The New Palgrave Dictionary of Economics

2018 Edition
| Editors: Macmillan Publishers Ltd

Output Fall – Transformational Recession

  • Barry W. Ickes
Reference work entry


A significant decline in GDP has been a common feature in transition economies. This sharp drop in output has been seen as a surprise and puzzle to many observers. Understanding the nature of the output fall is crucial to understanding transition. Analysis is complicated by measurement issues associated with moving from plan to the market. Theoretical models of the output fall are examined, including those that see the output fall as a natural consequence of the legacies of the Soviet-type economic system.


Arrears Asset specificity Command economy Coordination problems Double marginalization model Hold-up problem Incomplete contracts Incomplete information Monopoly National income measurement Output fall in transition economies Over-Industrialization Planning Price controls Price liberalization Privatization Search frictions Second economy Technical complementarities Trade dependency Uncertainty problems Value destruction Wage rigidity 

JEL Classifications

A significant output decline has been a common feature in transition economies. To some extent this is a surprise: transition represents the removal of (highly significant) distortions. See, for example, Blanchard (1997, p. v): ‘The fact that the transition came with an often large initial decrease in output should be seen as a puzzle. After all, the previous economic system was characterized by a myriad of distortions. One might have expected that removing most of them would lead to a large increase, not a decrease, in output.’ Or as Svejnar (2000, p. 8) notes, ‘The depth and length of the early transition depression was unexpected.’ Similarly, Robert Mundell has written:

The first and most obvious conclusion is that output contracted by a cumulative percentage never before experienced in the history of capitalist economies (at least in peacetime). Early denials that the contractions were occurring have proved to be incorrect. We observe that cumulative contractions over the 1990–4 period ranged widely, from a low of 18% to a high of more than 80%. (Mundell 1997, pp. 97–8)

Hence, a simple neoclassical argument would predict that output would rise rather than fall as the transition starts. Yet output fell in each transition economy, and quite significantly. The officially reported cumulative output decline for 26 transition economies from 1989 to 1995 was 41 per cent. Of this, the average decline in central Europe was 28 per cent and in the former Soviet Union it was 54 per cent (Fischer and Sahay 2000, Table 1). By comparison, output in the United States during the Great Depression declined by 34 per cent. The ubiquitous nature of the output fall thus represents an important puzzle for transition economics, and understanding the causes and nature of the output fall is crucial.

Analysis of the output fall is complicated by important measurement issues. In the change of economic systems from plan to market, the valuation of goods and services changes dramatically. This makes it important to distinguish official measures of the output fall from welfare-based measures.

Stylized Facts

It is useful to begin with some stylized facts about the output fall. Official GDP measures of output are given in Figs. 1 and 2. It is evident from these figures that in all transition economies output follows a U-shaped pattern. This represents another interesting puzzle. A theory based on the chaotic nature of the collapse of planning might predict that output would collapse at the start of transition, but would rise from that point. The pattern displayed by the transition economies, on the other hand, suggests that the peak output fall occurs with a lag of several years. So an additional part of the puzzle is to explain why the output fall intensifies in the early transition.
Output Fall – Transformational Recession, Fig. 1

Official GDP growth in central and eastern Europe (Source: International Monetary Fund Dataset)

Output Fall – Transformational Recession, Fig. 2

GDP in the former Soviet Union, 1989–2000 (Source: International Monetary Fund Dataset)

Measured output fell in all transition economies. Generally, the declines are larger in the former Soviet Union (FSU) than in central and eastern European economies (CEEs). For example, using 1989 as the starting point, the falls in Poland (15 per cent), Hungary (18 per cent), and the Czech Republic (21 per cent) were relatively moderate compared with Russia, where from 1991 GDP fell by 40 per cent. Later reformers appear to have larger falls: Ukraine has had a very significant fall in output. (There is, however, a puzzle concerning the output path of Uzbekistan. The output fall was smaller there than in any former Soviet republic, yet it reformed the least. For an analysis, see Zettelmeyer 1998.)

If we look at industrial output, rather than GDP, the observed declines would be even larger: about 40–50 per cent in central Europe and 50–60 per cent in the FSU. The reason, of course, is that most of the negative value added under planning was in industry, so we would expect a larger contraction there.

The decline in investment, especially in inventories and housing, was even greater than the decrease in GDP. This is especially true for defence. Hence, consumption has fallen less than GDP. In a sense, this is not a surprise as investment is more volatile than output in market economies. Yet transition as an economic process involves restructuring, and this does require investment. The fact that investment absorbed so much of the shock means that the resumption of growth was delayed even further. But it also means that living standards have not fallen as much as GDP. This is important for considering the welfare effects of the output decline.

Measurement Issues

Perhaps the output fall is overestimated. (Aslund 2002, p. 121, considers the output fall to be a myth.) There are many problems with interpreting official data in the context of transition, especially with respect to living standards: too many, indeed, to discuss here. Tracing output dynamics in the transition is complicated by the measurement issues that arise as the economic environment changes from central planning to market forces. Hence, an important issue in understanding the output fall is to gauge the extent to which it is a statistical rather than a real phenomenon.

Some observers (Aslund 2002; Campos and Coricelli 2002) argue that the size of the output fall is overstated because of the growth in the size of the shadow economy in early transition. It is argued that the hidden economy grew substantially during the transition period. Hence, actual production fell by less than measured output. The factual basis of this claim is controversial, however. It is also suspect theoretically. The biggest incentive to growth in the second economy is price controls. Hence, price liberalization should result in an immediate drop in the size of the shadow economy. The countervailing pressure could come from tax incentives, but it is hard to believe that this force is stronger than the impact of price controls.

The typical evidence cited in support of the proposition that the hidden economy grew in transition is that measured output fell by more than electricity production. Estimates based on comparing electricity consumption and GDP assume that the elasticity is close to unity. But this elasticity is well below unity in market economies during recessions, so employing the unit elasticity assumption amounts to assuming away the phenomenon to be measured. In Finland, for example, real GDP fell by about 11 per cent from 1990 to 1993 while electricity consumption rose by 5.5 per cent (Statistics Finland). By the logic of the advocates of the power consumption thesis we are led to conclude that the hidden economy exploded in size over these three years. For example, if the hidden economy initially was five per cent of total output, then for electricity consumption to rise with no change in intensity of use the hidden economy would have had to grow by 319 per cent! This seems hard to believe. A more likely explanation is the decline in capacity utilization that occurs in recessions causes kilowatt hours of electricity per unit of GDP to increase.

Moreover, as shown by Alexeev and Pyle (2003) the frequently cited estimates of Johnson et al. (1997) assumed no growth in the size of the shadow economy of the Soviet Union from the late 1970s to the collapse of the system. (The same error is made by Aslund 2002, p. 122.) This assumption is rejected by all observers of the Soviet economy. Hence, these empirical estimates of the growth in the shadow economy are based on too small an estimate of its initial size.

A second measurement problem in assessing the output fall arises because of the inadequacy of the inherited statistical system to cope with a market economy. Command economies, by their nature, focused on population statistics with regard to output. This is natural in a planned economy where the output produced was the result of a central plan. Indeed, the very nature of command required the planners to coordinate output, hence the statistical system needed to record what each enterprise produced. (Of course, in practice, this was difficult, as discussed in command economy.) The demise of the planning system weakened the authority of central statistical systems. More importantly, new entry became increasingly important in market economies, and the inherited statistical systems are not organized effectively to capture this.

It is also argued that under command systems enterprises had an incentive to overstate output in order to achieve bonuses, while firms in market economies want to hide output in order to avoid taxes (for example, Shleifer and Treisman 2004). It is thus argued that much output is simply missed by the change in the incentive to report. While it is certainly the case that firms have an incentive to hide output – especially when the financial system is undeveloped so they cannot seek external finance – the incentive to over-report under planning is less clear. Enterprises in planned economies were subject to the notorious ratchet effect. Higher production today meant higher output targets in the future – essentially a highly progressive dynamic tax system. The typical response to the ratchet effect was to produce only as much as needed to satisfy the plan. Hence, it is not at all clear that enterprises over-reported output in the command system.

A more important reason to question the magnitude of the output fall is the contraction in value-destroying activities. Because prices were distorted in planned economies, a portion of economic activity actually destroyed value at market prices. The contraction in these activities represents an increase in welfare, and correctly measured represents an increase in national income as well. The problem is that at the prices that prevailed in command economies this output appeared to be valuable; hence the contraction is measured as a fall in output.

There are two aspects to this decline. First, the separation of domestic from world prices means that activities that produce value added at domestic prices could destroy value at world prices. Given the underpricing of raw materials and overpricing of industrial goods characteristic of planned economies, this was more than a theoretical possibility. External liberalization then leads to a contraction of these activities (McKinnon 1991). The second aspect is that domestic prices were similarly distorted so that domestic price liberalization has a similar effect. This is discussed below.

To the extent that a reduction of value-destroying activity occurs at the same time as output falls, it is clear that movements in measured output are not consistent with movements in welfare. Indeed, if a greater measured output fall is associated with a faster removal of value-destroying activities, then it is likely that welfare is enhanced by the output fall. In this case the output fall is associated with more reform and quicker removal of welfare destroying activities. (This also means that output recovery could mean a resurgence of value-destroying activities, in which case the upward-sloping part of the U shape is welfare decreasing. Unlikely, but it might be relevant for Belarus under President Lukashenko). Of course, for this to be the case there must a serious distortion in national income measurements. To the extent that output measurements use base-weighted prices this is possible.

It is difficult to measure the extent to which the output fall is overstated by the contraction of value destroying activities. For example Aslund (2002, p. 126) estimates that about 20 per cent of GDP was value destroying in the last years of Communism. He uses, as an indicator, the decline in the share of industry in GDP. Soviet-type economies were over-industrialized, and liberalization led to sectoral shifts as services, which were previously undersupplied, expanded. Moreover, shifts in relative prices, discussed below, also lead to a reduction in the share of value added produced by industry. Thus, one cannot infer value destruction from the change in industrial shares. The general problem is that output may be falling for various reasons so one cannot consider all of the contraction to be previously value destroying. One valuable indicator of the importance of value destruction is given by the comparison of the contraction in industrial output with the rise in consumption that occurred in transition economies. In Russia, for example, industrial output contracted by roughly 35 per cent from January 1992 to January 1994. Real disposable income, on the other hand, increased by almost 70 per cent in the same period (albeit from depressed levels). The fact that real disposable income was growing at the same time as industrial output was contracting suggests that the cessation of value-destroying activity was an important process, and that some of the output fall may be overstated.

A related problem is the shift in preferences. Gaddy and Ickes (2003) argue that a specific index number problem leads to an overstatement of the output fall – the camellia effect. The argument is easily understood in terms of an analogy. Consider a flower shop that specializes in the sale of extremely rare camellias. Cultivating these plants is inordinately expensive, but this activity is profitable because the shop has a customer willing to pay very high prices for camellias. Now suppose this customer passes away. The shop can no longer sell rare camellias at a price that covers the cost of production. So camellia cultivation ceases. Resources that were previously devoted to camellia production will now be used for something else, say, roses. Profits at the flower shop fall because camellias were very profitable as long as their special customer lived. But given that there is no longer a market for rare camellias (while there is a market for roses, everyone is better off with rose cultivation than if they continued to cultivate camellias as if nothing had changed.) In the Soviet regime defence output was demanded despite the enormous cost. It had value as long as the Communist Party had command over resources. The special customer of Soviet times made it ‘valuable’ to produce defence output. When the Soviet system collapsed, so did the special customer. Output thus fell – valued at Soviet prices – because at those prices defence output was valued far above cost. After the fall this output is not valued sufficiently and production declines. This is an output fall, but welfare is certainly higher with lower defence production given that the Communist Party is no longer the measure of value.

To see this, suppose that we have two final goods, (x1, x2), and that the pre-transition production bundle is \( \left({x}_1^A,{x}_2^A\right) \), where good 2 is defence output, and A represents planners preferences. The post-transition allocation is \( \left({x}_1^B,{x}_2^B\right) \), and reflects social preferences. We might consider, for example, that at point A there is large military production and little civilian production, reflecting planners’ preferences (UP). The new production bundle is at point B, based on society’s preferences. Note that using pre-transition prices to value output, GDP is \( {Y}^A={\sum}_i{p}_i^A{x}_i^A \).

Now suppose that liberalization causes the production bundle to move to point F in Fig. 3. This is the most pessimistic outcome – demand for x2 declines with almost no increase in x1. Measured in real terms, at the old prices, output falls approximately by the distance AF in units of x2, or \( {\sum}_i{p}_i^A{x}_i^F-{\sum}_i{p}_i^A{x}_i^A \). But this greatly overestimates the welfare change, because it places a high value on the output that has fallen in valuation.
Output Fall – Transformational Recession, Fig. 3

The camellia effect

Although output has fallen precipitously at planners’ prices, measured at the new prices welfare has clearly increased. The minimum expenditure to achieve the old welfare level \( e\left({P}^B, {U}_A^P\right) \) is less than the cost of purchasing bundle F at the new prices. It is evident that welfare is higher at point F than at point A. Output has risen at the new prices but has fallen at the old prices.

From Fig. 3 we can also distinguish the fall in output due to coordination-type failure and that due to measurement. If resources are fully utilized we would be at point B. Hence \( {\sum}_i{p}_i^B{x}_i^B-{\sum}_i{p}_i^B{x}_i^F\equiv \Omega \) measures the fall in output due to coordination-type failure. The measured fall in output, could be larger or smaller than this. The key point, however, is that the measured fall does not measure Ω at all.

Notice that, if the resources devoted to defence production are highly specialized, then there may be great inertia in response to the demand shift. It may be very hard to find alternative uses for these inputs. Output may remain depressed for quite a while. There may also be interesting behavioural issues to think about. A Russian defence enterprise director may expect that the government will soon restore orders and that cuts were temporary. This would lead to inertia in shifting to new activities. Both of these inertial forces could prolong the decline in output.

The importance of the camellia effect for thinking about the output decline is especially important in comparative terms. The camellia effect explains why transitional recessions are observed. But the size of this drop will be proportional to the share of ‘camellias’ in GDP, and this clearly differs across the post-Communist world. (Even for the former Soviet Union the differences are dramatic, as Russia had a much larger than average share of Soviet defence industry; see Gaddy 1996.)

In a country like Russia the size of the defence sector was especially large. This exacerbates the size of the output drop that is due to transitional factors. To measure the pure transition effect we should compare what would have been produced under central planning had planners’ preferences not determined production decisions with what happened during transition. Ignoring the camellia effect mixes the two sources of output fall.

Theories of the Output Fall

Theories of the output fall in transition generally fall into one of two classes. The first class of theories treats this phenomenon as a sign of inefficiency. The output fall is thus welfare decreasing. The second class treats the output fall as a natural feature of liberalization but does not consider the fall to be welfare reducing. (One could also consider the specific negative shocks that have caused output disruptions. For central Europe there is the breakup of Council for Mutual Economic Assistance (CMEA) trade plus the end of subsidized energy from the Soviet Union. For the former Soviet Union there is the disruption in trade caused by the breakup of a common economic space into 15 independent countries. For Russia, there is the decline in oil prices. The importance of movements in the oil price for Soviet and Russian output has been emphasized by Gaddy and Ickes 2005. The power of this explanation has been fortified by the close timing of the recovery of Russian output with the increase in oil prices starting in the later 1990s.)

A basic framework for thinking about the output fall is the reallocation problem. Consider an economy with two sectors, state (S) and private (P). Initially all labour is employed in the state sector. It is assumed that labour productivity in the private sector (β) exceeds that in the state sector (α), α < β. The reallocation process occurs as labour moves from the state to the private sector. Per-capita output, yt, is thus given by
$$ {y}_t=\alpha \frac{L_t^S}{L_t}+\beta \frac{L_t-{L}_t^S}{L_t}; $$
it is immediately apparent that rather than decline, output will increase monotonically in the transition. Hence, to obtain an output fall some unemployment of resources is necessary. If the private sector cannot absorb all the labour released from the state sector then labour will be unemployed, LU. In that case per-capita output is given by
$$ {y}_t=\alpha \frac{L_t^S}{L_t}+\beta \frac{L_t-{L}_t^S-{L}_t^U}{L_t}. $$

This simple framework suggests that to produce an output fall some rigidity or friction is required that prevents smooth reallocation of the labour released from the state sector. The essence of transition suggests that this will be likely. In addition to the normal culprits such as wage rigidity, institutional features play a critical role. For example, prior to the privatization of state sector assets, capital is immobile between sectors. This naturally limits the absorption rate of the private sector. Hence, the exit rate from unemployment will depend on the rate of growth of the private sector. What is important to understand are the determinants of the exit rates from these states. Notice that the growth of the private sector may depend on what is happening in the other sectors. This dependence can occur for several reasons. First, following Aghion and Blanchard (1994), unemployment can cause fiscal deficits which must be financed at the expense of the private sector, limiting its growth. Second, the growth of the private sector may depend on the rate at which complementary resources are released from the state sector. This is especially true for the most basic of resources for production, space. Until privatization of fixed capital takes place it is difficult for new private enterprises to obtain space for production, let alone to lease equipment.

At the most basic level, unemployment can be due to rigidity in real wages. But it is hard to understand how this can explain the output falls that were actually observed, as real wages fell in most transition economies once prices were liberalized. Hence the need for more fully developed theories.

Double Marginalization

Li (1999) develops a theory of the output fall in transition based on double marginalization. The basic idea is that the dismantling of central planning or centralized organization of production permits monopolistic and vertically interdependent enterprises to pursue their own monopoly profits by restricting output and intermediate trade to the detriment of the economy as a whole. The basic idea is that the collapse of planning institutions removes constraints on intermediate producers’ activities. Intermediate producers now have monopoly power, so they raise prices. This happens all along the supply chain, and results in an increase in the cost of producing final output. So there is less final output available and government output falls. The essential reason is that the enterprises do not consider the consequences of their price increases for the profits of the other enterprises. Since there is less left over for consumers, it is equivalent to a decrease in real wages, and hence labour supply falls.

The essential idea of the double marginalization theory is that output falls because liberalization precedes the development of competition. Entry is a process that takes time. Hence, the theory would predict that output falls would be greater in economies that are less able to ‘import’ competition through opening the economy. This roughly fits the picture of larger output falls in the FSU than in the CEEs. But the theory also predicts that the output fall should be largest when liberalization first takes place, since that is when market power is most potent. The effect of double marginalization should wane over time. This is harder to reconcile with the paths of output in Figs. 1 and 2.

The double marginalization model also predicts that each enterprise will face a contraction in demand and an increase in input prices relative to wage rate. The contraction in demand is attributable to the following factors in this model: the decline in real wage rate, the decline in the government’s real income and the decline in input demand. The increase in input prices relative to wage rate is attributable in this model to monopoly pricing by a ‘web of monopolies’. The more complex is the web of inter-industry production, the greater the propagation of the price shock. Hence, complexity magnifies any intermediate price markup throughout the economy, resulting in higher input prices relative to wage rate. The sharp increase in input costs is indicative of a sharp supply contraction. This prediction is also consistent with empirical observations.


Blanchard and Kremer (1997) (see also Blanchard 1997) have developed a model of disorganization that has had great impact. Their argument is that the output fall is a result of the chaos that surrounds the elimination of central planning. They focus on three mechanisms (hold-up problems, coordination, and uncertainty problems) that are greatly magnified as the result of missing institutions likely to be important at the start of transition. The basic idea is that the collapse of planning causes performance to decline during the period when alternative market mechanisms have not yet developed.

The basic idea can be understood in terms of a simple example presented by Blanchard and Kremer. Consider a vertical chain of production. Assume that each step is carried out by a different enterprise. A unit of a primary good is needed at the first step. At the end of the n steps one unit of the final good results, and we normalize the price of this good to unity. The value of the intermediate output, at each step, is zero. The supplier of the primary input has an alternative use, which is c. This could be much lower than one. It is a private opportunity that could be exporting the good, or selling it for a less fabricated use. Under planning the relations in the chain were directed from above. With liberalization alternative activities may be considered.

The end of planning thus leads to n bargaining problems. Each unit must bargain with a supplier and a customer. They assume that there is Nash bargaining at each step, so that the surplus is split given the symmetry of the situation. To see what happens start with the last step. The value of the surplus in the last stage (bargaining between the final producer and the last intermediate producer) is 1. This follows because the value of the good at stage n is still zero. So the last intermediate producer gets one half of the surplus. Similar bargaining takes place at all the upstream stages. At the n − 1 stage there is one half to split ... Continue in this fashion and it follows that the first intermediate producer gets \( {\left(\frac{1}{2}\right)}^n \). The surplus available to split at the first stage is \( {\left(\frac{1}{2}\right)}^n-c \), since the first producer must purchase the primary input to produce. It is thus clear that unless \( c<{\left(\frac{1}{2}\right)}^n \)the raw material will be diverted and production will cease. Moreover, c does not need to be all that large to trigger defection that results in a fall in output that could be as large as \( 1-{\left(\frac{1}{2}\right)}^n \). Thus rather meagre private opportunities can cause a rather large fall in output.

Blanchard and Kremer interpret n as the level of complexity of production. As n increases, the likelihood of defection increases exponentially. This is a hold-up problem. Each producer in the chain must produce before bargaining with the next in line. This suggests that the problem would go away if each of the producers could sign an enforceable contract before production takes place. As long as c < 1, defection could be avoided and production could take place, if the intermediate producers could sign a contract to split the 1 − c before production. The problem is thus one of asset specificity and incomplete contracts. Eliminating the ministry before institutions that support contracts are developed is the source of the problem. Vertical integration could help, but this requires ownership to be specified, another problem early in transition. The notion that producers in transition could suffer from this problem is not far-fetched. (It is interesting to compare this outcome with the double marginalization case. Notice that in that case the raw materials producer has market power and thus a higher share of the surplus than is the case in the bargaining problem. This makes production in the state sector more likely. Of course, what is not explained is why the producer is able to extract monopoly rents in a situation of bilateral monopoly.)

Blanchard and Kremer consider other examples based on incomplete information. A state-owned enterprise must negotiate with many suppliers that may have outside options. Each of the suppliers produces a key input without which production is impossible. With uncertainty over the magnitude of outside options a state-owned enterprise must guess how much to pay for the inputs. When outside opportunities are low the possibility that the state-owned enterprise offers too low a price is negligible. But as these outside opportunities rise this probability increases. Even if it is still efficient to sell to the state-owned enterprise because of uncertainty over the size of these options, the price offered may be too low and production falls. The interesting feature of this model is that it produces a U-shaped output path. The key assumptions are technological complementarities and inefficient bargaining.

A coordination example can also be constructed. Suppose that the firm needs n workers (it could be supplying firms, but this is easier), and the technology is Leontief. If all workers stay, the firm produces one unit of output per worker. If a worker leaves, a replacement is hired with output per worker equal to γ < 1. Here again n measures the degree of complexity, while γ is an inverse measure of the specificity of the production process or job-specific human capital.

Each worker has an alternative opportunity given by c, distributed on \( \left[0, \overline{c}\right] \), where \( \overline{c} \) represents the maximum outside opportunity, which is of course a function of the state of the transition. Draws from this distribution are independent across workers. The distribution is known, but the specific realization is private information. This could be thought of as alternative employment, perhaps in a Western multinational. The firm pays a common wage, w, to all workers, equal to output per worker. This simplifies the analysis, but is probably not crucial.

The key assumption of the model is that workers must decide whether to take up the alternative before they know the decision of the other workers. This creates the coordination problem. Workers are risk neutral, so that all we need to look at is expected output. There are thus two potential outcomes: (a) all workers stay, output per worker and thus the wage are equal to unity, or; (b) one or more workers leave, output per worker and the wage are equal to γ.

The decision problem for the agents boils down to determining some threshold level of outside opportunities, c*, such that if c < c*, workers stay and vice versa. If a worker leaves he receives c. If he stays his expected earnings will depend on what the other n − 1 workers do. Assume symmetry so that the other workers also have the same c*. Then the probability that they all stay is (F(c*))n − 1, where F( ⋅ ) is the distribution function so that F(0) = 0 and \( F\left(\overline{c}\right)=1 \). Expected output per worker is thus equal to (F(c))n − 1 + γ[1 − (F(c))n − 1].

The key point is that there may be multiple equilibria, depending on the level of outside opportunities. If alternative opportunities are very low, workers always stay in the firm, and output equals 1. As outside opportunities increase there are two equilibria; in one of these output falls close to γ. With very high outside opportunities production in the state sector ceases. Note the problem here is coordination, not uncertainty. If the outside opportunity were common knowledge, with \( \gamma <c<\overline{c} \) there would still be two equilibria.

The essential feature of the disorganization model is that central planning is replaced before the infrastructure of markets is created. The lack of central organization leads to disorganization, and the development of outside opportunities makes this problem more severe. Over time, market infrastructure develops and disorganization problems are lessened.

Roland and Verdier (1999) develop a related model of disorganization, focusing on search frictions rather than bargaining problems. In their model liberalization means that enterprises can search for new suppliers and customers. There are good matches and bad matches. If too many bad clients are searching the productivity of potential matches may fall. What is critical in their model is that relationship-specific investments take place only after long-term matches are formed. If search continues this will not happen, investment demand will fall, and output can fall.

Investment specificity is crucial in this model. Without it output would not fall even with bad matches, since the partners could produce this period and keep on searching. It is the asset specificity that introduces the cost of bad matches.

The Roland–Verdier model is interesting from a theoretical point of view, but one may wonder how relevant it really is for explaining the output fall. The problem is that the initial output fall was associated with very little search for new suppliers. The predominant behaviour was a relationship-conservatism. Agents tried to maintain their relationships as much as possible. Networks of suppliers already had relationship-specific investments. The problem is that they had no customers who would purchase the goods at a price that covered their new costs.


A more subtle, but equally important explanation of the output fall focuses on the micro distortions due to Soviet pricing rules. Ericson (1999) has analysed this problem. His focus is on structural problems with Soviet pricing – the arbitrariness and non-uniformity of producers’ prices across users of the product within standard commodity aggregates. Ericson shows that Soviet pricing rules hid inefficiency and waste, creating an illusion of capacity and output that wasn’t there. The advantage of this theory is that it can explain why prices exploded when output fell. His argument is that post-Soviet ‘stagflation’ is, to some extent, a consequence of the irrational structure of production hidden in apparently consistent (adjusted) input–output (I–O) matrices and economic statistics.

Soviet pricing rules contained three systematic distortions: (a) basic factors were seriously undervalued (land was free, and capital-in-place virtually so); (b) raw materials and natural resources were undervalued; and (c) highly processed goods – in particular investment products and services – were seriously overvalued. These distortions in the principles of economic valuation used in centrally planned economies systematically hide tremendous waste, exaggerating both net outputs and net income (economic value) produced, while understating the productivity of that most seriously mismeasured factor of production, capital. This implies that the size of the apparent initial collapse in industrial production is evidently exaggerated, even if one ignores new economic activity generated in the wake of the reforms. However, the wasteful production structure can also spur a continuing and deepening collapse, as it is not economically viable in a market environment.

Ericson shows that embedding these distortions in the input–output tables that are used to create national income statistics results in lower prices for inputs than for final uses, and generates an understatement of the share of gross output used in the production process. Thus, it leads to an overstatement of the share of net output. Furthermore, these distortions cannot be revealed by any consistent input–output framework derived from the ‘value’ of transactions between sectors; the methodology itself imposes a consistency that hides those distortions. This means that the true nature of the system cannot be revealed until price liberalization takes place. Until then, intersectoral relationships are hidden. This is what creates the ‘circus mirror’ effect discussed by Gaddy and Ickes (2002). (A circus mirror distorts size and shape. Soviet pricing rules had the same effect, making value added look larger and intermediate input use look smaller). Just as an individual may look taller and thinner in a circus mirror, the Soviet-type economy appeared more productive under Soviet pricing rules. Liberalization revealed the true nature of the economy.

Ericson shows that for the case of Russia the 1991 input–output coefficients were substantially understated, hiding significant materials input use and waste, and hence obscuring much of the inherited inefficiency in the industrial structure. This inefficiency became of consequence for producers when liberalization released them from ministerial tutelage and constraints, and made them primarily responsible for covering their own costs. Because enterprises are initially constrained by existing technological structures, the first impact of liberalization is typically seen in the move to raise prices to cover their full material costs and to compensate for any increases. This led to increases in industrial prices that far exceeded the general rate of inflation, raising the real price of industrial output and consequently real materials costs. As in the double marginalization theory, price increases in the intermediate sector propagate through the economy and result in less final output. But the impulse is different. Ericson’s theory does not require any market power on the part of intermediate producers. Price increases are solely due to price liberalization itself in the context of Soviet pricing. Of course, at those increased real prices, demand for many products, now not supported by plan requirements, falls dramatically; producers find they are unable to sell at higher prices and hence unable to recover the full costs of production. Yet they continued to operate and ship output to traditional users of their product.

Ericson’s theory is thus consistent with several important aspects of the output fall that are hard to explain in other models. First, his theory explains why the output fall is associated with a rise in the price level. Second, it is consistent with higher wholesale price inflation than consumer price inflation. Third, it is consistent with the explosion of inter-enterprise arrears. Supply and disorganization type theories make no prediction with regard to overall inflation and they are inconsistent with the latter two observations.

Empirical Analysis

Most empirical analyses of the output fall has been focused on assessing the role of policies (primarily, stabilization and liberalization) and initial conditions in determining the size of the fall in output. This literature is too large to summarize here (a good summary is Campos and Coricelli 2002), but a few points can be made. First, results are very dependent on how policies, especially the speed and extent of liberalization, are measured, and how initial conditions are proxied. Measures of liberalization that rely on expert evaluation are subject to performance bias: that is, the liberalization score that is assessed is often inferred from economic performance. The set of initial conditions that are important include the degree of over-industrialization, repressed inflation, dependence on CMEA trade, distance from Frankfurt, years spent under Communism, initial income, and the rate of urbanization. Depending on the set used results can differ dramatically.

One of the most comprehensive studies of the impact of policies versus initial conditions is by Berg et al. (1999). They use a sample of 26 transition economies and use a general to specific modelling approach that allows for differential effects of policies and initial conditions and for time-dependent effects of initial conditions. They find that structural reforms are more important than either policies or initial conditions in explaining the cross-country variation in performance. Initial conditions play the predominant role in explaining the output fall, while structural reforms explain the recovery. The most important initial conditions appear to be the degree of over-industrialization and trade dependency.


Although the size of the output fall indicated by official measures is clearly overstated, the fact that output and incomes did fall in the aftermath of liberalization is not disputed. Moreover, the fact that output followed a U-shaped pattern has had important consequences for transition. Not least of these is the negative effect it had on the political support for many economic reformers. The output decline made it politically difficult to stick with reforms. Hence, the output declines may have altered the course of policy reform in transition. Ironically, it seems that reform reversals were often associated with longer output declines.

See Also


  1. Aghion, P., and O. Blanchard. 1994. On the speed of transition in central Europe. In NBER macroeconomics annual. Cambridge, MA: MIT Press.Google Scholar
  2. Alexeev, M., and W. Pyle. 2003. A note on measuring the unofficial economy in the former Soviet republics. Economics of Transition 11: 153–175.CrossRefGoogle Scholar
  3. Aslund, A. 2002. Building capitalism: The transformation of the former Soviet bloc. Cambridge: Cambridge University Press.Google Scholar
  4. Berg, A., E. Borenzstein, R. Sahay, and J. Zettelmeyer. 1999. The evolution of output in transition economies: explaining the differences. Working Paper No. 99/73, International Monetary Fund.Google Scholar
  5. Blanchard, O. 1997. The economics of transition in Eastern Europe. Oxford: Oxford University Press.Google Scholar
  6. Blanchard, O., and M. Kremer. 1997. Disorganization. Quarterly Journal of Economics 112: 1091–1126.CrossRefGoogle Scholar
  7. Campos, N., and F. Coricelli. 2002. Growth in transition: What we know, what we don’t know, and what we should. Journal of Economic Literature 40: 793–836.CrossRefGoogle Scholar
  8. Ericson, R.E. 1999. The structural barrier to transition hidden in input–output tables of centrally planned economies. Economic Systems 23(3): 199–224.CrossRefGoogle Scholar
  9. Fischer, S., and R. Sahay. 2000. The transition economies after ten years. Working Paper No. 7664. Cambridge, MA: NBER.Google Scholar
  10. Gaddy, C.G. 1996. The price of the past. Washington, DC: Brookings Institution Press.Google Scholar
  11. Gaddy, C.G., and B.W. Ickes. 2002. Russia’s virtual economy. Washington, DC: Brookings Institution Press.Google Scholar
  12. Gaddy, C.G., and B.W. Ickes. 2003. How to think about the post-Soviet output fall. Working paper, Brookings Institution and Pennsylvania State University. Online. Available at Accessed 20 May 2007.
  13. Gaddy, C.G., and B.W. Ickes. 2005. Resource rents and the Russian economy. Eurasian Geography and Economics 46: 559–583.CrossRefGoogle Scholar
  14. Johnson, S., D. Kaufmann, and A. Shleifer. 1997. The unofficial economy in transition. Brookings Papers on Economic Activity 1997(2): 159–239.CrossRefGoogle Scholar
  15. Li, W. 1999. A tale of two reforms. RAND Journal of Economics 30: 120–136.CrossRefGoogle Scholar
  16. McKinnon, R. 1991. The order of economic liberalization: Financial control in the transition to a market economy. Baltimore: Johns Hopkins Press.Google Scholar
  17. Mundell, R.A. 1997. The great contractions in transitions economies. In Macroeconomic stabilization in transition economies, ed. M.I. Blejer and M. Skreb. New York: Cambridge University Press.Google Scholar
  18. Popov, V. 2005. Shock therapy versus gradualism reconsidered: Lessons from transition economies after 15 years of reforms. TIGER Working Paper No. 82.Google Scholar
  19. Roland, G., and T. Verdier. 1999. Transition and the output fall. Economics of Transition 7: 1–28.CrossRefGoogle Scholar
  20. Shleifer, A., and D. Treisman. 2004. A normal country. Foreign Affairs 83(2): 20–38.CrossRefGoogle Scholar
  21. Statistics Finland. Online. Available at Accessed 28 June 2007.
  22. Svejnar, J. 2002. Transition economies: Performance and challenges. Journal of Economic Perspectives 16(1): 3–28.CrossRefGoogle Scholar
  23. Zettelmeyer, J. 1998. The Uzbek growth puzzle. Working Paper No. 98/133, International Monetary Fund.Google Scholar

Copyright information

© Macmillan Publishers Ltd. 2018

Authors and Affiliations

  • Barry W. Ickes
    • 1
  1. 1.