Preliminary results

This section presents the results from our preliminary modeling based on the base case values outlined above. These are not intended to provide a definitive answer, but rather to test the feasibility of the approach, define and scope the data collection requirements and further model developments necessary for more robust estimates of the potential impacts of an open access archiving mandate for federally funded research, such as that proposed in the FRPAA. 

Modeled impacts on returns to R&D

Table 3 presents the preliminary modeled estimates of the impacts of a one-off increase in accessibility and efficiency on returns to 2008 R&D spending based on total federal expenditure on R&D and R&D expenditure by the 11 FRPAA agencies, with percentage changes in accessibility and efficiency shown cumulatively. For illustrative purposes, we present ranges of rates of social return to R&D of 20% to 60% (Arundel and Geuna 2003) and increases in accessibility and efficiency of 1% to 10%.[1]

With a 20% return to FRPAA agency R&D spending of $61 billion in 2008, a single percentage point increase in accessibility and efficiency would have been worth around $245 million (per annum) in increased returns to R&D. Of this total, perhaps some $160 million might be expected to accrue within the US, with the remainder spilling over to other countries.

Table 3: Estimates of the impacts of a one-off increase in accessibility and efficiency on returns to R&D (USD millions, 2008)

Source: Authors’ analysis.

These are recurring annual gains from the effect of one year’s R&D spending, such that if the change that brings the increases in accessibility and efficiency is permanent (e.g. the adoption of open archiving as proposed in the FRPAA) they can be converted to growth rate effects.

Comparing costs and benefits

In this section we attempt to compare the costs and benefits associated with the proposed FRPAA archiving mandate. Details of the model and its operationalization can be found in Annexes I and II.

One thing to note is that we are modeling the transitional impact of open access archiving over 30 years. Because of the lag between research expenditure and the realization of economic and social returns to that research, the impact on returns to R&D is lagged (by 10 years in the base case) and the value of those returns are discounted accordingly. This reflects the fact that the impacts of open access archiving would be prospective and not retrospective, and that the economic value of impacts of enhanced accessibility and efficiency would not be reflected in returns to R&D until those returns are realized. Put simply, this has the effect that over a transitional period of 30 years we are comparing 30 years of costs with 20 years of benefits (Figure 1).

Figure 1: Indicative distribution of impacts over a transitional period of 30 years (USD millions in years 1 to 30)

Source: Authors’ analysis.

Figure 2: Indicative distribution of impacts over 30 years in a simulated steady state period (USD millions in years 1 to 30)

Source: Authors’ analysis.

An alternative approach would be to model a hypothetical ‘steady-state’ system in which the benefits of historical increases in accessibility and efficiency enter the model in year one. This would reflect the situation in an alternative system, after the transition had worked through and was no longer lagging returns to R&D. Put simply, in such a model one would be comparing 30 years of costs with 30 years of benefits (Figure 2).

We took the view that it was more realistic and of more immediate concern to model the transition. Nevertheless, it must be emphasized that a transitional model returns significantly lower benefit/cost ratios than would a hypothetical alternative ‘steady-state’ model.

Potential impacts of an open archiving mandate

The base case model parameters, their sources and rationale are outlined in Table 1 (above) and Annex II, and input data values are summarized in Table 2 (above) and Annex II. With these base case values, we model the impacts relating to R&D spending by the 11 departments affected by the FRPAA using reported arXiv, NIH and the upper bound LIFE² lifecycle archive costings. These archive costings vary significantly, but the mid-range NIH costing might provide a reasonable guide.

Table 4: Modeled estimates for the base case parameters (USD millions over 30 years in Net Present Value & Benefit/Cost Ratio)

Notes: Using the base case parameters outlined in detail in Annex II and assuming a six-month embargo and 100% compliance with the proposed mandate.
ource: Authors’ analysis.

Over a transitional period of 30 years from implementation, the potential incremental benefits of an open access archiving mandate for all FRPAA agencies’ funded R&D[2] might be worth around $1.6 billion (Net Present Value), around 4 times the estimated cost using the higher end lifecycle costing, 8 times the cost using NIH costing and more than 24 times the cost using arXiv costing. Perhaps some $1 billion of these benefits would accrue within the US, with the remainder spilling over to other countries. Hence, the US national benefits might be around 5 times the costs.[3] The overall impacts of openly archiving all FRPAA agencies’ funded R&D article outputs would be greater than these incremental impacts, with likely US national benefits of around 8 times the costs (Table 4).

These estimates assume a six-month embargo period between publication and open accessibility. If there were no embargo, we estimate that incremental returns might be closer to $1.75 billion. Hence, a six-month embargo reduces the returns by around $120 million (NPV). Of course, the impact of an embargo delaying open accessibility will vary significantly between fields of research and disciplines, having greater impact in faster moving fields of research and practice than in those where the progress of knowledge, application and practice is slower.

It should be noted that these estimates are based on increased returns to R&D through increased accessibility and take no account of the potential activity cost impacts of more open access to federally funded research (e.g. possible savings in such areas as researcher time spent in search, discovery and access) or of potential efficiency impacts (e.g. in reducing duplicative research). They also focus on the transitional period following implementation and, as noted above (Figures 1 and 2), once established the benefits of open archiving would be substantially greater than immediately following implementation and during the transition. In a hypothetical ‘steady-state’ scenario, for example, estimated US national benefits might be more than 50 times the cost.

However, it should be noted that benefits in the form of increases in returns to R&D are diffuse in nature, occur throughout the economy and, indeed, throughout the world. They also accrue over time, sometimes lagging research expenditure and publication by many years. In contrast, the costs are both local and immediate. Hence, the costs must be met up-front, in order to maximize the return on public investment in research. It should also be noted that these estimates are preliminary in nature, intended to test the feasibility of the approach, define and scope the data collection requirements and further model developments necessary for more robust estimates of the possible impacts of an open archiving mandate for federally funded R&D. They come with many caveats (See Annexes I and II for details).

Sensitivity in the model

Among the caveats is the model’s sensitivity, which we examine here in order to prioritize areas for further data collection and model development. Using the mid-range NIH reported archiving costs and changing individual parameters one-at-a-time we find that:

• The number of articles produced is an important driver of archiving costs. Nevertheless, at base case values the US national benefits would exceed the costs if current annual article output where more than one million – almost 6 times current article output – at the same level of R&D funding.
• Archiving costs are also important and further work is required to establish exactly what those costs would be. Nevertheless, the base case model returns net national benefits with per article submission costs of more than $375 (excluding author deposit costs).
• The potential increases in accessibility and efficiency resulting for open archiving are an important input, as the greater the increases the greater the benefits. However, a combined total change in accessibility and efficiency of less than 1% returns net benefits.
• The average rate of social return to publicly funded R&D is an important parameter, but average rates of as low as 4% produce US national net benefits.
• The rate of depreciation of the stock of research knowledge is set according to the established formula used by the US Bureau of Labor Statistics, and while the rate chosen makes a substantial difference (with lower rates producing higher benefits) the benefit/cost ratios remain greater than 1 for rates higher than 16% per annum – a rate that is higher than conventionally used for publicly funded R&D.
• The lag between research spending and its economic impacts and the distribution of those impacts over time have been set to work in the range of 5 to 15 years average lag, with a normal distribution of impacts over approximately 10 years. The shorter the lag the greater are the impacts, but a lag of 15 years still gives benefits that are more than double the costs.

Overall, even during a transitional period, the benefits appear to exceed the costs over a wide range of values and it is difficult to imagine any plausible values for the input data and model parameters that would lead to a fundamentally different answer.