Putting it off for later

The first paper of my PhD considers the issue of end of fiscal year government spending. Governments regularly spend a large portion of their annual budgets in the last few months of the fiscal year. As an example 16% of UK government capital expenditure is done in the final month of the fiscal year. The Northern Ireland government spends more than 3 times as much on capital goods in the final month as compared to the average month. In 2014-2015 UK government capital spending rose in the last month of the year by £3.9 billion which is more than the cost of the U.S.S. Ronald Reagan Supercarrier.

A plot showing time in year against amount of spending. This shows greater spending later in the fiscal year


This figure shows average Northern Ireland departmental spending from 2008-2009 to 2012-2013 (including all Northern Ireland government departments). On the x axis are the fiscal year months from April to March while on the y axis is spending for that month divided by the monthly average. There are two series for capital expenditure and current expenditure. It can be seen that both are increasing over the fiscal year with capital exhibiting particularly large spending spikes.

This kind of spending is regularly criticized for being wasteful, for instance a 1980 US Senate subcommittee estimated that 2% of US government contract expenditure was wasted due to rushing at the end of the fiscal year. Recent research by Liebman and Mahoney (2013) (henceforth I will refer to this paper as LM) has also backed up this view using more modern data on the value of US government IT spending.

But what is causing it? We can note that any explanation needs to have two features. The first is that there is some force which encourages departments to hold back some money in the early months of the year. The second is some force which encourages departments to spend this held back money at the end of the year.

The LM view is that this kind of spending is caused by government departments building up a rainy day fund (or "precautionary savings") to use in case of sudden expenses that occur. In this model the force for holding back funds is uncertainty. The force to spend the money at the end of the fiscal year comes from expiring budgets that mean departments cannot save money from one fiscal year to the next. The LM solution to the problem is to allow government departments to "rollover" funds from one fiscal year to the next. They argue that with this policy in place departments will build up a rainy day fund once and then hang onto it and thus there will be no end of fiscal year spending spikes.

This is a interesting paper and the explanation for end of fiscal year spending that they propose is intuitive. My analysis suggests however that precautionary savings and expiring budgets provide at best a partial explanation. The first bit of evidence is that the UK did implement rollover budgeting starting in the 1998-1999 fiscal year and heightened end of fiscal year spending did not seem to change as a result. Indeed in every fiscal year since final month government capital spending has been greater than the monthly average (it is usually a bit more than 50% higher). This is quite significant as it suggests that if rollover does not abate end of year spending then expiring budgets are not underlying the problem. This can be seen in the below figure.

A plot of end of year spending amount for each fiscal year from 1990-1991 to 2013-2014. It shows higher spending at the end of every fiscal year in this period.


Here you can see UK final month spending for every year from 1990-1991 to 2013-2014. On the y axis of this figure is March spending divided by the average month of each fiscal year while on the x axis are all of the fiscal years. There are three spending series: government cash outlays; capital expenditure and current expenditure. There are three budgetary regimes in this period: No rollover where money could not be saved into the next fiscal year; End of Year Flexibility (EYF) where full rollover was allowed and finally Budgetary Exchange System (BES) which allowed rollover but with more restrictions than the EYF system. The key takeway from this figure is that the implementation of rollover budgeting did not seem to impact end of fiscal year spending spikes.

The second bit of evidence I assemble comes from a cross section of Northern Ireland government departments from 2008-2013. The LM theory would suggest that departments that have the most uncertainty would build up the most precautionary savings. These departments would then exhibit the greatest spending spikes at the end of the fiscal year. I test this with my Northern Ireland government departments sample and find that actually the departments with the least uncertainty have the greatest spending spikes which is opposite to the implication of the LM model. The third bit of evidence comes from the amount of uncertainty you need to have in order to explain the amount of money departments hold back to spend at the end of the year. When this is examined closely the implication is that government departments face demand for their services which is many times less predictable than inflation, unemployment or the prevailing interest rate. The implied level of uncertainty seems unrealistic in the real world.

I propose a different explanation based around procrastination. Here public servants hold back some money because there is effort in spending it and they would prefer to put off the effort. At the end of the fiscal year they are judged (in part) by their usage of their budget and so they rush to spend money in order to look good for their managers. I show this explanation is more consistent with the three features of spending noted in the preceding two paragraphs than the precautionary saving model.

Whether precautionary savings or procrastination is the correct explanation is important for determining how to fix the problem. While the precautionary savings explanation would suggest end of year spending could be disincentivised by allowing government to rollover their funds between fiscal years this approach will not help if end of year spending is caused by procrastination. So instead I propose a more direct way in which a parliament can encourage departments to not procrastinate.

My solution is a kind of tax on departmental spending that increases throughout the fiscal year[1]. This means that it is relatively cheaper for government departments to spend in the early months of the fiscal year relative to the last months of the fiscal year. To give a simple example consider a government department with a budget of £120 that faced a tax of 0% in the first six months of the year and 20% in the second six months of the year. This means that if the government department spent their entire budget in the first six months of the year they could buy £120 worth of stuff but if they spent it in the second six months of the year they could only buy £100 pounds worth of stuff. This gives an incentive for these departments to avoid procrastination. It also means that if departments do procrastinate then at least some of this rushed spending is returned to treasury through these "taxes".

So how beneficial could such a tax be? This is a hard question as there are two things to consider. One is some money wasted at the end of the fiscal year could be saved with a mechanism like this tax. The second is that some value might be getting lost if departments have a good spending opportunity but the parliament will not give funds to the department for fears they will procrastinate and subsequently spend the money poorly. That is to say there is an opportunity cost of end of year spending in that it deters parliaments from giving budgets when it would be better for them to do so (in the absence of end of year spending problems).

To try and get an idea of the total value that a tax could deliver I calibrated an economic model of departmental procrastination with spending data from Northern Ireland. I then inserted my tax into this model and saw how government departments reacted to it. What I found is that the value of government spending increased by more than 10% in every variation of the model that I considered. Thus the recommendation of this paper is that such a tax should be implemented to deter end of fiscal year spending spikes and encourage more even spending throughout the fiscal year.


[1] Of course this is not really a tax as the taxed money goes from a government department's budget into general treasury. It could alternatively be thought of as a budgetary reduction through time but I personally find that thinking about it kind of like a tax is more intuitive.