Modern Portfolio Theory

Question A:

Harry Markowitz, the Nobel Prize-winning pioneer of modern portfolio theory, passed away earlier this year:
https://afajof.org/news/in-memoriam-harry-markowitz-past-president-of-the-american-finance-association-1927-2023/

Application of the principles of modern portfolio theory allows investors in practice to achieve substantial improvements in the risk-expected return trade-off relative to naive strategies such as equal-weighting that do not take account of return covariances.

Responses weighted by each expert's confidence

Question B:

Widespread adoption of modern portfolio theory by investors has substantially improved the efficiency of capital allocation in financial markets.

Responses weighted by each expert's confidence

Question A Participant Responses

Participant University Vote Confidence Bio/Vote History
Campbell
John Campbell
Harvard
Strongly Agree
9
Bio/Vote History
Although there are pitfalls in using historical covariances to predict future covariances, thoughtful application of modern portfolio theory can indeed improve the risk-return tradeoff.
Cochrane
John Cochrane
Hoover Institution Stanford
Agree
8
Bio/Vote History
Actual mean-variance calculations don't work well in practice. But portfolio theory tells us all to diversify, and what that means exactly. Investors today, via funds, hold portfolios that are much better diversified than the individual stocks they held when Markowitz wrote.
Cornelli
Francesca Cornelli
Northwestern Kellogg Did Not Answer Bio/Vote History
Diamond
Douglas Diamond
Chicago Booth
Agree
7
Bio/Vote History
Duffie
Darrell Duffie
Stanford
Disagree
10
Bio/Vote History
Surprisingly, empirical evidence suggests that equal-weighted portfolios have done about as well as market weights, which are Markowitz-optimal in equilibrium. But equal weights are obviously not consistent with market clearing (See Sharpe)!
Eberly
Janice Eberly
Northwestern Kellogg
Strongly Agree
9
Bio/Vote History
Fama
Eugene Fama
Chicago Booth
Strongly Agree
10
Bio/Vote History
In the early 1950s Markowitz laid the foundations for portfolio selection. Building on Markowitz, Sharpe (mid-1960s) produced the first asset pricing model. The work on market efficiency in the 1960s and thereafter provided empirical justification for passive investing.
Gabaix
Xavier Gabaix
Harvard
Strongly Agree
9
Bio/Vote History
Goldstein
Itay Goldstein
UPenn Wharton
Agree
7
Bio/Vote History
Graham
John Graham
Duke Fuqua
Strongly Agree
8
Bio/Vote History
Harvey
Campbell R. Harvey
Duke Fuqua
Agree
8
Bio/Vote History
Correlations should be taken into account in diversifying portfolios. However, in applying MPT, we need to be careful. 1952 paper assumes we exactly know expected returns & covariances. It also assumes variance=risk p92. It is impt to take uncertainty & other risks into account.
Hirshleifer
David Hirshleifer
USC
Strongly Agree
8
Bio/Vote History
Hong
Harrison Hong
Columbia
Disagree
6
Bio/Vote History
Studies go back on forth on this issue
Jiang
Wei Jiang
Emory Goizueta
Strongly Agree
9
Bio/Vote History
Kaplan
Steven Kaplan
Chicago Booth
Agree
9
Bio/Vote History
Kashyap
Anil Kashyap
Chicago Booth
Agree
5
Bio/Vote History
Important to remember what kind of advice people were following before him
Koijen
Ralph Koijen
Chicago Booth
Strongly Agree
7
Bio/Vote History
While it is challenging to estimate covariances and expected returns using historical return data alone, using characteristics helps to outperform the 1/N benchmark (Section 6.1 of the first paper; Table 1 of the second paper; Table 5 of the third paper)
-see background information here
-see background information here
-see background information here
Kuhnen
Camelia Kuhnen
UNC Kenan-Flagler
Strongly Agree
10
Bio/Vote History
Lo
Andrew Lo
MIT Sloan Did Not Answer Bio/Vote History
Lowry
Michelle Lowry
Drexel LeBow
Strongly Agree
8
Bio/Vote History
Ludvigson
Sydney Ludvigson
NYU
Agree
9
Bio/Vote History
Maggiori
Matteo Maggiori
Stanford GSB
Agree
8
Bio/Vote History
Matvos
Gregor Matvos
Northwestern Kellogg Did Not Answer Bio/Vote History
Moskowitz
Tobias Moskowitz
Yale School of Management
Strongly Agree
10
Bio/Vote History
Theoretically this is true. When having to estimate inputs with substantial error, the results can be noisy, but innovations in how to improve estimation have made this a big success.
Nagel
Stefan Nagel
Chicago Booth
Agree
9
Bio/Vote History
There are a variety of practical hurdles (e.g., estimation of means and covariance, non-tradable wealth) but an investor certainly benefits from applying the basic principles (e.g., that covariances matter for portfolio risk, not individual asset variances)
Parker
Jonathan Parker
MIT Sloan
Strongly Agree
10
Bio/Vote History
Parlour
Christine Parlour
Berkeley Haas
Uncertain
8
Bio/Vote History
In practice many "optimized" portfolios are subject to constraints (rebalancing costs/short selling). Depending on the number of constraints, outcomes will look like the simple heuristic.
Philippon
Thomas Philippon
NYU Stern
Agree
7
Bio/Vote History
Puri
Manju Puri
Duke Fuqua Did Not Answer Bio/Vote History
Roberts
Michael R. Roberts
UPenn Wharton
Strongly Agree
9
Bio/Vote History
Sapienza
Paola Sapienza
Northwestern Kellogg
Agree
6
Bio/Vote History
It is an important framework but there are many limitations based on the assumptions (returns have to be normally distributed, investors allocate all portfolio assets to a single timeframe, past data predict future data). These assumptions are often not verified in practice
Seru
Amit Seru
Stanford GSB
Strongly Agree
9
Bio/Vote History
Stambaugh
Robert Stambaugh
UPenn Wharton
Strongly Agree
10
Bio/Vote History
Key points of Markowitz (1952, JF): Consider the E-V tradeoff (vs just max E), diversify to reduce V, and covariance's role in V makes diversifying across industries important. Equal vs non-equal weights is a lesser point, missing the broader message of modern portfolio theory.
Starks
Laura Starks
UT Austin McCombs
Strongly Agree
8
Bio/Vote History
Stein
Jeremy Stein
Harvard
Strongly Agree
9
Bio/Vote History
Stroebel
Johannes Stroebel
NYU Stern
Strongly Agree
5
Bio/Vote History
Sufi
Amir Sufi
Chicago Booth
Agree
5
Bio/Vote History
Titman
Sheridan Titman
UT Austin McCombs
Agree
10
Bio/Vote History
Van Nieuwerburgh
Stijn Van Nieuwerburgh
Columbia Business School
Uncertain
6
Bio/Vote History
The equal weight 1/N rule has been shown to be hard to beat
-see background information here
Whited
Toni Whited
UMich Ross School
Agree
6
Bio/Vote History

Question B Participant Responses

Participant University Vote Confidence Bio/Vote History
Campbell
John Campbell
Harvard
Agree
6
Bio/Vote History
This is a tricky question because passive investors opt out of price discovery; however, active investors using modern portfolio theory do improve the efficiency of capital allocation - aided by the analytical tools and cheap trading made possible by information technology.
Cochrane
John Cochrane
Hoover Institution Stanford
Uncertain
5
Bio/Vote History
Investors haven't adopted portfolio theory. (Some hedge funds do, but not most institutions and individuals). Plus, it was never about investment or market efficiency.
Cornelli
Francesca Cornelli
Northwestern Kellogg Did Not Answer Bio/Vote History
Diamond
Douglas Diamond
Chicago Booth
Agree
5
Bio/Vote History
Duffie
Darrell Duffie
Stanford
Strongly Agree
10
Bio/Vote History
Before Markowitz and Sharpe finally took a grip on practice (with diversification and index funds), many investors paid large amounts of transactions costs to do "stock picking," which is clearly risk-return inferior except for investors with excellent private information.
Eberly
Janice Eberly
Northwestern Kellogg
Strongly Agree
9
Bio/Vote History
Fama
Eugene Fama
Chicago Booth
Uncertain
1
Bio/Vote History
Gabaix
Xavier Gabaix
Harvard
Agree
7
Bio/Vote History
Goldstein
Itay Goldstein
UPenn Wharton
Strongly Agree
7
Bio/Vote History
Graham
John Graham
Duke Fuqua
Agree
6
Bio/Vote History
Harvey
Campbell R. Harvey
Duke Fuqua
Agree
8
Bio/Vote History
It is hard to measure. However, in my experience in talking to very large allocators such as pensions and SWFs, they focus a lot on correlation structure. It is hard to measure "substantial" given we don't have a lot of examples of funds ignoring MPT.
Hirshleifer
David Hirshleifer
USC
Strongly Agree
8
Bio/Vote History
Hong
Harrison Hong
Columbia
Uncertain
5
Bio/Vote History
no studies on this issue
Jiang
Wei Jiang
Emory Goizueta
Strongly Agree
9
Bio/Vote History
Kaplan
Steven Kaplan
Chicago Booth
Agree
2
Bio/Vote History
Kashyap
Anil Kashyap
Chicago Booth
Uncertain
3
Bio/Vote History
So many other factors matter for this.
Koijen
Ralph Koijen
Chicago Booth
Agree
7
Bio/Vote History
Kuhnen
Camelia Kuhnen
UNC Kenan-Flagler
Strongly Agree
10
Bio/Vote History
Lo
Andrew Lo
MIT Sloan Did Not Answer Bio/Vote History
Lowry
Michelle Lowry
Drexel LeBow
Disagree
6
Bio/Vote History
Ludvigson
Sydney Ludvigson
NYU
Agree
8
Bio/Vote History
Maggiori
Matteo Maggiori
Stanford GSB
Agree
8
Bio/Vote History
Matvos
Gregor Matvos
Northwestern Kellogg Did Not Answer Bio/Vote History
Moskowitz
Tobias Moskowitz
Yale School of Management
Agree
4
Bio/Vote History
Should be true, but as a field we have not done a great job at measuring resource allocation.
Nagel
Stefan Nagel
Chicago Booth
Uncertain
6
Bio/Vote History
Adoption should have helped ensure that risk that is in principle diversifiable doesn't command risk premia, which in turn should improve capital allocation efficiency, but I'm not sure there is empirical evidence that this has been a substantial effect.
Parker
Jonathan Parker
MIT Sloan
Strongly Agree
10
Bio/Vote History
Parlour
Christine Parlour
Berkeley Haas
Strongly Agree
9
Bio/Vote History
Portfolio theory led to the design of ETFs and mutual funds which increased participation and capital market depth.
Philippon
Thomas Philippon
NYU Stern
Disagree
7
Bio/Vote History
Puri
Manju Puri
Duke Fuqua Did Not Answer Bio/Vote History
Roberts
Michael R. Roberts
UPenn Wharton
Agree
7
Bio/Vote History
Question is a little ambiguous.
Sapienza
Paola Sapienza
Northwestern Kellogg
Agree
2
Bio/Vote History
For the reasons, that I exposed in the previous answer, in practice the assumptions are not satisfied, so empirically the results of the model depend on those assumptions being met or violated.
Seru
Amit Seru
Stanford GSB
Agree
8
Bio/Vote History
Stambaugh
Robert Stambaugh
UPenn Wharton
Uncertain
5
Bio/Vote History
Index funds apply the advice of Markowitz (1952, JF): diversify, especially across industries. Active quant funds often make covariance-conscious portfolio decisions. Net effects on price discovery and capital allocation seem hard to identify..
Starks
Laura Starks
UT Austin McCombs
Strongly Agree
8
Bio/Vote History
Stein
Jeremy Stein
Harvard
Uncertain
5
Bio/Vote History
Stroebel
Johannes Stroebel
NYU Stern
Agree
3
Bio/Vote History
Sufi
Amir Sufi
Chicago Booth
Uncertain
5
Bio/Vote History
I am not sure what "efficiency" means in this statement.
Titman
Sheridan Titman
UT Austin McCombs
Disagree
10
Bio/Vote History
The modern portfolio approach has probably improved the informational efficiency of markets but the effect on capital allocation is likely to be modest.
Van Nieuwerburgh
Stijn Van Nieuwerburgh
Columbia Business School
Agree
7
Bio/Vote History
Consideration of the best risk-adjusted investment opportunities in capital budgeting should have improved the efficient allocation of capital
Whited
Toni Whited
UMich Ross School
Uncertain
1
Bio/Vote History