Real Wages and Output Growth
Having established that in pre-industrial Japan wage growth was not associated with widening wage differentials, we now ask how changes in output were related to trends in real wages.
Only two attempts have so far been made to estimate overall growth of output in TokugawaJapan.
One is for the farm sector only, and was derived from Tokugawa data on taxed land. Under the Tokugawa system every piece of taxable land was assessed in terms of its productive capacity and expressed in koku of rice equivalent (called kokudaka). Thus the kokudaka was a measure of farm output covering all agricultural products, and the national aggregates exist for 1598, 1645, 1697, 1721, 1830, and 1872. But it is believed that from the 1697 compilation onwards the gap between the officially assessed and the actual production capacity tended to widen, due to productivity growth that undoubtedly took place from the mid-seventeenth century. Satoru Nakamura compared the 1872 kokudaka figure with total farm output from early Meiji government statistics. By assuming that the productive capacity of the land was enhanced by land-improvement investments made by samurai administrations, he allocated the difference between these output measures over the entire period in question according to the period-by-period numbers of land-improvement projects recorded during the Tokugawa era (Nakamura 1968: 168—74).4 The estimates thus derived for two benchmarks of 1700 and 1872 are set out in the second column of Table 3.3. In view of a notion widely held in the Tokugawa period that 1 koku of rice was enough to feed one person for one year, it is worth noting that the availability of farm products in the benchmark year of 1700 was above the level of 150 kg (1 koku)∕person and increased over time. Judging from several case studies of agricultural land productivity in the Tokugawa period, it is not inconsistent with a similarly gradual rise in the level of rice yields (Yasuba 1987: 299; Yagi 1990).On the basis of these Nakamura estimates, Angus Maddison (2001) recently put forward hypothetical GDP figures by topping up non-farm products which were simply assumed to have grown somewhat faster than farm output (column 3 of Table 3.3, expressed in 1990 international dollars: Maddison 2001: 254—8, 264). Given the fact that most industrial crops, cocoons, and even raw silk were covered by the early Meiji ‘farm’ statistics and hence included in Nakamura's calculation,
Table 3.3 Comparisons between wage and output growth, 1700—1870
| Growth of per capita output | |||
| (1700=100) | Wage growth Farm sector (kg) | GDP (1990 dollars) | |
| 1700 | 100 | 169 | 570 |
| 1870 | 118 | 201- | 737 |
| 1700—1870 (% per annum) | 0.10 | 0.10 | 0.15 |
‘ Figure is for year 1872.
Notes: Wage growth is calculated from Table 3.1, on the assumption that from 1700 to 1820 real wages increased at 0.6% annually, the same pace as for the weighted average of Kami-Kawarabayashi's carpenters and agricultural labourers (weights are taken from the Choshi case, that is, 0.14 for the skilled and 0.84 for the unskilled), then from 1820 onwards at the same pace as for the Choshi series of soy-sauce makers, that is, at -1.1% until 1870.
Sources: Table 3.1 and Maddison (2001: 255, 264).
it is debatable if growth in the non-farm sector was appreciably higher than that of the farm sector. It is true, however, that textiles expanded substantially. According to a recent survey of non-agricultural activities in Tokugawa times (Odaka 2001), for example, the output of cotton cloth in three rural districts exhibited a fourfold increase from the late eighteenth century to 1840, stagnated for three decades, then started growing again from the 1870s onwards.
But in other sectors growth was far less spectacular. In the brewing of sake, miso paste, and soy sauce, undoubtedly the largest single manufacturing sector of the day, output growth seems to have been modest. It is of course extremely difficult to quantify how modest it was, but judging from recent studies in other research areas, such as works on consumption and the world of commodities (Hanley 1997; Koizumi 1999), the overall impression is that industrial output must have increased marginally faster than farm output. In that case, the degree of overestimation by Maddison was not significant.It is not easy to give comparable evidence concerning wage growth because there is no single real wage series running through the entire period. Given the nature of the quantitative information set out in Table 3.1, my judgement is that the linked series of Kami-Kawarabayashi and Choshi real wages probably represent the general pattern of change better than other combinations. Since the Choshi series is a composite of skilled and less skilled workers' real wages, the Kami-Kawarabayashi series should also be a composite of carpenters' and agricultural day labourers'. Using the Choshi weights (0.86 for the skilled and 0.14 for the less skilled) implies that real wages in Kami-Kawarabayashi rose at the average annual rate of 0.6% from the first benchmark of 1700 to 1820, then declined at 1.1% in real terms until 1870—a decline like that experienced by Choshi's workers from 1818 to 1867. It is likely that rural wages in other regions rose at a slower rate than the Kami-Kawarabayashi wages in the eighteenth century and peaked much later in the early nineteenth century (Saito 1998: 31, 45—6). In this respect, the Kami-Kawarabayashi data probably overstates the degree of wage increase in the eighteenth century. However, the decline exhibited by the Choshi series after 1818 may also have overstated the degree of decline. The errors offset each other, so it is probably safe to say that this calculation is not off the mark when the start is compared with the end point.
The results of this exercise are shown in column (1) of the table.This table tells us that the growth rate of real wages during Japan's age of rural development was between 0.1% and 0.15% per annum. If the end point were set at 1890 or 1900, the growth rates would become slightly higher. Whatever the end point, growth of that magnitude was neither very impressive nor was it a miserable achievement. More important, however, is that real wages rose at exactly the same pace as farm output grew. This cannot be a coincidence. As we have already seen, rice yields exhibited a tendency to increase. There were regional differences too. The rise in the productivity of agriculture in the Kinai started early and appeared to taper off, or even to dip, sometime in the early nineteenth century, whereas that in the eastern provinces started late but continued until about 1850, which is a pattern almost identical to the one for real wages as suggested above. That said, however, the secular trend was unmistakably upward, and it is demonstrated that the
long-term movement of rents in kind per unit of land was very much similar to that of rice yields. In other words, the proportion of rents to yields remained stable at the level of 60—70% during the latter half of the Tokugawa period (Tomobe 1996). The samurai government's tax rate on land was high but remained static in relation to the official kokudaka assessment (Smith 1968), which meant that after-tax rent earnings of landlords increased slightly. The wage—rent ratio, by implication, would have increased in the first phase and declined in the second, but it is probable that the ratio showed a marginally downward trend over the long-run.
The tempo of the wage increase was not so different from that of per capita GDP growth, either.5 Clearly there was thus no divergence between wage growth and output growth. As mentioned earlier, we may not be able to equate this evidence with that on the income gap between the peasantry near the bottom and the population group near the middle of the income distribution. But given the unambiguous evidence in Table 3.3, it is probably safe to suggest that income inequality did not increase during the period concerned. This is a conclusion which is consistent with what we have seen in Section 2.
4.