Moving on with my series on the Greek crisis I “d like to take a brief statistical look on consumption behavior during the last few years. Since the Euro era provided most of the population with a large supply of consumer credit, it is reasonable to try and test the Permanent Income Hypothesis. According to it, most people maintain a “permanent” view of their lifetime income and walk on a stable consumpion path smoothed by use of of savings and credit. One way to represent such a path is:
Ct+1 = Ct + ε
meaning that real consumption per head is equal to last year’s consumption plus a random variable. Regressing the above equation on the 1995 – 2011 period (using “Real Private final consumption expenditure per head”) leads to:
Ct+1 = 1.0114Ct + ε with very large R² and highly significant statistical testing.
In other words, it seems that consumption lead a smooth path with a 1.14% annual rise. Looking into residuals there are some interesting results:
The residuals are large and positive, especially after 2000. Ever since 2009 a “regime change” seems to have occured with very large negative residuals. Taking residual statistics for the 1996 – 2008 period, they have a mean of 236.88 and standard deviation of 115.28. Assuming they correspond to random errors which are normally distributed, all residuals should be positive with 95% probability. As a result, 2009 – 2011 residuals are clearly outliers and point to a large change.
The main possible reasons for this shift are:
- Credit constraints on households which are no longer able to smooth their consumption through credit.
- Permanent changes on future income expectations which are updated with new data each year and lead to even lower expectations.
Answering which of the above explains a larger part of the shift is quite important since, in case (1) is highly significant, a change in credit supply can provide for a short-term rebound in consumption. Unfortunately, no detailed data is available in order to reach into a conclusive view. Large shifts in permanent wages and pensions, 25% unemployment and more than 25% loss in property value points to a large change of permanent income and credit worth which should probably make the above changes long lasting.