We don’t break rules, we create new ones!

Retailing is hard. The evidence is incontrovertible. In Australia alone, just look back over the last couple of months we have witnessed Herringbone, Rhodes and Beckett, Marcs and David Lawrence brands all being the subject of voluntary administration. Casting our minds back just a little further, Dick Smith, Masters, Payless Shoes and Pumpkin Patch have all been wound up. Then, how about the recent Oroton downgrade in January (share price is down 30% since) and The Reject Shop downgrade (on 7 April, share price down 46% since), and two RCG downgrades (most recently 1 May, but share price down over 50% since first downgrade in February)^{1  , 2} .

“Retailing is a tough business. During my investment career, I have watched a large number of retailers enjoy terrific growth and superb returns on equity for a period, and then suddenly nosedive, often all the way into bankruptcy. This shooting-star phenomenon is far more common in retailing than it is in manufacturing or service businesses. In part, this is because a retailer must stay smart, day after day. Your competitor is always copying and then topping whatever you do. Shoppers are meanwhile beckoned in every conceivable way to try a stream of new merchants. In retailing, to coast is to fail.” – Warren Buffett

As we started writing this in late March, “Retail Trade” businesses listed in the United States (US) featured prominently amongst our top ranked stocks according to our investment systems? Why is this so? The short answer is that our process is backward looking. In order to avoid biases, our models intentionally apply only historic data.

So this begs the question – how appropriate is our investment process when applied to the whims of retail businesses? Mr Buffett’s quote highlights (and our experience shows) retail businesses can be subject to rapid change where high earnings on invested capital can quickly turn negligible or negative – particularly exacerbated in recent times by an increasing pace of change of industry conditions, for example the rising influence of Amazon.

To address industry changes, some of the world’s most astute retail property players have evolved by focusing on high-end experiential and lifestyle centres. Scentre (owner of the Westfield malls in Australia), Westfield Holdings (owner of the Westfield malls in the UK and US), and Simon Property Group (owner of shopping malls in the US) have all divested their lower quality malls over recent years. Vornado Realty Trust exited the mall business altogether, with the Chairman writing in the 2016 Annual Report:

“I do not believe we can grow our way out of this mess. I believe the only fix for brick and mortar retailing is rightsizing by the closing and evaporation of, you pick the number, 10%, 20%, 30% of the weakest space. This very painful process will surely take more than five years… I believe the decision to exit the mall business will look better and better as each year goes by.” – Steven Roth, Chairman and CEO, April 2017

Retail space has grown at levels exceeding income growth, and in many regions (particularly in the US), there is clearly an abundance of retail space³ . In most cases retailers have been left footing the bill for unproductive assets in the form of long term lease commitments.

Back to the Fund, in mid-March we held significant positions in US retailers. In fact we were “full up”. Our rules allow us to invest up to 15% in any one sector, conditional on a maximum of 8% being listed in any one country. We held a 7.7%⁴  position in three US listed retailers – Buckle, Wal-Mart, and a small position in Cato. However, in implementing these diversification rules, we strictly applied independent industry classifications (consistent with our aim of minimising behavioural biases). But, Vera Bradley (another Fund holding at the time) is not classified “Retail Trade” despite having substantial retail operations (presumably because it also manufactures the products it sells). If we were to include our Vera Bradley position, that 7.7% weighting would increase to 10.7%. So, arguably we were “overflowing” with US retailers.

These positions (with the exception of Wal-Mart) have performed extremely poorly. Time for further introspection.

All of these companies were set to release their final annual reports in the last week of March. Many of the other US listed retailers that ranked highly in our analysis were also about to release (or had recently released) their final annual reports. These final reports provided substantially more information than the preliminary results data. So, during late March and early April we were able to review comprehensive recent information in relation to not only the four retail businesses we owned, but also another six US retailers of interest.

Given the peculiarities of the retail sector, and the availability of additional data from the final reports, we undertook a more fundamental analysis than our process would ordinarily allow. We did so with some hesitancy, as we recognised that it greatly increased the likelihood of biases entering into our decisions – but then again, this is another reason we are not “Quant”.

Before delving further, let us first explain a mea culpa. Last year we added some new rules designed to avoid investing in companies with substantial off balance sheet liabilities, such as lease commitments. (Because rent obligations from lease commitments are like interest obligations on debt, we were/are very concerned about the implicit leverage). Whilst that intention was good, the execution was not, and yours truly is solely to blame. I made the fatal (but not uncommon) mistake of assuming that my testing sample was reflective of the universe. I was wrong – it was not. I discovered that although the process I developed worked perfectly for all companies in the sample, there are a number of other companies where information was lacking and my work-arounds failed. I learnt an important lesson – this is not an exercise suited to automation (further evidence of us not being “Quant”). So one important update was to ensure we correctly identified and removed from consideration those companies that failed to meet that additional lease commitment criteria.

However, we also extended our analysis for retailers by considering trends in:

• sales productivity (sales per square foot);
• returns on capital (to a greater extent than previously);
• earnings and profit margins (to a greater extent than previously); and
• online revenues and how they have changed over recent years.

The outcome, in relation to the businesses we owned, was that:

• we sold Buckle its sales productivity has declined by 20% over three years, and we began to view its earnings per share declines to be more symptomatic of structural rather than cyclical issues. The rules allowed (but did not require) its sale due to its declining financial strength score;
• we sold Vera Bradley for a few reasons – principally on evidence of poor capital allocation where the data suggests that over the recent four year period it has more than doubled its number of stores but seen sales productivity fall by over 40%. The rules allowed (but did not require) its sale due to its significantly declining earnings;
• we sold Cato because, as a result of its decreasing revenue, its lease commitments relative to earnings breached our thresholds, and the rules required its sale;
• we initially maintained our shares in Wal-Mart, however due to us not purchasing additional shares after the cash inflows in April our position was diluted from 2.7% of the portfolio to 1.3%. Just very recently, with the rules allowing its sale (due to recent price strength) we divested that remaining position.

As an aside, although Wal-Mart has not been a great performer, its indirect contribution to the portfolio has been quite pronounced in that by holding it, the diversification rules prevented us from acquiring other US retailers which have fared much worse. Unfortunately, working to offset that benefit was our “effective” over-weighting to US retailers as a consequence of the Vera Bradley industry classification. Whilst cognisant of potentially introducing biases, we have since implemented a discretionary rules overlay for industry classification to avoid future occurrences of this nature.

Upon divesting these businesses, our rules “freed up” 6.2% investment opportunity in US retailers – and remember these companies are amongst the top ranked stocks on our systems.

Whilst we want to reduce the chance of investing in “value traps” (seemingly attractive businesses on a path to extinction), by the same token we don’t want to discard opportunities simply on the recognition that our process does not seem to work well in the rapidly changing retail sector. After all, it would be easy to admit “it is retail – it’s just too hard”.

So, we adopted the alternative approach, appreciating that to a large extent our process is about investing in cheap, unloved, out of favour businesses. These companies are often cheap for good reason, but conversely they are also often oversold and represent good value. We won’t always get it right, so yes there are risks. We recognise these risks but accept them in the context of our overall portfolio risk being mitigated by strictly adhering to our diversification rules.

Although our process is rules based, the rules are continually refined along with our experiences such as these recent retailing woes. When we weigh up the merits of refining our rules, we are very wary of introducing biases into the process. For example, we do not foresee ever introducing any changes to the rules that would allow us to override other rules in order to facilitate investing in an otherwise unacceptable business.

“I think that a life properly lived is just learn, learn, learn all the time. And I think Berkshire’s gained enormously from these investment decisions by learning, through a long, long period…. and the one feature that comes through is the continuous learning. If we had not kept learning, you wouldn’t even be here. You’d be alive probably, but not here.” – Charlie Munger

We have refined our rules in relation to investments in retail businesses and with the newly “freed up” capacity we accumulated positions in two other US retailers in late March and April.

Keep evolving,

Stuart and Alexis

¹ Company names shortened in interests of readability.

² From announcement of downgrade to 3 May 2017.

³ TVornado Realty Trust 2016 Annual Report which refers to International Council of Shopping Centers (ICSC) publication quoting 24 square feet of shopping center space per capita in the US compared to Canada (17), Norway (10), and all mature European countries in single digits. Further, the 24 square foot number does not include total retail space, which would equate to 55 square feet per capita.

⁴ To allow for price fluctuations, we will not purchase when our holdings are above 7.5%.