Friday, October 16th, 2020

FAANGs can bite you

Like most other professionals, investment analysts love complexity. It’s what makes them valuable – in their own eyes – if not everyone else’s. But sometimes, the important questions are very simple, almost frighteningly so. Right now, the only question that matters is whether the FAANGs can continue to outperform.

The Q3 earnings season has just started and these companies will publish all the financial information which investors could reasonably expect, but this may not provide any valuable insight. Most of our clients long ago abandoned any conventional valuation framework when assessing these companies and the level of uncertainty surrounding earnings forecasts and the future political, economic and fiscal regimes has never been so wide.

In these circumstances, we believe that simple technical and quantitative processes can help investors decide what to focus on. When it comes to the US FAANGs and their fellow travellers, there are two sector charts which give us cause for concern: the first is China Technology relative to the China index; the second is US Communications relative to the US index.

We have long argued that Chinese Technology’s relative performance is a lead indicator for the US Tech sector, so we republish a chart we have used before showing how the relative weight of Tech in China tends to crack before the US and then lead it lower. This time China, cracked in late July and is course for a downgrade to neutral sometime in November. We would expect the US sector to follow the same trajectory with a gap of about five weeks. As we noted, in our recent note (Party Like It’s 1999 – August 7th) the US Tech sector does not normally bottom out until it is well into underweight territory.

The second problematic chart is US Communications, which we have downgraded to neutral this week. The sector is dominated by Facebook and Google, two of the FAANGs, which are front and centre in the great policy debate about privacy, media bias and market dominance. We have no insight into which party is going to control which part of the US government apparatus or what they will be able to do when they get there. But we do know that both companies have dropped through important technical signals, which suggest that there is further downside. In addition to the politics, we suspect that investors may be starting to discount another more fundamental risk.

Depending on which forecast you believe, digital advertising will account for between 55-60% of total ad-spend in the US in 2020. Google and Facebook are already dominant in this category and are starting to face stiff competition from new entrants, including Amazon. There is a rising probability that their period of super-normal growth is coming to an end. There may be other revenue streams available to them, but they will be small in relation to their current turnover and may not be enough to prevent a material de-rating as we move into 2021. If Google were a sector on its own it would be rated underweight, with a high-conviction negative lead indicator. Facebook is just about to be downgraded to neutral, with a medium conviction, negative lead indicator. These are not signals normally associated with market leadership.

We have models for another 28 companies in the Tech sector, of which 12 are below their MAV, 13 above, and 3 in line. If we look at the level of our probability score, not the trend, 16 out of the 28 are below their level in late July, when we last wrote about the sector, and only 10 are above, with 2 the same. This suggests that the sector is heading for a downgrade.

Conclusion: We re-iterate our call to take profits in US Tech. Among the FAANGs, only Apple, and possibly Amazon, still look robust. Google has been downgraded twice, Microsoft once, and Facebook is just about to suffer its first. In the rest of the sector, there are individual companies which still look attractive, but the majority do not and their number is growing.

The fact that we are negative on Google and Facebook does not necessarily mean we should be negative on all the other FAANGs but we should at least ask the question. The most important signal we use is the probability of an individual stock outperforming the index on a risk-adjusted basis and how it relates to its own 26-week moving average.

Google broke down through its 26-week MAV in early August, while it was still overweight. Facebook did the same two weeks ago. Netflix did the same 5 weeks ago and Amazon did last week. Of the FAANGs, only Apple is still above its MAV, but Microsoft, the other trillion-dollar company dropped below its MAV in August and is now rated neutral.

Leave a Reply

Friday, October 2nd, 2020

Dropping Bunds as the Benchmark

The time has come to say good-bye to an old friend and faithful servant. Our euro-denominated asset allocation models will no longer use 7-10 year German bunds as their ultimate safe asset. The decision is prompted by two considerations. The first is well-known: negative yields. The second is more abstruse: they no longer offer the best combination with other fixed income assets to create risk-efficient portfolios.

Instead of Bunds we propose using a pan-Eurozone index, weighted by the amount of bonds in issue – not GDP. There are several well-established benchmarks which can easily accessed. For the record, we are going to use the ICE BofA series and would be happy to include Greece even though it is not investment grade. In practice, it has little direct impact on overall performance, even though its indirect impact via Italy and Spain can be substantial. The 7-10 bucket for the Eurozone also has a negative yield, but this is a relatively recent development and it is only -0.13% as opposed to -0.62%

The big issue, as far as we are concerned, is the interaction with other fixed income assets. Over the last two years a systematic portfolio including bunds would have produced a total return of 17.0%, compared with 18% for a portfolio using the same process but substituting a spread of euro-sovereign bonds. These returns are more risk-efficient (1.44 vs 1.33) and have a smaller drawdown (-3.7% vs -4.4%). None of these are huge differences, but they all point in the right direction. We would take the idea seriously, even if we were only interested in incremental improvements, but there are other reasons as well.

Something happened in May. Politicians in Germany and the Netherlands may refuse to admit it to their voters, but the countries of the core effectively agreed to guarantee the credit of the countries of the periphery. The machinery for shared eurozone issuance may not exist yet, but debt-mutuality is the assumption on which the ECB’s polices are based. German bunds had a critical role to play in the portfolio when there was a realistic possibility that the Eurozone could break up. If investors no longer believe this is possible, we should move to a different benchmark. By way of corroboration, our model for eurozone bonds has failed to generate any significant alpha for the last two years and has almost exactly the same volatility and drawdown as the pan-euro benchmark.

Subject to all the usual caveats, we believe investors should move in this direction as soon as practical, ideally by the start of next year. We want to get this move out of the way, so that we can begin to think about the next stage of restructuring our fixed income portfolio, which is to lengthen the maturity of our government bond holdings. Our last report (How to Hedge an Equity Sell-Off, September 18) showed that 7-10 year eurozone bonds no longer provide any offsetting gains when European equity markets decline. Given that they too have a negative yield, it is time to move out along the curve.

In due course, we intend to adopt the 15+ years index as our new benchmark. Over the last two years it has produced higher and more risk-efficient returns than the 7-10 year benchmark. This change cannot be done quickly. It depends on the stock of available bonds and investors may want greater clarity about the legal arrangements. But purely from a portfolio construction perspective, this is a desirable and sensible direction of travel.

Leave a Reply

Friday, September 18th, 2020

How to Hedge an Equity Sell-Off

Ever since the since the first asset allocation text book was written, the fundamental idea behind all the theoretical models is that when equities go down, government bonds will go up. If this relationship stops working, most balanced portfolios are in deep trouble. And yet that is exactly what our data suggest is already happening in the key 7-10 year maturity. Even if the relationship continues to work for longer maturities, it covers a much smaller part of the total bond universe than investors were previously able to rely on.

There are two ways of approaching this issue, the first is to look at current nominal yields to maturity and speculate how deeply negative they could go in a period of crisis. The problem with this approach is that it requires investors to make a big call about inflation over the next most of the time. The only scenario in which the offset could work is one of massive deflation. If it works in the next bear market, it can only do so by driving nominal yields so negative that they guarantee its failure in any subsequent sell-off.

The second is approach is to look at the data over the last 20 years and draw a trend line. Our first chart shows the weekly drawdown of US Equities on a total return basis. Most of the time, the reading is at or close to zero, which means that equities are rising, but there are three significant lows in 2002, 2009 and 2020. (We also look at a fourth episode in early 2016, which was not significant for the US, but was for many other countries.)

The other line tracks the cumulative return on bonds while the equity market is below its previous high, measured from the start of each equity sell-off. For each episode we take the maximum drawdown and compare it with cumulative return on bonds on the same day. In 2002, the worst weekly drawdown was 45.8% and the return on bonds was 28.7%, giving an offset ratio of 62%. We calculate this ratio for all four episodes for the US and six other developed equity markets: Japan, Australia, Canada, Switzerland, UK, and the Eurozone.

Our second chart plots this data over time and draws a trend-line which is quite terrifying in its simplicity. The average offset ratio was 50% in 2002, 32% in 2009, 18% in 2016 and 0% in 2020. As expected, there is some dispersion around the trend, but the overall R-squared is 0.68, which is good enough, and every country on its own has a downward sloping trend. This is despite the fact that we are dealing with a denominator effect. Recent equity sell-offs have been smaller than earlier ones, which means that a smaller bond gain would have generated a higher offset ratio.

We should always be wary of simplistic extrapolation. Our argument is not that all government bonds will go down in the next equity bear market, though this is exactly what happened this year in Australia, Switzerland and the Eurozone (including Germany and France). We argue that there will be a net zero offset in most countries, with some very small bond gains in the US and possibly Canada (see Chart 3).

Even here, the best we can realistically hope for is that bonds hold their value, rather provide a significant positive return. US investors used to be able to offset about 50% of their equity losses by owning 7-10 year Treasuries. It’s a big deal if this ratio falls to 5%.

There are (or were) three main reasons for holding long-dated government bonds: income, volatility reduction and negative correlation with equities (particularly when equities go down). The first has effectively vanished across most developed markets (especially if management costs are taken into account). This exercise suggests that the third reason won’t work very well in the next bear market. Only the second reason – volatility reduction – still works, but government bonds are not the only way of achieving this – cash or derivatives can do the same.

We will explore the implications of this in future notes, but here are some ideas.

  • It makes sense for investors to separate these three objectives, to look for income and negative correlation in different asset classes. Gold can provide some negative correlation, but has no income and can be hugely volatile; real estate will provide income but has a horrible positive correlation with falling equities; derivatives can hedge volatility, but only at a cost.
  • The only way investors can achieve all three objectives in government bonds is to go much further out on the yield curve. If the 7-10 tenor no longer fulfils their objectives, they will have to own 20, 30 and even 50-year maturities.
  • Investors will also have to vacate the 2-7 year space. This means there will be large quantities of medium-dated government debt with fewer natural holders, which could make them more volatile and more susceptible to international rather than domestic factors – i.e. the foreign exchange markets.

Conclusion: The key message is that for most developed markets, in the next equity sell-off, the part of the yield curve traditionally owned by balanced-portfolio investors cannot do what the text-books say it does – mitigate equity losses.

Leave a Reply

Friday, September 4th, 2020

Prices Move Before the News

Junior analysts sometimes have to be reminded about the old stock market saying that “If it’s in the press, it’s in the price.” Nobody can know everything all the time, but when enough people know enough information about a particular security, investors can be confident that the current price is fair and reasonable. The reason why investors do research is not to get an informational advantage – though sometimes this happens. It is to pool their knowledge so that there is greater transparency in the price formation process and greater liquidity as a result.

All this sounds like an introduction to a textbook about capital markets theory, but we want to make the opposite point: the importance of having a process which highlights changes in price for which there is no apparent explanation and which prompts investors to search for one. Over the course of the summer we have had a classic example of why this matters.  From early April to the end of June, our global equity models had an overweight recommendation on Japan which fluctuated in a narrow band between +25 to +35%. At the beginning of July this cracked suddenly. By the middle of the month, we were below benchmark. By the end of the month, we were underweight, with a recommended weight of -31%.

There was no apparent justification for the move – no change of policy by the BoJ, no sudden deterioration in Japan’s management of Covid, no big corporate scandal or earnings downgrade relative to the rest of the world. When asked by clients, the best we could come up with was the disappointing AGM season which had seen the majority of shareholder-friendly resolutions defeated.

However, there was a very big story, which was being deliberately suppressed: the resignation of Prime Minister Abe. Looking back, the timetable fits perfectly. According to the FT, “a regular medical check-up revealed signs of illness in June, from mid-July he began to suffer with bouts of exhaustion and his diagnosis was confirmed in early August.” From then on, it was an open secret because the LDP – his political party – was canvassing opinion as to who should succeed him. Our recommended weight hardly moved and has recovered slightly since the announcement on August 28th.

We make no apologies for not knowing the reason. The point is that we knew something was going on, because our process is sensitive enough to identify these situations and our weighting was adjusted automatically. Instead of fighting the change in prices because it wasn’t justified, we adapted and mitigated our risk. In the same spirit, here is a list of things we have noticed over the last month, which we don’t fully understand. All of them relate to a sudden change in direction in one of our sector or country charts – not an acceleration in a pre-existing trend.

For the bulls, we have a sudden improvement in the recommended weighting of UK Consumer Services relative to the UK index i.e. retail, media and leisure. This is strong enough to influence the Pan-Europe sector model, even though there is no equivalent move in the Eurozone model. We also have sudden upward moves in Chinese Consumer Services and Consumer Goods, which are offset by downward moves in Chinese Financials and Chinese Tech. There is also a downward move in Japanese Tech. We have frequently commented on the fact that big moves in Chinese Technology can be a lead indicator for the US Tech sector.

At the country level, we have sudden upward moves in Norway and Peru and downward moves in Hungary, Brazil, Portugal and Russia. We don’t argue that all of them are unexplained – or even important – but if you are thinking of doing something in these areas, you should make sure you know what’s going on.

Leave a Reply


    Full Blog Archive:


  • Search Blogs by Category