Tuesday, February 8, 2011

Estimating the size of the multiplier

There has been a fair bit of discussion in places like the Wall Street Journal and the Financial Times about the Keynesian multiplier. Economists like Robert Barro insist the multiplier is insignificant and below one. He bases his argument in part on analysing data from the Second World war era to make his case. But there are a number of problems implicit in that data which alters significantly the value of the multiplier depending upon which years are subject to analysis and which years are used as the base year from the point of view of inflation. Quite clearly the greater the slack in the economy and higher the initial rate of unemployment the more likely it is for the multiplier to be larger since at lower rates of unemployment and at full employment there is the likelihood of  state expenditures  competing with private sector investment and possibly raising rates of interest.

However if we begin our analysis using 1939 as a base and examine the years 1939 to 1942 when unemployment was still elevated we can get a fair idea of what the multiplier was by examining the GDP data and establishing a base year for inflation. The rate of unemployment from 1939 to 1945 in the U.S. was as follows:

1939 17.2 %
1940 14.6
1941 9.9
1942 4.7
1943 1.9
1944 1.2
1945 1.9

So by focusing on the years 1939 to 1942 we can evaluate the multiplier in a more unbiased way because for three of those years unemployment was very elevated. The years 1943 to 1945 were years of extremely low unemployment and because of this we can expect the multiplier to be much lower in those years. If we use 1942 as the base year 100 then a dollar in 1939 was worth 1.11 in 1942 prices since inflation averaged 3.45 % per year from 1939 to 1942. Adjusting the GDP data to this base year of 1942 yields the following;

federal government expenditures in constant 1942 dollars  

1939     6.06 billion
1942    52.027

difference an increase of 45.97 billion in constant 1942 dollars.

U.S. GDP in constant 1942 dollars

1939  100.34 billion
1942  159.6

difference an increase of 59.26 billion.

The multiplier 59.26 /45.97 is 1.29.

Data source: U.S. Department of Commerce for GDP data and U.S. Bureau of Labour Statistics for unemployment data.


  1. Hey Dr. Chorney,

    Keep doing what you're doing.

  2. Harold, I thought you might be interested in knowing that, according to TD Bank, 3 out of the 4 Canadian regions with the most vulnerable household sector also happen to be the provinces with the lowest debt/gdp ratios (http://www.td.com/economics/special/db0211_householddebt.pdf)

    (see bottom of page 2 for a list of provincial debt/gdp ratios: http://www.td.com/economics/budgets/govt_budget_feb11.pdf).

    In my view, it's looking like "sound" government finances is being achieved at the expense of household balances.

    Any thoughts?



  3. Giosofat, thanks for the comment. Unfortunately the url you give does not work as the docs have been moved somewhere. but the general principle you describe makes some sense, although one has also to check the rate of unemployment in the region concerned. If its higher than the norm that indicates pressure on household income and therefore the liklihood of higher household indebtedness. but generally if you pursue a policy of fiscal austerity in an economy with higher unemployment the result will likely be slower growth and greater difficulties for households.

  4. Interestingly enough, unemployment is pretty low in these provinces. But more surprisingly is the fact that the personal saving rate in one of the provinces identified as most vulnerable (Alberta) now stands at over 15%!

    I was thinking more in terms of the approach used by Wynne Godley (and other post-keynesians) on sectoral flows and financial balances. Martin Wolf of the FT has also been using a lot recently. The framework is based on the source and use of funds, and it holds that in a context in which trade is low, corporate savings are high, and government debts and deficits are low, household financial balances have no choice but to be effected negatively.

    (Regarding the links, they work fine if you remove the brackets.)