Announcement

Collapse
No announcement yet.

San Diego Historical Solar Pricing

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • San Diego Historical Solar Pricing

    I captured the following stats about San Diego county solar pricing from the NEM data set for my own purposes earlier. Hopefully others will find this type of stat interesting like I do. The entries from 2015 to present are from the NEM database. I only included residential rooftop installs, without loans/PPA/Pace, and not self installs. I also did not include some entries that were incomplete or appeared to be errored. The entries prior to 2015 are from the CSI database. The CSI database has more limited filtering options, so I was not able to exclude all of the same categories as the 2015+ NEM entries. I expect this relates to why the change in trend occurs between 2014 CSI and 2015 NEM. Cost per DC Watt appears to have had a sharp drop from 2008 to 2013, dropping by roughly 10% per year. The drop in price appears to have slowed down since then, particularly since 2017, as summarized below.

    2015 -> 2016 -- ~6% drop in cost per watt DC
    2016 -> 2017 -- ~6% drop in cost per watt DC
    2017 -> 2018 -- ~2% drop in cost per watt DC
    2018 -> 2019 -- ~2% drop in cost per watt DC
    .
    Date Cost / DC Watt: 3 to 5 KW DC Cost / DC Watt: 5 to 10 KW DC
    25th Median 75th 25th Median 75th
    2019: Q2 3.42 3.93 4.40 3.20 3.61 4.25
    2019: Q1 3.47 4.02 4.62 3.28 3.71 4.32
    2018: Q4 3.44 4.02 4.85 3.26 3.69 4.27
    2018: Q3 3.43 4.20 5.12 3.26 3.72 4.59
    2018: Q2 3.50 4.25 5.12 3.32 3.86 4.68
    2018: Q1 3.33 4.06 5.12 3.23 3.74 4.65
    2017: Q4 3.39 4.06 5.10 3.19 3.72 4.33
    2017: Q3 3.45 4.14 5.01 3.22 3.71 4.30
    2017: Q2 3.55 4.21 5.12 3.27 3.87 4.69
    2017: Q1 3.70 4.65 5.12 3.29 3.97 5.03
    2016: Q4 3.71 4.31 5.12 3.35 3.95 4.69
    2016: Q3 3.76 4.32 5.12 3.54 4.02 4.66
    2016: Q2 3.92 4.46 5.09 3.68 4.05 4.64
    2016: Q1 3.95 4.50 5.12 3.70 4.24 4.77
    2015: Q4 4.00 4.53 5.12 3.76 4.27 4.89
    2015: Q3 4.13 4.76 5.12 3.85 4.36 5.00
    2015: Q2 4.20 4.81 5.12 3.91 4.39 4.93
    ---- ---- ---- ---- ---- ---- ----
    2014: CSI 3.91 4.46 4.95 3.65 4.05 4.69
    2013: CSI 4.04 4.59 5.34 3.76 4.23 4.79
    2012: CSI 4.52 5.20 6.32 4.13 4.66 5.38
    2011: CSI 5.65 6.58 8.23 5.10 5.74 6.66
    2010: CSI 5.80 6.51 7.32 5.43 6.02 6.72
    2009: CSI 6.59 7.24 8.02 6.24 6.91 7.47
    2008: CSI 7.39 7.86 8.40 7.08 7.60 8.20

  • #2
    I've done similar analysis w/ somewhat different criteria, but the results look similar, particularly the price declines.

    Any max/min, std. deviation information for any of your data ?

    The installs in my HOA generally ran/run about 6 - 8 kW and, according to the database, about 5-10 % more costly/STC W in the same zip (92026) since 2008. The few who take my advice and use the database seem to have done better on price, but I've got no data to back that up.

    The NEM, or the old CSI database is a very valuable tool for price comparison. There's no B.S., and the database is full of exactly the type of information needed for pricing information for comparison purposes, and exactly the stuff vendors would rather not have public. You get to see exactly what folks bought and exactly what they paid. Probably one reason why vendors never talk about it. When I designed my system, it helped me know exactly where I could go in terms of price with no guesswork.

    Comment


    • #3
      Any max/min, std. deviation information for any of your data ?
      Max and min are difficult to confirm because the extremes usually involve incorrect entries and/or non-standard arrangements. For example, quite a few NEM entries list a net cost of under $50. so they have a very low cost per DC watt. This includes installs from some reputable companies like Milholland. Maybe they are listing system cost in thousands of dollars, rather than dollars, so a $18 = $18,000? It's hard to say for sure. I excluded the min and max extremes for this reason. If I limit the min and max to system cost is > $1000, DC provided by 1st group of PV panels matches listed DC watts, was not obvious power of 10 typo, used an installer who I believe is likely to enter things correctly in the database, and with the filters previously mentioned above; then the min and max I see are:

      Min -- 6.87 KW DC for $4,777; Baker Electric install in Chula Vista of 21 Sunpower 327NE (lowest 4 Bakers all used Sunpower panels... perhaps customer provided panels)

      Max -- 0.36 KW DC for $11,966; Baker Electric install in Del Mar of 1 LG 360Q1C

      The incorrect entry outliers also make standard deviation misleading, which relates to why I chose to list percentiles above. One can crudely estimate SD without the errored outliers as (75th percentile - 25th percentile)/1.35.

      The NEM, or the old CSI database is a very valuable tool for price comparison. There's no B.S., and the database is full of exactly the type of information needed for pricing information for comparison purposes, and exactly the stuff vendors would rather not have public. You get to see exactly what folks bought and exactly what they paid. Probably one reason why vendors never talk about it. When I designed my system, it helped me know exactly where I could go in terms of price with no guesswork.
      I agree. I found the database useful, and the quotes I initially received earlier this year generally fit well with the database. It was also helpful in confirming some non-obvious trends in prices of specific installers. For example, the installer I selected appears to have different default cost/watt after crossing thresholds in number of panels. 11 panels averages much lower cost/watt than 10, but 10 does not show much difference from 9. Other installers appear to have a standard cost/watt rate that occurs in the bulk of their installs (with a specific type of panel), regardless of number of panels.
      Last edited by NCSD; 09-05-2019, 05:47 PM.

      Comment


      • #4
        Originally posted by NCSD View Post
        Max and min are difficult to confirm because the extremes usually involve incorrect entries and/or non-standard arrangements. For example, quite a few NEM entries list a net cost of under $50. so they have a very low cost per DC watt. This includes installs from some reputable companies like Milholland. Maybe they are listing system cost in thousands of dollars, rather than dollars, so a $18 = $18,000? It's hard to say for sure. I excluded the min and max extremes for this reason. If I limit the min and max to system cost is > $1000, DC provided by 1st group of PV panels matches listed DC watts, was not obvious power of 10 typo, used an installer who I believe is likely to enter things correctly in the database, and with the filters previously mentioned above; then the min and max I see are:

        Min -- 6.87 KW DC for $4,777; Baker Electric install in Chula Vista of 21 Sunpower 327NE (lowest 4 Bakers all used Sunpower panels... perhaps customer provided panels)

        Max -- 0.36 KW DC for $11,966; Baker Electric install in Del Mar of 1 LG 360Q1C

        The incorrect entry outliers also make standard deviation misleading, which relates to why I chose to list percentiles above. One can crudely estimate SD without the errored outliers as (75th percentile - 25th percentile)/1.35.


        I agree. I found the database useful, and the quotes I initially received earlier this year generally fit well with the database. It was also helpful in confirming some non-obvious trends in prices of specific installers. For example, the installer I selected appears to have different default cost/watt after crossing thresholds in number of panels. 11 panels averages much lower cost/watt than 10, but 10 does not show much difference from 9. Other installers appear to have a standard cost/watt rate that occurs in the bulk of their installs (with a specific type of panel), regardless of number of panels.
        Understood. Thank you. I had luck and what looks like good results for making sense of CSI/NEM data combing the database(s) for obvious errors/outliers, particularly when first sorting by panel and then by vendor after eliminating self installs. After a while, it gets a bit like herding cats. I've been away from it for about a year now, but what I've done with the databases in the past seems a lot like what you present here.

        FWIW, in addition to a general analysis of all the SDG & E stuff I looked at, I found every install in my HOA (~ 150 and counting) and that database information matched what I saw every time as the guy in on the Arch. Rev. Comm. that reviews/recommends to the ARC and then monitors construc. for conformance to the CC & R's. By that experience I'm of the opinion the data is reasonably reliable. There's so much specific data avail. for each install that I had no trouble ID'ing every job in my HOA in less that 5 min.

        Comment

        Working...
        X