Reserve Studies - A Policy Analysis

Twelve states require community associations, such as homeowners associations or HOAs, to conduct periodic reserve studies. A reserve study is a financial planning tool for estimating an association’s ability to fund current and future planned maintenance costs for large projects, such as roofs, pools, and asphalt roads. An underfunded reserve may lead to special assessments. Yet, due to restrictions on how reserve accounts can be used, an over funded account denies funds that may be better used by the community or the individual homeowners. In this post, I describe the reserve study methodology, its strengths and weaknesses, and my recommendations for improvements.

Tree roots damaging sidewalk because the trees were misplanted
Sidewalk damage from tree roots

Conducting a Reserve Study

Community associations have used a variety of methods to maintain shared assets. In 1964, the Federal Housing Administration funded the Urban Land Institute to study the organization and running of American community associations. As documented in section 14.52 of (ULI, 1964), about half of the surveyed associations held “reserves,” financial accounts dedicated to cover the depreciation and replacement cost of certain assets1. Homeowners would pay into these reserves every year as part of their assessment fees, and the associations would draw upon the account to pay for major projects.

The study found that some associations used a “special assessment” mechanism as an alternative, or at least a complement, to a reserve account. When the organization needed large amounts of funds to replace an asset, the homeowners could vote to oblige everyone to pay a special, single year fee. Although the study did not advocate for special assessments becoming common, it did advocate for the option to be available and included special assessments in its model covenant agreements.

Given the existence of a reserve account, how much money should go into it? Although (ULI, 1964) does not mention a reserve study per se, it does describe a model based on depreciation from accounting. The study uses the example of a tractor that needs periodic replacement. The cost for a replacement is spread evenly (“straight line”) over the estimated useful life. By changing the term from depreciation to deterioration, which removes tax law quirks, we get an inventory and one of the financial models used in a reserve study.

A basic reserve study consists of:

Periodically, after the inventory is formed, the practitioner conducting the study will visually inspect components within the inventory and update the remaining life. Component’s life and costs are tabulated on a yearly basis and, based on inflation, interest, and reserve contributions, the yearly balances of the reserve fund will be estimated.

Variation and Standardization

Founded in 1973, the Community Associates Institute (CAI) provides and develops resources for community association management. In 1998, they published the Reserve Study (RS) Standards to establish common language and methodology for the practice. Updated in May 2023, these standards are influential but do not bear legal weight in most states. Other professional organizations, such as the Association of Professional Reserve Analysts and International Capital Budgeting Institute have developed their own, parallel standards.

Statutory requirements for reserves and reserve studies differ state by state (Florida, for example, in 2022 added a structural integrity reserve study), but they share a common approach, often inspired by the CAI standard. Utah, for example, first adopted laws requiring HOAs to conduct a reserve analysis in 2010, but left open the contents of that analysis. (The bill originally called for a ‘reserve study,’ but the language was changed to ‘reserve analysis’ in committee.) In 2021, Utah updated the law and specified the contents of the reserve analysis, paralleling the CAI definition.

As another example of a state law, Nevada’s Reserve Study Requirements (NRS 116.31152) are:

  1. The study of the reserves must include, without limitation: (a) A summary of an inspection of the major components of the common elements and any other portion of the common-interest community that the association is obligated to maintain, repair, replace or restore; (b) An identification of the major components of the common elements and any other portion of the common-interest community that the association is obligated to maintain, repair, replace or restore which have a remaining useful life of less than 30 years; (c) An estimate of the remaining useful life of each major component of the common elements and any other portion of the common-interest community that the association is obligated to maintain, repair, replace or restore identified pursuant to paragraph (b); (d) An estimate of the cost of maintenance, repair, replacement or restoration of each major component of the common elements and any other portion of the common-interest community identified pursuant to paragraph (b) during and at the end of its useful life; and (e) An estimate of the total annual assessment that may be necessary to cover the cost of maintaining, repairing, replacement or restoration of the major components of the common elements and any other portion of the common-interest community identified pursuant to paragraph (b), after subtracting the reserves of the association as of the date of the study, and an estimate of the funding plan that may be necessary to provide adequate funding for the required reserves.

Note that the Nevada law does not require any specific methodology for how estimates are obtained nor how they are calculated. While not specifying a specific methodology per se, a recent 2024 update to Tennessee law, certainly treats the CAI as a de jure standard by defining a reserve study as:

“Reserve study” means an analysis, prepared in conformity with the latest edition of the Reserve Study Standards published by the Community Associations Institute, or similar standards by another nationally recognized organization, by a reserve specialist who is credentialed through the Community Associations Institute or a similarly recognized organization, or a licensed engineer or architect,

While most states allow anyone to conduct a reserve study, including the community board, some states apply restrictions. Florida requires a licensed engineer or architect to conduct the visual inspections of the components, but does not require a license for the other sections. Hawaii requires the study to be reviewed by an “independent reserve study preparer.” Maryland requires the preparer to either be:

  1. A preparer of at least 30 reserve studies within the past 3 years,
  2. A participant in the preparation of at least 30 reserve studies within the past 3 years while employed by a firm that prepares reserve studies,
  3. Holds a license with the State Board of Architects or State Board for Professional Engineers, or
  4. Is a designated reserve specialist by the Community Association Institute or professional reserve analyst by the Association of Professional Reserve Analysts.

To be a designated reserve specialist, one needs three years of practice preparing at least 30 reserve studies, a bachelors degree, and the payment of an application fee plus annual renewal fees. A professional reserve analyst requires five years of experience, at least fifty reserve studies, a membership fee, plus eight hours of annual continuing education.

An Example

If you are interested in playing with this model or simulating your own inventory, our simulation tool is available online. To keep costs low, the hardware is limited. Running the model will take at least twenty seconds to complete. Larger models may take a couple minutes to complete. Serious use cases should fork the source code and run on their own hardware. Locally, models can complete in a few seconds.

Below is an example inventory, used as an example by CAI. UL is the Useful Life of each component and RUL is the Remaining Useful Life, both in years. A RUL of zero indicates the need for replacement this year. Costs are in USD and “present dollars.” We’ve extended the model by adding lower and upper bounds for the cost of each project/component. (We’ve kept RULs constant, but our tool allows low and high RUL values.)

Component UL RUL Cost_Original Cost_Low Cost_High
Pool Furniture - Replace 5 0 4600 3680 5520
Pool - Resurface 10 5 10000 8000 12000
Roof - Replace 20 18 80000 64000 96000
Asphalt - Seal 5 2 5000 4000 6000
Asphalt - Resurface 20 2 25000 20000 30000
Building - Repaint 10 1 50000 40000 60000
Elevator - Modernize 20 5 80000 64000 96000
Hallways - Refurbish 8 6 24000 19200 28800

If we simulate this model over 30 years, assuming an inflation rate between 2% and 4%, we can plot the expenses per year using a box plot:

Expenses per year, showing many spikes
Simulated Expenses Per Year

Although expenses may seem rather bumpy in this model, it is likely smoother than real inventories from medium and larger sized associations. This model also lacks the clutter from the many smaller projects that populate a real association’s inventory.

In order to meet current and future costs, associations are expected to fund reserves to meet cash flow needs while keeping assessments stable (e.g. increasing with inflation). The standards refer to a Fully Funded Balance (FFB) as an indicator of a conservative and healthy level of funding. For each component in the inventory, the FFB is calculated as:

$$ FFB = C \frac{a}{u} $$

where C is the replacement cost, a is the effective age, and u is the useful life. FFB assumes a linear degradation of the asset, until the remaining life hits zero, at which point the full replacement cost is realized. For a site, the aggregate FFB is the sum of the FFB for each component. If an association’s reserve balance is equal to the FFB (or within a few percentage points), then the balance is considered “fully funded”.

Continuing the example, we compute the (most likely) expenses per year and plot the FFB at the beginning and end of the year for a ten year span. The contribution is the amount necessary to fill the gap.

Category Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10
FFB (Initial) 154.1 174.8 148.4 142.1 169.0 197.5 114.5 113.3 135.9 166.3
Expenses 4.6 51.5 31.8 0.0 0.0 109.7 28.6 6.1 0.0 0.0
Contribution 25.3 25.1 25.6 27.0 28.5 26.7 27.3 28.8 30.4 32.1
FFB (EoY) 174.8 148.4 142.1 169.0 197.5 114.5 113.3 135.9 166.3 198.4

Although expenses are bumpy (e.g. no expenses in years 4 and 5), contributions are much smoother because they are the aggregate of many individual “glide slopes”.

This table assumes the association is already fully funding their fund. A practitioner would need to examine the association’s current funding and contributions, as well as any existing deferred maintenance or debts in order to complete a financial analysis. Since an association may not start their fund fully funded, a practitioner would provide advice on how to smoothly escalate contributions over time while avoiding a negative reserve balance.

Strengths of the Methodology

Annually, a community uses a reserve study for two primary purposes:

  1. Decide how much budget to contribute to the reserve fund, and
  2. Decide what components/projects to fund this year

The first purpose is covered by the financial plan of the study, while the second purpose is handled via the inventory. In some states like Florida there may also be a safety inspection, but we will ignore that in this analysis.

Easy to implement, audit, and understand

The reserve study methodology is freely available2 and, except for the remaining life estimates, requires no skills beyond high school math. The primary output of the study is a table that can be easily audited by walking around the community and useful life numbers can be compared (or retrieved) from publicly available sources or references like the The Whitestone Facility Maintenance and Repair Cost Reference 2009-2010.

Estimating the remaining life of components requires specialized knowledge and is sometimes called an art rather than a science. However, these estimates simply populate a column in the table and most estimates are not safety-critical. With further development of video-based prognostic systems, we can expect computer-aided decision tools and further automation of the process.


The community board can easily switch to a different provider of the study. As a work for hire, the results of previous studies are owned by the community association and can be used to ‘jump start’ a new study. Vendor lock-in is difficult because the study does not include ‘black box’ or proprietary features, although a study may include ‘value add’ elements.

Based on member directories for the professional organizations and listed companies by internet search, this appears to be a competitive market, although one with limited price transparency. In most states, associations are free to balance vendors' prices, experience, certifications, and other factors. States with licensing requirements or considering measures to restrict trade should carefully consider the evidence for the claim that restrictions promote quality and safety. The report Occupational Licensing: A Framework for Policymakers is a good resource, detailing the costs and benefits of licensure and the many policy alternatives.

Reduced need of forecast accuracy via inexpensive updates

Reserve studies need to be updated periodically and some states mandate an update frequency, typically from one to six years. Studies have different “levels” of detail and updates in many years may require less effort. Due to these updates, uncertainties in the model are less critical because there are more opportunities for review and revision, particularly as components approach their end of useful life and estimates become more accurate.


Although reserve studies are claimed as a best practice (CAI, 2020), we have failed to find any study of their efficacy. This weakness was identified in (Winkour, 1998) and stems from a lack of public data. Our preference would be an assessment of their contribution to efficient fiscal management, but lacking data for such a study, our criticisms are based on our experiences in modeling and simulation, rather than historical analysis or comparative analysis of outcomes.

In terms of efficacy, associations have voiced their desires in various surveys. A recent representative one is documented in (Foundation, 2020):

“More than three-quarters (80%) of those surveyed felt it was critical that their association have adequate reserves in the event of a major infrastructure failure or construction need. Nearly half (40%) of those surveyed considered deteriorating infrastructure as a top-ranked concern. More than two-thirds (70%) of survey respondents indicated that maintaining property values was of primary importance.”

Restated, the desired outcomes of the reserve study process are:

  1. Protection of property values, by avoiding deferred maintenance or a weak fiscal position,
  2. Avoidance of special assessments3, and
  3. Discovery, prioritization, and remediation of imminent infrastructure failures and decay.

An additional outcome in some states is fulfilling regulatory reporting requirements, but paperwork is not a good by itself.

We identify four weaknesses in how the methodology addresses these key desired outcomes. They are:

  1. Lacks a top-down measure or “smell test”
  2. Fails to model risk
  3. Lacks sensitivity analysis
  4. Fails to stress need to mitigate inflation

Lacks a top-down measure or “smell test”

Reserve studies use a purely bottom-up approach. By bottom-up, we mean that the practitioner catalogs individual items and then sums them up to estimate the overall deterioration. While this is a detail-oriented approach, there is always a methodological risk that a component may be forgotten or underestimated. In contrast, a top-down approach will start at the community and make estimates from the top, often parameterised by broad questions such as “does the community have a pool?”. The top-down estimate may have greater uncertainty and will lack the detailed inventory to satisfy the year-by-year maintenance cycle, but it can be directly compared to other communities to ferret out gross inaccuracies or signal that a community is out-of-line with its peers.

We are not advocating the abandonment of the bottom-up inventory approach for a top-down approach or model, but rather that they should be used together. With two measurements, an association will have better confidence that an analysis is complete and valid. For an example of an existing top-down model, (Whitestone, 2009) includes cost profiles for a variety of building types and typical 50 year maintenance and repair costs and tasks which can be used as benchmarks.

In (CAI, 2020), they state “THE MOST CRITICAL decision in conducting a reserve study is selecting which components to include. This requires analysis of the association’s governing documents, application of the National Reserve Study Standards four-part test, and applicable state statutes and civil codes in addition to the physical inspection of each component.” The addition of top-down methods will diminish the criticality of the inventory selection.

Fails to model risk

One may be confused by the claim that reserve studies do not model risk. After all, (CAI, 2023) speaks of three funding goals and their respective levels of risk:

  1. Baseline Funding - meet immediate cash flow requirements and minimize reserve contributions; highest risk
  2. Threshold Funding - maintain a reserve balance above a specified percentage or dollar amount
  3. Full Funding - maintain a reserve balance within a few percent of the fully-funded balance (FFB); lowest risk

Aside: Contrary to their definitions, threshold funding is really two separate goals. A minimum dollar balance can have a basis as a true risk model, while, as we argue below, a goal based on a percentage of the FFB is not a true risk model but rather an acknowledgement that associations can use cash flow to their advantage.

However, the standards do not define risk nor attempt to quantify it. In this section, we will define risk and uncertainty using generally accepted definitions, describe how we might translate FFB into risk terms and why we disagree with that approach, and then ways we can enhance the method to manage risk.

Before going into the reasons why the reserve model is not a risk model, we need to ask why should an association care? An association is responsible for all community expenses, but the CAI definition of the reserve study inventory, and some states, is a subset. In the CAI’s three-part test (CAI, 2023), a project is only included in the study if it can be “reasonably anticipated” and the cost “reasonably estimated.” While this restriction limits the scope of the inventory to a more grounded list, it fails to account for realistic obligations of the association. Even when a project can be reasonably anticipated and the cost estimated, the execution of projects has risks and uncertainties that the literature does not address. Thus, reserve studies will underestimate financial costs.

In our opinion, ideally, a reserve study would inform the association of known planned maintenance items, unplanned maintenance items that are likely to occur and contingencies (i.e. risks), and the ranges of variables (i.e. uncertainty). The methodology, however, often restricts the scope to just the known planned maintenance items. A restriction in scope can make sense if reserve studies are part of a portfolio of tools, but if a portfolio is the intent, the literature and legislation are oddly silent on the issue.

How do we define risks and uncertainty? First, risks and uncertainty are often conflated, but they are distinct concepts. A risk is a potential event that, if it happens, will have negative impact (e.g. cost the association money). Uncertainty is the range (or distribution) of possible values for some variable. A component requiring highly premature replacements due to contractor negligence is a risk. The range of potential bids for a job is an uncertainty. Inflation is not a risk as it will happen, but the magnitude of inflation is uncertain 4. A special assessment is not a risk; it is a consequence of insufficient budget.

If we sum all the risks (their probabilities times their magnitudes), we will arrive at a dollar value that expresses the probabilistic cost to the association. An association can bear this risk as part of their reserve, part of their operational budget, or as a potential cost to homeowners and collected as a special assessment.

Risk Mitigation and Threshold Funding

Central to CAI’s claim is that Threshold Funding, based on a percentage of the FFB, provides a proportional amount of “risk management.” As described above, the FFB for a site is the aggregate of the FFB for every component. Let’s explore this concept with a simple model of three components with zero uncertainty:

Component Repl. Cost Degradation %
C1 $5 10%
C2 $3 50%
C3 $7 80%

The FFB for C1 is \(5 * 0.1 = 0.5\), C2 is \(3 * 0.5 = 1.5\), and C3 is \(7 * 0.8 = 5.6\). Summed, the FFB is $7.60.

Within the coming year, we can compute the probabilities for each to fail (using the degradation percentages as probabilities and assuming failures are independent), the summed magnitude (cost of replacement), and the weighted magnitude (cost of replacement times the probability).

C1 Fail? C2 Fail? C3 Fail? Prob. Magnitude Weighted M
0 0 0 0.09 0 0
0 0 1 0.36 7 2.52
0 1 0 0.09 3 0.27
0 1 1 0.36 10 3.60
1 0 0 0.01 5 0.05
1 0 1 0.04 12 0.48
1 1 0 0.01 8 0.08
1 1 1 0.04 15 0.60

If we sum the weighted magnitude, we get $7.60, which is equal to the FFB. Thus, the FFB is the weighted probability of replacing the entire inventory. (This assumes there are no out-of-inventory costs and all projects complete within their allotted budget.)

Let’s now sort the probabilities in declining order to arrive at a cumulative distribution function. Because some components are nearing end-of-life and they are more degraded than not, the lowest probability items do not correspond to the highest magnitude items.

Prob. Magnitude Weighted M Cum. P Cum. M
0.36 7 2.52 0.36 2.52
0.36 10 3.6 0.72 6.12
0.09 0 0 0.81 6.12
0.09 3 0.27 0.9 6.39
0.04 12 0.48 0.94 6.87
0.04 15 0.6 0.98 7.47
0.01 5 0.05 0.99 7.52
0.01 8 0.08 1 7.6

As it is unlikely for many assets to fail ahead of schedule (as a reminder, a site-wide disaster would be covered by insurance, not the reserve fund), we can capture 95% of the probability range with $6.87, rather than the full $7.60. Since $6.87 is 90% of the money but is yielding 95% of the coverage, this seems we are ahead and accepting that 5% “risk” is rational.

However, this logic is fallacious. First, if we simplify the model to a single component, or make one component significantly larger than others (e.g. roofs versus patio furniture), there will be a short-fall. This is because the approach is really a model of cash flow; having multiple independent inflows can “free up” enough cash to cover a small number of outflows. Depending on the composition of the inventory, a skillful association can make this approach work for a long time, but it is not robust.

Secondly, note that of the eight results in the table, four of them have magnitudes greater than $7.60, the FFB amount. Realizing any of these scenarios would exhaust the reserve fund. If the association reserved an additional $0.40, then it would cover one more potential result. Ideally the method would provide a rational basis to balance the possibility of a special assessment versus the costs of saving the extra $0.40 or $2.40.

Third, the approach is a “closed universe” as it does not seek to cover risks. For example:

  1. Methodological risk - a component is missed from the inventory, but must be covered by the association.
  2. Infrastructure risk - a component prematurely needs repair or replacement
  3. Project risk - the execution of a repair or replacement causes other damage or needs to be redone by other vendor due to quality issues (which are not fully mitigated by contractual terms)
  4. Inflation / Exterior Factor risk - availability or cost of material soars beyond that captured in the cost uncertainty range due to supply chain issues
  5. Uncertainty Methodology - If the low and high marks for a cost represent two-sigmas of the range, then 5% of the costs will be outside the given range

Alternative - Value at Risk

Value at Risk (VaR) is a commonly used, although controversial, risk framework. VaR estimates the risk of loss of capital, under normal conditions within a set amount of time. For instance, looking at the historical movement of a stock, a practitioner might say that 95% of the time, the loss is 4.6% or less on a daily basis, so over the next day, we can expect with 95% likelihood the stock’s value at risk is the (current value) times 4.6% or less. The amount of potential loss will inform how much money needs to be in reserve or hedged.

If we use this framework and lack historical data, we can try a projection instead. The value is the association’s property value. As a deteriorating asset, the expected minimal loss is equal to the FFB for the year. However, there are additional uncertainties and potential risks that may incur additional expenses that are not mitigated by insurance.

Based on the inventory, we know the expected costs for each component in nominal dollars. Sorting the components in order by cost, an association can pick the n-th percentile cost from the list. An association is unlikely to choose an outlier, but rather a higher-end yet typical project. The association adds the cost for this project as an additional buffer to the reserve fund. Thus, the desired budget is not the FFB, but the FFB plus this n-th project cost. The sum is the value-at-risk.

The effect is that the budget accounts for all planned projects plus one unexpected but typical-sized project. An association will still likely need a special assessment if a large capital project becomes urgent several years ahead of schedule or similar high magnitude item. Unfortunately, the more long-tail the component costs are, the less amenable to straight-forward risk management approaches.

Alternative - Double Declining Balance / Non-Linear Deterioration

If an association is mostly concerned about uncertainties in component’s remaining life, one possible technique lifted from accounting is “double declining balance.” Rather than modeling deterioration as a linear function, this model front loads the deterioration. If the component ends up needing replacement early, the impact will be less because more has been paid up-front. For a component with a five year useful life and a replacement cost of $1000, the impact of replacing at four years or one year early would be $200, but about $50 in the double-declining model. Other non-linear models are certainly possible, although an association might only feel it useful for particularly high cost components. Residents may also object as the burden of payment is shifted to those with more time before the replacement is due.

Lacks sensitivity analysis

Although some practitioners include some form of sensitivity analysis (SA) within their studies, the standards do not prescribe it. Indeed, this follows since the standards do not address uncertainty in estimates. In reality, practitioners do not know every number to the third significant figure. Sensitivity analysis is an important disclosure of the practitioners confidence in the data. Associations can use SA to prioritize their own efforts.

Sensitivity analysis (SA) is a set of techniques for understanding what “drives” a model. (Saltelli, 2000) states:

“Modellers conduct SA to determine: (a) if a model resembles the system or process under study; (b) the factors that mostly contribute to the output variability and that require additional research to strengthen the knowledge base; (c) the model parameters (or parts of the model itself) that are insignificant, and that can be eliminated from the final model; (d) if there is some region in the space of input factors for which the model variation is maximum; (e) the optimal regions within the space of the factors for use in a subsequent calibration study; (f) if and which (group of) factors interact with each other.”

For our purposes, a reserve study benefits from (b), understanding the drivers of uncertainty and variability, and (c), the significance of elements on the result.

Quantifying (c) or the impact of each component (and parameter) on the model is easiest; many existing studies include a breakdown of the costs over the time horizon using a pie chart or a treemap. In the example model, the primary contributions to total cost are:

Based on this summary, we can tell that r, the inflation rate, is the primary driver of costs. This is expected since we are looking over a 30 year period. Although the inventory shows “Elevator - Modernize” as having the same cost as “Roof - Replace,” the elevator will need to be replaced twice during the study period as the roof.

Associations should find these lists useful to prioritize their efforts, particularly where to not spend their time.

The other relevant form of sensitivity analysis is to quantify the impact on uncertainty or variability, rather than the overall cost. If the practitioner does not provide any estimates of uncertainty, then this analysis is impossible. Associations should be suspicious of any study that presents all the estimated costs as having no significant uncertainty and practitioners being certain within a year of every component’s remaining life.

Within the model, there are four variables with uncertainty: the remaining useful life of components, their cost for repair or replacement, the inflation rate, and the interest rate. In our experience, although not required nor encouraged by the standards, practitioners often provide some estimate of uncertainty for the costs and will plot low and high error bounds of their financial estimates. Uncertainty in remaining useful life can be very significant (e.g. a +/- one year uncertainty for a ten year component is equal to 10% of the nominal cost), but seems to be rarely recorded.

That said, computing contributions to uncertainty requires advanced mathematics and, in practice, specialized software. For our tool, we used the SALib library for its implementation of Sobol Sensitivity Analysis (Saltelli, 2006). Using this technique, we find that the drivers of uncertainty within the example model are:

This list is similar to the one prior, although inflation (r) is significantly larger. Although the bounds on r are relatively small versus the +/- 20% of each component’s costs, inflation is more impactful due to its corrosive effect.

If a component is a major driver of cost, but not uncertainty, then it is well understood and the project should be easy to manage. Contra-wise, if a component is a minor driver of cost but a major driver of uncertainty, then it is poorly understood. An association may be free to ignore it until the project comes due. If a component is both a major driver of cost and uncertainty, then an association should invest effort to reduce the uncertainty, perhaps by seeking contractor advice or seeing how other associations have managed a similar situation.

With many practitioners already providing a summary by cost impact, it should be a standard element of reserve reports. The summary prioritizes projects and helps the association board communicate the need for reserve funding to their members.

As an advanced technique, we are hesitant to mandate quantifying impact on uncertainty. Yet without it, or a similar technique, the practitioner is failing to disclose fundamental limitations of their knowledge and ability. This is not an admission of negligence or incompetence, but rather limitations of the state of the art.

Fails to stress need to mitigate inflation

In the examples so far, you may have noted the high impact of inflation on costs and uncertainty. Inflation is highly variable, and thus the impact uncertain, as this table of historical inflation illustrates:

Decade Annual Average Cumul. by Decade
1913-1919 9.80% 92.86%
1920-1929 0.38% 4.82%
1930-1939 -1.80% -18.60%
1940-1949 4.86% 68.57%
1950-1959 1.82% 24.58%
1960-1969 2.45% 28.23%
1970-1979 7.25% 103.45%
1980-1989 5.82% 64.41%
1990-1999 3.08% 33.47%
2000-2009 2.54% 28.31%
2010-2019 1.75% 19.38%

The “annual average” is computed as the geometric mean of each year’s inflation (CPI-U) within that decade. The cumulative inflation demonstrates the constant erosion effect as, even in the 1950s where the average annual inflation rate was below the 2% target, prices grew by 25%. Over the thirty year time frame of most reserve studies, inflation provides a huge range of possibilities.

As something so variable, one might be inclined to eliminate any projections of inflation as it cannot be “reasonably estimated” per the three-part test. In fact, an alternative described in (Winokur, 1998) is to revise cost estimates as inflation numbers are known, but otherwise keep the study in nominal dollars. However, in the language of risk management, rather than accept inflation, we can work to mitigate it.

Associations should invest savings to mitigate inflation. (Frumkin, 2009) states that “Meeting or potentially beating the inflation rate through safe, secure investments should be the board’s ultimate goal." The financial analysis of a study, which has a multi-decade horizon after all, should similarly advance that goal.

By “safe, secure investments”, (Frumkin, 2009) advocates the following prioritized elements in an investment policy:

  1. Safety of Principal: Safety of income and safety of principal.
  2. Liquidity
  3. Yield

The book then goes on to describe some specific investment strategies that fit this policy, although, as usual, it recommends boards to work with a qualified financial advisor. (Some associations use alternative systems, such as selling bonds, to provide revenue. Associations can also own profit-generating assets, so any government seeking to legislate in this area should be aware that there is ongoing innovation.)

The 2023 reserve standards simply require practitioners to describe “assumptions utilized for interest and inflation” in their report. In sample reports, we have witnessed interest rates often far below assumed inflation rates. Given the massive deleterious impact of inflation on savings without mitigating interest, it seems a dereliction of duty for the practitioner to not, at least, refer the association for some professional help.

Conclusion and Recommendations

Since all models are wrong the scientist must be alert to what is importantly wrong. It is inappropriate to be concerned about safety from mice when there are tigers abroad. (Box, 1976)

Reserve studies, at least from first principles, purport to be a straight-forward financial planning tool for planned maintenance. Professional organizations advocate that their practice leads to better long-term maintenance and more predictable member assessments. Some states have agreed and have mandated their practice. However, data is sparse on their efficacy.

Research into community associations is often restricted by the lack of public data. Although associations and managers are often surveyed, researchers lack access to data to make efficacy measurements or track decisions. If associations shared reserve studies, they could all benefit from tracking trends in costs and actual life expectancies of key components.

We propose the following changes in the spirit of maintaining the good qualities of the method while mitigating its weaknesses. Association boards can be proactive, rational, and adaptable. Per Box’s quote above, faults and limitations in a model can be overcome when they are understood and the board has freedom of action.

  1. Associations share data and assist in advancing knowledge
  2. Practitioners quantify impact by cost
  3. Practitioners include uncertainty estimates on costs and remaining life
  4. Professional bodies advocate mitigation of inflation
  5. Associations work to mitigate the impact of inflation on their reserve accounts

We also recommend reviewing (Winokur, 1998) which focuses on association’s financial governance.


(Box, 1976) Box, George E. P. (1976) “Science and statistics”, Journal of the American Statistical Association, 71 (356): 791–799, doi:10.1080/01621459.1976.10480949 PDF

(CAI, 2020) Foundation for Community Association Research (2020) Best Practices: Reserve Studies/Management. PDF

(CAI, 2023) Community Associates Institute (2023) Reserve Study Standards. PDF

(Foundation, 2020) Foundation for Community Association Research (2020) Breaking Point: Examining Aging Infrastructure in Community Associations. PDF

(Frumkin, 2009) Frumkin, M., March, N. (Eds.) (2009) Reserve Funds: How & Why Community Associations Invest Assets, 2nd ed.

(Saltelli, 2000) Saltelli, A., Chan, K., Scott, E.M. (Eds.) (2000) Sensitivity Analysis, Probability and Statistics. John Wiley & Sons, Ltd.

(ULI, 1964) Urban Land Institute (1964) The Homes Association Handbook Rev. and enl. ed. Washington: Urban Land Institute. Archive

(Whitestone, 2009) Abate, D., Towers, M., Dotz, R., Romani, L. (2009) The Whitestone Facility Maintenance and Repair Cost Reference 2009-2010, 14th ed. Whitestone Research.

(Winokur, 1998) Winokur, J. (1998) Critical Assessment: The Financial Role of Community Associations. Santa Clara Law Review Vol 38, Iss 4. PDF

  1. In some cases, the reserves only existed to cover inflation, rather than the principal plus inflation costs of a project. ↩︎

  2. ICBI requires their practitioners to use certified software, while CAI and ARPA impose no similar restrictions. ↩︎

  3. This is not a fully rational goal as an association could excessively mitigate the possibility of a special assessment. Ideally, this goal would be “two sided” and be expressed in terms of efficiency. ↩︎

  4. In the risk management literature, “inflation risk” is the risk that inflation will exceed some defined threshold. ↩︎