The Butterfly Effect
The Price Impact of Fund Flows in Inelastic Markets
By: Isabel Chomnalez
How much does the aggregate value of equities change if an investment fund allocates an additional $1B to stocks? A survey of 300 economists found that the consensus answer was “zero.”
Most financial economists believe that the idea that fund flows might impact market prices is “sadly illiterate,” according to one recent paper. This is because the efficient markets theory presumes that markets are highly elastic—that is, demand is highly reactive to prices and thus stock prices shouldn’t diverge from fundamental value just because large investments flow in or out of the security.
But in a controversial new paper, Harvard’s Xavier Gabaix and Chicago’s Ralph Koijen propose an unorthodox and even heretical idea: that the stock market is so inelastic that $1 invested in the stock market increases aggregate valuations by $5. Their “Inelastic Markets Hypothesis” presents structural inelasticity as the answer to the long-debated excess volatility problem.
The paper is so controversial relative to academic consensus that we decided to replicate the authors’ methodology to better understand the empirical support for their hypothesis. Replicating the paper’s methodology was no simple task. The paper is chock-full of phrases like “intuitively, we use the sector-specific, or idiosyncratic, demand shocks of one sector as a source of exogenous price variation to the demand elasticity of another sector,” and when we emailed the authors to ask their advice, they replied that we’d need a graduate-level course in econometrics to understand their logic. Not a promising start, especially given our skeptical view of ideas that require complex models or academic degrees to understand. But, after substantial effort, we think we understand their empirical argument well enough to discuss it in simple terms and to conclude that we find the argument credible but are not yet willing to say the empirical support is substantial enough to be definitive.
To replicate the study, we compiled the flows and the aggregate levels of equity from 1990 to the present from the Federal Reserve’s Flow of Funds database. The database records the purchase of and the aggregate prices of corporate equities owned by different sectors of investors (households, pension funds, mutual funds, the federal government, etc.). We then compared flows into equity markets to the returns on the S&P 500 index to attempt to understand the relationship between flows and prices.
Since elasticity captures how price responds to demand, intuition would direct us to simply regress returns onto flows. If stock market demand is elastic, the multiple by which flows drive up prices should be no greater than 1. To support the broadest version of the inelasticity hypothesis, we must get a multiple above 1 that is statistically significant.
A regression of the total market value versus aggregate inflows and outflows yields a multiple of 2 but an r2 of 0. By lumping all the investor sectors together, we lower the statistical significance of the multiple. Some investor sector flows do not predict changes in aggregate prices and thus should be excluded from the regression. So we broke flows down by sector, as Gabaix and Koijen do in their paper.
The next challenge is more technical: how do we make sure that:
Our regressors (in this case, flows from each sector) are not predicting each other.
That some other unknown factor is not driving the same change in both flows and returns.
For every buyer there has to be a seller, so when households sell their equities, an investment fund will buy them, and the net change in flows will balance to around 0.
To escape this problem, Gabaix and Koijen construct a variable that captures demand shocks that are produced neither by the demand of other sectors nor by general economic pressures common to all sectors (e.g., economic growth/contraction). The multiple on this so-called instrumental variable should show by what factor demand shocks drive up aggregate prices and thus show how elastic or inelastic the market is.
The multiple we get is slightly higher than Gabaix and Koijen’s, but our results are broadly confirmatory. Returns are highly inelastic: a 1% increase in flows into the stock market will result in anywhere from a 5% to 9% increase in valuations! Neither Gabaix and Koijen’s paper nor our regressions reached a highly statistically significant level, however.
Reservations aside, the regression suggests that investor preferences impact market valuations “in a quantitatively important way.” This is in keeping with our work on the extraordinary negative effects of crises on illiquid asset classes and the extraordinarily high returns investors who buy into those asset classes immediately following such crises can thus earn. And it’s also in keeping with our work on how over-priced markets can trend for extended periods of time as valuation increases attract new flows which, in turn, support higher prices.
US Equity Markets have been the most obvious flow winner in the past couple of years. In 2020, portfolio equity inflows in the US climbed to $725 billion, which accounted for 84.04% of the world’s equity net inflows. US equity funds have received more than $900 billion in net inflows in the first half of 2021, pushing the S&P 500 up by 17% (in comparison to a 14% rise of Germany’s DAX and a 2.2% increase in Japan’s Nikkei Stock Average).
While a recent phenomenon, this surge in inflows might not ebb soon. Not only are American equities receiving the lion’s share of global inflows, but investors are also increasingly allocating their capital to inelastic vehicles. Demand for equities is rising, and that demand is flowing through investment funds which our regressions found had the highest multiple impact on aggregate valuations.
The past 30 years have witnessed a dramatic shift in the share of total equity owned by households and investment funds. In the 15-year period between 1990 and 2005, investment funds’ share of total equity increased by 0.25% per quarter, stabilizing around 2008 at 28% of total equity. Household share of equity moved inversely to that of investment funds, declining steadily until leveling out a little below 40% in 2008.
Figure 1: Evolution of Equity Ownership by Investment Funds and Households
Source: US Flow of Funds
This change is compounded by a mix shift within investment funds from active to passive.
Passive investing is perfectly inelastic, as demand is insulated from changes in price. Academics Valentin Haddad, Paul Huebner, and Erik Loualiche expect that an increase in passive investing of 40% over the past two decades should have driven down the stock market’s aggregate elasticity by as much as 65%. Jesse Livermore of O’Shaughnessy Asset Management paints a similarly dystopian portrait of where this might lead, with investors perceiving no alternative but bidding on equities, thus locking in multiplicative effects on prices.
If this hypothesis is true, inelasticity further obscures price movements in the equity market, but, in Gabaix’s and Koijen’s words, it replaces “the mystery of apparently random movements of the stock market” with “the more manageable problem of understanding the determinants of flows in inelastic markets.”
Acknowledgement: Isabel Chomnalez interned with Verdad this past summer. She is a rising junior studying philosophy and comparative literature at Yale, where she chairs her debate society and plays on the polo team. She will be interning at Blackstone next summer.