The Evolving Pattern of Internal Market Integration in Russia
Daniel Berkowitz* and David N. DeJong** Department of Economics University of Pittsburgh Pittsburgh, PA 15260.
Paper prepared for joint AEA/ACES panel on Fiscal Policy in Transition Economies, Boston, MA.
January 2000
*Tel: 1-412-648-7072; Email:
[email protected]. **Tel: 1-412-648-2242; Email:
[email protected]. We thank Mitch Mokhtari for useful discussions and providing the price data used in this study, Jean-Francois Richard for his comments and Michelle Haigh and Steven Lehrer for research assistance. Berkowitz gratefully acknowledges financial support from the National Science Foundation under grant SBR9730499.
“As winter looms and food runs low, Russia’s regions are growing hungry – for power. Behind worries that Russia might not have enough to eat lies a more menacing shortage. The central government doesn’t have enough authority to stop tree lines and hedge lines and hedgerows from turning into borders between rival fiefs and would-be statelets… What haunts Russia today isn’t the passionate separatism that galvanized the Baltic Republics into revolt in the last years of the Soviet Union and fired Chechen fighters in the 1994-96 war with Moscow. It is instead the banal and often bungling defiance of Soviet-bred barons eager to keep bread cheap, sausages plentiful and their own interests secure.” (Andrew Higgins, Staff Reporter of the Wall Street Journal, October 16, 1998)
1. Introduction The economic reforms initiated by Russia in the early 1990s were designed in part to help establish strong internal market linkages across its far-flung and diverse regions. However, recent evidence suggests that the Russian economy more closely resembles a collection of fragmented internal markets, with fiefdoms controlled by regional politicians. For example, roughly 30 percent of the Russian regions erected official trade barriers of various types following Russia’s financial collapse in the summer of 1998 (Serova, 1998). And since the early 1990s, conservative regions such as Bashkortostan and Ulyanovsk are reported to have established border controls and issued ration coupons to residents in order to limit the export of consumer goods to other regions, and to curtail non-residential consumption of low-priced goods (see Koen and Phillips, 1993; Mitchneck, 1995; and Freinkman, Treisman and Titov, 1999). While examples of regional trade barriers can be found within Russia, the pervasiveness and persistence of such barriers is unclear. Here, we use a statistical model of commodity trade to analyze the extent to which distance and regional borders account for observed differences in commodity prices across 47 regions in Russia. Monthly time-series data on regional commodity prices spanning 1994 through 1998 are used to analyze the extent to which the importance of these explanatory variables evolved over this period. Our results indicate substantial time variation: an initial period of widespread interconnectedness – signaled by the relative statistical unimportance of regional borders in accounting for regional price variation – gradually gave way to a period of disconnectedness in 1996 and 1997, which seems to have subsided by May of 1998.1 Having described these temporal fluctuations in internal market interconnectedness, we analyze how they relate to temporal fluctuations in international trade, internal transport costs, public discontent, and standards of living. We find a weak relationship between interconnectedness 1
Our time series ends before the financial crisis in the summer of 1998. As our epigraph suggests, the resurgence in integration we document here may have been short lived.
1
and international trade, but strong relationships between interconnectedness and internal transport costs, public discontent (proxied by data on strike activity and wage arrears), and standards of living. A caveat with this aspect of our analysis is that all of the variables we consider are measured at the aggregate level. Our goal in future work is to analyze how regional measures of economic performance, public discontent, etc. interact with the aggregate data considered here in influencing internal market integration.2 A growing literature has attempted to measure and explain market integration in transition economies. Gardner and Brooks (1994) studied this problem for the first seven months of the Russian reform, a period in which the federal government attempted to rapidly transform the Russian economy from a system with fixed prices and highly administered inter-regional relations, to one in which inter-regional resource allocations were guided by a flexible price system. Their analysis focused on differences in the level of food prices across cities; for a wide variety of goods, they found large differences in prices that could not be explained by distance. Since arbitrage opportunities for tradable goods do not persist in inter-regional markets within developed countries (Engel and Rogers, 1996; Parsley and Wei, 1996; Rogoff, 1996), Gardner and Brooks concluded that Russia’s market integration was limited. De Masi and Koen (1996) studied Russian price data through 1994. They also found inter-regional price dispersion to be quite high by international standards, and concluded that market integration was not advanced. On a more positive front, we concluded in another paper (Berkowitz and DeJong, 1999) that the primary source of high interregional price dispersion in Russia during the period 1992-1996 was a relatively small group of pro-Communist and anti-market-reform regions called the Red Belt. Finally, after analyzing regional production patterns in China, Young (1996, 1999) concluded that the Chinese internal market has been poorly integrated since the inception of market reforms in 1978. Naughton (1999), however, used regional trade-flow data to argue that Young overstated China’s market integration problem.
2
Regarding the weak relationship between interconnectedness and international trade we find, it bears mentioning that the widespread internal disconnectedness we observe between 1996 and 1997 coincides with a period in which Russia had a strong export performance, and attracted a tremendous amount of international portfolio investment. This timing appears coincidental: conditional on the additional variables we consider, the statistical relationship between interconnectedness and international trade is weak. Young (1996, 1999) has argued in the case of China that their unusually high trade-to-GDP ratio could signal weakening inter-regional trade relations; this does not appear to be true in Russia.
2
The studies cited above analyzed market integration over set time periods. Our goal in studying temporal changes in market integration is to begin to understand causes and consequences of these changes for Russia.
2. Methodology and Results In integrated market economies, arbitrageurs and traders can quickly move tradable goods from regions in which sales prices are cheap to regions in which consumers are willing to pay higher prices, so long as transport costs are not prohibitive. This criterion implies that within a group of integrated regional markets, price differentials for similar tradable goods sold in different regions will be increasing in the distance separating the regions. More formally, let Qij(t) denote the percentage price differential for some tradable good sold in regions i and j at some date t: Qij(t) = abs(log(Pi(t)/ Pj(t))), where throughout the paper log denotes the natural log. Also, let dij represent the distance separating regions i and j, where inter-regional transport costs are increasing in dij. In the absence of border taxes, highway robbery, breakdowns in the transport system, and other barriers to inter-regional trade, arbitrage opportunities exist at some date t when Qij(t) is greater than or equal to transport costs. However, when Qij(t) is less than transport costs, moving the good between these two regions is not profitable, and inter-regional trade does not occur. In order to pin down the relation between distance and transport costs, we follow Engel and Rogers (1996) in employing Krugman’s (1991) transport cost model, in which 1 - 1/(1+ dij) is the share of a good that depreciates when it is moved between regions i and j. Arbitrageurs can profitably buy goods in region j and resell these goods in region i when Pi/ (1 + dij) ≥ Pj; the same arbitrageurs can profitably buy goods in region i and resell in region j when Pj/ (1 + dij) ≥ Pi . So, given an integrated internal market, there will be trade between regions i and j when the relative price (Pi/ Pj) fluctuates within the band [1/(1+ dij), (1+ dij)]. This band is increasing in dij, thus a testable implication of this model is that the variance of Qij is increasing in inter-regional distance when internal markets within a country are integrated. Our goal is to measure temporal movements in inter-regional price dispersion. This goal is pursued using a regional data set that measures the cost of a uniform basket of basic food goods. The data span the period March 1994 through December 1998, and were compiled by the Russian statistical agency Goskomstat. During 1994-1996, the basket includes 19 food goods; starting in January of 1997, the basket was modified to include 26 food goods (see Goskomstat Rosiyi 19951998, various issues). The data set includes observations from 47 Russian cities: Moscow, St.
3
Petersburg, and all the major regional capital cities.3 The data are monthly, and cover 56 out of the possible 58 months during March 1994 through December 1998. Let t = 1, 2, … , 56 denote a particular month and year, and let σij(s) denote the standard deviation of Qij(t) calculated over the twelve-month sub-period indexed by s. To measure temporal movements in price dispersion, we calculate σij(s) for every possible (i,j) combination such that i ≠ j, and for every possible twelve-month sub-period in the time period spanned by our data. Since there are 47 regions in our sample, there are N = (47*46)/2 such (i,j) combinations for each time sub-period; and since there are 56 months spanned by our sample, there are 45 possible twelvemonth time sub-periods.4 Table 1 lists each rolling sub-period and its approximate median date (month and year), which is the sixth month within sub-period s. To measure market integration for region i in sub-period s, we use OLS to estimate σij(s) = α i(s) + β i(s)log(dij ) + u ij(s),
i ≠ j,
(1)
where α i(s) is the estimated intercept, β i(s) is the coefficient for log distance from region i to all other regions, and u ij(s) is an error term. As previously argued, inter-regional price volatility in integrated economies should be increasing in log distance, thus we deem region i to be integrated during sub-period s when β i(s) is estimated as positive and statistically significant.5 We define significance at the 10-percent level in the results reported below; the temporal patterns identified in this manner are robust to alternative choices. 6 Table 2 summarizes Russia’s market integration trajectory during the 45 rolling subperiods by reporting the share of regions for which β i(s) is statistically significant in sub-period s; Figure 1 illustrates the movement of these shares over time. The pattern of interconnectedness that emerges is “U-shaped”: between 1994:I and 1996:I, 60.4 percent of the regions were integrated, and the share of integrated regions fell below 50 percent only once during this period. However, at the beginning of 1996:II the share of integrated regions fell to 42.6 percent, and continued to fall until it bottomed out at a remarkable 8.5 percent in February and March of 1997. Integration then
3
A major provincial capital contains a population that accounts for at least 30 percent of the provincial population. 4 Experiments with shorter and longer sub-periods produced results similar to those reported below. 5 This test is used by Parsley and Wei (1996) to analyze market integration in the United States. 6 Standard errors used to evaluate significance throughout the paper are heteroscedasticity consistent (White, 1980).
4
jumped in December of 1997 to 68.1 percent, and continued to rise throughout out 1998: 78.7 percent of the regions were integrated during December 1997 through May 1998.7
3. Market Integration and the Aggregate Economy Our interest here is in analyzing how the temporal fluctuations in regional market integration characterized above correspond with related economic and political variables. In Berkowitz and DeJong (1999, 2000) we showed that voting patterns in June 1996 were important in explaining regional isolation observed between 1994 and 1996. Regions within Russia’s Red Belt supported the Communist Party candidate Zyuganov in the final round of elections; the rest of Russia supported Yeltsin. Zyuganov’s platform included an emphasis on price controls, price subsidies, and administered resource allocation. In contrast, Yeltsin’s platform called for further price liberalization, a cutback in price subsidies, and a deepening of the privatization process. We found that this difference in platforms was an important predictor for differences in policies between the Red Belt and the rest of Russia. Specifically, we observed significantly more retail price regulation, budgetary and agricultural-price subsidization, and less entrepreneurial activity (measured by the number of new small private firms per capita) in the Red Belt during 1995-1996. As a possible explanation of this pattern, regional governments that use subsidies and price controls to keep basic consumer goods cheap have an incentive to erect borders to limit the outflow of their cheap goods via exporting or non-residential consumption. And the presence of price restrictions and border controls is likely to result in an economic environment that is not conducive to entrepreneurial activity. Unlike our previous findings, the Red Belt is not closely associated with market integration over the longer period of 1994:II-1998:II considered in this study. In particular, the propensity for Red Belt regions to be deemed integrated according to our measure in a given period (48 percent) is statistically insignificantly different from the propensity for non-Red-Belt regions to be integrated (50 percent). One reason for this is that regional preferences and policies have evolved over during the course of Russia’s reform. For example, some regions that had rapidly liberalized prices in the early 1990s subsequently introduced price controls in 1995-1996; other regions followed exactly 7
A potential problem with this method of measuring integration is that a region may fail to appear integrated not because it chose to be this way, but because a sufficiently large proportion of its neighbors did. To explore this possibility we considered an alternative procedure that involves two steps. The first step is exactly as explained above. In the second step, we re-estimated (1) for each region i by measuring its price volatility vis-à-vis only those regions deemed to be integrated in the first step. Use of this alternative procedure had a minor impact on the shares reported here.
5
the opposite pattern (Berkowitz, DeJong, and Husted, 1998). Likewise, some regions that initially supported anti-reformist politicians subsequently voted them out of office and replaced them with market-oriented politicians (McFaul, 1997). In short, the adoption of reforms in Russia has evolved over time, and this evolution has varied across regions. This evolution, and its many regional variations, may be important for understanding the evolution of market integration. We consider the interaction of our measure of market integration with four classes of variables: international trade, internal transport costs, public discontent, and standards of living. All variables are monthly, and are measured at the aggregate level. Regarding international trade, Young (1996, 1999) has argued that high trade shares may signal weak internal trade relationships. As noted previously, Russia’s internal market was highly disconnected during the latter half of 1996 and all throughout 1997, a period in which Russia exported heavily on world markets. This was also a period during which regions used discretionary tax and subsidy polices to attract large domestic and foreign firms (Shleifer and Treisman, 1999), and when many regions sought to develop independent trade relations on world markets. To quantify trade activity, we use trade shares: imports plus exports divided by GDP (source: SITE, 1999). Regarding internal transport costs, these directly impinge on the ability of arbitrageurs to profit from regional price differences, and thus weaken regional price linkages. We measure transport costs using a freight-transport price index. This was converted to a real index using aggregate CPI data (source: SITE, 1999). Regarding public discontent, there is evidence that regions dissatisfied with federal economic policies or with their political status within the federation have taken measures to isolate themselves from the internal market, including measures designed to limit trade flows across their borders (Mitchneck, 1995). We consider two measures of public discontent: strike activity, and over-due wage payments (or wage arrears), measured in real terms (source: SITE, 1999). 8 Finally, many theoretical models predict linkages between regional standards of living and market integration. For example, Rivera-Batiz and Romer (1991) argued that improvements in integration should cause standards of living to increase: increased flows of goods, ideas and people across economically integrated regions should foster economic growth. In contrast, in a theoretical study of center-periphery relations in transition economies, Berkowitz (1997) derived conditions under which increased incomes in peripheral regions can weaken internal market linkages by 8
Beyond fuelling public discontent, another reason wage arrears may coincide with breakdowns in integration is that they depress liquidity, which in turn raises the importance of barter trade, and increases the difficulty of conducting inter-regional trade. We thank Sergei Perov for raising this point.
6
encouraging secessionism. To quantify standards of living, we use the share of the population living above the poverty line (source: SITE, 1999). Figure 2 presents time-series plots of our variables.9 Transport costs initially rise sharply, peak in 1995:II, and then fall during the rest of the period. Living standards gradually rise throughout the period. Regarding the public discontent variables, wage arrears are generally increasing during the period, while strikes have no systematic temporal tendency. Finally, trade shares are relatively low during 1995, increase dramatically during 1996:III, and then maintain their higher level during the remainder of the period. In studying the relationship between these variables, timing is an issue: our measure of market integration at any given point in time is based on a twelve-month span of price behavior, thus linking our measure with a specific date is not straightforward. As noted, the dates we assign in reporting our measure of integration in Figure 1 and Table 2 correspond to the median month in the relevant twelve-month span. In linking this variable with the additional monthly variables, we consider several alternatives. First, we align this median date with the date at which the variables were observed. Thus for example, we define June 1997 as the median month for the measure of integration calculated for the period January 1997 – December 1997. This is then linked with June 1997 values of the remaining variables in evaluating the correspondence of these variables over time. Next, we link our measure of integration at each date with the twelve-month average of the additional monthly variables calculated over the same time span. In this case, there is complete overlap in the time spans over which the variables are measured. Finally, we link our measure at each date with the average value of each of the additional variables calculated over the first six months of the relevant time span; and then the last six months of the relevant time span.10 For each method of linking the variables, we characterize conditional correlations observed among our variables using the model log(integrationt) = γ 0 + γ 1log(trade sharet) + γ 2 log(internal transport costst) +
(2)
γ 3 log(wage arrearst) + γ 4 log(strikest) + γ 5 log(standard of livingt) + ut.
9
For ease of presentation, the variables are presented in log terms and normalized on a 0-1 scale In addition to six-month spans, we also calculated averages over three-month spans (for all four quarters of the relevant time span), and obtained results similar to those obtained using the six-month spans. 10
7
OLS estimates of (2) always yield residuals that are serially correlated, thus we assume an AR(1) specification for ut: ut = ρut-1 + εt,
(3)
and estimate the system (2) – (3) via maximum likelihood.11 ML estimates of this system obtained for each method of measuring right-hand-side variables are reported in Table 3. Since all variables are measured in logs, the coefficients reported in Table 3 represent elasticities. However, some care is needed in comparing these elasticities across methods of measuring right-hand-side variables, since a one-percent increase in a twelve-month average of a variable is presumably more important quantitatively than a one-percent increase in a single observation of the variable. This is clearly reflected in Table 3: the longer the time span used to calculate average values of right-hand-side variables, the larger in absolute value is the corresponding parameter estimate. When we use observations at median dates to measure right-hand variables, the variables have weak explanatory power for market integration: no variable is significant at the 10-percent level (although the coefficients on transport costs and wage arrears are nearly significant), and the R2 statistic is only seven percent in this case. 12 In contrast, when we use twelve-month averages to measure right-hand-side variables, the R2 statistic increases to 22 percent, and transport costs, wage arrears, and living standards are significant at the 5-percent level. As expected, transport costs and wage arrears are negatively associated with market integration, and their quantitative importance is striking: their elasticity measures are –20 and –4.8. Even more striking is the quantitative importance of living standards, which has an elasticity measure of 36. The positive sign in this case is consistent with the notion that living standards are enhanced by openness, although the remaining estimates do not support this direction of causality. To see this, note first that in measuring variables using average values calculated over the first six months of the relevant time span, the fit of the model we obtain is virtually identical to the twelve-month case (the R2 statistic is 23 percent in this case). Also, transport costs, wage arrears and living standards continue to be statistically significant, and although the absolute values of their coefficients are
11
Durbin-Watson statistics obtained using ML estimates of (2) – (3) consistently support the AR(1) specification in (3). 12 In each set of estimates in Table 3, the coefficients on trade shares and strikes are insignificant. Moreover, dropping these variables from (2) has a trivial impact on the remainder of our results.
8
lower in this case (for the reason noted above), their relative sizes are strikingly similar as compared with the twelve-month case: the coefficients on transport costs and wage arrears are both 2.7 times larger in the twelve-month case than in the six-month case, and the coefficient on wage arrears is 2.2 times larger. In contrast, when variables are measured using average values calculated over the last six months of the relevant time span, no variable is significant at even the 40-percent level, and the R2 statistic drops to six percent. The correspondence of market integration with the behavior of transport costs, wage arrears, and living standards measured over the first six months of the relevant time period, and the lack of correspondence of market integration with these variables measured over the last six months, suggests that these variables have influenced the behavior of market integration. Ideally, we would investigate this further by conducting tests for Granger causality in a vector autoregressive framework, but we have far too few time-series observations to do so. Instead, we re-estimate (2) – (3) using six-month averages of right-hand-side variables computed using a time span that begins three months prior to the beginning of the twelve-month time span used to calculate our measure of market integration. In other words, the right-hand-side variables used to produce the third set of estimates of (2) – (3) in Table 3 are simply lagged by three months, so there is a three-month time overlap between them and the measure of market integration. For completeness, we also re-estimate (2) – (3) using six-month averages of right-hand-side variables measured for six-month time spans that end three months following the twelve-month time span over which market integration was calculated. This amounts to leading by three months the set of right-hand-side variables used to produce the last set of estimates in Table 3. Each set of estimates is presented in Table 4. These estimates fail to provide additional support for a causal relationship between variables. In lagging right-hand-side variables, the R2 statistic drops to eight percent, and while each coefficient retains its original sign (compared with the third set of estimates in Table 3), the absolute value of the three previously significant coefficients drops considerably, and each becomes insignificant. And when we lead the right-hand side variables used to produce the fourth set of estimates in Table 3, we again fail to obtain a single statistically significant parameter estimate. In sum, we have found evidence of a statistically and quantitatively significant conditional correspondence between movements in our measure of market integration and our measures of transport costs, wage arrears, and standards of living. We failed to find similar evidence for our
9
measures of international trade and strikes. Also, we have found some evidence that movements in market integration come in response to movements in the additional variables, but this evidence is fragile: it does not hold up to a three-month change in the timing used to measure the additional variables.
4 Conclusions We have quantified temporal fluctuations in the extent of regional-market interconnectedness in Russia. Based on the behavior of inter-regional differences in commodity prices observed across 47 Russian regions, we found substantial movements in interconnectedness. Specifically, a widespread pattern of integration observed between 1994:I and 1996:I subsequently gave way to a period of disconnectedness that persisted until mid-1997. This in turn was displaced by a resurgence in interconnectedness that has persisted through the end of 1998. We also analyzed interactions observed between our measure of internal market integration and a set of related variables measured at the aggregate level. We found significant negative relationships between internal integration and measures of internal transport costs and public discontent; a strong positive relationship between integration and standards of living; and a weak relationship between integration and international trade. In future work, we plan to analyze how regional measures of economic performance, public discontent, etc. have interacted with the aggregate data considered here in influencing internal market integration. The goal is to better understand how a given region’s ties to the aggregate economy are related to its own economic conditions as well as those prevailing at the aggregate level. For example, are regions that are relatively prosperous vis-à-vis the economy as a whole more or less likely to have close ties with the aggregate economy? If so, is there a clear pattern of causality behind this relationship? We are working to compile a regional data set that will enable us to make progress in addressing questions such as these.
10
References Berkowitz, D. (1997). “Regional Income and Secession.” Regional Science and Urban Economics 27: 17-45. Berkowitz, D., DeJong, D.N., and S. Husted (1998). “Quantifying Price Liberalization in Russia.” Journal of Comparative Economics 26: 735-760. Berkowitz, D. and D. DeJong (1999). “Russia’s Internal Border.” Regional Science and Urban Economics 29: 633-649. Berkowitz, D. and D. DeJong (2000). “Russia’s Internal Incoherence.” In Press, von Braun, J, Frohlberg, K. Serova, E. and P. Wehrheim, P., editors, Russia's Agro-food Economy: Towards Truly Functioning Markets. Dordrecht: Kluwer Academic Publishers. DeMasi, P. and V. Koen, (1996). “Relative Price Convergence in Russia.” IMF Staff Papers 43: 97-122. Engel, C. and J.H. Rogers (1996). “How Wide is the Border?” American Economic Review 86(5): 1112-1125. Freinkman, L. and D. Treisman and S. Titov (1999). “Subnational Budgeting in Russia.” World Bank Technical Paper No. 452, Washington, D.C. Gardner, B. and K. Brooks (1994). “Food Prices and Market Integration in Russia: 1992-93.” American Journal of Agricultural Economics 76 (3): 641-646. Goodwin, B.K., Grennes, T.J., and C. McCurdy (1999). “Spatial Price Dynamics and Integration in Russian Food Markets.” In Press: Journal of Policy Reform. Goskomstat Rosiyi (1995-1999), Sotsial’noye Ekonomicheskoye Polozheniye Rossiyi. Goskomstat Rossiyi: Moscow, various issues. Higgins, A. (1998). “Odd Borders Appear in Russia as Regions Face Poor Harvest.” Wall Street Journal October 16th, p. 1. Koen, V. and S. Phillips (1993), “Price Liberalization in Russia: The Early Record.” International Monetary Fund Occasional Paper #104, June. Krugman, P. (1991). “Increasing Returns and Economic Geography.” Journal of Political Economy 99(3): 483-499. McFaul, M. (1997). “Russian Regional Elections: Presidential Primaries?” Analysis of Current Events 9(9): 1, 3-4. Mitchneck, B. (1995), “An Assessment of the Growing Local Economic Development Function of Local Authorities in Russia.” Economic Geography 71(2): 150-170.
11
Naughton, B. (1999). “How Much Can Regional Integration Do to Unify China’s Markets?” University of California, San Diego working paper. Parsley, D. and S-J Wei (1996). “Convergence to the Law of One Price Without Trade Barriers or Currency Fluctuations.” Quarterly Journal of Economics 111: 1211-1236. Rivera-Batiz, L.A., and P.M. Romer (1991). “Economic Integration, and Endogenous Growth.” Quarterly Journal of Economics 106: 458-70. Rogoff, K. (1996). “The Purchasing Power Puzzle.” Journal of Economic Literature 34: 647-68. Serova, E. (1998). “Federal Agro-Food Policy in the Conditions of the Financial and Economic Crisis.” Russian Economy: Trends and Perspectives, November. Shleifer, A. and D. Treisman (1999). “Without a Map: Political Tactics and Economic Reform in Russia.” Harvard University and UCLA working paper. SITE (Stockholm Institute for Transition Economies) (1999). Russian Economic Trends, monthly update database, available online at http://www.hhs.se/site/ret/exceldb/default.htm. White, H. (1980). “A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroscedasticity.” Econometrica 48: 817-838. Young. A. (1996). “The Razor’s Edge: Distortions, Incremental Reform and the Theory of the Second Best in the People’s Republic of China.” Boston University working paper. Young, A. (1999). “The Razor’s Edge: Distortions and Incremental Reform in the People’s Republic of China.” Fifth Nobel Symposium in Economics: The Economics of Transition, Stockholm.
12
Table 1: Sample Sub-periods and Dates Sub-period
Median Date
Sub-period
Median Date
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Sept-94 Oct-94 Nov-94 Dec-94 Jan-95 Feb-95 Mar-95 Apr-95 May-95 Jun-95 Jul-95 Aug-95 Sept-95 Oct-95 Nov-95 Dec-95 Jan-96 Feb-96 Mar-96 Apr-96 May-96 Jun-96 Jul-96
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
Aug-96 Sep-96 Oct-96 Nov-96 Dec-96 Jan-97 Feb-97 Mar-97 Apr-97 May-97 Jun-97 Jul-97 Aug-97 Sept-97 Oct-97 Nov-97 Dec-97 Jan-98 Feb-98 Mar-98 Apr-98 May-98
13
Table 2: Integration Dynamics Median date
Number of
Share Median date
Integrated Regions
Number of
Share
Integrated Regions
Sep-94
27
0.574
Aug-96
15
0.319
Oct-94
28
0.596
Sep-96
16
0.340
Nov-94
29
0.617
Oct-96
16
0.340
Dec-94
33
0.702
Nov-96
16
0.340
Jan-95
35
0.745
Dec-96
7
0.149
Feb-95
34
0.723
Jan-97
6
0.128
Mar-95
36
0.766
Feb-97
4
0.085
Apr-95
32
0.681
Mar-97
4
0.085
May-95
24
0.511
Apr-97
5
0.106
Jun-95
16
0.340
May-97
12
0.255
Jul-95
28
0.596
Jun-97
17
0.362
Aug-95
27
0.574
Jul-97
18
0.383
Sep-95
26
0.553
Aug-97
18
0.383
Oct-95
27
0.574
Sep-97
20
0.426
Nov-95
28
0.596
Oct-97
18
0.383
Dec-95
29
0.617
Nov-97
9
0.191
Jan-96
29
0.617
Dec-97
32
0.681
Feb-96
25
0.532
Jan-98
32
0.681
Mar-96
26
0.553
Feb-98
37
0.787
Apr-96
20
0.426
Mar-98
40
0.851
May-96
19
0.404
Apr-98
39
0.830
Jun-96
15
0.319
May-98
42
0.894
Jul-96
17
0.362
14
Table 3: Regression Results Timing Method Contemporaneous R2 : 0.070
rho: 0.803
12-month average R2 : 0.224
Variables
Coefficient
S.E.
p-value
Coefficient
Trade share
-0.314
0.578
0.589
Transport costs
-2.669
1.826
Wage arrears
-0.722
Strikes Living Standards
rho: 0.707
S.E.
p-value
2.978
4.447
0.507
0.153
-20.006
8.051
0.018
0.442
0.111
-4.784
1.907
0.017
-0.001
0.019
0.938
0.021
0.144
0.847
2.173
2.985
0.472
35.957
17.878 0.052
Timing Method Average over first six months
Average over second six months
R2 : 0.231
rho: 0.697
R2 : 0.061
rho: 0.790
Coefficient S.E.
p-value
Variables
Coefficient
S.E.
p-value
Trade share
2.400
1.867
0.207
-0.715
0.995
0.477
Transport costs
-7.366
2.518
0.006
-1.263
3.729
0.737
Wage arrears
-2.215
0.706
0.003
-0.299
0.961
0.758
Strikes
-0.096
0.073
0.199
-0.005
0.081
0.944
Living Standards
13.423
5.094
0.012
-5.343
8.061
0.512
15
Table 4: Regression Results Timing Method 3-month lag of average over first six months R2 : 0.085 rho: 0.830
3-month lead of average over second six months R2 : 0.217 rho: 0.751
Variables
Coefficient
S.E.
p-value
Coefficient S.E.
Trade share
1.578
2.306
0.498
-0.874
2.112
0.682
Transport costs
-1.690
1.969
0.396
0.334
3.119
0.915
Wage arrears
-0.824
1.065
0.444
-0.520
0.908
0.570
Strikes
-0.180
0.134
0.189
0.266
0.188
0.166
Living Standards
1.729
3.633
0.637
-7.540
6.451
0.250
16
p-value
