A Model of Learning in Sequential Bidding: Voucher Privatization in the Czech Republic: Wave I

Jan Hanousek* and Eugene A. Kroch**
January 1995

By the end of February, 1995, five years after the fall of the communist government, over 80 percent of the assets of the Czech Republic will be in private hands. Apart from the unification of Germany, this transfer of public property amounts to the largest of its kind ever*. It is at the very core of the transformation of the Czech Republic to a modern capitalist state. And its success makes it a model for other Eastern European countries. Among the many dimensions of the Czech privatization program, by far the most important in terms of asset value is the voucher privatization scheme, which is responsible for transferring all or part of over 1650 large enterprises formerly owned by the state. In this process citizens were given voucher points that they could use to bid for shares of designated firms in a series of price-administered bidding rounds. The recent implementation of this scheme, as the main instrument of ownership transfer, offers a unique opportunity to study the privatization process and the behavior of new investors in an emerging capitalist state.

Czech privatization has proceeded on several fronts. In the early years of the transition (1990 and 1991) the former Czechoslovak parliament authorized the restitution of small properties to their former owners. Most of these were shops and restaurants, housing, and farm land, valued between 75 and 125 billion Czechoslovak crowns (CsK). A second program followed in mid 1991 under the designation "small-scale" privatization. These were small enterprises that were offered in public auction, the last of which occurred in late 1993 and reached sales of 31 billion CsK. Other programs during this period included the transformation of cooperatives (principally agricultural) and the transfer of about 350 billion CsK of property to about 6000 municipalities. But the vast bulk of the transfer has been occurring under the "large-scale" privatization of state-owned enterprises worth almost 900 Czech crowns (CzK), or $30-35 billion in the Czech Republic alone (*. Firms could remain whole or be separated into subsidiaries. Each proposed unit could then lic in one of five ways: (1) transformation into a joint stock company with shares offered to voucher bidders, (2) direct sale to a designated buyer, (3) public auction, (4) public tender, and (5) transfer to a public institution. As shown in lic in one of five ways: (1) transformation into a joint stock company with shares offered to voucher bidders, (2) direct sale to a designated buyer, (3) public auction, (4) public As shown in Table 1 over 86 percent of the large privatization projects (in terms of asset value) approved by the Czech Ministry of Privatization and the Federal Ministry of Finance by the end of 1993 were of the first type; and almost all (93.6%) of the 1774 joint stock compa nies participated in the voucher privatization scheme, offering almost half (45.5%) of their shares to voucher bidders. The book value of these shares was 343 CzK, hence making voucher privatization by far the most important single type of asset transfer .
This paper examines Czech large-scale voucher privatization, which for the first wave took place over five rounds of bidding in 1992. It examines the voucher mechanism from the standpoint of investors, pricing, and the allocation of shares. Investors co uld participate as independent individuals (IIs) or by assigning some of all of their voucher points to investment privatization funds (IPFs).
The transition from a command economy to a market economy requires the establishment of private capital markets. As a consequence of distributing the wealth of the state to the citizenry, the voucher privatization scheme guides the formation of these cap ital markets. Moreover, studying the voucher process and its outcomes can shed light on the dynamics of these emerging markets. It not only determines market prices and allocates ownership shares of enterprises, but it establishes the relationships between the public and these markets*. The focus of this study is on investor participation in the bidding process in order to understand whether individuals (IIs) behaved differently from funds (IPFs)*. The principle hypothesis o that IIs, in particular, could benefit by learning from one round to the next and by observing the behavior of others.

Voucher Scheme

The voucher scheme provided for the allocation of shares of these firms under the following procedure: Each adult citizen was entitled to buy a voucher book with 1000 investment "points" for 1000 crowns (about a week's wages). Seventy-five percent of tho se eligible (about 6 million Czechs) participated, which (as a result of dividing total book value by the number of participants) gave each voucher book an accounting value of about 35,000 crowns. Citizens used these voucher points to buy shares of eligi ble enterprises in a series of bidding rounds. Before the bidding began individuals had the opportunity to allocate some or all of their points to any of over 400 IPFs. About 72% of all voucher points were deposited into these private funds. For many I Is these funds were a way to tap into professional investment expertise, as well as a way to diversify their portfolios.
At the start of the process the public was given basic financial information about each enterprise to be transferred. At the beginning of each round participants were told the administered price of the shares of each firm and the number of shares offered . The first round started on May 18, and ended on June 8, 1992. During that time participants submitted written bids in voucher points for the number of shares desired of any firm. For the first round the share price was set uniformly across firms at 3 shares per 100 points, because the number of shares assigned to each firm was based on accounting valuation so that each share represented the same book value (about 1200 crowns) for every enterprise.
The process was by no means a standard auction, since investors bids were quantities and the prices were in fact administered by the privatization authority. The rules for awarding shares in any round were as follows: If bids for a firm did not exceed its supply of shares, these demands were satisfied and the remaining shares were deferred to the next round. But if the demand exceeded supply, then no bids were accepted and all shares were deferred to the next round*. Although th rity never revealed its algorithm for adjusting share prices between rounds, but it was generally observed that prices would rise for shares in excess demand and fall for shares in excess supply.

Approach and Methods

The basic approach to answering the questions raised in the introduction is to cse suitable econometric techniques to analyze the first waves of voucher privatization that took place in the Czech Republic from May until December, 1992. As shown in Table 1, the first wave involved 988 enterprises and shares valued at 212.5 billion CzK. *. The central working hypothesis is that IIs take into account the previous bidding behavior of IPFs. A simple dynamic model of share demand, which takes into accoun past bidding behavior and fixed firm effects (NUi) has the following form:

DII(i,t)=RHO(t)*DII(i,t-1) + BETA(t)*DIPF(i,t-1) + GAMMA(t)*P(i,t) + SIGMA(t)*S(i,t) + NU(i) + U(i,t)

where Latin letters indicate logarithms*. DII(i,t) and DIPF(i,t) denote demands of IIs and IPFs, respectively, corresponding to the i-th firm at round t;
P(i,t) is the price of a share, and
S(i,t) is the supply of shares of the ith firm at round t.
The first term is the autoregressive term, accounting for the influence (RHO) of past information; the second term reflects the extent that IIs learn from or follow the IPFs (BETA); the third term is the price effect (price elasticity GAMMA; and the fourth term accounts for the number of outstanding shares in the bidding strategy (SIGMA). By working with first differences and allowing variation across rounds, we eliminate the firm level nuisance variables (NU(i)).
In a dynamic model such as this with fixed effects, classical estimation techniques yield biased (inconsistent) estimates, as demonstrated by Nerlove (1971) and Nickel (1981). To eliminate this bias we employ an instrumental variable estimator, where the instruments come from the static (firm-level) variables and twice lagged dynamic variables from prior bidding rounds. Efficient estimation requires that we treat the round-by-round bidding equations as a system and employ the "generalized method of mome nts" (GMM) technique, which can handle the appropriate set of instruments and impose the proper orthogonality conditions. We designate this standard efficient model Model 1. But in our application there are two problems with this model. One is that we lose observations because 170 firms (whose shares represent over 20 percent of the total) are sold out before the final round of bidding, as shown at the bottom of Table 2. The other is that the firms that sell out may not be full entire sample, hence causing selection bias in our estimates. This bias would tend to increase as the bidding rounds proceed and more firms sell out. One way to handle the first problem is to estimate the model separately for each round, as Model 2, inc reasing in the effective sample size. To deal with second problem (selection bias) we implement the Heckman (1979) two-step procedure as Model 3, where the selection bias correction term (lamda) is based on the static firm characteristics.

Results

The most obvious way to compare the behavior of IIs and IPFs is to look at the outcomes of their bidding round by round. Table 2 shows that IPFs spent their voucher points more quickly than IIs. For example, by the end of the sec virtually 80 percent of their points, while IIs spent less than two-thirds. In each of the first three rounds the IPFs had a substantially higher proportion of their bids accepted. It appears from the downward trend of prices over the rounds that IPFs p aid a premium for the shares they received. The price patterns of Table 2 confirm this impression, showing that round by round the mean price that IPFs paid for accepted bids was about 50 percent higher than the mean price paid by strengthens the suspicion that either (1) IPF bidding was motivated by other considerations than price or (2) that IPFs had better information than IIs and could assume the leadership in bidding.
This leadership role of the IPFs appears to be recognized in the bidding behavior of the IIs, according to our three estimated dynamic models, tabulated in Table 3. The three models give similar results. Model 2 closely resembles e selectivity parameter in Model 3 is statistically significant in rounds 4 and 5. Most striking is that in all models the learning parameter (BETA) is significant in every round. It falls abruptly from the third to the fourth round. Not coincide is in the fourth round that the autoregressive parameter (RHO) is significant and larger than the learning parameter; but in the other rounds r is not significantly positive and in these rounds (especially round 3) learning dominates. The price elasticity (GAMMA) is quite large, but falls by the fifth and final round, reflecting the narrowing of choices and the fact that voucher points are useless after the final round of bidding. In this respect, as well, the results from Models 2 and 3 are more plausible, since we would expect in the fifth round that bidders would display unitary elasticity if they follow the authority's exhortation to resubmit identical bids from the previous round's excess demand firms*.

Conclusions

The first wave of voucher privatization in the Czech Republic was remarkably successful in allocating shares of targeted state enterprises. The bidding process was crude in many ways, especially in the administration of share prices and in the attempts b y the privatization authority to artificially speed the process by overadjusting prices. But in spite of the artificial price jolts, the market reacted logically, even predictably. In five short rounds over a few months almost all shares were allocated and almost all voucher points were spent. IIs,taking their cues from the mutual funds (to whom they attributed better information), tried to get the most value for their vouchers. And the IPFs, guided by considerations other than short-term portfolio m aximization, tried to acquire shares even at premium prices.
The openly public way that shares were transferred from the state to private hands ensured that no individual or group of investors could reap windfall gains at the expense of the general populace. The experience convinced many skeptics that even in a co untry unaccustomed to free enterprise, the market is a powerful force that can be more equitable than other paternalistic restitutions. For this reason many other Central and Eastern European countries, such as Romania, Bulgaria, Poland, and Ukraine, are examining the Czech model very seriously to adapt it to their own privatization plans.
References