MicroTest Performance FY2005 An Analysis of Lending Program Data Welcome to a summary analysis and presentation of selected MicroTest program performance statistics for FY2005. The data were collected in 2006, from 69 MicroTest member microenterprise development organizations. The report is broadly organized around the AEO Numeric Microenterprise Accreditation Standards. In addition, additional information is provided around program scale and self-sufficiency two subjects of considerable interest among practitioners within the field. Generally, statistics are presented in multiple ways. First, the statistics will often be presented in summary, using standard statistical summary values of mean, median, minimum, and maximum. Next, a chart will graphically show the distribution of values among the participating programs. Finally, in some cases, the connections between statistics will be shown using statistical correlations. In addition, in selective cases the statistics are presented according to different subsets of programs, to illustrate different approaches or perspectives among programs.
Lending Programs Analysis Industry Comparisons: MicroTest Summary Statistics and Distributions AEO Numeric Accreditation Standards for Lending Programs (Scale, Portfolio Quality, Efficiency) Industry Topics of Interest: For many programs, the financial return to the program from the loan fund is very important. 2 Lending programs offer loans to program clients for their microenterprises. Often the lending programs offer some form of training and technical assistance along with their loans. This credit-related training and TA is often designed to prepare borrowers for taking on debt for their business, and to improve the likelihood of repayment. The AEO accreditation standards evaluate credit programs by looking at the scale of lending, the portfolio quality (in terms of delinquencies, restructures, and writeoffs), and the efficiency of operations. Some of these standards will vary, depending upon the methodology of the program. Also included in this analysis is understanding the self-sufficiency of the credit program.
Lending Programs in Data Set 27 (39%) MT programs report being credit-led 50 (72%) MT programs report disbursing 4,845 loans for $38,630,610 in FY2005 Approximately 23% of clients served by MT programs in FY2005 received loans directly from the participating programs The total cost of providing lending services was $21,687,206 (n=46) 3 About 39% of the programs in the MicroTest FY2005 dataset described themselves as credit-led, meaning that the majority of their program activities involved lending. About 72% of MT members reported making loans. While lending was a significant activity of the programs, less than one-quarter (23%) of the clients served in FY2005 received a loan that year.
Lending Scale Portfolio Scale Measures Total Capital Available Number and $ Amount of Loans Outstanding Deployment Ratio Loans Disbursed in Fiscal Year MT Summary Statistics AEO Numeric Accreditation Standards 4 Lending scale can be measured in multiple ways.
Total Capital Available Al l Credit-led Training-led Mean $1,719,583 $2,970,566 $349,459 Median $670,683 $1,406,742 $197,200 Top Perf. $2,047,026 Low $7,100 $187,500 $7,100 High $13,292,798 $13,292,798 $1,805,666 Total $75,661,657 $68,323,012 $7,338,646 N 44 23 21 5 Here are the summary statistics for total capital available, with detail by the methodology of the program. As you can see there are significant differences between credit-led and training-led programs. The 44 participating programs have a total of over $75 million available for credit. Of this, the vast majority (90%) is held by the 23 credit-led programs.
Portfolio Size and Methodology 30 25 Median 20 15 10 5 0 < $1M $1 M => $2M $2M => $3M $3 M => $4M $4M => $5M $5M => $6 M $6 M => $7 M $7 M => $8M $8M => $9M $9 M => $10M $10M => $1 1M > $ 11M Cred it-le d Training-led 6 The ski slope distribution of portfolio size shows programs clustering around $1M, with most training-led programs having portfolios less than $1M. From there we then see a series of programs, all credit-led, that are sized significantly larger than the rest.
Deployment Ratio Al l Credit-led Training-led Mean 73% 76% 46% Median 57% 72% 49% Top Perf. 79.8% Low 0% 43% 0% High 98% 98% 93% N 44 23 21 7 Deployment ratio is the percentage of the portfolio lent out to clients. Money lent out is more likely to be fulfilling the mission of the programs, and, as we shall see later, is more likely to help promote program self-sufficiency. Overall, 73% of the funds available had been lent out to microentrepreneurs at the time of survey.
Deployment Ratio and Methodology 10 9 Median 8 7 6 5 4 3 2 1 0 <0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.5 0.6 0.6 0.7 0.7 0.8 0.8 0.9 > 0.9 Cred it-led Tra ini ng-l ed 8 The distribution of the deployment shows quite a range of deployment ratios. Overall, the training-led programs (with a couple of exceptions) have the lowest deployment ratios. Most of the credit-led programs are in the highest categories. While higher deployment ratios mean that more money is meeting its programmatic purpose, very high deployment ratios signal limited resources available for new borrowers, and potentially greater risk to the programs.
Portfolio Outstanding, Large and Small Scale Lenders $50,000,000 $45,000,000 $40,000,000 $35,000,000 $30,000,000 $25,000,000 $20,000,000 $15,000,000 $10,000,000 $5,000,000 $0 1 2 3 4 9 Here we see the size of the aggregate outstanding portfolios of programs divided into quartiles, from the smallest (1) to the largest (4). As we can see, clients are more likely to be served by large programs than small ones. The 11 largest lending programs have over $44M in outstanding loans, or about 4.2 times the outstanding loans of all other lenders. The 12 smallest have about $622,671. Looked at another way, the half of the programs with the smallest portfolios outstanding (the two smallest quartiles of lenders) have outstanding loans of $2.8M. In comparison, the half of the programs with the largest portfolios (the two largest quartiles) have nearly 19 times the loans outstanding of the smaller half with approximately $52.6M outstanding.
Unused loan capacity in FY2005 $14,00 0,0 00 $12,00 0,0 00 $10,00 0,0 00 $8,00 0,0 00 Dollar Value of Loans Outstanding Idle Funds $6,00 0,0 00 $4,00 0,0 00 $2,00 0,0 00 $0 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 10 Where is the unused loan capacity? As you can see from the distribution above there is considerable variation across programs. Programs with bigger portfolios overall hold a larger share of the unused funds, with the top 10 largest portfolios accounting for 63% of the unused funds. The program with the largest portfolio has unused funds greater than the aggregate of the smallest 50% of the dataset.
Loans Disbursed in FY2005 All Programs Credit-Led Training-Led Mean 161 270 33 Median 54 111 22 Top Perf. 142 Low 0 29 0 High 1777 1777 180 Total 8044 7284 760 N 50 27 23 11 Here we will look at the number of loans disbursed by programs in 2005. As can be seen, there is a big difference in the summary statistics of lending volume between credit-led and training-led programs.
AEO Accreditation Standards for Lending Scale Credit-led programs should disburse at least 40 loans in a year Training-led programs should disburse at least 12 loans per year 12 The AEO Accreditation Standards for lending scale are different for credit-led and training-led programs. Overall, 76% (38 of 50 lending programs) met this AEO Numeric Accreditation standard.
Loans Disbursed in FY2005 by Program Methodology 25 20 15 Median 10 5 0 <20 20 40 40 60 60 80 80 100 100 120 120 140 140 160 160 180 180 200 200 Credit-led Training-Led 13 Here is the distribution by methodology.
Portfolio Quality Indicators Delinquencies (Portfolio at Risk) Restructures Losses (Write-Offs) 14 The next group of measures involves the quality of the loan portfolio, specifically concerning the likelihood of loans being repaid. Intuitively, a lender would want all of these rates to be as low as possible, to maintain capital and maximize earned income. Ideally the rates should be zero. However, there are costs associated with minimizing risk and loss, and programs have a mission of lending to riskier borrowers. As we will see, among the most sophisticated lenders these rates are managed at levels that balance repayment risk with mission and the costs of origination and collection. This presentation will look at each of the indicators separately. However, in analyzing portfolio quality, it is best to look at all 3 indicators together.
AEO Numeric Standards for Portfolio Quality Portfolio at risk at or below 16% Loan loss rate at or below 7% Restructured loan rate at or below 12% 15 Here we see the AEO Accreditation Standards for each of the portfolio quality indicators. Two-thirds of the programs (33 programs, or 66% of lenders) met the standard for delinquencies. None had missing data for the statistic. For loan loss rate, Two-thirds of the programs (34, or 68% of lenders) met this standard. None had missing data for the statistic. Sixty-six percent of the lenders met the standard for restructured loan rate, with 6 programs (12% of lenders) having insufficient data to report on this measure.
Portfolio at Risk Al l Credit-led Training-led AEO STD 16.0% Mean 10% 9% 22% Median 11% 10% 13% Top Perf. 3% Low 0% 2% 0% High 83% 43% 83% N 50 27 23 16 Portfolio at risks deals with those loans that are more than 30 days delinquent. Here we see the summary statistics across all programs.
Portfolio at Risk and Methodology 25 20 Median 15 10 5 0 <10% 10% - <20% 20% - < 30% 30% - < 40% 40% - < 50% 50% - < 60% 60% - < 70% 70% - < 80% 80% - < 90% >90% Credit-led Training-led 17 Here we see a chart of the distribution. For both credit-led and training-led programs, the largest category includes those with PARs of less than 10%. For credit-led programs, the next category is still fairly high. Some training-led programs have very high rates of delinquencies.
Loan Loss Rate Al l Credit-led Training-led AEO Std 7.0% Mean 6% 6% 5% Median 4% 5% 0% Top Perf. 0% Minimum 0% 0% 0% Maximum 61% 20% 61% N 50 27 23 18 Loan loss rates reflect write-offs, or loans that are not valued as assets on a program s books. The program has not forgiven the loan, and in some cases the loans may eventually be repaid. By writing off the loan, instead of forgiveness they ve internally stopped accruing interest on the loans and on assuming recovery of principal on their financial statements. Here we see the LLR summary statistics. Note the zeros. In some cases a zero LLR represents complete confidence in and a practice of repayment. In other cases, however, it may represent inconsistent or non-existent write-off practices.
Loan Loss Rate and Methodology 20 18 16 14 Median 12 10 8 6 4 2 0 0% 0 %- < 5% 5% - <10 % 10% - < 15% 15% - < 20 % 2 0% - < 25% 2 5%- < 30% 30%- < 3 5% 35%- < 40% >= 4 0% Cr edit-led T raining- led 19 Here is the distribution of loan loss rates. Most programs are clustered on the lower end. And there are some outliers way off to the right.
Median Portfolio Quality Indicators for Various Program Initiatives All Lenders Large Lenders High OSS High % Start-Up Portf olio at Risk 11.2% 6.6% 10.3% 11.2% Loan Loss Rate 4.3% 5.5% 5.3% 1.5% Restructured Loan Rate 3.8% 4.1% 3.5% 0% 20 Here we see the median portfolio quality indicators of programs with different program emphases. Large lenders (lenders disbursing 100 or more loans in FY04), who one could assume have more standardized products and more sophisticated lending processes, on average have lower loss and delinquency rates. Interestingly, while delinquencies are considerably lower than the median of all lenders, losses are actually higher. Those with a mission of targeting a high percentage of start-up businesses (high percentage of start-ups are defined as those with 45% or more of their outstanding portfolio loaned to start-ups), where one would assume there would be more risk, do show higher risk rates in all categories. One would assume that lenders pursing self-sufficiency (High OSS is defined as an OSS of 40% or more) would have very low rates. After all, they want to be repaid. However, their delinquencies are about average, and their losses are higher. This could be explained by the fact that for very cost-conscious programs, the financial benefit of recovery is balanced by the costs of increased due diligence at the front end, and recovery work with the borrower at the back end.
Lending Efficiency Operational Cost Rate Cost to manage each dollar outstanding in the portfolio (includes current and prior year loans) Formula: Total Credit Program Operating Expenses / $ Average Outstanding Portfolio AEO Numeric Accreditation Standard 1.4 or less for programs with fewer than 50 outstanding loans 0.7 or less for programs with 50 or more outstanding loans 21 Lending efficiency is measured in two ways in MicroTest. The cost per loan measures the cost to make each new loan disbursed in a fiscal year. It measures efficiency in terms of the number of loans, regardless of loan size, given the costs of the lending program. There is not an AEO accreditation standard for cost per loan, and the statistics for this indicator are not included in this presentation. The Operational Cost Rate (OCR) measures the cost to manage each dollar outstanding in the lending portfolio. It measures efficiency in terms of the total amount of loans outstanding, regardless of the number of loans, and can include loans from current and prior years. The AEO Accreditation Standard varies according to the scale of loans outstanding with two standards: One with total loans outstanding at 70% of lending program costs, and a second with total loans outstanding at 140% of the costs of the lending program. 58% of lending programs (29 programs) met their respective OCR standard. Twelve programs (24%) did not have sufficient data to determine compliance.
Operational Cost Rate Al l Credit-led Training-led Mean 0.34 0.30 0.92 Median 0.45 0.31 0.92 Top Perf. 0.26 Low 0.17 0.17 0.29 High 20.82 2.91 20.82 N 38 23 15 22 This is the standard that the AEO accreditation standards uses to measure credit program efficiency. It is the ratio of the total program costs over the total amount of loans outstanding. For programs with 50 or fewer loans, the standard is 1.4. This means that loans outstanding should be at least 70% of total program costs. In other words, if your lending program costs you $100,000 to operate, you need to have at least $70,000 in loans outstanding, on average, in a fiscal year. For programs with more than 50 loans, the standard is 0.7. In this case the average amount of loans outstanding should be more than 1.4 more than the cost of the loan program. Using the above example, with a $100,000 lending program, average loans outstanding should exceed $140,000.
OCR and Methodology 14 12 Median 10 8 6 4 2 0 <0.3 0.3 0.6 0.6 0.9 0.9 1.2 1.2 1.5 1.5 1.8 1.8.1 2.1 2.4 2.4 2.7 2.7 3 3 Credit-led Training-led 23 Here is the distribution, by methodology. Again we see credit-led programs among the highest in efficiency (those furthest to the left). However, we also see very efficient training-led lending programs, and very inefficient credit-led programs.
OCR Correlations Portfolio outstanding Dollar amount of loans outstanding Not Correlated Current year scale activity Number of loans disbursed Not Correlated Dollar amount of loans disbursed Not correlated Deployment ratio Not strongly correlated -0.534 (p=0.001) Correlation disappears when two largest OCRs are remov ed 24 What factors tend to be associated with more efficient lending programs, e.g. those with lower OCRs? The only variable correlated with OCR is Deployment ratio. Even then, when the two largest OCRs (18 and 20) are removed, the correlation disappears.
Portfolio Size and OCR 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00 <$ 200,000 $200K - <$ 400K $400K - <$600 K $600K - <$800 K $800K - <$ 1M $1M - < $ 1.2M $1.2M - < $1.4M $1.4M - < $1.6M $1.6M - < $1.8M $1.8M - < $2M >$2M Median OCR 25 Here we compared portfolio size (total capital available) with OCR. Note the big change in OCR after the threshold of $200,000 in portfolio size. After this point the slight trend of lower OCR with increasing portfolio size is comparatively modest.
Financial Return from Loan Portfolio Operational Self Sufficiency: Income from loan fund / Cost of lending program Affected by lending scale, efficiency, portfolio quality, pricing 26 In the next set of statistics we are going to look at the financial return to lending programs from their loan portfolio. Some programs are pursuing aggressive selfsufficiency strategies in order to reduce their need for subsidy. We will also look at the relationship of various lending performance statistics to OSS. Intuitively one would assume that lending scale, efficiency, portfolio quality, and pricing would all have an affect on OSS.
Operational Self Sufficiency Al l Credit-led Training-led Mean 43% 51% 9% Median 21% 37% 7% Top Perf. 45% Low 0% 4% 0% High 100% 100% 54% N 43 23 20 27 Operational self-sufficiency looks at the ratio of the amount of earnings from the loan fund to the cost of the lending programs. Note the significant differences between training-led and credit-led programs.
OSS and Methodology 14 12 10 Median 8 6 4 2 0 <10% 10% - <20% 20% - < 30% 30% - < 40% 40% - < 50% 50% - < 60% 60% - < 70% 70% - < 80% 80% - < 90% >90% Cr ed it- led Tr aining -led 28 Here is the graphic representation of that. Most of the training-led programs are clustered to the left. And some of the credit-led programs are clearly in different territory.
OSS Correlations Measure Scale (# of loans disbursed) Value 0.643 (p=0.000) Comment Correlated Scale ($ amount of loans disbursed) 0.705 (p=0.000) Strongly Correlated Overall Program Efficiency (Cost per client) Efficiency in Making New Loans (Cost per loan) Ratio Cost of Lending Program/Loans Outstanding (OCR) Pricing Portfolio Quality - Write-Offs Po rtfoli o Qu al i ty - Del in qu en ci es -0.153 (p=0.327) -0.320 (p=0.050) -0.363 (p=0.025) 0.325 (p=0.041) 0.508 (p=0.001) 0.629 (p=0.000) No Correlation Weak Correlation Weak Correlation Weak Correlation Correlated Correlated 29 What program indicators are correlated with high rates of Operational Self- Sufficiency? Here we look at some likely suspects. Making more loans is correlated with higher rates of operational self-sufficiency. Since income is derived on loans outstanding, having more loans outstanding results in more income. The dollar amount of the loans is more significant than the volume of loans. The efficiency measures are either weakly correlated or not correlated. Pricing is only weakly correlated. We ll look at that in more detail momentarily. Portfolio quality is correlated.
OSS and $ Amount of Loans Outstanding 120% 100% 80% 60% 40% 20% 0% 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 30 Here is the distributed relationship between the amount of loans outstanding and OSS. Note the upward trend in the majority of, but not all, cases.
Effective Portfolio Yield Al l Credit-led Training-led Mean 13% 13% 8% Median 9% 10% 7% Top Perf. 13% Low 0% 5% 0% High 32% 32% 26% N 49 27 22 31 Here are some summary statistics on Effective Yield from the lending portfolio. What is shown here is the ratio of the financial return from the portfolio in terms of interest and fees to the total amount of the lending portfolio. It is based upon what is achieved, rather than what is estimated up front (e.g. the stated interest rates on loan products). Note the difference between credit-led and training-led programs.
Effective Portfolio Yield and Methodology 12 10 Median 8 6 4 2 0 <0.02 0.02 0.04 0.04 0.06 0.06 0.08 0.08 0.10 0.10 0.12 0.12 0.14 0.14 0.16 0.16 0.18 0.18 0.20 0.2 Credit-led Training-led 32 Here is a chart to show the graphic distribution.
OSS and Effective Yield 100% 80% 60% 40% 20% 0% 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 33 Here is a distribution showing the relationship between OSS and effective yield. In this graph, the each bar represents a program, with those with the highest effective yields to the right. Generally we see that higher effective yields result in higher OSS rates. However, we see more cases where this generalization does not work than we did with the correlation with the amount of loans outstanding. This is less a strong of fit than the number of loans outstanding, but still shows some correlation.
For More Information Contact MicroTest Tel: (202) 736-1089 E-mail: microtest@aspeninstitute.org On the web: http://fieldus.org/li/microtest.html