As a followup to my earlier interest in a trial version, I thought it would be worthwhile to capture the things that are frustrating me about Pralana Online as I do the "build" following the manual and Bill's book (which is very helpful). Some of these are nice-to-have's, others seem likely to directly affect the accuracy of computation.
In the nice-to-have's...
- As mentioned in my question about a trial version, I assumed that the magic "R" variable could be used for both miscellaneous and expense periods (outside healthcare where it is supported). I have several use cases, particularly work-related expenses that stop with R (licenses, continuing education) and expenses that start with R (increased travel, healthcare expenses). This can be worked around by adjusting values to match retirement dates, but that involves remembering all the dependent values across many tabs.
- An "L" variable as Bill suggests would be very helpful for LTC costs. Again, can be worked around manually.
- Advanced portfolio modeling wants asset breakdowns by percentage of total accounts. This is annoying, because the initial balances section allows me to enter multiple accounts, but here that information is unused. The statements I am drawing the balances from usually have an asset mix percentage pre-calculated, factoring in the content of mixed funds (target retirement funds, etc.). But on this page I have to externally compute those percentages and enter here. It should be easier to enter data by account.
- Very similar concerns with the "Account Fees" section. These are strongly correlated with the account, not just the asset class, so entry by account would be nice.
- In an attempt to workaround the asset breakdown I tried to create more asset classes ("Account A Stocks", "Account A Bonds", "Account B Stocks", etc.). But the limit here is 12 classes, which feels very arbitrary in the online product.
Starting to affect accuracy...
- Misc expenses and period expenses cannot be associated with a particular inflation rate. Since I have some medical expenses listed here (see below for why) they will not track healthcare inflation.
- The LTC expenses page assumes (per manual) that all LTC expenses are deductible, but this is not true for LTC insurance premiums. I have moved those to misc expenses, as suggested in Bill's book, but now they do not inflate with LTC inflation.
Finally, my largest accuracy issue is that the healthcare modeling is too restrictive. Beyond the bug I discussed here, there are not enough periods to reflect my possible plans, particularly when modeling early retirement.
- Relocation plans are not considered the start of a period, but they will affect the ACA premium modeling if moving to a lower COL area.
- Periods are based solely on parent ages, child age is not considered. So while I can enter the family premiums, in practice that means I'm modeling too high costs once my child is eligible (or required) to get separate healthcare. This date is unrelated to my R or medicare eligibility date.
I have not yet figured out a workaround for this, period expenses could do it manually, except that gets inflation wrong, and loses many (all?) of the benefits of Pralana's ACA modeling.
I should note that even if periods were more flexible I'd still want to associate medical inflation with period expenses as I expect to need to pay for some of my (future) adult child's medical expenses and premiums past 26, but in that case Pralana's ACA calculations aren't necessary as that will be out-of-pocket.
In summary, Pralana online is very impressive in many ways, but healthcare modeling seems limited to work best for two partners close in age, with kids already outside the home and without any disability. I'm glad I bought Pralana to try it, but I'm not sure how to get it to reflect my major cost (healthcare).
Thanks for reading, and suggestions for workarounds on these issues are welcome,
Axel
For advanced portfolio modeling, it sounds like you are using Mode 1. Unless you hold the same allocation in all accounts, that will seriously mess up and Roth Conversion calculations. I suggest you use Mode 2, where you enter the priority for which account holds stocks vs bonds. If you really want to, say, hold some bonds in taxable, you specify that and it will hold it until it can't do it anymore.
For Account fees, those are set by account under Financial Assets-Accounts and the Account Fee tab.
It does sound like some programming would be needed to meet what you'd like to do. My 2 cents worth - Perhaps the simplest solution on the LTC is for the program to add "% tax deductible" entry to each expense category? For childrens' healthcare, including ACA, perhaps the easiest programming thing would be to have a tab under Children for healthcare.
Your note sparked me to look up healthcare inflation and surprisingly (to me at least), since 2012, it has been no faster than general inflation at an average of 2.57%/year. Who knew!
@ricke I am using Mode 1, but I don't think Mode 2 helps. If anything it seems to make the math harder? AIUI mode 2 seeks to distribute the entered asset allocation across the account types by prioritization. The total balance of asset types across all accounts of a type is obviously very important, but I hoping to calculate it from the actual asset distribution from the real accounts, not allow mode 2 to try to force it. For example, I have two taxable accounts one of which is less stock heavy because I hold more stocks in a different account. I can compute the actual asset percentages of the taxable account myself, but it feels like something Pralana could do. In particular since its already aware of the particular accounts balances, unrealized gains, etc.
Fees are roughly the same, I have accounts of the same type with low fees and others with higher fees. Again, I can work out the fraction of each account type comes from the particular accounts (and then the fees inside that account, etc.), but Pralana already knows what fraction of each account type is coming from each account already.
Yes, I can do this on a scratch sheet (and I did), but as a new user coming in cold it felt very forced by the tool.
I'm not sure why mode 1 would cause Roth errors, could you elaborate? Or maybe I'm not that far in the manual (and book)? I'm under the impression that roth calculations are already aggregated across multiple accounts of the same type so I'm expecting that if the Roth calc wants to sell a certain percentage of an asset type from Roth it will be up to me (eventually) figure out which real-world account to pull that from?
I'm aware of the flattening of healthcare inflation, but because of my particular situation, I think I'm better off assuming a higher rate to be conservative.
Mode 1 (which is the only method that other commercial tools support) will mess up Roth Conversions if your stock allocations are different between accounts. To take a typical example, let's say your Roth is all stocks and your traditional IRA is all bonds. When a conversion happens in Mode 1, the program takes money from the IRA and puts it in the Roth, but that removed bonds from the portfolio and added stocks. That changing bonds to stocks is a lot bigger deal in the calculation of future wealth than the tax benefits of a Roth Conversion. So the tool will incorrectly recommend gigantic Roth Conversions to get stock returns instead of bond returns.
Mode 2 will hold the overall portfolio allocation constant from year to year and eliminate the problem.
It's generally unattractive to hold bonds in taxable, you might want to check out the bogleheads.org wiki on the subject of tax efficient fund placement:
https://www.bogleheads.org/wiki/Tax-efficient_fund_placement
Thank you for the explanation, it makes sense, though since I have not used any of the conversion optimization tooling yet, I would not have noticed. I now see its mentioned in the manual as well (citing you I believe).
My newbie objection is mostly around the data entry required to get the portfolio modeling to reflect my starting state which seems like a ground truth problem, regardless of whether I have previously considered tax efficiency. If Mode 2 is more accurate (at least for Roth conversion computations), then I think that frustration of external computation still seems valid?
To get to Mode 2, I will need to do more external calculation to basically back-compute my portfolio percentages from current balances and current allocations so that the prioritization step will approximate actual allocations. The tool could do this, let me enter my current balances by account and their percentages off my statements, and then prefill the Mode 2 first period based on that data?
But again, this is nice to have, and perhaps only affects the approachability of the tool at first. The healthcare modeling is my larger concern.
That said, thanks again for the explanation!
Versions of the Mode 1 problem would plague your results even without Roth Conversions as spending from taxable would skew your asset allocation one direction and then RMDs would skew it a different direction. It might be hard to spot that the results didn't make sense though.
Maybe the question should be turned around to ask whether you should keep the asset locations that you have? It's costing you a lot of money, are you sure it's worth it? Bonds belong in tax deferred to reduce annual tax drag. Annual tax drag is a depressingly large lifetime cost. Holding your bonds in tax deferred also slows the growth of tax deferred, reducing your future RMDs.
Folks' typical reason for wanting bonds in taxable is they want income, but if they think it through, when they need income, they could sell a stock in taxable and pay the capital gains. Not all of the sale price is gain, so not all of the proceeds would be taxed and the the portion that is taxed is taxed at a more favorable rate than ordinary income. To keep the portfolio in balance, they could then sell some bonds in tax deferred and buy stocks there. So it's not as easy to hold things tax efficiently, but it's profitable.
Thanks for the warning about mode 1, since I haven't fixed healthcare spending yet, I'm not putting too much weight on the future modeling (or any other aspect of the results). I'll try to move to mode 2 before I trust it too much.
As for turning the question around, I think that's focusing too much on assumptions about the root cause of my feedback. I'm trying to provide the same feedback I would in my professional capacity (in software) about the onboard experience, and what new users encounter vs what experienced users encounter. Its a common problem for onboard to be poor because the developers understand too much about how the tool works, or assume a particular mental model on the part of users.
Yes, users could change their allocations to match the tool's expectations, and that might be the best thing for them to do financially, but I tend to think that's an unreasonable precondition for onboard. Starting with ground truth seems like a reasonable place to begin. Or perhaps the user's goal is to model less tax efficient allocation in one scenario and a better one in another. Both seem like usecases that Pralana can support.
To be clear, I don't think any of my nice-to-haves are fatal, nor do I expect the Pralana folks to treat them any differently than any other feature requests. I'm just pointing out ways in which the initial experience (and any future trial mode) will potentially come across to at least some new users. In particular, I suspect some limitations are a result of the Excel heritage of the product, and may no longer be relevant for online.
That said, I realize your concern is that my problems with the onboard come from the possibility I have a particularly tax inefficient allocation, and the real-world implications of that. In practice, I'm much more close to tax efficiency in the account types than I think you suspect, and yes, bonds are held more in tax advantaged accounts. I expected that, because that's what the planner I work with has previously indicated, but I think the process of making Pralana understand that could be made easier.
Axel, I suggest that you first define your priorities, what is your main goal, and then what are the important steps that will bring you there. For me Asset Allocation (overall Stocks/Bonds ratio i.e. managing risk/reward) is step #1 and Asset Location (which investment to put in which account category, i.e. tax management) is #2. Asset Allocation is NOT ‘nice-to-have’; it is probably the most important investment decision to have. Healthcare modeling is lower priority compared to the other two. You are probably losing much more money (or take unintentional risk) if you do not manage #1 and #2 compared to healthcare. I 100% agree with Rick.
Axel, I suggest that you first define your priorities, what is your main goal, and then what are the important steps that will bring you there. For me Asset Allocation (overall Stocks/Bonds ratio i.e. managing risk/reward) is step #1 and Asset Location (which investment to put in which account category, i.e. tax management) is #2. Asset Allocation is NOT ‘nice-to-have’; it is probably the most important investment decision to have. Healthcare modeling is lower priority compared to the other two. You are probably losing much more money (or take unintentional risk) if you do not manage #1 and #2 compared to healthcare. I 100% agree with Rick.
@hecht790 please read the top post, "nice-to-have" in this context refers to the desirability of software feature improvement and the experience with it during onboard. It is not, in any way, a reference to whether real-world asset allocation considerations are important or not
I feel like this conversation is becoming dominated about assumptions about the reasons I have frustrations with the tool, and not the actual feature suggestions. So, just to be completely clear... I am aware of the general implications of asset class allocations, though not specifically how they impact the modeling in this particular tool. In that way, the mode 1 vs mode 2 distinction a was a helpful education, though one I am relieved to see is actually mentioned in the manual.
More generally, I'm not, at least in this thread, seeking particular advice on strategy. I'm sure I may have questions about that in the future, but right now, today, that discussion isn't my focus. My allocations are not anywhere as poorly considered as seems to be assumed.
Again, more than anything else, I want to understand how to model my healthcare costs, for which I have no workaround. I'd rather focus on that vs the "nice-to-have" feature requests which seems to have come to dominate the discussion.
I hope that Stuart and Charlie read the thread, a lot of your suggestions/frustrations look like things that can be improved without major re-writes. Having been a customer for a few years now, one of the things I really like is the responsiveness and speed that bugs are fixed and features added.
@axel You're absolutely correct that the tool isn't designed to handle every conceivable case. We've tried very hard to provide the best capabilities for most people, but we undoubtedly come up short for some folks. As you say, there's no capability to model changing ACA insurance costs due to a relocation during the middle of a period, the periods are based strictly on the parents' ages, and there's no provision for the tapering of costs in period 5. Medical costs for a dependent child that extend into that period would probably need to be captured elsewhere. You can associate medical inflation with the period expenses on the Healthcare expenses page, just not uniquely to specific periods. Is this what you're wanting to see (period-specific inflation rates)?
We will take your inputs to heart and consider enhancements, but please understand that we have lots of other enhancements in the works too and these things have to be prioritized based on perceived benefits to the broadest base of our users. Maybe other users can suggest some workarounds to help you and I'll circle back soon and maybe I can offer some suggestions.
Stuart
@smatthews51 I completely understand that there are many feature requests, and that no tool can fit all users. That's why I was asking about a trial in the prior thread. Since I decided to give it a try with the paid tool, I figured the least I should do is point out some areas for potential improvement. And yes, I understand you cannot prioritize these fixes unless they are a general benefit.
That said, no, the healthcare periods don't really solve the complexity of modeling my healthcare expenses. As we're discussing in the other thread, the periods don't completely align with my expectations even for just the parents and different retirement/medicare dates. Adding the child's potential ACA enrollment to the puzzle increases the permutations in a way that I think would overcomplicate the way Pralana wants to flatten the parent's healthcare expense into periods.
The relocation change in parent medical costs seems like an oversight that I'm surprised hasn't come up before, I suppose I should just assume a relocation aligned with one of the period starts.
For the rest, the simplest thing I can suggest is just to allow line item expenses in the period and miscellaneous tabs to override the inflation rate, either by naming the rate to use (i.e "healthcare") or entering an offset from general inflation. This is an option in other tools.
Per expense inflation overrides would allow me to start one or more "miscellaneous" expenses in the year my child can no longer be covered by my insurance to represent their separate insurance premiums, out of pocket medical expenses, etc. If each of these line items could be assigned to the "healthcare" inflation rate that would better reflect the expense growth in the plan. I'd still need to externally compute the child's insurance premium and tax implications, but that's fine.
This approach would mean I will be overestimating some family premiums in one of the parent healthcare periods because my child will no longer be covered, but I can understand that as a side-effect of simplification of parent healthcare periods.
Line item inflation would also help with the LTC premiums. I can split the premium into two expenses now, one portion deductible and one non-deductible, and then both could be assigned the LTC inflation rate.
So far the only workaround I can think of is to list all these expenses as miscellaneous items and then boost my general inflation rate to match my healthcare inflation rate based on the proportion those expenses make up of my total expenses. The downside of this is that it makes the model a little more brittle. For example, it probably means I should use nominal instead of real rates of return, and I'm guessing consumption smoothing will break the implicit expense/inflation relationship. Any other concerns with this approach?
@axel Charlie and I have discussed the issues you're facing in trying to model your case and are planning to make the following enhancements:
1. On the LTC expenses page, add a control that allows you to specify whether the expense of a given line item is tax deductible or not. It already has a control that allows you to associate each line item with the healthcare or LTC inflation rate, so I think we're good to go there.
2. On the Miscellaneous expenses page, replace the "Is Amount Adjusted for Inflation?" column with two columns: one for "Annual Expense Adjustment %" which defaults to a blank and another for "Real or Nominal" which defaults to Real. This is a generic approach and if these fields are left blank, the line item will inflate at the general inflation rate; otherwise, it will inflate at the specified (constant) rate. This will be transparent to existing users since these fields default to blanks.
3. We're still considering any changes to the Healthcare period implementation because it's very complex and I think we've agreed (via the other thread) that there is a viable workaround available for your case.
Stuart
@smatthews51 thanks for detailing these potential enhancements!
1. Yes, this sounds reasonable, though for LTC premiums its not quite that simple ( https://apps.irs.gov/app/vita/content/00/00_25_005.jsp). However, since LTC expenses can be associated with an age range I expect LTC premiums can still be modeled by breaking up the deductible/non-deductible component over the IRS age ranges.
2. Sounds great. As I've implied elsewhere, I expect some of my healthcare expenses to rise faster than even general healthcare inflation, so finer control is welcome. Are you thinking of this only for the misc expenses page or also for line items on the phased expenses? I probably can do everything I need with misc expenses, but could see this being useful in general in both places.
3. Ageed, I can manually adjust my spouse's medicare date manually. From what I see the major consequence of this is that I need to be cautious with some of the automatic analysis, particularly "earliest safe retirement", since there are now dependent variables not adjusting relative to the analysis. That's already the case for some of the other expenses (no expense ranges based on "R"), so its not a major loss.
Thanks again considering my feedback, I'll keep an eye out for the changes. (I work in software, so I know not to ask for any sort of ETA 😀 )
