Bogleheads

[Paul Douglas Boyer]

What asset classes correlate closest with inflation (CPI)?

I took the CPI numbers and added them to the correlation matrix (1972-2008) in Simba's backtest portfolio spreadsheet (see below), did a sort, and here is how it turns out:
(1 and -1 means perfect correlation, 0 means no correlation. Note that a negative correlation means the investment goes down when inflation goes up, right? So bet against it in times of inflation. )

[haberd]

Forget the mumbo-jumbo - and don't bet on inflation any time soon.

Stocks generally do well after inflation subsides, as the release of fear lets pent-up valuations catch up to the previous inflation. (See the 1950s and 1980s.) In the shorter term, they don't respond well to unsual inflation, as equity markets don't like uncertainty.
Commodities and cash do well for obvious reasons.
TIPS will probably do well relative to nominal bonds, though prices (not yields) are driven by inflation expectations, not necessarily actual inflation.

[grayfox]

[grayfox]

If I rather arbitrarily called those assets with magnitude of correlation less than 0.10 "Uncorrelated", then here is the exact same table broken into three tables: Positive, Uncorrelated, Negative. (So Uncorrelated means 10% or less of the return is x-explained by CPI-U.)

[richard]

[market timer]

I'd be interested to see how various asset classes correlate with the change in inflation expectations. It's the arrival of new information that influences asset prices.

[grayfox]

[stratton]

From Table 13-1 in The Handbook of Inflation Hedging Investments

[Jack]

I assume CPI correlations are calculated at the end of each year, but TIPS dividends lag by six months.

[craigr]

[Lauren Vignec]

Hello All,

Well, these data are certainly a little bit disappointing. Cash and gold have very low expected real returns (close to zero and zero) and over the time period they were the only two asset classes to moderately correlate with inflation. There was no free lunch if your concern was inflation protection.

The past doesn't predict the future, but in this case I think the past is probably a pretty good general guide. Or maybe I'm just pessimistic. The only bright spot I can really see is that maybe if you looked at monthly correlations the picture would look better for a few of the asset classes.

L

[Tramper Al]

[Lauren Vignec]

[Tramper Al]

[haberd]

I suspect TIPS, like nominal bonds, correlate to anticipated, not actual, inflation in the short term. For example, in the 1970s, when inflation was unexpectedly high, Treasuries had very low total returns. In the 1980s, the market expected continued inflation, so bond total returns were 5% over real. It would be interesting to see the historical "correlation lag", if it indeed exists.
I think that nominals, as well as TIPS, catch up to inflation over time, which is part of the reason I am a bit skeptical of TIPS as being some kind of revolutionary break-through for very long term investments. They are nice as an inflation insurance policy for people who are in the withdrawal stage and need a steady stream of constant real income year after year, but probably not as advantageous as some appear to think for long-term accumulators. Or so I think.

[Paul Douglas Boyer]

I have done some more analysis on this and posted the results on my blog here:

Investments for Inflationary Times (to Come?)

http://madmoneymachine.com/200....s-to-come/

I examine the performance of the Harry Browne Permanent Portfolio during the inflationary years of 1973 - 1981. I also data mined and came up with a more optimized portfolio for those years, the "Inflation Portfolio."

I could keep typing or you could just go read it...

Not a recommendation to buy.
_________________
Paul Douglas Boyer, host, MadMoneyMachine.com

[plake15]

The most positively correlated with CPI-U are TBILL and GOLD

Then how over the last 6 months when we have had this big deflationary period,has Gold done so well and just surged? it's doing very well in strong deflationary period.

[Paul Douglas Boyer]

Wholesale inflation takes biggest jump in 6 months

http://finance.yahoo.com/news/....10311.html
_________________
Paul Douglas Boyer, host, MadMoneyMachine.com

[tarnation]

I looked the TSP G-fund, it has a correlation coeff to CPi-U of 0.621 (and a CAGR of 6.22%).

Since G-fund returns are only income returns, It made me wonder how a TIPS fund (VIPSX) income returns would correlate with CPI-U. I know it is only N=8 data points, but for returns from 2001-2008, I get a coeff of 0.5.
_________________

[Bulldog Bond]

A couple of points. I believe that the CPI-U series used by the OP was the December-December percent change. Different numbers arise if the Avg-Avg changes are used. The inflation in 2008 for example was 3.84% (Avg-Avg) or 0.09% (Dec-Dec). I prefer the Dec-Dec metric.

I have examined how the S&P 500 index behaves (with NO dividend reinvestment) thru eras of high and low inflation. Taken in decade-long data sets, the REAL Yr-Yr change in the S&P 500 Index is seen to be surprisingly constant, rarely straying far from 3.3% over the 49 year interval ending in 2002. The Lowest Real Yr-Yr Change was 2.5% in decades ending in the mid 90's. It wasn't above 3.8% in decades ending after the mid 70's. The peak was near 5% in for decades ending in the late '60's. Taken in decade-long time horizons, the S&P 500 Index seems to do a prety good job of providing a relatively consisten Real return.

The corelations in other assets depend rather strongly on the interval selected. It is important to differientate between the effect when inflation is increasing vs when it is decreasing. Here is a table of Correlations for selected assets in 3 different intervals:
...........................................S&P 500.....5 Yr T......LTGB.....S-TIPS
............................................VFINX....... VFITX......VUSTX...VIPSX
1972-2008 All data................ -0.09.........-0.24.......-0.40......0.06
1983-2002 Falling Inflation..... -0.08.........-0.05.......-0.16.....-0.04
1972-1982 Rising Inflation...... -0.26.........-0.51.......-0.63......0.59

VFINX (tracking S&P 500 total returns) was essentially uncorrelated with CPI-U changes, except when inflation was rising from '72-82.
VFITX (the 5 Yr Treasury fund) was negatively correlated when inflation was rising, but not correlated when inflation was falling.
VIPSX (or Synthetic TIPS) was positively correlated with inflation, but only when inflation was rising.
I think that is what TIPS were intended to do.