A great deal of analysis of fiscal policy is based upon analysing the trend in the primary budget balance. I explain in this post why this is a mistake, as the primary budget balance gives a misleading view of fiscal settings…
What Is the Primary Budget Balance, And Why People Use It
The primary budget balance is the government fiscal balance excluding interest payments. As an equation,
Overall Fiscal Deficit = (Primary Deficit) + (Government Interest Payments).Alternatively,
Primary Deficit = (Non-Interest Spending) – (Taxes).The implication is that interest payments are singled out as a special category of the budget.
Breaking out interest payments appears to make sense if you are interested in modelling monetary policy - which directly affects the level of interest payments - and are less interested in fiscal policy. This is a good description of the motivation for the creation of most Dynamic Stochastic General Equilibrium (DSGE) models, as these models are typically created for the use of central banks.
Within the context of many of these models, the primary budget balance is taken to be exogenous – determined outside the model. (Given that hundreds if not thousands of DSGE macro models have been churned out, there may be exceptions to that statement.) The interest component is determined by the monetary policy rule embedded within the model, based on the evolution of the model economy.
The standard working assumption is that the primary balance is the result of fiscal policymakers, and that they do not wish to depart from this set policy. For example, they do not want to be forced to raise or lower taxes, as that has political consequences. The same holds true for programme spending. The usual interpretation of holding the trajectory of the primary balance fixed is to see whether the current fiscal policy settings are "sustainable". (In practice, fiscal settings change regularly as the result of the political process.)
If we follow the assumption that the path of the primary budget balance has been fixed, we can then simulate what happens to debt levels based on interest rate scenarios. A typical result is that if one sets nominal rates growing greater than nominal GDP growth, debt levels can spiral out of control unless primary surpluses are run (“explosive debt dynamics”). As a result, this style of forecasting is an excellent way of generating scary headlines about fiscal policy. A cynic might insinuate that is exactly why the primary surplus has been used as an analysis tool.
Problems With The Primary Budget Balance
I will now list some of the problems associated with using the primary budget balance.
Primary Budget Balance Is A Function Of Economic Activity
The basic problem is that it makes very little sense for fiscal policymakers care about the primary balance. Taxes are not imposed in the form of absolute levels; they are almost always imposed as percentages of nominal incomes and activity (e.g., income and sales taxes). Media discussions may refer to dollar amounts that will be raised by a tax measure, but those dollar amounts are based on forecasts of taxable activity. As such, the tax component of the primary balance will shift based on the economic cycle, even if policy settings are unchanged. (Note: a progressive income tax system, which has tax rates rise as incomes increase, creates a greater sensitivity to nominal GDP than is the case for the simplified flat tax rate used in my simulations.)
Governments have slightly more control over programme spending, but even that will break down on horizons longer than a few years. Salaries of civil servants will presumably rise along with private sector salaries and costs of goods procured will rise with inflation. (Or actually drive inflation, if you are a Chartalist.)
However, transfer payments such as unemployment insurance are completely dependent upon the economic cycle. Even payments to retirees may vary according to the cycle; a weak economy could force many older workers into early retirement. (Although what we are seeing now is that the older cohorts are being forced to work longer, and that unemployment has been shifted towards younger workers. This is unfortunate is that these younger citizens end up being excluded from the labour market. This is a particularly perverse outcome given the lack of workers that is allegedly being caused by demographics.)
The net result is that the primary budget balance moves in a counter-cyclical fashion with the business cycle (deficits rise during recessions). The usual reaction function of central banks means that interest payments generally move in the opposite direction (they cut the policy rate during a recession). The time preference effects of an interest rate cut is generally viewed to override the impact of lower interest income, but there is a debate over that (see this theme article for a list of posts discussing this topic). In any event, developed market governments have a good deal of fixed coupon debt, so that overall interest payments only follow the average of the policy rate with a lag.
Interest Spending Is A Form Of Stimulus
The second area of analytical weakness is how interest payments are dealt with. It makes sense to do a stand-alone sensitivity analysis of an individual or firm’s finances with respect to interest costs. If their interest expenses rise, there will only be a negligible impact on national GDP, so it is safe to make an “all else equal assumption” with regards to interest rate scenarios. Such an assumption makes no sense for a central government; changes in its interest payments will have a measurable impact on GDP.
Interest payments are a form of stimulus to the economy. Interest spending probably has a low short-term multiplier, given that interest mainly flows to those with a high saving rate, but at least some of that money will be spent. As such, increasing interest spending reduces the need for other social transfers that act to stimulate the economy. (Demonstrating this was the point of my previous article on interest costs.)
One could argue that the consumption function in my model was too simplistic; a “proper” inter-temporal optimisation would generate different results. (See this post by Nick Edmonds which offers a summary of a recent controversy in the blogosphere over consumption functions.) However, I am unsure whether this criticism has merit with regards to my scenario involving a shock to interest payments.
I believe that it would be possible to generate a similar pair of scenario outcomes even if you replaced my Stock-Flow Consistent (SFC) model with a DSGE model. The exact economic trajectory might vary slightly, but the steady state condition that I focus on could presumably be generated by a suitably chosen set of model parameters. (It would certainly be difficult to prove that the contrary; i.e., that there exists no model that reflects standard DSGE assumptions that could generate a trajectory “close” to my simulation results.) The rise in interest rates in my model was a permanent level shift, and would have reflected a change in the time preference parameter. As such, the change in interest income was a permanent increase, and so this increase would be largely spent.
(If one attempted to re-create the simulation trajectories with a DSGE model, the very interesting question of Ricardian Equivalence would come up. Given the complexities associated with Ricardian Equivalence, I will discuss it in a later post.)
The paper “Fiscal Policy In A Stock-Flow Consistent Model”, by Wynne Godley and Marc Lavoie, gives a more formal explanation of how to approach fiscal policy using SFC models. They have a number of results within that paper, but here is one interesting point with regards to "explosive" debt dynamics:
Our simple SFC model can, however, provide us with a more surprising result. It is usually asserted that for the debt dynamics to remain sustainable, the real rate of interest must be lower than the real rate of growth of the economy for a given primary budget surplus to GDP ratio. If this condition is not fulfilled, the government needs to pursue a discretionary policy that aims to achieve a sufficiently large primary surplus. We can easily demonstrate that there are no such requirements in a fully-consistent stock-flow model such as ours.
They additionally argue that the emphasis on monetary policy is misplaced, and that fiscal policy could theoretically achieve the inflation and output targeting that has been currently delegated to central banks. However, that topic is well beyond the scope of this post.
(c) Brian Romanchuk 2013