How Wide are European Borders? New Evidence on the Integration Effects of Monetary Unions Guenter W. Beck Axel A. Weber Goethe University Frankfurt Goethe University Frankfurt, CFS ➀ Motivation and Background ➁ ➂ ➃ ➄ Data Cross-Sectional Evidence for EMU and GEMU Cross-Sectional/Time-Series Evidence for GEMU Summary and Conclusions 56th European Meeting of the Econometric Society, August 25-29, 2001 1 M OTIVATION /B ACKGROUND Empirical literature on goods market integration based on: ➜ trade data see e.g. McCallum (AER, 1995) ➜ price data see below Theoretical background: ➜ PPP/LOOP Two major strands of literature: ➜ Compare intra-country and inter-country price dispersion Engel and Rogers (AER (1996), JIE (2000)), ... ➜ Determine half-lives of devations from the PPP/LOOP Wu (JMCB, (1996), Frankel and Rose (JIE (1996)), ... M OTIVATION /B ACKGROUND 2 M OTIVATION /B ACKGROUND Objectives of our paper: ➜ quantify the effects of EMU on intra-European price dispersion use primarily new data determine inter-national price dispersion without exchange rate noise ➜ determine the evolution of inter-German price volatility over time ➜ quantify the speed of price convergence between East and West German regions M OTIVATION /B ACKGROUND 3 DATA "EMU" data: ➜ CPI data: All items and 7 major groups (not available for all countries) ➜ 81 locations in 6 European countries 12 German locations, 20 Austrian locations, 4 Swiss locations, 20 Italian locations, Spanish locations, 7 Portugese locations ➜ Data period: January 1995 - June 2000 (updated to June 2001) "GEMU" data: ➜ CPI data: All items and 7 major groups ➜ 33 locations in Germany, Austria and Switzerland 5 West-German locations, 4 East-German locations, 20 Austrian locations, 4 Swiss locations ➜ Data period: January 1995 - June 2000 (all items and major groups), all items: January 1991 - June 2000 (updated to June 2001). DATA 4 DATA DATA 5 EMU C ROSS -S ECTIONAL E VIDENCE Objective: ➜ analyze determinants of intra- vs. international regional price dispersion Alternative Measures of Price Dispersion ➜ standard deviation of monthly changes in relative prices (base specification) ➜ standard deviation of 1-month-ahead in-sample forecast error ➜ spread between 10th and 90th percentile Some characteristics of the data ➜ see tabe1 and figure 1 EMU C ROSS -S ECTIONAL E VIDENCE 6 EMU C ROSS -S ECTIONAL E VIDENCE Some data properties Table 1: Descriptive Statistics, Relative Price Volatility and Distance, Measure 1 Index All Items, Mean Stdv. Food, Mean Stdv. Clothing, Mean Stdv. Furniture, Mean Stdv. Fuel, Mean Stdv. Health, Mean Stdv. Transport, Mean Stdv. Education, Mean Stdv. Exrate Var. Distance Relative Locations, Indicated by Country Names ge-ge ge-au ge-ch ge-it ge-sp ge-po au-au au-ch au-it au-sp au-po ch-ch ch-it ch-sp ch-po (6) (7) (3) (4) (5) (8) (9) (10) (16) (17) (18) (19) (20) (21) 1,66 0,37 4,13 0,38 9,92 0,13 7,35 0,31 8,74 0,18 9,97 0,35 8,78 0,41 9,64 11,77 1,22 13,21 13,66 16,46 2,01 0,32 0,44 0,12 0,26 0,21 0,29 0,34 6,44 0,30 7,56 0,81 1,92 0,59 8,80 0,55 4,17 1,59 2,78 0,38 6,46 11,57 8,65 10,87 12,54 6,08 11,59 9,74 11,54 13,15 4,88 14,37 15,67 17,75 3,75 0,81 0,31 0,39 0,81 1,16 1,02 0,64 0,67 0,80 1,05 0,67 0,57 0,78 0,94 0,65 - 8,60 1,81 9,68 1,50 30,84 25,29 10,98 12,69 - 1,64 0,60 - - 7,89 1,03 9,04 0,61 11,59 0,66 - - - - 4,63 1,70 - 1,01 1,50 1,01 1,50 9,48 4,92 1,18 - - - - 3.61 0.75 6.39 11.36 8.82 10.00 13.65 4.29 11.25 8.26 0.58 0.28 0.62 0.39 0.94 0.87 0.30 0.60 3.08 20.55 0.66 1.18 - 10,68 9,71 10,23 12,80 0,76 1,41 1,50 1,01 1,50 - - 20,80 20,63 22,09 6,35 5,67 5,41 - 15.60 16.75 18.47 4.78 1.63 0.73 1.54 4.43 - 0.49 9.08 18.68 8.44 176 309 289 519 10.89 1.010 1.409 EMU C ROSS -S ECTIONAL E VIDENCE 103 - 20,44 20,71 36,57 13,21 11,93 11,57 (15) 8,94 10,23 5,18 11,06 7,99 0,77 1,90 1,75 1,48 3,11 - - - 3,87 1,27 6,77 32,52 2,35 29,77 27,26 0,72 12,88 0,99 11,34 13,67 - - - - 4,05 1,83 7,13 1,06 2,66 15,07 14,17 17,02 6,59 0,60 1,14 1,15 0,66 1,76 - 5,87 1,48 9,60 10,49 5,14 10,86 6,32 1,65 1,57 2,61 1,49 2,68 4,29 1,01 9.59 11.85 2.03 12.85 14.36 17.15 3.54 0.49 1.17 0.45 0.48 0.33 0.74 1.30 7.60 0.81 9.09 1.88 2.17 10.16 6.19 1.12 1.12 3.19 - 1.000 1.438 75 - - 9,04 0,92 9,34 1,49 - - 3,25 1,14 7,75 1,11 8.98 18.67 8.42 10.88 - 8,38 1,14 4,62 2,02 - 317 (14) - 23.08 26.50 26.71 2.24 1.16 2.77 271 (13) it-po sp-sp sp-po po-po (2) 4,68 17,87 3,05 12,65 (12) it-sp (1) 11,32 3,31 0,43 0,74 (11) it-it - 22.86 13.56 16.27 308 729 1.150 6.50 3.22 230 9,83 1,15 6,34 1,48 8.50 10.19 4.84 11.15 9.65 1.84 2.07 1.65 1.46 2.05 12.72 10.91 847 1.294 - 7.32 - 321 477 348 7 EMU C ROSS -S ECTIONAL E VIDENCE Figure 1: The Variances of Average Relative Price Changes and Nominal Exchange Rate Changes, Overall Period (1995:1-2000:6) it-ch 24 22 20 18 16 14 12 10 8 6 4 2 0 ge-it au-it it-sp it-po de-at Intra-national 0 2 4 6 8 10 12 14 16 18 20 22 24 Key to Figure: The variance of nominal exchange rate changes is plotted on the vertical axis, and the variance of average relative price changes (across locations) are on the horizontal axis. EMU C ROSS -S ECTIONAL E VIDENCE 8 EMU C ROSS -S ECTIONAL E VIDENCE Estimation: Basic Equation: ➜ ➜ also included: city dummies and monthly dummies Total Period: 1995.01 - 2000.06 ➜ see table 2b and 2a Subperiods: 1995.01 - 1998.12 and 1999.01 - 2000.06 ➜ see figures 3, 4 and 5 and table 3 Conclusions: ➜ intra-national volatility is initially low and does not decline significantly ➜ intern-national volatility is initially high and falls drastically for all EMU cross-border city pairs ➜ however: even in EMU distance and the border continue to matter! EMU C ROSS -S ECTIONAL E VIDENCE 9 EMU C ROSS -S ECTIONAL E VIDENCE Is there a border effects? YES! Table 2b: Estimation Using Log Distance, Border and Exchange Rate Volatility, European City Pairs, Overall Period (Jan 1995-June 2000), Measure 1 Variable Coeff. Log(Distance) Border Nom. Exrate Volatility 1,33 301,8 R2 R 2 (adj.) 0.871 0.868 1.199 SEE(*1000) Stde. 0,04 4,99 t-Stat. 29,3 60,5 Coeff. 1,04 1,44 249 Stde. 0,05 0,10 5 t-Stat. 22,5 14,5 45,8 0.880 0.877 1.158 Notes: All regressions contain as explanatory variables a dummy for each of the included individual in addition to the variables listed in the cells. Heteroscedasticity-consistent standard errors (White, 1980) are reported. Coefficients and standard errors on log distance and border are multiplied by 103 . The dependent variable is the standard deviation of the one-month difference in relative prices. Standard deviations are computed over the sample period from January 1995 to June 2000 (Exceptions: Italian Data are available from January 1996 only, Swiss Data end in May 2000). There are 3240 observations in the overall CPI regression. EMU C ROSS -S ECTIONAL E VIDENCE 10 EMU C ROSS -S ECTIONAL E VIDENCE How many borders matter? ALL! Table 2a: All Items Estimation Using Log Distance Function, Overall Period (Jan 1995-June 2000), Volatility Measures 1, 2, 3 and 4 Variable Constant Log(Distance) Distance Distance sqr. Border ge-au Border ge-ch Border ge-it Border ge-sp Border ge-por Border au-ch Border au-it Border au-sp Border au-po Border ch-it Border ch-sp Border ch-po Border it-sp Border it-po Border sp-po R2 R2 (adj.) SEE(*1000) Coeff. 0,065 1,587 8,417 5,452 6,858 8,292 7,637 6,074 6,907 7,902 11,540 11,987 13,639 4,402 4,367 5,730 0.9915 0.9912 0.3087 Stde. 0,022 0,035 0,038 0,032 0,046 0,118 0,060 0,030 0,047 0,121 0,036 0,053 0,135 0,036 0,115 0,110 EMU C ROSS -S ECTIONAL E VIDENCE Measure 1 t-stat. Coeff. Stde. t-stat. 3,03 44,73 220,10 169,10 148,75 70,25 127,32 204,63 146,78 65,54 319,07 225,01 101,04 122,12 37,95 52,12 0,421 -0,400 1,604 8,438 5,488 7,044 8,718 7,662 6,092 7,096 8,364 11,565 12,082 13,913 4,522 4,727 5,765 0.9920 0.9918 0.2989 0,000 0,000 0,034 0,036 0,032 0,052 0,112 0,060 0,029 0,053 0,116 0,035 0,051 0,125 0,039 0,107 0,091 5,11 -9,95 47,03 232,23 173,24 136,10 77,62 128,31 209,99 134,68 72,32 330,80 238,02 111,75 116,58 44,30 63,67 Measure 2 Coeff. Stde. Measure 3 Coeff. Stde. 0,154 0,056 0,071 0,022 4,031 20,755 11,585 9,353 8,787 17,569 11,179 8,786 8,677 28,856 23,547 23,735 9,739 10,466 6,773 0.9655 0.9644 0.1192 0,088 0,167 0,104 0,124 0,297 0,215 0,095 0,133 0,303 0,143 0,172 0,388 0,112 0,285 0,271 1,436 8,011 5,323 5,071 6,831 7,334 6,145 4,941 6,338 11,878 9,071 11,256 4,483 4,122 4,814 0.9854 0.9849 0.3532 0,035 0,051 0,031 0,047 0,129 0,070 0,031 0,049 0,132 0,052 0,062 0,151 0,038 0,128 0,120 Measure 4 Coeff. Stde. 0,03 0,11 1,58 0,008 5,415 6,836 8,353 7,632 6,068 6,885 7,965 11,509 11,959 13,698 4,375 4,403 5,876 0,04 0,000 0,042 0,058 0,113 0,061 0,032 0,054 0,113 0,042 0,065 0,132 0,045 0,108 0,094 11 EMU C ROSS -S ECTIONAL E VIDENCE EMU has reduced volatility! Figure 3: Volatility of Relative Price Changes in the ERM (1995:1-1998:12) and EMU (1999:1-2000:6) Periods, this graph: look at 15 country pairs 20 18 16 14 12 EU-Switzerland 10 8 Intra-national 6 4 2 0 0 5 10 15 20 Key to Figure: EMU Sub-Period (1999:1-2000:6) on the vertical axis, pre-EMU sub-period on the horizontal axis. next graph: look at 3240 city pairs EMU C ROSS -S ECTIONAL E VIDENCE 12 EMU C ROSS -S ECTIONAL E VIDENCE Figure 4: (a) Cross-city Relative Price Volatility, Pre-EMU, Jan. 1995- Dec. 1998 Switzerland vs. Portugal Switzerland vs. Spain/Italy Within Country EMU C ROSS -S ECTIONAL E VIDENCE 13 EMU C ROSS -S ECTIONAL E VIDENCE (b) Cross-city Relative Price Volatility, EMU, Jan. 1999- June 2000 moving closer! but have borders become irrelevant? Switzerland vs. EU Within EMU EMU C ROSS -S ECTIONAL E VIDENCE 14 EMU C ROSS -S ECTIONAL E VIDENCE Estimating the border effects of EMU Table 3: Estimation Using Log Distance Function, Overall Period (Jan 1995-July 2000) and EMS/ERM-Subperiods (Jan. 1995-Dec. 1998, Jan. 1999-July 2000), Volatility Measures 1 and 2 Variable Log(Distance) Distance sqr. Border ge-au Border ge-ch Border ge-it Border ge-sp Border ge-por Border au-ch Border au-it Border au-sp Border au-po Border ch-it Border ch-sp Border ch-po Border it-sp Border it-po Border sp-po R2 R2 (adj.) SEE(*1000) Overall Period Coeff. Stder. Specification 1 Subperiod 1 Coeff. Stder. Subperiod 2 Coeff. Stder. 0,065 0,022 0,084 0,025 0,029 0,020 1,587 8,417 5,452 6,858 8,292 7,637 6,074 6,907 7,902 11,540 11,987 13,639 4,402 4,367 5,730 0.9915 0.9912 0.3087 0,035 0,038 0,032 0,046 0,118 0,060 0,030 0,047 0,121 0,036 0,053 0,135 0,036 0,115 0,110 1,744 9,128 6,840 8,173 9,972 8,441 7,710 8,224 9,540 13,468 13,607 16,040 5,353 5,504 6,872 0.9913 0.9911 0.3671 0,041 0,045 0,038 0,054 0,123 0,065 0,036 0,055 0,126 0,043 0,062 0,154 0,043 0,118 0,112 1,210 6,391 0,909 1,123 2,000 5,074 0,944 1,682 1,897 5,576 5,751 3,989 1,043 1,338 1,367 0.9918 0.9915 0.3573 0,041 0,074 0,036 0,042 0,139 0,100 0,031 0,042 0,141 0,077 0,076 0,252 0,033 0,137 0,133 EMU C ROSS -S ECTIONAL E VIDENCE Overall Period Coeff. Stder. 0,421 -0,400 1,604 8,438 5,488 7,044 8,718 7,662 6,092 7,096 8,364 11,565 12,082 13,913 4,522 4,727 5,765 0.9920 0.9918 0.2989 0,000 0,000 0,034 0,036 0,032 0,052 0,112 0,060 0,029 0,053 0,116 0,035 0,051 0,125 0,039 0,107 0,091 Specification 2 Subperiod 1 Coeff. Stder. 0,515 -0,400 1,765 9,154 6,878 8,369 10,426 8,474 7,732 8,423 10,037 13,498 13,709 16,335 5,479 5,889 6,909 0.9501 0.9485 0.3674 0,000 0,000 0,040 0,045 0,037 0,060 0,119 0,065 0,035 0,061 0,124 0,044 0,061 0,146 0,045 0,112 0,093 Subperiod 2 Coeff. Stder. 0,135 -0,100 1,221 6,404 0,925 1,184 2,129 5,090 0,955 1,745 2,038 5,590 5,788 4,078 1,083 1,448 1,379 0.9502 0.9487 0.3669 0,000 0,000 0,040 0,072 0,038 0,052 0,143 0,098 0,030 0,050 0,144 0,075 0,075 0,251 0,040 0,139 0,129 15 GEMU C ROSS -S ECTIONAL E VIDENCE Data ➜ as discussed above Procedure ➜ add a shadow-border dummy for East-Germany ➜ further procedure analogous to EMU analysis Results ➜ see table 5 GEMU C ROSS -S ECTIONAL E VIDENCE 16 GEMU C ROSS -S ECTIONAL E VIDENCE How Wide Are European Borders: Cross-Sectional Results Table 5: Estimation Using Log Distance Function, Overall Period (Jan 1991-June 2000) and EMS/ERM -Subperiods (Jan. 1991-Dec. 1994, Jan. 1995-Dec. 1998, Jan. 1999-June 2000), Vol. Measure 1 Measure 1 Variable Log(Distance) Border we-ea Border we-au Border we-ch Border ea-au Border ea-ch Border au-ch R2 R 2 (adj.) SEE(*1000) Overall Sample Subperiod 1 Subperiod 2 Subperiod 3 Coeff. Coeff. Coeff. Coeff. 0.096 11.567 2.139 8.477 11.412 15.308 8.276 0.9967 0.9965 0.0003 GEMU C ROSS -S ECTIONAL E VIDENCE Stde. 0.029 0.241 0.043 0.051 0.244 0.247 0.050 0.173 18.407 2.627 8.712 18.476 21.530 9.426 0.9977 0.9975 0.0004 Stde. 0.029 0.366 0.055 0.083 0.371 0.378 0.068 0.088 0.542 2.261 8.721 1.794 8.580 8.258 0.9837 0.9824 0.0004 Stde. 0.044 0.097 0.053 0.070 0.099 0.115 0.082 0.022 0.130 1.037 8.169 0.940 8.040 6.702 Stde. 0.035 0.115 0.065 0.074 0.098 0.107 0.105 0.9690 0.9665 0.0005 17 GEMU PANEL E VIDENCE Procedure ➜ panel unit root test based on Levin/Lin (1992) ➜ adjust for individual-specific and time-specific means ➜ adjust for serial correlation (ADF) Estimation equation ➜ Results ➜ see table 6 GEMU PANEL E VIDENCE 18 GEMU PANEL E VIDENCE How Wide Are European Borders: Panel Evidence Table 6: Time Series (ADF) and Panel (Levin & Lin) Estimates of AR(1) Coefficients from 528 City Pairs, Overall Sample (1991:1-2000:6) Location of City Pairs WE-WE WE-EA WE-AU WE-CH EA-EA EA-AU EA-CH AU-AU AU-CH CH-CH Mean AR(1) Coef. from ADF tests 0.964 0.923 0.866 0.930 0.901 0.913 0.895 0.896 0.895 0.948 Min. AR(1) Coef. from ADF tests 0.881 0.917 0.469 0.904 0.865 0.900 0.888 0.409 0.860 0.869 Max AR(1) Coef. from ADF tests 1.008 0.934 0.992 0.959 0.957 0.929 0.902 1.003 0.950 0.985 Mean Implied Halflife in Month 19 9 5 9 7 8 6 6 6 13 0.994 0.932 0.970 0.984 0.878 0.918 0.920 0.954 0.959 0.974 -0.743 -10.141 -8.410 -2.923 -8.972 -19.933 -10.653 -15.864 -10.319 -2.952 AR(1) Coef. from Levin & Lin tests t-Stat. from Levin & Lin tests Critical Value of t-Stat, (Based on Monte Carlo Simulation) Implied Halflife in Month Cross Section Dimension (N) Time Series Dimension (T) GEMU PANEL E VIDENCE 124 10 22 42 5 8 8 15 16 26 10 113 20 113 100 113 20 113 6 113 80 113 16 113 190 113 80 113 6 113 19 S UMMARY AND C ONCLUSIONS ➜ EMU has drastically reduced inter-European price dispersion (only due to elimination of exchange rate volatility?) ➜ however: national borders and distance continue to be important determinants of relative price volatility ➜ convergence to PPP/LOOP seems to be nonlinear S UMMARY AND C ONCLUSIONS 20
© Copyright 2026 Paperzz