Assets for Independence Act Evaluation:
Design Phase Final Report
August 9, 2000

7.	Overview
	7.1	Purpose
	7.2	Data Collection Plan
	7.3	Data Analysis Plan
	7.4	Cost Estimate
		References

7.1 Purpose

A primary question of the evaluation is whether AFIA achieves the intended results cost-effectively. To inform an answer, the overall evaluation measures results, both in terms of changes in cash flows and in terms of changes in non-financial outcomes. Furthermore, the overall evaluation estimates the costs to produce these results. The ultimate aim is to compare IDAs to other means to achieve the same goals.

This section focuses not on the overall evaluation but rather on one specific, limited component, financial benefit-cost analysis. Specifically, this analysis responds to Section 414(b) item (5) of AFIA:

"The potential financial returns to the Federal Government and to other public-sector and private-sector investors in individual development accounts over a 5-year and 10-year period of time."

To this end, the evaluation will include a financial benefit-cost analysis from the points of view of seven groups:

• IDA participants
• IDA non-participants
• Private donors
• Employees and administrators of IDA programs
• Federal government
• State and local governments
• Society as a whole

The public sector includes the federal government and state and local governments (Exhibit 7-1). The private sector includes participants, non-participants, private donors, and employees and administrators of IDA projects. Society as a whole is the union of the other six groups.

The benefit-cost analysis must include all of these groups of stakeholders because each group has its own roles and its own goals, so each group experiences different benefits and costs. If IDAs are to succeed, then each group must play its part, and for a given group to play its part, its own benefits must exceed its own costs (Schreiner, 1997). Suppose, for example, that IDAs, if used, would have social benefits in excess of social costs. If benefits would not exceed costs from the point of view of the participants, however, then no one would participate, and then no other group of stakeholders nor society as a whole will receive any benefits. In essence, each group has some measure of veto power over the success of the entire project, and so the analysis checks whether benefits exceed costs not only for society as a whole but also from the point of view of each of the other groups.

Exhibit 7-1
Seven groups of stakeholders

7.1.1 Financial benefit-cost analysis in the context of the overall evaluation

Financial benefits and costs are quantitative, but in general, benefits and costs are both quantitative and qualitative (Exhibit 7-2). The property of "qualitativeness" or "quantitativeness" inheres not in the benefit or cost itself but rather in the ability of the analysis to measure it. Although qualitative results (such as changes in hope) are not measured in units comparable to other results, they are still real benefits and real costs that matter to the various groups of stakeholders. The overall evaluation gauges these, for example, through in-depth interviews with participants. Quantitative results, in contrast, are measured, and they include both results that can be expressed in dollars (such as changes in income) and results that cannot be expressed in dollars (such as changes in civic engagement). The survey, for example, produces quantitative measures.

As mandated by AFIA, the benefit-cost analysis planned here is "financial," considering only benefits and costs measured in dollars. The "impact" analysis will look at non-financial quantitative outcomes. Thus the financial benefit-cost analysis, although useful in many ways, does not pretend to encompass all of the many and varied benefits and costs engendered by IDA programs. This is not because non-financial results do not matter, but rather because it is too difficult to measure them in dollar units. The rest of the overall evaluation, and indeed the final verdict about whether "a permanent program of individual development accounts should be established" (AFIA, 414(b)(6)), will attempt to consider all benefits and costs, qualitative as well as quantitative.

Exhibit 7-2 below illustrates the relationships between the financial benefit-cost analysis and the other types of analysis in the overall evaluation. The final overall judgment will, as in all evaluations, be subjective, but the aim is to make the judgments and assumptions that underlie the subjective verdict as explicit as possible because explicitness makes the verdict susceptible to review, discussion, and improvement. Accordingly, this section lays out the judgments and assumptions that gird the financial benefit-cost analysis.

Exhibit 7-2
Relationship between overall evaluation and financial benefit-cost analysis

7.1.2 Measurement of financial costs as an input to cost-effectiveness analysis

IDAs are an example of "strong policy," that is, a single intervention with myriad benefits (Sherraden, 1999; Yadama and Sherraden, 1996). In addition to looking at multiple points of view, the overall analysis will attempt to capture multiple effects through cost-effectiveness analysis from the point of view of society. Whereas financial benefit-cost analysis compares financial benefits with financial costs, cost-effectiveness analysis compares quantitative results—both financial and non-financial—with net financial costs (financial benefits minus financial costs). Thus, the measure of financial costs from the financial benefit-cost analysis described here will serve as an input into the overall analysis.

For example, suppose that average financial benefits from the point of view of participants are $50 and that average financial costs are $100. Then suppose that IDA participation increases the probability of voting by 5 percentage points, the probability of talking to a neighbor by 3 percentage points, and the probability of expecting a child to attend college by 5 percentage points. By themselves, each effect might be small, and although the results cannot be added together mathematically, the results can be combined for purposes of decision-making. Furthermore, each effect did not cost $100 by itself; rather, all of the effects together cost $100. Thus the final evaluation will hinge not on the judgment of whether $50 is greater than $100 but rather on the judgment of whether $50 and more voting and more neighborliness and more hope for the future of children (and whatever other quantitative and qualitative effects are documented) are greater than $100.

The survey will capture changes between treatments and controls for the quantitative outcomes listed in Exhibit 7-3. The survey-measured impacts, in addition to the results from the financial benefit-cost analysis described below, will be incorporated into the overall analysis.

Thus, the cost-effectiveness analysis will include the financial benefit-cost analysis but will extend it to these quantitative outcomes whose effects cannot be measured in terms of dollars, as mandated in AFIA (414(b)(3) and 414(b)(4)).

7.1.3 Analysis framework: net present value of cash flows

The heart of the analysis is the estimation of the net present value of changes in cash flows due to IDAs for each of the six basic groups and then for society as a whole. In essence, the framework counts outflows of cash from a given stakeholder as a cost to that stakeholder and inflows of cash as a benefit to that stakeholder. Cash flows are discounted to account for the fact that they take place at different points in time. Benefits net of costs for society as a whole is the sum of benefits net of costs for the other six groups of stakeholders. The rest of this section describes the various assumptions needed to implement a net-present-value framework in the evaluation of IDAs.

Time frame

Assets are resources that last through time, and the effects of assets, like assets themselves, are likely to accumulate through time (Sherraden, 1991). Therefore, AFIA mandates that the financial benefit-cost evaluation encompass five- and ten-year time frames. Of course, IDAs probably have effects that last far longer than a decade, and the longer the time frame of evaluation, the more effects of IDAs will be considered. Still, policy choices will be made long before any evaluation could attain perfect knowledge of the effects of IDAs.

In fact, AFIA will come up for renewal before Congress in 2003. If the evaluation is to inform that vote, then the time frame for the evaluation will be from the moment of the baseline survey until 2003. Thus, the time frame in which real measurements will have been made will be shorter than the 5-years planned for the experimental site. Likewise, after 5 years, the mandated 10-year time frame will include real data only for the first 5 years.

Results may be extrapolated either from a time frame of less than 5 years to a time frame of 5 years, or from a time frame of 5 years to a time frame of 10 years. There are two simple ways to extrapolate results, and the evaluation will report results under both assumptions. Under extrapolation of levels, the total net present value of benefits net of costs in the short time frame is simply multiplied by the ratio of years in the long time frame to the years in the short time frame. For example, if net benefits in a 3-year time frame were !10, then net benefits in a 5-year time frame would be !10 @ (5/3) = !16.67. To go from a 5 years to 10 years would then be !16.67 @ (10/5) = !33.34. Thus, extrapolation by levels does not change the sign of net benefits and thus adds little to the policy process.

The second possible assumption is extrapolation of changes. In this case, the change in net benefits that takes place in final year of the time frame is assumed to be the change in net benefits that takes place in all future years beyond the time frame. For example, suppose annual net discounted benefits in a 3-year time frame were -6, -4, and -2. The total net benefit is -12, but the change in the net benefit in the final year of the time frame is (-4) - (-2) = 2. Given a change in net benefits of 2 units per year, then the assumed net benefits for year 4 would be -2 + 2 = 0, and the assumed net benefits for year 5 would be 0 + 2 = 2. The total net benefit for a 5-year time frame would be -6 + (-4) + (-2) + 0 + 2 = -10. For a 10-year time frame, total net benefit would be -10 + 2 + 4 + 6 + 8 + 10 = 20. Extrapolation by changes can switch the sign of net benefits; lengthening the time frame might change policy choices.

Whatever the time frame, it will start with the baseline survey at t = 0. Each period will last a year, and the analysis will end T years after the baseline survey.

Social weights of benefits and costs

To the individual or group that receives it, a dollar of net benefit is worth a dollar. To society as a whole, however, a dollar of net benefit may be worth more or less than a dollar, depending on which member in society receives it. For example, if society has a preference for the poor or disadvantaged, then a dollar of net benefit that accrues to an impoverished, non-white female probably has more social benefit than a dollar of net benefit that accrues to a rich, white male (Schreiner, 1999a). The analytical concept that describes social preferences across different people is the social-welfare function (Deaton, 1997).

To keep matters simple and because no one knows exactly what is the true social-welfare function, this analysis will assume that a dollar has the same social worth regardless of who receives it.

Discounting

Discounting matters for the financial benefit-cost analysis of IDAs for two reasons. First, the benefits and costs of IDAs do not take place at a single point in time. Unlike the purchase of a loaf of bread which entails one cash flow at one point in time, IDAs affect cash flows over an entire range of times, perhaps over decades in the life of a single person or over centuries in the lives of generations of a family. Second, cash flows that take place today are more important in a real sense than cash flows that take place sometime in the future. Even in the absence of inflation, resources would have a time value, in part because people know that they may die before the future comes, in part because of imperfect capital markets, in part because of uncertainty combined with risk aversion, and in part because imperfect human imagination tends to place more importance on current benefits and costs today than on future benefits and costs.

Thus, a dollar in the hand today is not worth half of two dollars in the bush tomorrow. In fact, a dollar today plus a dollar tomorrow is not two dollars of anything (Boulding, 1962). Resource flows at different times have different units, in much the same way as a pound of copper and a mile of copper wire have different units.

Discounting puts cash flows that take place at different times in a common unit so that they can be compared (or added or subtracted) meaningfully. In essence, the net-present-value framework to be employed discounts resource flows according to when they take place so as to make them comparable to resources at a single point in time.

The analysis takes the discount rate r as 10 percent per year in real terms for all years. Of course, no one agrees on the exact value of the true discount rate, but the United States government (U.S. Office of Management and Budget, 1972) and the World Bank (Belli, 1996) both use 10 percent. These are the two biggest entities in the world that conduct financial benefit-cost analysis. In practice, the question of the "correct" discount rate is often moot. Suppose, for example, that the results of financial benefit-cost analysis are used to select among alternative projects to be funded from a fixed budget. Then the choice of projects to fund is not affected by the choice of a discount rate as long as the same rate is used to evaluate all alternatives (Belli, 1996).

Given a discount rate r, the annual discount factor is * = 1/(1 + r). If a cash flow took place at the end of year t, then the relevant discount factor would be *t, where the t is not a notational superscript but rather a mathematical exponent. In fact, the analysis will not have access to information about the exact timing of cash flows within a year. A reasonable assumption is that the cash flows take place in a single lump halfway through the year (or, almost equivalently, in a constant stream throughout the year). In this case, the relevant discount factor is approximately *t-0.5 (Schreiner, 1997). Note that 0 # * # 1, so given a cash flow xt, then *t @ xt < xt. Furthermore, for any , > 0, *t+, @ xt < *t @ xt. This fits with the idea that a given cash flow now is worth more than the same given cash flow in the future. Furthermore, as the future cash flow takes place further and further in the future, it is worth less and less compared to the same cash flow in the present.

Apart from the "pure" time value of money reflected in discounting, inflation also changes the real purchasing power of a dollar between two points in time. To counteract the effects of inflation, all cash flows to be discounted will first be converted to constant-dollar units. Given a nominal dollar amount at time t (dt), the consumer price index at time t (CPIt), and the consumer price index at time T (CPIT), then the constant-dollar value of dt in units of dollars as of time T is dt @ (CPIT/CPIt) (Schreiner, 1999b).

Net present value versus return on human investment

This analysis is based on a net-present-value framework; the only other attempt to measure the financial benefits and costs of IDAs (Clones et al., 1995) uses a return-on-investment framework. What is the difference, and why choose net present value?

In return-on-investment analysis, the result is an annual rate of return, computed as ((Benefits ! Costs) / Costs) / Years (Brizius (1991), as cited in Clones et al., 1995). Return-on-investment analysis has three advantages. First, the formula is simple. Second, the output is a rate of return, and most people believe that they understand rates of return. Third, and not unimportantly, the name of the framework contains the word investment, which sounds better than cost. Although no one likes costs, few would dare to speak out against investments.

The main disadvantage of return-on-investment analysis is that it does not discount. Thus, for projects such as IDAs in which most costs are bunched early in the time frame and most benefits are bunched late in the time frame, return-on-investment analysis overestimates true net benefits. For short time frames, discounting may not matter much, but in long time frames, it does matter a lot. IDAs are most likely to be relevant in long time frames.

Thus, although the net-present-value framework is slightly more complex (because each cash flow is multiplied by a discount factor), it also produces a more meaningful output (discounted benefits net of costs). If the user of the analysis prefers to work with rates of return then the appropriate measure is not the annual rate of return produced by return-on-investment analysis but rather the internal rate of return produced in a net-present-value framework. (The internal rate of return is the discount rate that would make discounted net benefits exactly zero.) Also, because the net-present-value framework looks only at cash inflows or outflows, whether those flows are seen as "expenses" or "investments" is irrelevant.

Measurement of changes in cash flows

For both benefits and costs, the quantities that enter the net-present-value analysis are changes in cash flows caused by IDAs. For all stakeholders except participants, these quantities are simply the cash outflows to the IDA program or to IDA participants and the cash inflows from the IDA program or from IDA participants. Because none of these cash flows would have taken place in the absence of IDAs, the analysis assumes that the presence of IDAs caused the cash flows that did in fact take place.

For participants, the quantities that enter the net-present-value analysis are measured as the difference between the cash flows for the treatment group in one survey period compared to the previous survey period, minus the difference between cash flows for the control groups in one survey period compared to the previous survey period.

Changes in cash flows caused by IDA participation are measured as the average cash flow for the treatment group at the experimental site in a given survey period (xt) minus the average cash flow for the treatment group in the previous survey period (xt-1), minus the average cash flow for the control group in the same given survey period (ct) minus the average cash flow for the control group in the previous survey period (ct-1). In symbols, the change in cash flows caused by IDA participation is (xt ! xt-1) ! (ct ! ct-1). Thus the quantities that enter the analysis are difference-in-differences; the difference between the treatment and control groups of the difference in the cash flows for one group between two survey periods.

As a simple example, consider the measurement of the effect of the IDA program on the cash outflows into own IDA deposits during the first survey period. This flow is a cost from the point of view of participants because it is a cash outflow from the participant to the IDA account. (Later, of course, when participant make withdrawals from the IDA account, deposits become benefits.) For both the treatment group and the control group, the cash flow was zero in the baseline period before the IDA program, so x0 = c0 = 0. For the control group, the average cash outflows for IDA deposits for the time period of the first follow-up survey are still zero (c1 = 0) because controls, by definition, cannot make IDA deposits. Treatments, on the other hand, can and do make deposits into IDAs, so their average cash outflows are positive (x1 > 0). The change per participant in own IDA savings caused by the IDA program in the first follow-up period is then simply the amount deposited in the period, or (x1 ! 0) ! (0 ! 0) = x1. Of course, this is a simple example; in general, the average cash flows for the two groups in the two periods will not be zero.

Appropriateness of a framework based on cash flows

In the evaluation literature, the appropriateness of a framework based on discounted cash flows is unquestioned. The theory behind the framework is well-established, incontrovertible, and its use in practice is standard. Indeed, most stakeholders themselves tend to count their own benefits and costs largely—if often implicitly—in terms of cash flows.

For participants, however, a cash-flow framework may not be the best way to measure the benefits of IDAs. Evaluations, especially of the so-called manpower programs, focus almost exclusively on changes in employment and in wages. In his seminal work on assets and the poor, however, Sherraden (1991) argues that assets are much more than mere factors of production and stores of potential consumption. In his view, the ownership of assets may produce "asset effects", that is, non-economic psychological and social changes in expectations and behavior that improve long-term well-being. Indeed, the most oft-quoted passage of the book states that "while income feeds peoples' stomachs, assets change their heads." Thus, an ideal evaluation of IDAs would consider much more than just effects on employment and wages.

The dilemma—and the irony—is that the standard net-present-value framework measures the benefits and costs of IDA participation in terms of changes in cash flows, that is, in terms of income, not in terms of assets. Thus the cash-flow measure is indifferent between additional cash inflows that are saved to improve future well-being versus additional cash inflows that are consumed to improve current well-being. In other words, cash-flow measures completely ignore all effects of IDAs on asset accumulation, even though the possibility of asset accumulation and of non-economic "asset effects" is the chief reason why IDAs might be a better way to help the poor than, for example, simple increases in the amount of means-tested cash assistance.

Despite these issues, the evaluation of AFIA will use a cash-flow framework because that is the only way to compare net benefits for participants and for other groups of stakeholders and then to combine them all in a measure of net benefits for society as a whole. A conceptual framework that could guide attempts to measure "asset effects" still does not exist. Still, it is interesting to speculate about the rough contours of such a framework, one that could measure asset accumulation and its effects.

Income is a change in the level of resources in a given time frame, whereas assets are resources kept through time. Thus a measure of changes in asset accumulation (as opposed to a measure of changes in income) would explicitly incorporate the passage time in the unit of measurement. For example, the effects of IDAs on asset accumulation might be taken as the change in dollar-years of financial assets held in a year, where a dollar-year of assets is a dollar's worth of resources kept for a year. For example, $12 of assets kept for a month is equivalent to 1 dollar-year and $2 of assets kept for three months is equivalent to 0.5 dollar years. Dollar-years of assets can be discounted much like dollars of income are discounted (Schreiner, 1997). The framework would resemble a standard discounted-cash-flow framework, but the quantities in the net-benefit equation would be dollar-years instead of cash flows, and the final result would be discounted dollar-years instead of discounted dollars.

Such a framework would differentiate between the use of extra income to fund assets and the use of extra income to fund consumption. Furthermore, to detect whether "asset effects" are real, discounted dollar-years of assets derived from the framework could be compared with long-term non-economic psychological and social changes in expectations and behavior.

7.1.4 Experimental design considerations

The financial benefit-cost analysis will likely be confined to the experimental site. Even without the expense and complications of an experimental design, it is unlikely that the financial benefit-cost work could be extended to extra sites. To measure benefits and costs, the analysis requires longitudinal surveys of participants and non-participants (if not of treatments and controls) as well as site visits to the IDA program. It is highly unlikely that the evaluation budget could cover the costs of a financial benefit-cost evaluation at even two sites, to say nothing of evaluation at as many as 40 sites, as implied by AFIA section 414(a) that states that the demonstration projects should be evaluated "individually and as a group". Cost considerations dictate that the benefit-cost analysis component of the evaluation be confined to a single site, the site with the experimental design.

If there is no experimental design at any site, then the "experimental" group will be participants, and the "control" group will be non-participants. The weakness of this design is that differences in outcomes between treatments and controls may be due not to the IDA program but rather to differences between the two groups that existed before the IDA program came into existence.

7.1.5 Site selection

Careful site selection matters because in all likelihood only one site will be analyzed. For the purposes of financial benefit-cost analysis, the ideal site would have the following characteristics:

• Many participants. In ADD, recruitment of applicants for the treatment and control groups was difficult and time-consuming. Because a large sample size can only improve the analysis and because statistical theory cannot guide the choice of sample size other than to suggest that more is better, a site that can quickly enroll many applicants is best.
• Few donors. Cash flows from 20 donors are more difficult to track than cash flows from 2 donors. Likewise, the cash value of volunteer time is easier to track for fewer volunteers.
• An experienced, reputable organization. Cash-flow data at the program level are usually more reliable for organizations that are used to being formally accountable. In particular, a site that maintains formal financial statements and formal budgets is preferred.
• A single-purpose organization. The mixture of IDA programs with non-IDA program within the same organization complicates the analysis of IDA-program cash flows. Of course, most large, experienced organizations will not be single-purpose.
• Variation in key IDA design features. A site with variation in match rates, monthly savings goals, and non-IDA services such as financial-literacy classes will reveal more about optimal IDA design than will a site with a one-size-fits-all IDA contract. Ideally, randomization could be applied not only to the assignment of applicants to treatments or controls but also to the assignment of IDA-design features to treatments, but this possibility is unlikely. If two sites were to be analyzed, then they should be chosen based on the variation between them of key IDA design features.
• Location far from state lines. Benefits and costs for two state governments will be more difficult to measure than for one. For the same reasons, it would be more convenient for all participants to be in a single city rather than spread across several municipal jurisdictions; it is easier to measure the benefits and costs that accrue to one local government than it is to measure the benefits and costs that accrue to many local governments.
• Staff commitment to IDA rules. To test the effects of IDAs requires that staff not bend the rules to allow "approved" withdrawals for "unapproved" uses. If IDAs were expanded universally, discretion in the use of withdrawals would be allowed no more than discretion is currently allowed for Individual Retirement Accounts.
• Staff commitment and understanding of the goals and worth of financial benefit-cost analysis. Data collection will rely largely on cooperation from IDA staff. A lot of data is derived from MIS IDA, and data in MIS IDA is mostly self-reported by participants by way of IDA staff. Furthermore, site visits by evaluators will draw staff away from their normal duties and impose "extra" work on them. Cooperation and commitment can only follow from a clear understanding the purposes of the benefit-cost exercise.
  

 top of page

---

7.2 Data Collection Plan

The financial benefit-cost analysis will draw on six basic sources of data:

• Survey of treatments and controls
• MIS IDA monitoring instrument
• Government-program administrative data
• Desk review of tax laws
• Site visits to IDA programs
• Interviews with government and private donors

The survey instrument is described in detail in Chapter 5. For the purposes of the benefit-cost analysis, respondents must be surveyed at least twice, once at baseline and again later. More survey rounds are better than fewer. The ideal time between rounds is one year (to aid the accuracy of respondent recall) and should never exceed two years. The survey will capture changes in financial and non-financial outcomes for participants.

The MIS IDA Monitoring Instrument is built into MIS IDA. IDA staff update participant-account data monthly or quarterly, and they self-report resource inflows and outflows for the IDA program itself every six months.

In principle, it might be possible to use government-program administrative data to measure changes in the use of means-tested public assistance, and this data might be more accurate than survey data. Arrangements for access to administrative data from state and local governments, however, are likely to be time-consuming and thus expensive. Ultimately, the decision to attempt to gain access to administrative data will be based on the likely costs of such an attempt, and it seems likely that budget constraints will dictate the use of survey data exclusively.

Changes in taxes are a large part of financial benefits and costs for participants and for federal, state, and local governments. MIS IDA does not record tax payments, and survey respondents probably are neither able nor willing to give accurate responses. Even if MIS IDA did record tax payments, it would do so only for participants and not for members of the comparison group. Thus the analysis will estimate taxes based federal, state, and local tax law. When possible, these estimates will use already-estimated relationships between income or profits and tax paid.

Site visits will measure resource flows—both in cash and in kind—between private donors, government, and IDA programs. Although MIS IDA already records self-reports for flows in-cash and in-kind, experience suggests that regardless of the effort to self-report accurately, the conceptions of resource flows held by IDA staff rarely match perfectly with the conceptions required for the financial benefit-cost analysis.

Thus the site visit will function not as an audit but rather as a cross-check and as a clarification of definitions. Furthermore, upon examination of budgets, financial statements, and bank records, an evaluator may notice resource flows that the IDA staff forgot to include in the self-reported data. In particular, it is very easy to inadvertently overlook inflows in-kind. The annual site visit should last about one week. In time, IDA staff will learn what measurements are needed for the evaluation, and later site visits will require less time and effort.

The site visit will also serve to price in-kind resource flows received at a discount or as a gift. For example, evaluators will query landlords during the site visit about the market price of discounted office space, volunteers about the market value of their time, and program partners about the cost of free services provided to IDA participants.

Finally, interviews with government and private donors will act as cross-checks on disbursements to IDA programs as recorded in MIS IDA. These interviews, perhaps conducted by mail or phone, will also gather the data on the administrative costs of donors. Furthermore, taxable private donors will be asked about the tax write-off claimed for their contributions to IDA programs. Again, the purpose is not to audit but rather to ensure that all resource flows are recorded as accurately as possible for the purposes of the financial benefit-cost analysis.

Data at the program level should be gathered annually, even if the survey that gathers data at the participant level is administered only two years. Organizations work on annual cycles, so knowledge is freshest if collected each year.

 top of page

---

7.3 Data Analysis Plan

The analysis plan is described in detail in the accompanying Appendix G.

 top of page

---

7.4 Cost Estimate

This section presents the estimated costs for benefit-cost analysis as a component of the AFIA evaluation. Exhibit 7-4 shows these costs, by year.

The estimated costs are based on the following assumptions:

• The benefit-cost analysis will take place at the experimental site and will involve four annual visits to the site by an Associate-level researcher, for data collection. These visits are assumed to occur during January-March of 2001, 2002, 2003, and 2004.
• Planning will involve organizing personnel and coordinating work activities. This will involve approximately five person-days of effort at startup and follow-up planning during each of the annual data collection periods.
• At each site visit, the researcher will collect data on resource flows in and out of the IDA program. This involves a visit to the program site and thus expenses for airfare, lodging, ground transport, and meals.
• At each visit, data on resource flows in the previous calendar year are collected. The first visit requires 2 person-days planning, 6 person-days on site, and 8 person-days post-visit to compile and integrate results. Subsequent site visits will require less time because the IDA program will have learned better to collect the relevant data and to have it ready. Thus, the subsequent visits will take 3 days each, with 1 day for planning and 5 days post-visit.[24]
• Data collection will also include an annual "desk review" of tax laws at the federal, state, and local level. It also requires a review of IDA-related resource flows to and from public and private donors. In particular, it includes estimates of the funds disbursed to the IDA program and of the administrative time and expense used to manage relationships with the program. This annual review requires 3 person-days per year at the Intermediate staff level. Communication with government and private donors will require an additional 3 person-days per year.
• The actual computation of differences between treatment and control groups in their resource flows will involve statistical regressions as described in the Evaluation Design Plan. This analysis will take place in each year of the evaluation. Each round of processing will cover data collected for the previous year. In the first round, processing will require 30 person-days at the Associate level. Processing in the subsequent rounds will require 12 person-days.
• The findings from the benefit-cost analysis will address the financial benefits and costs from the points of view of seven groups of stakeholders: IDA participants, non-participants, the federal government, state and local government, employees of IDA programs, and society as a whole.
• Interim findings from the benefit-cost analysis will be presented in the September 2003 Interim Report. These findings will be based on program data covering the period through calendar year 2002 and on the first-round follow-up data from the experimental sample. The interim findings require 25 person-days of effort.
• The final benefit-cost analysis will be presented in the September 2004 Final Report. These findings will be based on program data covering the period through calendar year 2003 and on both rounds of follow-up data from the experimental sample. This report will build upon the interim findings and will require 20 person-days of effort.

Exhibit 7-4 Benefit-Cost Analysis -
Estimated Costs by Year

Item			Rate	Year 1		Year 2		Year 3		Year 4		Total
				Units	Cost	Units	Cost	Units	Cost	Units	Cost	Units	Cost
Staff Labor
		Class I - Senior		40	$4,138	6	$621	106	$10,966	66	$6,828	218	$22,552
		Class II - Associate		392	$19,541	172	$8,574	272	$13,559	316	$15,753	1152	$57,427
		Class III - Intermediate		48	$2,244	48	$2,244	48	$2,244	4	$187	148	$6,918
		Class IV - Junior		0	$0	0	$0	0	$0	0	$0	0	$0
		Class V - Clerical		0	$0	0	$0	0	$0	0	$0	0	$0
	Labor Inflation Adjustment		4%		$1,037		$933		$3,342		$3,867		$9,180
	Subtotal Staff Labor				$26,960		$12,372		$30,111		$26,634		$96,077
	Fringe and Overhead				$29,699		$13,629		$33,170		$29,341		$105,838
Total Staff Labor				480	$56,658	226	$26,001	426	$63,281	386	$55,975	1518	$201,915
Other Direct Costs
	Survey Direct Costs				$0		$0		$0		$0		$0
	Travel				$2,348		$1,769		$1,769		$1,769		$7,655
	Telephone and Computer				$1,913		$939		$1,671		$1,537		$6,060
	Duplicating and Delivery				$84		$84		$91		$91		$348
	Payments to Respondents				$0		$0		$0		$0		$0
	ODC Inflation Adjustment		3%		$130		$170		$327		$426		$1,054
Total Other Direct Costs					$4,475		$2,962		$3,858		$3,823		$15,118
G&A and Fee					$15,980		$7,571		$17,550		$15,631		$56,732
Total Estimated Costs					$77,113		$36,533		$84,689		$75,429		$273,765

 top of page

---

References

Belli, Pedro. (1996) Handbook on Economic Analysis of Investment Operations, Operations Policy Department, World Bank: Washington, D.C., pbelli@worldbank.org.

Bendick, Marc Jr.; and Mary Lou Egan. (1987) "Transfer Payment Diversion for Small Business Development: British and French Experience", Industrial and Labor Relations Review, Vol. 40, No. 4, pp. 528-542, egan@gwu.edu.

Berger, Allen N.; and Timothy H. Hannan. (1998) "The Efficiency Cost of Market Power In the Banking Industry: A Test of the 'Quiet Life' and Related Hypotheses", Review of Economics and Statistics, Vol. 80, No. 3, pp. 454-465.

Boulding, K.E. (1962) "Economics and Accounting: the Uncongenial Twins", in W.T. Baxter and Sidney Davidson (eds.), Studies in Accounting Theory, Richard D. Irwin: Homewood, Illinois.

Brizius, Jack A. (1991) What Are the Returns on Human Investments? Exploring a New Way of Budgeting, Washington, D.C.: Corporation for Enterprise Development.

Browne, Lynne Elaine; and Joshua Gleason. (1996). "The Saving Mystery, or Where Did the Money Go?", New England Economic Review, Federal Reserve Bank of Boston, pp. 14-27.

Clones, Daphne; Friedman, Robert; Grossman, Brian; and Cicero Wilson. (1995) The Return of the Dream: An Analysis of the Probable Economic Returns on a National Investment in Individual Development Accounts, Washington, D.C.: Corporation for Enterprise Development, ISBN 1-883187-07-9, cfed@cfed.org.

Deaton, Angus. (1997) The Analysis of Household Surveys: A Microeconometric Approach to Development Policy, Baltimore and London: Johns Hopkins University Press, ISBN 0-8018-5254-4.

Gittinger, J. Price. (1982) Economic Analysis of Agricultural Projects, Second Edition, Baltimore: Johns Hopkins University Press, ISBN 0-8018-2912-7.

Inter-American Development Bank. (1994) "Technical Guide for the Analysis of Microenterprise Finance Institutions", Microenterprise Division, Washington, D.C.

Rosenberg, Richard; Christen, Robert Peck; and Brigit Helms. (1997) "Format For Appraisal of Micro-Finance Institutions", Washington, D.C.: Consultative Group to Assist the Poorest, www.worldbank.org/html/cgap/tools/format.doc.

Schreiner, Mark. (1999a) "Aspects of Outreach: A Framework for the Discussion of the Social Benefits of Microfinance", Center for Social Development Working Paper No. 99-3, Washington University in St. Louis, gwbweb.wustl.edu/users/schreiner/.

_____. (1999b) "A Cost-Effectiveness Analysis of the Grameen Bank of Bangladesh", Center for Social Development Working Paper 99-5, Washington University in St. Louis, gwbweb.wustl.edu/users/schreiner/.

_____. (1999c) "Self-Employment, Microenterprise, and the Poorest Americans", Social Service Review, Vol. 73, No. 4, pp. 496-523.

_____. (1997) A Framework For the Analysis of the Performance and Sustainability of Subsidized Microfinance Organizations With Application to BancoSol of Bolivia and to the Grameen Bank of Bangladesh, unpublished Ph.D. dissertation, The Ohio State University, gwbweb.wustl.edu/users/schreiner/.

Schreiner, Mark; and Jacob Yaron. (1998) "Two Measures of the Social Cost of Subsidized Development Finance Institutions", manuscript, World Bank.

Sherraden, Michael. (1999) "Youth Service as Strong Policy", paper presented at the International Conference on Youth Service, San José, Costa Rica, January 2000,

_____. (1991) Assets and the Poor: A New American Welfare Policy, Armonk, NY: M.E. Sharpe, ISBN 0-87332-618-0, csd@gwbssw.wustl.edu.

United States Office of Management and Budget. (1972) "Discount rates to be used in evaluating time-distributed costs and benefits", Circular No. A-94 (rev.), Washington, D.C.

Yadama, Gautam N.; and Michael Sherraden. (1996) "Effects of Assets on Attitudes and Behaviors: Advance Test of a Social Policy Proposal", Social Work Research, Vol. 20, No. 1, pp. 3-11, sherrad@gwbssw.wustl.edu.

 top of page

---

Notes

[24] These estimates are based on a pretest conducted by the Center for Social Development during April 18-21, 2000 at the Community Action Project of Tulsa County (CAPTC), the experimental IDA program site for the American Dream Demonstration. [Return to Text]