 I want to make a sharp distinction between the provision of social insurance that in principle you would like to reach the whole population, this is very, very important. Social insurance really has to do by risks, so it has to do about everybody being ill or longevity risk or losing your job, and redistribution programs or social assistance that are by almost by definition focused only on a subset of people. The basic problem that I want to deal with in this chat is basically captured by the fact that the interface between social insurance and redistribution is an interface that in some countries, particularly in the countries of Latin America that I know better, is an interface that works very badly. And for that, think of the rose in blue as the division of the population between poor and non-poor by whatever metric, and you clearly want to redistribute from the upper row to the lower row as part of any poverty program. And then think of the columns, which is separate, a distinction between the labor force that can be labeled formal or informal that is covered by other contributory or non-contributory programs, and that in principle you want everybody to be covered by the same social insurance because they're facing the same risks, but that because of the architecture of the system is divided into two separate segments. Ideally the red column should only be one, the whole population covered against risks, and then the redistribution, the blue, should be two from rich to poor. But in Latin America, and maybe this is useful for other countries in the world, the segmentation of the labor market into formal and informal implies that the provision of insurance is made very difficult. The key concepts here is that by construction, by law, the institutions and the laws of most countries in Latin America make some kind of distinction between salaried workers, sometimes are called dependent workers, sometimes paid workers. These are workers that usually have contributory quote-unquote social insurance, which is paid from a wage, and depending on the country that you look at, there's always health insurance, retirement insurance, disability insurance, unemployment insurance, and in some countries you have child allowances like Argentina, in some other countries you add housing like in Mexico, in some other countries you have labor training programs like Colombia. It's all bundled, and we call this contributory social insurance because it's paid from a contribution based on the wage. But over the last 20 years, what's an emerging Latin America, and this was shown partly in the figure shown by Armando, is the emergence of a parallel set of programs that are also providing insurance that we call non-contributory mislabel, simply because they're not financed from a wage-based contribution, but they're financed from general revenues. And these are targeted to all workers not covered by contributory social insurance. We tend to associate that with informal workers being less income than formal workers, but in fact the data shows that that's not really true, because the wage distribution of informal workers overlaps quite substantially with the wage distribution of formal workers, even though the mean of the distribution of formal workers is to the right to the mean of distribution of formal workers, but there's quite a bit of overlapping the distributions. So you have workers of same income levels receiving insurance through different mechanisms. The three big problems with contributory programs, first by design, by construction, they're never going to be universal, because the law does not obligate a whole bunch of workers to actually be enrolled in social insurance program. The risk pooling is limited, and so many workers are left out legally so, even though of course what you also observe in the data is that many firms violate the law and there are many salaried workers that are not covered by contributory programs because the law is being violated. So the first problem that you see is that coverage is incomplete, and I can show you the data there. For Latin America, less than half of the population is covered by contributory programs, but even in some middle income countries like Colombia, like Peru, like Mexico, more than half of the population is not covered by contributory programs. A second big problem is that these programs implicitly generate attacks on salaried employment. Firms are going to pay W plus alpha W, where alpha is the contribution rate, usually a number between 30 and 45% in Latin America. But workers might not think that alpha W, that 45% of the wage, 35% of the wage is really worth to them, 35% of the wage in terms of benefits because of monopoly provision, because they don't trust the pension system, because of quality issues, or because simply the bundle of goods that are being delivered for them is not really what they want. So workers get W plus something less than alpha W, firms pay W plus alpha W, and there's some kind of an implicit tax on formal employment associated with the design of these programs. And then the third problem, perhaps the most difficult problem of all and the most important, is that we tend to think about formal workers as informal workers as separate subsets of the population. But in fact, the data shows that there's a lot of transits of the same individuals between formal and informal status, so that it's much better to say when workers are formally employed rather than to say formal workers, because the same worker sometimes is formally employed and sometimes is informally employed. The same person with the same human capital with the same abilities sometimes is covered by contributory social insurance and sometimes covered by non-contributory social insurance. And here's some data of transits of the same individual, this is panel data, between formal and informal status within one year for various countries of Latin America. And what this does is that it really makes social insurance to be very inefficient. Because if I am an individual that finished high school and I'm working and sometimes I'm covered by some risks and sometimes the same individual is not covered from some risk, the kind of insurance that you're providing me is not really very good. So suppose I sell to you house insurance and I tell you that your house is covered against fire in the winter and in the spring, more done in the fall, or I sell to you car insurance and I'll tell you that your car is insured Monday, Wednesday and Friday, but your car is not insured Tuesday, Thursdays and Saturdays. So if you're a worker in Latin America, sometimes you're in a contributory program and you're insured against disability and then you move, you change jobs and then you're not covered against disability. And what this will do is that the efficacy of insurance is much diminished because most workers will not contribute sufficiently to actually get a pension so that most pension systems in Latin America are actually not going to deliver a pension for most people saving for a pension. And the medical treatments that they're getting are going to be very incomplete because you're shifting from one type of medical coverage to another type of medical coverage. In addition from not being a very efficient system from the social point of view, this system of contributory social insurance is going to induce changes in firms and changes in workers. What firms are going to do is they're going to change the nature of the contract that they offer workers to avoid the tax. And the change in the contract might be perfectly legal because there's some firm worker contracts that don't associate with a salary in which workers are paid commissions or there's rent sharing or there's profit sharing typically in a family firm that are not covered by this so firms will change the structure of the contracts. Of course, firms will also cheat and they will evade. So there's a mixture of illusion and evasion. And to evade, firms must remain small. So what you see is that the size distribution of firms will be affected by the social insurance system. Firms will tend to be small, underexploited economies of scale, economies of scope. And there's a lot of firm churning because firms are very small and very unproductive and workers are actually changing a lot from job to job so it's also affecting workers. So this is an interesting slide that gives you a sense of the tax on formality associated with contributory programs. Colombia in 2012 made a reform in which we deduced the contributions to contributory social insurance from 12.4 to 4%. This is only the health component of contributions. In Colombia, the contributions are much larger. But what is interesting is that formality increased quite substantially after that was changed. If in fact there was no implicit tax, there should have been no change in formal employment associated with the change but there is a very important change. So by the early 1990s, the failure of contributory social insurance in Latin America was evident as shown by the fact that more than half of the labor force was actually being left out from social insurance. And if you think about people transiting between formal and informal jobs, in fact individuals were less covered because part of the spells of their life, they were out of social insurance. So in the early 1990s, little by little in Latin America, what began to happen is that lo and behold, different programs to provide social insurance to informal workers began to emerge. Sometimes a health program here, sometimes a pension program there, sometimes a disability program, sometimes a daycare program. But a parallel set of benefits began to emerge. These programs with finance from general government revenues, that's why the name of non-contributory, they very much vary from country to country. In some countries, you know, they're more generous than in others. But now the country that I know best, you have almost the same health benefits. A pension, not contributory, that is 0.6 of the pension of contributory, you have access to disability insurance and you have access to daycare centers pretty much in the same way that you have your contributory programs. So evidently, these programs from a social point of view are very welcome because they're going to be providing protection against workers who otherwise would not have it. But if you think about it from a point of view of firms, what they're going to do now is they're going to subsidize informal employment because now the value of a contract between a firm and a worker is worth more than the wage paid by the firm to the worker. Because it's an outside benefit that is being paid from general revenues. It's exactly the opposite of what is happening with a contract that is covered by contributory programs. So this will end up subsidizing informality which adds to the tax on formality that is being there. This is an interesting slide from Mexico. On the left, you see the coverage of a non-contributory health program that started in the early 2000s in Mexico. The red line is the coverage of the contributory health program. And on the right, this is work for some colleagues of mine at the IDB. This is a difference in difference estimator of what happened to firms hiring decisions in Mexico when the non-contributory program was put into place. And by looking at firms that were earlier in the program versus firms that were later in the program, what you can see is that firms' decisions were biased against hiring workers legally with contributory programs and actually cheating by firms was much encouraged by the introduction of this non-contributory program. So I've spoken a lot about the problems of contributory and non-contributory programs and I've said very little about social assistance. What I'd like to do now is instead of talking about the columns of my first slide in which I was discussing the problems with the contributory non-contributory programs, I now want to see the interface between the columns of my first slide and the rows of my first slide in which how do these social insurance programs interface with poverty programs? Ideally, all these poverty programs should be targeting workers only on the basis of their income levels. What has happened in Latin America has been a bit of a confusion between social insurance and social assistance and poverty alleviation programs and what has ended up happening is that sometimes these poverty programs are being targeted on the condition of being informal. So there's a confusion of the rows and the columns in terms of the terminology that I was introducing before and what this does is that it affects poor workers' choices in terms of the status that these workers would have in the labor force given that they're poor. And what you might have end up happening and I'll show you a slide that this is actually happening is that workers who would otherwise be formal even though they are poor end up being informal because of the way that the social assistance program is structured. So this is an interesting example of a program in Ecuador. Ecuador has a bono de sarroyo humano which think of it as a conditional cash transfer program in Ecuador. And what happened in Ecuador is that they conditioned access to the poverty alleviation program on the informal status of the recipients. And this is work also by some colleagues at DIDB in which basically what they show is in 2002 to the left of the vertical line are people below the poverty line to the right above the poverty line. And then what you see is the rates, the share of women that should be in social insurance covered by social insurance. And what you can see is that six years later after the conditional cash transfer program was put into place but was being made conditional upon the informal status of the workers, there was a discontinuity and the rates of formal employment of women in Ecuador actually fell by about 15% on points. So this is substantial. And what this is doing is exactly the opposite of what you wanted to do, which is to induce formal employment. And this is because the interface between the incentives of the two sets of programs are not being put together right. So to go back to my columns and to my rows, think that the formal sector has on average higher productivity than the informal sector. This is actually true if you make the productivity distributions of firms in the formal sector, they are to the right of the productivity distributions of firms in the informal sector, even though there's overlapping the productivity distributions, the mean is to the right of the formal sector. So ideally what you would like is to shift workers from the left column in my, no, from your point of view, from the right column to the left column because you would like them to shift from the informal sector to the formal sector to get higher productivity. And ideally what you would like to do is to move workers from being poor to being non-poor. So if you're at point A and you're a worker who is poor and informal, ideally the incentive structure should be shifting you towards point B. And if point A, OI was your own earned income, ideally by shifting from point A to point B because you're more productive with your own labor, perhaps because you acquired more education with the CCT and you have better health, now your own income is going to be higher and now you're gonna be shifting from A to B. But I showed you slides before that in fact the combination of contributory and non-contributor social insurance is actually pushing workers in the other direction. And rather than shifting from point A to point B, what you're actually doing is you're shifting from point A to point C. You're getting the transfer because your transfer is conditional on informality, you're actually getting a higher level of income but not because of more own earned income but because of the transfer while you're remaining in the informal sector. And of course if the distribution of the labor force is much affected by these forces, the average productivity of the economy is gonna be punished by the fact that a large share of the labor force in the economy is gonna be pushed or remaining in the informal sector rather than shifting to the formal sector. So the basic punchline of this sort of chat is to say that the current architecture of social insurance programs and interface between social insurance programs and poverty programs is creating a vicious circle between informality and low productivity. You have a lot of people with informal jobs. The policy response to that is I'm going to create a parallel system of social insurance through non-contributory programs, more health programs, more pension programs, more decker programs. Sometimes I'm going to increase the size of the transfers of my poverty programs. But the fact of what this is actually going to do is it's going to subsidize more the informal sector. This has to be paid somehow so you're gonna tax more of the formal sector or you're gonna incur more debt. What this will do is it's actually lower the productivity of the economy and if you lower the productivity of the economy what that's gonna do is it's gonna induce more informal jobs. So Miguel was very nice to mention a book that I just finished on Mexico and I've been looking at the data from Mexico over the last two decades. Mexico is a country that over the last two decades had zero total factor productivity growth. Zero. And while this is not the only reason tax considerations and other considerations are also very important parts of the explanation clearly what we have in Mexico is a system of social insurance and social protection that is acting despite its best interest in terms of what really matters for increasing productivity and creating better jobs for the workers that should be getting from that. So let me conclude a little bit. I think that in Latin America and maybe this might be useful some other countries in the world this dichotomy between contributory and non-contributory social insurance is really bad. We really need to unify regimes and we really need to transfer to universal systems. This of course has many fiscal implications but I think the transit to universal social insurance is really essential from the social point of view because the system is not efficient. Workers are not being well protected. But in addition the system is very costly from the productivity point of view. So universalism I think is a direction where we have to move. And then when we think about the interface between that and poverty we clearly have to ensure that we don't condition any benefits on the poor upon their situation in the labor market. And in fact poor workers should have the right to the same social insurance as all other workers with the same quality of health the same pensions, the same taker, the same systems like that. Of course this implies a big fiscal change but clearly if we really wanna break the intergenerational transmission on poverty we really have to change from one system to the other and not continue in the vicious circle that I was there before. So I think by now there's a large literature that has been accumulated. I just showed a few slides here from one experiment in Ecuador sort of one case in Ecuador, one case in Colombia, one case in Mexico. But there are quite a bit of papers here. Guillermo Cruz is here. He's written some nice papers on Argentina. Other people have written papers from many other countries. By now there's sufficient empirical evidence to be showing that the combination of contrariatory programs and non-contrariatory programs is really distorting heavily firms and workers choices within the labor market so that we need to pay attention to that. And this is a problem of architecture. This is not a problem of enforcement. This is a problem that the laws and the institutions are not well defined, are actually not well thought out. More enforcement is not going to solve this problem. And as an academic issue I think we now need to move beyond the impact evaluation of individual programs. Over the last 20 years we've learned a huge amount by doing all these very careful control treatment impact evaluations with randomized controls about what is the impact of this program, this program, this program, this program. There's a lot that we have learned over the last two decades, but I think now we can use all that learning and all that evidence to begin to think about systemic pieces. Not about individual programs, but about how all these things fit together and to be an integral view of policy. And I think this is really essential because what I think is happening in Peru, in Colombia, in Mexico, in Ecuador, in many countries in Latin America, is that we're constructing welfare states in economies that have stagnant productivity and that are actually not gonna be viable over the meeting term. Stop here, thank you. Thank you.