A common question asked in the factor investing field is the following — “how much of the model’s performance is driven by sector allocations, and how much is driven by security selection?”

Get our full guide on value investing in PDF

Get the entire 10-part series on value investing in PDF. Save it to your desktop, read it on your tablet, or email to your colleagues.

Also read:

Our answer is to simply buy Value stocks or Momentum stocks, regardless of sector constraints.

Why?

Well a nice anecdote (but not data) is that investing in “cheap” technology stocks was not a great idea in the internet bubble crash.(1)

However, not everyone likes our answer and some prefer to have sector-neutralized portfolios to reduce tracking error. We understand that tracking error to many is bad, as this can cause unfortunate events to occur (such as getting fired for recommending the portfolio with no sector constraints).

But what does the data say?

An interesting working paper titled, “Portfolio Allocations Using Fundamental Ratios: Are Profitability Measures Effective in Selecting Firms and Sectors?” by J. Christopher Hughen and Jack Stauss examines this question.

They specifically use multiple profitability measures to form portfolios in three ways:

  1. At the sector level (buying the whole sector)
  2. At the security level (for sector-neutral portfolios)
  3. At the stock and security level

The paper finds that building portfolios using profitability measures works in all 3 cases, but works the best in case 3 above, where the model is allowed to pick the top sectors and top stocks (almost unconstrained).

Below we dig into the paper.

The model and paper setup

There are many papers highlighting that using fundamental ratios, cheaper firms outperformed expensive firms in the past. One of the original papers by Fama and French (1992) highlighted the Value premium by examining the Book-to-Market ratio. However, while a DFA favorite, there are other measures one can use to sort stocks into Value / Growth (or Profitable / Unprofitable). As mentioned above, we are fans of the Enterprise Multiple, as we show the performance of the measure here, updated data here and give some theory behind the measure here. The paper studies 7 metrics: cash flows (CF), earnings (EP), operating pro fit (OP), gross pro fit (GP), book value (BM) to market value, enterprise multiple (EV/EBITDA) and composite variable (COM).

On the data side, the paper examines only stocks within the SP500 from 1975 to 2014. Using SP500 stocks helps to reduce concerns regarding shorting, as the paper does recommend this (discussed more below). The paper splits the universe into 10 sectors using GICS, and examines quarterly data with a one-quarter lag to avoid look-ahead bias.

The model used in the paper is below:

Source: Portfolio Allocations Using Fundamental Ratios: Are Profitability Measures Effective in Selecting Firms and Sectors? Accessed 8/14/17 from SSRN.

So the model forecasts the next quarters multiple, using past multiples (both at the stock and sector level). The portfolios are rebalanced quarterly using the model’s forecasts.

The Results

As mentioned above, the paper examines the model in 3 ways:

  1. At the sector level (buying the whole sector)
  2. At the security level (for sector-neutral portfolios)
  3. At the stock and security level

Following that order (slightly different than the paper), we examine the results below.(2)

Exhibit 2 shows the performance of the model at the sector level. The “average return” below is for each quarter, as the portfolios are rebalanced quarterly. Since there are 10 sectors, the top quintile (20%) is the top 2 sectors, while the bottom quintile is the bottom 2 sectors. As shown below, the forecasts from the model show a spread between the top and the bottom. The long/short portfolio goes long (150% of NAV) the top quintile and shorts (50% of NAV) the bottom quintile. So this L/S portfolio has a net equity position (on a $ basis) of 100% and is somewhat comparable to the SP500 or SP500 EW. Over this same time period, the SP500 returned 3.2% per quarter with a Sharpe ratio of 0.52 and an ending value of $5,455, while the SP500 EW returned 3.3% quarterly with a Sharpe ratio of 0.59% and an ending value of $7,017. Compared to the SP500 and SP500 EW, using the model to forecast sectors has (on average) higher portfolio ending values (before trading costs).

Portfolio Allocations Using Enterprise Multiples

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Exhibit 3 shows the performance of the model at the firm level, while maintaining an equal sector weighting.  Again the paper examines the top and bottom quintiles, while forming a 150/50 L/S portfolio. Compared to the SP500 and SP500 EW, using the model to forecast firm allocations (while holding sectors equal) has higher portfolio ending values (before trading costs). Comparing the model using either (1) sectors or (2) firms, one notices that the model generates more benefit at the firm level than the sector level.

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Last, what happens if we allow the model to pick both the top sectors and the top firms? Exhibit 1 examines the results to portfolios using either the top (bottom) quintile of sectors (2 sectors), and within the top (bottom) sectors, selects the top (bottom) quintile of firms on the valuation measure. The paper similarly forms the 150/50 L/S portfolios, with a slight difference from the L/S portfolios described above (from footnote 5 in the paper):(3)

A 150/50 strategy takes short positions worth 50% of the portfolio value and uses the proceeds from the shorts to fund long positions worth 150% of the portfolio value. As GICS has ten sectors, short positions are taken in the two sectors in the bottom quintile of valuation so the sectors each have weights of -25%. This funds overweighting the sectors in the top quintile of valuation, and these two sectors have weights of 37.5%. The remaining six sectors are equally weighted with each comprising 12.5% of the portfolio.

As such, this long/short portfolio actually allows for allocations to all sectors, while over-weighting the top and under-weighting the bottom, and picks the top (and

1, 2  - View Full Page