Kerry Sloan is a summer intern who has been placed at SPICe as part of the Economic Futures program which is run by the Fraser Allander Institute. The Economic Futures programme aims to encourage economics graduates to get involved in applied economics by placing them in an institution to do a research project. Kerry’s project at SPICe has been to study the behavioural responses to income tax changes and she wrote the following blog as part of her project.
Scottish income tax rates diverged from the rest of the UK in the 2017-18 tax year after non-savings and non-dividend (NSND) income tax was devolved to the Scottish Parliament. The Scottish Government and the Scottish Fiscal Commission (SFC) regularly publish estimates of how policy changes will affect income tax receipts. These estimates usually consider behavioural responses to changes in income tax policies. Increases in income tax rates tend to raise additional revenue, but not as much as might be expected. This is because taxpayers sometimes change their behaviour in response to changes in income tax policy. This blog will explain what these behavioural responses are and how they are forecasted.
What are behavioural responses?
There are three types of behavioural responses to a change in income tax:
- Changes in the intensive margin are where taxpayers change their efforts to earn more money or their engagement in tax avoidance schemes. For example, this type of response can include working fewer hours. Typically, changes in the intensive margin are a response to changes in the marginal tax rate (the rate of tax paid on the next pound earned).
- Changes in the extensive margin are where a taxpayer changes where they live for tax purposes, or they exit the labour market. Examples of this type of response include migration, choosing to study or becoming an unpaid carer. These types of responses are typically driven by changes in the average rate of tax paid on all income earned.
- Forestalling is when a taxpayer has an opportunity to change when they receive their income in response to changes in the income tax rates. For example, a taxpayer who knows that income tax will increase next year may choose to bring forward some of their income to this current tax year so that they can pay a lower tax on it.
Is there evidence of behavioural responses in Scotland?
If employees have any control over their income level, hours worked or methods of payment, then we may see evidence of a behavioural response via “bunching”. Bunching is where there is a large mass of individuals earning incomes just below an income tax threshold. Bunching can be identified by examining the distribution of incomes earned in Scotland.
Taxpayers with the ability to change their behaviour in response to changes in the tax rates might bunch around tax thresholds because there is a higher tax on the next pound earned when the employee moves into the next tax band. This increase makes working more hours less appealing compared to alternatives like increased leisure time. If the tax rate did not increase, then these individuals may have chosen to work more hours. However, the increase in the tax rate induces a behavioural response to reduce their working hours so that they do not earn more than the next income tax threshold. Alternatively, if the individual has the capacity to do so, they may choose to receive this additional income via other methods of payment, like dividend payments which are subject to a different, UK controlled, tax rate.
These responses are most likely in sectors of work where individuals have more control over their hours worked, income levels or methods of payment. For example, we would expect smaller behavioural responses in public sector work, due to the nationally negotiated pay bands, and larger behavioural responses in part-time work, as there may be more control over the hours worked. Bunching only captures the changes in the intensive margin as it does not identify if someone has left the labour market, or the tax jurisdiction, due to income tax changes.
The graph above shows the distribution of incomes in the construction sector in Scotland in the 2019-2020 tax year and is based on data from the Survey of Personal Incomes. In this sector, we can see some evidence of bunching around income tax thresholds, particularly at the £12,571 threshold where they tax rate moves from no tax to 19%. There are also smaller bunches before the £25,689 threshold and the £43,663 threshold. This indicates that there are behavioural responses to income tax changes.
Why are measures of behavioural responses so uncertain?
The SFC forecasts income tax revenues to inform the setting of the Scottish budget. However, there is a crucial difficulty involved in measuring the behavioural response. This is because there are not two scenarios to compare. The first scenario, where there is a change in tax, does occur. But the opposite scenario, where taxes are not changed, obviously does not happen simultaneously, so cannot be directly observed.
Therefore, comparisons between the two situations cannot be made. This means that if tax rates are increased and then taxable income decreases, it is hard to isolate how much of that decrease was the result of individuals responding to the tax change, and how much of it was due to other factors, meaning that it would have occurred in the absence of the tax change too.
One of the most common examples here would be when taxable income may have also decreased due to an economic shock, like job losses. In this case, any reduction in taxable income is not purely a result of individuals responding to the higher tax rates. This difficulty is at the heart of why measuring these behavioural responses is so uncertain.
How are behavioural responses measured?
To forecast revenue raised by Scottish income tax policies, the SFC quantifies Taxable Income Elasticities (TIEs). A TIE is the percentage change in the taxable income that occurs in response to a one percent change in the net-of-tax rate (one minus the tax rate). If the TIE is 0.45, then a 10% increase in the net-of-tax rate leads to a 4.5% increase in the taxable income, because 0.45 multiplied by 10 is 4.5. The TIE is expected to be positive, as this means that as the tax rate decreases (meaning the net-of-tax rate increases), taxable income is expected to increase. Larger TIEs indicate larger behavioural responses to income tax changes. The TIE allows the SFC to forecast how much revenue will be raised as a result of changes in the income tax rates in Scotland.
A lot of research has been done on TIEs. In all this research there are three areas of agreement:
- First, that the TIE is highly uncertain and highly sensitive to the methods used to measure it.
- Second, that the behavioural response is significant and therefore must be included as part of the costing for a policy change.
- Finally, that the behavioural response is largest for those with the highest incomes.
However, the exact value of the TIE is highly uncertain within the research, with prominent papers having measured TIEs ranging from 0.1 to 1.
The SFC chose the TIEs that they would use by comparing Scotland to the UK. HMRC had already done statistical work to calculate TIEs for the UK before the devolution of NSND income tax rates to Scotland. The SFC set out a few distinguishing characteristics that would differentiate Scotland’s behavioural response in comparison to these estimates done by the HMRC for the UK.
The opportunity to migrate from Scotland to the rest of the UK is considered to be larger than the opportunity to migrate from the UK to abroad. This would increase the behavioural response in Scotland relative to the TIEs estimated for the UK, as the ability to migrate in response to a tax change is larger. Further, the Scottish Parliament’s powers in respect of income tax relate to NSND income only. Any income shifting behaviour away from NSND income, like being paid in dividends, would result in a total loss in tax revenues for Scotland, as the UK government receives the tax revenues on dividends. From the perspective of the SFC, this increases the behavioural response in Scotland, relative to the UK. This is because there is a larger loss in tax revenue from this income shifting behaviour.
As a result of these characteristics, the SFC decided to use larger TIEs for Scotland than the TIEs used by the HMRC for the UK. It is therefore the SFC’s judgement that the behavioural response to a given policy change is larger when considering Scotland alone than it is when the UK is considered as a whole. The SFC, and HMRC, use different TIEs for different levels of income. High income individuals are more responsive as they often have more opportunity to change their behaviour, so the TIEs for that group are higher. In Scotland, the average of these TIEs used by the SFC for the top rate taxpayers is 0.51, which is higher than the UK’s estimate, by HMRC, of 0.48. The bigger TIE in Scotland reflects the SFC’s prediction of a larger behavioural response in Scotland.
How sensitive are revenues raised by policy changes to the assumptions of the researcher?
Using the SFC’s assumptions for TIEs we can estimate the expected revenue raised by a change in tax policies with and without behavioural responses being accounted for. Based on survey data from the Family Resources Survey, if the top rate of tax was increased from 47% to 48%, we would expect to raise around £30 million if there was no change in taxable income. However, these TIEs would predict that around £25 million of that revenue would be lost due to behavioural responses. This leaves an estimated revenue from this tax policy of around £5 million. This large drop in expected revenue highlights the importance of accounting for behavioural responses to income tax changes, especially where policy changes are directed at the high-income groups.
As noted earlier, there is a large amount of uncertainty around these estimates of the TIEs. If each income group’s TIE was 0.05 higher than the TIEs used by the SFC, this would reduce the expected revenues raised by this tax policy to zero, as the behavioural response would be expected to fully offset the revenues raised. This shows how sensitive the expected revenues of a tax policy are to the TIE used.
This sensitivity to the researcher’s assumptions about the size of the behavioural response, and the uncertainty around what the true behavioural response is, means that estimates of expected revenues from policy changes that include behavioural responses should be taken with a large degree of caution.
Kerry Sloan, Economic Futures Intern, Financial Scrutiny Unit