
Robert Townsend on Development, Poverty, and Financial Institutions 03/14/2011
What this covers
This is a conversation between Russ Roberts and MIT economist Robert Townsend, built mostly around Townsend's household surveys in Thailand and a village study in India. Townsend asks how well informal arrangements and formal institutions handle risk. In the Indian village data, most households were insulated from income shocks, and only the poorest and wage earners passed through roughly 20 cents on the dollar of income swings. That fits Milton Friedman's Permanent Income Hypothesis, though Townsend notes that smoothing depends on borrowing, lending, and insurance arrangements that may be cultural. The weakness shows up at the long-run investment margin: some households earn real returns on assets of 40 to 50 percent while capital fails to flow to them, a gap Roberts finds puzzling given that such households could profitably borrow at 20 to 30 percent.
The conversation then moves to measurement. Townsend argues that national income accounts treat households only as consumers and suppliers of labor and savings, which misses household production. He also argues that cross-sectional data can mislead about mobility and inequality, because demographic shifts and entry into and exit from work change who is being counted. He uses this to dispute claims that income gains went mostly to the top 1 percent, and he prefers panel data that follow the same households. Related topics include the limits of Kuznets-style inevitability about inequality, and decomposition studies of occupation, education, and financial access.
The more contested material is on policy. Townsend credits financial deepening with about 75 percent of the movement in Thai total factor productivity over 30 years, because finance reallocates existing resources instead of creating new ones. He discusses the Thai village fund and evaluations of microfinance that found mixed results. He argues against replacing local moneylenders with subsidized cheap credit, and he says open entry matters because incumbent banks rarely innovate. He also agrees with Roberts that US policy channeled trillions into housing.
Townsend argues that household-level panel data and the deepening of formal and informal finance (more borrowers, savers, and branches) drive productivity and welfare gains, and that policymakers and researchers often misread aggregate data and misallocate capital.
- Panel data tracking the same households reveal transitions that cross-sectional medians and quintile snapshots hide
- Financial deepening explains about 75% of the movement in Thai total factor productivity over 30 years
- Villages informally pool risk and smooth consumption, but credit constraints block high-return investment, and well-intended interventions can undercut local finance
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Risk-sharing in villages does not translate into long-run investment: return on assets varies enormously across households, with some earning real returns of 40 to 50 percent per year while others earn low returns, and capital does not flow from low-return to high-return uses, which signals a poorly functioning financial system.
“the rates of return are so large as to raise doubts at least among people who have not studied developing economies you can have real rates of return that are in the order of 40 50 percent per year”
As transport costs fall and the financial system improves, intra-national trade grows like trade between countries (Portugal and England): labor-abundant rural areas with low wages export labor-intensive goods to capital-rich Bangkok, so rural wages rise gradually as infrastructure improves.
“as a transportation cost fall as the financial system improves you get these predictions that people for for intra-regional trade that people are more used to thinking about in terms of inter-country trade”
Incumbent banks have little incentive to innovate, and when researchers present case studies of underserved areas, banks regard those areas as too risky and decline to serve them; this is why the system must stay open so new entrants can innovate.
“if banks don't have an incentive to innovate then if you bring up a case study from an area they do they don't service there inevitably going to regard it as very risky and and not do it”
As financial systems and transport improve, transaction costs fall and intra-regional trade expands, so rural areas with abundant labor export labor-intensive goods to Bangkok, raising rural wages gradually rather than all at once because infrastructure improves slowly.
“as a transportation cost fall as the financial system improves you get these predictions that people for for intra-regional trade”
In an Indian village study, most households pooled risk remarkably well, achieving an efficiency standard in which consumption is largely insulated from income shocks; only the poorest, and those who earn wages rather than farm, pass roughly 20 cents on the dollar of income fluctuations into consumption, far less than the 70 cents or more one would expect in the absence of insurance.
“the relatively poor the poorest among the poor and those they with wage earnings as opposed to engage in agriculture actually were somewhat vulnerable but not anywhere near 70 cents on the dollar”
In developing countries, informal financial transactions among family, friends, moneylenders, gifts, and remittances can be as large as or larger than measured transactions in the formal financial sector, even though each individual transaction is small.
“those financial transactions are just huge they're typically small you wouldn't think it would mount up too much but it often is as big or bigger than the measured transactions”
Microfinance in India went into crisis because successful microfinance institutions became profitable, went public, and drew political backlash, and then over-indebtedness and suicides in Andhra Pradesh led the state to issue a loan repayment moratorium that damaged the culture of repayment and drove investors away.
“the provincial government the state I guess in mother possession issued a moratorium on repayment of loans so that which basically destroys the culture of repayment”
Cross-sectional snapshots (such as income quintiles or medians) can give a misleading picture of mobility and inequality, because demographic changes such as the large rise in US divorce rates and the increase in the number of households change who is being measured; panel data that track the same households over time are needed.
“there have been a lot of changes demographically in the United States over the last four decades huge increase in divorce rate a big increase in the number of households as a result so that we're looking at changes in mobility or quintiles snapshots give a very different picture”
Households largely smooth consumption against idiosyncratic income shocks rather than letting consumption track income, consistent with Friedman's permanent income hypothesis, though smoothing relies on borrowing, lending, and insurance that may be cultural.
“it could well be they have ways of stabilizing consumption and consumption is largely immune to these large idiosyncratic income fluctuations”
National income accounts treat households only as suppliers of labor and savings and as consumers, which is inappropriate for developing economies where households produce much output themselves; household production not sold in markets is hard to measure, so growth and inequality may be mismeasured.
“the conception of households and the national income accounts would be as suppliers of labour and savings suppliers of factors of production and in turn the demanders of of consumer goods and and that's definitely inappropriate for developing economies”
Very high returns on capital in poor households, if they were real, should attract outside investors, so the fact they do not suggests a poorly functioning financial system and the unexploited gains from trade come from credit constraints rather than from lack of opportunity.
“if such households exist out there surely there would be investors getting them the money but the answer seems to be by and large no”
Thai total factor productivity moves with financial deepening because finance does not create new land, labor, or capital but reallocates them, moving savings from inefficient households who should not run their own farm or business to those who should, so that lenders and savers share in the gains.
“it's not creating more factors of production it's reallocating them it's taking money through lending and interest bearing deposits from those inefficient households who shouldn't be running their own farm and business”
Households with very high returns on assets could profitably borrow at 20 to 30 percent, so the large gap in returns implies very large unexploited gains from trade in credit, which Townsend finds puzzling.
“there appears to be very large unexploited gains from trade there”
Entry into and exit from the labor force changes the sample composition of cross-sectional data: people not working are coded as zero and dropped, then appear once they find jobs, so medians can be distorted by changes in the size and composition of the pool.
“we take someone who's not working we call them a zero we don't put him in the sample and they get a job suddenly they're in the sample”
Overlabeling financial institutions by prescribing what each kind should do, rather than letting them coordinate, can cause serious problems, which echoes Hayek's point that economics shows how little designers know about what they can design.
“you can definitely cause some problems or as FA Hyack said the curious task of economics is to illustrate to men how little they really know”
In rural agricultural settings, household production that never reaches the market is hard to measure, which makes it difficult to gauge true growth and inequality; the World Bank's Living Standards Measurement Surveys include extensive questions on household production.
“household production that is stuff that isn't brought to market is going to be hard to measure and therefore it might be very hard to get a feel for what actual growth really is”
Decomposition studies that attribute changes in inequality to occupation transitions, education, and access to the financial system now quantify which sources are large or small, offering a map for where to look even though they do not prove causation.
“people have been doing decompositions of the sources of changes of inequality in to say occupation transitions or education or access to the financial system ... it's not a proof but it's it kind of tells you where to start looking”
Snapshots of formal financial institutions (branch counts, balance sheets, funds mobilized) give a distorted picture of developing-country finance; in rural areas commercial banks are often absent, and informal finance (moneylenders, family loans, gifts, remittances) can be as large as or larger than measured formal transactions.
“those financial transactions are just huge they're typically small you wouldn't think it would mount up too much but it often is as big or bigger than the measured transactions in the formal financial sector”
Measuring growth and inequality in agricultural economies is hard because much income is produced and consumed within households and never reaches the market, so household production can be underrecorded in national income accounts.
“household production that is stuff that isn't brought to market is going to be hard to measure and therefore it might be very hard to get a feel for what actual growth really is or inequality”
A simulation of Thailand's last 30 years, built on Townsend's economic models and household data, shows that financial deepening explains about 75% of the movement in macro total factor productivity.
“we can actually explain about 75% of the movement and macro total factor productivity with this deepening of the financial system”
Snapshots of existing formal financial institutions, such as branch counts, balance sheets, and funds mobilized, give a distorted picture of a developing country's financial system, because commercial banks are often absent in rural areas and the real financial activity is informal.
“the you know snapshot pictures of the say existing financial institutions in a developing country provide at best a distorted picture”
Data quality is a central obstacle in studying development: many developing countries lack reliable income data, and South Korea, a highly successful economy, lacked detailed household surveys until relatively recently, so its transition from poor agriculture cannot be traced at the household level.
“the irony Korea South Korea that we know is quite successful for decades doesn't have detailed household surveys until relatively recently”
Changes in labor force participation, such as women with children entering the workforce in the US over the last half of the twentieth century, make medians hard to interpret, because people previously coded as zero income enter the sample and shift the median even when the median is otherwise robust to outliers.
“if someone who's not working we call them a zero we don't put him in the sample and they get a job suddenly they're in the sample”
Financial systems do a good job smoothing transient consumption shocks but do much worse at financing longer-run investment, so households remain sensitive to cash-flow timing when they invest.
“the system doesn't work nearly so well for these longer run investment possibilities”
Informal financial markets such as kinship networks and local moneylenders are by and large helpful in Thailand, and policies replacing local moneylenders with cheap subsidized credit are probably a big mistake because the cheap credit is unsustainable and undercuts local systems.
“trying to replace local money lenders with sources of cheap credit that in my mind is probably a very big mistake not only is the cheap credit not sustainable”
Financial systems smooth consumption and protect well-being against bad luck, but their more important role for growth is in long-run investment, which the system serves poorly: households' investment is sensitive to cash flow, and the financial system fails to fund high-return long-run projects.
“one of the big findings and one might say puzzles is that the system doesn't work nearly so well for these longer run investment possibilities”
In an Indian village study (Risk and Insurance in Village India), households largely achieve the efficient risk-sharing benchmark, and poor households are not unable to organize mutual insurance as policymakers had presumed.
“which show that for the most part with some exceptions households in a village we're doing remarkably well at achieving this efficiency standard”
Financial deepening in Thailand is measured on the extensive margin: the share of households with a transaction at a government agricultural bank or commercial bank in the previous month rose threefold from about 7 to 8 percent in 1976 to a much higher level by 1996, a period of very high growth.
“basically only seven or eight percent of the household dance answered the question affirmatively but over the years between 76 and say 96 a period of very high growth for Thailand that number rises by threefold”
The Thai village fund raised investment in agriculture and in repair of vehicles used for trade, and consumption rose too, arguably because households had a reliable borrowing source and needed fewer precautionary buffer stocks.
“investment actually does go up there's investment in agricultural activities in repair of vehicles that are used for trade consumption also goes up arguably because they now have a reliable source of borrowing and they don't need to have as many buffer stocks”
Kuznets curves (inequality rising then falling with development) are not inevitable; there is broad agreement that countries need not experience rising inequality, and decomposition studies can quantify how occupation transitions, education, and financial access drive changes in inequality.
“I think there is pretty much agreement that we don't have to experience whose Nets curves and rising periods of inequality or not inevitable”
Milton Friedman's permanent income hypothesis predicts that people smooth consumption over time against windfalls and shocks, and this is largely borne out by data, for example in US temporary tax rebates that tend to be saved rather than spent.
“Milton Friedman in his theory the consumption function made that argument right and it is largely borne out I think by data and certainly in the United States it comes up in policy discussions because temporary tax rebates tend to get saved”
Some US commentators claim income has grown dramatically while most of the gain goes to the top 1 percent, but Townsend argues this reflects a distortion from comparing different people in cross-sectional data rather than the same households over time.
“there's a claim out there I'm that I don't agree with but there's a claim that average income per capita income the United States grown dramatically”
The poorest households and those with wage earnings (rather than agriculture) are more vulnerable, with consumption fluctuations on the order of about 20 cents for each dollar of income fluctuation, rather than the full pass-through seen without insurance.
“the poorest among the poor and those with wage earnings as opposed to engage in agriculture actually were somewhat vulnerable but not anywhere near 70 cents on the dollar”
Thailand's previous government set up a village savings and loan in every village by seeding it with a grant, and because villages vary in household count, the per-household treatment varied, which allowed the researchers to identify the effect of this financial innovation on investment and consumption.
“the previous government established a local savings and loan for every single village in the country by seeding it with a grant”
Supporting the informal financial market is arguably a bad policy because it addresses symptoms, so policymakers must first find and remove the underlying distortions rather than adopt policies that deal only with symptoms.
“we have to sort of get to the bottom of where the distortions were without adopting policies that really dealt with the symptoms and not the underlying disease”
Well-intended bipartisan policy that treats home ownership as a right created incentives for federal and local institutions to expand mortgage lending to people who were not qualified to borrow, which fueled a construction boom in suburban and rural areas.
“housing has been thought to be some not exactly a right but it would be a good thing if many more people and low-income people could could own their own home ... many people who really weren't qualified to to borrow money who had mortgages”
The US Panel Study of Income Dynamics (PSID) gives a very different picture of the last 30 to 40 years than the economy-wide median does, but it asks few consumption questions, so researchers must splice it with the Consumer Expenditure Survey (a cross-section) to assess consumption.
“the PSID the panel study of income dynamics gives a very different picture of the last 30 or 40 years compared to the economy as a whole measured by say the median”
Evaluations of microfinance in Hyderabad by Banerjee and Duflo found it was not a resounding success: many people do not borrow, some borrowers use loans for consumption, and only a subset of entrepreneurs expand businesses.
“there have been evaluations of some microfinance lending my colleagues Banerjee and DUPLO have done this in Hyderabad and again it's it's not a resounding success for all the borrowers”
Replacing local moneylenders with subsidized cheap credit is probably a big mistake: the cheap credit is not sustainable and undercuts the local financial systems that are, by and large, helpful even though they are not uniform.
“the notion that somehow money lending is evil and I'll and unfortunately has taken the route of trying to replace local money lenders with sources of cheap credit that in my mind is probably a very big mistake”
US financial markets channeled trillions of dollars into housing over the last 10 to 15 years through policy errors, and researchers have misallocated capital rather remarkably, a claim Townsend agrees with.
“I do think it's true and I agree with the the example of housing”
Well-intended political goals of expanding home ownership, bipartisan across administrations, created incentives for institutions and programs that pushed mortgages to borrowers who were not qualified, and the resulting construction boom in suburbs and rural fields was a hindsight-visible mistake.
“for well intended reasons perhaps both through Democratic and Republican administrations housing has been thought to be some not exactly a right”
Panel data (following the same households over time) is essential for measuring income mobility and consumption, and the US has very little of it, relying mostly on the Panel Study of Income Dynamics, which asks few consumption questions in some years.
“we don't have very much of it here in the United States we have some and but for those listening you know panel data”
Wide US housing finance over the last 10 to 15 years, shaped by policy errors and a political goal of widening homeownership, misallocated trillions of dollars of capital into housing, and researchers have largely missed this misallocation.
“through policy errors and some stupidity but aided by a lot of policy errors are highly wanted and developed financial system managed to channel trillions of dollars into housing”
The Kuznets curve, where inequality rises and then falls with development, is not inevitable: there is broad agreement that countries need not go through rising periods of inequality, and researchers now decompose the sources of inequality change into occupation transitions, education, and access to the financial system.
“I think there is pretty much agreement that we don't have to experience Kuznets curves and rising periods of inequality or not inevitable”
National income accounts conceive of households only as suppliers of labor and savings and as demanders of consumer goods, which is definitely inappropriate for developing economies where much production is done by households themselves.
“the conception of households and the national income accounts would be as suppliers of labour and savings suppliers of factors of production and in turn the demanders of consumer goods and that's definitely inappropriate for developing economies”
Panel data that follow the same households over time reveal transitions and mobility that repeated cross-sectional snapshots and medians obscure, and recent US inequality claims that top incomes grew sharply are largely a distortion caused by changing sample composition and demographics.
“compared to panel data so I totally agree with you that panel data is extremely important”
Microfinance in India ran into trouble because its success attracted attention: successful microfinance institutions that began as not-for-profits became privatized with IPOs and enriched owners, which fed the preconception that lending to the poor is exploitative.
“some of the microfinance institutions as you were suggesting or we're making a lot of money they start as not-for-profits in Mexico”
In Hyderabad, evaluations of microfinance lending by Banerjee and Duflo found it is not a resounding success for all borrowers: many do not borrow, some use loans for consumption, and only a subset of entrepreneurs set up or expand businesses.
“there have been evaluations of some microfinance lending my colleagues Banerjee and DUPLO have done this in Hyderabad”
The real winners of the Thai village microcredit program were credit-constrained households, and an equivalent lump-sum grant would have had the same budget cost and consequences, meaning the lending program's gains come at taxpayers' expense.
“the real winners in this were households who were constrained in the amount of investment that they could do and other households would have preferred to take a lump sum grant”
Low-return households running farms and businesses should exit and redeploy their labor under a well-functioning financial system, but in these villages exit happens slowly if at all, which is a symptom of poor financial functioning.
“if things were working better you would expect these households to exit what they're doing and do something else with their labor and their resources but by and large that happens slowly if at all”
Thai finance companies were relatively unregulated compared with commercial banks, and they fed the housing bubble that burst in Thailand and then spread to the rest of Asia, helping produce the Asian financial crisis.
“the finance companies in Thailand were relatively unregulated unlike the commercial banks and they were the culprits”
The early microfinance worry that lenders would take advantage of borrowers was based on the premise that borrowers would be over-indebted, but Townsend's data show most households have fairly low debt and insufficient funds to take advantage of profitable opportunities.
“that's not the picture by the way of course that I'm painting which is largely households with fairly low levels of debt and insufficient funds”
Latin America also had similarly high growth rates before the early 1980s, but these did not persist; countries like Mexico grew, then suffered financial crises, so growth went 'one step forward, two steps back'.
“Mexico will grow and then have a financial crisis and take you know it's almost one step forward two steps back”
Thai finance companies, which were relatively unregulated unlike commercial banks, fed the housing bubble that broke and triggered the Asian financial crisis.
“the finance companies in Thailand were relatively unregulated unlike the commercial banks and they were the culprits”
Financial systems can go off track, so institutions should be rated by the services they provide using the same outcome benchmarks, and the largest cost of the housing misallocation is not taxpayer cost (possibly hundreds of billions or trillions) but the unseen misallocation of scarce capital into oversized houses.
“the unseen cost that's transfer mostly the unseen cost is the misc allocation of scarce capital and how what we don't have because we built so many and so much such bigger houses”
The US also has areas with no commercial bank and with payday loans and high-interest informal lending, so asking how 'developed' a financial system is may be the wrong question; the better approach is to evaluate financial systems by their outcomes, such as how household consumption responds to shocks.
“I look at outcomes you know I looked directly first of all at the at the consequences of the financial system not taking a stand on whether it's good or bad”
Townsend's extension of Friedman's permanent income idea is insurance across people at a point in time: in a village, households transfer money through remittances and gifts from lucky to unlucky, so consumption moves together across households but not in lockstep.
“I'm actually adding to that the idea of insurance across people at a point in time if you had Friedman's world there would be a lot of smoothing going on”
The better way to assess a financial system is by its outcomes rather than its institutional form: observe how households' consumption responds to adverse income shocks such as unemployment or illness, without taking a stand on whether finance is good or bad in general.
“I look at outcomes you know I looked directly first of all at the at the consequences of the financial system not taking a stand on whether it's good or bad”
Townsend distinguishes Friedman's smoothing over time from insurance across people at a point in time: villages can pool income so that the unlucky poor receive transfers from the lucky, which produces consumption that is smooth but not coincident across households.
“I'm actually adding to that the idea of insurance across people at a point in time”
Total factor productivity growth in Thailand reflects reallocation of existing land, labor, and capital rather than more factors of production: finance moves money from inefficient households that should not run their own farm or business toward households that should, and savers share in the gains.
“it's not creating more factors of production it's reallocating them it's taking money through lending and interest bearing deposits from those inefficient households”
The hidden cost of financial crises and bailouts is not only the taxpayer cost, which may be hundreds of billions or trillions of dollars, but the misallocation of scarce capital into oversized houses, which is harder to see.
“the unseen cost that's transfer mostly the unseen cost is the misc allocation of scarce capital”
The efficiency standard for village risk sharing is measured relative to each household's long-run average, so persistent differences in income levels across households are compatible with efficient insurance; efficiency does not imply egalitarianism.
“I'm not arguing for equality I'm not taking a stand that we that efficiency means an egalitarian system you can have long run persistent differences across households”
Simulated economic models based on Thai household data explain about 75 percent of the movement in macro total factor productivity over the last 30 years through the deepening of the financial system.
“we can actually explain about 75% of the movement and macro total factor productivity with this deepening of the financial system”
In Thailand, inequality first rose because wages were pinned at subsistence-agriculture levels while capital earned large industrial profits; around 1990 to 1994, rising capital per worker raised labor productivity and wages, and inequality then fell, a rising tide that lifted most boats.
“eventually you see this this increasing level of capital increasing the productivity of Labor which around 1990 to 1994 made its way into wages and then the picture just basically reversed itself”
Thai development is not just rural-to-Bangkok migration; regional villages are growing into towns through construction, and a large share of national income comes from hundreds of thousands of small household businesses without hired labor, most commonly middlemen who lower transaction costs.
“a lot of it is truck barter in exchange you know you'll buy a pickup truck which is of course not a small expense”
Households earning very high returns on assets reinvest their earnings, raise their savings rates, and even cut consumption to obtain more capital, so the high returns cannot be a measurement artifact; the gains from lending to them appear to be large and unexploited.
“the strongest evidence that we're getting the numbers right is that these households with very high rates of return do take the money that they earned and put it back in the business”
Many policy consequences that look unintended are in fact intended: Roberts argues that consumer protection rules often make consumers worse off, and that the financial sector is extremely politically savvy and powerful, so its influence shapes policy outcomes.
“I tend to think that there are fewer unintended consequences than people think that a lot of consequences are intended”
Thai development is not only rural-to-Bangkok migration: villages are growing into towns through local construction, and small household businesses without hired labor contribute a substantial share of national income and are expanding in regional agglomerations.
“there's a lot of development going on in the regions you're seeing collections of villages grow with construction”
The financial system does not work nearly as well for long-run investment as for transient shocks: return on assets varies enormously across households, with some earning 40 to 50 percent annually while others run farms and businesses at low returns that persist without exit.
“one of the big findings and one might say puzzles is that the system doesn't work nearly so well for these longer run investment possibilities”
Thai inequality first rose because wages were pinned low by subsistence agriculture with little agricultural technological progress, which generated large business profits reinvested in capital; around 1990 to 1994 rising labor productivity pushed up wages, and inequality then fell.
“at first inequality was increasing the wage rates were kind of staying low basically pinned down by the return in subsistence agriculture”
Latin America had similarly high growth rates to Asia before the early 1980s, but the growth did not persist; Mexico grows, then has a financial crisis, so growth is 'one step forward, two steps back'.
“latin america if you go back in its history to the before say the early 80s they also had similarly high growth rates but they didn't persist”
Because some households earn 40 to 50 percent returns and some for-profit lenders charge rates above that, large unexploited gains from trade exist; a lender could earn 20 to 30 percent by funding a borrower earning 50 percent and both would gain.
“there appears to be very large unexploited gains from trade there”
Thailand's previous government gave every village a government-seeded savings and loan fund; because villages differ in household counts, the per-household amount varied, and Townsend exploits this variation to estimate that the fund raised investment and consumption.
“ironically the same amount of money was put in every village but villages vary quite a lot by the number of households that live there so in some sense the per person or per household treatment varied”
High-return households do not sit idle: they save at higher rates and reinvest earnings, sometimes cutting consumption to add resources to their businesses, and they are far from steady state, so the gains are not quickly exhausted by diminishing returns.
“their savings rates are higher they will even drop consumption in order to sure try to to get yet more resources to put in the business”
Supporting informal financial markets as a policy may be bad policy, and drawing rigid plans that assign each type of financial institution a set role is risky unless the institutions can coordinate with one another.
“over labeling financial institutions in the sense of sort of drawing a plan and saying this kind of institution should be doing X and another kind of institution can be doing Y”
Village credit funds in Thailand raised agricultural investment and trade-vehicle repair investment, and consumption also rose, plausibly because households no longer needed large buffer stocks of savings against bad years.
“an investment actually does go up there's investment in agricultural activities in repair of vehicles that are used for trade”
Microfinance in India fell into crisis partly because successful microfinance institutions converted from not-for-profits to publicly listed for-profit firms, whose owners made large sums; press coverage and a Mexican congressional investigation fed a narrative that profiting from the poor is evil.
“eventually the more successful ones privatized a they have IPOs and the owners make tons of money and then in Mexico there's an investigation from the Congress”
The main lessons from Thailand are to identify transitions into the financial system, occupational change, and children's education, and to improve markets and institutions, not to give money away; entrepreneurial households were already employed in small businesses even after the crisis.
“I think you learn to locate transitions into the financial system changing occupation getting the kids educated but not necessarily you know do it by giving money away”
Although there is no consensus yet, the causes of growth can be resolved by combining micro and macro evidence: start with household and small-firm data (including own surveys), identify binding constraints and which households overcome them, then extrapolate to the macro level.
“my own view is one half to dig pretty deep down and get existing data or even do one's own surveys and really look at households and and small medium-sized firms”
After a banner story about a microfinance firm's profits, reports of over-indebted households in Andhra Pradesh committing suicide led the state government to impose a moratorium on loan repayment, which destroyed the culture of repayment and made outside investors, including venture capital, wary of funding Indian microfinance.
“the provincial government the state I guess in mother possession issued a moratorium on repayment of loans so that which basically destroys the culture of repayment”
Banks in many developing countries have little incentive to innovate, so when researchers present case studies of underserved areas, banks regard lending there as too risky and do not enter; open entry by other institutions is therefore the necessary complement to government facts-gathering.
“if banks don't have an incentive to innovate then if you bring up a case study from an area they do they don't service there inevitably going to regard it as very risky”
The lesson from Thailand is to locate transitions into the financial system and occupational change, and to improve markets and institutions rather than give money away, because Thai households are entrepreneurial and remain employed even in crises.
“I think you learn to locate transitions into the financial system changing occupation getting the kids educated but not necessarily you know do it by giving money away”
Reports that over-indebted households in Andhra Pradesh committed suicide led the state government to impose a moratorium on loan repayment, which destroyed the culture of repayment and deterred outside investors, so venture capital drying up is a lasting cost even beyond the immediate crisis.
“the provincial government the state I guess in mother possession issued a moratorium on repayment of loans so that which basically destroys the culture of repayment”
The real winners of the Thai village fund experiment were households constrained in how much they could invest, and a lump-sum grant would have had the same consequences; the fund's advantage is only that it avoids politics, not that a financial institution is essential.
“there are alternative ways to to give away money other than establishing a financial institution”
Financial deepening on the extensive margin means more households transact with formal finance: in Thai survey data asking about transactions with the Bank for Agriculture or a commercial bank in the previous month, only about 7 to 8 percent answered yes in 1976, and that share rose roughly threefold by 1996, a period of high growth.
“basically only seven or eight percent of the household dance answered the question affirmatively but over the years between 76 and say 96 a period of very high growth for Thailand that number rises by threefold”
The hidden cost of the US housing misallocation is mostly the opportunity cost of scarce capital that was not invested elsewhere, which may be far larger than the taxpayer cost, estimated at hundreds of billions or possibly trillions of dollars.
“the unseen cost that's transfer mostly the unseen cost is the misc allocation of scarce capital”
Regulators should not tell every financial institution what it ought to do or which services to offer; instead they must keep the system open so that someone else can step in, since gains from specialization in banking come from allowing entry.
“we shouldn't be in the business of telling every institution they audit what they ought to do”
Policymakers should not tell every institution what services to offer; if there are gains from specialization in banking, the system must remain open so that another institution can step in and innovate.
“we just have to make sure the rest of the system is open so that somebody else can can step in and and innovate”
Poor rural households are diversified across wage work, palm-tree toddy tapping, livestock, and small businesses, and they pool these diverse incomes as if in an implicit mutual fund.
“they're pooling all of that together as if they were in an implicit mutual fund and in sort of eating the the dividends”
Economists should be humble about what they can design: the curious task of economics is to show how little people know about what they imagine they can design (Hayek).
“as FA Hyack said the curious task of economics is to illustrate to men how little they really know about what they imagine they can design”
Financial system assessment by outcomes implies that a better system can produce large, not modest, gains in well-being, because misallocation of capital is severe.
“isn't that suggestive or is it not that that that a better financial system would lead to big increases not modest increases but big increases in well-being yes that's quite true”
Although there is no consensus on the causes of growth and stagnation, Townsend is confident the question can be resolved by combining micro and macro evidence, starting from household and firm-level data (surveys, panels) to identify which households and firms are constrained and then extrapolating upward.
“my own view is one half to dig pretty deep down and get existing data or even do one's own surveys and really look at households and and small medium-sized firms”
The real beneficiaries of government-funded village credit were households constrained in how much they could invest; a lump-sum grant would have had similar budgetary consequences and might have been preferred, so microcredit is not the only way to deliver these gains.
“there are alternative ways to to give away money other than establishing a financial institution”
Friedman's permanent income hypothesis argues that people smooth consumption over time by borrowing and saving, and temporary tax rebates are largely saved rather than spent, which is consistent with the data.
“Milton Friedman in his theory the consumption function made that argument right and it is largely borne out I think by data”
Thailand's transformation is not only rural-to-Bangkok migration: villages are growing into towns, farmland is paved over, and urbanization is occurring across the regions, as shown by soil moisture and land-use data.
“I would dispel the image of lots of poor farmers migrating to the Bangkok that has happened to a large degree it's a big part of the story but it's not the only story”
The US is a highly developed financial system on the surface, yet parts of it have no commercial bank nearby and rely on payday loans and informal high-interest lending, so its financial development is also uneven.
“there are parts of the United States where there's no commercial bank anywhere in sight as is the example you just gave and where payday loans are common very high rates of interest”
Rural households in the village study diversify income sources (wage work, tapping palm trees for toddy, livestock, small shops), and pool these different incomes as if in an implicit mutual fund.
“they're pooling all of that together as if they were in an implicit mutual fund and in sort of eating the the dividends”
Many developing countries field large cross-sectional socio-economic surveys that can go back decades, but they are not panels (they do not follow the same households over time), and survey quality varies widely by country.
“some countries have been fielding socio-economic surveys fairly large cross-sectional surveys they're not panels they don't necessarily survey the same households”
The most common small business in the Thai household data is middleman work, helping others make transactions more cheaply than they could alone, and many households buy pickup trucks costing thirty to forty thousand dollars to haul goods.
“perhaps the most common business in the in the data being and being a middleman”
The US, despite appearing highly developed financially, has parts with no commercial bank nearby where payday loans and informal high-interest lending are common, so a highly developed surface does not imply universal access.
“there are parts of the United States where there's no commercial bank anywhere in sight as is the example you just gave and where payday loans are common”
Standard sources of growth in industrialized economies are productivity gains rather than increases in land, labor, or capital inputs, and this is also true of emerging economies like Thailand.
“then industrialized economies a lot of the growth is associated with productivity growth and not with changes of the factors but that's likewise true for emerging market country like Thailand”
Panel data, which tracks the same households over time, can measure transitions in and out of work and production that cross-sectional snapshots miss, and Townsend's household panel data show income transitions that medians hide.
“if we have panel data and we can track what's going on in the household over time then you can actually measure these transitions”
Wage earners have become the largest category of Thai workers, displacing farm and non-farm self-employment, and corporate profits are only about 22 percent of household income in the national accounts versus 40 to 50 percent for other income categories.
“the number of people who are say earning wages is going up it's become the single that largest category of people”
The share of Thai workers earning wages has grown to become the largest employment category, displacing farm and non-farm self-employment, which was largest in 1960; corporate profits are only about 22 percent of household income versus 40 to 50 percent for other categories.
“the number of people who are say earning wages is going up it's become the single that largest category of people”
Thailand's rural small businesses are often truck-based barter and middleman operations: households buy pickup trucks costing around 30,000 to 40,000 US dollars, financed by previous savings, and the most common business is middleman services that make transactions cheaper for others.
“perhaps the most common business in the in the data being and being a middleman yeah helping others make transactions more cheaply”
Milton Friedman's wit about a windfall: when asked how an unexpected windfall would affect his consumption, he replied 'who said it was unexpected', meaning he would fold expected future gains into spending plans.
“the wit replied well who said it was unexpected with meaning that his life he was being you know facetious in part”
Data quality is a central obstacle in studying development: the transition from agriculture to manufacturing or information work happens in economies where income data are poor, and even good surveys are often not panels.
“we don't have very good data on income and in agricultural economies”
The causes of growth stagnation are unresolved because researchers hold strong prior beliefs (financial system, politics, education) and the evidence has not yet fit together, so confirmation bias is a serious obstacle.
“many people have prior beliefs about it it's the financial system it's politics it's it's education unfortunately these pieces just don't come together”
Thailand's microfinance-like village funds are capitalized by government grants, while the banking sector in Thailand is dominated by commercial banks and government agricultural banks rather than microfinance institutions.
“Thailand actually doesn't have very many of those apart from these village funds”
Statistical analysis tries to hold other factors constant to isolate the effect of one variable, but no consensus on growth causes has emerged from it.
“people try to use it I try to answer the questions they try to hold things constant and look at the effects of one variable or another but I wouldn't say any consensus has emerged”
Decomposition of inequality into sources such as occupation transitions, education, and financial access is not proof of causation, but it shows where to start looking and makes it possible to quantify which components are large or small.
“it's not a proof but it's it kind of tells you where to start looking to understand the sources of changes and inequality”
Policymakers and others presumed that poor Indian households near poverty could not organize mutual insurance among themselves, so Townsend's finding that they do was controversial.
“it really raised a controversy because there were presumptions by policy makers and others that these Indian households near or below poverty would not be able to help themselves and organize among themselves”
Cultural entrepreneurship varies across societies, but Townsend's focus is how cultures create local institutions such as kinship networks, which are robust and have helped mitigate poverty risks.
“I look at how cultures create local institutions that can help mitigate these problems these kinship networks for example are quite lively and and robust”
Microfinance-based rejection of the 'evil' narrative is important because the Indian backlash shows how misconceptions mixed with politics create barriers to financial innovation.
“these sort of conceptions and misconceptions of mixed in with you know politics have often been a barrier to innovation”
Panel data means following the same households over time, asking the core questions repeatedly, and comparing the trajectories of those households rather than comparing medians across different samples at different dates.
“panel data by panel day we mean you go to the same household year after year after year”
Under the rule of 72, growth of 4 percent doubles income roughly every 18 years, 6 percent every 12 years, and 10 percent about every 7 years, so Thai growth implies very rapid doubling of living standards.
“you double every seven years and it's actually the rule of seventy more than 72 but more things going to 72”
Introduction: Russ Roberts hosts EconTalk and welcomes Robert Townsend, Elizabeth and James Killian professor of economics at MIT, author of Financial Systems and Developing Economies.
“my guest is Robert Townsend the Elizabeth and James Killian professor of economics at MIT”
Using the rule of 72, at 4 percent annual growth income doubles about every 18 years, at 6 percent about every 12 years, and at 10 percent about every seven years, so compounding growth rates imply very fast doubling of living standards.
“you're going to double income at that rate double things at four percent and we can get six percent you double every 12”
Early critics of microfinance worried that lenders would not be repaid, or that poor households would become over-indebted, but Townsend's evidence shows most borrowers have fairly low debt and insufficient funds to pursue profitable opportunities.
“that's not the picture by the way of course that I'm painting which is largely households with fairly low levels of debt and insufficient funds for some of them to take advantage of profitable opportunities”
Thai households are highly entrepreneurial and rarely unemployed even during the financial crisis, setting up low-return businesses that still earn more than returning to unemployment in the village.
“they're very entrepreneurial they're very successful some of them so they don't sit around unemployed rarely even in the financial crisis household set up businesses”
Good income data are lacking in agricultural economies; some countries field large repeated cross-sectional socio-economic surveys that go back decades (as in Mexico), but South Korea, a successful economy, lacked detailed household surveys until recently, so its transition from agriculture cannot be documented household by household.
“the irony Korea South Korea that we know is quite successful for decades doesn't have detailed household surveys until relatively recently”
Thailand's growth typically exceeds 4 percent per year, so a year below 4 percent counts as bad, and the country has continued high growth despite the Asian financial crisis, the tsunami, and political turmoil.
“for them it's a bad year when the growth rate is below four percent and that's still the case”
In Mexico and many developing countries, banking is thin: the top five banks account for nearly all financial activity, and large sectors of the population lacked savings and financial services as of about five years earlier.
“if you count the the top five say banks in Mexico they say pretty much account for all of it”
The World Bank's Living Standards Measurement Surveys include extensive questions on household production, showing that measurement of non-market production is feasible if surveys are designed for it.
“some institutions such as the World Bank began these living standard measurement surveys”
The US Panel Study of Income Dynamics is valuable but thin on consumption data in many years, so researchers splice it with the Consumer Expenditure Survey, which is mainly cross-sectional, to exploit the panel structure for consumption measures.
“they asked relatively few questions about consumption for example so you have to get creative and combine say the PSID with the Consumer Expenditure survey”
Thailand has grown rapidly for three decades, interrupted by the 1997 Asian financial crisis, and although its long-run average growth fell below the earlier five to six percent, its growth typically remains above four percent, which is the envy of many countries.
“the growth rate never got up to the you know five six percent on average that it was prior so there's kind of like shift in terms of the long run averages”
In many developing countries the banking system is thin: the top five banks in Mexico account for essentially all of it, yet conventional financial access surveys revealed large parts of the population without savings accounts or other financial services.
“if you count the the top five say banks in Mexico they say pretty much account for all of it”
Policymakers must understand what is happening on the ground before drawing conclusions, and cultural kinship networks often provide robust local institutions that help mitigate financial shocks, so culture matters for which institutions can work.
“these kinship networks for example are quite lively and and robust and have been helpful on some if not all of the dimensions”
Informal arrangements such as family assistance, private agreements, and gifts count as financial institutions when assessed by their outcomes, because they make life more pleasant by smoothing well-being against bad luck.
“you're talking now about the potential to smooth either through organized financial transactions or these informal methods of assistance from families and friends or even some agreements”
Even after the financial crisis, Thai households rarely sit unemployed and set up businesses, some with low returns, that still earn more than returning to the village would, a pattern visible in the pre- versus post-crisis business data.
“even in the financial crisis household set up businesses some of them with low rates of return but still they're employed to make that in earning more money than they would be”
A rural 'lending for profit' argument: households with high returns could pay high interest and still profit, so lenders could earn large returns lending to credit-constrained households.
“you've got somebody who's got an asset that's returning 50 percent a year they would like to have twice as much of it”
Adam Smith's pin factory idea of specialization is part of the story of Thai development, but growth came from people integrating into a larger division of labor through cheaper trade, not from learning to make things better.
“it's not that you're learning to create pens better though it's just as a transportation cost fall”
Financial deepening means more bank branches, more borrowers and savers, and, on the extensive margin, households moving from no formal-sector financial relationship into it.
“deepening the financial system means that there are more branches of banks there are more borrowers as well as savers”
Confirmation bias is a serious obstacle in development research because most researchers arrive with prior beliefs about causes of growth, though Townsend believes the causes can ultimately be determined by combining micro and macro evidence.
“confirmation bias is a big problem”
Consumer protection regulation often makes consumers worse off when poorly designed, and people underestimate how many policy consequences are intended; the financial sector is extremely savvy and powerful, and it shapes policy through access rather than just money.
“I tend to think that there are fewer unintended consequences than people think that a lot of consequences are intended”
Consumption is largely insulated from idiosyncratic income shocks in many households, because they use informal insurance and smoothing mechanisms, though this is an empirical question that varies by country.
“consumption is largely immune to these large idiosyncratic income fluctuations”
Policy and political misconceptions about profitable lending to the poor, not the economics of microfinance itself, have been a barrier to financial innovation in developing countries.
“these sort of conceptions and misconceptions of mixed in with you know politics have often been a barrier to innovation”
Informal financial markets in Thailand are largely helpful despite not working everywhere, whereas in the US policy is stifling financial innovation, a parallel that is provocative but reflects the same problem of distorted policy.
“these sort of informal financial markets in Thailand are by and large quite helpful it's not uniform it doesn't work everywhere whereas in the US we seem to be going in the direction of stifling financial internet innovation”
The Federal Reserve contacted Townsend about fielding a panel survey of US households motivated by the financial crisis, reasoning that tracking households and their mortgages might have shown the crisis coming.
“I've been actually contacted by the Federal Reserve you know like what would it take to field a panel in the US instigated by motivated by the the recent you know financial crisis”
The US financial system is moving in the direction of stifling financial innovation, unlike Thailand's informal markets, which are by and large helpful, and policies that support the informal market may be bad policy in some contexts.
“whereas in the US we seem to be going in the direction of stifling financial internet innovation”
Hindsight aside, if the US had tracked households and their mortgages in panel form before the financial crisis, policymakers might have seen the problems earlier.
“if we had been tracking some of those households and in houses and their mortgages and so on we might have done a little better”
The 1997 financial crisis temporarily raised inequality in Thailand for a year or two, but it fell back and is now lower than before the crisis.
“there was a bit of a back step in the financial crisis inequality went up again pretty sharply for a year or two but then relatively quickly it fell back down”
Better policy requires enough data and enough analytical frameworks to assess the flow of funds and rates of return institution by institution, and to call a spade a spade.
“my heart felt you know agenda is to have enough data and enough frameworks to be able to really assess the flow of funds and rates of return”
The needed agenda is enough data and frameworks to assess the flow of funds and rates of return, and to evaluate financial institutions one by one, since policy depends on measured outcomes rather than assumptions.
“my heart felt you know agenda is to have enough data and enough frameworks to be able to really assess the flow of funds and rates of return”
Reading the Acemoglu point, the host says a policy should be judged in terms of its outcome for consumers, and he suspects the political influence of the financial industry is the main driver of policy.
“in the United States the financial sector is pretty politically savvy not pretty it's extremely savvy and extremely powerful”
The US financial sector is extremely politically savvy and powerful, giving money and access, which affects financial policy in ways that go beyond campaign contributions.
“in the United States the financial sector is pretty politically savvy not pretty it's extremely savvy and extremely powerful”
Roberts argues the Federal Reserve should be less involved in systemic financial risk, and that its interest in household panel research is mission creep.
“personally I would like to see the Federal Reserve get less involved in the systemic financial risk but I know that's not the trend”
Explanations for why some poor regions grow and others stagnate (financial system, politics, education) are many, but researchers have not yet fit the pieces together, and confirmation bias from prior beliefs is a major obstacle to finding the answer.
“many people have prior beliefs about it it's the financial system it's politics it's it's education unfortunately these pieces just don't come together”
Townsend has found the work of uncovering the data and understanding what is happening on the ground enjoyable, and believes many countries would be better off with both the data and an analytical framework to base policy on conclusions.
“I really think many countries would be much better off if if they not only had these kinds of data but they had the framework”
The Federal Reserve should get less involved in systemic financial risk, and its interest in funding household panel research is further evidence of mission creep.
“personally I would like to see the Federal Reserve get less involved in the systemic financial risk”
Financial innovation was possible in Thailand: the 'village funds' worked because household borrowing needs were met locally, and microcredit can be assessed without political noise, showing a positive effect.
“there is an instance where you know something like microcredit could be assessed without all the the politics that comes with and it looks on that as if it was helpful”
The most remarkable recent development is the growing presence of India and especially China with very high growth rates, which continues the earlier high-growth episodes in Japan, Korea, and Singapore, even if the official statistics are somewhat unreliable.
“the most remarkable thing now is the growing presence of India and and especially China with with very very high rates of growth even if you think they're exaggerated a bit”
Reserve Bank of India is trying to set up an insurance pool to keep microfinance going, but much damage has already been done to the sector.
“the Royal Bank of India now realizes the problem and is trying to set up some kind of insurance pool”