 Welcome to us at the beginning of the conference this morning. Is it privatization is very much on the agenda here? And I was eager to come to this conference in part to learn more about what Vietnam has already done and is planning. So let me jump into this. Let's see, so what I'm gonna be focused on here are firm level outcomes of privatization with a special emphasis on firm performance issues. So productivity, profitability, output as consequences of being privatized. I may have a few words to say also about what happens to workers and what happens to the creative destruction process. I'm also going to talk a little bit about the consequences for firm performance of different methods of privatization to the extent that the data permit. And I will also talk briefly about some variation in the effectiveness of productivity at raising firm performance, what we would call the heterogeneity of the effect in terms of a few different dimensions. Now of course there are lots of questions that one could ask about privatization. So while the emphasis here is on estimating the direct performance effects. And so what this means is a change of ownership from a shareholding which is dominated by the state to shareholding including or expanding the role of private owners, private investors. One could also talk about other kinds of privatization about land and housing, about premises for shops. These are certainly very important issues over Bia. In the center of town last night, I met someone who was telling me about problems of land ownership for enterprises in Vietnam and how it limits investment. But they're not going to be the subject of my talk today. I could also talk about new entry. So there are lots of definitions of privatization. One of them encompasses entry of new private businesses. Some people have argued that this is a more important process than privatization. Actually it's not clear that the empirical evidence supports that but in any case I'm not talking about it today. I'm not talking about sequencing, what should be privatized, how quickly or what firms first. Or in general about the relationship with other policies hardening of budget constraints, financial sector reform, et cetera. Nor about the political economy of privatization. The consequences of privatization for political economy and decentralization. Or about the political economy causes of privatization. And in light of the title and the themes of this conference, I'm really not talking about inclusion or sustainability. Although those could be potential topics. I've written quite a bit about the effect on workers of privatization. But one could also analyze issues of inequality and what happens to workers who are laid off and what happens to work conditions. Also there's some research on environmental aspects. So none of those things are part of what I am planning today. So I'm trying to limit my topic but even so limited there's a vast ground to cover. First of all, I'm just going to mention briefly what is probably the prevailing view now about privatization theoretically, the way it works, is by raising the cost to politicians of intervening in enterprise behavior. Under state ownership, the minister can simply call the enterprise manager and order what the director should do. Hire more workers, raise wages or start a new product or whatever. But under privatization, it's more costly for the politician to do that. The politician can still make the phone call but the manager says, well look, we're private, I have owners, I have to answer to. And so what are you going to do for us? And so the idea is that that will then tend to lead to a change in the firm's behavior as a consequence of that. Okay, so there was a lot of early research on privatization that I just wanted to mention. When these topics were hot in the 1990s, there was enormous speculation about the effects of privatization and there were also many empirical studies but they suffered from substantial problems in estimating effects. The sample sizes were generally tiny, a couple hundred observations on firms observed for a very short period of time. And so many questions can be raised about those studies and I include my own studies from that time raising those questions. So it's only more recently when all the furor has died down about East European and post-Soviet privatization that we actually have managed to put together the data sets that allow us to estimate, I would say with some greater degree of reliability. So larger numbers of observations, longer time periods and it's possible now with the larger data sets to compare firms within narrow industries within the same year. And so we get rid of some important aspects of heterogeneity arising for instance if firms from particular industries tend to be privatized more rapidly than from others and there are different performances across those industries. So many studies of privatization are plagued by that problem. It's also possible using longer panel data sets to introduce firm fixed effects and to add firm specific trends also to the model. So we'll be looking at results that use universal data that is for every firm in four countries. There are Hungary, Lithuania, Romania and Ukraine and for Russia the data are more limited in a number of ways that I don't really have time to get into here but it's a caveat for the analysis of Russia. But in general these are comparable samples and variables for these five countries. These five countries were not chosen as a result of some random selection of transition economies, I wish I could tell you that but it's simply on the basis of data availability. Data of this high quality and quantity type that enables us to estimate in a comparable way. And we are using the data on old firms. So old firms we define as firms that are inherited from the planning system and their measure is either existing before 1990 or ever having any state ownership. This is something about the coverage. We have about 600,000 firm here observations and on average in the time series we have 14 observations. So we have a chance to observe these firms from a long time before they're privatized until a long time afterwards which is a basic difference from lots of papers published in say the Journal of Finance that look at privatization all around the world and generally have only very few years of observation on each of them. Here's a quick graph showing the evolution of privatization within this old sector. You can see that we start off zero private. Initially there's a jump there in Hungary where you see some jumps down it means that state enterprises have been split up and this is just simply the fraction of the sample with a private designation. So split ups lead to more state enterprises in the sample but in all the countries there's a big rise in the private share among the old firms. Here private is defined as 50% or more of the shares owned by private owners. We're gonna look a little bit at some alternative definitions of that threshold. So here are some initial results. I said the only table I'm actually gonna show you everything else is maybe more user friendly graphs but the main takeaway here is we can estimate with ordinary least squares which functions only as a baseline maybe comparison with earlier studies that were forced to use ordinary least squares for due to lack of information. We can add firm fixed effects or we can add firm specific trends which turns this into what is essentially a random growth model. But regardless of what we do all of the estimated effects are positive. They're highly statistically significant. Well that's not difficult with 600,000 from your observations but we're also putting in we're taxing this regression with a lot of added variables. And so in the most highly taxed regression we're still getting some substantial positive effects. Let me turn to some heterogeneity issues. This is for the domestic versus foreign ownership and in the rest of the results I'm dropping the OLS just to save space and I'm gonna always be showing you firm fixed effects and firm fixed effects plus trends. The trends of course reduce all of the estimates. They just take a lot of variation out of the data. But in all cases we estimate at least where they're statistically significant we estimate much larger effects from foreign investors compared with domestic. So the bang for labor productivity, for return on sales, a profitability measure and for output are larger for foreign investors. I don't consider these three dependent variables to be equally reliable. My preferred one would be the labor productivity. I also have results with multi-factor productivity which are similar to these but so the foreign effect is much larger which is something for policy makers to think about when they're designing future privatizations. Now I'm going to distinguish three levels of privatization, a 100% privatization. So all the shares are vested in private hands. Then we have majority but not 100% privatizations and finally minority or partial privatizations which have been pretty common in some countries and some people think that they have substantial positive effects. Well, what we find is actually different from that received wisdom. We find by far much stronger effects of privatizing all the shares in the firm. In fact, for all of these, the typically the effect of privatizing all 100% compared with privatizing a majority but not 100% is double, sometimes more than double. So getting the state out entirely is having a much bigger effect than the majority. The minority privatizations are all estimated to be negative, some of them are statistically significantly different from zero. So this mixture, at least for these countries and these data of state and private where private is not in control is actually negative for these firms. As anybody who was reading the newspaper in the 1990s would know there was a great controversy over different methods of privatization, voucher or mass method, MEBO's management employee buyouts, which is typically not really a buyout but a giveaway to the insiders of the company with varying levels of non-managerial and managerial employee participation. Then we have outside domestic investors and we have foreign investors. So once again, foreign investors are dominating here but we also see the second best, so second best to domestic investment implied by these estimates is for an outside domestic investor and MEBO is actually not far behind. So giving the firm to the employees on the basis of these results is not a much worse than giving it to outside domestic investors. Voucher privatization, and I have to caution that here, this distinction among methods, we can only actually measure in Romania. And so Romania had a voucher privatization program that yielded extremely dispersed individual ownership and that effect is typically the smallest among all the different methods. So methods of privatization do matter or that's the implication of this analysis. When we look by country, so we have five countries here, these are the five, Russia, if you're especially interested in Russia is the fourth one over and Russia is frequently having a negative effect overall. It might be because those privatizations were voucher or mass privatizations. I'm gonna come back to talk about Russia a little more in just a second. But in countries that were joining the European Union like Hungary and Romania, the effects are typically where they're estimated precisely, they are, so that's the first and the third of these columns here, typically large and positive, actually slightly larger in Romania compared with Hungary. There's international heterogeneity in terms of how effective privatization has been. Finally, if we look at the business environment and what I'm doing here is using the EBRD's average score of the progress and transition to a market economy made by these countries and seeing how the privatization effect varies with respect to that. The data are unambiguously implying that a better business environment is associated with a stronger effect of privatization. You might say, well, that's obvious, but actually it hasn't been obvious for a lot of people. It was frequently argued that privatization could actually substitute for the business environment. So in the absence of a good environment, so with other kinds of constraints, other disciplinary devices on management, such as financial markets, well-functioning legal system, et cetera, it was argued that privatization might be more effective. So what we're finding here is different from that. Okay, Russian anomalies, Russia's performance effect of privatization tends to be negative. The wage effect also tends to be negative. The employment effect tends to be positive. I'm not talking much about wages and employment here, although I'd be happy to if you have questions. But Russia is basically an outlier in just about any analysis that one does. So it merits some special attention. Well, one thing we can do is, Russia also has the advantage of being a very large country and it consists of 75 odd subjects of the Federation like states, oblasts, they're typically called. So we can estimate a privatization effect separately for each of them. If we do that, we draw a histogram of those. What we find is that, yes, the distribution is centered around zero or a little bit to the left of zero, but we have some oblasts, some parts of Russia where there's a positive effect of privatization that actually looks like central Europe. And then there's some parts where the estimated effect is large and negative. Where are those different parts of Russia with those different effects? Well, it turns out they're all over the place. So here, a darker, this is the estimated effect. So here, a darker color implies a larger effect and it kind of looks here just because we have these huge Siberian oblasts that much of it is dark, but in fact, the dark ones are scattered everywhere here. And so geographically, it's not true that European Russia on average has a stronger effect than Asian Russia. Okay, well, so the results which we published on this topic, which got a lot of attention generated some controversy, used data only through 2002. Unfortunately, Russia stopped the systematic collection and release of data and it was much harder to get, but we've obtained data through 2005. And here what we're doing is we're allowing that effect to vary by calendar year. So most of the Russian privatization took place in 92 to 94. So if we allow the effect to vary by calendar year, what we obtain is, and in fact we had already reported, was this big dip in the late 1990s, then coming back close to zero, but what we hadn't seen was this big rise afterwards. So if you plot this for all the countries, the other countries are going to look like this. So Russia in the early 2000s was essentially catching up to where Hungary and Romania were in the mid 1990s. Foreign privatization effect is basically always positive. We have a small blip down, but not statistically significant with some drop off towards the end. So I'm really hoping we're going to be able to get some more recent data and see what has happened with this, but one interpretation could be this was around the time where Horokhovsky was being imprisoned or first brought to trial. And so foreign investors were feeling less comfortable, less willing to invest than they had been formerly. Okay, so just to conclude some lessons. If you asked me for the average effect of privatization on productivity in these economies, these would be conservative estimates. So these are from the estimates including firm specific trends, so-called those the best estimates. The foreign effects are much stronger and the foreign effects also tend to be more uniform. So even in Russia, there's a strong foreign effect. So it's a domestic privatization effects that vary much more. Methods matter. So sales seems to produce a stronger performance effect than mebos do and those are better than voucher or mass. The more that's privatized, the stronger is going to be the effect on firm performance. Of course, they're trade-offs with speed and politics that are involved here because it's gonna be much faster to do a voucher privatization than to engage in case by case sales. By the way, the employment and wage effects are usually small while the Russian average is masking large regional variations. So maybe it's not really a function so much of the method of privatization, but instead of aspects of the local environment and that's a topic also that we have written on. Finally, the privatization of the business environment are compliments rather than substitutes so it does make sense to try to improve institutions before or while privatizing. And I have some supplementary slides which I'd be happy to refer to if you have questions. Thank you.