Keynesians vs monetarists A Granger Causality match Guy Lion April 2006 Hey Keynes, your [Tax] Cuts are no uppercut. Milton, where is your [Money] Velocity? I am Clive Granger who decides Causality
Introduction <ul><li>We will use Granger causality to referee a match in economic theory between Keynesians and monetarists. </li></ul><ul><li>Both claim their respective tools are best to manage GDP. </li></ul><ul><li>We will test which tools better “Granger cause” quarterly GDP changes. </li></ul>
Keynesian economics <ul><li>John Maynard Keynes (1883 – 1946) stated that a counter-cyclical fiscal policy was effective in managing the business cycle. </li></ul><ul><li>This means running a Budget Deficit when the economy is decelerating. And, running a Budget Surplus when economy is accelerating. </li></ul><ul><li>Keynes suggested government (net) spending affected aggregate demand without affecting price (inflation). </li></ul>
Monetarism <ul><li>Milton Friedman (1912 - ) stated the Money Supply has a direct impact on economic growth and inflation. </li></ul><ul><li>Milton relied on Irving Fisher’s (1867–1947) Quantity Theory: MV = PQ or </li></ul><ul><li>Money Supply x Velocity of Money = Price x Quantity. </li></ul><ul><li>The best way to manage the economy is to target a conservative Money Supply growth and stick to it. This way, you control inflation. </li></ul>
Keynesians vs Monetarists fighting stands <ul><li>Keynesians do not believe monetary variables (M1, M2, M3) affect macroeconomic variables (GDP, CPI). This may be due to change in Velocity of Money being a countervailing force to change in Money Supply. Thus, for them “money does not matter.” </li></ul><ul><li>For monetarists, fiscal policies are not effective because interest rate is a countervailing force to fiscal stimulus (the “crowding out” theory). And, because of the Quantity Theory [MV = PQ] Money Supply does have a direct impact on macroeconomic variables (GDP, CPI). Thus, for them “money is everything.” </li></ul>
A Budget Deficit proxy variable <ul><li>Budget Deficit data is annual. That’s a problem. </li></ul><ul><li>As an alternative we looked at Treasury levels from one quarter to the next. </li></ul><ul><li>Disaggregate Interest refinancing from Deficit financing as shown below. </li></ul>
Granger Causality review <ul><li>Develop a Base case autoregressive model using dependent variable and its lagged values as independent variable. </li></ul><ul><li>Develop a Test case model by adding a second lagged independent variable you want to test. </li></ul><ul><li>Calculate the square of the residual errors for the two models and run a t test (unpaired) to check if the residuals are significantly lower when you add tested second variable. [We will run the non-parametric Mann-Whitney test to observe P value differences]. </li></ul><ul><li>Redo steps 1 through 3, but reverse the direction. By comparing the tests significance or P value, you can see if A Granger causes B more than B Granger causes A. </li></ul>We will skip step 4 because we are not interested on whether GDP impacts fiscal or monetary policies, but on whether fiscal or monetary policies have a greater impact on GDP.
Selecting the lagged variable for GDP Data source: quarterly data since second quarter 1959 to 3d quarter 2005. Source: BEA (GDP), Flow of Funds (Treasury -> Deficit Financing), Federal Reserve (M2).
Granger Causality GDP Output P values for both test models are too high. In the Mann-Whitney test, the Milton model is going in the wrong direction. The avg. rank went up. That’s bad.