Last week I had the opportunity to speak with Randy Bullard, Founder and EVP of Business Development at Placemark Investments. I planned to ask him a series of detailed questions about his company’s tax management and optimization services.
Here is the list of topics I wanted to cover:
- Tax Aware Transition of Pre-Existing Securities
- Tax Aware Withdrawals
- Tax Efficient Rebalancing
- Short Term Gain Deferral
- Systematic Tax Loss Harvesting
- Tax Lot Management Between Managers
Well, we didn’t get to all of these issues, but we managed to knock off a few. We went into such a deep dive that I lost track of time and before I knew it we had spent 90 minutes on the phone and both had to run to other meetings.
Randy’s Rules:
There are two things that all tax management systems should do: Absolutely eliminate wash sales & keep all capital gains as long-term.
Tax management is not tax avoidance. You should always realize gains smartly and as tax efficiently as possible.
Tax-Aware Transition of Pre-Existing Securities
When I was designing the rules for the rebalancing engine at Standard & Poor’s, I spent untold hours contemplating the myriad of issues surrounding transition of pre-existing securities (also referred to as “legacy assets”) into customer accounts. This includes brand new accounts that are funded with a mix of cash and securities and existing accounts that are being transitioned from one product to another.
According to Randy, Placemark takes the legacy assets and evaluates the risk characteristics using multi-factor risk data (provided by Northfield) in order to measure the correlation and affiliated risk of the securities versus other securities or an index.
There are many questions that need to be addressed before the transition can be properly evaluated. What is the tax impact to the client if we sell the stock(s)? What risk reduction do they see? Cost vs Risk? What is the risk of not selling? What is the appropriate measure of risk? What is the appropriate risk target?
What is the risk position going to be after the initial set of trades to begin the transition process? A client may have a bunch of managers he likes, so Placemark compares the correlation of the customer’s holdings versus the manager portfolios. They then choose the one with highest correlation. Even if they’re highly correlated, there is always a stock-specific risk. Multi-factor risk only explains 80-90% of a stock’s performance. The remaining 10% are due to factors beyond your control.
Randy then walked me through a widely seen scenario: A customer wants to open a new account by transferring his existing assets; 30 stocks, many with big long-term capital gains, with a total market value of $1 mil plus another $1 mil in cash equivalents.
One approach is to transition the customer’s assets into a Benchmark-Constructed Portfolio. Let’s assume that the customer has to keep 20 securities out of 30 and the target benchmark is the S&P500. Take the liquid portion of customers portfolio and buy securities that move the portfolio towards the S&P500. You can buy other securities in the S&P500 that have directly opposite offsetting risk factors versus the 20 stocks the customer owns. This creates a “completion portfolio,” which is something that institution wealth managers have been doing for a long time.
Any risk-model-based optimizer does this well, Randy added. What basket of securities should I buy with the $1 mil that combined with the 20 legacy stocks has the highest correlation and lowest tracking error to the target benchmark? If the 20 legacy securities are all in financial services, then you should use the $1 mil to buy a basket of securities that are negatively correlated to financial services. The basket is mathematically constructed to offset the risk factors of the portfolio. You can then create a new portfolio that has a tracking error within 5 bps of target benchmark. Of course, this is only one approach.
A different approach would have to be used if the client didn’t want an index as the target benchmark. Let’s say he wants to go for alpha and wants some actively-managed benchmark instead.
If the manager has 70 stocks in her portfolio and the client has 20 legacy stocks and over time wants to move into a portfolio that matches the manager. Now, you have a much more constrained portfolio and have to stick to the universe and relative weights of manager. This approach is more constrained, less flexible and incurs more risk to client.
Using the same basic optimizer, the 70 stocks in the manager’s model have correlations that are compared to the 20 legacy stocks that the client holds. A decision must be made on a liquidation schedule based on the correlation to the manager’s model. Perhaps the client wants a three year selldown, with no more than $50K in long-term gains realized per year, but still wants to minimize the tracking error vs the active benchmark.
So, which positions should be sold first? Some stocks have lots of risk associated depending on the target benchmark. This becomes one of the variables to optimize with, balancing risk versus tax.
For example, selling stock X might generate a $10K tax event and selling stock Y might generate a $20K tax event. But this decision is not as black and white as it seems. How much gains have already been realized? How much budget is left? You should usually hold highly-correlated stocks longer unless they have minimal tax impact.
Tax Loss Harvesting
Some overlay managers follow the strategy of generating losses even if the client has no offsetting capital gains. This is called “banking” the losses. However, they’re only worth $3,000 in the absence of offsetting gains, so you have to carry them forward to future tax years.
This strategy is bullshit, Randy claimed. This strategy involves banking losses that may or may not be of any value in the future. Randy’s opinion is that since the loss will still be there in the future, why bank it now?
This strategy bears some amount of tracking error and incurs transaction costs. Even parking the cash in an ETF will result in stock-specific risk (what if the stock you sold goes up during the 30-day wash sale period?). There is no actual value to do this today if there are no offsetting gains, Randy concluded.
Tax Alpha Claims
One thing in particular that I wanted to take a deeper dive into was Randy’s claim that Placemark’s discretionary overlay managent service can deliver between 70 -80 basis points in tax alpha. (from a meeting we had on 02/09/09)
We agreed that there is no GIPS-compliant way to measure tax alpha, but you can compare the taxable gains of your tax managed accounts versus the non-tax managed at the end of the year.
You could run a monte carlo simulation on any set of drivers, such as portfolio turnover.
Short term gain avoidance is by far the biggest way to generate tax alpha (80-90%), but requires an accurate method of tax code modeling.
In over 95% of client accounts, Placemark completely eliminates short term capital gain realization. In their non-tax optimized accounts, 50% of capital gains are short-term. They have to make some assumptions based on the projected amount of gains, such as what is the most appropriate ST/LT mix for each client?
For example, look at a SMA program over a 10-15 year period (before 2008!), Randy offered. Assume a market return = 8%, with 50% long-term and 50% short-term gains. The average SMA manager has 70-80% turnover, which means that short-term gains must exist!
If client X has a long-term capital gains rate of 20% and a short-term capital gains rate of 40% and the return was 8%, break out the 8% at these two rates. What if you only have gains at the long-term rate? What is value added for this client?
There aren’t good tools for doing tax optimization, Randy asserted. There are two classes of optimizers; mathematical vs rules-based. You can get 75% of value by using rules-based, so using an optimizer is often warranted.
Some product notes:
- Placemark built their tax management process 10 years ago. There was only one piece of software on the market at the time that did it right. They were an institutional financial optimizer engine. Would create an optimal blend of trades to solve problem.
- Placemark’s engine solves complex tax problems in seconds vs hours for other tools. They use Northfield’s multi-factor risk-base data, but not their optimizer.
