Can the simple act of lowering taxes stimulate growth? We now know (thanks to Liz Truss) that, when unfunded, tax cuts can certainly trigger economic chaos. But even if they are properly funded, the question remains, will it really foster growth?
There are many who would argue that it would. Some even go so far as to present a policy of low taxation as a silver bullet – a golden ticket to growth and prosperity!
Bold claims indeed – but is it true? What is the evidence?
The search for evidence
How would we test the veracity of this concept? It is not uncommon to see people quote anecdotal examples to support the contention. But highlighting a particular instance where reduced tax has been followed by positive growth in a single country is potentially no more than a propaganda exercise. What about the big picture?
There are many countries in the world. Collectively they represent a wide variety of different economies and different tax regimes. In a great many cases we have access to a lot of historic data on growth, tax policy and so on. Surely it cannot be beyond the wit of man to compare tax policy to outcomes across many countries over time. Can it?
This is not as simple as it sounds but it is possible. The main problem is to make sure we compare apples with apples.
What is growth?
Firstly, we need to agree some kind of sensible definition of what we mean by ‘growth’. At a simple level we might look at growth in terms of GDP. However, GDP just tells us the total monetary value of an economy. Growth in GDP is, of course, good. However, it tells us very little about how wealthy people who live in that country are. That is because it takes no account of population size.
Think of it this way:
- 100 people living on an island earning $5k each per year,
- So collectively they earn $500k in a year.
- Only one person lives on the neighbouring island. He earns $100k a year.
- Which island is wealthier?
One island generates five times the money of the other. However, clearly the guy living alone on the second island is wealthier. For this reason, it is often better to look at per capita GDP (GDP per head of population) as a more accurate measure of wealth.
It might seem like a simple thing to compare taxation between one economy and another until you come to try and do it. It isn’t.
If you think about it, any given country has a wide variety of different taxes and tax rates. One might have high sales taxes and low income taxes. Some might have forms of taxes that few other countries have. In some countries individuals might pay limited personal tax but companies pay a lot (or vice versa).
Hence, what we need to do is look at the overall tax burden when all these different elements are bundled up. i.e., the proportion of wealth generated that is being taken in tax.
In order to compare like with like we probably ought not to look at particularly odd or weird situations. Ireland is a case in point. So much so that Nobel Prize winning economist Paul Krugman labelled the phenomenon ‘Leprechaun economics’. So, what happened?
In 2015, Apple changed the way it reported its accounts. It shuffled a large chunk of revenue, previously reported elsewhere, into Ireland. Now, suddenly, in a single quarter, Ireland’s national GDP jumped up 26%! This had nothing to do with Ireland and everything to do with an Apple accounting policy change.
This distortion makes it very misleading to look at the Irish economy in the same way as other economies.
Of course, the biggest ‘oddity’ of all in recent years has been covid. The pandemic has had a massive effect on global economies since it first struck in early 2020. Attempting to measure the impact of tax policy on growth in the period after 2020 would therefore be very difficult to say the least.
We often look at growth quarter by quarter or year on year. That is fine but it is nevertheless potentially just a snapshot. It can be distorted by unusual events that are unique to a particular country or year. This can present a picture in a particular period that is very different from the underlying trend.
To measure the impact of taxation on growth we really need to be thinking in terms of longer-term impacts – performance over a longer period of time than just a year. The sun coming out on one day does not a summer make.
One of the biggest problems in measuring growth between different economies is known as ‘base effect’.
So, what is base effect?
Imagine the following scenario:
- Brian earns $10k per year in his very first job.
- After his first year, he gets a pay rise and a promotion. He now earns $15k per year.
- Brian’s salary has grown by 50%!
- His friend’s mother Anne earns $200k per year.
- Anne also gets a pay rise at the end of the year. Her salary increases to $240k.
- Her pay has only gone up by 20% but Brian’s has gone up by 50%!
- So … is Brian doing better than Anne?
Of course not.
This is a classic example of base effect. It is the main reason why developing / emerging economies grow very fast compared to developed economies. They are starting from a very small base!
If we are going to meaningfully compare economic performance, we need to be mindful of distortions like base effect. Unfortunately, virtually all economies are of different sizes. Nevertheless, for a meaningful comparison we ideally need to compare economies at a similar stage of economic development.
The reality is that more economic information is available for some countries than others and some report slightly differently. The OECD strives to record comparable metrics for its member states and for several other key countries as far as is possible.
However, the fact remains that reasonably comparable information on tax and growth is not always available.
This naturally places limitations on the extent to which we can meaningfully compare different countries.
Designing a test to measure the impact of tax
So, how might we go about attempting to measure the impact of tax policy on growth? That, of course, sounds like a good title for an economics PhD thesis, which is not what we can realistically attempt in a short blog.
Nevertheless, it is possible to look at some high-level measures for a basket of countries to see if we can observe any patterns over time. So, with this in mind, let’s define our parameters:
- Time period: 2010-2019. This gives us 10 years of data to look at that should capture trends over a reasonably long period. It also has the merit of being the most recent period we can pick before covid starts to distort growth figures.
- We’ll measure the overall tax burden in comparison with growth in per capita GDP.
- To avoid the worst impacts of Base Effects, we’ll focus on a basket of developed economies (defined as having a per capita GDP in excess of $30k in 2010 at 2015 prices).
- This would potentially include Ireland, but due to the distortions unique to this country already mentioned, we will exclude them.
What tools can we use to find a pattern?
Let’s say we have data for 25 countries over 10 years. It tells us, in each case, the tax burden and the growth experienced. How do we know if there is a pattern, i.e., that they are inter-related in any way?
We have a couple of tests we can apply in the initial instance:
- Regression analysis
These tests measure slightly different things. You have probably heard of them and may have some familiarity with them. However, perhaps you are one of the many people who is not entirely clear on the specific differences between the two.
What is correlation and how does it work?
Correlation is a statistical technique that measures the strength of the relationship between two sets of data. It generates a number between -1 and +1 to indicate the strength of a relationship. Technically we call this the coefficient (or simply ‘r’).
A positive number indicates that a pattern exists and that, as one number increases, the second number was also observed to increase.
A negative number indicates the reverse – that one number declines as the other increases.
A number close to 1 indicates a very strong relationship. Close to 0 indicates a virtually non-existent relationship.
So, for example, we might compare number of ice cream sales to average temperature. If we see a correlation of +0.8, this tells us that there is a strong relationship and that as temperature rises, ice cream sales also rise.
Correlation and causation
Correlation is not causation of course. Correlation might tell us that ice cream sales rise when polar melt increases. But that does not mean that buying ice cream causes the ice caps to melt! In this example it just means that both are impacted by a third variable that we have failed to take into account – i.e. temperature.
In the case of what we’re trying to do here, a negative high correlation between tax burden and growth indicates a strong relationship between a high tax burden and poor growth.
As a general guide to the strength of a relationship we’d typically consider:
|Correlation coefficient||Strength of the relationship|
|1 or -1||Perfect!|
|0.7 or -0.7||Strong|
|0.5. or -0.5||Moderate|
|0.3 or -0.3||Weak|
What is regression analysis and how does it work?
Correlation seeks to measure the strength of a relationship but no more. Regression analysis goes one step further. It seeks to build a model to predict how one factor might change as a result of a change in another.
So, in this case, a regression analysis would seek to predict how an increase or decrease in the tax burden might impact growth.
Typically, regression produces two things. The first is a formula that you can use to predict an outcome. Hence, you can use a regression formula to tell you what growth you might expect if you set the tax burden at a particular level.
The second key output from a regression analysis is a measure of how reliably this formula can predict an outcome. The technical term for this measure is ‘R2’ (not to be confused with the ‘r’ we use in correlation). Even more confusingly, also like correlation, that number can be between 0 and 1. However it means something quite different.
If the number is 1, the equation is a strong fit to the data. However, if the number is zero, the equation is pretty much junk. So R2 simply tells us how well we can we predict growth rates based on the tax burden.
Comparing taxation levels with growth
In this analysis we compared a total of 25 OECD countries in the period 2010-2019. We looked at the average overall tax burden in each case over the period and compared it to growth GDP per capita over the same period. Ireland and countries with a GDP per capita under £30k were excluded.
The countries were: Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Finland, France, Germany, Iceland, Israel, Italy, Japan, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Slovenia, South Korea, Spain, Sweden, Switzerland, UK, USA.
This results from this analysis are shown in the scatter plot below:
So, do high levels of tax impede growth?
The correlation between tax burden and growth rates experienced is -0.4. This indicates there is a relationship between the two. High tax levels do tend to correlate with weaker growth. However, the relationship is moderate to weak. We can see this clearly from the plot. There are countries with a tax burden above 40% that experienced stronger growth than some with a tax burden under 30%.
In terms of regression, it is possible to plot a trend line that demonstrates how taxation in general might impact growth. This indicates, for example, that an economy with a taxation level at around 25% might expect a growth of around 15% over ten years. In contrast, the model suggests a tax rate at 45% might expect growth at only around half this level. However, this model is a terrible fit, with an R2 of only 0.16! We only need to compare the scatter plot to the trend line to see there are numerous exceptions to the rule. There are plenty of examples of countries experiencing half or double the predicted growth at various different tax levels.
All this suggests that the tax burden only has a limited depressing effect on growth. So an obsession with lowering taxes as a panacea remedy for delivering high growth is clearly naïve. It is just one of several factors that need to be considered and, quite possibly, not even the most important.
Strong growth is clearly possible in countries with high levels of tax. By the same token, having low tax rates does not guarantee strong growth by any means.
A more balanced view might be simply to say that tax has a limited depressive impact on growth. For this reason, we could argue that it is better to keep it lower than higher. But, by the same token, increasing the overall tax burden by 3% or even 4% would not necessarily depress growth. Growth might even be stimulated if the additional revenues raised were invested wisely.
The idea that lowering tax is a silver bullet for stimulating growth is therefore unsupportable.
Synchronix is a full-service market research consultancy. We work with data to generate market insight to drive evidence-based decision making.
We specialise in serving b2b and niche consumer markets in Science, Engineering, Technology, Construction, Distribution, and Agriculture. Understanding and dealing with technical and industrial markets is our forte and this enables us to draw a greater depth of insight than non-specialist agencies.
Take a look at our website to find out more about our services.