Category: Economics

There are two standardized mathematical methods of measuring raw performance (non-risk adjusted) and that is time-weighted and dollar-weighted returns. I won’t get into the definitions (go to investopedia for a good primer), but either method is legitimate. Typically funds that have lots of inflows and outflows use dollar-weighted returns, while time-weighted returns allows for an “apples to apples” performance comparison between funds.

Something we almost never question is the unit of measure that we use to evaluate such performance. This is most typically in the currency of the portfolio construction. For Canadians, this is usually in Canadian dollars, although it can be acceptable to use US Dollars as well.

However, is this yardstick, the currency, a correct one? It is very easy to overlook as this core assumption is baked into everybody’s consciousness that the currency is stable. Times are changing, however, where we now have to question the currency we use.

I looked at a couple alternative measures of past performance.

One alternative measure is the portfolio value in relation to the price of gold – i.e. performance in how many ounces of gold you could purchase at quarter-end. Using this metric, despite the fact that I have fared reasonably well in the COVID crisis, I have not done that much better today in relation to gold since the September 2018 quarter. In fact, going back as far as I could with some moderately reliable records, my performance, as strictly measured in gold, for the past 15 years has been about 7% compounded annually (compared to roughly 19% if you use a standard time-weighted return). This is a huge difference in performance – it shows that I am barely treading water.

If I use crude oil as a benchmark, I’m making out like gangbusters.

Another common measure is benchmarking to an index – relating your portfolio performance above or below the overall TSX Composite or S&P 500 is popular. Indeed, when I do my regular performance reports, I include those indices as a benchmark, simply because if I could do just as well (or better) sticking my money in an index fund, there’s no point in me bothering to do any investment research at all (indeed, I’d be adding negative value with every minute spent on investment research!).

I’m always cognizant that my past performance could purely be a function of luck. While depressing to think about, it is a possibility.

Another measure is benchmarking your portfolio to real dollars (instead of nominal dollars). This doesn’t change too much since reported CPI figures are under-reported (typically around 1.5% per year), so I do not consider this a useful measure. However, for marketing purposes, fund managers can claim victory in that most of them will show “your purchasing power will increase” – a 1.5% friction to nominal performance is not too difficult to overcome when central banks are applying a category 4 hurricane tailwind to your portfolio, especially if you invested in Tesla and Nio.

Is using real dollars to benchmark absolute performance correct?

I’d view M2 as a better barometer, as it reflects the fraction of currency that you control of the entire float of currency (that gets created through loans and central bank actions). This represents cash and cash-like instruments at banks. Over the past five years, it (USA) looked like this:

Very roughly, M2 has increased about 6% a year from 2016 to 2020. Going back earlier, we have the following:

We see during the economic crisis that the overall trend did not really change (there was a mild inflection) but during the COVID crisis, it really ramped up. Assuming the year ended today, M2 in the USA has increased by a factor of 24%. Needless to say, this is unprecedented in the past 40 years of financial history.

When applying M2 from 2005 to present, the growth has been about 7% compounded annually. Hence, my “real” performance as measured by USA M2 is about 12%.

Canada (Bank of Canada Statistics) has a similar M2 curve, please forgive my thirty second excel job on this dataset:

While their M2 is only refreshed up to September 1, 2020 (the USA data is updated weekly and is current), Canada’s M2 growth over the past 15 years is roughly 5% compounded. From January 1 to September 1, M2 growth in this year alone has been about 15%.

I generally believe that measuring absolute performance in relation to monetary aggregate growth is a more realistic measure than using CPI. There are still deficiencies using this as a yardstick, but the overall point of this exercise is to use different ways of measuring absolute performance.

Interestingly, if you relate USA M2 to the S&P 500 index, the price index return of the S&P 500 is roughly equal to the growth of the M2 over the past 15 years. The 15 year price return of the TSX Composite has lagged the growth in Canada’s M2 – using this metric, an index investor has barely kept afloat when you factor in dividends.

I deeply suspect we are all doing a lot worse in absolute terms than our nominal and real returns indicate – something to keep in mind when measuring our financial objectives.

I try to keep politics out of this site, except when politics has an impact on the marketplace. As long as this is the case, I’ll continue to comment about politics and this week definitely warrants some comment about it.

I’d like to preface this by explicitly outlining my level of emotional involvement in the presidential election. There is none. Whether Biden, Trump or anybody else wins the presidency, doesn’t matter much to me personally. I’m not like a spectator cheering for my preferred sports team; I’m a detached observer that is trying to predict who will win for the purposes of determining what set of policies will get implemented, and these policies will give certain sectors tailwinds that are worthy of investment, and certain sectors headwinds that you will want to steer clear. Look no further than any natural resource investment in Canada, especially post-Bill C-69, which is the incumbency protection bill for any company with an existing pipeline or natural resource project. Pipeline? Nope, unless if the government nationalizes it! Coal mine expansion? Sorry, you’re toast!

Again, winners and losers come out of any political changes in office. It is not my job to get happy or upset at any particular candidate or party, but rather to figure out what is likely to happen as a consequence and take appropriate action. My prediction of Trump winning is not out of any love for him but how I see it (although I will confess that he is ten times more entertaining to watch than Hillary Clinton or Joe Biden will ever be – it is undeniable he is a truly unique character in politics).

The point of this – some people confuse a prediction of a winner as some sort of endorsement. It is not.

Anyhow, my 2020 prediction was essentially correct. It doesn’t matter that Trump actually wins or loses the electoral votes in the states in question (the relevant ones being Arizona, Michigan, Pennsylvania, North Carolina, Georgia, Florida), the fact that Trump was polling 5-8 points behind most of these states should have made them all easy Democratic states. The real story was totally different. The public polling data was complete garbage (much more so than in a typical Canadian election) and one of the reasons for this was that the polling data was not generated for the purpose of ascertaining the voting outcome, but rather for fundraising. The other that I outlined was that they simply did not capture the composition of the actual people voting in elections.

Right now we have all the media outlets (which all are blinded by their hatred for Trump, which is ironic since the hatred amplified his political persona and elevated him to the presidency) claiming or nearly about to claim (I post this on a very late Thursday evening) that certain states have been won by Joe Biden, bringing him above the 269 electoral vote threshold, which should be sufficient for the presidency. The numbers on the electoral map will indeed show it. Half of the country will cheer thinking that Biden has won the presidency.

The reality, however, is going to be a lot more complex due to how the American electoral system works.

It is the electoral college that votes for a presidency. States approve the electors on the basis of the vote. This process is not automatic.

There will be a mountain of evidence assembled of various non-compliances of the election. This will get appealed up, and will get taken to the Supreme Court. The Supreme Court, despite being stacked with three Trump appointees, will want to wash its hands as thoroughly as possible, but they will give states some options. This is definitely a different scenario than Bush v. Gore – analogies will be made, but this analogy will be inappropriately used.

In the case of Pennsylvania, Georgia, Wisconsin, Arizona and Michigan, the state legislatures are Republican controlled. If the final reported vote margin is within a fraudulent margin provided for by evidence, it appears possible that the state can simply fail to certify the results of the vote and not choose electors. There is some precedent for this in 1864, which notably was in the middle of the civil war, which seems to be an appropriate term of what is going on in the USA right now.

If Biden was already up 300 to something, the remaining states wouldn’t have mattered. But if things are very tight (e.g. within 20 electoral votes), the closeness of the result is going to matter significantly if it is determined that one large state (e.g. Pennsylvania) is deemed to be fraudulent with their handling of mail-in ballots. With 20 electors abstaining, nobody can achieve 270 electoral college votes, and hence the vote for the president becomes a contingent election. It gets kicked to the House of Representatives, and they will have a choice of choosing the top three receivers of electoral college votes. While it is likely to be Trump, there could be a non-Biden, non-Trump elected president as long as there is a faithless elector willing to nominate a potential third person as a compromise president.

If the media claims Biden won, Biden gives his victory speech, and half the American public believes he won, and on the other side you have claims of election fraud, and also states clearing the way to (legally) not certifying the results or sending electors to Washington, DC, it amounts to a very contentious scenario which is probably to involve a lot of social unrest, especially on the Democratic side.

At one point after the election, British gambling sites were offering 15:1 odds on Trump winning the presidency given what we have seen numbers-wise. While I would not take such a bet simply because they would probably close it in Biden’s favour once the victory speech is given, I would not count Trump’s chances as being over yet. There is a very viable scenario where it becomes a contingent election, and then you’ll have to dust off the 1824 playbook, where this last occurred for presidency. This is the year 2020, and it is most certainly a year for strange things. Keep your mind open to the possibility.

With respect to the impact on the markets, the orgy of buying that we have seen in the past three days has been immense, but extreme caution is warranted in my opinion. This is not over at all.

I write this in the early afternoon of the Pacific Time Zone so the only true data I have to work with at present is the 16 votes for Trump and 10 votes for Biden in New Hampshire, which is not exactly representative of the national vote.

For reference, here was my November 3, 2016 presidential election prediction, where I predicted Trump 295, Clinton 243 (actual: Trump 304, Clinton 227 and 7 protest votes). The big “value-added” to that prediction, which the vast majority of others did not predict, was the breakout in Michigan.

Although all elections claim to be unique, in a sense they can be predicted to some extent given the historical correlations that various interest groups have with the red team or the blue team affiliations. Analytical minds in the major political parties try to ascertain which nudges (i.e. payouts to marginal special interests) to make to generate a winning coalition of voters. You get the numbers associated with each group, plug them into a paper napkin formula, and then build the campaign around such messaging. In 2016 it was the revival of the rust belt vs. establishment backroom politics. Trump’s team basically knew what he had to do, and aimed for a target in plain sight and got his deserved victory. Clinton’s campaign was basically “I’m not Trump”.

In 2020, times have changed. While in 2017 to early 2020 Trump was fighting the civil war in the bureaucracy (consider that a significant portion of the public service is Democratic and just because you’re the top guy in the top seat doesn’t mean that you’ll actually be able to implement policy instantaneously) and fulfilling various campaign pledges, COVID-19 completely destroyed the obvious re-election campaign pitch of “Keep America Great” to something a bit more diffuse. In this respect, COVID-19 was a net negative for Trump not because of the administration’s response to it, but because it destroyed the messaging narrative of the re-election campaign. In regards to the actual COVID-19 response, similar to Canada, the national government has limited control over the situation while provinces/states have a significant ‘say’ on what happens on the ground, which the public interprets as a national matter.

We look at the polling data. If you believe the polling data, you get an electoral map looking something like this:

In addition, congress will go Democratic, with both the House and Senate receiving Democratic majorities.

The question is – is this polling data correct? If it is, then this isn’t even going to be close. Biden wins by a mile. Indeed, in terms of the popular vote, states like California overwhelmingly will go Democratic, to the point of skewing the nation-wide popular vote by some wild margin towards the Democrats. The only reason why a Republican in California should bother up to vote is if their congressional race is in contention, or if they wish to vote in a state-wide initiative.

What polling data does not capture very well is motivation. Almost everybody you survey claims they will be voting and in reality the number is around 55-60%.  In elections where voters have the choice to not vote, it is just as important to model the cohort that vote versus those that do not.

From a more fundamental perspective, I ask what numbers of voters that voted for Donald Trump in 2016 would want to either sit this one out, or to vote Democratic – and the only people that would be in this category are the ones that believed all of the promises in 2016 (which they’d most likely turn into non-voters), or establishment Republicans that a restoration of the previous order (the Jeb Bushes, Mitt Romneys, Carly Fiorinas and the like). The first cohort is sizable; the second cohort would likely have not supported Trump in 2016.

On the flip side, there is evidence that the Republican coalition is expanding to include more of the ethnic voters (especially in the Latino/Black communities) and this has an impact in states like Florida, Arizona and Nevada. Indeed, these numbers are probably going to have a significant impact in states like Georgia, which marginally are polling Democratic this election.

Although impacted by COVID-19, you can also see that Trump has crowd-gathering abilities still, despite 4 years in office (which tend to depress voter enthusiasm, similar to how Obama’s re-election did not bring nearly the crowds that came in 2008). In fact, the crowds that still come to Trump rallies can only be described as insanely high – it is a politician’s wet dream to see such numbers coming (just imagine Trudeau trying to set something up like this).  This is contrasted with Biden, where even the most positive videos released by the Democrats don’t show that much (Bernie Sanders was much more successful in this metric).

The choice of Joe Biden is as close to a paper candidate as it gets – the whole world knows that he is not mentally functioning at a regular capacity, and that he would basically be a regent.  The VP selection did not perform very well in the campaign trail, despite hitting all of the ‘political correctness’ check-boxes.  Essentially, this renders the election as a referendum on Donald Trump – vote Democratic if you don’t like Trump, vote for Trump if you do. Are there more people upset with his presence today than they were in 2016?

Putting a long story short, to answer the original question, I believe the polling is skewed because they are not sampling the right cohorts. Here’s my guess, and it is awfully similar to 2016’s electoral map and projection:

The one state I would focus on for the Latino vote is Nevada – if the vote is relatively close (it was 47.9% Democratic and 45.5% Republican in 2016) then this state could potentially flip.

Unlike 2016, I have no money directly on the result of this election. The value received is too thin (Trump is +189, Biden is -215), compared to 4 years ago where Trump was being given away at +800 after the revealing of his politically incorrect comments.

Market-wise, no matter who wins, things are going to be in extremely rough shape. There will be a limit to the borrowing power of the US Government, and there will be a day of financial reckoning which will be extremely painful. An all-Democratic congress and presidency would be the worst outcome, while a divided congress historically is the best outcome for stability, which the markets like. But don’t believe any of these pundits that believe that Biden or Trump will be the best for the stock markets – I’d be playing the safety card.