for a rise in real wages to be justified by an equivalent increase in productivity. And only if real wages grow faster than productivity with this be inflationary.
This explains why the Reserve Bank’s unspoken ‘line in the sand’ for annual nominal wage growth is 4 pc: an inflation rate of 2.5 pc (mid-point of the inflation target) plus 1.5 pc for trend labour productivity growth. Only when nominal wage growth exceeds 4 pc is it likely to worsen inflation.
A point to note is that these are macro economy-wide figures. It’s neither necessary nor desirable that the real wage increases granted at each individual workplace be matched by productivity improvement at that workplace. This is the notion of ‘productivity bargaining’ which for a period of the Accord proved a useful device for encouraging unions to give up impractical demarcations and restrictive work practices, but provides no sound basis for continual bargaining at the enterprise level. This is because there are some firms with much scope for the removal of inefficiencies, but others with very little. Is it seriously suggested that workers at already-efficient firms receive no real wage rises? As well, there are high-productivity industries (usually those that are capital-intensive) and others than are low-productivity (often in the services sector). Should a clerk working in a high-productivity industry end up being paid a lot more than one in a low-productivity industry?
Fortunately, neither wage bargaining nor the labour market work like that. There are differences between the wages of workers with similar skills working in different industries, but they aren’t great - particularly when you remember than truck drivers in the Pilbara are being paid a special premium for working long hours at a remote and unpleasant location. In other words, workers in particular occupations are reasonably mobile between industries and so tend to be paid similar rates. Because workers’ wages don’t differ much according to the productivity of their industry, this operates as a mechanism by which improvements in national productivity are spread reasonably proportionately between labour and capital and reasonably proportionately between workers generally.
Productivity and employment
Especially when viewed from the perspective of particular industries or firms, efforts to improve productivity by automation and other labour-saving investments can be seen as job-destroying. This may seem true from a short-term, micro perspective, but it can’t be true economy-wide. It can’t be true because industries have been replacing men with machines for more than 200 years and the unemployment rate is only up to 6 pc, not 94 pc.
The explanation is simple: replacing workers with machines leads to increased productivity and increased productivity increases real income. The community is better off because outputs have increased relative to inputs. This increase in real income will be reflect in some combination of: higher profits for the owners of the business, higher wages for the remaining employees of the business or, if competitive pressures are strong, lower real prices for the products of the business. To the extent that real prices fall, customers are able to choose between using the increase in the purchasing power of their income to buy more of the business’s products or to buy more of other firms’ products.
The point is: improved productivity leads to increased real income; when that higher income is spent, jobs are created somewhere in the economy. This is why more than 200 years of investment in labour-saving equipment have not led to mass-unemployment. It’s also why, though non-economists speak of new technology ‘destroying’ jobs, economists prefer to say the jobs have been ‘displaced’ - moved from one industry to other industries.
If you look back at the history of our economy over the past century, you see technological advance making first agriculture and mining more capital-intensive, so that they are able to increase their output while accounting for an ever-falling share of the workforce, and since, the 1960s, a similar process occurring in manufacturing. All the employment growth has been in the services sector, with most of it in the more highly skilled occupations.
Productivity and immigration
The commitment of business people, politicians and even most economists to growth means almost all of them are believers in a high rate of population growth through high levels of immigration. It’s understandable for business people and politicians to believe in growth for growth’s sake - the bigger our economy the better. From a business perspective, the bigger the economy, the bigger their domestic market and so the easier it should be for them to increase their sales and profits.
But it’s not rational for economists to believe in growth for growth’s sake. They believe in faster economic growth only because it’s expected to lead to a higher material standard of living. But a higher standard of living requires higher income per person - that is, GDP has to grow faster than the population. If it doesn’t, nothing is gained. And if population were to grow faster than GDP, average living standards would fall.
Empirical research by the Productivity Commission and others over the years has concluded that immigration - even skilled immigration - does little or nothing to raise income per person. Why not? Because of the problem of productivity. An increased population requires increased provision of housing and public infrastructure: roads, public transport, schools, hospitals, police stations and all the rest. If this increased infrastructure investment fails to occur, the productivity of our infrastructure falls. This will be manifest in housing overcrowding, greater traffic congestion, longer hospital waiting times and so forth. Similarly, an increase in the workforce that’s not matched by an increase in the equipment workers are given to work with will lead to a decline in the average productivity of labour. In other words, and increase in the population requires a greater increase in capital investment - public and private - if it’s to lead to higher income per person.
This raises the old distinction between capital deepening and capital widening. Capital deepening involves the provision of greater capital equipment per worker - an increase in ‘capital intensity’ or the ratio of capital to labour - which will raise the productivity of labour. Capital widening involves increased investment to prevent an increase in the workforce causing a decline in capital equipment per worker - a decline in capital intensity - and a fall in the productivity of labour. Immigration does little to increase income per person because it requires a lot of investment just to stop average labour productivity falling.
Productivity and other advanced economies
It is well known to the economic managers, but not widely reported, that Australia is not alone in suffering slower productivity improvement since the turn of the century. Dr Philip Lowe, deputy governor of the Reserve Bank, observed in a speech in March 2014 that ‘productivity growth in many advanced economies has trended lower for some time now’. Labour productivity in the OECD economies averaged 1.9 pc over the 14 years to 2000, but 1.1 pc over the 13 years to 2013. Australia’s figures were virtually the same as this average, except that whereas the OECD average for the three years to 2013 fell further to 0.8 pc a year, Australia’s average recovered to 1.5 pc.
This common experience around the advanced economies suggests the underlying problem may be a worldwide slowdown in technological advance, but this is only conjecture.
Productivity and recessions
In my experience, the measures businesses take to lift their productivity - particularly by trying something new, reducing waste and abandoning unsuccessful projects and products - aren’t evenly distributed across the business cycle, but tend to come in a bunch, like busses. In the upswing of the cycle when profits are strong an optimism abounds, firms tend to expand, experimenting with new products and processes, and generally allowing staffing levels to creep up.
When the cycle turns and sales and profits slow down or fall, however, times get tougher and pessimism sets in. Firms seek to minimise the decline in profits by looking for ways to cut costs and reduce staff numbers. Unsuccessful processes, products and divisions, which should have been abandoned much earlier, are finally brought to an end. The least profitable firms in an industry may collapse or exit. Following Schumpeter, economists often refer to this as a process of “creative destruction”. What this means is that economic downturns often lead to surges in productivity.
When you remember that Australia hasn’t experienced a serious recession since the early 1990s, it make makes you wonder if a price of our good fortune isn’t a slow build-up of inefficient practices in the absence of a great reckoning that would have made our productivity figures look better. Along similar lines, however, it may well be that the pressure imposed on our tradables sector - particularly manufacturing and tourism - by the years of a high exchange rate has obliged those firms that have weathered the storm to become a lot more efficient than they were. The productivity debate has been led by people who imagine greater productivity to be a warm and cuddly thing. In reality, it’s often the product of adversity - firms forced to lift their game if they want to survive.
Productivity and microeconomic reform
It has become fashionable among economists to see a strong link between our productivity performance and microeconomic reform. Most economists attribute the surge in labour productivity and MFP in the second half of the 1990s to the effects of the Hawke-Keating government’s extensive micro reforms. But there’s no way they can prove this is the case. Examining productivity performance by industry doesn’t show much correlation with the major reform initiatives. And since the main reforms occurred in the second half of the 1980s, we’re being asked to believe the benefits of those reforms took about a decade to show up.
When the reform push was in full swing it was hoped the reforms would, by making the functioning of various industries more efficient and flexible, bring about a lasting increase in the rate of productivity improvement and economic growth. As we now know, however, the best the proponents of micro reform can claim is just a few years of above-average improvement, bringing about a once-only lift in the level of our productivity.
I find this a bit disillusioning. If micro reform to remove impediments to growth can’t lead to a lasting increase in the rate of growth, the benefits of micro reform are less than we initially imagined them to be. If the only way we can achieve even a temporary increase in economic growth is to come up with another dose of micro reforms, we’ll soon run out of reforms and willingness to pay the political and social price of reform. If repeated doses of reform are the only way we can maintain a reasonable rate of productivity improvement, there must be something badly amiss in our economy that no amount of reform is fixing.
My belief is that the peculiarities and limitations of the economists’ conventional model have warped their attitude towards productivity improvement. They know well that national productivity is a function of what goes on in each of the many firms that make up the economy, yet it is no part of the economists’ approach to study what happens within the individual firm or to imagine there is anything they or governments should say or do to affect what happens inside firms directly. Anything done may affect firm behaviour only indirectly via the monetary incentives firms face. Because the model assumes business always responds rationally to the incentives it faces, if there’s something lacking in the economy’s productivity performance the only available possibility is that something government is doing must be distorting the incentives facing private firms, causing them to operate less efficiently than they otherwise would. So governments should examine and reform their interventions in markets, reducing them in ways that encourage greater competition. So, problem with productivity = more micro reform.
But the model’s largely unnoticed effect on economists’ thinking about productivity doesn’t end there. Because the standard model’s unit of analysis is the individual - whether individual consumer or individual firm - it contains an inbuilt bias against collective action, including actions by governments. The model leads economists to the conclusion that, since the role of government is suspect, the desire for improvement in the economy’s functioning invariably leads in the direction of less government rather than more. Government intervention in particular markets may be reducing their efficiency, or the high tax rates needed to pay for a big public sector may be reducing incentives to work, save and invest.
By contrast, we know that with a knowledge economy, much productivity improvement should be coming from investment in human capital. The rise of the knowledge economy is prompting more of us to crowd into big cities where we can learn from face-to-face contact with each other. But this increases congestion and other infrastructure problems. So adequate - and effective - spending on infrastructure is an important part of maintaining productivity improvement.
This suggests there are two rival approaches to government efforts to improve productivity: measures that involve making government smaller, and measures that make it bigger. Guess which approach economists tend to focus on?
