Matt Nolan on the Top 10 misleading economic facts

Today's Top 10 is a guest post from Matt Nolan, an economist at Infometrics, and an author at the blog site TVHE.

As always, we welcome your additions in the comment stream below or via email to david.chaston@interest.co.nz.

And if you're interested in contributing the occasional Top 10 yourself, contact gareth.vaughan@interest.co.nz.

See all previous Top 10s here.

Economics is a discipline that can be used to make sense of things going on in the social world. However, sometimes the economic facts we rely on to describe what is happening can be misleading - or may have changed since they were first noted. Here I’m going to chat about 10 economic facts that can be a bit misleading - not necessarily false, but misleading without context. 

There are two things to keep in mind from this process. Firstly, we must be careful not to jump on single facts uncritically, the world and social issues are too complicated to be stuck in one sound bite. But secondly, this does not imply that our feelings or beliefs should replace facts - anecdotes are merely biassed and single pieces of data that we use to understand the world in the absence of any other way of thinking. As soon as we have data, and a critical way of investigating it, this should take precedence over our biases. Ultimately, both these things tell us not to expect certainty when it comes to understanding the world around us - and to accept the limits of our own knowledge. 

1) Labour’s share of income, relative to capital owners, is in decline due to rising amounts of capital. 

Once we accept that the share of income that goes to either labour or capital can vary, it has become common for people to state that labour’s share of income in the economy has fallen. And with physical capital per worker rising while this has occurred it has led many to say that capital is replacing labour. However, this is an argument we need to be careful of (something we will discuss in a later point). 

Although many people say that labour has been in decline since the 1970s, there are significant data comparability issues that need to be taken into account with this work. More recent work which does account for this suggests that - in the case of the United States - the actual decline is more recent with labour's share of national income starting its decline during the 2000s. Now this was not a period of significant policy or market power change around the world, but instead a period of significant technological change.

The above paper suggests that it is labour-augmenting technological change that has led to a lower share of income going to labour. Now ignore the terminology and have a look at what has happened around us - the increasing utilisation of computers has meant that people can now do a number of tasks more quickly than they could previously, this is the type of change the econospeak represents. This is something many of us have experienced in our daily lives. 

Why does this lead to a lower labour share? Well it implies that the amount of “effective labour” is growing more quickly than the amount working age population/hours of work suggests. Given the difficulty substituting between labour and capital, this implies that this technical change must lower the relative price of labour - wages relative to the return on capital. 

But isn’t this labour augmenting technical change largely just a form of capital? We could suggest looking at it this way, with this type of capital highly substitutable for labour. However, even in this way having some capital that is not substitutable and some that is makes policy considerations a lot more complex - given that taxing capital reduces labour’s share in the first case and increases it in the second case. In practice the tax will largely fall on physical capital, which would in turn lower labour’s income share further! 

What about New Zealand? The decline in the income share has been more moderate here - and in the 2000s virtually nothing happened! Also remember that New Zealand has had a poor productivity performance and that our utilisation of ICT has been poor - this is consistent with the above argument for why labour shares have declined.  

And finally, remember that these are discussions of the shares of income. If New Zealand was to utilise ICT more fully we may almost all become more wealthy, but the wealth may become more unequally distributed - these are consequences society is trading off between. 

2) 47% of jobs are at risk from automation. 

After Frey and Osborne (2013) suggested that 47% of jobs were at risk from automation, journalists and talking heads went into a frenzy.

Although that made great headlines, a follow up study noted that there were significant issues with this result. The initial study took the idea that entire occupations would be replaced by automation, when it is actually specific tasks that will be automated. Furthermore, they focused only on the possibility of automation - not whether such automation is economically feasible! By considering the automation of tasks, Arntz, Greogry, and Zierahn (2016) found that it was 9% of jobs that were at risk of automation. Furthermore, for this automation to occur there would need to be significant changes in the organisation of firms before such automation was economically feasible.  

More deeply, even if 9% of the tasks humans currently do at work could be replaced in an economically feasible manner, this tells us nothing about new tasks that can be created. The introduction of the loom didn’t create permanent unemployment when it made many handweavers non-economical, but it did lead to a temporary period of unemployment as social structures had to change to employ these people to meet different needs and wants. 

Technological change will lead to significant adjustments in the way we do things, but it is not going to knock out the jobs half of us do overnight! 

3) Migration pushes down wages. 

One of the “accepted facts” behind current events in the US and UK is the idea that immigrants are driving down wages for people within these countries. However, there are two reasons why the truth is more complicated than this: 

1. It depends on whether migrant labour acts as a complement to native labour - or a substitute.
2. The short-term and long-term impact differs - over the long term the organisation of firms, the types of jobs on offer, and amount of physical capital adjust to the higher population due to inward migration. 

There have been many many studies into the short-term impact of migration on wages - with a myriad of studies providing empirical evidence of whatever conclusion you would like. As a result, I think it is appropriate to look at a “meta-analysis” that averages over a large series of these individual studies. According to this by Longhi, Nijkamp, and Poot the weight of evidence shows a very small negative impact on native wages from migration (eg increasing the proportion of the labour force that is made up of migrants by 1 percentage point - which is huge - would reduce wages by 0.119%), furthermore most of the wage competition is in jobs that migrants already have rather than native workers. 

Given what is going on in Britain though, I was interested in Britain specific information. Looking on the Oxford University website I found this cool write-up. Again, the average wage effect was small.

UK research suggests that immigration has a small impact on average wages of existing workers but more significant effects along the wage distribution: low-wage workers lose while medium and high-paid workers gain. 

Low income workers in the UK tend to work in jobs where migrant labour are substitutes (they perform the same tasks), while the highly skilled work in London that migrants perform tends to be complementary to native labour (they perform tasks that increase the productivity of native born workers tasks) - as a result, the wage impact is very different. In the UK, inward migration has been of largely skilled labour, but the short-term economic insecurity of those who complained does not appear to be without merit. 

Now, in the long-term job opportunities change and firms seeing a larger pool of labour increase capital investment. In that way, the long-term impact on wages is likely to be positive. But as Brexit shows, ignoring that there can be short-term losers from migration doesn’t change the fact that there can be - a situation New Zealand has more than anyone else taken into account with our fairly restrictive points system. 

How is this “fact” misleading? The evidence shows that, when migration involves complementary skills then even in the short term it can provide a win-win scenario. With New Zealand’s focus on pulling people in who fill skill gaps, migration's impact on our labour market seems like an undeniably positive force!

4) Underlying inequality in access to resources leapt up in the 1980s. 

The time was the 1980s. A period of social and economic upheaval in the Western World had taken place. And policy makers radically changed the nature of the contract between the state and its citizens. 

This change in policies has been blamed on the sudden surge in income inequality during 1987-88. And although policy changes will have been partially behind this change, there is something that has long vexed me about those figures. For some reason the fact so much of the lift in the level of inequality by varying measures occurred in a single year, or over a short series of years, is never reported - and it screams for an explanation. 

Why, when looking at gross incomes, does most of the change happen in a single year? People keep blaming the tax cuts that took place in that year … but this isn’t net of tax income, this is gross and market income. The burden of tax in these cases is shared between an employer and employee - so in theory those who received bigger tax cuts (all other things equal) should have seen lower market wages.

However, between April 1985 and April 1987 we had seen significant changes to the income tax system. Fringe benefit taxes were introduced, definitions of taxable income were broadened, a comprehensive Goods and Services Tax came into place, and the top tax rate was slashed. 

This will have led to a bunch of changes to how people reported their own income in a household survey:

1. High income earners will substitute previously untaxed fringe benefits (cars and holidays) for cash. As a result, this will turn up as new income for high income earners.
2. With many tax loopholes closed and top tax rates lower there is less scope and incentive for tax avoidance. Although the survey is confidential it is still run by government - and so previously high income earners would have been more interested in understating their income earned. 

These are significant changes that will have influenced the reporting of income data. If this, or other survey based changes (sampling for the survey also changed during this period), was the reason why the Gini suddenly rose, then a lot of the increase in income inequality we keep talking about didn’t actually happen. 

This is reinforced by what has happened to inequality in reported consumption. Ball and Creedy (2015) discuss the inequality in both incomes and expenditure according to the Household Economic Survey. Using the Gini coefficient, inequality in expenditures does not rise over the last 30 years - unlike the sharp increase in market and disposable income inequality during the late 1980s. As expenditure by households gives us an indication of what they believe their lifetime income is expected to be, this tells us a very different story regarding changes in the inequality of resources in New Zealand during this period. 

5) Virtually no-one was unemployed during the 1950-70s.

It is factually true that, during the 1950s to the 1970s the number of people collecting the unemployment benefit, and as a result the unemployment rate, was much lower than it has been over the past 30 years. But does this mean that the pre-distributionalist, and nuclear family centric, policies of the past were better - or even better for those out of work? I’m not so sure. 

The common joke about the early 1950s was that the Minister of Labour knew every member of the unemployed by name, and with only 2 people collecting the unemployment benefit on 31 March 1952 this could well be true! 

However, the social model of 1950-70 that social policy, including the definition of unemployment, was set against was completely different to that of today. A good outline of this can be found in A Civilised Community. 

The reforms of the 1980s are often painted as some “neo-liberal revolution to extract resources for the rich” - whatever that is supposed to mean. But one of the core changes was that of removing loopholes and treating individuals equally. Prior to reforms during the 1970s and 1980s, individuals were treated very differently based upon the industry they worked in and whether they were in the “right” type of family. 

Can someone be counted as unemployed when they were not going to be given the unemployment benefit? In Maori communities people were often ineligible for “living communally”, many women were ineligible for support when they came out of abusive relationships because they “should be trying to make it work with their husband”. A low unemployment rate because the state is unwilling to support those in need isn’t a positive thing.

However, this is not the only reason reported unemployment was low. Prior to the the mid-1970s New Zealand was strongly “pre-distributionalist”. With government job schemes used to keep those at risk of unemployment in work. Although some of the schemes were interested in transitioning people into employment and developing skills, by the 1970s most were accused of simply being a payment for not really working - in other words they were the unemployment benefit under another name. Looking at unemployment figures and excluding these people artificially lowers the measured unemployment rate - in a way it was data manipulation. 

With the New Zealand economy struggling in the late 1970s, and household formation switching from typical “nuclear families” to allow a larger variety of household types, the unemployment benefit (and newly founded domestic purposes benefit) started to be used at levels that are more comparable to that of today. This occurred prior to the Fourth Labour Government coming into power, and represents a change in social attitudes to taking up these benefits - a change policymakers have been reacting to, rather than “causing” ever since. 

The question of how we work with those who are long-term unemployed is an important one. But the myth that this didn’t exist in some golden age in the past and was created by policy makers does not help us answer this. 

6) Comparing LCI (labour cost index) to Inflation to show little real wage growth. 

Every time the labour market data is released one of the newspapers compares the labour cost index to the consumer price index to point out that the average person hasn’t experienced a pay rise over the past year. Sometimes consumer prices rise more than labour costs, sometimes less, but the key concern is that it just isn’t enough - all us wage earners are merely moving sideways. 

Except this isn’t what the data means at all. 

The labour cost index is a measure of the cost of a fixed quantity of labour for a firm - so it excludes productivity improvements and changes in the types of tasks and roles the average individual undertakes. The LCI is meant to measure “inflation expectations in the labour market” - by definition it is supposed to grow in a similar way to the CPI! 

If we are actually interested in considering real wage growth, we take a measure of the wage growth faced by households, such as average hourly wages in the Quarterly Employment Survey. We would then divide it by CPI to get real wages. This gives us the following graph of annual average real wages.

Average hourly wages have been rising almost consistently - with the recent dip partly the result of the increase in GST. As taxes were cut to compensate for this, net of tax wages will not have seen the same dip. This tells us what has actually happened to average real wages in the economy - no matter how much people don’t want to believe it.

7) The minimum wage will always cause lower employment. 

For economists the idea that setting a price floor would create unemployment is incredibly natural and intuitive - as there will be people who are willing to work at that wage, but firms who are unwilling to utilise their labour at that price. Here it is important to remember how unemployment is defined - it isn’t just that people aren’t working, but that people who want to work at a given wage can’t find work. The first part of considering this is to ask if minimum wages do constrain employment opportunities. 

However, like all things this is complicated. It depends upon whether this minimum wage is actually binding and whether firms that hire minimum wage workers have significant “market power”. If it is not binding or firms do have market power, then the minimum wage need not be a cause of unemployment or lower employment. 

To think about the issue we will focus on where most of the research has been - the United States. Although early evidence pointed to minimum wages reducing employment, Card and Krueger (1995) found that there was a publication bias towards this result. 

Applying meta-analysis (as we did above with migration) there is a suggestion that, at current levels of the minimum wage the negative impact on employment is negligible. 

8) The minimum wage will not always cause lower employment. 

However, in turn the suggestion that minimum wages do not reduce employment appears overused. David Neumark notes (here, here, and here) that the estimated impact of the minimum wage does appear to be negative - and it the loss of employment primarily falls upon those with low skill/training levels.

Note how complicated the discussion of the impact of minimum wages is when only discussing what happens initially. We haven’t talked about the long-run effects (as firms substitute away from low skilled workers, or low skilled workers get an incentive to train). And we haven’t talked about situations with much higher minimum wages (eg New Zealand). 

In the end, the debate about the appropriateness of the minimum wage has to be about more than whether it will lead to lower or higher employment. As stated here, we need to know “why” this is the case to know what minimum wages actually mean for the welfare of citizens - and to know whether these historic patterns would continue to hold if we were to change minimum wage policy. 

9) Inheritance payments drive income inequality.

Following the release of Thomas Piketty's Capital in the Twenty-First Century, many felt that the key driver of rising inequality in incomes was inheritance - especially since the text made such a big deal of discussing the inequity of inheritance. 

However, I fear there is some confusion here. Income inequality that we believe is unjust may well be the result of inheritance - but most of the income inequality we observe and talk about is not the result of this process! 

Wealth transfers (much broader than just inheritance) in the US makes up less than a 1/5th of a persons wealth, and this is less for wealthy people than for the middle classes! In relation to income inequality, wealth transfers tend to go to poorer members of the family - and so have been found to reduce income inequality. 

When we talk about the issue of inheritance we are often considering something else - the fact that people with high inherited social status are able to earn income and experience lift with lower risk and higher returns, all other things equal. Discussions of this can be found here, here, and in this book. However, the discussions tend to focus on inherited ability - rather than inherited status. As individuals we often feel positive about people using inherited ability (we do not tax people based on their height or strength) but feel uncomfortable that two people with the same ability could be treated differently based on status. 

This is the inherited inequality that most of us find socially unfair, and it is not inherently clear there is anything policy could do about it. Ultimately, income inequality is a very poor indicator of this (after all, people who are high status are more than willing to throw away income opportunities to keep status) - and we should be aware of this when discussing these issues.

10) The provinces are in decline.

 Everywhere I seem to go in Wellington nowadays I get told that the “provinces are in decline”. Personally I think the provinces are lovely, having grown up in provincial New Zealand, so whether there are issues is something that interests me.

Intuitively it makes sense that the forces of agglomeration, specialisation, and globalisation are making cities more and more relatively attractive - convincing people to move from the provinces to cities. 

This point was made most forcefully by Shamubeel Eaqub in his book “Growing Apart”. But for the rhetoric given in the book, something didn’t quite add up for me - yes some regions are going to be in decline, as their set of skills and resources become less valuable. But is it true that the provinces as a whole, or on average, are in trouble? Especially when we have seen a sharp increase in the price of agricultural products - which are goods made by provincial regions. 

Turns out that my colleagues at Infometrics, who focus significantly on regional economics, had some answers - with Benje Patterson, Gareth Kiernan, and Andrew Whiteford all penning articles on the issue in a short space of time. The takeaway seems to be that, in terms of economic indicators, the majority of provinces seem to be outperforming urban regions - furthermore, the disparity between urban and rural areas is low relative to other countries. 

There will always be structural change in a dynamic economy, and there is a case for policy makers to help support the transition for those who lose. But we need to be careful not to oversimplify by making large “winners and losers” groups - especially when those groups don’t necessarily match up with the needs that are required.