Thursday, August 22, 2019

Workers’ compensation Insurance Arrangements: Does the model make a difference? Part 2


 [For the first part of this discussion, see “Workers’ Compensation Insurance Arrangements:  Does the model make a difference?  Part 1” at  http://workerscompperspectives.blogspot.com/2019/07/workers-compensation-insurance.html    or   https://workerscompperspectives.wordpress.com/2019/07/11/workers-compensation-insurance-arrangements-does-the-model-make-a-difference-part-1/ ]   

In my previous post, I described the range of public policy insurance arrangements governments use in the workers’ compensation insurance market.  As noted, the market for workers’ compensation is not a free market but one that is altered by government interventions and oversight.  The most common active intervention in the market (after mandating compulsory workers’ compensation insurance coverage for most employers) is the legislative creation of a “state” fund for workers compensation. 

Analysis of the state, provincial and federal jurisdictions in the US, Canada, and Australia shows that the majority of jurisdictions use competitive or exclusive state funds to provide some or all of the workers’ compensation insurance needs within their respective jurisdictions.  Provision of workers’ compensation insurance solely by private or mutual insurance is common in many jurisdictions either as the only option or in a market competing with state funds. 

I concluded the last post by grouping jurisdictions into three categories:
  • Private and mutual insurers only jurisdictions
  • State Fund and Private Insurance competitive jurisdictions
  • Exclusive “State” Fund Jurisdictions




The purpose of this examination is to determine if the categories of workers’ compensation insurance arrangement are associated with different worker benefits, employer costs, or the ratio between these important measures. 

For this analysis, “worker benefits” refers to benefits paid per $100 of payroll (payments to injured workers and to providers of medical care in the calendar year, regardless of year of injury); “employer costs” are represented as costs paid by employers per $100 of payroll referring to calendar-year insurance premiums paid plus any annual deductible or self-insurance costs including administration.  

Note:  Although the ratio between worker benefits paid and employer costs is often expressed in dollar terms [e.g., “Total Benefits per $1 Employer Cost”], mathematically the ratio between these two measures is a unitless value, independent of currency or consideration of exchange rates. Whether derived directly from cash flow analysis or indirectly from published rate per $100, the ratio is the same:


[(Worker Benefits Paid)/payroll*100]/ [(Employer Cost)/payroll*100]=
(Worker Benefits Paid)/ (Employer Cost)    

This ratio represents the proportion of employer costs (mainly premiums paid for workers' compensation insurance) in a year that are paid out in cash benefits (compensation and medical costs, regardless of year of injury) in that same year.    


If the workers’ compensation insurance arrangement is irrelevant, the ratio of worker benefits paid to employer cost should be similar regardless of the insurance arrangement category.  Individual states within each category may vary but collectively, the ratio of “worker benefits paid” to “employer costs” over time should be similar for each of the three categories noted above.  However, this analysis shows substantive difference associated with these three models. 

Main Findings



US jurisdictions in the exclusive state funds category are associated with a higher worker benefits paid to employer cost ratio [WBP/EC ratio] than categories of jurisdictions without a state fund (relying on private workers’ compensation arrangements), or jurisdictions with a competitive state fund and private insurance.  The category of jurisdictions with competitive state funds has a WBP/EC ratio greater than the private provision category but lower than exclusive state fund category. 

A similar calculation methodology was applied to Canadian workers’ compensation boards, all of which are in the exclusive category. Collectively, Canadian workers’ compensation boards have a WBP/EC ratio similar to the US exclusive state category.  A sample of Australian exclusive, publicly underwritten workers’ compensation insurers collectively produce a similar ratio.  

The category or US jurisdictions served by private or mutual insurance with no competitive state fund had the lowest WBP/EC ratio, i.e., the lowest proportion of current year employer costs going to pay workers cash compensation, related medical bills and other benefits. The WBP/EC ratio was also below the average for the three categories considered and the ratio for the exclusive state fund category.  The private  category also had the lowest worker paid benefit per $100 payroll and the lowest employer cost per $100 payroll.  

Some background on workers’ compensation insurance finance

Workers’ compensation insurance has both similarities and differences with other insurance lines.  Insurance transfers much of the financial risk of rare but costly events from the insured to the insurer in exchange for a premium.  Workers’ compensation insurance has the added feature of being an exclusive remedy in most jurisdictions; employers are protected from legal action for most work-related injuries—a feature not present in most other insurance lines.  Workers may lose the right of tort or common law action but (in theory if not practice) gain assured access to defined benefits for work-related injury, illness and death.  Access to benefits is based on “work-relatedness” [injury arising out of and/or in the course of employment] rather than fault.

As with most other lines of insurance, premiums are paid to cover a specific period of time, typically a policy year (usually a calendar year); claims against the policy must meet well-defined events.  Workers’ compensation differs from many other lines in that benefits are paid in accordance with levels set by legislation. 

A major difference between workers’ compensation and other lines of insurance relates to the time it takes for the full cost of a claim to be paid out.  The full cost of most property and casualty insurance claims usually developed quickly.  For example, you pay insurance for your home or your car on an annual basis; if your home burns down or your car is stolen in the coverage period, you can claim the loss to the agreed upon value of your loss shortly thereafter.  The insurer makes the payment and the claim is closed. The full claim cost under these types of insurance agreement develop quickly and definitively.  Not necessarily so for workers’ compensation insurance.

With workers’ compensation, work-related injury, illness or disease may or may not occur quickly; disability may carry on for many months or years beyond the policy year, the period covered by the insurance agreement.  The premium must be sufficient to take into account both the immediate and long-term liabilities associate with a claim.  In workers’ compensation insurance terms, every accepted claim “incurs” costs. The money is not paid out immediately but the liability can be estimated (and notionally funds “reserved” to cover these future costs).  The true, full cost of the claim will develop over time and known definitively once the claim is concluded or "finalled".  The conclusion of an accepted clam may be days, weeks, or years after the date of injury (when the workers’ compensation insurer’s financial liability for the claim was incurred). 

From the workers’ compensation insurer’s perspective, cash flow accounting is not really the most appropriate way to look at accounts.  Liabilities that are incurred need to be projected well into the future; premiums collected in the policy year must meet those projected costs.  Factors such as the rate of inflation, required rate of return on investments, changes in claim or claimant life expectancy, and potential increases in medical costs can have major impacts on future costs of claims incurred.  The current premium must cover all current year and expected future costs associated with claims that arise from the policy year.

I mentioned investment returns as a factor. The “long tail” of costs associated with a work injury means the eventual value to be fully paid out may be discounted to take into account anticipated gains.  Notionally, the premium collected in a year will be used to pay medical and indemnity costs arising from injuries in the current year and in many years to come.  Insurers set aside or “reserve” funds from each policy year’s premiums to pay these future costs.  Those reserves may be invested to generate income and growth in value that can be used to pay claims.  An expected rate of return is taken into account in premium development but excess gains or losses often occur.

On an annual operating basis, considering all cash payments made in the policy year to workers or their families, a portion of those payments will relate to injuries that arose in that year while the rest of the cash paid will relate to claims for injuries that occurred in the prior and previous years.  Notionally, these payments are accounted for by drawing down reserves established each year for “unfinalled” claims. 

This disconnect between the benefits paid to workers in the year and the employer costs including premiums paid in the year can lead to divergent patterns of expenditures.  For this reason, the National Academy of Social Insurance (NASI.org) advises the following in its reports on Workers’ Compensation: Benefits, Costs and Coverage (October 2018) [page 41]:

The reader is cautioned that the ratios represent benefits and costs paid in a given year, but not necessarily for the same claims. The benefits measure includes payments for all injuries/illnesses that occurred in the given year and for some injuries that occurred in prior years. The costs measure (premiums paid to insurers and state funds) includes projected future liabilities for injuries/illnesses that occurred in the given year. In other words, the costs and benefits paid in a given year are not tracking the full costs of a particular set of claims.

NASI cautions against using individual state differences in their reported measures as a way of identifying relative differences as favorable or unfavorable.   The NASI report provides data on workers benefits paid, employer costs and the ratio between them on a jurisdictional level.  It also provides and charts the data at an aggregate level as an indicator of trends. 

The following analysis acknowledges the cautionary advice and does not examine individual jurisdictional data.  Instead, it disaggregates the NASI data and re-aggregates the same data in accordance with the three main categories of workers’ compensation insurance arrangements described in detail in the previous post.  This allows a direct comparison with additional jurisdictions outside the US.  Canadian data from the Association of Workers’ Compensation Boards of Canada (AWCBC.org) and from a sample of Australian jurisdictions reflect two additional sub-groupings in the exclusive state fund category and can be compared with the US data.

Employer Costs and Worker Benefits Paid

Workers’ compensation insurance collects premiums from employers and pays compensation and benefits (essentially the cost of medical diagnostics, treatment, rehabilitation, and a portion of worker financial losses) to workers (families or estates in the event of a work-related fatality).  There are other insurance costs, of course, including the cost of administration, underwriting, adjudication, adjusting, and loss prevention to name a few.  The human and social costs associated with work injuries are real and significant but are not consistently calculated, tracked or reported; consequently, these important and real costs are not included in this analysis. 

Generally, workers’ compensation insurance premiums are the main employer cost.  As noted above,  employers pay the premium to transfer to the insurer a defined portion of their financial risk associated with work-related injury, illness, disease and death for a specified period; each work-related injury, illness and death occurring in that period (typically a policy year) creates a liability-- an “incurred” claim expense that will be paid out as worker compensation (amounts toward income loss or non-economic losses based on temporary and permanent disability) and benefits(including medical treatment, rehabilitation, vocational rehabilitation, prosthetics, medications, funeral expenses, survivor benefits) in the year of injury and potentially for many years beyond that.  Also as noted above, those benefits recorded as being paid out to workers on a cashflow basis in a given year in the NASI report, AWCBC data and Australian annual reports (cashflow statements in financial reports), relate to claims from many prior years as well as the current year. 

This actuarial “long tail” of payouts means insurers must have sufficient funds set aside to cover the future costs of the worker compensation, benefits and administration.  These funds and reserves are invested with the returns providing revenue and securing entitlements.  Excess returns may allow for lower premiums or rebates/dividends to employers (effectively reducing employer costs); less than required income from premiums and investments may also require premium rate increases (reflected in higher employer costs).

There are other employer costs.  Some states allow for employer deductibles, effectively a form of self-insurance.  These are included in employer cost estimates in the US.  Some Australian jurisdictions require employers pay the initial time-loss compensation and medical costs before workers’ compensation takes over, although the amounts and durations vary by jurisdiction. [Note:  there are no similar employer deductibles in Canada].  Employers also have many costs associated with investigation of injuries, disability management, and return to work.  These are not included in the estimates of employer costs.  

Workers also have costs that are not directly accounted for in this analysis.  Workers and their families pay the human cost in pain, suffering, and quality of life; they also pay in terms of differential between earnings above the compensation rate and  “worker deductibles” such as the non-reimbursed waiting period common in the US and two Canadian provinces (New Brunswick and Nova Scotia).  [Australian jurisdictions have no worker waiting periods].

Drivers of worker benefits and employer costs

Legislatures set the parameters of compensation and benefits—the main drivers of overall worker benefit expenses and, consequentially, workers’ compensation premiums.  The parameters include maximum insurable earnings (or maximum benefit that may be paid), compensation rate, duration of waiting periods, provisions for retroactive periods, cost of living adjustment provisions, entitlements to vocational rehabilitation, funeral expenses, and many other elements of workers’ compensation coverage.  These provisions vary widely in the US (see NASI data Table C;  IAIABC/WCRI, Workers’ Compensation Laws as of January 1, 2019, April 2019. WC-19-22;  U.S. Chamber of Commerce,  Analysis of Workers' Compensation Laws (2018 Edition)).  Analogous provisions in Canada also vary but less widely (see AWCBC Summary Tables). Theoretically, jurisdictions with identical industrial mix, wage distribution and injury rates could have vastly different employer costs and worker benefits paid as a result of differences in legislative provisions. 

Workers’ compensation jurisdictions in Canada tend to have higher maximum insured earnings,  benefits payable,  and compensation rates than their US counterparts.  Compensation rates in the US are typically 66 2/3rds of gross earnings while Canadian compensation rates are typically 85-90% of net (spendable) earnings—the latter often providing a greater level of income replacement across all income classes, particularly where progressive tax regimes are in place.  Most US workers’ compensation jurisdictions have a waiting period of three to seven days while most Canadian provinces have no waiting period.  These differences in legislative design of worker compensation and other benefits are reflected in both employer costs and worker benefits per $100 of payroll. 

Industry mix is a big driver of differences in workers’ compensation cost differences among states.   States vary in their concentration of covered employment in sectors such as agriculture, manufacturing, construction, transportation, medical services, and advanced technologies.  Each industry has its own risk of injury and loss potential.  States with higher concentrations of work-related injuries in skilled employment with higher earnings within a high legislated insurable earnings or benefits payable cap may have higher individual claim costs compared to a state with a higher proportion of work-injured in lesser skilled and compensated categories.

Many other factors may also influence these costs including the severity and frequency of injuries, demographics such as workforce age, distribution of wage earner incomes, medical fee costs, etc.   Given this background, it is not surprising that the value of worker benefits paid and employer costs vary from state to state and even from year to year in the same state. 

Comparing employer costs across jurisdictions

Few studies have examined the how employer costs vary with the category of workers’ compensation insurance arrangement across jurisdictions.  Challenges to analysis include the difficulty of controlling for industry mix, demographic differences, and the many variables related to legislated worker benefits.  In theory, for jurisdictions where all external conditions (industry mix, demographics, medical cost structure, etc.) and experiential conditions (injury rate, duration, severity, earnings distribution, access to medical care, etc. ) are similar,  the jurisdiction with the legislation that defines greater worker benefits will likely have both greater worker benefit and employer cost per $100 payroll.  Employer costs per $100 payroll will likely be higher for jurisdictions with unfunded liabilities and lower for jurisdictions using excess market returns to moderate or discount premium costs for policy holders. 

Complex analysis that controls for most of these factors is limited but exists.  Work by the late Terry Thomason and John F. Burton Jr. in particular stand out.  For more information on their approach and findings, see Terry Thomason, Timothy P. Schmidle, and John F. Burton, Jr., Worker’ Compensation: Benefits, Costs, and Safety under Alternative Insurance Arrangements, WE Upjohn Institute, January 2001 and Terry Thomason and John F. Burton, Jr, The Employers’ Costs of Workers’ Compensation Insurance in Ontario and Selected Other Canadian and U.S. Jurisdictions, Workers’ Disability Income Systems, Inc., December 2001 [Revised June 2002].

NASI and AWCBC Data

The National Academy of Social Insurance (NASI.org) publishes an annual report that covers most of the US jurisdictions noted in the above table (Puerto Rico is not currently in the NASI dataset).  The Association of Workers’ Compensation Boards of Canada (AWCBC.org) publishes data on all the Canadian jurisdictions.  In the most recent NASI report, several tables provide a standardized measure for both costs and benefits. 

In the NASI report, Table 12 Workers' Compensation Total Benefits Paid Per $100 of Covered Wages, by State   provides a standard comparator for 2012 through 2016.   Each year’s figure for each state is an “observation”.  Any one observation is subject to a lot of variability so multiyear data for each state tends to smooth out some of these issues.  For example, a court ruling or legislative change could result in large cash payouts one year.  Across many observations, these anomalies tend to “average out”.  

Table 14 Employer Costs for Workers' Compensation Per $100 of Covered Wages, by State provides a similar set of observations.  The figures in this table represent a cost for each state for each year based on a standard $100 payroll basis.  Note, the values in Tables 12 and 14 are actual dollar amounts calculated against a standardized $100 payroll base (essentially, a percentage expressed in dollar terms). The values are standardized to the but not normalized, that is, the values for each jurisdiction do not control for differences in legislation, wages, industry, risk, demographic, or any other factors. 

These two tables are limited in what they reveal about the jurisdictions they describe.  Together, these tables are more akin to statements of cash flows against a common denominator of $100 of payroll.  Table 12 figures represent outgoing dollars to workers toward covering lost wages and medical costs.  Table 14 represents incoming dollar amounts paid by employers (cash expenditures in a typical financial statement) relative to the same $100 of payroll base. [Note:  Source data for each states in these tables are contained in the Sources and Methods document on the NASI.org website].

Individually, these figures tell us nothing directly about administration costs, sufficiency of reserves, return rates on investments or system performance.  At best, on an individual jurisdiction basis, the ratio between the worker benefits paid and employer cost figure for each year create an “observation pair” that can be tracked over time, grouped with other observations from other jurisdictions and used in analysis like this.  The observation pair ratio reports the proportion of premium income (employer cost paid for workers’ compensation coverage) that equates to cash payments made to workers.  The complement of this ratio represents the proportion of employer costs that are expended on things other than worker benefits.  This ratio is a cash-flow rather than an actuarial calculation. 

Any one observation or even a set of observations from one jurisdiction has limited value because it does not track the full costs of any set of claims.  Across all states, like the results for employers’ costs and worker benefits, the ratio provides a reference amount that helps visualized the long-term relationship between benefits and costs for workers’ compensation in the US.  Disaggregating the national averages into the three categories of insurance arrangements provides an opportunity to determine if any of the insurance arrangements are associated with a higher or lower WBP/EC ratio.  
If the arrangement is irrelevant then this ratio should be similar across the three categories.

Figure 1 in the NASI report summarized the aggregate data from these two tables into points on in a timeseries chart.  The reported line includes federal workers’ compensation.  Disaggregating the combined data for the five-year period into the Private, Competitive and Exclusive categories and adding a line that excludes the federal results produces the following:



The Competitive state fund jurisdiction category mirrors the overall NASI totals for both worker benefits and employer costs.  Removing federal employees from the NASI total shifts both employer costs and worker benefit lines marginally lower.  The Exclusive state fund jurisdiction category is associated with markedly higher worker benefits and accordingly higher employer costs.  The Private insurance jurisdiction category tracks well below the average NASI result with or without federal employees. 

It is important to note that the points on the chart tell only part of the story.  The range of employer costs and worker benefits per $100 payroll is very wide.  For 2016, the range in worker benefits per $100 payroll was $0.26 to $1.57 and employer costs ranged from $0.48 to $2.32.

AWCBC provides a publicly available data reporting tool that allows for the extraction of data necessary to create similar ratios from the provincial workers’ compensation system.  Using Key Statistical Measures 5.1-Benefits Paid During the Year ($ millions)($), 10-Assessment Revenue for Assessable Employers ($ millions)($), and 12-Assessable Payroll ($ billions)($)for Canada for 2012 to 2016, values for worker benefits and employer costs per $100 payroll were calculated and added to the NASI data for comparison.   Note, for this analysis, worker benefits and employer costs related to federal employees are excluded. 



The Canadian results track higher than the US results for both employer costs and worker benefits and are most similar to the results from the US Exclusive state fund jurisdictions.
SafeWork Australia (safeworkaustralia.gov.au) advises that similar data are not directly available for comparison. 

Ratio of Worker Benefits to Employer Costs

The primary source of cash paid to a workers’ compensation insurer is through the premiums (and/or assessments) paid by employers.  The primary outflows of cash are for worker compensation (indemnity, permanent disability payments) and medical expenses (including payments to physicians, hospitals and for medical-related expenses including pharmaceuticals, physiotherapy, etc.) paid to workers or to healthcare and rehabilitation providers for their treatment and care.   The NASI data expresses the worker benefits paid and employer costs relative to $100 payroll, mitigating many of the effects of system and jurisdictional economic differences.  The ratio between the two is analogous to the claims loss ratio often used as an indicator of the financial health of an insurer.  More complex ratios such as the combined ratio require additional data not included in the NASI data set. 

The ratio between these worker benefits paid and employer costs per $100 payroll will differ from jurisdiction to jurisdiction and fluctuate from year to year.  Premiums are generally prospectively determined while expenditures resulting from work injuries are retrospective, covering claims that occurred from many past years.  Economic cycles can also play a role in the relationship between the benefit paid out and the premiums received.

There is no established benchmark or threshold for the workers’ compensation benefits paid to employer cost ratio.  While paid benefit to cost ratios are frequently used in financial analysis, their applicability to workers’ compensation insurance are subject to the cautions and context outlined in the NASI report (2018, page 41-45).  One possible comparator reference could be the Medical Loss Ratio for healthcare insurance in the US.    

The US federal Affordable Care Act of 2010 (ACA) set first the Medical Loss Ratio (MLR) standard. The ACA-MLR standard requires insurers spend at least 80% (85% for larger insurers)  of premium income on medical care and health care quality improvement. The remainder of premium income is not constrained and goes to cover 20% (or 15%) for administration, promotion, and shareholder profits.  As noted earlier, there is no comparable standard in workers’ compensation insurance.  In the absence of any other benchmark, the MLR may provide a reasonable external reference value.  

Although workers’ compensation insurance medical benefits expenditures now exceed other cash 
benefit payments in the US [see NASI 2018, page 20], workers’ compensation insurance covers more than medical and healthcare-related costs.  The multi-year timeframe of more serious work-injuries may mitigate against a similar 80/20 rule.  That said, the relatively narrow range of observed ratios in the following analysis suggests a 75/25 split is typical for workers’ compensation insurance.

The ratio between these two measures at the aggregate level across all workers’ compensation systems in the US has fluctuated in a relatively narrow range over time.  Looking at the historical data in the NASI report, the ratio between Employer Costs and Worker Benefits has averaged    0.71  within a range of 0.54 to 0.88.  The higher the ratio, the greater the proportion of collected employer premiums go to worker benefits as opposed to other costs (including administration, underwriting costs, advertising, etc.).  The ratio may be an indicator of overall system efficiency.  


Each pair of observations (Worker Benefits Paid and Employer Cost) from the five years of data for the three categories of insurance arrangements in the US, the Canadian Boards and a sample of Australian publicly underwritten workers’ compensation insurers provide the data for the table at the top of this post.  

Note:  The observations from Arizona 2012 were omitted from the observation data for the US as the state fund was transitioned to a private insurer Jan 2013.  Arizona data 2013-2016 are included in the private jurisdiction analysis.   Data points from Ontario 2013 and Prince Edward Island 2012 are omitted from the Canadian observations because of incomplete data for those particular years.

As noted above, SafeWork Australia advises that similar data are not directly available for comparison.  Annual reports from the all but one of the jurisdictions that fall into the exclusive category (Queensland, Victoria, Comcare and South Australia, omitting New South Wales) do contain enough information in roughly parallel financial statements to calculate a similar ratio.  Using 2012-2016 published results from financial statements of cash flows and averaging the ratio of the workers’ compensation payments made (less recoveries) over premium income for each fund and year produces a ratio of 0.76.   Payroll data was not stated in these annual reports, therefore, no cost per $100 could calculated. 

The Australian data excluded the value of the “employer excess” applied in most jurisdictions (all jurisdictions in this sample except Comcare). Employers are responsible for the direct payment of the first 5 days (or 10 in Victoria) of time-loss or the initial set amount of medical cost ($667 in Victoria in 2016) that varies by jurisdiction.  This value is excluded from the denominator (employer costs) and the numerator (worker benefits).  This sort of individual claim employer deductible may contribute to a slightly higher ratio of benefits paid to employer costs as many claims within the excess level end up being paid directly by the employer.

The ratio between worker benefits and employer costs for workers’ compensation jurisdictions in the US is trending lower than the historical average, a possible indication of increasing costs of workers’ compensation insurance beyond the main cost of worker compensation and benefits. 

Exclusive and competitive state funds are associated with higher worker benefits and consequently higher employer costs but the ratio between worker benefits and employer costs for exclusive and competitive state fund jurisdictions tracks higher than for jurisdictions served by private insurers only.  Canadian workers’ compensation boards have both higher worker benefits and employer costs than US jurisdictions but have a WBP/EC ratio similar to US exclusive state funds.  Australian data for a sample of publicly underwritten schemes produces a ratio similar to the exclusive category in the US and to the Canadian workers’ compensation boards, also in the exclusive category. 

Concluding comments

The five years of data across US demonstrate higher ratios of workers’ compensation benefits to employer in jurisdictions with state funds.  As a group, US exclusive state funds had the highest WBP/EC ratio with Canadian workers’ compensation boards and the sample of Australian publicly underwritten schemes closely behind.  US Jurisdictions with competitive state funds were more similar to jurisdictions with private markets for workers’ compensation insurance for this ratio.  

The apparent similarity of competitive state fund to privately underwritten workers’ compensation jurisdictions is somewhat expected.  As competitors in the same market, both competitive state funds and their private counterparts face similar environmental factors.  The cost of land, labour, systems, transportation, medical services, prescriptions, etc. are part of the operating context; tax status and unique features of the state fund mandate are the only things separating state funds from private provision competitors in their respective marketplaces.  The slightly lower ratio for the private provision jurisdictions may relate to taxation effects that are diluted in the competitive state fund category by the market share held by the state fund.  To the extent that competitive state funds may be intended to widen the competitive field of private insurers, as a group, competitive state fund jurisdictions appear to be very similar to the private category. 

This analysis does not examine the underlying reasons for the observed differential.  The not-for-profit nature, exempt tax status, and the economies of scale of exclusive state funds may be significant factors contributing to this association.  This analysis did not address industry mix or benefit levels.  Given that exclusive state fund jurisdictions often provide higher benefits than many of the private jurisdictions, it is also possible that the higher WBP/EC ratio is at least partially related to the efficiency of making larger payments per unit of administrative effort.  Further research would be necessary to examine these possible explanations.  

It is not clear why the category of jurisdictions without a state fund have significantly lower worker benefits paid than the other two categories with state funds.  It may be that the existence of a state fund in the market is a cause or consequence of higher public interest in workers' compensation in these jurisdictions.  Further research into this association is necessary.  

The methodology applied to the Australian and Canadian data is analogous to but not identical to the NASI approach.  The Australian data does not have payroll denominators for comparative calculations of worker paid benefits or employer costs per $100.  Canadian values for these parameters may be higher than the NASI methodology determines for US jurisdictions because of differences in the denominator.  Canadian jurisdictions rely on "assessable payroll", which is analogous to "reportable payroll" defined in many jurisdictions.  Assessable or reportable payroll typically includes wages and other forms of compensation and remuneration but may also be subject to a "payroll limitation" or individual "payroll cap".  If the definition of payroll used in these calculations captures all wages and compensation without limitation, the lower the dollar value of worker benefits paid and employer costs per $100 payroll.  Conversely, a definition of assessable payroll that takes into account maximum insurable earnings, limitations or caps generates a denominator that is smaller than full payroll (wages and compensation).  As a result, the worker benefit paid and employer cost per $100 payroll noted in the table for Canada appear higher.  This topic requires additional research and may result in revisions if a common payroll definition applicable to both Canada and US can be devised and a reliable data source determined.  

The selection of an insurance arrangement to further the public policy objective is not simply a matter of one measure nor is it necessarily a decision that is set in stone.  As noted in part one of this series, jurisdictions have changed arrangements from time to time.  By disaggregating the NASI data into exclusive, competitive and private insurance categories, the data suggest exclusive and competitive state fund arrangements for workers’ compensation insurance provide higher ratios of worker benefits paid to employer cost than the private insurance provision category. Data from Canada’s exclusive workers’ compensation boards and a sample of Australian publicly underwritten schemes produce similar ratios to the US exclusive state fund category.

What this analysis suggests is that insurance arrangements are worth considering in comparative analysis.  Jurisdictions may find benchmarking against other insurance arrangements in the same category provides additional insights and more accurate assessment of their performance.  The insurance arrangement lens may be useful to policy makers and stakeholders in evaluating system performance and considering alternatives to their own public policy.  

This analysis looked at worker benefits paid, employer costs, and the ratio between these two measures.  Other parameters such as worker outcomes, denial rates, appeal or litigation rates, and stakeholder satisfaction may also be examined in a similar fashion in future research.