Java HotSpot Virtual Machine

We are using the system property, file.encoding, to set the default character encoding for the JVM. On an English NT OS we start the Java process using -Dfile.encoding=UTF-8. This seems to work fine and sets the default encoding. But on a Japanese NT OS, it does not pick up the encoding setting. The following is a test java program I am using......

import java.io.PrintStream;
import sun.io.*;
import java.io.*;

public class TestEncoding {

public static void main(String argv[]) {

String encProperty = System.getProperty("file.encoding");
String byteToCharClass =
sun.io.ByteToCharConverter.getDefault().getClass().getName();
String charToByteClass =
sun.io.CharToByteConverter.getDefault().getClass().getName();

System.out.println("encProperty == '" + encProperty + "'");
System.out.println("Default ByteToChar Class == '" + byteToCharClass + "'");
System.out.println("Default CharToByte Class == '" + charToByteClass + "'");

ByteArrayOutputStream buf = new ByteArrayOutputStream(10);
OutputStreamWriter writer = new OutputStreamWriter(buf);
System.out.println("OutputStreamWriter encoding = " + writer.getEncoding());
}
}

When I run this from the English NT OS using I get the following results:
encProperty == 'UTF-8'
Default ByteToChar Class == 'sun.io.ByteToCharUTF8'
Default CharToByte Class == 'sun.io.CharToByteUTF8'
OutputStreamWriter encoding = UTF8

BUT when I run it from a Japanese NT OS I get:
encProperty == 'MS932'
Default ByteToChar Class == 'sun.io.ByteToCharMS932'
Default CharToByte Class == 'sun.io.CharToByteMS932'
OutputStreamWriter encoding = MS932

Environment:
English NT 4.0 JDK1.2.2
Japanese NT 4.0 JDK1.2.2 and/or JDK1.3.1_01

Can anyone tell me why the results are different for the Japanese OS? Does anyone know of a workaround?