Archive Compliance Home Office Marketing Misc Regulatory Product Debuts Retirement Returns Sales
The Home Office - Financial Theory - pricing & hedging
Managing Agents & Employees  Financial Theory - pricing & hedging
Newest To Oldest

A Better Index?  (3/16)
Low Yields Foretell Difficult Year (2/15)
Hypothetical FIA Returns Are More Real Than Real Returns (7/14) 
Managing FIA Volatility: 1995 to Today (3/14) 
How Well Do Cap Rates Correlate With Bond Yields & The VIX? (12/13)
Hedging Longevity Risk (3/13) 
Buy This Annuity (and win a chance at a new car) (2/13) 
Annuity Rates Will Move Up Even If Overall Bonds Yields Don’t (1/13) 
Annuity Unbundling – Cost vs Benefit (2/12) 

VA Carrier Seeds in the FIA Garden (10/11) 
Effect Of Low Returns On GLWBs (9/11) 
Black-Scholes Is A Lie (5/11) 
Serial Annuity Exchanges (8/09) 
CD Rates And Market Movement Both Affect Fixed Annuity Sales (2/09) 
GLWB – Pricing Considerations & Reinsurance (5/07) 
Index Methods Are Priced To Perform The Same (4/07) 
Index Annuity Product Feature Bloat (5/06) 

Option World Glossary (7/05) 
Comparing Crediting Methods (7/05) 
Can Index Annuities and Variable Annuities Compliment Each Other? (6/05)


Annuity Rates Will Move Up Even If Overall Bonds Yields Don’t (1/13) 
Annuity Unbundling – Cost vs Benefit (2/12) 

A Better Index?  (3/16)
Black-Scholes Is A Lie (5/11) 
Buy This Annuity (and win a chance at a new car) (2/13) 
Can Index Annuities and Variable Annuities Compliment Each Other? (6/05) 
CD Rates And Market Movement Both Affect Fixed Annuity Sales (2/09) 
Comparing Crediting Methods (7/05) 
Effect Of Low Returns On GLWBs (9/11) 
GLWB – Pricing Considerations & Reinsurance (5/07) 
Hedging Longevity Risk (3/13) 

How Well Do Cap Rates Correlate With Bond Yields & The VIX? (12/13)
Hypothetical FIA Returns Are More Real Than Real Returns (7/14) 
Index Annuity Product Feature Bloat (5/06) 
Index Methods Are Priced To Perform The Same (4/07) 

Low Yields Foretell Difficult Year (2/15)
Managing FIA Volatility: 1995 to Today (3/14)
Option World Glossary (7/05) 
Serial Annuity Exchanges (8/09) 
VA Carrier Seeds in the FIA Garden (10/11)

A Better Index?  (3/16)  
The capital market theories of Markowitz and Sharpe are used by millions in the financial world to determine the relative value of securities. They are simple to understand and compute. They favor a story of passive investing through capitalization weighted indices. And, a growing chorus says they are also...wrong. The major assumptions of these theories are that markets are efficient, that there are no such things as inflation, taxes, transactions costs or changes in interest rates, that investors always maximize expected utility, and that knowing the portfolio's mean and standard deviation is all you need to know. There is no risk management. However, acknowledging that the real world does not operate like the assumptions means managing for risk can produce better results than passively waiting.  

A study by Wojciechowski and Thompson captures the results of thousands of random portfolios they created where they then compared their performance within a thirty three year overall period with a market capitalization approach, such as the S&P 500 Index uses.    

What they found was in twenty two of the years over half of these random portfolios had returns above the capital market line. What this means is a non-passive approach can "beat the market". Sharpe’s theory says the reason why some of these random portfolios have higher returns is because they probably have higher risk than the market as a whole. However, Wojciechowski and Thompson found that there was no correlation between the portfolios with higher returns and risk (defined as volatility). Their conclusion was that a risk based  approach can generate higher returns than a market capitalization one.    

 Applying a more active rules-based approach appears to offer the possibility of the higher returns than simply using the S&P 500

A new study by Ernst, Thompson and Miao says the S&P 500 can be beaten, precisely because it is a market capitalization weighted portfolio. As an example, they compare the performance of buying one share each of the stocks of the S&P 500 versus buying stocks based on their market cap. From 1958 to 2014 the annualized geometric return for the one-stock non-market cap    approach was 10.4%, for the market cap weighted S&P 500 the return was 7.1%. The   authors did another study where they picked the top twenty stocks each year with the highest medians from the S&P 500. The geometric mean return was 10.8%. Indeed, their simple 20 stock rules-based portfolio beat the S&P 70% of the time. Neither return includes dividends.  

Tying this into a rules-based VCI Although many of the volatility controlled indices use the stocks of the S&P 500 Index as their catalogue, many of them also apply rules based strategies and only select a small percentage of the overall universe for their own index and     these stocks are then moved in and out based on the rules in place. The 20 to 100+ stocks    selected, depending  on the index, are reallocated based on performance or dividends or  individual volatility or undiscovered value, based on the rules applied. These indices are not actively managed, like a mutual fund, but simply move based on the formulas in place.  

Regardless of the strategy used, if the results of the studies I mentioned are not flukes, the implication is one or more of these rules-based indices could outperform the passive S&P 500 over the long term. In other words, everything else being equal, an index picking a subset of S&P 500 stocks could outperform the S&P 500 over the long-term. In the volatility-controlled world of FIAs these rules based strategies are restrained by the volatility controlled mechanism, but could still generate more interest.  

P. Ernst, J. Thompson & Y. Miao. February 2016. Portfolio Selection: The Power of Equal Weight. arXiv:1602.00782 [q-fin.PM]  
W. Wojciechowski & J. Thompson. 2006. Market truths: theory versus empirical simulations. Journal of Statistical Computation and Simulation. 76(5):385–395.

Low Yields Foretell Difficult Year (2/15)  
The 10 year Treasury was exactly 1% lower at the end of January than it was a year earlier. The precise opposite of where most predicted it would be and where the economy implies it should be.   The U.S. economy has recovered from the 2008 Crash. GDP set a new high. The S&P 500 has tripled and Nasdaq has almost quadrupled. The unemployment rate is 5.6% and the supply of labor is tightening, putting upward pressure on wages. By historical standards, the bond yield increases that started in the spring of 2013 should have the 10-year U.S. Treasury around 3.5%; instead it ended January at 1.67%.   FIA Caps & Rates Are Unlikely To Head Higher   This continues to be a weak recovery and confidence about the future is thin. Rates are low because businesses aren't borrowing to expand. Rates are also low because the recovery in much of the rest of the world has stalled and investors are putting money into government bonds for safety, not yield.    

The U.S. dollar is strong. A year ago it took $1.35 to buy 1 Euro, today it takes $1.13. That's great news if you're a tourist in Paris, but if you're a U.S. company selling in Europe your goods got a lot more expensive.   Concerns over future growth have ramped up stock market volatility, as demonstrated by the big market swings in recent weeks. Looking at a 20-day moving average, the S&P 500 Volatility Index (VIX) has increased from the 12-ish level from last summer to over 19, meaning option prices are higher. In addition, after a six-year run there doesn't seem to be much room in the stock market's plaza de toros.  

The problem in setting index annuity participation rates is not only are the current yields low and option prices higher, but the future looks worse. If a carrier thinks bond yield will go up later in the year they are often more generous on current-yield based pricing – giving up a little profit today might maintain the pace of sales and you'll make back the profit with the future sales. However, if one thinks rates will stay low or head even lower, it makes for a frugal situation.    

What it would take to get bond rates heading higher this year would be a continuation of falling unemployment as well as rising GDP aided by an improved outlook for the global economy. Unfortunately, this is an unlikely hat trick. Current FIA pricing probably won't get better this year. This scenario also means a greater use of managed volatility indices. While they aren't a panacea for low bond yields they do offer a better return story than more traditional methods.  

In spite of a pricing environment indicating lower index participation FIAs remain attractive, because the alternatives look worse. Rates on  intermediate term multi-year annuities (MYGAs) from higher rated companies are paying in the low to mid 2% range, the best 1-year certificate of deposit rates are barely over 1%, bond returns are low and if rates do rise, can turn negative, and gambling that the stock market's run will extend for seven years is looking dicey. Even with lower caps FIAs still offer the potential for more interest, protection from stock market loss, and a low opportunity cost when compared to MYGAs and CDs or doing nothing.

Hypothetical FIA Returns Are More Real Than Real Returns (and so is a Bayesian model)   (7/14)
In the financial world statistics are used to try understand the past and predict the future. People look at past numbers and attempt to see patterns. At its simplest, these numbers are expressed as an average such as "the long-term annual stock market return is 10%" – the Ibbotson return numbers often quoted are examples of this. In the last twenty years or so this has been replaced or at least augmented by looking at past returns using the Monte Carlo method that came out of the world of physics. The Monte Carlo method does a far more nuanced job of showing how the past looks than merely reporting the average by looking at the distribution of returns. What makes using a descriptive frequentist inferential statistical* or classical approach easy is the basis for the historic returns does not change.  

* All this term means is observing (drawing inferences) how often (frequently) certain returns occur.

What this means is regardless of your data source for the S&P 500 or XYZ stock you will see the same return for the same past periods for each. However, the same cannot be said for past fixed index annuity (FIA) returns.  

                                                               phynances ain't physics (and fisics ain't finances)

The S&P 500 is the underlying index represented in the majority of past and current FIA purchases. However, for any given time period there were wide differences between the returns of different annuities using this index. The main reason was different crediting methods. At one time there were over 40 ways used to tweak the index values to produce a return and the differences produced by these tweaks could be huge – there were years when the difference between the highest and lowest S&P 500 index-linked return was over 30% for the same timeframe. The remaining variation in returns was largely explained by the amount of participation in the crediting method (two annuities could use daily averaging with a spread, but one used a 2% spread and the other had a 3% spread). In addition, there would even be differences in returns for the same period   between owners of the same annuity product.  

Same Annuity – Different Returns
Whether you bought, say, the Vanguard 500 Index fund in 2006 or 2008 or 2010 your return for 2012 was the same for all. However, if you bought the exact same annuity index using the S&P 500 in 2006 or 2008 or 2010 you very likely earned three different returns for 2012. The reason why is index participation is strongly influenced by the bond yield environment when you purchased the annuity.   When you buy a fixed annuity the carrier typically buys bonds to provide the interest to cover the minimum return guarantee and provide additional interest to buy the index-link (the hedge) on the underlying index. Typically the maturity of the bonds roughly mirrors the anticipated life of the annuity – if you think the annuity will be around for 10 years you buy 10 year maturity bonds. What this means is the money available to buy index-links in future years is strongly influenced by the level of bond yields when the annuity is purchased. Thus, if yields are high when you buy the annuity there will be more money from interest available in future years than if yields are low at time of purchase. This means in future years the annuity bought when yields were high will have better caps, lower spreads and higher index participation. These differences narrow as years past and the underlying bonds are replaced, but the bond yield environment at time of purchase influences annuity renewal rates, caps and spreads for years and years.  

An annuity purchased in 2006 won’t have the same future returns as the identical annuity purchased in 2008 or 2010   Even so, if you’re only looking at the past classical statistics works for both investments and FIAs. In other words, if your goal is to find out the average return or the return patterns of either an investment or an FIA for previous time periods the same statistical approach works for both. But what if you're trying to predict the future?   If one believes that the warning on every piece of investment literature is incorrect and that past performance really does predict the future, then both averaging and Monte Carlo models would provide indications of future returns on investments. After all, reversion to the mean would provide evidence that if current returns are lower than the historic average then the return is bound to move up, or vice versa. Or our 95% confidence level derived from some Monte Carlo model on the investment might means the odds of a bad return are very low and this would encourage us to invest more. However, even if past performance did predict future investment returns, past FIA returns would not predict future index annuity ones.    

When looking at historic FIA returns the bond yields in effect at time of purchase play a bigger role than actual index performance. Even if the future daily index movements were identical to the past the FIA returns would be different – unless the pricing patterns also repeated and annuity carrier management set the new rates, caps and spreads exactly as they had the last time around. The bottom line is past historic index  annuity returns bear no relationship to future return predictability and the classical statistical approach does not and cannot work with FIAs because there are too many other factors in play besides the movements of the index. However, using hypothetical models and not actual performance may create a more accurate picture.  

A Bayesian Approach
For the period from September 2003 to September 2008 the average annualized FIA return on the 28 annuities reviewed by Advantage Compendium was 5.6% and a few annuities posted annualized returns for this 5 year period of over 7%. However, even if index movements over the next five years were identical to this period it'd be almost impossible to earn 5.6% using those past crediting methods because the current pricing environment is worse – bond yields are lower so index participation is lower. In addition to the current pricing environment, the Federal Reserve Board has made it very clear that their goal for the foreseeable future is to keep interest rates low. In the light of this, using past returns to predict the near future will produce the wrong      answers. Or, to put this in Monte Carlo model terms – I have a 95% confidence level that the returns produced by my classical statistical model will be 100% wrong. However, Bayesian statistics provide a solution.  

Bayesian statistics combines new information with existing information to try to determine   future outcomes; it does not simply use past data or assume that the probabilities of past data will be true for the future, without first looking to see whether conditions have changed. It then determines the odds that these outcomes will happen. A race track is a good place to see how Bayesian is different from classical statistics.    

A Bayesian Horse Race
The horse Bayes Bay has won nine out of its last ten races. A classical statistician would look at the past and say that the probability of winning today's race would be 90% or that the odds are 9 to 1 that Bayes Bay will win and bet accordingly. However, a Bayesian tout would look to see if things were different for this race when compared to the previous wins and the loss. Perhaps the only loss Bayes Bay suffered was on a muddy track and it rained all through this morning. In addition, two of the horses in this next race have winning records on muddy tracks. In light of this information the Bayesian tout will conclude that the true odds that Bayes Bay wins are not 9 to 1. In fact, Bayes Bay may not even show in this race, and so the tout bets against Bayes Bay.   A Bayesian approach uses subjective probabilities – what are the results based on what we think is going to happen. With this method actual    return data is treated as additional information and not the sole source of information. A Bayesian approach to FIAs also relies on what is happening and what we think will happen instead of relying on the past to predict the future. Since hypothetical return FIA models typically reflect current pricing, the results generated should prove more accurate reflections of the future   results than using prior actual returns.  

Hypothetical returns are more truthful than actual returns   Say that current pricing permits a 4% cap. We know that the Federal Reserve hopes to keep   interest rates down for at least the next couple of years and has said their long-term goals are a low interest rate environment. Even if each of the next five years of the underlying index has double digit gains it is highly unlikely our annuity would credit even 5.6% annualized (that previous 5-year average return I mentioned) because that would require at least a 6% cap in years two through five and the future interest rate/bond yield forecast won't provide enough interest to be available to buy the 6% index-link.    

Hypothetical FIA returns based on a current cap of 4% would provide a much more realistic future return picture than a picture painted from past return colors. Simply put, creating a hypothetic illustration based on today's rates, caps or spreads provides a more realistic view of possible returns than using actual historic returns for the same annuity.  

Creating Bayesian FIA Models
Although index annuity hypothetical models generate results that are more truthful than real returns in predictive power they are typically static; the participation in the index is assumed to remain the same over time. One can attempt to create a model of what future index participation would look like based on assumptions of future bond yields and index-link (hedging) costs – the further assumption being that the carrier's future setting of caps, rate, and spreads will reflect only these external factors. The other subjective aspect would be assigning probabilities to the occurrence of past segments of actual historic index movements. In other words, we would take a look at what is happening in today's stock market, speculate on what we think might happen tomorrow, and then assign odds that the next few years will look more like a particular past time period than another.  

A Bayesian FIA hypothetical return model would calculate index participation (caps, rate, spreads) on historical movements of the index. This participation would be based on where we think   interest rates will be, what hedging costs will be, and what part of the index history is more likely to occur. This isn't as far out or difficult as it sounds. As examples, the odds of investment grade bond yields staying under 7% for the next five years have a far higher probability than yields shooting to 10%; the odds of a bear stock market occurring in the next five years are greater than the current bull market extending until 2019. The math formulas to create these Bayesian hypothetical models are available and the probabilities can be estimated.  

Actual FIA returns are essentially worthless in predicting future returns    
Hypothetical FIA illustrations do a better job because the index participation reflects current pricing
A Bayesian approach to FIA hypothetical  illustrations would do an even better job  

A Less Inaccurate Forecast
The warning is true – past performance does not predict future results, except in very broad arcs. Even if it did, classical statistics would still be useless as predictors for fixed index annuities if actual index annuity historic returns are used. The reason why is FIA returns are driven by participation in the index which is dependent on bonds yields and hedging costs. Hypothetical models offer a more realistic picture of possible FIA returns – at least over the next five to ten years – because today's caps, rate, and spreads are a known and provide a base for what future index participation will be.  

The predictive power of hypothetical illustrations could be increased (or, better stated, major inaccuracies could be decreased) by modeling FIA hypothetical returns using a Bayesian approach that adds probabilities to future bond yields,    future hedging costs and even which parts of the index history should be used in the predictive model. These Bayesians FIA illustrations should provide a more accurate future picture than static hypothetical illustration or attempting to prophesize based on looking at actual returns.   Even though S&P 500 and Vanguard are mentioned no fixed index annuity is sponsored, endorsed sold or promoted by either. This information is for educational use and is not intended as financial or investment advice or to persuade or induce anyone to do anything.

Managing FIA Volatility: 1995 to Today (3/14)
A new direction in fixed index annuity interest crediting involves the use of methodologies that attempt to actively manage or control volatility. The reason behind this is if you lower the volatility you may lower the cost of generating the index annuity interest and make it possible to earn index-linked interest during periods of low interest rates. The "kick-start" behind the search for new ways of crediting interest was the severe drop in bond yields over the last few years that shrunk the amount of money available to produce index-linked interest. What these new methods share is creating the potential for an annuityowner to do better when bond yields are down. These index innovations make for exciting times.

It's also important to know that annuity carriers have been attempting to manage volatility since the inception of the FIA. Why they have done this and how it has affected annuitybuyers provides a foundation that helps to understand the new innovations. To begin, we need to look at what volatility is.

Volatility means the degree to which returns sway from the average or Mean return. Here are two investments with different degrees of volatility:

Investment A has a Mean return of 5%. The worst expected return is 4% and the best expected return is 6%.
Investment B has a Mean return of 5%. The worst expected return is -15% and the best expected return is 25%.

Investment B has far greater volatility than Investment A – B's return in any one year could be 20% higher or lower than the average return while A's range of returns is 1% in either direction. Which is the better choice? Over the long term they are a match – both average 5%; over the short-term it depends to an extent on your expectations and risk aversion. If your goal is to never to earn less than 4% then "A" is best, but if your goal is try for double digit returns then "B" is the only choice – keeping in mind "B" could also incur double digit losses.

The investment world tries to lower or manage high volatility by combining instruments with different levels of individual volatility. However, the ultimate goal in the investor's world is usually not to lower volatility – no investor would prefer a best possible return of 6% rather than 25%. The true goal is to reduce the possibility of investor losses and this is attempted by lowering overall volatility. The traditional method of reducing volatility is by adding bonds to the mix. This 19th century rule has received a 21st century twist with the use of computer models that shift back and forth between different equities and bonds based on the expected volatility of the asset (which I'll talk about at the end).

FIAs, Volatility & Hedging
A fixed index annuity removes the risk that negative volatility creates a loss, but it does not remove volatility. High volatility increases the cost of hedging the index-linked interest because it increases the risk to the person that he or she may have to pay out much more interest than they planned. Here's what this means:

The index-linked interest received is based on the movement of the underlying index. We could buy the index itself, but that exposes us to an unknown amount of losses if the index goes down. Instead, we want someone else to take the risk of a loss so that we only get the gains. We can do this by buying a call option that gives us the right to buy the index upside. Since we're buying an option – and not the index itself – our loss is limited to the cost of the option, a predetermined and known cost. The risk of unknown loss is all on the side of the option seller. This is why most carriers buy call options to provide the index-linked interest potential.

Why would someone sell us an option where we get the upside? The seller will charge us for the option. If the index goes down during the period the index is good for – and all options have a time limit in which they can be exercised – we won't use the option and the seller gets to keep our money. Or, the index may only go up a little and so the seller nets more money from the buyer after taking into account the money paid in for the option and the money paid out for the gain, than he or she would make from simply holding onto the index.

Say you could buy a call option that would let you buy the index at the level of 100 at anytime over the next twelve months. If the index finished the year at 110 you would use the option to buy the index at 100 and make a 10% return (ignoring the cost of the option). What would that option have cost us when we bought it a year earlier?

It depends to a large extent on the past and expected volatility of the index. If the index typically only moves around 3% or 4% a year, and that pattern is not expected to change, the cost for the option would be very low. However, if the index might be down or up 30% in any given year – and it is expected in the next twelve months the index will be up – the option cost will be much higher.

If we owned 100% of that option we would receive the 10% return. However, there may not be enough money available from the index annuity to buy 100% of the upside of the index; one way to reduce the option cost is to restrict the return. Instead of getting 100% of the upside we might only have the money to buy 50% of the upside, or we might agree to get 100% of the upside, but only to a ceiling or cap of, say, 5%. If the index goes up 10% we would earn 5% with either method. Who gets the remaining 5%? The option seller that retained the other part of the option (not the insurance company). However, limiting the gain (interest received) by cutting the participation rate or capping the total interest are not the only ways to deal with the higher option costs that result from volatility. We can also incorporate volatility into the pricing of the option.

Effective & Nominal Rates

Averaging a series of numbers moves the sum to the middle of the series. As an example, if you average the daily closing values of the S&P 500 over the last 50 years you find the average annual return is 50% of the return without averaging (if you use an annual reset approach wherein negative years are treated as zeros the mean average return is 54% of the unaveraged index). In other words, if 100% of the actual long-term unaveraged index returned 10%, 100% participation in the daily averaged index would return 5% on average The nominal or stated daily averaged participation rate is 100%, but the effective actual real-world participation rate is 50%.

The effective participation rate by using averaging is half the unaveraged rate, therefore the price of a call option where the gain is determined by using averaging would cost roughly half that of a call option for the unaveraged index, since the option seller knows, over time, that the averaged gain will be roughly half of the unaveraged gain. Does that mean if the index goes up 10% next year that daily averaging 100% of the index values will produce a 5% gain? Probably not, because the actual daily movements can be quite volatile, thus the effective participation rate can differ greatly from year to year.

In the 12 months ending 31 August 2012 the S&P 500 gained 15.4% and a 100% nominal rate using daily averaging of the index produced a gain of 7.4% for an effective participation rate of 48%. However, a 31 August 2011 year closed with the index up 16.2%, but daily averaging of the S&P 500 produced a gain of 19.9% for an effective participation rate of 123%.

The previous example is one version of averaging, and this type of averaging has been computed over daily, monthly and quarterly index values. Another way to average and limit volatility is to cap the size of the interim gains.

Ten years ago the next great FIA method introduced to tweak volatility was monthly-cap-gains-not-losses, shortened to monthly cap or monthly sum. With this method each month's index gain or loss is calculated during the period and added together, with the proviso that monthly gains are capped and losses are not. This method is a counterweight to high upside volatility because one bad month can offset many good months. As an example, a monthly cap of 2.5% produced an average annual (reset) return of 3.8% over the last 50 years, making it competitive with, say, annual point-to-point methods with 5.8% caps (a 5.8% annual cap also generates a 3.8% average annual return over the last 50 years). So, if an option seller would accept a deal giving 100% participation in the unaveraged index up to a gain of 5.8%, they should just as readily accept a 2.5% monthly cap deal, ceteris paribus.

The option seller might even allow a higher monthly cap than the long-term returns would justify if the seller thinks volatility will be very high over the next period and thus increase the odds of having at least one big loss month.

We said that from August 2011 to August 2012 the S&P 500 gained 15.4% and daily averaging the values produced a gain of 7.4%. However, a 2.5% monthly cap only produced a 12-month gain of 2.0%, due to the index dropping 7.2% in September 2011 (in fact August 2012 was the first period in a year that produced any gain using a 2.5% monthly cap formula). By contrast, for the 12 month period ending August 2013 the same 2.5% generated a 9.3% return because although there was volatility during this period, there were no really bad months.

Unaveraged methods try to lower option costs by lowering the effective participation in the index and crediting less than 100% of index gains, or capping the gains that will be credited. Averaging may also put a cap on the gain. Both averaging and monthly cap methods use negative volatility as way to limit gains and thus lower costs. Another way to deal with volatility is to avoid it altogether or lessen exposure to it through asset allocation.

Asset Allocation
One can try to avoid volatility. Early on in FIA evolution there were products that required a portion of the premium to be placed in the fixed interest account; this idea was revived several years ago. These balancing methods can produce higher overall returns when index returns are very high and lower returns in middling performing index markets.

This volatility avoidance has taken on a new twist within the last year in the FIA space wherein a managed index is used that shifts between an asset class with higher volatility to one with lower volatility. This has involved moving between an equity index and a bond index or between different asset classes. In a non-index annuity setting this approach reduces the probability of big losses. Inside an FIA it reduces the probability of big gains, but that does not mean a low or managed volatility index inside an annuity cannot be competitive.

As a hypothetical example, an S&P 500/bond managed index that shifts to lower volatility produces a 10% return in 2008 versus a 23% return for the S&P 500 index alone, but the "low volatility" approach also produces a 4.15% return in 2011 when the S&P 500 finished flat. In you look back over the last 20 years a managed equity/bond index such as this – even with a 2.9% spread (the actual computed return minus 2.9%) – produces the same long-term return as using the S&P 500 index with an annual point-to-point (APP) method and a 7.5% cap. The key point is whether managed volatility methods are competitive with other FIA crediting methods. Well, the managed index with the 2.9% spread is available today, but I cannot find an index annuity today using an APP method with a cap that is close to 7.5%.

To Sum Up
Although low/managed/dynamic volatility index strategies are being introduced it is important to note that attempting to control volatility is not a new concept. Managing for volatility has been used since the fourth FIA hit the market in 1995. The ongoing concern is that managed volatility methods require more explanation to ensure the annuitybuyer understands what they may earn, but these methods can and have produced, at times, superior returns to other methods.

Creating Realistic FIA Return Expectations

The reality of every fixed index annuity crediting method is each one may produce a more competitive return when compared with fixed rate annuities. Another reality is that under certain circumstances averaging, monthly cap and managed volatility methods can produce significantly higher returns than methods that attempt to restrict high volatility with lower participation rate or caps. However, these controlled volatility methods require more explanation to ensure the annuitybuyer has realistic expectations of how their FIA will perform.

Annuitybuyers quickly understand what they will earn if they are told "the index interest will be based on 40% of any index gain for the year" or "the index interest will be based on 100% of any index gain for the year up to a cap of 4%". However, whenever the nominal participation rate is different than the actual participation rate more explanation is needed.

When averaging was introduced the tag line was "You never have to worry about your year ending on the lowest index closing value of the year." But the annuitybuyer also needs to realize that since averaging drives all numbers to the middle that their index annuity return will never be what the financial web site shows the index did for the period.

The 2.5% monthly cap can get mentally transmogrified into "I can make up to 30%." Although numerically true, the annuitybuyer needs to understand that the long-term hypothetical past return of a 2.5% monthly cap generates average interest of 3.8% a year.

Dynamic or managed volatility indices may be uncapped, offer 100% participation, and have only a small spread deducted from calculated performance. The problem here is the annuityowner has no benchmark to hang their return hat on and may believe uncapped automatically means high interest. Because of this it is very important to use hypothetical examples of past index movements to give some indication of the range of possible interest credited with managed volatility FIAs.

The catalyst for averaging was to get away from caps and offering nominal participation rates that were less than 100%. The catalyst for creating the monthly cap method was high volatility that had significantly increased the option cost used for existing methods and monthly caps offered a way to deal with this volatility. The most recent crediting method additions were created to cope with a low bond yield world. The industry has continued to creatively meet challenges to lower annuity rates; the agent needs to ensure the buyer’s return expectations resulting from this creativity is realistic.

How Well Do Cap Rates Correlate With Bond Yields & The VIX? (12/13)
This chart takes the actual first year average cap rate for annual point-to-point index annuities with an 8 to 10 year surrender period from 1997 to 2012 and compares changes in those caps to changes in the Advantage Insurer Bond Yield Index (a proxy for the yields insurers earn on new bond purchases) and changes in the S&P 500 implied Option Volatility Index as a reference for option prices. There is 82% correlation between changes in caps and changes in new bond yields and 84% correlation between changes in caps and changes in the VIX.

Strong correlations are not surprising. The amount of index annuity participation you can buy depends largely on the amount of money coming in from the bonds and the cost of the options you buy to get the index-linked return. The chart shows when bond yields go down caps go down and when yields go up caps go up. You can pick out a similar story on the VIX (when volatility goes down, option prices typically go down and so you can buy more options to support a higher cap). I did some modeling to see if there was any predictive relationship – if rates or VIX moved to X did caps became Y – but it was very weak. Thus the conclusion is movement in bond yields or the VIX will cause caps to move, but by how much cannot be predicted (at least with this methodology).

Hedging Longevity Risk (3/13)
An annuity that provides lifetime income – whether immediate or deferred with a lifetime withdrawal benefit – creates a risk for the insurer that the consumer may not die fast enough, thus requiring the insurer to shell out money from their own pocket. How do you handle this risk? First, you insure a lot of people – that way the savings from the people that die early help pay for the added cost of the people that live too long.

But what if too many people live too long? A problem for annuity carriers is life expectancy has been steadily increasing for the last half century. What if new drugs are found that increase life expectancy by, say, adding on another decade? It would be difficult to even calculate how high the rider fee would need to be to cover this unknown risk or how low to set the payouts. One way to deal with mortality unknowns is to hedge the risk in case the annuityowners live too long.

The ideal hedge would be something that gave the insurer more income if life expectancy increased. One way to do this is through the natural hedge of using life insurance.
If a consumer buys a life insurance policy and an annuity with a lifetime benefit, the insurer is collecting premiums to pay off if the consumer lives too long and dies too early; only one of these can happen. What this means is the risk for the event that didn’t happen can help pay for the risk that did. So, if mortality improves and the consumer dies later than expected the insurer has more years to profit from the premiums on the life insurance and this extra money could be used to help pay for the annuity; if mortality gets worse and the consumer dies sooner than expected the insurer will have to pay a death benefit, but will save the future expense of additional annuity payments. An insurer that provides both life insurance and annuity lifetime benefits effectively creates a natural hedge with the premiums on one side helping to offset the costs on the other. It has been argued that insurers that offer both life insurance and annuities with lifetime income benefits have a competitive advantage because they can charge lower costs.

However, natural doesn’t mean easy. If the longevity and mortality timelines don’t match up, or if your business model is too concentrated in either life or annuities, then you may be unable to create a workable natural hedge. If you can’t build a hedge you can try to buy one.

Carriers offering both annuities with lifetime income benefits and life insurance have the capability to create a natural hedge that may give them a competitive advantage, but longevity bonds or swaps are needed to more fully hedge the risk of annuityowners living too long

In December 2003 Swiss Re introduced mortality bonds that paid a handsome yield to investors if there wasn’t a catastrophic increase in deaths during the life of the bond, but the investors could lose some or all their money if future mortality (life insurance death claims) were far above the historic norm. The bonds were snapped up by investors. Someone at BNP Paribus looked at this concept and wondered, “why couldn’t we create an un-mortality bonds?” And so the first longevity bond was offered for sale in 2004. It didn’t sell well (for a variety of reason), but it got a lot of people thinking.

No one has developed the definitive longevity bond and the market will ultimately develop a variety of solutions. One concept would be for the annuity carrier to issue bonds that pay a certain coupon rate, but if the longevity of the covered group of annuity living benefit buyers increases more than a certain amount, the bond could be structured to reduce the coupon or the maturity value.

One might approach this from the other end by producing bonds the annuity carrier would buy because these bonds would increase in value, either in coupon rate or in maturity value, if longevity increased. Who benefits financially from increased longevity and, thus, might be interested in hedging longevity? Life insurance carriers are aided when payment on the death benefit is delayed and they could issue bonds that paid a higher coupon when longevity increased. One report suggested that drug companies could issue these bonds because increasing life expectancy means consumers would buy their Lipitor® longer.

The wonder of an annuity is the consumer is able to transfer their risk of living too long to the insurer. The problem is how does the insurer handle the added risk? Although natural hedging is a partial solution the development of different types of longevity bonds as tradable securities in capital markets could protect insurers and possibly increase lifetime payouts to the consumer. The reality is lifetime income benefits pose certain distinct unknowns. One is not knowing how many annuitybuyers will use the benefit (a topic discussed in other articles) but the biggest unknown is how long will the users live. If the carrier has a strong book of life insurance policies these can work as a natural hedge, for others it may come down to the creation of a viable actively traded market in longevity bonds.

My thanks to Noel Abkemeier for his help with this article.

Blake, D., T. Boardman & A. Cairns (2010).The case for longevity bonds  Center for Retirement Research  Number 10-10

Cox, S. and Y. Lin (2007). Natural hedging of life and annuity mortality risks. North American Actuarial Journal. 11, 1: 15

Luciano, E., L. Regis & E. Vigna (2011). Natural delta gamma hedging of longevity and interest rate risk.  ICER Working Paper No. 21/2011

Wang, J., H. Huang, S. Yang, and J. Tsai (2010). An optimal product mix for hedging longevity risk in life insurance companies: the immunization theory approach. The Journal of Risk and Insurance. 77, 2: 473-497

Buy This Annuity (and win a chance at a new car) (2/13)
Government lotteries were popular in the 19th century, but were turned out due to corruption. It wasn’t until 1964 that New Hampshire re-entered the numbers racket with a state lottery. Today over 40 states have lotteries and they are extraordinarily successful for the states – Americans spend more on lottery tickets than milk. They also have terrible odds and terrible payoffs for the consumer.

Private lotteries have been offered for a number of years in several countries. Perhaps the most successful one ever was started in 2005 by First National Bank of South Africa. Savers received 1 lottery ticket for every 100 rand ($16) deposited. The savers earned a slightly less than market saving rate and 114 prizes were awarded each month (with payoffs up to $160,000). Within three years the bank had opened 1.1 million new accounts and collected $170 million in new deposits. The private savings lottery worked well – too well – which is why the South African Lottery Board successfully sued to shut it down. And competition is the reason that all U.S. states have laws that prohibit private lotteries.

People like the idea of prizes being associated with savings. A survey conducted in 2009 asked consumers ‘Would you be interested in a savings account that awarded chances to win prizes based on the amount of money you save?’ Only 26% of the people said they weren’t interested. In 2009 a few Michigan credit unions offered the ‘Save to Win’ program with the requirement that money be deposited in a savings account. The grand prize was $100,000. In eleven months 11,600 new accounts were opened. A similar program offered by in Indiana was also successful.

The reason the Indiana and Michigan credit unions were able to offer prizes is that technically the Indiana one was a “sweepstakes” and the Michigan one was a “raffle”. Most states permit a sweepstakes where a purchase is not required (but they may not limit the difficulty required for a person to enter that hasn’t made a purchase). There are 39 states that allow raffles (source: that may legally require a purchase (usually a non-profit is involved). The reasons for the Indiana and Michigan experiments was to see whether in a low interest rate environment the addition of a “lottery prize” would increase deposits. It did. The other reason was to see if an interest-paying savings vehicle that also offered a prize would get people to save that normally did not. This too was a success with 56% of the new accounts opened by people that had never before had a savings account.

The Lotto Annuity
This article is based on a study1 and was first intended to be strictly tongue-in-cheek. Something like “Buy the new Lotto Fixed Annuity, earn 1.5% and a chance at a new Lexus” ha-ha. However, perhaps this should be given more serious consideration. Over $60 billion a year is spent on state lottery tickets; low-income households spend a larger percentage of their wealth on lottery tickets than other US households. An annuity that offered prizes would induce consumers that have never saved to begin saving. A “prize” annuity may mean more consumers looking at a supplemental income at retirement instead of a trash can full of losing state lottery tickets. And, less altruistically, a prize annuity may attract somewhat wealthier people to buy annuities because of the attraction of the prize.

Current state laws do prohibit annuity carriers from offering lotteries and may ban raffles and sweepstakes as well (although in the few state codes I checked I didn’t see specific bans on insurance carrier raffles, but if prizes are considered “rebates” that would ban them in most states). In any event, any carrier that managed to get a “prize” annuity approved by the insurance department would almost certainly be sued by the lottery commission (perhaps the solution is for the annuity to offer state lottery tickets). However, from a behavioral viewpoint a prize annuity would result in more people saving for retirement resulting in less financial strain on future state services; a prize that benefits everyone.

 1) Kearney et al. 2010. Making savers winners” An overview of prize-linked savings products. NBER Working Paper 16433.

Annuity Rates Will Move Up Even If Overall Bonds Yields Don’t (1/13)
The 10-year U.S. Treasury note closed out 2012 yielding 1.76%. However, based on history it should have been 3.25% because the average Aaa corporate bond yield was 3.62%. Treasury Yields will increase in 2013 even if overall rates do not go up because they are abnormally low. If you look back over the last half of the 20th century you find that the yield on the 10-year U.S. Treasury note was 90.5% of the average yield of Aaa corporate bonds with a standard deviation of 2.9%. What this means is when the Aaa corporate bonds were yielding 5.0% the 10-year Treasury was at 4.5%; if Aaa corporates were at 10% the 10 year treasury was at 9%. Even in the early ‘50s when the yield on Aaa corporate bonds was at 2.8% to 3.0% the yield on the 10-year U.S. Treasury Note was 2.5% to 2.7%.Relative to Aaa corporate bond values Treasury yields tracked very closely through both good times and bad. Until the Asian Credit Crisis in 1998 10-year Treasury yields were never less than 79% of Aaa corporate yields and even that ratio was quickly back over 80% by 1999.

The early ‘00s were a different story. The millennium recession, although mild in the U.S., caused credit problems in foreign markets, and that fear fueled an unjustified fear about U.S. corporate debt. Between the summers of 2002 and 2003 Aaa corporate bond yields averaged 6.1%, which should have meant a 10-year Treasury yield of 5.5% based on history. However, Treasury Notes instead paid 4.0%, or 66% of the Aaa bond yield instead of 90%. Based on fear of foreign contagion, and a possible double dip recession, investors did not so much as build in a risk premium for owning corporate bonds as they instead created a safety rate discount for owning government debt. This caution proved unwarranted and those that had purchased Treasury Notes instead of Aaa corporate bonds experienced relative losses as the correlation between the two steadily increased to 88% by 2007. To put a head on it, during this period while 10 year Treasury yields rose from 3.9% to 5.1% (and T-notes market value declined) Aaa corporate bond yields rose from 5.7% to 5.8% (and market values barely shuddered).


The financial crisis of 2008 shook investor confidence and the credit markets to the core. Once again there was a flight to Treasuries, but the safety discount became even more drastic with 10-year Treasury’s yielding less than 50% of Aaa corporate debt yields by the fall of 2008. This ratio steadily improved in 2009 and reached 73% by the spring of 2010, before slowly falling back again as worries increased about the strength of the recovery. However, the next stumble was strictly due to the summer 2011 folly in Washington and the nonsense with the debt ceiling.

When the politicians almost caused the U.S. to default in August of 2011 the ratio again dipped below 50% and then got worse. By the time of the 2012 election the yield on 10-year Treasuries was half of its long term ratio – 45%. To put this into perspective, based on 60 years of history the 10-year U.S. Treasury Note should have yielded 3.25% in November 2012. Using the Asian financial Crisis as a worse case the T-Note should have been yielding 2.4%. Instead it was at 1.59%. Simply put, based on the actual risk of default on corporate debt the 10-year Treasury yield is far too low and should be much higher. Even if we approach this from the other side and say corporate yields are too high and should come down to the 2.7% range only briefly seen 60 years ago  it would still mean 10-year Treasury notes should be yielding 2.5% today.

Different Standards For Today?
Because today is a different world than that of the last half of the 20th century it may be that the old metric of 90% is obsolete because Aaa corporate debt is much riskier than U.S. Debt (although the evidence doesn’t support this contention). However, even if the new ratio is 66% - which was the bottom a decade ago – this means 10 year Treasury’s should yield 2.4% today, and if the new ratio is a much more likely 80% it means 10 year Treasury note should gradually increase to yield 2.9% - even if overall interest rates do not increase.

Once again it appears the financial markets have severely overreacted to the possibility of default of Aaa corporate debt. However, rather than increasing the yield on the corporate debt the market’s reaction has been to give a safety rate discount for U.S. debt, which is unwarranted. The end result is that even if rates on corporate bond do not improve the yields on Treasury debt will go up over the next couple of years bringing a measure of relief to insurance carrier investment portfolios.

Annuity Unbundling – Cost vs Benefit (2/12)
As bond yields began to collapse in 2010 and continued their fall annuity carriers reacted by cutting commissions, bonuses, and minimum guarantees. However, by autumn 2011 bond yields had reached a point where there was little left that could be cut.There have always been additional costs to an annuity beyond the mortality expense, administrative expenses and minimum guaranteed return, but these extra costs have usually been hidden in the renewal rate credited to the annuityowner in subsequent years. The most visible examples of additional costs are higher sales commissions and premium bonuses that must be repaid from future annuity earnings. Although some of these costs may be offset by lengthening the surrender period, thus allowing the acquisition cost of the annuity purchase to be more efficiently amortized and also permitting the annuity to benefit from the yield curve that rewards longer bond maturities, the reality is the annuityowner is still paying much of the freight. 

Bonuses and commissions are obvious costs; others are more subtle. Most products include a waiver of surrender charges if the annuityowner is confined to a nursing, some extend this waiver to cover a terminal illness, a few even cover unemployment. All of these benefits impose a cost on the annuity, but anecdotal evidence suggests that these benefits are almost never used. A less subtle cost is offering an initial interest rate or interest cap that can’t be supported by carrier earnings and so it is recaptured with lower renewal rates to the annuityowner, but the initial unsupported rate or cap is not disclosed as such.The reality of the lowest bond yield environment in over 50 years is there simply isn’t enough spread to recapture all of these costs from the renewal rates, therefore carriers are being forced to unbundle the benefit, state the cost, and allow the annuitybuyer to decide whether the benefit is worth the cost.

The first unbundling occurred before rates dropped and it was charging for guaranteed lifetime withdrawal benefits. The first two carriers to offer GLWBs charged specific fees for this income benefit and almost every carrier followed their lead in treating GLWBs as an optional benefit. A little over a year later Aviva was the first to charge an annuityowner an explicit fee* for guaranteeing a minimum death benefit, but almost no one followed that time around. It wasn’t until autumn 2011, as bond yields continued to fall, that carriers began to seriously consider charging annuityowners fees for benefits that had heretofore been seen as standard equipment. I feel the reason for this was wariness about breaking the sanctity of the annuity promise, but this promise had already been broken.

Redefining The Annuity Promise
A defining point of the fixed annuity story has always been a promise that you can’t lose what you have whatever amount your account balance shows at the end of one year will at least be there the next, and will probably increase too. However, most GLWBs broke that promise by charging a fee that could cause and in an index-annuity context almost certainty would cause at some point next year’s account balance to be less than the previous year. Although the contract of the first carrier to offer a GLWB, American National, explicitly stated that the charge would never be more than the interest credited, the second entrant, Aviva, made no such guarantee and almost every carrier followed Aviva’s example. With the advent of GLWBs most fixed annuities offering this benefit could no longer promise that the annuity balance would never fall.

Within the last two months a couple of carriers have taken unbundling to the next level. Midland National’s MNL RetireVantage series offers a package that includes a higher bonus, higher penalty-free withdrawals, annuitization bonus and return of premium all for an annual cost of 0.6%. Allianz’s new 365i offers an annuity rider that, at present, raises an index cap by 2% for a 1% rider charge.

These new examples of unbundling have been caused by a very low bond yield world that will continue to challenge carriers for at least the next few years, but I view it as a positive step because it brings greater transparency to the fixed annuity world and permits the consumer to decide whether a benefit is worth the cost. While I believe bond yields will again head up and carrier spreads will increase,  the unbundled approach could become the new annuity promise of greater transparency for consumers.

* Index annuities may talk about an “asset fee” that reduces index-linked interest credited, but it really isn’t a fee because it can never reduce the interest earned to less than zero. An index annuity asset fee is simply another way of saying participation rate.

VA Carrier Seeds in the FIA Garden (10/11)
A decade ago, as the millennium bear market worsened and variable annuity sales slid, most of the major variable annuity carriers had me come in and talk to them about the index annuity market. The purpose of having me in was to help them decide whether they should add index annuities to their product line-up. The decision in all cases was no. The reason in all cases was they felt adding index annuities would cannibalize their VA sales and cause disruption in their distribution channels.

Interest in FIAs resulted from weakening VA sales and fear of GLWB in-the-money risk 

VA sales did recover and rose to set an all-time record in 2007 of $179 billion, aided in large part by strong living benefit riders that helped to shield investors from loss. Since that time sales have slipped. In addition, it was discovered that those living benefit riders could create losses for the VA carriers, instead of profits, because investors were inclined to use them when they presented the greatest risk of loss for the carrier. A Milliman study (discussed in the July 2011 issue) found withdrawals increased when the VA’s income account value was higher than the actual cash value. Essentially, the structure of a VA guaranteed lifetime withdrawal benefit (GLWB) means it is most likely to be utilized at the worse possible time for the carrier – when the annuityowner’s cash value is down he or she is more likely to withdraw and use up all of their own money and thus start using the carrier’s money.

The result was that some VA carriers began to look at the index annuity world as an alternative. After all, FIA sales had steadily risen since 2007 setting new records. In addition, a GLWB on an index annuity coaxes the annuityowner to not take withdrawals because “things can only get better” due to guaranteed roll-up rates or increases in the factors that increase income the longer the withdrawals are delayed (it’s an entirely different mindset and one that works in favor of the index annuity carrier). In addition to this – though it hard to believe today – bond rates had started to move up in the fall and winter of 2010 making fixed products more attractive.

I don’t know how many VA carriers drew up plans for index annuities. I do know four carriers filed fixed index products with the regulators. Phoenix (PHL Variable Insurance Company) was early to sow releasing FIAs in 2010. Hartford Life, Mass Mutual and Pacific Life filed their fixed products within the last 6 months. However, Hartford reportedly pulled their FIA filing in August (Contract: LA-FIA-A-11. SERFF Tr Num: FRCS-127383797). Although Mass Mutual filed in May (FL DOI Initial Filing May 5, 2011) it has not yet come to market. And Pacific Life’s FIA has a requested implementation date of 1 November 2011 (SERFF Tracking Number: PACL-127538215). We know how the index annuities of Phoenix are structured and it might be instructive to see what the other VA carriers were thinking in their product design.

The three unreleased products are pretty straightforward. All offer a fixed account option and all offer a link to the S&P 500. Hartford also filed a Global Allocation (50% S&P 500, 20% S&P 400, 15% Russell 2000, 15% MSCI EAFE Index) while Pacific Life filed to offer the MSCI ACWI All Country World Index. All have an annual point-to-point with cap method. Both Hartford and Pacific Life also offer the Trigger method where a declared rate is paid if the index doesn’t lose ground. Hartford filed to offer monthly averaging less a spread and a monthly cap design. The Mass Mutual design has a 7-year surrender period, with no MVA, with the option of adding 5 and 9 year products if desired. Hartford filed for 5, 7 and 10 year periods and Pacific Life filed for 6, 8 and 10 years; both Hartford and Pacific Life have an MVA. The only GLWB specs I could find were on the Pacific Life annuity and it is competitive. Even though the FIA designs are basic, they can be competitive – there’s nothing wrong with them. Whether any of the three come to market will depend on the interest rate environment.

Effect Of Low Returns On GLWBs (9/11)
The first index annuity carrier to offer a lifetime withdrawal benefit (GLWB) was American National Insurance Company in June 2006 and it appears roughly half of current index annuity sales now include a GLWB. The math of the benefit parts has changed over the years; the most noticeable recent change has been the fee for the GLWB. Up until 2010 the most common percentage charged for a GLWB was 0.40%; however in 2011 the most common rider charges are 0.90% or 0.95%. In addition, there has been a trend towards lowering the roll-up growth rate and/or cutting the payout factors.

When index annuity GLWBs were first introduced the roll-up rates were modest – 4% to 5%, but the race escalated so that by 2008 several were guaranteeing 7% to 8% annual payout increases, for each year withdrawals were delayed, for up to 20 years. At age 65 most of the carriers used 5% or 5.5% single payout factors and a 6% payout at age 70. However, these higher roll-up rates when combined with strong payout factors amplified the longevity risk to the carrier if long-term returns were poor. 

Low returns have a moderate effect if withdrawals begin soon after purchase

Many retirement models try to get the money to last until age 95 – 30 years of payouts for a 65 year old. Realistically this wasn’t a problem for a GLWB where payments began immediately because taking 5% when earning as little as 3% means money lasts 31 years – even at 2% the money lasts 26 years or until age 91.

But high and long payout growth guarantees are strongly affected 

Say the original 5% payout is guaranteed to double in 10 years – a 7.2% compound or 10% simple growth approach. A 55 year old putting in $100 could take out $10 at age 65 ($200 x 5%), but how is the carrier affected if the original premium only earns 3%? At 3% net the annuity cash value (ignoring any surrender charges) is worth $134.39 after 10 years. If the annuity continues to earn 3% with the annuityowner taking our $10/year the money runs out in 18 years or age 83. If the annuity only earns 2% from age 55 on, and $10 withdrawals begin at age 65, the money runs out at age 80.

Rider fees have increased from 0.4% to 0.9%

The problem is made worse as the payout growth guarantee period increases. If the annuitybuyer was 45 and put in $100, using the same 7.2% growth and 5% payout assumptions at age 65 the guaranteed payout would be $20 ($400 payout account value x 5% payout factor). At a 3% return the annuity cash value would be $180.61, and keeping at a 3% return the money runs out in 11 years or age 76.

How to solve this?
The cleanest way is earn a higher return. In the age 55 example if you earn 4% the money lasts until age 88, or until age 100 if 5% is earned. However, the carrier can’t control future returns.

Cut Roll-up Rate – if you drop the guarantee from 7.2% to 6% (or from 10% to 8% simple interest) the payout at age 65 drops to $9. If you only take out $9/yr and earn a steady 3% the money lasts to age 86 instead of age 83.

Cut Payout Factor – if you still guarantee 7.2% (10% simple), but cut the payout factor from 5% to 4.5%, and earn a steady 3% the money also lasts to age 86 instead of age 83.

Cut Both – a 6% growth rate and a 4.5% payout means a 3% return makes the money last until age 89. By cutting back either the payout or growth percentage you can get to where the customer’s annuity doesn’t run out of money until after age 95 as long as 4% is earned. And, when you take into account the number of GLWB buyers that won’t use the benefit or die early, combined with the rider fee that has roughly doubled to nearly 1% on average, my scratchpad math says the carriers is still okay as long as around 3% is earned on the annuity with either cutback.

Shift More Risk To The Annuityowner – When Lincoln Financial Group and Lincoln Benefit Life came out with their original GLWBs they did not get involved in the roll-up race, but instead increased the payout factor based on the number of years withdrawals were delayed. What this meant was if the annuity returns were poor the two Lincolns generated lower payouts than most other GLWBS, but if returns were in the 5% or 6% range they did very well (the new Allianz 360 Annuity compared on the next page is a descendant of this approach). The “but” in the increasing factor approach is if the annuity actually earns 5% or 6% for 30 years you won’t need the insurance carrier’s money.

Annuity purchased at age 55 with Premium of $100 

How many years the money lasts based on these assumptions

Acct Value
@ Age 65
5% Payout
7.2% Growth
5% Payout
6% Growth
4.5% Payout
7.2% Growth
4.5% Payout
6% Growth
2% $121.90 15 yrs 16 yrs 16 yrs 19 yrs
3% $134.39 18 yrs 21 yrs 21 yrs 24 yrs
4% $158.02 26 yrs 31 yrs 31 yrs 30 yrs
5% $162.89 35 yrs 49 yrs 49 yrs Infinity

Historic Odds Say Future Returns Should Be Better  
At current 2% to 4% caps the index annuity returns are not enough to ensure the annuityowner’s money lasts. Looking at consecutive 30 years periods going back to 1951 an index annuity with a 3% cap generated an average 30 year S&P 500 index-linked return of  2.01%. Bumping the cap to 4% only increased the average return to 2.63%. However, today’s caps are produced by today’s interest rates, which are low by historic standards.  

Using the Advantage Composite Bond Index you find today’s bond portfolio yield is 4.6%. When you subtract out carrier costs and agent commissions there simply isn’t much left over to buy the options to get higher index annuity caps. Could we have a 30 year period where bond yields stayed this low? Yes.  

From 1935 until 1965 there were 26 years where composite bond yields were 4.6% or less. However, the average portfolio bond yield from 1919 though now was 6.5%. Looking back 90 years bonds rates were high enough to support 5%, 6% or much higher caps 72% of the time. At even a 5% cap the long-term hypothetical average index annuity return is 3.22%, meaning that the carrier can support current GLWB promises.  

GLWB product changes in 2011 assume that the current low interest climate stays with us for the long, long term. This is prudent since the carrier is insuring the longevity risk and needs to still be in business in 2046. Changes in fees, factors and roll-up rates are meant to ensure the carrier can insure even in worse case scenarios. However, history says that bond yields should increase at some point, meaning that index annuity participation will also go up, meaning annuity returns will also increase. This could result in much higher GLWB payouts than are currently anticipated.

Typical Index Annuity GLWB

2007 2009 2011
Roll-up (Compound Intr) 4% - 5% 7% - 8% 6% - 8%
Rider Fees 0% - 0.40% 0.40% - 0.50% 0.65% - 0.95%
Payout Factors Age 65:  5%-5.5% Age 65:  5%-5.5% Age 65:  4.5%-5%
Age 75:  6%-6.5% Age 75:  6%-6.5% Age 75:  5%-6%

Black-Scholes Is A Lie (5/11)
Black-Scholes-Merton not only is not an options formula but it is “fragile to jumps and tail events” and it is “highly invalid mathematically”. It is built on dynamic hedging which “is dangerous in practice as it subjects you to blowups.” In short, Black-Scholes-Merton theory is not real and Sweden should take back the Nobel Prize. Thus opens an article by Talib and Haug published in the Journal of Economic Behavior & Organization. Talib, the creator of the Black Swan theory that says we cannot predict unpredictable events, castigates the whole notion that the Black-Scholes option formula is the basis for option trading since the 1970s by saying that option traders never use it, but the media and academia jumped on the term “Black-Scholes” because of excellent marketing by its creators and mistakenly gave it the credit for the tremendous growth of derivative trading that has taken place in the last four decades. 

The argument is option traders use rules of thumb developed over many years. These rules work, but because academia didn’t create these rules they pretend they don’t exist, and then create their own theories. When presented with evidence that theories like dynamic hedging don’t work (such as the failure of “portfolio insurance” in the crash of 1987) they ignore the evidence.

The first myth is Black-Scholes made proper option pricing possible. However, De La Vega and De Pinto over 200 years ago developed option hedging techniques. Indeed, a recent study concludes, “traders in the 19th century appear to have priced options the same way that 21st century traders price options.” The authors cite many, many cases where rules for proper option pricing have been in place for many, many years. The second myth is traders use Black-Scholes to “value” options. The reality is traders “price” options, and the difference between “valuation” and “pricing” in the option world is akin to the difference between a real estate appraisal and the actual selling price of the house. The article basically accuses Black-Scholes-Merton and their proponents of either being liars or stupid. I can’t wait to see the responses.

Why Wall Street Gets It Wrong (1/11)
I believe there are two main reasons why Wall Street screws it up from time to time costing investors billions of dollars, and also why they’ll never stop having an ongoing parade of crises. The first reason is they treat finance as physics, meaning they think finance is a science, and as such expect past results to predict the future and unforeseen outcomes to not occur. The other reason Wall Street won’t stop blowing up at regular intervals is they assume all of the decisions made in the financial markets are completely rational at all times. I’d like to talk about this second point and what they missed.

Wall Street thinks finance is a science and that people are always rational, but neither is true. 

Almost three centuries ago the cousins Nicolas and Daniel Bernoulli came up with a theory. What it said was people will choose the outcome that provides the maximum utility for their situation, taking into account their tolerance for risk and that the decision is based on the information they have – which may not be perfect – but that rational people will decide to choose the greatest benefit based on the information they have. This expected utility theory was strengthened by a book written by Frank Knight in 1921 and almost codified by von Neumann and Morgenstern in 1947 who concluding that a person’s decisions were always completely rational as long as some basic rules were followed.

Today’s Wall Street is built on modern portfolio theory, the capital asset pricing model and others but ignores the fatal flaw revealed by behavioral economics which is people don’t always act as the model says they should.

However, folks noticed the theory didn’t always work because there were many situations where people did not choose the best outcome. As an explanation for these contradictions Leonard Savage in 1954 offered a modification saying the reason decisions were not completely rational at times is that people create their own probabilities of an outcome, which may not reflect the true odds and may be subjective. His subjective utility theory concluded that rational people make rational decisions, but they may rationally apply different subjective odds, so two people could come up with different decisions, but both would still be rational, therefore people still made fully rational decisions. Savage tried to offer a rational reason why people are sometimes irrational.

Rational Decision Theory and Behavioral Decision Theory

Adam Smith

Bernoulli Cousins








von Neumann & Morgenstern











Kahneman & Tversky






1750 1920 1940s-1950s 1960s-1970s 1980s-1990s 2000s-2010s

Around this time Harry Markowitz came up with something called modern portfolio theory that said since investments have different return distributions and different levels of risk it was possible to model and compare these on bell curves and find the optimum mix of assets at a prescribed level of risk. The assumption behind this was that people would always make rational decisions to maximize returns at their accepted risk level. This led to the capital asset pricing model created by Treynor, Sharpe, and others that says we can determine the systematic risk, or beta, and then look at the additional potential return of an investment, or alpha, above a risk-free security, and thereby compare totally different assets. And this was followed by a flood of financial formulas including the Sharpe Ratio (risk premium), Black-Scholes model (option pricing) and many others.

Wall Street believed by creating better financial math models they could create above-average returns forever by applying scientific methods to reduce financial risk, and so for the next quarter century Wall Street spent their time and money hiring science grads and creating more formulas...but it didn’t always work (Asian Financial Crisis, Internet Bubble, Real Estate Bubble, Subprime Mortgage Crisis) and the reason was because people weren’t always rational and didn’t always try to maximize utility.

After the last financial crash almost pushed the world into a depression – although the math models said that could never happen – a few folks at Wall Street are finally admitting that maybe there are other factors influencing decisions, therefore in the last couple of years there have an increasing number of articles saying that behavior may be a factor in decisions, and topics such as loss aversion, mental accounting, herding behavior, confirmation bias, and others have been appearing in the financial press. Wall Street acts as if knowledge of a behavioral factor in decision-making is a new discovery, but economists have been writing about behavior since the beginning of financial theory.

During that 18th century time when the Bernoullis were creating utility theory, economist Adam Smith was writing that loss aversion affected rational decisions and thus needed to be taken into account. Although Frank Knight’s 1921 book said that maximizing expected utility would be the decision process in a rational world, he also admitted that a psychological variable is a part of the decision process but that he didn’t know how to study it so he decided to ignore it. Even so, Knight said our “instincts” may account for the difference between rational decision outcomes and actual outcomes.

Finally, during the last half century as Wall Street modelers kept insisting on a rational world, their theories failed to explain results such as the Allais Paradox (1953), the Ellsburg Paradox (1961) and the kink in the utility curve demonstrated by Kahneman and Tversky’s Prospect Theory (1979). The knowledge that people’s decisions contain both rational and emotional elements has been known for 250 years, and the research amassed over the last 30 years has convinced even the biggest doubters that people’s decisions are seldom completely rational. Wall Street has been forced to talk about the behavioral factor because of the repeated failures of their computerized math formulas. 

The problem is if Wall Street accepts that all of their financial models are flawed because human behavior is an unpredictable variable, meaning the rational decision will not always be made, then Wall Street would be forced be admit that they can do little more than guess about the future, but one doesn’t pay millions of dollars in annual advisory fees for “guesses”. And this is why the role of annuities will increase in the years to come as consumers move to a retirement vehicle based on guarantees and realize they do not want to base their retirement income on Wall Street’s guesses.

Serial Annuity Exchanges (8/09)
Commissions and premium bonuses are costs that reduce the yield paid to the annuityowner. These costs are amortized and deducted from future yields until they are recaptured. While yields lost because of commissions are gone for good, the bonus is more of a tradeoff with the annuityowner receiving, essentially, prepaid interest that is recaptured in future years. However, my research indicates that premium bonuses are almost never shown to the agent or consumer as this zero-sum game, but are instead presented as a “free gift” that may be used to offset surrender charges that are incurred when one annuity is sold and another purchased in a tax-free exchange. The story told is that because of these bonuses the annuityowner benefits from switching policies.

What I’ve attempted to do is look at the effects of these policy exchanges on an annuityowner by seeing what the annuity value becomes over time when the annuityowner stays with the original annuity and when policies are exchanged for new annuities with bonuses. To do this I’ve made several assumptions:


  • The commission and bonus are recaptured over the surrender period. To wit, if an annuity has a 5 year surrender period and paid a 5% commission the gross yield available to the annuityowner would be reduced by 1% for each of 5 years reflecting the recapture of the commission. After the recapture period ends the annuity is credited with the full gross yield.
  • The surrender charges initially reflect the full cost of the up-front premium bonus and commission and then decline proportionally over the life of the surrender charges. In other words, if an annuity with a 15 year surrender period paid a 10% premium bonus and an 8% agent commission the first year surrender charge would be 18%, the second year would be 16.8% and the fifteenth year would be 1.2%. The surrender charge is calculated on the accumulated value.
  • Market Value Adjustments (MVA) are not used.
  • The gross yield available to the annuity after carrier profits and other expenses is 5%. The recaptured cost of commissions and premium bonuses are amortized and deducted.

The assumptions reflect the issues in annuity yields and exchanges, and are not intended to be an exact mathematical model; however, the real world is not far off. The initial surrender charges of many products do approximate the amount of the up-front vested premium bonus plus the first year agent commission and they typically decline over time, but real world surrender charges may also be calculated on 90% of accumulated value or on the premium only. Another difference is the real world charges tend to remain level longer while mine decrease each year.  

I played with different surrender charge assumptions doing multiple comparisons, but they did not affect the ultimate conclusion so I tried to be simple and consistent in my treatment. Regarding MVAs, altho the idea behind an MVA is that the annuityowner receives a higher yield because they are sharing some of the risk of interest volatility I have never seen any proof of this on  renewal rates, so I ignored MVAs. While it is true that an MVA has, at times, reduced or eliminated surrender charges because interest rates have fallen since the policy was issued, this could also mean the old annuity is backed by existing bonds with higher yields and the new annuity would be backed by lower yielding bonds, resulting in less interest paid on the new annuity. In addition, there are offsetting times when an MVA results in higher charges than the surrender schedule listed.

I am also comfortable assuming that all insurance companies have the same amount available on a gross product yield basis. Having looked at annuity carrier investment yields over many years I find that they clump in a narrow band, and even ones with higher than average yields in one year tend to have lower than average yields in another, so it all seems to even out. If a carrier could show that they had discovered the annuity philosopher’s stone that would enable them to offer consistently above average yields, then this would be a legitimate reason to transfer to this insurer. Alas, no carrier is making this claim.

Base Annuity  
The annuityowner put $100,000 into a deferred annuity. The commission paid was 5% and there was no bonus. The 5-year surrender schedule is 5%, 4%, 3%, 2%, 1% and the annuityowner finished the third year of the policy so the current surrender charge is 2%. The upfront cost – the 5% commission – is being recaptured over the surrender period. Or, 5% divided by 5 years means 1% is currently being deducted from the 5% gross product yield. The policy has and will credit 4% for five years and then credit 5% thereafter. The following chart reflects the accumulated and surrender annuity values if the policy is held 20 years.  

The annuity is out of both the surrender period and the deduction for the upfront charges (in this instance only the commission) in two more years and will earn 5% thereafter.  

Second Annuity
At the end of year three the annuityowner exchanges the Base Annuity for a new annuity with a 5% premium bonus and a 5% commission. The 10-year surrender schedule is 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%. The upfront cost of 10% is recaptured over the 10-year period meaning the annuity credits 4% for the first 10 years (5% - 1%) and 5% thereafter. The value transferred from the Base Annuity is $110,236 reflecting the $2,250 surrender charge. The $110,236 is credited with a 5% bonus resulting in a new accumulated value $115,748. The following chart reflects the accumulated and surrender annuity values for the original 20 year period. 

The bonus provides a higher accumulated value (Ex2 AV) for the next 5 years, but with a lower surrender value. By the end of the second annuity’s surrender period (Ex2 SV) the second annuity’s value is $8,420 worse than the Base Annuity, or $171,335 versus $179,755. At the end of the 20 year period the base annuity value is $252,933 and the second annuity value is $241,086.  

Third Annuity  
By the end of year 6 the second annuity surrender value is $121,087 and the surrender charge has reduced to 7%. The annuityowner exchanges the second annuity for a new annuity with a 10% premium bonus and an 8% commission. The 15-year surrender schedule is 18%, 16.8%, 15.6%, 14.4%, 13.2%, 12%, 10.8%, 9.6%, 8.4%, 7.2%, 6%, 4.8%, 3.6%, 2.4% 1.2%. The upfront cost of 18% is recaptured over the 15-year period meaning the gross product yield is reduced by 1.2% a year to 3.8%, but will return to 5% after this annuity begins its sixteenth year. The value transferred from the second annuity is $121,087 reflecting the $9,114 surrender charge. The $121,087 is credited with a 10% bonus resulting in a new accumulated value $133,196. The following chart reflects the annuity values for the original 20 year period of the Base Annuity Compared with the third annuity.  

 The bonus provides a higher accumulated value for the next 4 years, but with a much lower surrender value. At the end of the 20 year period the base annuity surrender value is $252,933 and the third annuity value is $221,826.  

Final Annuity  
By the end of year ten the third annuity surrender value is finally above the initial value transferred at the end of period year six. The annuityowner exchanges the third annuity for a fourth annuity that has a 10% commission and adds 20% to the payments if a lifetime withdrawal schedule is chosen. The 10 year surrender schedule is 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%. The upfront commission cost of 10% is recaptured over the 10-year period reducing the gross product yield by 1% a year. The 20% added to the income account is not a true bonus and is only realized if the annuityowner chooses to receive withdrawals over their lifetime and lives long enough to burn through their own money, therefore the cost of this “bonus” is 0.1% a year. Together the recapture of the commission and the cost of increasing a possible income payout are 1.1%; therefore the annuity is credited 3.9% interest. The value transferred from the third annuity is $134,215 reflecting the $20,411 surrender charge. The 20% income bonus does not affect the accumulated value so the new accumulated value remains at $134,215. The following chart reflects the accumulated and surrender annuity values for the original 20 year period of the Base Annuity Compared with the final annuity.  

After transferring to the fourth annuity at the end of year ten the original premium of $100,000 has a cash surrender value of $120,794. The annuityowner has earned $20,794 in interest for an annualized return of 1.91%.  During the same ten years commissions of $34,865 have been paid to the selling agent, and this does not include the additional commissions paid to any marketing companies in the commission hierarchy.  

Doing 1035 Exchanges cost the annuityowner $56,164 in lost account value

It may be argued this is not a fair comparison because it does not reflect the effect of income payout growth guarantees. However, if the exchange is not made the payout produced by the 20% income payout bonus in this comparison is exceeded by the actual account value of the first three annuities within one or two years. If all offer the same age based payout factors the value of the 20% income bonus is quickly matched or exceeded. At the end of 20 years the cash value of the original annuity would have been $252,933. If the annuity owner had stopped after one exchange the cash value would have been $241,086, after the second exchange the cash value would have been $221,826 and after the final exchange the cash value is $196,769. The serial exchanges cost the annuityowner $56,164.

CD Rates And Market Movement Both Affect Fixed Annuity Sales (2/09)
In an article last September I showed arrogance by saying, “If you do your research you find there is little historic correlation between stock market movement and fixed annuity sales, but a strong correlation between fixed sales and CD rates.” I went back, redid my research, and found out I was wrong. Altho fixed and indexed deferred annuity sales have tracked inversely with CD rates over the last decade – annuity sales fall when CD rates go up and sales rise when CD rates go down – fixed annuity sales also appear to correlate with stock market movement – annuity sales rise when the stock market is falling and sales fall when the market is rising. Neither correlation is simultaneous. In the annuity world it appears to take roughly six months before consumers notice or believe that CD rates and market movements have changed, and then alter their annuity buying habits.

Both stock market movement and CD rate levels strongly appear to have a direct cause and effect relationship on fixed annuity sales

I also said this correlation pattern was consistent and showed causality, meaning that falling or rising bank rates or stock market movement were causing the changes in annuity sales. I can build a strong case for causality going back a dozen years, especially for index annuities, but the evidence saying either CD rates or market movement affect annuity sales is much sketchier if I go back another decade

Part of the problem for the earlier years relates to differing definitions of what a fixed deferred annuity is and how sales should be counted. Some sales data I found separated group and individual annuity sales, others broke the world into flexible and single premium, and included group and individual sales on both sides. In many cases I was seeing three different totals with each labeled “total fixed annuity sales.” What I think I see f rom the mid ‘80s to the mid ‘90s is there was no apparent correlation between fixed annuity sales and either CD rates or the stock market. For example ,from 1984 thru 1990 the stock market rose and faltered and CD rates were higher than the previous year half of the time and lower the over half. Regardless of what was happening, annuity sales edged up. 

I think the reason is because individual annuity sales were so tiny – $12.5 billion in 1985 and still only $19.3 billion in 1990 – that any correlation was undetectable. So chastened, I will now restate my theory and say that both stock market movement and CD rate levels do have a direct cause and effect relationship on fixed annuity sales, but that the effect of CD rates is stronger. I’ll let you know if the future data supports my theory.

GLWB – Pricing Considerations & Reinsurance (5/07)
A cost arises for the insurance company if a policy’s account value is exhausted, yet the company has guaranteed that a periodic payment will continue to the owner. The risk comes down to:

How well does the account value perform?
How large is the periodic payment that is guaranteed?
How long will people be around to collect the guaranteed payment?

A simple starting point could be to look at an issue age 65, male, 5.5% guaranteed lifetime payment, annual total S&P 500 return of 8.0%, with a normal lapse assumption. The benefit cost in this instance is projected to be zero, as the credited interest always exceeds the guaranteed payment. But what happens if the account value performs poorly, or buyers of the benefit are older, because the guaranteed payment increases with age or the valuable benefit causes more people to stick around longer?

The projection is now changed to be an issue age 75 male, 6.5% guaranteed lifetime payment, annual total S&P return of 3%, with no lapses. The benefit cost goes up dramatically, to a projected cost of 64 basis points of account value annually.  

The projected benefit costs are dramatically different, but neither scenario is unrealistic. This simple example illustrates the risks inherent in evaluating the expected cost of a newly created benefit. Index linked performance is not under the insurance company’s control. The benefit is inherently more valuable, and costly, as age increases, which can lead to a higher average age among those electing the benefit. The existence of the benefit can alter historical lapse patterns.  


Age 65
Normal Lapse

Age 75
No Lapse

 A more dynamic pricing exercise should include consideration of:    

  • Stochastic modeling – The use of hundreds or thousands of randomly generated equity returns, which are used to generate the level of the indexed credited interest.  

  • Lapse rate sensitivity testing – Could involve testing a variety of selected lapse levels, or developing a dynamic lapse formula that automatically reduces lapses as the value of the GLWB benefit increases.  

  • Demographic risk – This is the actuarial term that means that you could end up selling more to high cost people than you originally anticipated.  

  • Tax qualified business – Since most benefits allow withdrawals to be larger if such is needed to satisfy Required Minimum Distributions, costs could be higher for tax qualified business.  

  • Mortality levels – The people who utilize a GLWB may be in better general health than people who don’t elect the benefit, resulting in higher guaranteed payments.  

  • Joint life payouts instead of single life – Most GLWB’s provide an option for guaranteed payments to continue as long as at least one of the two people are alive.  This should be tested separately.  

Reinsurance support is available as a risk management tool for the direct writers of these benefits. The most classic use of reinsurance for a direct writer is to substitute certainty for uncertainty, and this is of course the case here.  Reinsurance of the GLWB is available in the marketplace. The direct writer can achieve a known and guaranteed cost of providing the benefit. Another common reinsurance tradition is for the direct writer to receive a ceding commission from the reinsurer. That is, the cost of the reinsurance is less than the retail charge for the benefit, leaving a balance with the direct writer to cover expenses and contribute to profit. 

Based on the retail pricing that I’ve observed on GLWBs and reinsurance levels available, most direct writers should be able to reinsure their GLWB and retain an amount to cover their expenses and contribute to their profit. GLWBs for indexed annuities are rapidly gaining acceptance and availability from various direct writers is spreading. Competition will typically lead to improved benefits for the consumer, often accompanied by lower fees. Which means if the GLWB follows this typical course, expected margins will reduce over time. Reinsurance is available today, but I wouldn’t exactly call it abundant. So, the development of capacity and relationships now, while the environment is favorable, is likely to pay dividends in the future when the margins are more difficult.

Index Methods Are Priced To Perform The Same (except when they’re not) (4/07)
If policies using an annual point-to-point, monthly average, or cap forward crediting methods each paid 4 cents out of a dollar of premium to buy index-linked gain potential on that dollar of premium, the person guaranteeing to pay any index-linked gain would want to make sure that the probability of a payout was equal for all methods, or else everyone would flock to the crediting method with the best odds and the “gain payer” would lose over time. To use a gambling analogy, regardless of which number you place a bet on at the roulette wheel the odds are always 35 to 1. The reason why is that the casino has enough historical experience to show that over time the probability of each number coming up is the same and the odds are priced to reflect this probability.

The index annuity “gain payer”, who is usually the seller of the options covering the hedge, uses past index performance history and a guesstimation of how other factors like yields on alternative investments, volatility and such will affect the index movement, and then says what potential upside will be offered on the 4 cents received. The gain payer’s task is to try to ensure that everyone gets the same ‘spin of the wheel’ for their 4 cents. 

Just as a casino wouldn’t give better odds on one number at the roulette wheel the index annuity gain payer tries to ensure that the odds on the different crediting methods are also the same

What this means is that one annuity might offer, say, 100% of the point-to-point gain up to a 7% cap, while another gave 80% of the monthly average gain and still another used a 2.5% monthly cap forward-not loss structure to measure performance, because the gain payer believes that over the next year these crediting methods at these rates have the same odds in producing the same return. Does this mean that different methods perform the same each year? No. Different methods react differently to different markets. In 1998 averaging methods produced the highest index annuity interest, while in 1999 annual point-to-point with cap structures best rewarded their annuityowners. What the gain payer is betting on is that if their analysis, for example, showed that an 80% monthly averaging rate and an annual point-to-point with 7% cap produced the same returns over the long haul in the past that this will hold true for the future. The roulette wheel might come up showing 16 four times in a row, but over time the other numbers will come up and it will all even out.

Does this mean all index annuities will perform the same over time? No. Different products have different pricing structures based on carrier profit goals, commissions paid, and length of surrender period. This means one annuity might have 4 cents available from each premium dollar to use for the index-link and another has 3 cents and yet another has 5 cents. Ultimately, it is the carrier that decides how much money is available to fund the index-link. This is why I continue to believe that the most important factor in index annuity returns is not the crediting method, but how the carriers will treat the annuityowner when they reset that cap or rate in the future.

Suppose you find a carrier you can trust with an annuity offering a variety of crediting methods. My advice has been to split up the premium between two or three different methods, and put the policy in the drawer. My hypothetical backtesting models over the years have shown that the carriers set the rates each year so the yield potential odds are the same for each method over time. However, for the last several months when you plug in current rates for different methods and do a long time analysis for some carriers the returns are not all similar.

This chart backtests three actual products from three different carriers that offer multiple crediting methods within the product. I have applied each method’s current rates to calendar year S&P 500 Index movements from 1956 through 2006. Zeros have been substituted for negative years to reflect an annual reset methodology. The returns listed are the average yearly returns for each method at the current rate. I have not shown the specific rates or caps because the intent is not to compare the returns between different products, but to highlight return differences within each product.

  #1 #2 #3
Annual Point-to-Point – no cap 5.83% 5.83%  
Annual Point-to-Point  – 100% to cap 4.11% 3.67% 4.17%
Monthly Averaging – no cap 5.04% 5.83% 4.90%
Monthly Averaging – 100% to cap 4.61%    
Monthly Cap Forward 4.59%   3.94%

In all cases the uncapped methods had average returns that were higher than the methods paying 100% of index gain up to a specified cap. The uncapped point-to-point option had average returns that were 41% higher in the first product and 59% higher in the second than the capped choice. What this says is that if rates stayed the same and history repeated – although both assumptions are unlikely – the straight rate choice would produce a much higher return over time than the capped option. Product returns using averaging methods were grouped more tightly, but the uncapped monthly average return was still much higher than the annual point-to-point with cap selections. Another finding of interest was the monthly cap forward options had returns similar to the annual cap methods and much lower than the uncapped choices.

Today the uncapped methods offer the best potential returns for next year (based on history)...
What this appears to say is the gain payers are predicting low index volatility for the coming year and pricing methods accordingly. Essentially, their strong belief is a “more of less” capped method will produce the best returns. However, this means if they are wrong and the index goes up more than 12% to 14% over the next twelve months that selecting an uncapped method would produce an extraordinary advantageous return.

The difference in rates on different carriers offering identical methods is also higher than it usually is. Normally, products with similar surrender terms, commissions, minimum guarantees, and bonuses offer similar rates and caps when the underlying methods are the same. Today, I am seeing differences in carrier rates that do not seem to be justified by pricing or history. This means even if two products have identical features one could offer significantly higher rate potential for the coming year.

...but it is the carrier that ultimately determines annuity performance
Picking the best crediting method over the short term usually means being able to predict the future. The current market shows some methods and some products that appear to have better performance odds than others in the next year. Today’s index annuity marketplace is somewhat akin to a roulette wheel paying paid 35 to 1 on some numbers, 45 to 1 on others and 25 to 1 on the rest. But it should be remembered that unlike the casino the index annuity carrier can change the odds while the wheel is spinning – during the surrender period, and so regardless of any temporary advantage today’s pricing may give the ultimate game is still determined by the carrier on subsequent spins.

Index Annuity Product Feature Bloat (5/06)
The index annuity world a decade ago was a simpler place. The most adventurous product, the Integrity Omni Pathway, offered three different crediting methods within one annuity, but the rest of the field was content to offer one method, one surrender period, one index, and no bells & whistles. And then there was ELI.

In the fall of 1996 Jackson National Life introduced the Equity Linked Index annuity, or ELI for short. Although only one crediting method was used, ELI offered five different surrender periods and five allocation choices, including the first international index. Whether because of, or in spite of ELI’s complexity Jackson National sales steadily rose and they became the top selling index annuity carrier for 1999. ELI was the industry’s first step into the “Swiss army knife” approach to index annuity product segmentation, but definitely not the last.

Does offering more and more features create more index annuity sales or simply dilute?

During the next eight years carriers attempted to outdo the competition by offering products with new bonuses, new crediting methods, new indices, multiple surrender periods, and insurers introduced features borrowed from other annuity lines such as nursing home, terminal illness and unemployment surrender charge waivers. How far can it go? Midland National Life offers an incredible 253 different strategies spread among their now 23 products – and that total does not include choices of different levels of initial premium or annuitization bonuses, and even products with three different commission schedules from which to choose. A producer looking at the Midland lineup has an almost infinite number of ways to configure the choices ultimately presented to the consumer. Most carriers have not stressed product differentiation to the degree demonstrated by Midland. Only one product is offered by fourteen of the carriers, and an additional eight to ten multiple product carriers tell essentially the same story with only minor modifications. However, the reality is the majority of index products do offer a multitude of choices and features that are all designed to make each standout from their neighbor. The question is – Does offering more and more features and variations, also known in marketing circles as “feature bloat” create more index annuity sales?

One method to determine whether the addition of new features cause product sales to jump would be to measure sales before and after the features were added. However, there are usually many other variables in play and I find it impossible to sort out the direct effects of adding a particular feature. For example, if a product adds two new indices and the product sales go up 10%, did the new indices cause sales to increase, or did sales increase because there was more advertising of the new product, or did the marketing support people spend more time talking about the product, or did sales rise because the product’s main competitor coincidently lowered their rates just as the indices were added. To examine the effect of feature bloat on sales I found it necessary to look at academic research.

Feature Bloat Does =  More Sales  (even when the feature isn’t needed)

Product differentiation is a classic marketing strategy wherein one attempts to create a meaningful competitive advantage on the basis of attributes that are relevant, meaningful and valuable to the consumer. Suppose a potential annuity buyer had a history of health problems that pointed to the strong likelihood of a nursing home stay and the annuity value would then need to be tapped. An annuity offering a nursing home waiver making cash value available without surrender charges would be a meaningful feature. But what if the potential annuity buyer was in good health, already owned long term care insurance, and had sufficient resources so that the annuity value would never be needed – would the nursing home waiver feature still be attractive? Based on my research the answer is yes, even though the feature is irrelevant in this situation it probably is positively valued by the consumer.1 

Studies have also found a higher price could often be charged as more irrelevant features were added. The implication is an index annuity customer that only wants and needs a basic index product without bells and whistles would buy a different index annuity because the new product had more features, even though the features would not be used, and even if the potential return was lower. However, this does not mean that complexity in and of itself increases sales. A product viewed as too complex is less likely to be purchased.2

The message that needs to be presented is that the complexity is manageable, or that the product is not too complex when used as directed.  In addition, the extra features must be related to the product in some fashion to be valued.3 Letting a consumer choose from amongst three different crediting methods, as opposed to one, is perceived as meaningful. But throwing in a color television with each annuity purchase is unrelated and not valued by the consumer. Many studies indicate that consumers favor products with high levels of features, even when they feel the overall usability is decreased. But after using the product the initial positive feeling fades and dissatisfaction is higher for buyers of the complicated product and lower for buyers of simpler ones. A recent study concluded “Our research suggests that too many features can encourage initial purchase but damage satisfaction and reduce repurchase probabilities”.4

The product with the most features wins...once.

What does all of this mean for index annuity providers? If your goal is repeat business from your annuity customers then you should avoid offering products with so many variations that the consumer will suffer from feature fatigue, or else they will probably select someone else the next time around. However, if the consumer purchase is viewed as a one-time event the product with more features should win against the product with fewer features, but shouldn’t count on repeat business.

The Message
Adding more related features (indices, crediting methods, surrender penalty waivers) will make the consumer more likely to buy a particular annuity, as long as it can be presented in an easy-to-use fashion, but the chances for repeat business are better if the annuity is simpler.

1. Carpenter, Glazer, and Nakamoto (1994), “Meaningful Brands from Meaningless Differentiation: The Dependence on Irrelevant Attributes,” Journal of Marketing Research, 31 (August), 339–50

2. Mukherjee and Hoyer (2001), “The Effect of Novel Attributes on Product Evaluation,” Journal of Consumer Research, 28 (December), 462–72

3. Simonson, Carmon, and O’Curry (1994), “Experimental Evidence on the Negative Effect of Product Features and Sales Promotions on Brand Choice,” Marketing Science, 13 (Winter), 23–40 

4. Thompson, Hamilton, Trust (2005), “Feature Fatigue: When Product Capabilities Become Too Much Of a Good Thing,” Journal of Marketing Research, 42 (November) 431-442

Option World Glossary (7/05)
The lexicon of option terms within the index annuity pricing world does not end with “calls”.

American-style Option – may be exercised at any time before it expires. 

Asian Option – options that use averaging in determining either the strike or final price.

At The Money – the interest value is at the exercise price.


Basket Option – an option for which the underlying is more than one index. Example, an option that pays out based on the performance of the FTSE 100, Russell 2000 and S&P 500 would be a basket option.


Best-of Option – an option exercisable against the best performing indices. Example, a best of option on the Nasdaq and S&P 500 would pay out on the index that rose most during the term.

Call Option – gives the option buyer the right to buy an underlying security or index at a specified price on or before a given date.

Digital – one that pays out a fixed amount if the underlying index is above (or below) a specified level on a given date, usually the maturity date of the product.

European-style Option – may only be exercised on its expiration date

Everest – a structure that typically returns principal plus a bonus equal to a fixed amount plus the worst-performing gain from a given basket of underlying indexes.

Himalaya – the return is calculated as the average of the performances of the best index in each specified period during the term of the product. Once selected; however, the best performer is  removed from the basket for subsequent periods.


Hindsight Option – a type that pays out at maturity the difference between the strike price and the highest (or lowest) level of the underlying index during the term. Also called a look-back option.

Intrinsic Value – the degree the option is in the money.

In The Money – the market value of the optioned interest is above the exercise price of the option.


Knock-Out – an option that matures early if the underlying index rises to a specified level during the term.

Option – a contract which conveys to its buyer the right, but not the obligation, to buy or sell something at a specified price on or before a given date; after this given date the option ceases to exist. Insurers typically buy options to provide the index linkage for the excess interest potential.

Out Of The Money – the interest value is below the exercise price

Put Option – gives the buyer the right to sell an underlying security or index at a specified price on or before a given date.

Rainbow – an option basket whose best-performing indices are weighted more heavily than those that perform less well. 

Strike Price – the specified share price or index value at which the shares or interest may be bought or sold. If the strike price of a call option is less than the current market price of the underlying security the call is said to be in-the-money. Conversely, if the strike price of a call option is greater than the current market price it is out-of-the-money and would not be exercised.

Time Value – the cost or premium of the option above the intrinsic value.

Worst-of Option – a call option that is exercisable against the worst-performing index in a basket of indices.

Comparing Crediting Methods (7/05)
Index annuity crediting methods are often presented as enabling the consumer to participate in increases of the underlying equity index. To be accurate; however, with an index annuity the consumer is participating in the crediting method of the annuity, and not the index. To wit, 100% participation in an index annuity crediting method often bears little relationship to realizing 100% of the index performance. One crediting method may be used instead of another to give the illusion of earning higher potential returns, and it is this “smoke & mirrors” effect that may result in one annuity being purchased instead of another even though the potential returns of the former are no better, and may be even worse, than the later. 

The index annuity buyer participates in the crediting method—not the index

No one method is inherently better or worse than another method. An Oldsmobile, Plymouth or Studebaker will all get you from point A to point B although each one rides and handles a little differently during the journey. There are no bad methods, only perhaps, a bad understanding of how a particular method really performs. The method that most closely mimics that of the actual index is Annual Point to Point multiplied by a participation rate. How does it work?  If the index goes up 10% measured from the start of the contract year to the end, and the participation rate is 55%, the annuity would credit 5.5% interest. If the index went up 20% the annuity would credit 11% interest. That’s it – no caps, no spreads, no averaging. However, only two carriers offer annuities using this crediting method.

Why don’t more carriers offer this simple method? Based on my talks with producers it is because most producers say it is difficult to sell 55% participation. If that is so, let’s get an idea of the potential participation that is being sold. I calculated the annualized return for ten-year periods ending during the 1980s (71-81, 72-82...80-90) and 1990s (81-91...90-00) by applying a 55% participation rate to calendar years wherein the S&P 500 rose and crediting a 0% return in calendar years that fell. I then added up the ten 10-year period returns, divided the total by ten, and came up with average annualized returns for the ‘80s and ‘90s. I then worked backwards using the average returns produced by the 55% participation rate for each decade and determined what caps, rates or spreads were needed by the different crediting method to reach the same identical returns. If you applied a 55% participation rate annual reset method to the S&P 500 the annualized return for ten-year periods ending in the 1980s averaged 6.38%; the average for ten-year periods ending in the 1990s was 7.83%. The chart shows what other crediting methods needed to be to equal these returns.

All of These Different Crediting Methods Produce The Same Period Return

1980s 1990s
55% Annual Point-to-Point 55% Annual Point-to-Point
90% Monthly Average 86% Monthly Average
94% Daily Average 92% Daily Average
100% Annual Point-to-Point with 10.4% cap 100% Annual Point-to-Point with 12% cap
100% Biennial Point-to-Point with 20.5% Cap 100% Biennial Point-to-Point with 18.55% Cap
100% Monthly Average with 13.8% cap     100% Monthly Average with 13.55% cap    
100% Monthly Average less 1.35% spread        100% Monthly Average less 1.80% spread       
4.28% Monthly Cap on gain/not loss 3.42% Monthly Cap on gain/not loss
6.00% Monthly Biennial Cap on gain/not loss 4.25% Monthly Biennial Cap on gain/not loss
8.25% Trigger Binary Nonnegative Index  9.5% Trigger Binary Nonnegative Index  
206% Monthiversary/Monthly Term Averaging 123% Monthiversary/Monthly Term Averaging
2.40% Term Yield Spread 4.45% Term Yield Spread

What Does This Mean?
Whether you apply a 55% annual point-to-point or 94% daily average method to index movements in the 1980s you get the same overall return, and when you apply a 55% annual point-to-point or a 3.42% Monthly Cap method to the ‘90s you get the same return for that decade.

The illusion of higher potential returns

It also means that if your monthly average method had a cap of less than 13.5% or your monthiversary method kept a rate of less than 123% – looking at both the ‘80s and ‘90s  – your effective participation rate is less than 55%. In fact, you would be hard pressed to find an annuity that would have given you as much as 55% effective participation in these two decades when you plug in today’s current rates. At today’s rates and under most back-testing “what ifs” index annuities would deliver 40% to 60% of index performance. If consumers get roughly half of the index return is that bad?

Consumers Need Realistic Expectations
If we take today’s index annuities I would speculate that in a half-ways decent stock market, or when long-term interest rates push participation up a bit, that could mean long-term index annuity returns in the 5% and 6% range, which would place them roughly 2% higher than CD rates in a similar kind of interest rate environment, and that 2% more than the bank pays should be acceptable, based on the consumers I’ve talked with. 

What if consumers are being lead to expect “stock market like” returns from their index annuity? Then most of those doing the leading should pray for a really rocky market period, because if the future confounds the past and the stock market soars, the only high many index annuities might see are new lawsuits filed. The preponderance of the statistic forecasts and analysis I’ve conducted (guessing via computer) show index annuities performing like fixed annuities, which is what they were designed to do, and not like equity investments. Could index annuities generate “stock market like” returns if we had a repeat of the ‘90s? Yes, but long-term bond rates would need to significantly increase and those increases would need to translate into higher renewal participation on issued contracts. 

Alternatively, a couple of currently offered term end point crediting method products that do not lock in index gains until the end of the surrender period (to be specific, ING MarketSmart and Jackson National Life Elite 90), could offer 75% or more of actual index gain in a ‘90s scenario if there was even a modest increase in long-term bond rates. The flipside is term end point methods also perform like the index in down times, so a ‘70s market scenario would mean returns for these products near the minimum guarantee. 

The Reality
Regardless of what you think you see, annual and biennial reset index annuities are currently priced to return around half of what the index does over the longer term – producers today are usually selling around a “50% participation rate” but many do not realize it. The good news is index annuities can still be very competitive against CDs and other fixed annuities even at current rates if the index cooperates. Return measurement reflects the end of the year value for the S&P 500 Index. The results obtained should not be taken as a representation of actual returns for any product or period. Neither Standard & Poor’s nor Advantage Compendium market, sell or endorse any index product.

Can Index Annuities and Variable Annuities Compliment Each Other? (6/05)
by Rich Tucker, Vice President                                                            Ruark Insurance Advisors

I heard Mr. Tucker present this elegant and usable concept and asked him if he would allow me to share it with my readers. I encourage those involved in hedging on both sides of the annuity aisle to contact Rich Tucker at 973-783-3168 or –  Jack Marrion   

Proponents of index annuities and variable annuities are typically mutually exclusive. Strongly held beliefs are found among distributors of each camp and these beliefs tend to migrate upwards to the insurance company manufacturers. The results are insurers that focus only on their particular world of variable annuities or carriers seeing only their own universe of index annuities. However, a look at possible synergies from a broad perspective and joint management of index annuities and variable annuities can create cost savings.

On the one side you have carriers creating variable annuities. Variable annuities are being enhanced to offer additional guaranteed benefits such as guaranteed minimum death benefits, guaranteed minimum income benefits, and guaranteed minimum withdrawal benefits. All of these guaranteed benefits are “put-based” benefits, meaning they produce value if the underlying funds decline.  Insurance companies typically use reinsurance or hedging strategies to manage the financial risks of these variable annuity guaranteed benefits.

On the other side you have insurance companies manufacturing index annuities using “call-based” derivatives to manage the index participation formula. Call-based means they produce value if the index increases.  

Chart 1 is an example of the expected cash flow to the insurance company of a put-based variable annuity guaranteed benefit, in this case a guaranteed minimum withdrawal benefit. The insurance company makes money through 95% of the stochastic scenarios generated, but can lose significant money ($30 million for each $1 billion of sales) the remaining 5% of the time.

Index annuities, being call-based, have the opposite effect. Benefit costs to the insurance company arise when the underlying index increases.  

Chart 2 shows the expected cash flow to the insurance company under the same stochastic scenarios used previously. The shape of the curve is reversed, indicating that indexed annuity participation is high when variable annuity benefits are low, and vice versa.

What happens if an insurance company can combine the index annuity call-based participation formula and the variable annuity put-based guaranteed benefits?


Chart 3 shows the combined result, with $5 billion of GMWB combined with $1 billion of index annuity. The probability of a loss has decreased by four-fifths, from 5% in Chart 1 to only 1% in Chart 3.  The worst case loss has declined 86%, from $350 million to only $50 million. 

An insurance company that can create this type of combination should be able to significantly reduce  hedging costs. Although there are not many companies that currently have significant blocks of both index annuities and variable annuities, this may change over time if companies evolve to create balanced product offerings of indexed annuities, variable annuities, and fixed annuities. 

Reinsurance can be a mechanism to create this balance immediately. Companies with a heavy concentration in indexed annuities could assume reinsurance of variable annuity benefits, or companies with a heavy concentration in variable annuities could assume reinsurance of index annuities. I have found that this powerful risk management technique is not being utilized by the insurance industry today and could provide a win-win scenario for both variable and index annuity manufacturers.





Copyright 1998-2013 Jack Marrion, Advantage Compendium Ltd., St. Louis, MO (314) 255-6531.  All information is for illustrative purposes only, does not provide investment or tax advice. No index sponsors, promotes, or makes any representation regarding any index product. Information is from sources believed accurate but is not warranted. Advantage Compendium neither markets nor endorses any financial product.