Optimizing the ACA PTC (Premium Tax Credit)

By Alan Silverstein, Fort Collins, Colorado. Email me at ajs@frii.com.
Last update: December 14, 2020

Here are some thoughts and free advice on optimizing the ACA PTC for early retirees. This webpage might interest you if (like me) you have an unnatural fascination with financial economics...

...or if you are early-retired in the USA, meaning no employer health insurance and not yet on Medicare (65 or disabled) and you buy health insurance through the ACA marketplace.

...and if you have a low enough taxable income (never mind your assets, or actual spending if much of it comes from post-tax resources); that is, below 400% of the federal poverty line, about $68K for MFJ (married filing jointly) in the continental US in 2020; to be eligible for the ACA PTC (premium tax credit); whether you get it post-facto, or monthly as an APTC (advanced PTC).

How to optimize your situation:

Let me see how simple I can make this. I won't repeat background information you can find on websites like this one under Obamacare Facts.

Preplan taxable income: If you want to do 401k or traditional IRA withdrawals (distributions) or Roth conversions, but not cross 400% FPL (federal poverty line) with your MAGI (modified adjusted gross income for ACA purposes) and lose thousands of dollars of ACA PTC (see below for details), you must model your tax situation before the end of the year and act accordingly.
Note well: Use the FPL number for the correct tax year (and of course family size and location). For some reason, each tax year they use the FPL from the previous calendar year. Don't repeat my mistake for TY2015 where I got very close -- to the wrong target, higher than I expected because I used the calendar 2015 FPL not TY2015 FPL. (Which then revealed a bug in TurboTax, but that's another long story!)
Reduce MAGI: There are many ways you can reduce your Modified Adjusted Gross Income (for ACA purposes) through deferring optional income (see Schedule 1 Additional Income section) or increasing above-the-line adjustments to income = deductions (see Schedule 1 Adjustments to Income section).
But remember if you're already getting Social Security benefits, even the non-taxable portion (TY2020 Form 1040 line 6a - 6b) counts toward your ACA MAGI! Along with tax-exempt interest (TY2020 Form 1040 line 2a) plus foreign earned income and housing expenses for Americans living abroad (Form 2555).
After the end of the tax year, I know of only four, err, now three, err, add one more, ways you can retroactively reduce your MAGI to maintain your PTC eligibility:
1. Make a retroactive deductible traditional IRA contribution, if you have eligible earnings and haven't already max'd out for the previous tax year. (Increases TY2020 Schedule 1 line 19.)
2. Make a retroactive deductible HSA (health savings account) contribution, if you are in an HDHP (HSA-eligible high-deductible health plan) and haven't already max'd out for the previous tax year. (Increases TY2020 Schedule 1 line 12.)
3. Hunt for any overlooked deductions (Schedule 1 Adjustments to Income section).
4. (Later, Bruce Blondina also suggested:) If you just started Social Security within the last 12 months, consider paying it back from non-taxable sources to reduce your ACA MAGI. This is called a withdrawal of your Social Security retirement claim. You can only do this once, and must repay all amounts received (no interest charged). When you reapply in the future, you receive a larger monthly amount.
(One other trick was to recharacterize a Roth conversion from last year, but that went away in 2019.)
Other PTC-retaining strategies to consider:
1. If newly married during the tax year, try the alternative marriage calculation to prevent or reduce repayment for the months before marriage. (Web search for details.)
2. Consider the Married Filing Separately status to disallow PTC, but apply the APTC repayment cap to only one spouse's income.

Anecdotal ideas from our own situation:

Late each tax year, my wife and I refill our short term spending buckets (for the next 1-2 years), partly by doing some traditional IRA withdrawals (distributions). (Early in retirement I did only Roth conversions because I wasn't 59.5 yet, and had enough other post-tax resources to live on.) However while buying ACA health insurance through the marketplace, that is until age 65 or disability puts us on Medicare, we are constrained by the ACA PTC (MAGI) limit; plus a desire not to exceed the Colorado $20K/person retirement funds exemption either, which would cost us another 4.63% on any extra withdrawals or conversions.
Ideally a Roth IRA should hold "bottom dollars" that you touch only in an emergency or much later in retirement life (so that it grows tax-free as long as possible). When you're younger, you can't withdraw converted amounts without penalty for 5 years, but you can always withdraw older contributions. After age 59.5 though, you can withdraw any money you like, all tax and penalty free.
This suggests that now I should not withdraw anything from my traditional IRA for direct spending; instead do a Roth conversion again for 2015 (and in later years). But from now on if I do need it for spending, take the equivalent amount back out of the Roth in the next 1-2 years guilt-free.
The downside of an "extra" Roth withdrawal would be just a bit more accounting work and one more 1099-R for tax reporting. The upside is I have another "safety valve" to reduce our MAGI retroactively. I can recharacterize (although it's another accounting pain) if we discover in February that we accidentally crossed the MAGI limit somehow despite our best planning.
In fact someone not bothered by the accounting overhead could do a much larger Roth conversion and retroactively recharacterize it down to be just $1 below the MAGI limit! (Don't know if this is considered an "abusive tax shelter".)

More thoughts about relevant details:

The federal EMTR (effective marginal tax rate) for someone in this part of the ballpark (300%-400% of FPL) isn't just the 10% federal marginal rate, it's another 9.5% (in 2016, 9.56%) for lost PTC too. Just another variable to consider. (So the AMTR = additional MTR is 9.5%, and the EMTR is, say, 10% + 9.5% = 19.5%.)
Your personal PTC depends both on your personal ACA MAGI and your "personal" SLCSP (second lowest cost silver plan), which depends on your zipcode and/or county, age(s), gender(s) and tobacco status. See the calculation details in Form 8962, Part II.
If you're lucky, your annual (new for 2014) Form 1095-A just contains the magic SLCSP value you need; otherwise you must hunt it down. In Colorado at least, I recently discovered that HealthCare.gov eventually sends you to the state marketplace site ConnectForHealthCo.com, which in turn ends up requiring you to call a live agent, whose software in turn forces you to provide your estimated MAGI as if you were asking for APTC when all you want to know is your SLCSP for PTC estimation or calculation. But I did get our personal SLCSP number.
And I suspect that since for 2016 my wife and I are buying (identical) individual plans instead of a family plan to avoid aggregate-policy deductible and out of pocket maximum penalties (you can web search for more on this), our 2016 1095-A's won't contain useful SLCSP numbers, and we'll have to obtain them as-if in a family policy.
One website I found suggested you should also consider breakeven points, where for example falling off the PTC cliff (losing it completely at MAGI > 400% of FPL) is a smaller hit than your net increase from, say, starting SS sooner. Or conversely if you're in the PTC boat, you can use the loss-of-PTC argument as another support for deferring SS at least until age 65 (with Medicare).

Why to consider skipping APTC and just seeking PTC:

When my wife's early retirement health coverage ran out (really didn't, just jumped 14x in cost and become technically unaffordable) in September 2014, we went through the Colorado marketplace and, not knowing any better, got APTC for four months.

During 2015 open enrollment we had to again obtain a "Medicaid denial number" to proceed with APTC through the marketplace, but the Colorado website sent us to a Medicaid site (Colorado Peak) that, being poorly designed, required us to enter (and attest under penalty of fraud to) a lot of irrelevant asset-based information to get the denial. (Medicaid is means-tested, but the ACA is supposed to be only income-tested.) It was crazy considering my wife has some rental houses. And there was no workaround, not even appealing to a phone-in agent's supervisor to just take our word for it. In disgust I gave up.

And then -- we experienced that paying up-front the full insurance premium cost, a considerable amount, and aiming for a whopping PTC refund at tax time instead, was much simpler. No need to pre-estimate our MAGI, qualify for APTC, notify them (painful) if our MAGI estimate changed by 10%, etc.

This approach requires us to carry the total annual premium costs out of pocket and "gamble on the come" -- but we'd have to pay back excess APTC anyway if ending up with MAGI too high -- but within payback limits unless you go over 400% of FPL and lose it completely, in which case you must return it all.

I think the APTC system is set up for the average low-income taxpayer who (a) needs premium help and (b) cannot afford to defer PTC until the following February-April timeframe.

Bear in mind that for low-enough MAGI relative to FPL, you can also get CSR (cost sharing reductions) that lower your deductibles, and which you do not have to repay if your MAGI estimate ends up too low (assuming they don't come after you for fraud). But fortunately we're not low enough to worry about CSRs... Just getting PTC will be good enough.

Thoughts on ACA PTC EMTR:

The EMTR (effective marginal tax rate) for ACA PTC (premium tax credit) loss with increasing MAGI (modified adjusted gross income) is 9.5% at higher MAGI levels (in the 300-400% of FPL = federal poverty line range), and turns out to be higher percentages (EMTRs) at lower MAGI values. (To be clear, by EMTR here I mean just the AMTR = additional MTR, added to the base MTR on ordinary income.)

However if you examine the Form 8962 instructions (such as for 2014), table 2, the Applicable Figures (AF) start at 0.0200 for 133% of FPL, rising to 0.0950 for 300% or higher (for 2014). On the face of it, the EMTR for PTC loss at lower MAGI should be just this marginal rate, which is less than 9.5%, right?

However, exploring with a PTC estimator (marketplace or provider, in this case under kp.org because it's easier than ConnectForHealthCo.com but gives consistent results) does show much higher EMTRs for lower MAGIs; say in the 14-17% range. (Really EMTRs inferred from the loss of PTC/year versus increased MAGI/year, for various sample points.) How can this be?

It bugged me, and it took a while, but I figured it out...

First let me say that I don't know why "they" wrote the ACA to use a progressive rather than flat applicable figure (multiplier). What's wrong with saying that for everyone, their maximum expected spending on health care premiums should be 9.5% (or whatever) of their MAGI? Double your MAGI and your required contribution doubles, etc. But no, if you make 150% of FPL you must spend only 4% on healthcare premiums, but if you make twice that (300%) then you must spend 9.5%, or really more than four times as much as the guy earning half as much as you.

The problem with any progressive scale like this is that there's a hidden extra EMTR for the rate itself changing as you climb the ranks. (By the way, this is similar for Social Security taxability.) This arises from the fact that even for a 1 ppt step (like 240% to 241% of FPL), the small increase in AF is retroactively applied to the entire MAGI base amount before the step.

In this case it's a huge effect, approximately doubling the AF rate in most cases. In other words, for lower MAGIs:

  (MAGI * (AF delta)) ~= ((MAGI delta) * AF)

I intuitively expected the EMTR to just be the AF, but of course it's the sum of the AF plus the effect of the small AF delta on the entire base MAGI. Now once you hit 300% of FPL the AF freezes at 9.50% (or for 2015, 9.56%, yes even the rate goes up), so the AF delta vanishes and further MAGI increases do have an EMTR of just 9.5%, as expected.

If like me your main concern is not crossing the 400% FPL point and losing PTC completely (thousands of dollars), you don't care as much about the EMTR at lower MAGI values. But it does add to your ordinary income tax rate. Say in Colorado, in this scenario you might be in the 10% federal bracket (assuming much of your income is not preferential like long term capital gains) + 9.5% PTC loss + 4.63% state = nearly 25% marginal tax rate!

I wrote a little Perl script to examine the ACA PTC EMTR, and here's the output for the 2014 constants. The PRODUCT column just multiplies the %FPL times the AF (applicable figure) from the table for easy viewing. You can pretend 1% of FPL = $1 for this purpose, it doesn't matter since locations, household sizes, and MAGIs are "normalized" to %FPL for all taxpayers. (These variables all distill down to a %FPL index.)

The EMTR2 column just looks back over two %FPL steps instead of one to average out (a little) the raggedy nature of the EMTR% for single steps that ultimately arises from AF's being rounded to just three decimal points.

%FPL AF    PRODUCT   EMTR%   EMTR2
 131 0.0200   2.62    -       -
 132 0.0200   2.64   2.000    -
 133 0.0300   3.99 135.000  68.500
 134 0.0306   4.10  11.040  73.020
 135 0.0312   4.21  11.160  11.100
 136 0.0318   4.32  11.280  11.220
 137 0.0324   4.44  11.400  11.340
 138 0.0329   4.54  10.140  10.770
 139 0.0335   4.66  11.630  10.885
 140 0.0341   4.77  11.750  11.690
 141 0.0347   4.89  11.870  11.810
 142 0.0353   5.01  11.990  11.930
 143 0.0359   5.13  12.110  12.050
 144 0.0365   5.26  12.230  12.170
 145 0.0371   5.38  12.350  12.290
 146 0.0376   5.49  11.010  11.680
 147 0.0382   5.62  12.580  11.795
 148 0.0388   5.74  12.700  12.640
 149 0.0394   5.87  12.820  12.760
 150 0.0400   6.00  12.940  12.880
 151 0.0405   6.12  11.550  12.245
 152 0.0409   6.22  10.130  10.840
 153 0.0414   6.33  11.740  10.935
 154 0.0418   6.44  10.300  11.020
 155 0.0423   6.56  11.930  11.115
 156 0.0428   6.68  12.030  11.980
 157 0.0432   6.78  10.560  11.295
 158 0.0437   6.90  12.220  11.390
 159 0.0441   7.01  10.730  11.475
 160 0.0446   7.14  12.410  11.570
 161 0.0451   7.26  12.510  12.460
 162 0.0455   7.37  10.990  11.750
 163 0.0460   7.50  12.700  11.845
 164 0.0464   7.61  11.160  11.930
 165 0.0469   7.74  12.890  12.025
 166 0.0474   7.87  12.990  12.940
 167 0.0478   7.98  11.420  12.205
 168 0.0483   8.11  13.180  12.300
 169 0.0487   8.23  11.590  12.385
 170 0.0492   8.36  13.370  12.480
 171 0.0497   8.50  13.470  13.420
 172 0.0501   8.62  11.850  12.660
 173 0.0506   8.75  13.660  12.755
 174 0.0510   8.87  12.020  12.840
 175 0.0515   9.01  13.850  12.935
 176 0.0520   9.15  13.950  13.900
 177 0.0524   9.27  12.280  13.115
 178 0.0529   9.42  14.140  13.210
 179 0.0533   9.54  12.450  13.295
 180 0.0538   9.68  14.330  13.390
 181 0.0543   9.83  14.430  14.380
 182 0.0547   9.96  12.710  13.570
 183 0.0552  10.10  14.620  13.665
 184 0.0556  10.23  12.880  13.750
 185 0.0561  10.38  14.810  13.845
 186 0.0566  10.53  14.910  14.860
 187 0.0570  10.66  13.140  14.025
 188 0.0575  10.81  15.100  14.120
 189 0.0579  10.94  13.310  14.205
 190 0.0584  11.10  15.290  14.300
 191 0.0589  11.25  15.390  15.340
 192 0.0593  11.39  13.570  14.480
 193 0.0598  11.54  15.580  14.575
 194 0.0602  11.68  13.740  14.660
 195 0.0607  11.84  15.770  14.755
 196 0.0612  12.00  15.870  15.820
 197 0.0616  12.14  14.000  14.935
 198 0.0621  12.30  16.060  15.030
 199 0.0625  12.44  14.170  15.115
 200 0.0630  12.60  16.250  15.210
 201 0.0634  12.74  14.340  15.295
 202 0.0637  12.87  12.400  13.370
 203 0.0641  13.01  14.490  13.445
 204 0.0644  13.14  12.530  13.510
 205 0.0648  13.28  14.640  13.585
 206 0.0651  13.41  12.660  13.650
 207 0.0655  13.56  14.790  13.725
 208 0.0658  13.69  12.790  13.790
 209 0.0662  13.84  14.940  13.865
 210 0.0665  13.96  12.920  13.930
 211 0.0669  14.12  15.090  14.005
 212 0.0672  14.25  13.050  14.070
 213 0.0676  14.40  15.240  14.145
 214 0.0679  14.53  13.180  14.210
 215 0.0683  14.68  15.390  14.285
 216 0.0686  14.82  13.310  14.350
 217 0.0690  14.97  15.540  14.425
 218 0.0693  15.11  13.440  14.490
 219 0.0697  15.26  15.690  14.565
 220 0.0700  15.40  13.570  14.630
 221 0.0704  15.56  15.840  14.705
 222 0.0707  15.70  13.700  14.770
 223 0.0711  15.86  15.990  14.845
 224 0.0714  15.99  13.830  14.910
 225 0.0718  16.16  16.140  14.985
 226 0.0721  16.29  13.960  15.050
 227 0.0725  16.46  16.290  15.125
 228 0.0728  16.60  14.090  15.190
 229 0.0732  16.76  16.440  15.265
 230 0.0735  16.90  14.220  15.330
 231 0.0739  17.07  16.590  15.405
 232 0.0742  17.21  14.350  15.470
 233 0.0746  17.38  16.740  15.545
 234 0.0749  17.53  14.480  15.610
 235 0.0753  17.70  16.890  15.685
 236 0.0756  17.84  14.610  15.750
 237 0.0760  18.01  17.040  15.825
 238 0.0763  18.16  14.740  15.890
 239 0.0767  18.33  17.190  15.965
 240 0.0770  18.48  14.870  16.030
 241 0.0774  18.65  17.340  16.105
 242 0.0777  18.80  15.000  16.170
 243 0.0781  18.98  17.490  16.245
 244 0.0784  19.13  15.130  16.310
 245 0.0788  19.31  17.640  16.385
 246 0.0791  19.46  15.260  16.450
 247 0.0795  19.64  17.790  16.525
 248 0.0798  19.79  15.390  16.590
 249 0.0802  19.97  17.940  16.665
 250 0.0805  20.12  15.520  16.730
 251 0.0808  20.28  15.580  15.550
 252 0.0811  20.44  15.640  15.610
 253 0.0814  20.59  15.700  15.670
 254 0.0817  20.75  15.760  15.730
 255 0.0820  20.91  15.820  15.790
 256 0.0822  21.04  13.320  14.570
 257 0.0825  21.20  15.930  14.625
 258 0.0828  21.36  15.990  15.960
 259 0.0831  21.52  16.050  16.020
 260 0.0834  21.68  16.110  16.080
 261 0.0837  21.85  16.170  16.140
 262 0.0840  22.01  16.230  16.200
 263 0.0843  22.17  16.290  16.260
 264 0.0846  22.33  16.350  16.320
 265 0.0849  22.50  16.410  16.380
 266 0.0851  22.64  13.810  15.110
 267 0.0854  22.80  16.520  15.165
 268 0.0857  22.97  16.580  16.550
 269 0.0860  23.13  16.640  16.610
 270 0.0863  23.30  16.700  16.670
 271 0.0866  23.47  16.760  16.730
 272 0.0869  23.64  16.820  16.790
 273 0.0872  23.81  16.880  16.850
 274 0.0875  23.97  16.940  16.910
 275 0.0878  24.14  17.000  16.970
 276 0.0880  24.29  14.300  15.650
 277 0.0883  24.46  17.110  15.705
 278 0.0886  24.63  17.170  17.140
 279 0.0889  24.80  17.230  17.200
 280 0.0892  24.98  17.290  17.260
 281 0.0895  25.15  17.350  17.320
 282 0.0898  25.32  17.410  17.380
 283 0.0901  25.50  17.470  17.440
 284 0.0904  25.67  17.530  17.500
 285 0.0907  25.85  17.590  17.560
 286 0.0909  26.00  14.790  16.190
 287 0.0912  26.17  17.700  16.245
 288 0.0915  26.35  17.760  17.730
 289 0.0918  26.53  17.820  17.790
 290 0.0921  26.71  17.880  17.850
 291 0.0924  26.89  17.940  17.910
 292 0.0927  27.07  18.000  17.970
 293 0.0930  27.25  18.060  18.030
 294 0.0933  27.43  18.120  18.090
 295 0.0936  27.61  18.180  18.150
 296 0.0938  27.76  15.280  16.730
 297 0.0941  27.95  18.290  16.785
 298 0.0944  28.13  18.350  18.320
 299 0.0947  28.32  18.410  18.380
 300 0.0950  28.50  18.470  18.440
 301 0.0950  28.59   9.500  13.985
 302 0.0950  28.69   9.500   9.500
 ... (up to 400%)

Note how there's a huge hidden EMTR for people crossing the 133% FPL point in states that didn't extend Medicaid to the 138% level. Also how the EMTR actually rises slowly until it suddenly levels off at 300% FPL. And while not shown above, your next MAGI dollar over 400% of FPL has a virtually infinite EMTR as you lose your remaining PTC entirely!