I have a question about implementing tax credits (as a policy intervention) to promote low-carbon technologies over the conventional emission-intensive technologies. Our case study includes ammonia production industry with the aim of achieving net-zero emission by 2050 through implementing several low-carbon technologies (steam reforming+CCUS, water electrolysis, ...). As a relevant policy intervention, we want to evaluate the role of legislated tax credits to be applied to low-carbon technologies: this policy says facilities/technologies with emissions lower than 0.7 ton-CO2/ton-ammonia can be eligible for tax credits of 100$/ton-ammoniaup to 500$/ton-ammonia (depending on the level of their emissions). This tax credit can be applied only to facilities that begins construction before 2032 and it can be received for 10 years after the facility starts operation. To implement this tax credit in our model for low-carbon technologies, we have added the corresponding tax credits (defined in terms of $/t-ammonia) as variable operating costs but in the negative form (e.g., the tax credits of 100$/ton-ammonia is defined through considering the value of -100$/t for the VAROM entry of the target technology). In this way, the tax credits results in lower production cost for low-carbon technologies promoting their contribution compared to the case where not such a policy is implemented. However, we are not sure whether it is proper to define such tax credits as variable operating cost in the model. Does using this attribute sound reasonable for this purpose? Besides, we have difficulty in defining the conditions required for this policy which we would be grateful to have you advise on: How we can specify that this tax credits can last only for 10 years from the start of the technology operations and then will be removed for the rest of the technology's lifetime?
Your advise on these points would be highly appreciated. Thanks.
>Does using this attribute sound reasonable for this purpose?
Not ideal with respect to reasonableness in my view, due to the following:
• Policy mechanisms should be best be modeled as transparent add-on type scenarios
• A negative ACT_COST would easily be mixed up with any standard ACT_COST
• In the results, costs and fiscal policy measures should be easily distinguishable
Therefore, I would suggest using FLO_SUB instead, which defines a subsidy on a process flow. It has the following benefits:
• Transparency
• Easy add-on modeling of tax credits
• Easy shaping of tax credits according to age (e.g. credit only if age less than 11 years)
• No impact on model size (even if credits shaped), apart from the vintaging overhead
Shaping those tax credits is relatively easy: Just define a dummy commodity (say DUMSUB) of type ENV and with LimType N. Then define FLO_EMIS for this commodity from the ammonia output of the processes prc eligible for the credit, and define FLO_SUB(prc,DUMSUB)=100. Finally, define a SHAPE index N with SHAPE(N,1)=1; SHAPE(N,10)=1; and SHAPE(N,11)=0, and a FLO_FUNCX(r,y,prc,DUMSUB,DUMSUB) = N. That should work well, but note that those prc with shaping by FLO_FUNCX must be all defined vintaged. I just tested it, and it worked as expected.
03-03-2023, 05:26 AM (This post was last modified: 07-03-2023, 05:04 AM by jabarivelisdeh.)
Hi Antti,
Many thanks for your reply and the advices.
Exactly right, it is much more reasonable to try FLO_SUB instead of having tax credits as VAROM.
We tried to implement the tax credit based on the points you mentioned. However, it seems we did not define the tax credit properly, as the results is same as the case where no tax credit is considered for low-carbon technologies. We defined a dummy commodity for water electrolysis technology (WENH3_RNW) and defined FLO_EMIS and FLO_SUB for it accordingly, to be adjusted using SHAPE.
I would be grateful if you could have a look on the attached excel sheet to see whether we defined it correctly or not. Thanks a lot.
03-03-2023, 06:11 PM (This post was last modified: 03-03-2023, 06:25 PM by Antti-L.)
My corrections applied in Scen_TaxCredits:
• Leave a blank row/column around data tables, such as TFM_INS
• Define the dummy flow as a function of AMMONIA, with a shaped FLO_EMIS
• FLO_SUB expects a commodity, which must therefore be specified in CSet_CN
My corrections applied in VT_REG_PRI_V03:
• FI_Comm and FI_Process tables made contiguous (remove spurious blank columns)
• Row 67 removed (unnecessary there and FLO_EMIS incorrectly specified anyway)
• Column AF removed (unnecessary there and FLO_EMIS incorrectly specified anyway)
These were based on visual inspection (not tested as did not have e.g. SysSettings), and are in the attached.
TaxCredits.zip (Size: 151.34 KB / Downloads: 17)
Many thanks for your advises which solved the problem I had: The tax credit is now implemented for those low-carbon technologies eligible for it (water electrolysis with highest tax credit as 530 $/t-ammonia and CCUS-equipped technologies with a tax credit of 106 $/t-ammonia) for 10 years from starting their operation. However, we observed that the ammonia production in 2050 is much higher than the demand we specified for that year (we defined a constant demand for ammonia from 2020 to 2050), shown in the attached figure. I can track this behavior based on the levelised cost of water electrolysis technology (WENH3_RNW): the tax credit reduces the ammonia production cost of this technology to a negative value, therefore, more production (even exceeding the specified ammonia demand) cause less total system cost. I have added the corresponding DD and RUN files here: https://drive.google.com/file/d/1yjXWNsz...share_link
I would be grateful if you can have a look and let me know your thoughts on why this happens. Could we in that case constrain the total ammonia production not to exceed our specified ammonia demand? Besides, could we specify the tax credit to be applied only to technologies which are being built before 2032? as I think it may help in moderating the ammonia production costs by water electrolysis e.g. in 2045 and 2050 (right now the levelised cost of this technology is -353$/t in 2045 and -370$/t in 2050).
Thanks a lot.
04-03-2023, 10:57 PM (This post was last modified: 04-03-2023, 11:00 PM by Antti-L.)
Yes, you are running into another kind of problem. The tax credit is so high for WENH3_RNW that the model would actually be unbounded (because you could make infinite profits at government's expense by producing ammonia), if you had allowed the variable cost component of the objective function to become negative. (I just verified the model becoming unbounded.)
Fortunately (in this case), you had not allowed the variable cost component to become negative, because then you would have not obtained any solution. However, note that with the model you provided, the marginal price of your imported ELCRNWBL is only 2.1 USD(2020)/GJ in 2050, i.e. much lower than the exogenous import price. That is obviously caused by the model marginals being all distorted by the bounded variable costs. Likewise, the levelised costs are not really consistent in this case, because they are based on those distorted prices.
>Could we in that case constrain the total ammonia production not to exceed our specified ammonia demand?
Yes of course you could constrain the total ammonia production, if you like (e.g. with COM_BNDPRD).
>Besides, could we specify the tax credit to be applied only to technologies which are being built before 2032?
Yes, I think that would be reasonable. You can easily do that with the FLO_EMIS attribute. According to my quick test, the results will then indeed be much more reasonable (no unbounded profits), and no constraint on the ammonia production would be needed.
05-03-2023, 08:21 PM (This post was last modified: 05-03-2023, 08:52 PM by Antti-L.)
As indicated in my post above, the total discounted variable costs of the model (OBJVAR) are by default bounded to be non-negative. If, however, you would like to allow that the tax credits can exceed all the variable costs in the model, you can easily change that default and allow OBJVAR to take also any negative values. That can be done with the following setting: UC_RHS('OBJVAR','N')=-1;
See the following thread for an example, defining it in a TFM_INS: Export and FLO_SUB
But bear in mind that doing so may in some cases lead to an unbounded model, as demonstrated by your model, where the high tax credits were exceeding even the full production costs by 2050.
Thank you so much for your answers, Antti! Right, I think you exactly addressed the point which caused the problem, as I was not aware that variable costs can not be negative, unless we change the default of OBJVAR to take negative values. Super helpful advise, indeed!
Just one quick clarification about this suggestion you made: > Besides, could we specify the tax credit to be applied only to technologies which are being built before 2032? Yes, I think that would be reasonable. You can easily do that with the FLO_EMIS attribute. According to my quick test, the results will then indeed be much more reasonable (no unbounded profits), and no constraint on the ammonia production would be needed.
As mentioned, only those low-carbon based facilities who begin operations before 2032 can get this tax credits, applicable for 10 years. This means a plant which is for example built in 2030 can get this credit until 2040. however, I think what you suggested for modifying FLO_EMIS (to turn to 0 from 2035) in fact deactivates the tax credit implementation (which we defined already by the SHAPE attribute) in year 2035 while the example plant is eligible to receive it until 2040, right? Or am I missing something? Based on your suggestion, similarly one can define the FLO_EMIS attribute as 1 until 2041 and then switching that to 0 in 2042 (2032+10 years) to deactivate this tax credit at a certain time point. However, in that case a plant might be built in 2035 but still get the tax credit until 2042 (based on the defined FLO_EMIS values), while in reality it is not eligible to receive it (due to the starting year of after 2032). Thanks a lot, I hope I could explain my concern clearly!
06-03-2023, 03:12 PM (This post was last modified: 06-03-2023, 03:40 PM by Antti-L.)
>I think what you suggested for modifying FLO_EMIS (to turn to 0 from 2035) in fact deactivates the tax credit implementation (which we defined already by the SHAPE attribute) in year 2035 while the example plant is eligible to receive it until 2040, right? Or am I missing something?
No, it does not deactivate the tax credit implementation. It does what you wanted:
It specifies the tax credit to be applied only to technologies which are being built before 2035 (or a bit more accurately, those of having vintage of earlier than 2035). So, the last plant vintage eligible would be 2030, and because the tax credit is paid only 10 years, it would no longer be paid in 2040, because the vintage 2030 actually starts operation already in 2028. So, it may be that you would rather wish to shift that turning of FLO_EMIS to zero 5 years forward. But anyway, when the process is vintaged, the years in FLO_EMIS refer to vintages, and not operating years.
>Based on your suggestion, similarly one can define the FLO_EMIS attribute as 1 until 2041 and then switching that to 0 in 2042 (2032+10 years) to deactivate this tax credit at a certain time point. However, in that case a plant might be built in 2035 but still get the tax credit until 2042 (based on the defined FLO_EMIS values), while in reality it is not eligible to receive it (due to the starting year of after 2032).
Yes, right, if you want that plants starting after 2032 should no longer be eligible, you should not define the FLO_EMIS attribute as 1 until 2041, because doing so would give any plants built in 2035 the tax credit for 10 years of their operation, like you prescribed earlier. Sorry, but I don't fully see what your concern is here.
Regarding in implementing the tax credits, I have a question based on our results which I would be grateful to have your opinion on: although we know that water electrolysis is built in 2031 (which makes it eligible to receive the credit until 2040), the results indicate no tax credits in 2040. In fact, although based on defined the vintage constraint we assume that the plants being built before 2032 are eligible for the credit, we see that only plants built by 2030 could get the credit. And when we extend the vintage year up to 2034 (such that plants being built by 2034 can get the credits) still we see that no credits is being dedicated to plants which are built from 2031 to 2034 (although we believe that they are eligible to receive it). In contrast, if we extend the vintage year further to 2035, the results show that credits are dedicated to the plants being built by that time.
So my question is that: does it mean that our vintage constraint is only applied to milestone years? The file is attached here: https://drive.google.com/file/d/1yjXWNsz...share_link
05-04-2023, 03:31 AM (This post was last modified: 05-04-2023, 03:36 AM by Antti-L.)
Ok I see your dilemma. You have defined tax credits for the 10 first years of operation, and for vintages up to 2031 only. And yes, new capacity vintages are limited to the milestone years, and so you have chosen to restrict your new vintages to {2020,2021,2025,2030,2035,2040,2045,2050}, while it seems you would like to have the vintage 2031 distinguished in the results. However, as your model is now, there is no water electrolysis capacity of vintage 2031 in the results of your model. I think the vintage indexes are quite clearly shown in the results, and 2031 is not included.
The almost sole main purpose of using such model periods each representing several years, instead of modeling all individual years within the model horizon, is exactly this: to reduce the size of the model when the model is getting large. The number of equations and variables in the model increase directly along with the number of milestone years. But as your model is quite small, in your case you could well use single-year periods if you need results for individual years (and vintages), and you thus need more variables. But for now, you have deliberately reduced the vintage-specific variables to the periods you have chosen. Anyway, TIMES has been designed to be flexible in period definitions, and in such a way that the impacts of different period definitions on the model solution should, in general, remain small (apart from cases where you explicitly want sudden changes at individual years, like you seem to want here).
I hope that may to some extent answer your question?
05-04-2023, 03:38 PM (This post was last modified: 06-04-2023, 03:13 PM by Antti-L.)
You could, of course, also consider an approximate formulation, where the the tax credits for the investments in 2031 are approximated according to the five-year period lengths. It would thus mean that after the 2030 investments (which are granted the full tax credit), investment in one more year (2031) would be granted the credits, thus corresponding to one fifth (0.2) of the total investments of the vintage 2035 (which includes the 2031 investments). You could easily use this kind of an approximate approach by defining the value 0.2 for the FLO_EMIS that defines the vintages eligible for the tax credit:
While doing it on my own, I ran into an issue that PTC won't expire after a set period of time. Based on the example in current thread, I realised that this was due to the I/E rule. Tbh, I am a bit confused regarding application of the I/E rules to SHAPE. As far as I can see in the docs PART II (page 40, table 9), option 5 shouldn't have worked?
@Antti, any chance you could provide some feedback on slight differences in my approach:
- FIN instead of ENV for the PTC commodity
- didn't use LimType N
- actual amount of PTC as FLO_EMIS
- FLO_SUB = 1
To get this working I also had to use a TFM_TOPINS table to add the PTC commodity to the relevant processes. I guess, this is because ENV and FIN commodities are treated differently by Veda?
