Sorry Bears, We're In A Secular Bull Market [View article]
Jonathan - A secular bull market is generally defined as a period when price-to-earnings multiples are in an uptrend. Between 1900-2000, secular bull markets typically started after the cyclically adjusted P/E ratio was under 10 for a significant period and ended (eg. start of secular bear) with the CAPE above 20. In the current secular bear, the CAPE touched 13.3 briefly in March 2009, sits at over 23 right now and remains in a downtrend from the high of 44 in 1999. With corporate profit margins currently well above average as a percentage of GDP - making the TTM P/E and earnings yield calculation look better than they are, in my opinion - I think there may still be some valuation excess to work off yet in this secular bear cycle.
Who's Afraid Of The Big, Bad QE? M2 Growth, Inflation And P/E Ratios [View article]
Mubirasymbu - Over the last five years, Output Per Person has grown at 1.7% per annum compounded, similar to the early-mid 1970s and the mid 1990s. By contrast, the technological revolution of the mid-1990s-mid 2000s caused Output Per Person to routinely grow at rates above 2% (and even above 3%) over rolling 5-year periods. After trending up from the early 1980s until the mid-2000s (drive by women entering the workforce, entry of billions of people from China, India, Post-Soviet states into the workforce, technological advances, etc), productivity growth seems like it might be entering a period of stagnation or even secular decline.
My worry is that disinflationary forces prevalent from 1980-now may be dissipating at a time when we are also printing a lot of money. The effect will not be immediate, but would be longer term and secular in nature.
No Margin For Error: Margin Debt, Quantitative Easing, And Future S&P 500 Returns [View article]
Hello Dorky - The issue here is the transmission mechanism between the Fed quantitative easing and the stock market. The Fed purchases assets from various parties...the money eventually winds up as deposits in the banking system...the deposits are then used to purchase assets directly or make loans (eg. margin loans), which can then be used by market participants such as hedge funds to buy stocks.
My article did not explicitly argue for a collapse. It argued that increasing margin debt adds to risks of a systemic "de-margining" cycle, which could turn a normal market correction into something deeper.
Growth in leverage increases the tight coupling of the financial system, making it harder for markets to "self-arrest" while in decline because of forced selling. Physicists such as Didier Sornette, who did his early work developing a model for predicting rocket-fuel tank ruptures using fractal geometry, have actually had a good record of predicting what they call "critical events" in financial markets by looking for log-periodic sequences in price fluctuations that suggest increasing "self organization" in the little market fractures that make up a typical correction (read the following paper for more information on log-periodic power laws http://bit.ly/WTQ6uW and this one for recent thoughts from Sornette http://bit.ly/We3u0r).
(Uh) Oh Canada! A Developing Minsky Moment In Canadian Real Estate? [View article]
The RBC report I referenced definitely shows that BC prices (Vancouver especially) are more overextended than other metro areas. However, if I take BC (Vanc and Vict) and Toronto out of the house price index and take a simple average of the rest, price appreciation has actually been pretty similar, averaging across Winnipeg, Calgary, Montreal, Halifax, etc. As for prices dropping being good for first time home buyers, I agree. It sure is here. My younger first time homebuyer friends are getting some amazing deals in the Seattle metro area, if they have good credit and can get the loans to do so. I think renting right now, saving up money as much as possible for a downpayment in the future and biding your time is a great strategy in this type of real estate market.
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
Hello Stefan - Thanks so much for your comment. I think you make good points about deltaCAPE and Tobin's Q. I was trying to keep it pretty simple, but looking to add these, as well perhaps as the good old dividend discount model, would be a good idea.
With respect to the R^2 of the final model, I actually didn't put them all together into a multiple regression because I was concerned with multi-collinearity and overspecifying the model. I wanted to show the range and then put them together into a simple weighted average that could provide sort of an indicative point prediction, while allowing people to still make their own decisions based on the various indicators. That said, the correlation between actual future 10-year real returns and the expected return time series formed using the various models and weighted averages was over 0.50, higher than any of the single indicators.
Deleveraging Accelerating: Risk Rising Of Accelerating SPY Sell-Off [View article]
Buying index volatility in general, both calls and puts, has historically been a money loser across many different indexes because of the gap between implied and realized volatility, which likely has something to do with irrational investor preferences for the positive skewness of both out of the money calls and puts (eg. lots of little losses and a fat right tail of potential huge gains), as well as the rational explanation that you are essentially buying insurance and must pay a premium for that. However, during volatility clusters, subsequent realized volatility has rocketed above implied over short periods. The profit would depend on what regime we were in.
(Uh) Oh Canada! A Developing Minsky Moment In Canadian Real Estate? [View article]
Hi Chizquiyahu - Good point. Speaking from the US experience, residential REITs, like Equity Residential (EQR) or Avalon Bay Communities (AVP), for instance, really underperformed the broader market (S&P 500) as the Minsky moment began in US real estate in early 2007. They then outperformed as the major liquidity crisis hit in late 2008. My thought is that, if property prices turn down a bit more severely in Canada as I think they might, investors will add a risk premium to anything "real estate" related, as they seem to have in the US in 2007.
(Uh) Oh Canada! A Developing Minsky Moment In Canadian Real Estate? [View article]
Thank you, DirtyDozen! I agree. In Orange County, CA, where I grew up, the four bedroom townhouse that my mom bought in 1992 quadrupled in value by 2006. Here in Tacoma, WA, where I live, big new condo complexes were going up down by the waterfront, all of the apartment buildings were converting to condos, and people listing their houses would get 10 offers on the first day, some without the buyers even walking through the house. Now, after values declined 30% in three years (including my house, unfortunately), the half-built hulks of condo buildings are finally being finished (after having been bought by investors out of the portfolios of the failed local banks that financed them), many of the condo conversions have converted back to rentals, and people selling their houses are lucky if they can get asking price in a reasonable time frame.
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
Hello BeenAroundTooLong - Thanks so much for your comments on methods.
I visually checked the relationships for linearity using scatterplots and then regressed the indicator levels against the 10yr expected returns. I did not hold the intercept to zero or transform the indicators in the regressions, as none seemed to have an exponential relationship with the dependent variable (returns).
The regressions do suffer from serial correlation of the residuals (eg. Durbin-Watson statistics low enough to reject the null of no serial correlation) and, to a lesser extent, heteroskedasticity (Breusch-Pagan tests of varying significance). These issues primarily affect the standard errors and the confidence intervals, though, and should not have an impact on coefficients, meaning that the point predictions themselves should be valid, but the confidence intervals (wide as they are already) might actually be wider.
I was okay with this for this type of analysis because I wasn't trying to tell people that expected returns were absolutely going to be 2% per year over the next 10 years, but rather to try to use the analysis as a barometer to show people whether "pressure" was generally rising (eg. valuations stretched) or falling (eg. valuations easy) and give a prediction context around that. I hope I made that point sufficiently clear in the text. If I did not, I apologise.
Where the standard error issue may have showed up, other than the confidence intervals, is in the weighted average, as I did use the R^2 values to construct the weights. I went back and did a simple average on the point predictions from the 8 indicators and got an average prediction of 2.3%, pretty close to the 2.1% I mentioned in the article.
With respect to normality, I went back and checked the residuals and found that some of the normal probability plots showed more of an S-shape than I would like, suggesting skewness and/or kurtosis. So, I did a Log10 transformation on the dependent variable and all of the independent variables to ensure normality and re-ran the regressions. All of the normal probability plots now fit pretty closely to the diagonal line. The simple average of the predictions from the Log10 regression is 1.9%, also pretty close to the 2.1%.
Finally, with respect to regime shifts in the data (your comment on 1890 vs 1971), in terms of valuations, I would argue that the CAPE, 10-yr returns and other series seem more stochastic over the 110 year period rather than having any huge regime shifts.
Thank you again for taking them time to question methods and add value and depth to the analysis!
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
By the way, I apologize for two typos that I apparently missed in my final proof-read.
The first sentence in the last paragraph of the sub-section titled "10-year Real Expected Returns of 2.1% Per Year" should read: "A 2.1% real equity return over ten years, or a 5.1% per year nominal return if we add 3.0% annual CPI inflation...". I had originally put 2.5%, which is roughly the rate implied by the 10-year TIPS, though I decided before publication to bump that up a bit to 3.0%.
Also, the first sentence in the last paragraph of the sub-section titled "Medium-Term Downside Risks" should read: "In the current secular bear market, we have experienced two major decreases in valuation with the most recent reaching a de-trended CAPE of 0.62...".
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
Hello Jim - I appreciate your comment. Please know that I did not intend at all to use the Lord's name in vain. I believe that I quoted Him faithfully in the beginning of the article and tried to apply His words to the context of investing. Even though they were originally spoken in a spiritual context, I think we can reflect on them in the practical contexts of our lives.
Credit Acceptance: Cruising For A Bruising [View article]
Hello Michael - Thanks so much for your good comments.
CACC's model rests on number of active dealers, sub-prime loan originations per dealer, and average loan value to replace the roughly 25% of its loans receivable that run off each year (eg. ~43 month average term from 2009-12).
With respect to continuing to grow active dealerships, and your point about a long runway in a fragmented market, I agree with you to a point. The annual NIADA Used Car Industry Reports (2011 was most recent that was available for now charge) show that the number of independent dealerships fell from 49,896 in 2004 to 37,717 in 2010. Of these, as of 2010, 37.8% had a relationship with a finance company, whereas the remainder were strictly buy-here-pay-here dealers or had relationships with banks. That means that the total, immediate addressable market for CACC in 2010 was a little over 14,000 dealerships, for which their 3,200 active dealers represented about 23% market share. As of end 2012, they had 5,300 active dealerships. Even if the total number of dealerships vaulted back up to 50,000 between 2010 and 2012, CACC's 5,300 dealerships would represent 30% market-share, substantial penetration into a fragmented market. They may be able to penetrate further into this market, or to convert dealerships that have traditionally worked with banks (or been BHPH and impacted by recent California legislation). However, my experience with selling into fragmented markets is that it is very difficult to expand out of your niche without an expensive sales and marketing effort. They have been running off an average of 1,250 dealerships per month since 2007 and adding an average of 1,760/month, enabling them to growth from a little under 3,000 active dealerships to 5,300 active dealerships. A slowdown in their ability to convert new dealers - due to increased competition, which Douglas Busk mentioned, or difficulty with their sales effort - would negatively impact their ability to replace the dealerships running off. That is when the decline in originations per dealer would start to hurt.
With respect to originations per dealer - and to your point about new vs used sales - growth in used vehicle sales is projected to slow as well. CNW Marketing Research, which produces the data used in the National Independent Automobile Dealers Association annual market review, projects used car sales will rise from 40,500,000 in 2012 to only 41,250,000 in 2014, or only 0.9% per year (http://bit.ly/SFtxf0). In February 2013, used car sales by independent auto dealers, which form CACC's main market, were increasing at a 4.4% year-over-year rate (http://bit.ly/12GVudO). However, the February 2013 report shows sub-prime demand for used cars to still be growing strongly, with sub-prime buyers up 53% year-over-year, which could be supportive for CACC's dealership enrollment effort and loan originations.
With respect to leverage and yield, I keep coming back to Busk's comment from the most recent earnings call comparing the current environment to 2006-07, in terms of abundant capital, increased competition and...increasing leverage ratios. Forecast collections are down, but dealer advances are up (ostensibly because of increased competition for originations), meaning that spreads - and by extension yields - are down. If cash flows underperform forecasts due to an increase in delinquencies, yield will fall further. It strikes me as an example of a company "dancing while the music is going." And financing this growth through greater leverage. Indeed, decomposing the increase in Return-on-Equity over the past 10 years using DuPont Analysis (eg. ROE= Net Margin x Asset Turnover x Financial Leverage), I found that the growth in financial leverage was the greatest contributor to the growth in ROE, followed by net margin (which is influenced by it, since increased leverage leads to increased revenues and contribution margin to pay for fixed costs).
Finally, if loans receivable growth simply stagnated for a couple of years and then returned to 10-year average growth - and yield, net margin, etc, were at 10-year average levels, the company would be worth around $110/share, below its current level.
You make sound arguments, but I see more downside risks from slowing used car sales, an eventual slowdown in growth of sub-prime buyers, the need to keep churning high numbers of dealerships in a finite market, the potential for yields to fall further due to increased delinquencies following a period of aggressive dealer advances, and the threat of financing becoming less available following a period of abundant, cheap capital.
False Prophets: Transports And Small Caps Outperform [View article]
Hi Berloe - It is partially because they are commission free. With any dynamic strategy, keeping transaction costs low (or nonexistent) is good. Also, these strategies are best in a tax-free account, as a taxable account would run into issues of short-term versus long-term gains (although, most of the holdings would be for more than 12 months, especially in the trend-following base strategy).
More broadly, for the momentum strategy, I wanted there to be a broad mix of asset classes. For instance, I have four U.S. equity styles (small cap value & growth, large cap value & growth), developed international equities, emerging intl equities, and several credit asset classes. The iShares themselves are not the secret sauce - if you can trade Vanguard or other ETFs commission free in your IRA platform, you can replicate this strategy using those. A Treasury ETF and a Commodities ETF would be good additions to this basic momentum strategy. Also, sector and country ETFs could as well if you have access to those, as I have separately found momentum rules to be profitable with those as well. However, there reaches a point of diminishing returns, as it is cumbersome to pull the adjusted price series on Yahoo Finance each month the more securities you use. I think 11-13 ETFs that span domestic and intl equities, bonds and commodities are a great universe from which to run the momentum strategy.
With regards to the trend following strategy, I chose those five asset class ETFs to mimic the portfolio proposed by the Mebane Faber article I linked above.
With regards to the trading rule window, I used one month because it conformed to Faber and others and seemed to strike a good balance between capturing the information in the trend or momentum signals while not trading too much (or requiring too much work). However, you can test shorter or longer periods, especially if you are using commission free ETFs and executing the strategies in a tax-free account.
As I said in my response above, it is important to remember that these strategies can underperform a simple buy-and-hold strategy for extended periods while markets are doing well. However, if you believe that risks are double-sided, I think making use of these strategies in your overall plan - especially when combined with a value equity buy-and-hold strategy - can be a great way to maintain exposure to equities if markets continue to trend up while limiting potential drawdowns if downside risks materialize.
False Prophets: Transports And Small Caps Outperform [View article]
Hello AlphaSeekerChris - Thanks for your question. On the first day of every month I pull the Yahoo Finance Adjusted Prices (daily frequency) for the following iShares ETFs that are available to trade at no cost on Fidelity: TIP Treasury Infl Protected Securities LQD Investment Grade Corp Bonds HYG High Yield Corp Bonds EMB Emerging Market Bonds MUB Municipal Bonds EFA Foreign Developed Mkt Stocks (MSCI EAFE Index) EEM Foreign Emerging Mkt Stocks IWF Large Cap Growth (Russell 1000 Growth) IWD Large Cap Value (Russell 1000 Value) IWO Small Cap Growth (Russell 2000 Growth) IWN Small Cap Value (Russell 2000 Value)
These ETFs represent four major styles of U.S. equities (Lg Cap and Sm Cap Growth and Value), foreign developed and emerging market equities, four credit asset classes (Munis, High Yield, Investment Grade and Emerging Market bonds) and one Treasury bond asset class (TIP). The one major whole is that there is no pure U.S. Treasury ETF on there to provide a true safe haven during a liquidity crisis.
I look at the return for each over the last 6 months (or, more accurately, over the last 120 trading days) and, by the end of the first trading day of the month, own the ETF with the best 6-month record as of the previous day's (end of month) close. As of end December 2012, the strategy owned EFA (foreign developed market equities), which it had bought at beginning of December and previously had owned EMB (emerging market bonds) before that.
The "oldest" two ETFs in the sample (Large Cap Growth and Value) have data only since May 2000, which means my backtest started only at the end of October 2000, approximately 12 years of history. From October 2000 to November 2012, this momentum strategy returned a compound 15.6% annually with an annualized standard deviation of 21.2%, beating the S&P 500 (IVV) return of 1.6% annually and 21.2% standard deviation. The strategy was not immune to drawdowns - its lowest rolling one-year returns were -24.1% in March 2003 and -21.3% in October 2008, better than the S&P 500 rolling 1 year returns of over 30% and 40%, respectively, in those periods.The backtest period is relatively short (eg. just 12 years) but the general performance was consistent with the finds of the Asness article mentioned in my article and of various other articles on momentum.
The strategy has also had periods of relative outperformance and underperformance during that time, with underperformance in recent times mainly coming during the strong snap-back rallies following large drawdowns, where the strategy was still positioned relatively defensively (eg. owned one of the bond funds) compared to a pure buy-and-hold equity strategy. It is currently coming off a period of relative underperformance stemming from the rally following the Aug-Oct 2011 correction.
I use this strategy as a counterweight to a value equity position - as recommended by Asness et al in the "Value and Momentum Everywhere" article I linked - and to a multi-asset trend following strategy (like the one discussed by Mebane Faber in the other article I linked above) that also mainly uses the commission free ETFs on Fidelity (IWV - Russell 3000, AGG - Barclays Total Bond, ACWX - Global ex US, and IYR - REITS), as well as DBC - Deutsche Bank Commodity ETF for the commodity portion, and the 200-day moving average as the trend. The trading rule intuition with the trend following strategy is similar to that of the momentum strategy - on the first day of each month, look at whether each ETF adjusted price was above its 200-day moving average as of the close on the previous trading day (eg. end of month). By the end of the first trading day of the month, own the ETF if above or equal to the 200-day and don't own it if below the 200-day moving average.
Each strategy (value, momentum, trend) captures a different source of expected return (see Antti Ilmanen, "Expected Returns" for a detailed treatment of risk factors and sources of expected returns) and is not perfectly correlated with the other, meaning that you can basically allocate between them like you would allocate between asset classes. Right now, I personally allocate 50% to the trend strategy, 25% to value and 25% to momentum. The back-testing history for this allocation is even shorter than for the momentum (because of the short history of commodities ETFs), but since November 2006 it achieved a compound return of 6.3% with a volatility of 13.6%, besting the 1.52% return and 24.2% standard deviation of the S&P 500 during that period.
Deleveraging Accelerating: Risk Rising Of Accelerating SPY Sell-Off [View article]
I stand corrected! I was referring more to the virtuous/vicious cycle dynamic and inadvertently used the feedback-loop term incorrectly. Thank you for bringing that up.
Sorry Bears, We're In A Secular Bull Market [View article]
Who's Afraid Of The Big, Bad QE? M2 Growth, Inflation And P/E Ratios [View article]
My worry is that disinflationary forces prevalent from 1980-now may be dissipating at a time when we are also printing a lot of money. The effect will not be immediate, but would be longer term and secular in nature.
No Margin For Error: Margin Debt, Quantitative Easing, And Future S&P 500 Returns [View article]
My article did not explicitly argue for a collapse. It argued that increasing margin debt adds to risks of a systemic "de-margining" cycle, which could turn a normal market correction into something deeper.
Growth in leverage increases the tight coupling of the financial system, making it harder for markets to "self-arrest" while in decline because of forced selling. Physicists such as Didier Sornette, who did his early work developing a model for predicting rocket-fuel tank ruptures using fractal geometry, have actually had a good record of predicting what they call "critical events" in financial markets by looking for log-periodic sequences in price fluctuations that suggest increasing "self organization" in the little market fractures that make up a typical correction (read the following paper for more information on log-periodic power laws http://bit.ly/WTQ6uW and this one for recent thoughts from Sornette http://bit.ly/We3u0r).
Best - Bard
(Uh) Oh Canada! A Developing Minsky Moment In Canadian Real Estate? [View article]
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
With respect to the R^2 of the final model, I actually didn't put them all together into a multiple regression because I was concerned with multi-collinearity and overspecifying the model. I wanted to show the range and then put them together into a simple weighted average that could provide sort of an indicative point prediction, while allowing people to still make their own decisions based on the various indicators. That said, the correlation between actual future 10-year real returns and the expected return time series formed using the various models and weighted averages was over 0.50, higher than any of the single indicators.
Best - Bard
Deleveraging Accelerating: Risk Rising Of Accelerating SPY Sell-Off [View article]
(Uh) Oh Canada! A Developing Minsky Moment In Canadian Real Estate? [View article]
(Uh) Oh Canada! A Developing Minsky Moment In Canadian Real Estate? [View article]
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
I visually checked the relationships for linearity using scatterplots and then regressed the indicator levels against the 10yr expected returns. I did not hold the intercept to zero or transform the indicators in the regressions, as none seemed to have an exponential relationship with the dependent variable (returns).
The regressions do suffer from serial correlation of the residuals (eg. Durbin-Watson statistics low enough to reject the null of no serial correlation) and, to a lesser extent, heteroskedasticity (Breusch-Pagan tests of varying significance). These issues primarily affect the standard errors and the confidence intervals, though, and should not have an impact on coefficients, meaning that the point predictions themselves should be valid, but the confidence intervals (wide as they are already) might actually be wider.
I was okay with this for this type of analysis because I wasn't trying to tell people that expected returns were absolutely going to be 2% per year over the next 10 years, but rather to try to use the analysis as a barometer to show people whether "pressure" was generally rising (eg. valuations stretched) or falling (eg. valuations easy) and give a prediction context around that. I hope I made that point sufficiently clear in the text. If I did not, I apologise.
Where the standard error issue may have showed up, other than the confidence intervals, is in the weighted average, as I did use the R^2 values to construct the weights. I went back and did a simple average on the point predictions from the 8 indicators and got an average prediction of 2.3%, pretty close to the 2.1% I mentioned in the article.
With respect to normality, I went back and checked the residuals and found that some of the normal probability plots showed more of an S-shape than I would like, suggesting skewness and/or kurtosis. So, I did a Log10 transformation on the dependent variable and all of the independent variables to ensure normality and re-ran the regressions. All of the normal probability plots now fit pretty closely to the diagonal line. The simple average of the predictions from the Log10 regression is 1.9%, also pretty close to the 2.1%.
Finally, with respect to regime shifts in the data (your comment on 1890 vs 1971), in terms of valuations, I would argue that the CAPE, 10-yr returns and other series seem more stochastic over the 110 year period rather than having any huge regime shifts.
Thank you again for taking them time to question methods and add value and depth to the analysis!
Best Regards - Bard
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
The first sentence in the last paragraph of the sub-section titled "10-year Real Expected Returns of 2.1% Per Year" should read: "A 2.1% real equity return over ten years, or a 5.1% per year nominal return if we add 3.0% annual CPI inflation...". I had originally put 2.5%, which is roughly the rate implied by the 10-year TIPS, though I decided before publication to bump that up a bit to 3.0%.
Also, the first sentence in the last paragraph of the sub-section titled "Medium-Term Downside Risks" should read: "In the current secular bear market, we have experienced two major decreases in valuation with the most recent reaching a de-trended CAPE of 0.62...".
Jesus, Bob Dylan And Expected Returns: 8 Indicators Of Long-Term Equity Returns [View article]
Best - Bard
Credit Acceptance: Cruising For A Bruising [View article]
CACC's model rests on number of active dealers, sub-prime loan originations per dealer, and average loan value to replace the roughly 25% of its loans receivable that run off each year (eg. ~43 month average term from 2009-12).
With respect to continuing to grow active dealerships, and your point about a long runway in a fragmented market, I agree with you to a point. The annual NIADA Used Car Industry Reports (2011 was most recent that was available for now charge) show that the number of independent dealerships fell from 49,896 in 2004 to 37,717 in 2010. Of these, as of 2010, 37.8% had a relationship with a finance company, whereas the remainder were strictly buy-here-pay-here dealers or had relationships with banks. That means that the total, immediate addressable market for CACC in 2010 was a little over 14,000 dealerships, for which their 3,200 active dealers represented about 23% market share. As of end 2012, they had 5,300 active dealerships. Even if the total number of dealerships vaulted back up to 50,000 between 2010 and 2012, CACC's 5,300 dealerships would represent 30% market-share, substantial penetration into a fragmented market. They may be able to penetrate further into this market, or to convert dealerships that have traditionally worked with banks (or been BHPH and impacted by recent California legislation). However, my experience with selling into fragmented markets is that it is very difficult to expand out of your niche without an expensive sales and marketing effort. They have been running off an average of 1,250 dealerships per month since 2007 and adding an average of 1,760/month, enabling them to growth from a little under 3,000 active dealerships to 5,300 active dealerships. A slowdown in their ability to convert new dealers - due to increased competition, which Douglas Busk mentioned, or difficulty with their sales effort - would negatively impact their ability to replace the dealerships running off. That is when the decline in originations per dealer would start to hurt.
With respect to originations per dealer - and to your point about new vs used sales - growth in used vehicle sales is projected to slow as well. CNW Marketing Research, which produces the data used in the National Independent Automobile Dealers Association annual market review, projects used car sales will rise from 40,500,000 in 2012 to only 41,250,000 in 2014, or only 0.9% per year (http://bit.ly/SFtxf0). In February 2013, used car sales by independent auto dealers, which form CACC's main market, were increasing at a 4.4% year-over-year rate (http://bit.ly/12GVudO). However, the February 2013 report shows sub-prime demand for used cars to still be growing strongly, with sub-prime buyers up 53% year-over-year, which could be supportive for CACC's dealership enrollment effort and loan originations.
With respect to leverage and yield, I keep coming back to Busk's comment from the most recent earnings call comparing the current environment to 2006-07, in terms of abundant capital, increased competition and...increasing leverage ratios. Forecast collections are down, but dealer advances are up (ostensibly because of increased competition for originations), meaning that spreads - and by extension yields - are down. If cash flows underperform forecasts due to an increase in delinquencies, yield will fall further. It strikes me as an example of a company "dancing while the music is going." And financing this growth through greater leverage. Indeed, decomposing the increase in Return-on-Equity over the past 10 years using DuPont Analysis (eg. ROE= Net Margin x Asset Turnover x Financial Leverage), I found that the growth in financial leverage was the greatest contributor to the growth in ROE, followed by net margin (which is influenced by it, since increased leverage leads to increased revenues and contribution margin to pay for fixed costs).
Finally, if loans receivable growth simply stagnated for a couple of years and then returned to 10-year average growth - and yield, net margin, etc, were at 10-year average levels, the company would be worth around $110/share, below its current level.
You make sound arguments, but I see more downside risks from slowing used car sales, an eventual slowdown in growth of sub-prime buyers, the need to keep churning high numbers of dealerships in a finite market, the potential for yields to fall further due to increased delinquencies following a period of aggressive dealer advances, and the threat of financing becoming less available following a period of abundant, cheap capital.
Best - Bard
False Prophets: Transports And Small Caps Outperform [View article]
More broadly, for the momentum strategy, I wanted there to be a broad mix of asset classes. For instance, I have four U.S. equity styles (small cap value & growth, large cap value & growth), developed international equities, emerging intl equities, and several credit asset classes. The iShares themselves are not the secret sauce - if you can trade Vanguard or other ETFs commission free in your IRA platform, you can replicate this strategy using those. A Treasury ETF and a Commodities ETF would be good additions to this basic momentum strategy. Also, sector and country ETFs could as well if you have access to those, as I have separately found momentum rules to be profitable with those as well. However, there reaches a point of diminishing returns, as it is cumbersome to pull the adjusted price series on Yahoo Finance each month the more securities you use. I think 11-13 ETFs that span domestic and intl equities, bonds and commodities are a great universe from which to run the momentum strategy.
With regards to the trend following strategy, I chose those five asset class ETFs to mimic the portfolio proposed by the Mebane Faber article I linked above.
With regards to the trading rule window, I used one month because it conformed to Faber and others and seemed to strike a good balance between capturing the information in the trend or momentum signals while not trading too much (or requiring too much work). However, you can test shorter or longer periods, especially if you are using commission free ETFs and executing the strategies in a tax-free account.
As I said in my response above, it is important to remember that these strategies can underperform a simple buy-and-hold strategy for extended periods while markets are doing well. However, if you believe that risks are double-sided, I think making use of these strategies in your overall plan - especially when combined with a value equity buy-and-hold strategy - can be a great way to maintain exposure to equities if markets continue to trend up while limiting potential drawdowns if downside risks materialize.
Thanks again for your question!
Best - Bard
False Prophets: Transports And Small Caps Outperform [View article]
TIP Treasury Infl Protected Securities
LQD Investment Grade Corp Bonds
HYG High Yield Corp Bonds
EMB Emerging Market Bonds
MUB Municipal Bonds
EFA Foreign Developed Mkt Stocks (MSCI EAFE Index)
EEM Foreign Emerging Mkt Stocks
IWF Large Cap Growth (Russell 1000 Growth)
IWD Large Cap Value (Russell 1000 Value)
IWO Small Cap Growth (Russell 2000 Growth)
IWN Small Cap Value (Russell 2000 Value)
These ETFs represent four major styles of U.S. equities (Lg Cap and Sm Cap Growth and Value), foreign developed and emerging market equities, four credit asset classes (Munis, High Yield, Investment Grade and Emerging Market bonds) and one Treasury bond asset class (TIP). The one major whole is that there is no pure U.S. Treasury ETF on there to provide a true safe haven during a liquidity crisis.
I look at the return for each over the last 6 months (or, more accurately, over the last 120 trading days) and, by the end of the first trading day of the month, own the ETF with the best 6-month record as of the previous day's (end of month) close. As of end December 2012, the strategy owned EFA (foreign developed market equities), which it had bought at beginning of December and previously had owned EMB (emerging market bonds) before that.
The "oldest" two ETFs in the sample (Large Cap Growth and Value) have data only since May 2000, which means my backtest started only at the end of October 2000, approximately 12 years of history. From October 2000 to November 2012, this momentum strategy returned a compound 15.6% annually with an annualized standard deviation of 21.2%, beating the S&P 500 (IVV) return of 1.6% annually and 21.2% standard deviation. The strategy was not immune to drawdowns - its lowest rolling one-year returns were -24.1% in March 2003 and -21.3% in October 2008, better than the S&P 500 rolling 1 year returns of over 30% and 40%, respectively, in those periods.The backtest period is relatively short (eg. just 12 years) but the general performance was consistent with the finds of the Asness article mentioned in my article and of various other articles on momentum.
The strategy has also had periods of relative outperformance and underperformance during that time, with underperformance in recent times mainly coming during the strong snap-back rallies following large drawdowns, where the strategy was still positioned relatively defensively (eg. owned one of the bond funds) compared to a pure buy-and-hold equity strategy. It is currently coming off a period of relative underperformance stemming from the rally following the Aug-Oct 2011 correction.
I use this strategy as a counterweight to a value equity position - as recommended by Asness et al in the "Value and Momentum Everywhere" article I linked - and to a multi-asset trend following strategy (like the one discussed by Mebane Faber in the other article I linked above) that also mainly uses the commission free ETFs on Fidelity (IWV - Russell 3000, AGG - Barclays Total Bond, ACWX - Global ex US, and IYR - REITS), as well as DBC - Deutsche Bank Commodity ETF for the commodity portion, and the 200-day moving average as the trend. The trading rule intuition with the trend following strategy is similar to that of the momentum strategy - on the first day of each month, look at whether each ETF adjusted price was above its 200-day moving average as of the close on the previous trading day (eg. end of month). By the end of the first trading day of the month, own the ETF if above or equal to the 200-day and don't own it if below the 200-day moving average.
Each strategy (value, momentum, trend) captures a different source of expected return (see Antti Ilmanen, "Expected Returns" for a detailed treatment of risk factors and sources of expected returns) and is not perfectly correlated with the other, meaning that you can basically allocate between them like you would allocate between asset classes. Right now, I personally allocate 50% to the trend strategy, 25% to value and 25% to momentum. The back-testing history for this allocation is even shorter than for the momentum (because of the short history of commodities ETFs), but since November 2006 it achieved a compound return of 6.3% with a volatility of 13.6%, besting the 1.52% return and 24.2% standard deviation of the S&P 500 during that period.
Deleveraging Accelerating: Risk Rising Of Accelerating SPY Sell-Off [View article]