VoxEU Column Global crisis Labour Markets Macroeconomic policy

Does fiscal policy matter? Is there a better way to reduce unemployment?

Can government spending help the economy recover from a recession by boosting job creation and lowering unemployment? Or is it a waste of money? This column addresses this question and others using a unique framework. It explains why fiscal policy was effective at ending the Great Depression but it argues that a big fiscal expansion may not be the best solution this time round.

Economic theories that lack an independent role for business and consumer confidence have difficulty explaining the cause of financial crises like the Great Depression or the Great Recession (e.g. Woodford 2003, Kydland and Prescott 1982).1 Both economic calamities were preceded by large drops in asset prices that were not associated with any fundamental trigger.

Because these models assume deviations of employment from the steady state are short-lived, they are characterised by unique steady-state equilibrium and as such cannot explain how today’s unemployment rate has remained above 8% for 2 years.

Farmer’s new model

For these reasons, we favour Farmer’s model in which recessions are caused by shocks to business and consumer confidence.2 Confidence is synonymous with the self-fulfilling beliefs of market participants and, within this framework, there is no unique natural rate of unemployment. Our model captures the intuition of David Altig of the Atlanta Fed:

...the current recovery seems so disappointing because we expect the pace of the recovery to bear some relationship to the depth of the downturn. That expectation, in turn, comes from a view of the world in which potential output proceeds in a more or less linear fashion, up and to the right. But what if that view is wrong and our potential is a sequence of more or less permanent "jumps" up and down, some of which are small and some of which are big?

  • Farmer’s (2009a) model codifies this idea by allowing for a continuum of steady-state equilibria, each of which is associated with its own unemployment rate. The economy's steady state can jump from one value to another.

Alternatively, the economy can get stuck in a socially inefficient low-employment equilibrium for an indefinite time period.

  • Farmer’s model does not contain a self-stabilising mechanism that restores full employment.

In contrast, new-Keynesian models, real business cycle models, and the Gertler-Trigari and Hall versions of search models all contain price adjustment mechanisms that automatically, and unrealistically, cause the unemployment rate to converge back towards a unique “natural rate.”

The standard way to close a model with unemployment is to use Nash bargaining to determine the wage. As in Farmer (2009a), we dispense with the Nash-bargaining equation and we assume instead that firms produce as much as is needed to meet aggregate demand. All agents in our model are price takers. They behave rationally and have rational expectations of future prices. The existence of a continuum of equilibrium unemployment rates is explained by matching frictions which lead to search externalities.

Figure 1. Unemployment and the S&P 500 index in the 1930s and 2000s

What makes the economy switch between different steady states? In our model, equilibria are selected by business and consumer confidence, reflected in the value of the stock market. When consumers lose confidence, asset values fall and there is a contraction in aggregate demand.

Figure 1 shows that there was a strong connection between the stock market and unemployment in the 1930s and the 2000s. In our interpretation, a loss of confidence, reflected in the 1929 crash, caused the Great Depression. Similarly, the Lehman Brothers bankruptcy in September of 2008 was the trigger that caused the Great Recession.

How we explain the role of government expenditure

In our paper, “Does Fiscal Policy Matter?” we extend Farmer’s (2009) model and use it to explain the Great Depression, 1929-1939, and the wartime recovery, 1940-1949 (see Farmer and Plotnikov 2011).

To operationalise our theory we assume that agents form an exogenous series of beliefs about future asset prices (Keynes called this the state of long-term expectations) and we proxy this sequence by the actual values of the S&P 500. We feed this sequence into our calibrated model and back out the implied values for unemployment, GDP, and consumption.

Curiously, at the end of the 1930s the relationship between unemployment and the S&P 500 disappeared. The S&P 500 kept falling but the unemployment rate recovered (see the left panel of Figure 2). How can this be possible in our model if the S&P 500 is the force that is driving the economy? Another variable must have changed substantially in the beginning of the 1940s to produce such a fast recovery. We argue that this variable was government expenditure. Between 1938 and 1945, government expenditure increased from 15% to 52% of GDP (see the right panel of Figure 2). As government expenditure increased, unemployment fell.

In earlier work, Farmer (2010a) showed that, in his model, a permanent fully anticipated fiscal expansion will have no effect on unemployment. In Farmer and Plotnikov (2011), we show that a temporary unanticipated fiscal expansion will reduce the unemployment rate. Since the fiscal expansion that paid for WWII was both temporary and unanticipated, it follows that Farmer’s model can account for the wartime recovery.

Figure 2. Role of government expenditures

Our main results

To check on the ability of our explanation to fit the facts from the WWII expansion, we input data on the S&P 500 and on government expenditures into our calibrated model and we compute the values of private consumption expenditures and the unemployment rate implied by the model, assuming that agents’ expectations are an independent driving force of recessions. The results can be seen in Figure 3. This Figure shows that, given its simplicity, the model does a good job at matching the actual consumption and unemployment data.

Figure 3. Model and data comparison

We also study the effects of fiscal stimulus on consumer welfare. We show that, holding fixed consumer and business confidence, a temporary unanticipated increase in government expenditure will decrease unemployment in the short run. However, if government purchases have no social value, a fiscal expansion leaves consumers worse off. As a group, households consume less but work more. Given the choice, households as a group would prefer a situation with higher unemployment but more consumption.


We find that Farmer’s model can account not only for the increase in unemployment during the Depression, but also for the recovery during WWII. But we offer a word of caution to those who advocate additional fiscal expansion as a solution to the current recession.

  • The fact that a temporary fiscal stimulus increased employment during WWII does not necessarily imply that it is the right policy in the current crisis.
  • In our model, welfare would decline in response to a fiscal expansion, even as unemployment falls, unless the government purchases goods that have significant social value.

This was clearly the case during WWII – but we do not advocate solving the unemployment problem today by increasing the size of the army. A clear case would need to be made that increased expenditures have social value, for example, by building new bridges or otherwise improving the public infrastructure.  

In our 2011 paper, it is critical to increase the value of confidence in the value of private wealth in order to restore jobs permanently. In the face of continued pessimistic beliefs about asset values, no amount of fiscal stimulus would be capable of restoring full employment. In our world, increased public spending can cause a reduction in private spending (economists call this crowding out) even when the economy is in the midst of a depression. For this reason, we believe that a better policy to reduce unemployment would be an asset market intervention of the kind suggested in Farmer (2009a, 2011b). Increasing business and consumer confidence by stabilizing the value of private wealth is an essential component of any recovery plan.


Farmer, Roger EA (2009a) “Confidence, Crashes and Animal Spirits”, NBER WP No. 14846, Economic Journal, forthcoming

Farmer, Roger EA (2009b) “What Keynes should have said”, VoxEu.org 4 February.

Farmer, Roger EA (2010a), Expectations, Employment and Prices, Oxford University Press

Farmer, Roger EA (2010b), How the Economy Works: Confidence, Crashes and Self-fulfilling Prophecies, Oxford University Press

Farmer, Roger EA (2010c), “Animal Spirits, Persistent Unemployment and the Belief Function”, NBER Working Paper 16522, CEPR Discussion Paper 8100

Farmer, Roger EA (2011a) “Animal Spirits, Rational Bubbles and Unemployment in an Old-Keynesian Model”, NBER WP # 13737, CEPR Discussion Paper 8439

Farmer, Roger EA (2011b) “A New Monetary and Fiscal Framework for Economic Stability”, 17th International Conference on Computing in Economics and Finance (CEF 2011), Plenary Lecture, Journal of Economic Dynamics and Control (forthcoming)

Farmer, Roger EA and Dmitry Plotnikov (2011), “Does Fiscal Policy Matter? Blinder and Solow Revisited”, NBER working paper #16644, and CEPR Discussion paper #6504, Macroeconomic Dynamics, forthcoming

Gertler, Mark and Antonella Trigari (2009), “Unemployment Fluctuations with Staggered Nash Bargaining”, Journal of Political Economy, 117:38-86

Hall, Robert (2011), “Large Employment Fluctuations with Product- and Labour-Market Equilibrium”, Stanford University mimeo

Keynes, John Maynard (1936), The General Theory of Employment, Interest and Money, MacMillan and Co.

Kydland, Finn E and Edward C Prescott (1982), “Time to Build and Aggregate Fluctuations”, Econometrica, 50:1347-1370

Woodford, Michael (2003), Interest and Prices: Foundations of a Theory of Monetary Policy, Princeton University Press.

1 The standard New-Keynesian and real business cycle models (Woodford 2003, Kydland and Prescott 1982) both contain a unique steady-state employment rate that is pinned down by fundamentals. Neither model contains unemployment and therefore cannot explain why unemployment has doubled in the current crisis. Gertler and Trigari (2009) and Hall (2011), have introduced unemployment into an otherwise standard macro model but they cling to the assumption that there is a unique natural rate of unemployment, explained solely by fundamentals. Models that maintain this assumption cannot credibly explain financial crises.

2 Farmer’s (2009, 2010a, 2010b, 2010c, 2011) model interprets Keynes’ General Theory (1936) in an original way. It reintroduces the idea, absent from new-Keynesian economics, that there is a continuum of steady-state unemployment rates. Farmer’s framework provides an independent role for business and consumer confidence. Confidence is treated as a fundamental that selects an equilibrium.

3,570 Reads