How can studying flows in mutual funds and ETFs improve my investment strategy? originally appeared on Quora: the place to gain and share knowledge, empowering people to learn from others and better understand the world.
Answer by Jiacui Li, Assistant Professor of Finance at University of Utah, David Eccles School of Business, on Quora:
This is an important question because mutual funds and ETFs currently hold around 40% of all U.S. stocks.
Before I provide a few approaches, let’s set up the context. When people invest money into these funds (inflows), the funds need to use this money to buy more stocks. Conversely, when people withdraw their money (outflows), the funds have to sell stocks to give the money back. This pattern of buying and selling is largely predictable; you can expect funds to buy more when they get more money from investors and sell more when investors are taking their money out. When a lot of funds are either buying or selling in the same way at the same time because they are experiencing similar inflows or outflows, it can push stock prices up or down significantly. However, these price changes tend to go back to where they were over time. This means that the price increase due to many funds buying will eventually decrease, and the price decrease due to many funds selling will eventually rise again.
Now, how can quantitative investors take advantage of these flows to improve investment performance? There are three approaches.
1) Trade on long-term reversion.
After large flows have happened, trade against those flows and bet on price reversion. This applies to slow-moving strategies as reversions can take years.
As an example, this study finds that fund flows often fluctuate wildly between different size and value/growth styles, creating large style-level price pressures that revert subsequently. I would recommend taking into account such flow-based signals in your asset allocation strategies. For instance, during the 2000s dotcom bubble, investors poured hundreds of billions into large-cap growth funds, and this made large-cap growth stocks more overvalued. On average, such overvaluation leads to substantially lower returns in the subsequent 1-2 years.
2) Front-run near-term flows. You can also predict fund flows and front-run them. This applies to shorter-horizon strategies, ranging from one week to one quarter.
This is slightly more complicated as it requires a deeper understanding of what drives fund flows. As an example, we find that fund flows are heavily determined by Morningstar ratings: funds with high (low) ratings receive large inflows (outflows). Further, Morningstar ratings are computed using mechanical formulas and are somewhat predictable. In follow-up research, we find that a Morningstar methodology change in rating calculation caused large, predictable style-level price fluctuations in June 2002 and also made momentum-type strategies less profitable afterward. Because the methodology change was announced ahead of time, if you understood fund flows and Morningstar ratings, you would have been able to take advantage of it back then.
3) Improve diversification.
If you conduct mean-variance portfolio optimization, you can use fund flows to better estimate return covariance and improve diversification.
Traditionally, quant funds use either factor-based approaches (e.g. BARRA) or Ledoit-Wolf type shrinkage to construct their covariance matrices. For reasons I cannot fathom, most of them do not use information beyond historical returns to estimate covariance matrices. However, fund flows can improve covariance matrix estimates, such as shown in this paper. The idea is very simple: if two stocks tend to be co-held by an overlapping set of funds, they tend to be traded together, and this leads to higher covariance between them.
In ongoing work, I find that incorporating flow- and trading-related data can improve covariance matrices estimation and lead to better diversification and higher Sharpe ratios. This work is still at an early stage and not ready for public consumption. If you are interested, and if you work in quantitative finance as a professional (this is not a topic for retail investors), feel free to email me.
This question originally appeared on Quora – the place to gain and share knowledge, empowering people to learn from others and better understand the world.
Is owning a house a risky investment? originally appeared on Quora: the place to gain and share knowledge, empowering people to learn from others and better understand the world.
Answer by Marco Giacoletti, Assistant Professor of Finance at University of Southern California, on Quora:
Owning a home is often seen as the American Dream, but it’s also a significant financial decision—arguably one of the largest an individual or family will make. Besides being the places that offer shelter to owners and their families, and the places where daily life unfolds, houses are also investment assets.
A house over its lifetime goes across multiple owners; a study over 2019-2021 calculated that the median time while a homeowner holds onto the same house is 13 years. Given that the median home put up for sale in the U.S., as of 2019, was built in 1978, we can see that homes change hands multiple times over their useful life. At the time of resale, homeowners realize price gains (capital gains), or losses. These price gains have been positive and large on average across the United States over the last two decades. The S&P Case Shiller Index, which tracks the price growth of resold homes, showed an average annual growth rate of 4.8% from 2000 to 2022, culminating in an impressive 280% increase over the entire period.
However, this growth is accompanied by substantial volatility. Timing of purchase and sale, as well as location, can dramatically affect the return on this investment. For instance, between 2006 and 2012, annual price appreciation rate for the US index was -3.2%. Disparities in growth rates are even more pronounced when comparing different metropolitan areas and neighborhoods. As an example, within the Los Angeles metro area over 2000-2022, house prices have grown by more than 7% annually in Culver City, but only by 5.6% in Montebello. Overall Los Angeles housing prices grew by 6% annually and in comparison, Detroit housing prices grew by 2.3% annually between 2000-2022.
Beyond these time and location components, there is a significant element of “idiosyncratic housing risk.” This is the variance in price gains that cannot be attributed to market-wide fluctuations (even when considering very narrow local markets, such as zip codes), but is rather specific to the individual sale of a house.
Work in this area has shown that idiosyncratic risk might be the most important driver of housing risk. In my own research, when I decompose the total risk for a house into local market risk (zip code-level fluctuations) and idiosyncratic risk, I find that idiosyncratic risk outweighs local market risk, especially for shorter holding periods. I show evidence that idiosyncratic risk is tied to the uniqueness of each home and transaction, and influenced by factors such as buyer interest and market liquidity at the time of sale. The diversity in buyer valuations and the limited pool of potential buyers at any given time can result in significantly different sale prices for the same property. For example, John who likes spending time outdoors with his dog may value a swimming pool and a garden at the property, and not care about a smaller living room, kitchen and dining area. Another potential buyer may have opposite preferences. Whether a house listing with a swimming pool ends up being visited by John, or not, may drive substantial differences in the sales price. This buyer “matching risk’’ component is a key driver of idiosyncratic risk.
I also find that the importance of idiosyncratic risk diminishes when the house is held for a longer holding period. This is because idiosyncratic risk is specifically driven by the matching risk effect that takes place at the time of sale. Thus, the longer the time span between resales, the lower the average impact of this risk per year of holding period. The figure below illustrates this point in depth:
The left panel of the figure plots the median, top quartile, and bottom quartile of the share of idiosyncratic risk across California zip codes, for house holding periods between 2 and 15 years. First, idiosyncratic risk determines the larger share of risk for most holding periods. Second, the share is decreasing with the holding period, across all zip codes. For the median zip code, the idiosyncratic share is close to 70% if a house is held for 2 years, and less than 50% if a house is held more than 15 years.
As mentioned above, this is because of some peculiar properties of idiosyncratic risk, as opposed to local market risk. Over the holding period (time between resales) of a house, the amount of risk (or variation) per year of the local market component is roughly the same, no matter the length of the holding period. However, the amount of idiosyncratic risk per year is decreasing.
The right panel of the figure then depicts the median of total annual risk for a house, across holding periods. The solid line shows estimates from the data. Since idiosyncratic risk is a large fraction of total risk, total risk per year is also decreasing with the holding period. It is 15% for a 2-year holding period, and roughly 11.5% for a 15-year holding period. The dotted line in the same panel shows the values of total housing risk if the idiosyncratic component was “forced’’ to be the same across holding periods. We can see that this would underestimate total risk at short holding periods, and overestimate at long holding periods.
The numerical estimates in the figure can be interpreted by making some simplifying statistical assumptions. Say that for the median zip code, average annual price growth over a certain holding period has been 5%. If the holding period is 2 years, 7 out of 10 homes in the zip code will earn capital gains between -10% and 20%. If the holding period is 15 years, 7 out of 10 homes will earn capital gains between -6.5% and 16.5%. Thus, price gains’ risk is quite large, once we account for the idiosyncratic component. However, homeowners can reduce the annual risk of their investment by holding onto their houses for longer.
References:
- Data on the characteristics of houses in the US is available trough the American Housing Survey.
- Data on homeowners tenure (holding periods).
- S&P Case Shiller Indices from Federal Reserve Bank of St. Louis.
- Zip code-level indices from Zillow.
- Giacoletti, Marco, 2021, “Idiosyncratic Risk in Housing Markets”, Review of Financial Studies, 2021, Volume 34(8), pages 3695-3741.
- Landvoigt, Tim, Monika Piazzesi and Martin Schneider 2015. “The Housing Market(s) of San Diego.” American Economic Review 105(4), 1371-1407.
- Piazzesi, Monika, Martin Schneider and Selale Tuzel 2007. “Housing, Consumption and Asset Pricing.” Journal of Financial Economics 83, 531-569.
- Sagi, Jacob, 2021, “Asset-Level Risk and Return in Real Estate Investments”, Review of Financial Studies, vol. 34(8), pp 3647-3694.
This question originally appeared on Quora – the place to gain and share knowledge, empowering people to learn from others and better understand the world.