Of families and inheritance: law and development in England before the Industrial Revolution

Abstract

We examine how pre-Industrial-Revolution English caselaw development on land, inheritance, and families affected, and was affected by, economic and demographic outcomes. Our yearly measures of caselaw development are derived from existing topic-model estimates that reflect a comprehensive corpus of reports on pre-1765 court cases. We estimate a structural VAR model using these caselaw time-series in combination with measures of real per-capita income and vital rates. Pre-industrial caselaw development profoundly shaped economic development. Strikingly, the areas of caselaw that stimulated real-income growth are on families and inheritance, not land. Caselaw on families and inheritance was especially important as a driver of real income and birth rates after 1710. Caselaw developments were spurred primarily by changes in real income, not by changes in vital rates. Incorporation of endogenous caselaw development leaves intact the findings of the existing literature that examines pre-industrial economic-demographic interactions. However, our findings do imply that any Malthusian trap that was present in pre-industrial England was made less severe as a result of developments in caselaw on families and inheritance.

Appendices

Appendix A

See Figs. 8 , 9 and 10

Fig. 8

The time series of the attention to topics comprising the Land theme

Fig. 9

The time series of the attention to topics comprising the Inheritance theme

Fig. 10

The time series of the attention to topics comprising the Families theme

Appendix B

See Fig. 11

Fig. 11

Accumulated responses of demographic and economic variables to law shocks. 90-percent confidence intervals computed using Kilian's (1998) bootstrap method

Appendix C

This appendix presents the results concerning the effect on economic development at the sectoral level of legal developments in the three areas of caselaw under consideration (land, inheritance, and families). To this end, we use Broadberry et al.'s (2015) data on real output in the agriculture, industry, and service sectors, expressed in the form of index numbers (the base year is 1700), and on England's population, expressed in millions. (The data in electronic format are available on the Bank of England's website under the heading A Millennium of Macroeconomic Data.)

For each of the three sectors (agriculture, industry, and service), we compute the real output index per one million people. Specifically, let ${q}_{t}$ denote the (unobserved) real output of a particular sector in year $t$. Then, the (observed) index of real output in year $t$ equals $({q}_{t}/{q}_{1700})\times 100$, and the index of real output per one million people in the same year is

$$\frac{\frac{{q}_{t}}{{q}_{1700}}\times 100}{{pop}_{t}}\equiv \left(\frac{{q}_{t}}{{pop}_{t}}\right)\times \left(\frac{100}{{q}_{1700}}\right),$$

(C1)

where ${pop}_{t}$ is year $t$ population measured in millions. The first term on the right-hand side of (C1) is the (unobserved) real sectoral output per one million people. The second term on the right-hand side of (C1) is time-invariant. Applying natural logarithms to (C1) and first-differencing the resulting expression with respect to time, $t$, then shows that relative changes over time in (C1) perfectly reflect the relative changes over time in the (unobserved) real sectoral output per capita.

We estimate an expanded version of the VAR developed in Sect. 5. Specifically, we replace logged real per-capita income in vector ${{\varvec{y}}}_{t}$, defined by expression (2), with our logged measures of real per-capita sectoral output, ${agric}_{t}$, ${ind}_{t}$, and ${serv}_{t}$, as defined in (C1). We add all three variables to the model at once. Our new vector ${{\varvec{y}}}_{t}$ therefore contains eight variables. We retain all other assumptions of the model as developed in Sect. 5.

In the resultant expanded VAR, we approach identification as follows. Intuitively, agricultural output, both arable and pastoral, reflects activity (planting, acquisition of cattle etc.) from many months ago. Shocks to the industrial and service sectors will thus normally not affect agricultural output in the same year. Similarly, industrial production relies heavily on the use of physical capital (machinery and equipment), the stock of which increases only with investment. Shocks to the service sector will therefore plausibly not affect industrial output in the same year. Accordingly, for identification, we place the new variables at the very end of the vector ${{\varvec{y}}}_{t}$ (the position occupied by ${income}_{t}$ in the main model) and order them as ${agric}_{t}$, ${ind}_{t}$, ${serv}_{t}$. We have verified, however, that our findings are fully robust to alternative orderings within the sub-vector of the three newly-included variables.

The relevant subset of the estimated (non-accumulated) impulse-responses is shown in Fig. 12 . As emphasized in Sect. 6.1, innovations in caselaw on families and inheritance increase per-capita output of the industrial sector (Figs. 12e, f), but do not exert a discernible effect on per-capita output in agriculture and services sectors (Fig. 12b, c, h, i). Innovations in caselaw on land have a discernible negative effect on per-capita output of the agricultural and industrial sectors (Fig. 12a, b). However, only the negative effect on the agricultural sector is statistically significant at extended horizons.

Fig. 12

Responses (non-accumulated) of sectoral real per-capita output variables to law shocks. 90-percent confidence intervals computed using Kilian's (1998) bootstrap method

Appendix D

This appendix presents and discusses the (non-accumulated) impulse-responses of attention to different areas of caselaw to law shocks. The results are summarized in Fig. 13.

Fig. 13

The non-accumulated responses of law variables to law shocks. 90-percent confidence intervals computed using Kilian's (1998) bootstrap method

The response in attention to law in one domain as a consequence of a shock in attention to law in another domain shows the coevolutionary character of caselaw. For example, when innovations in one area of law require complementary adjustments in another area of law, then shocks to the first area of law will lead to heightened attention in the second area of law. Law on inheritance and law on families evidence such coevolution (Fig. 13f, g). Innovations in one area of law can resolve issues in another area of law, implying that shocks to the first area of law will lead to reduced attention to the second area of law. In addition, reductions in attention in a given area of law following a shock in another area of law could be a consequence of substitution effects arising from constraints on the time and resources of the courts. While only a subset of the corresponding impulse-responses is statistically significant, this is the nature of the relationship between either law on families or law on inheritance on the one side and law on land on the other side (Fig. 13b, d, g and c).

Finally, following a shock to any single legal domain, later attention to that legal domain initially rises and then gradually declines (Fig. 13a, e and i). External shocks, such as legislation, thus do not immediately result in the resolution of all problems, but rather cause heightened attention to the relevant area of law for an extended time. This is especially true for law on land.

Appendix E

See Figs. 14 and 15

Fig. 14

Structural shocks (averages for five-year bins) to the three areas of law

Fig. 15

Structural shocks (averages for five-year bins) to demographic and economic variables

Appendix F

See Figs. 16 , 17 and 18

Fig. 16

The time series of the attention to the topic Non-Translated Latin

Fig. 17

Impulse-responses (non-accumulated) when replacing the Family Law series with the Non-Translated Latin (NTL) series. 90-percent confidence intervals computed using Kilian's (1998) bootstrap method

Fig. 18

Impulse-responses (accumulated) when replacing the Family Law series with the Non-Translated Latin (NTL) series. 90-percent confidence intervals computed using Kilian's (1998) bootstrap method