Unformatted text preview: 9/2/11 PADP 8670 POLICY ANALYSIS I Cost-Benefit Analysis Angela Fer7g, Ph.D. Today's plan Ra1onale for Cost-Benefit Analysis Rela1ve Efficiency & the Compensa1on Principle Kaldor-Hicks Compensa1on Principle Conduc1ng a CBA Iden1fying relevant impacts Valuing the costs and benefits in terms of dollars Market goods Nonmarket goods (e.g. clear air, life) Discoun1ng for 1me and risk Comparing CBA to cost-effec1veness analysis 1 9/2/11 Rela1ve Efficiency Rela%ve efficiency: the comparison of one alloca1on with that of another An alloca1on is Pareto Superior if at least one person is beWer off and no-one is worse off More interes1ng cases is when many people are made much beWer off but a few people are made worse off. Fundamental test of rela1ve efficiency is the compensa%on principle Essen1ally the compensa1on principle asks "Can the gainers from a change compensate the losers and s1ll come out ahead?" This idea serves as the cornerstone of Cost-Benefit Analysis (CBA) The Purpose of a Rela1ve Efficiency Standard Pareto Criteria is too restric1ve Why? We rarely have a circumstance where we can change a policy and some gain and no one is harmed. We want an analy1c rule that will allow us to capture large gains even when some small harm may be done to others. This measure is ethically problema1c and it is important to understand why and how. 2 9/2/11 Kaldor-Hicks Compensa1on Principle Kaldor-Hicks is the standard metric by which we evaluate policy An alloca1ve change increases efficiency if the gainers from the change are capable of compensa1ng the losers and s1ll coming out ahead. Each individual gain or loss is defined as the value of a hypothe1cal monetary compensa1on that would keep each individual indifferent to change. Kaldor-Hicks principle: controversial Buried inside Kaldor-Hicks are a series of equity judgments. Ra1onales for K-H as only criteria Some analysts want to ignore these and adopt the K-H as the only criteria for policy choice. Hope that we can deal with equity as a separate concern Concerns for equity are unfounded eventually if we adopt all the policies where K-H criteria holds - then all will gain Equity concerns are adequately represented in the policy process However, if a rich person gains $1.10 and a poor person loses $1, there is poten1al for Pareto improvement, but given that there is diminishing u1lity of wealth, the u1lity of the losses may be much greater than the u1lity of the gains because of how they are distributed. 3 9/2/11 Conduc1ng a CBA First step: iden1fy relevant impacts Whose perspec1ve maWers? Societal perspec1ve: all-inclusive, or may include geographic or ci1zenship restric1ons Business/Client perspec1ve: only costs & benefits relevant to the business/client Iden1fy and categorize costs and benefits Costs: nega1ve changes in social surplus Benefits: posi1ve changes in social surplus Free Clinic Ar1cle Discussion What perspec1ve does this CBA take? What would need to be included if it had a broader perspec1ve? What are categorized as costs and what are benefits? Why isn't the change in ALL hospital costs the es1mated benefit? 4 9/2/11 Conduc1ng a CBA Second step: Valuing the costs and benefits Let's focus on valuing changes in consumer welfare 3 ways to assign a dollar value 1. Calculate the change in consumer surplus using the demand curve P t Pconsumer P* D Qtax Q* Q S + t S Valuing Costs & Benefits 2. Compensa1ng Varia1on the size of the income effect necessary to maintain the same u1lity under the new price regime as before. Y Compensa1ng varia1on X Tax causes a price increase 5 9/2/11 Valuing Costs & Benefits 3. Equivalent Varia1on the size of the income effect necessary to maintain the same u1lity under the old price regime as the new. Y Equivalent varia1on X Tax causes a price increase Comparing the 3 methods CV and EV are exact measures of the welfare change; only differ based on what price level is used New prices compensa1ng varia1on Original prices equivalent varia1on The consumer surplus measure always gives a value between the CV and EV calcula1on 6 9/2/11 Exercise: Change in CS due to Rent Control Suppose that in the absence of controls, eqbm rent is $8000/year and quan1ty is 2 million apartments Assume that the government imposes a price ceiling of $7000/year for all apartments. This reduces the quan1ty of apartments rented to 1.8 million. Based on the graph, determine the effects on consumers. Calculate the actual $ value of the change in CS. Rent (thousands of dollars per year) Exercise (con1nued) S 9 8 7 D 0 1.8 2.0 Quan1ty of apartments (millions) 7 9/2/11 What if the public policy involves goods/ services that are not traded in a market? Then, we do not have a demand curve or an indifference curve to es1mate the change in CS Big problem for policy analysis because many public policies relate to nonmarket goods (environment, school quality, security, etc.) 3 approaches: Hedonic price models Survey assessments (con1ngent valua1on) Ac1vity surveys Valuing nonmarket goods 1. Hedonic Price Models: es1mate the nonmarket good by comparing market goods (one with the nonmarket aWribute and one without) Examples: Nonmarket good=school quality, compare 2 equivalent houses that only differ by quality of school district Nonmarket good=value of life, compare 2 equivalent jobs that only differ by probability of death (riskiness) Nonmarket good=air quality, compare 2 equivalent homes that only differ by quality of air Difficult to do (hard to find equivalent homes/jobs, requires a lot of informa1on) Many nonmarket goods can't be valued this way (e.g. na1onal defense no equivalent to compare it to) Limita1ons: 8 9/2/11 Value of Sta1s1cal Life (when the nonmarket good is life) Hedonic price models are osen used to value a "sta1s1cal life" Examples: Jobs that involve a risk of fatal injury by .001 percentage point raise wage by $6600/year 1000 people take job w/ .001 risk 1 person dies 1000*$6600 = $6.6 million paid for 1 to die Higher speed limit increases fatality rates by 35% but saved 125,000 hours per lost life Avg wage = $12.33 125,000*$12.33 = $1.54 million saved per fatality Valuing nonmarket goods 2. Con7ngent valua7on: just ask people how much they would be willing to pay to obtain a nonmarket good Examples: Would you be willing to pay $1/day to prevent x from occurring, thus preserving y? Suppose that a trust fund for the protec1on of x were on the next ballot and it would cost your household $y every year, would you vote in favor of it? People may not tell the truth (social norms, no real $ involved) Survey may be answered by those most interested in the nonmarket good, and other survey research problems Limita1ons: 9 9/2/11 Valuing nonmarket goods 3. Ac7vity Surveys: ask people about their ac1vi1es and what they have to invest to do these ac1vi1es, and then value their investments to es1mate their willingness to pay Examples: How osen do you go to X park? How many miles is X park from your home? Es1mate annual travel costs (gas, wear and tear, average wage value of 1me) = value of regional park for one household How far do people travel to get to a cleaner beach? Compare travel costs to various beaches and es1mate difference is costs by beach cleanliness. Need subs1tute sites if conduc1ng a comparison (like beach example) Those who value certain sites may choose to live nearby (have low travel costs but high values) Measures recrea1onal quality, but used to value environmental quality, and they may not be the same Limita1ons: Discussion of Exxon Valdez ar1cle 10 9/2/11 Net Present Value (or Discoun1ng) General formula: t indexes years and T is final year V is the nominal dollar value at 1me t r is the discount rate NPV = t =T Vt t t =1 (1 + r) Example 1: What is $100 given to me in 1 year worth to me today (assume r=0.05)? $100 NPV = = $95.24 1.05 Example 2: What is $300 given to me in 3 installments over the next 3 years worth to me today? NPV = $100 $100 $100 + + = $95.24 + $90.70 + $86.38 = $272.32 1.05 1.05 2 1.05 3 Relevance to CBA Discoun1ng is used for comparisons of different streams of benefits and/or costs over a number of periods/years. Assume discount rate=20%. Example: Investment A: returns $20 at end of Year 1 and $20 at end of Year 2 Investment B: returns $12 at end of Year 1 and $29 at the end of Year 2 A provides a total of $40 and B provides a total $41 so should we choose B? 11 9/2/11 Related concept: Conver1ng $ into the Same Base Year The NPV discussion assumes that all of the future values are not inflated by expected infla1on. However, when we use es1mates from the past to predict what our future costs or benefits will be, we must be careful to convert all dollars into the same base year. Example: If the cost of a doctor's visit in 2000 is $45, what is the cost of a doctor's visit in 2005? Use the all item Consumer Price Index (CPI), or appropriate component, to adjust costs: CPI medical care component in 2000 = a CPI medical care component in 2005 = b Dr visit in 2005 = Dr visit in 2000*b/a Expected Value Costs and benefits may not be known with certainty Can compare some aspects that are known with certainty to others that are uncertain by calcula1ng expected value of costs and benefits General formula: EV = iVi
i=1 N i indexes uncertain values (costs or benefits) N is the number of uncertain values V is the nominal dollar value of the cost/benefit if it occurs is the probability that the cost/benefit occurs 12 9/2/11 Group Exercise Read the Prevent Child Abuse America handout Review the calcula1ons carefully and try to find an error There are several but only one (big one) that you have the info to figure out. Cost-effec1veness analysis (CEA) vs. CBA CEA compares costs of achieving objec1ve of 2 alterna1ves Cost effec1veness ra1o =
Cost1 - Cost 0 Outcome1 - Outcome0 Difference from CBA: CEA does not convert benefit to dollars. where numerator is in $s but denominator is in life- years or some non-$ units. Some people (usually non-economists) are not comfortable with valuing nonmarket goods (especially life), so this avoids approach that. 13 9/2/11 Example Over a 5 year span: Annual cervical cancer screening increase life expectancy by .005 years One screening every 5 years increases life expectancy by .003 years Costs: $150 per screening in medical costs and lost produc1vity (2000 $) Assume 5% discount rate What is the cost-effec1veness ra1o of annual screenings rela1ve to one screening every 5 years? For next 1me Read Ch. 5 Case Study: Go to: hWp://www.ksgcase.harvard.edu/ Get the case called "Case Wars"; it is case #716.0; it requires payment ($3.45) Read it in prepara1on for class discussion Homework 4 14 ...
View Full Document
- Fall '10