Inclusive Finance Lecture Notes 2: Impact Evaluation in Microfinance 2024-2025 PDF

Inclusive Finance K. Czura Microcredit Miracle? Impact Inclusive Finan...

Inclusive Finance K. Czura Microcredit Miracle? Impact Inclusive Finance evaluation How to Measure Impact? Lecture Notes 2: Impact Evaluation in Micronance Econometric foundation Treatment & Control Group Analytical concepts Dr. Kristina Czura Microcredit Impact in Practice Access to Finance in University of Groningen (RUG) India Consumer lending Outlook microcredit Block 1.2 - Term 2024 / 2025 Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 1 The microcredit miracle? Inclusive Finance K. Czura Microcredit  High expectations and the Nobel peace prize for Miracle? Muhammad Yunus and the Grameen Bank in 2006 Impact evaluation  countless anecdotes for transformative impact How to Measure Impact?  early studies showed positive impact of microlending on Econometric foundation borrower welfare (e.g. Pitt and Khandker (1998)) Treatment & Control Group ⇒ Immense expectations for microcredit to reduce poverty Analytical concepts  Actual impact of microcredit on the welfare of the poor Microcredit Impact in  recent studies (2015) criticized early studies for their Practice Access to Finance in methodology ⇒ randomized evaluations India Consumer lending  microcredit impact views more nuanced now Outlook microcredit  BUT: Latest evidence shows large negative eects if Self-assessment micronance is reduced (Breza and Kinnan (2021)) Digression: Recap Empirical  wages, household earnings and consumption ⇓ Methods General OLS  due to ⇓ aggregate demand and business investments Di-in-Di Instrumental variable (IV) References 2 The microcredit miracle I Inclusive Finance K. Czura Clients' anecdotes Microcredit Miracle?  Rashidan Bibi: Impact  "My family and I used to sell milk worth $165 per evaluation month but with Livestock Finance from Asasah we now How to Measure Impact? sell milk worth $330 per month. I am Rashidan Bibi Econometric foundation from a small town Raiwind, Pakistan. I have been living Treatment & Control Group in this town for the past 18 years. My son and I sell Analytical concepts milk. Before becoming a loan member at Asasah we Microcredit Impact in had two cows, and then we bought another cow with Practice Asasah's credit. Since then every year we add a cow to Access to Finance in India our business by acquiring loans from Asasah. Currently, Consumer lending Outlook microcredit we own six cows. The increase in number of cows has Self-assessment served to an increased milk production and increased our Digression: prots. With the increase in income I have purchased a Recap Empirical Methods fridge, washing machine and a television for my house. General OLS Now we eat better and live well. Thank you Asasah." Di-in-Di Instrumental variable "http://www.micronanceconnect.info/stories (IV) References 3 The microcredit miracle II Inclusive Finance K. Czura Clients' anecdotes Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 4 Why measure impact? Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure  Cucumbers can kill! Impact? Econometric  99.8% of dead people ate cucumbers... foundation Treatment & Control Group  Knowing the causal impact of micronance (or any Analytical concepts other development program) is important to Microcredit Impact in  guide evidence-based policy Practice Access to Finance in  distribute scarce resources to the most ecient use India Consumer lending  improve micronance and other interventions and Outlook microcredit development programs Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 5 How to measure impact? Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 6 The ideal experiment Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 7 Reality interferes Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact?  We can only observe each individual in one state of the Econometric foundation world at the same time! Treatment & Control Group ⇒ Solution: Analytical concepts Microcredit  Estimate what would have happened to y in the Impact in absence of the treatment T (our explanatory variable x) Practice Access to Finance in  = the counterfactual India Consumer lending  Candidates for a counterfactual Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 8 Candidate counterfactual I Inclusive Finance K. Czura Candidate I: Before Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 9 Pre-post design Inclusive Finance K. Czura Concept Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 10 Pre-post design Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control  What are complications with this comparison? Group Analytical concepts  Before-after comparison only works if nothing else Microcredit Impact in changes absent the treatment. Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 11 Pre-post design Inclusive Finance K. Czura Complications Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 12 Candidate counterfactual II Inclusive Finance K. Czura Candidate II: Untreated Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 13 Treated vs. untreated Inclusive Finance K. Czura Concept Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 14 Treated vs. untreated Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation  What are complications with this comparison? Treatment & Control Group Analytical concepts  Treated vs. untreated only works if the two groups are Microcredit the same in all relevant respects. Impact in Practice  But often the treated and untreated are dierent! Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 15 Treated vs. untreated Inclusive Finance K. Czura Complications Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 16 Treated vs. untreated Inclusive Finance K. Czura Complications Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 17 What if there is no perfect match? Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 18 Distribution of outcome - Control group Inclusive Finance K. Czura Projected outcome for each individual based on distribution of outcome Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 19 Distribution of outcome - Control group Inclusive Finance K. Czura Density function Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 20 Distribution of outcome - Treatment group Inclusive Finance K. Czura Projected outcome for each individual based on distribution of outcome Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 21 Distribution of outcome - Treatment group Inclusive Finance K. Czura Projected outcome for each individual based on distribution of outcome Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 22 Comparing distributions Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 23 Comparing distributions Inclusive Finance K. Czura Analysis Microcredit Miracle? Impact evaluation  Compare averages How to Measure Impact? Econometric  yC = α foundation Treatment & Control  yT = α +β Group Analytical concepts  β is the dierence between the averages of the treated Microcredit and untreated group Impact in Practice  Regressions Access to Finance in India  yi = α + β Ti + εi Consumer lending  where Ti = 1 if individual i is treated, 0 if not Outlook microcredit Self-assessment  and εi is the amount by which i's outcome diers from the average of their group Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 24 Comparing distributions Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 25 Econometric implications Inclusive Finance K. Czura Microcredit Miracle? Impact  Simple OLS regression model evaluation How to Measure Impact? Econometric yi = α + β xi + εi (1) foundation Treatment & Control Group Analytical concepts  where y is the outcome (dependent variable) and x is Microcredit the independent variable. Impact in Practice  Problems with regular OLS regressions: Endogeneity! Access to Finance in India  Reverse causality Consumer lending Outlook microcredit  Omitted variable bias Self-assessment ⇒ We cannot establish a causal relationship between x Digression: and y! Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 26 Problems with OLS Inclusive Finance K. Czura Biased estimators  Endogeneity (E (u|x) ̸= 0) of the explanatory variable Microcredit Miracle? can lead to a biased estimator. Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) Source:Wooldridge (2013) References 27 Problems with OLS I Inclusive Finance K. Czura Main sources of endogeneity Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation  Relationship between x and y Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 28 Problems with OLS II Inclusive Finance K. Czura Main sources of endogeneity Microcredit Miracle? Impact  Omitted variable bias evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 29 Problems with OLS III Inclusive Finance K. Czura Main sources of endogeneity Microcredit  Reverse causality Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) ⇒we need exogenous variation to identify causal eects!!! References 30 Micronance impact Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 31 Control group Inclusive Finance K. Czura Microcredit  Recall: We can only observe each individual in one state Miracle? of the world at the same time Impact evaluation ⇒ use statistics on a treatment and a control group to How to Measure identify the counterfactual Impact? Econometric foundation  What is a good control group? Treatment & Control Group  Control should be as identical as possible to the Analytical concepts treatment group Microcredit Impact in ⇒ Identical in observable and unobservable characteristics Practice  Only dierence between treatment and control group Access to Finance in India should be the fact that the treatment group has credit, Consumer lending Outlook microcredit and the control group not Self-assessment ⇒ Control group should not be aected by spillovers Digression:  What happens if we do not nd a good control group? Recap Empirical Methods  Over- or under-estimate impact of intervention (see General OLS Di-in-Di biased estimator) Instrumental variable (IV) References 32 Selection biases Inclusive Finance K. Czura What can cause problems identifying treatment and control group? Microcredit  Program placement bias Miracle?  Treatment areas with a program dier from control Impact evaluation areas without this program How to Measure Impact?  Program areas dier in observable characteristics Econometric foundation because the program-designers had specic criteria in Treatment & Control Group mind → most social interventions are targeted! Analytical concepts  Program areas may also dier in unobservable Microcredit characteristics from non-program areas Impact in Practice  Endogenous selection Access to Finance in India  Individual participation is usually voluntary Consumer lending Outlook microcredit  Participants dier from non-participants in observable Self-assessment and unobservable characteristics Digression:  Nonrandom attrition Recap Empirical Methods  Over time people drop out of a program General OLS Di-in-Di  These dropouts are likely not random: those that do Instrumental variable (IV) not benet, leave the program References 33 Selection Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 34 Treatment vs. Control group Inclusive Finance K. Czura Analysis and assignment to treatment and control group Microcredit  Random assignment to treatment and control group Miracle? ⇒ Gold standard to measure impact Impact evaluation How to Measure https://www.youtube.com/watch?v=hLhXCo2jQkg Impact? Econometric foundation ⇒ There should be NO dierences between treatment and Treatment & Control Group control group due to random assignment to the groups! Analytical concepts  Treatment and control group have on average the Microcredit same individual characteristics (balance test) Impact in Practice  Since the treatment and the control group are Access to Finance in comparable on average before the treatment, we can India Consumer lending interpret the dierence as the causal impact of x (our Outlook microcredit treatment T) on y. Self-assessment Digression:  To be on the safe side (and increase statistical power): Recap Empirical Methods Control for dierences of treatment and control group General OLS Di-in-Di before and after the intervention Instrumental variable (IV) ⇒ Dierence-in-Dierences (DiD) estimation References 35 First best: Random assignment Inclusive Finance K. Czura Microcredit  Random assignment Miracle?  randomly assign individuals to treatment and control Impact group of a policy or program (like in medical trials) evaluation How to Measure  What does random mean? Impact? Econometric foundation ⇒ Randomized controlled trial (RCT) Treatment & Control Group  Why use random assignment? Analytical concepts  excludes any correlation between x (our treatment T) Microcredit Impact in and the error term u Practice  implies that the treatment assignment is orthogonal to Access to Finance in India other covariates that may inuence the outcome Consumer lending Outlook microcredit  treatment and control group are on average the same Self-assessment  balanced sample across treatment and control group Digression: (check this in balance test) Recap Empirical Methods ⇒ We have identied a causal eect of x (treatment T) General OLS on y Di-in-Di Instrumental variable (IV) References 36 First best: Random assignment Inclusive Finance K. Czura Simple Dierence Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 37 First best: Random assignment Inclusive Finance K. Czura  Cook-book procedure Microcredit Miracle?  Randomly assign treatment and control group.  Test whether randomization was successful, and the Impact evaluation treatment assignment is orthogonal to other covariates. How to Measure Impact?  Balance of Sample Test: there is no dierence in Econometric foundation individual characteristics between treatment and Treatment & Control Group control group. Analytical concepts Simple  If yes : analyze the data using OLS for a Microcredit Dierence evaluation between treatment and control Impact in Practice group. Access to Finance in India  If no: use Dierence-in-Dierence estimation Consumer lending Outlook microcredit methods to control for pre-treatment dierences. Self-assessment  Problems: Digression:  Implementation Recap Empirical Methods  External validity General OLS Di-in-Di  No general equilibrium eects, only zoom-in on policy Instrumental variable (IV) or program References 38 Dierence-in-Dierences (DiD) Inclusive Finance K. Czura Concept Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 39 Dierence-in-Dierences (DiD) Inclusive Finance K. Czura Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts  What are complications with this comparison? Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 40 Dierence-in-Dierences (DiD) Inclusive Finance K. Czura Microcredit Miracle?  Figure 9.1 above makes some assumptions: Impact evaluation  measured and unmeasured attributes do not change How to Measure over time Impact? Econometric  broad economic changes are the same for both groups foundation Treatment & Control ⇒ Dierences across groups that are not related to the Group Analytical concepts treatment are constant over time Microcredit  Same holds for Dierence-in-Dierences estimation! Impact in Practice  Estimates are only correct if observable and Access to Finance in India unobservable characteristics do not change dierentially Consumer lending Outlook microcredit over time and if both groups are aected similarly by Self-assessment broad economic changes. Digression:  In practice: attributes of groups will change over time Recap Empirical and broad economic changes may dier per group. Methods General OLS Di-in-Di Instrumental variable (IV) References 41 Compliance with treatment assignment Inclusive Finance K. Czura Perfect compliance Microcredit Miracle? Perfect compliance Impact evaluation  All people who are assigned How to Measure Impact? to the intervention, receive it Econometric foundation Treatment & Control  All people who are assigned Group Analytical concepts to not receiving the Microcredit intervention, do not receive Impact in Practice it. Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 42 Compliance with treatment assignment Inclusive Finance K. Czura Imperfect compliance Microcredit Miracle? Imperfect compliance Impact  How should we treat people evaluation How to Measure assigned to the intervention Impact? Econometric who do not participate in it? foundation Treatment & Control  Ignore them? Put them in Group Analytical concepts control group? Leave Microcredit them in treatment group? Impact in Practice  What about people in the Access to Finance in India control group who do Consumer lending Outlook microcredit participate in the Self-assessment intervention? Digression:  Ignore them? Put them in Recap Empirical Methods treatment group? Leave General OLS them in control group? Di-in-Di Instrumental variable (IV) References 43 Measuring impact Inclusive Finance K. Czura What eects can we measure? Microcredit Miracle?  Can we compare the treated Impact and the untreated? evaluation How to Measure Impact?  These groups look like Econometric they might be dierent! foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending  Can we compare the treated Outlook microcredit in the treatment group and Self-assessment the untreated in the control Digression: group? Recap Empirical Methods  These groups look like General OLS Di-in-Di they might be dierent! Instrumental variable (IV) References 44 Intent to Treat (ITT) Estimate Inclusive Finance K. Czura Compare all those in the treatment group with all those in Microcredit the control group Miracle? Impact evaluation  Randomization ensures that How to Measure Impact? statistically these groups are Econometric foundation the same!!! Treatment & Control Group Analytical concepts  Intent to Treat (ITT) Microcredit estimate Impact in Practice ⇒measures the average Access to Finance in India impact of being assigned Consumer lending Outlook microcredit to the treatment group Self-assessment Digression: Recap Empirical Methods  Treatment eect on the treated (TOT): use IV General OLS estimation and scale the ITT estimate by the inverse of Di-in-Di Instrumental variable (IV) the fraction of the sample who are aected by the References assignment 45 Microcredit impact Inclusive Finance K. Czura Well-identied studies - JPAL overview Where credit is due Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 46 Microcredit impact Inclusive Finance K. Czura Well-identied studies - JPAL overview Where credit is due Microcredit Miracle? Impact evaluation How to Measure Impact? Econometric foundation Treatment & Control Group Analytical concepts Microcredit Impact in Practice Access to Finance in India Consumer lending Outlook microcredit Self-assessment Digression: Recap Empirical Methods General OLS Di-in-Di Instrumental variable (IV) References 47 Access to Micronance in India Inclusive Finance K. Czura Methodological approach: Program expansion with random program placement Microcredit Miracle? Impact evaluation How to Measure  Randomizing by individual is not always feasible Impact? Econometric  implementation diculties foundation Treatment & Control  spillover eects Group Analytical concepts  Randomize at the next highest cluster level using an Microcredit expansion plan of a microlender Impact in Practice  identify clusters (villages, neighborhoods, etc.) into Access to Finance in India which the microlender would like to expand its business Consumer lending Outlook microcredit  randomly assign which cluster is serviced (treatment Self-assessment group) and which not (control group) Digression:  random phase in: after the impact study has been Recap Empirical conducted, former control clusters are serviced Methods General OLS Di-in-Di Instrumental variable (IV) References 48 Access to Micronance in India Inclusive Finance K. Czura Example Banerjee et al. (2014) I Microcredit Miracle?  Rely on a program expansion plan of Spandana Impact evaluation  in 2005, 52 of 104 poor neighborhoods in Hyderabad How to Measure Impact? were randomly selected for the opening of a branch Econometric foundation (treatment) Treatment & Control Group  in the remaining 52 neighborhoods no bank branch Analytical concepts opened until the end of the study (control) Microcredit Impact in  after the study, Spandana opened bank branches in the Practice Access to Finance in control neighborhoods India Consumer lending  Household data of 6,850 households was collected in Outlook microcredit two waves Self-assessment  15 to 18 months after the random program expansion Digression: Recap Empirical plan Methods  3 to 3.5 years after the random program expansion plan General OLS Di-in-Di Instrumental variable (IV) References 49 Access to Micronance in India Inclusive Finance K. Czura Example Banerjee et al. (2014) II Microcredit Miracle? Impact  Findings evaluation How to Measure  take-up of microcredit increased by 8.4 percentage Impact? Econometric points, mostly substituting informal borrowing foundation Treatment & Control  small business investment and prots of pre-existing Group Analytical concepts businesses increased, but no changes in the likelihood of Microcredit owning a business Impact in Practice  consumption did not change on average, but Access to Finance in India consumption of durable goods increased and of Consumer lending temptation goods decreased Outlook microcredit  no changes in health, education, women's empowerment Self-assessment Digression: ⇒ Access to microcredit helped some, but not most Recap Empirical Methods households General OLS Di-in-Di Instrumental variable (IV) References 50 Consumer lending in South Africa Inclusive Finance K. Czura Methodological approach: Experimental credit scoring Microcredit  Experimental credit scoring designs rely on exogenous Miracle? Impact shocks to the loan approval process evaluation How to Measure  Credit scoring: Impact? Econometric  approve or reject applicants based on their likelihood of foundation Treatment & Control default Group Analytical concepts  Randomization in experimental credit scoring: Microcredit  randomize clients "on the bubble" to either receive or Impact in Practice not receive credit Access to Finance in