My argument is based upon a generic analysis of Stock-Flow Consistent (SFC) models, or in fact any reasonable model of fiscal policy. For simplicity, I will cover the case of a “no-growth” economy; if the economy is growing at a steady rate, one could derive a similar result. This analysis is covered in more detail in an earlier post; as that post noted, this is just a restatement or generalisation of existing results given in the text by Godley and Lavoie.
What we see is that if a new steady state arises with higher government interest payments, if the government keeps its fiscal parameters otherwise fixed, the end result is that unemployment and welfare payments will fall by the same amount, implying that economy will just have moved closer to full employment. In other words, the interest income channel replaces welfare state stabilisation of the economy. There is therefore no reason for the government to view this as a negative outcome.
For those who are uninterested in the details of the derivation, I will now just give the interesting implications.
- If doing a fiscal risk analysis for a sovereign welfare state, the use of the “primary budget balance” is to be avoided. If analysing a household, one can ignore the impact of household decisions on the wider economy. This is not possible for a central government. Interest payments are a transfer payment, and have an impact on growth.
- The secular fall in interest rates seen since the early 1980s has coincided with slower growth, and rising debt. The weakening interest income channel is one potential cause of this slowing.
One could argue that my argumentation is weakened by being based on steady state analysis; but it should be noted that actual economic behaviour is characterised by steady growth rate periods interspersed by recessions. These average out to steady state growth rates across the cycle, and can be approximated closely as an economic model moving from one steady state to another. (I do not need to assume that economic variables are in equilibrium (fixed); just that average growth rates are steady across the cycle.) And if one uses modern mainstream economic models, those models are driven by the expectations of model entities. Those modelled expectations will have to converge to some steady state condition in any computationally tractable model.
Derivation Of Results
I will assume the model economy has the following characteristics:
- The population is constant over time.
- Model parameters are eventually stable. This includes fiscal settings like tax rates, unemployment insurance payments rates, although levels can move up and down in line with nominal GDP.
- The price level is constant, or reverts towards a constant level over time.
- There is no productivity increase due to technology, and productivity enhancements due to capital have an effective maximum.
- There are welfare state transfer payments made to out-of-work households (transfer spending increases as unemployment rises).
- Model entities have a limit to the amount of financial assets they are willing to accumulate relative to their nominal incomes. This either represents a stock-flow norm, or else the result of an inter-temporal optimisation of consumption.
It is easy to see that nominal GDP in the model will vary around a relatively constant level, where that constant level is based on the average level of employment. And it should be underlined that I am not referring to a particular model; the result holds for any well-posed model with the above characteristics.
As my previous article noted, under these assumptions, the budget will have to balance “across the cycle”.
- If surpluses are run, the government bond market and the monetary base will disappear. A monetary economy could not survive this condition.
- If deficits are run, then government debt (or money) holdings will eventually become arbitrarily large versus nominal incomes. That is incompatible with assumption (6).
(Note: If the economy is growing in nominal GDP terms, the condition is that deficits have to be run to keep nominal financial asset holdings in line with income.)
When we examine a simplified budget condition, the budget being balanced implies that the following equation holds (at least on average across the cycle):
Taxes = (Program spending) + (Transfer Payments),
Taxes = (Program Spending) + (Net Interest Payments + Social Transfer Payments).
(Note: The term “Taxes” above excludes tax on interest payments for simplicity.)
We then see:
(Taxes) – (Program Spending) = (Net Interest Payments) + (Social Transfer Payments).
Assume the government wants to keep (non-interest) taxes and program spending constant, for reasons of political expediency. This implies that the left hand side of the equation is constant (across the cycle). And imagine we move from one steady state to another, with an increase in interest payments in the new steady state. This could be the result of an “exogenous” increase in the interest rate on government debt. The implication then is that Social Transfer Payments have to be lower. Although this may sound to be unattractive, it only implies that unemployment insurance and welfare payments will be lower. If the amounts paid for these programs are frozen, the implication can only be that the unemployment rate is lower in the new steady state.
In other words, an increase in the spending mix towards interest payments when fiscal policy settings have not otherwise changed implies that the economy has to be closer to full employment, and that social spending necessarily drops in importance. The government should view this as being an attractive outcome, and not a problem. Therefore, it should not care whether interest payments are an increasing percentage of spending.
As a technical note, it would be hard to replicate this analysis within the standard DSGE model framework. The reason is that the typical assumption is that there are no economically sensitive transfer payments. These models assume that there is a single Representative Household within the economy who never is laid off; the Representative Household just raises and lowers hours worked based on whether it decides it wants more vacation time or not. This means that unemployment insurance (or welfare) payments make no sense within the model, and so they do not appear. I believe that a model in which welfare payments have been defined out of existence has obvious drawbacks for analysing a welfare state.
There are limitations of this analysis.
- Firstly, if welfare and unemployment insurance spending are too small relative to government consumption, they will not be able to fall enough to make room for interest payments. For example, we had this sort of situation during World War II, when the governments largely nationalised their economies to allow massive wartime production. However, economically sensitive transfer payments are now relatively large in most of the developed economies, and provide room for expansion of interest payments.
- A second issue is that this was a steady state analysis. For shorter term dynamics, it is likely that the multiplier for interest payments is much smaller than the multiplier for social transfers, and so deficits will tend to be higher if interest payments replace social transfers.
Conclusion: You need to account for the impact of things like unemployment insurance when you model a modern welfare state. Additionally, the multiplier on interest payments also have to be taken into account when doing fiscal simulations.
(c) Brian Romanchuk 2013