Is it better to spend time budgeting or choosing asset managers?

So I read an interesting blog post on Noahpinion today, regarding whether ’tis better to ditch active fund management (e.g. follow Jack Bogle’s advice and stick with index-tracker ETFs, which charge very little fees), or to save more, in terms of retirement savings.  The article gets a bit economist-like, involving some basic utility functions and the like, to come up with a tentative ‘better to ditch active management’ conclusion.  The short blog post + comments are worth a read.

Anyway, this got me thinking.  And when I get thinking, I get modelling.  So here’s a spreadsheet which models the analysis I think of, when considering this problem.  Method:

  • Hypothesis: I think the time spent figuring out asset managers (whether passive or active) is probably about the same as creating a basic budget.  The latter allows for increased savings.  So let’s find out whether time is better spent ditching active management for passive, or creating a budget, in terms of accumulated retirement saving.
  • Data: for simplicity, I use the total returns of the S&P 500, data which comes from Aswath Damodoran.
  • Assumptions:
    • Starting salary of $50,000.
    • A savings rate, without budget, of 5% p.a.
    • Salary growth of 2% p.a., which is reflected in increased savings (i.e. we save 5% of the new, higher salary).
    • Performance drag of active management over passive of 2% p.a. So I assume active managers under perform a passive S&P 500 fund by 2% each year.
    • Budgeting increases savings by 3% p.a.  So a person who currently saves 5% p.a. can up that to 8% p.a. by making a budget.
  • Results: 
  • Budget or bin the manager?  Source: Damodoran Data.

    Budget or bin the manager? Source: Damodoran Data.

    • A Monte-Carlo simulation of 100 lifetimes (44 years of accumulated saving), randomly choosing years of S&P500 returns with replacement, brings the above picture.
    • Each run charts the net benefit, in terms of wealth at retirement age, of choosing to budget rather than switch the active manager for passive.  See the worst line there?  That’s happened because the luck of the draw meant lots of good years for the S&P 500 portfolio: in that case having the fee drag is a really bad thing versus just saving a bit more (however you’d be very rich in either case).
    • It turns out the ‘break even’ is around 4% for budget savings to exceed fee savings.  So if the savings ratio can be bumped from 5% to 9%, in this example, better to budget.
    • In brief: just choose the method that saves more.  Unless your budget increases your saving by a fair bit more than active fees (2% in this example), focus on active fees first.
    • In Noahpinion’s favour, one of his problems with this type of conclusion is that, because we can’t know how much more $1 today means to someone than $1 at retirement, we can’t use phrases like ‘better off’ to characterise the result.  All we can say is the wealth is higher at retirement.
    • Go ahead and play with the assumptions and see the accumulated benefit on the line chart, if you like!

In sum: suppose you’re holding some actively managed mutual funds, and are considering dumping them all for passive ETFs to save on fees, or even whether to dump one manager/ETF for another.  Before spending time with the withdrawal forms and whatnot, consider creating a simple budget to increase your savings.  The latter may pay off more in future.

Advertisements

Ignore John Bogle???

I was reading this Marketwatch piece this morning, and find the topic quite interesting. There is plenty of opinion out there that equal-weighted indices outperform market cap-weighted indices.  And when you look at RSP versus SPY, you indeed see the result.

Why would you choose RSP over SPY?

  1. Why index? The conventional reason folks choose stock indices (especially broad-market indices, such as SPY) is to diversify away specific company risk.  We know that holding equity should pay a return in the long run; we just don’t want to get unlucky choosing a bad equity.  So we invest in everything, looking to average returns.  RSP, by investing equal amounts in each of the S&P 500 constituents, has more diversification than SPY.
  2. What about rebalancing? The recent article by Campbell Harvey et al. is instructive here.  It turns out that regular rebalancing increases risk versus keeping a static portfolio.  So perhaps RSP pays more than SPY because it’s taking the extra risk.  Indeed, this is reflected by RSP’s 1.11 beta versus SPY: the former takes roughly 11% more risk than the same $ investment in SPY.  At least for the past year, the return of RSP has been about 1.11x SPY, so I guess the risk/return level is about commensurate.  Anyway, I could just suggest buying more SPY than buying RSP.
  3. What about momentum? See the same article.  SPY, like other market cap-weighted indices, implicitly take a momentum approach to the market. Because there is no rebalancing in SPY, the fund will automatically allocate more capital to stocks with higher returns (and thus higher market cap).  Given momentum is a lasting source of return, you’re essentially getting a trading strategy for free in choosing SPY over RSP.

So what to choose? If capital were no issue, I’d probably just buy more SPY than going smaller in RSP.  I like the lower management fees (yes, I do agree with John/Jack Bogle on that point), and appreciate the implied momentum returns of SPY.  If capital were an issue, I’d think of RSP like IWM: a way to achieve higher returns for the capital than SPY, mainly due to overweighting smaller companies.

Trouble getting on the housing ladder? There’s an ETF (or several) for that.

Ah, the likely murmurs of stone-faced City workers I pass in the morning: just keep earning the paycheque; have to pay the mortgage.

I’ve written before about my general thoughts on buying a house.  I was burned in the bubble-pop of 2009, so perhaps I’m jaded.  In any case, here are my general thoughts:

  • Pros of buying a house
    • Decorate it as I (or more likely, my wife) see(s) fit
    • A nice feeling
    • A diversifying investment, with the opportunity for very high leverage
    • A real option: if things get very bad, I can try to convince my wife to allow a lodger or two
  • Cons of buying a house
    • Continuous repairs + taxes + insurance
    • Huge capital outlay
    • Extreme illiquidity, such that valuation is very tricky and transaction costs are high. These effects are multiplied by the leverage used
    • Reduced financial and physical flexibility: harder to move for whatever reason

I like the diversifying aspect of housing, so can I get the good without the bad?  The short answer is yes, I can get the investment characteristics of housing without buying a house.  A good example is the iShares Dow Jones Real Estate ETF (IYR), which contains a basket of real estate holding companies, REITs, and developers.  The performance of the fund tracks the Case-Shiller 20-city Housing Price Index quite well:

No need for bricks or mortar.  Source: Quandl and S&P.

No need for bricks or mortar. Source: Quandl and S&P.

The fund is a bit more volatile than the index, which is probably at least part due to the illiquidity of measuring house prices.  In any case, the correlation is reasonable, and the major trends are captured.  The ETF is liquid, with OK-ish liquidity on the options for those wanting a leveraged investment.

As a kicker, the dividend yield for IYR is about 3.5%; that’s probably a bit lower than the net rental yield on a buy-to-let, but at least no one calls in the middle of the night needing a toilet unblocked.

Latest strategy I’m trying: monetising vol-drag

I wrote before about why I love UVXY and other leveraged ETFs.  In a nutshell, they’re completely designed for day-trading; if you hold one longer than a day, you’ll likely be disappointed with the performance.

Away from the pretty straightforward UVXY trade I mentioned, I’m now trying a similar strategy with IWM & TZA.  Thesis is similar:

  • TZA is -3x the daily % return of IWM.  So if IWM goes up 1% in a day, TZA falls 3%.
  • Unlike volatility, I find no reason to believe IWM should mean-revert.  So I need to hedge the underlying exposure of TZA with IWM.
  • Over time, TZA has similar issues as UVXY: volatility drag due to Jensen’s inequality, as well as negative roll yield (being short IWM futures, which have pretty stable backwardation).
  • I am hedging just to start the trade – over time, the hedge will be less effective; however, I use long put options, which means gamma should work in my favour.
  • Overall, this looks to replicate a long variance-swap position on IWM.  I should be getting credit (i.e. decay) for IWM moving around prior to option expiration.  If IWM doesn’t move around enough to cause the decay in TZA, I’m out the option extrinsic value for both sides (ouch).

I’m interested to see how this moves over time.  Hopefully not walking into a quagmire.  Anyone have experience with this trade?  Let me know your lessons!

On scalability

Thoughts on the walk home from the City this morning:

A lot of trading & asset management, in particular the areas with which I’ve been involved, rely on the concept of scalability or capacity.  That is, how many $$ can we put to work in a given strategy?

  1. The main concern in an institutional context is the limited scalability of many types of strategies.  Any strategy with ‘arbitrage’ in the title fits in this context, as do many ‘convergence’ (i.e. purchasing/selling under/overvalued, related assets) strategies.  For example, funds targeting convertible, capital structure, or merger arbitrage have a limited universe of securities to access.  This limits fund size.
  2. The shorter the holding period for strategies, generally the lower the capacity.  This is both due to explicit trading costs and available market liquidity.  For example, latency arbitrage (a common high frequency trading strategy) has very limited capacity from an institutional perspective: one can put $millions to work, but not $billions.
  3. My experience is there’s a negative correlation between capacity and risk-adjusted returns.  While Sharpe ratios of unlimited capacity strategies – such as long-only equities – tend to average around 0.5 (e.g. 7.5% return for 15% annual standard deviation), arbitrage strategies commonly have ratios of 2.0 or above.  Several HFT strategies have Sharpe ratios so high the ratios lose meaning (is a ratio of 25 really that much worse than a ratio of 50?)

From a personal account perspective, scalability takes on a different importance.  In particular:

  1. Many strategies and asset classes are too large for individual investors to access.  For example, a diversified trend-following strategy should really have positions in 20+ futures markets to ensure adequate diversification; otherwise the strategy hinges on too large a proportion of the portfolio trending at the same time.  The lot size for futures markets is such that a reasonable trend-following programme is likely impossible with fewer than $1 million AUM.  Are you deep enough for this to be only a minority of your overall portfolio?  I’m not…
  2. The fixed costs of implementing many strategies is simply too large for small investors to access.  HFT is the most extreme example of this, in my opinion: data feeds, tech expenses, and co-location fees can run into the $millions/year.  That’s a big nut to cover before turning a profit.

What are the lessons?

  1. Hire others to do strategies you can’t.  If you can find HFT groups taking money (they’re basically non-existent), or can invest in trend-following funds (now freely available as mutual/UCITS funds), that’s probably the best/only way to access some diversifying strategies.
  2. Don’t pay others for strategies you can do.  Long-only investing seems to fit here.  Cheap ETFs, please!
  3. Take advantage of lower-capacity strategies which big funds have problems accessing.  I put many options-related strategies here, as market depth & costs are reasonable for an individual investor, but prohibitive for most funds.

Why I love leveraged ETFs/ETNs…particularly UVXY

In a phrase: vol drag.

Volatility drag (‘vol drag’) is a product of compounding a return series which is multiplied by a constant.  In the case of UVXY, the return series is the movement in the VIX index, which is in turn a measure of S&P 500 30-day implied volatility.  The ETN aims to provide 200% of the daily change in the VIX index.

The immediate appeal of the product seems easy, at first glance: the VIX tends to move in opposite direction to the S&P, so why not buy VIX to hedge an equity portfolio.  While I’m at it, why not cut my hedging costs by buying a 2x version of the VIX?  Well, that’s where vol drag comes into play.

It’s important to mention that UVXY was, apparently, never created for investors holding more than 1 day.  Why?  Well, the first thing to mention is the VIX, like almost all underlyings, moves up and down: there is no smooth path for the VIX:

Screen Shot 2014-08-27 at 10.41.31

In fact, the VIX moves a lot: realised volatility for the VIX is around 125% these days, or around 8-10x the volatility of the S&P.  When we multiply that return series by 2, we increase the daily volatility by the same factor.  When we compound those returns daily, to make the price series for UVXY, we find a significant drag.  The difference between arithmetic returns (e.g. the 2x daily return achieved by UVXY) and geometric returns (e.g. the UVXY price series, which is a daily-compounded return series) is equivalent to 50% of the variance of the underlying (VIX), which is a big number.  See here for more explanation.

There’s another reason I love UVXY, which is persistent futures contango.  Because the VIX is not directly investable, UVXY invests in a combination of front & 2nd month VIX futures to achieve its objective.  The persistent contango is, ironically, likely a result of so many folks buying VIX ETFs/ETNs – more equity hedgers than speculators.  Anyway, because later-dated futures contracts cost more than nearer-dated, or indeed cash VIX, UVXY (and other VIX ETNs) have a persistent negative drag from the contango.

How does it all add up?  Here is YTD for UVXY versus its underlying, the VIX index.  A reminder: the objective of the ETN is to provide 200% of the daily return of the VIX; the persistent decay of UVXY is due to the factors explained above (in addition to a third, small, factor: a 0.75% annual management fee):

Screen Shot 2014-08-27 at 10.58.23

So, what to do with this?  The main risk to mention with UVXY is massive skew and kurtosis: that is, the very real risk for a large upward move when the VIX spikes.  So while I look at a picture like the above and say ‘Go short!’, I think back to times like 2011 where UVXY explodes both because of a spike in VIX, and because of a swift change in futures contango:

Screen Shot 2014-08-27 at 11.02.22

Yes, UVXY can blow up.  So, the answer is a combination of going short UVXY with buying upside protection.  If the protection can be bought reasonably (it usually can), and the investor can stomach the big jumps (knowing max loss from inception is helpful here), it seems a good trade.  Extra credit: wait for a VIX spike before opening a trade.  

(Source for charts was Google Finance)