Great Walls of Cash

With stock markets still shaking our nerves and rattling our brains, it has become fashionable to look at the piles of cash still parked in savings and time deposits and money market funds and argue it can only be bullish for stocks.

The argument goes that, as fear continues to recede, investors will put their excessive cash holdings back into higher-yielding assets, which is clearly good for equities. Indeed, some go as far as to contend that this excess liquidity alone can provide support to equities in the near-term, even if economic data fail to (positively) surprise us in the coming months.

I’m about to take issue with this argument, but let me say it’s not exactly on theoretical grounds… From a macroeconomic perspective, an excessive increase in liquidity should affect asset prices by driving up the demand for less liquid assets—be it stocks, bonds, real estate or a Rodin sculpture.

In fact, this portfolio-rebalancing channel is partly the theoretical rationale behind the Fed’s quantitative easing: Flood banks with cash to the point that they are so overwhelmed by it that they decide to give some of it up in exchange of less liquid, riskier assets—like loans to the private sector.

Here is the problem, however: It’s not exactly clear when liquidity is “excessive”. Most measures of “excess liquidity” have flaws that are likely to be exaggerated in the current environment.

For instance, some compare the current stock of M2 (or the subcomponent of M2 that comprises savings deposits, time deposits and retail money funds) with what M2 would have been, had it grown at the potential nominal GDP growth rate. This latter is the combined rate of potential real GDP growth and the Fed’s “target” inflation rate (that is consistent with price stability).

[Incidentally, the rationale behind such a comparison is the so-called “quantity theory of money”—the idea that money supply should only expand to reflect the expansion in the demand for goods and services (aka GDP), assuming a constant financial “technology” and stable inflation expectations.]

The problems with such a comparison are clear: The first has to do with the inevitable “ad hocness” of having to pick a base year. Another has to do with the sensitivities of the results to different assumptions of what is a reasonable potential GDP growth.

One way to bypass these problems is to look at the ratio of M2 (or your favorite broad monetary aggregate) to GDP, and whether it deviates from “trend”—with a positive deviation pointing to “excess” liquidity. (In the same spirit, some look at deviations of credit/GDP from trend). In fact, the current M2/GDP ratio appears to exceed its trend levels, suggesting(?) that cash holdings are excessive.

But this is not good either… First, the resulting deviation from trend is sensitive to the way we choose to specify the trend itself (linear, etc), as well as the period over which it’s specified. Second (and most important in my view), this approach relies on fairly naïve assumptions about the demand for money—namely, that money demand is solely determined by the demand for goods and services, while everything else is constant (technology, expectations, etc).

The obvious (and my preferred) way to correct this is to compare current monetary aggregates with a theoretical estimate of money demand that is based on a more sophisticated specification of money demand than the one suggested by the quantity theory.

But good luck with that! Not only would the resulting estimates be dependent on your chosen specification; the main problem is that, in the present environment, any model estimated on historic data will likely produce misleading results, given the huge shock on investors’ liquidity preference brought by the financial crisis.

So is there a way around this?

One way is to tackle the portfolio-rebalancing argument head on: By looking at “imbalances” in the composition of investors’ portfolios. The idea is that, if such an analysis finds that investors are, on aggregate, heavy on cash and short of equities, they will have to rebalance by selling cash and buying more equities.

In fact, in this spirit (I guess), one often quoted ratio is that of M2 (or just the sum of savings & time deposits plus money funds) to the market capitalization of the stock market. This ratio is currently still very high by historic standards, suggesting an imbalance (excess cash) that, it is argued, is bullish for stocks.

I actually think that this is a bogus ratio to look at. Investors’ portfolios do not comprise just cash and equities, but also bonds and other forms of wealth including real estate.

Importantly, the whole idea of portfolio rebalancing hinges upon the idea that there is a “benchmark” (optimal) allocation that the rebalancing will try to achieve. But what determines that benchmark?

Many analysts just eyeball the data, take the long-term average and claim that current levels are excessive compared to that average… ergo, rebalancing (out of cash and into equities) will have to occur.

But looking at the long-run average (i.e. a constant) for a benchmark is naïve: The benchmark is not static, but time-varying, depending on the expected risk and return on each asset class: equities, fixed income, cash and so on.

So what do we get when we take this into account? One of the big investment banks on Wall Street conducted the analysis recently, looking at a global portfolio of stocks, bonds and cash and taking into account the time-variation in the benchmarks.

Evidently, the results depend on one’s assumptions about the expected returns for each asset class, and notably for equities: Per their assumptions, a great deal of expected future stock returns is influenced by past stock returns over a given (long) period of time... which, these days, translates into relatively low expected returns for equities, in light of the two stock market busts that we’ve had in the course of 10 years.

Accordingly, despite their recent increase, global cash holdings actually come out to be in line with the “benchmark” allocation. The “overweight” asset class turns out to be bonds, with equities on the underweight side.

Obviously, changing the assumptions (e.g. positing higher equity returns based on, say, forward-looking assumptions) would change the outcome.

But if there is a conclusion here is that simply looking at those great walls of cash says very little about whether stocks are about to be set on fire!